Defining the role of T cell dependent shaping of the microbiota to prevent dysbiosis and disease

Harmful shifts within the microbiota, termed dysbiosis, are associated with an array of diseases that plague modern society. Dysbiosis may include loss of diversity, expansion of pathobionts, and reductions of beneficial microbes that are required for symbiosis with the host. While one or more of these alterations are identified in correlation to many disorders, the impetus for these dysfunctional communities has remained largely uncharacterized. One pathway that has likely evolved to maintain healthy microbiota populations is secretory immunoglobulin A (IgA), which shapes commensal communities and maintains tolerant hostmicrobe responses. Intestinal IgA is produced at high levels as a result of T follicular helper cell (TFH) and B cell interactions in germinal centers along the gut, however, the molecular cues that govern T cell dependent IgA are largely unknown. This dissertation has uncovered a mechanism by which T cells use innate signaling through Toll-like receptors (TLRs) and Myeloid differentiation primary response gene 88 (MyD88) in order to detect shifts within the microbiota and direct IgA secretion. Loss of this pathway diminishes high-affinity IgA targeting of the microbiota and fails to constrain bacterial communities, leading to dysbiosis. Animals lacking MyD88 signaling within T cells are more prone to developing increased colitis severity and metabolic syndrome, both of which can be treated by rescuing the dysbiotic microbiota to a healthy state. These findings identify that T cells converge innate and adaptive immune signals to coordinate IgA against the microbiota. This maintenance of healthy microbiota communities has widespread impacts on immune development and metabolism, highlighting the wide variety of physiological pathways that have evolved to depend on symbiotic host-microbiota interactions in order to promote health and prevent diseases.

DEFINING THE ROLE OF T CELL DEPENDENT SHAPING OF THE MICROBIOTA TO PREVENT DYSBIOSIS AND DISEASE by Charisse N. Petersen A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Microbiology and Immunology Department of Pathology The University of Utah May 2017 Copyright © Charisse N. Petersen 2017 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The dissertation of Charisse N. Petersen has been approved by the following supervisory committee members: 03/10/2017 June Louise Round , Chair Matthew A. Williams , Member Janis J. Weis , Member 02/24/2017 Matthew A. Mulvey , Member 02/24/2017 John F. Valentine , Member 03/10/2017 and by Peter E. Jensen the Department/College/School of Date Approved Date Approved Date Approved Date Approved Date Approved , Chair/Dean of Pathology and by David B. Kieda, Dean of The Graduate School. ABSTRACT Harmful shifts within the microbiota, termed dysbiosis, are associated with an array of diseases that plague modern society. Dysbiosis may include loss of diversity, expansion of pathobionts, and reductions of beneficial microbes that are required for symbiosis with the host. While one or more of these alterations are identified in correlation to many disorders, the impetus for these dysfunctional communities has remained largely uncharacterized. One pathway that has likely evolved to maintain healthy microbiota populations is secretory immunoglobulin A (IgA), which shapes commensal communities and maintains tolerant hostmicrobe responses. Intestinal IgA is produced at high levels as a result of T follicular helper cell (TFH) and B cell interactions in germinal centers along the gut, however, the molecular cues that govern T cell dependent IgA are largely unknown. This dissertation has uncovered a mechanism by which T cells use innate signaling through Toll-like receptors (TLRs) and Myeloid differentiation primary response gene 88 (MyD88) in order to detect shifts within the microbiota and direct IgA secretion. Loss of this pathway diminishes high-affinity IgA targeting of the microbiota and fails to constrain bacterial communities, leading to dysbiosis. Animals lacking MyD88 signaling within T cells are more prone to developing increased colitis severity and metabolic syndrome, both of which can be treated by rescuing the dysbiotic microbiota to a healthy state. These findings identify that T cells converge innate and adaptive immune signals to coordinate IgA against the microbiota. This maintenance of healthy microbiota communities has widespread impacts on immune development and metabolism, highlighting the wide variety of physiological pathways that have evolved to depend on symbiotic host-microbiota interactions in order to promote health and prevent diseases. iv Dedicated to my three M’s Petersen. TABLE OF CONTENTS ABSTRACT ........................................................................................................... iii LIST OF TABLES ................................................................................................ viii LIST OF FIGURES ............................................................................................... ix ACKNOWLEDGEMENTS .....................................................................................xii Chapters 1. INTRODUCTION ............................................................................................... 1 References ................................................................................................. 4 2. DEFINING DYSBIOSIS AND ITS INFLUENCE ON HOST IMMUNITY AND DISEASE ............................................................................................................... 7 Summary .................................................................................................... 8 What Is Dysbiosis ....................................................................................... 8 Rebiosis: Establishing a Microbial Community Back to a Healthy State .. 14 Acknowledgements .................................................................................. 14 References ............................................................................................... 15 3. MYD88 SIGNALING IN T CELLS DIRECTS IGA-MEDIATED CONTROL OF THE MICROBIOTA TO PROMOTE HEALTH .................................................... 18 Summary .................................................................................................. 19 Introduction ............................................................................................... 19 Results ...................................................................................................... 20 Discussion ................................................................................................ 26 Experimental Procedures ......................................................................... 27 Accession Number ................................................................................... 28 Supplemental Information ......................................................................... 28 Author Contributions ................................................................................. 28 Acknowledgements .................................................................................. 28 References ............................................................................................... 28 Supplemental Information ......................................................................... 30 4. DIRECT RECOGNITION OF TLR2 LIGANDS INFLUENCES TFH DIFFERENTIATION AND GERMINAL CENTER DEVELOPMENT .................... 48 Introduction ............................................................................................... 49 Results ...................................................................................................... 51 Discussion ................................................................................................ 74 Experimental Procedures ......................................................................... 77 References ............................................................................................... 83 5. GUT IMMUNITY REGULATES INTER-MICROBIOTA INTERACTIONS THAT PREVENT METABOLIC DISEASE ..................................................................... 85 Introduction ............................................................................................... 86 Results ...................................................................................................... 88 Discussion .............................................................................................. 112 Experimental Procedures ....................................................................... 114 References ............................................................................................. 122 Appendix A ............................................................................................. 126 5. 6. DISCUSSION .............................................................................................. 135 References ............................................................................................. 140 vii LIST OF TABLES Tables 2.1. Antibiotic induced alterations to the microbiota……………………………....10 3.S1 Table of p-values for microbital community dissimilarities, related to Figures 4 and 5.........................................................................................................36 3.S2 9 OTUs identified as significantly different in separately housed animals as well as cohoused animals, related to Figure 5D..........................................37 LIST OF FIGURES Figures 2.1 A loss of beneficial microbes, expansion of pathobionts, and loss of diversity are events that encompass dysbiosis ……………………………………..…. 11 3.1 T cells utilize MyD88 signaling to coordinate germinal center responses .................................................................................................................... 20 3.2 Microbiota-dependent Tfh development relies on T cell-intrinsic MyD88 signaling ………………………………… ..............................………………... 21 3.3 IgA production against commensal antigens is perturbed in the absence of MyD88 signaling in T cells .......................................................................... 22 3.4 Innate recognition by T Cells influences the composition of mucosally associated microbial communities .............................................................. 23 3.5 T cell-intrinsic MyD88 signaling regulates IgA selection of the microbiota .................................................................................................................... 24 3.6 Phenotypic variation in immune responses correlates with microbial community similarity ................................................................................... 25 3.7 Increased intestinal disease observed in T-MyD88-/- animals is dependent on the composition of the microbiota .......................................................... 26 3.S1 T-MyD88-/- mice lack expression of MyD88 specifically within T cells (CD4+ and CD8+), related to Figure 1 ................................................................... 31 3.S2 Analysis of Th1, Th17, and Treg population frequencies within the spleens and mesenteric lymph nodes (MLNs) of WT and T-MyD88-/- mice, related to Figure 1 …............................................................................................... 32 3.S3 Analysis of Th1, Th17, and Treg population frequencies within cLP, siLP, and PPs of WT and T-MyD88-/- mice, related to Figure 1 ........................... 33 3.S4 Analysis of PP cellularity and Tfh populations within PPs, spleens, and MLNs, related to Figure 1............................................................................. 34 3.S5 Most significant taxonomic differences in mucosa, IgA bound and fecal communities between WT and T-MyD88-/-, related to Figure 4 and 5 …..... 35 4.1 T-MyD88-/- PPs have less ICOS+ T cells ................................................... 53 4.2 ICOS expression becomes increasingly defective on T-MyD88-/- T cells as they differentiate to Tfh cells ....................................................................... 54 4.3 TLR2 ligands alone promote ICOS expression in vivo in T cells with MyD88 signaling ...................................................................................................... 56 4.4 T cells directly recognize TLR2 ligands to increase ICOS expression in vitro .................................................................................................................... 57 4.5 OVA immunization of T-MyD88-/- mice results in significant defects in Tfh and GC responses ...................................................................................... 61 4.6 T cell intrinsic MyD88 signaling promotes GC responses through a combination of cell intrinsic and extrinsic mechanisms ............................... 63 4.7 MyD88 promotes Tfh differentiation in an ICOS dependent manner .......... 64 4.8 MyD88 signaling promotes proliferation to maintain Tfh populations ......... 66 4.9 Specific deletion of TLR2 within T cells was created by crossing a TLR2fl/fl mouse to a CD4-cre+ mouse ...................................................................... 68 4.10 TLR2 signaling on T cells promotes GC responses within the systemic compartment ............................................................................................... 69 4.11 TLR2 signaling promotes commensal specific immune responses in extraintestinal sites ............................................................................................. 72 5.1 Mice lacking MyD88 signaling within T cells develop obesity ..................... 89 5.2 Obese T-MyD88-/- mice have greater inflammation within adipose tissue 92 5.3 Dysbiosis within T-MyD88-/- mice is associated with spontaneous weight gain ............................................................................................................. 95 5.4 T cell shaping of the microbiota is associated with spontaneous weight gain .................................................................................................................. 100 5.5 Dysbiosis within T-MyD88-/- mice transfers obesity to WT animals ......... 105 x 5.6 Expansion of Desulfovibrio and reductions in beneficial spore-forming microbes are associated with obesity....................................................... 108 5.S1 Mice lacking MyD88 signaling within T cells develop obesity................... 126 5.S2 Obese T-MyD88-/- mice have greater inflammation within adipose tissue .................................................................................................................. 128 5.S3 Dysbiosis within T-MyD88-/- mice is associated with spontaneous weight gain........................................................................................................... 129 5.S4 T cell shaping of the microbiota is associated with spontaneous weight gain ...................................................................................................................131 5.S5 Dysbiosis within T-MyD88-/- mice transfers obesity to WT animals..........133 5.S6 Expansion of Desulfovibrio and reductions in beneficial spore-forming microbes are associated with obesity........................................................134 xi ACKNOWLEDGEMENTS I would like to acknowledge my mentor, June Round, for her support, inspiration and guidance throughout the last six years. She has lead by example through her work ethic and high standards, and it has been an honor to learn from such a successful scientist. I would like to thank my family for supporting me and giving me all the tools that I would need in order to succeed in graduate school. I would especially like to thank Mark for being there every step of the way. I would not have been able to do any of this without him. Lastly, I want to acknowledge all of those who have come before me to pave the way for more women in science. CHAPTER 1 INTRODUCTION 2 The origin of the adaptive immune system coincided with a dramatic increase in both the diversity and abundance of the microbiome, expanding the metabolic and genetic capacity of commensal communities (1). Despite traditional views that our immune system was designed to battle pathogens and maintain sterility, it has been hypothesized that the adaptive arm instead evolved to foster this increased symbiotic complexity while limiting infectious risks. Due to this relationship, a flexible dialogue has emerged between the host and resident microbes that shapes both immune development and the microbiota architecture. This communication is required to mold and maintain beneficial commensal communities while inducing tolerant immune responses to the microbiota. Seminal studies have provided evidence for the deleterious impact of failures within host-microbial communications that lead to perturbations to the microbiota, termed dysbiosis (2, 3). Indeed, dysbiosis has been implicated in allergic, autoimmune, and metabolic diseases, making involvement of the microbiota in disease development and persistence an ideal target for preventative and therapeutic treatments (4-6). While recent investigations have uncovered some elements of host-microbe interactions that initiate some of these disorders, the extent of the cross-talk between our immune system and microbiota is complex, and therefore largely unexplored. In order to ensure beneficial microbiota composition, the adaptive immune system has evolved a highly specific form of selection through the production of Immunoglobulin A (IgA). Secretory IgA (sIgA) is the most abundant antibody found within mucosal sites and performs a variety of valuable functions, including 3 barrier formation by sterically hindering microbe attachment to host cells, entrapment and elimination of pathogenic microbes, and modulation of antigenic expression within commensals (7, 8). All of these responses result in little activation of the host immune system, and are therefore thought to shape the microbiota while maintaining tolerance to it. Production of sIgA occurs via two mechanisms that differ by their dependence on T cells (9). Given the proper cues, B cells trafficking throughout the gut enact class switch recombination (CSR) to produce IgA capable of targeting a broad range of microbial antigens with moderate to low affinity. However, when T cells interact with B cells to direct antibody production, CSR within B cells is accompanied by rapid activation and cell death that results in a phenomenon called somatic hyper mutation (SHM). The end product of these reactions is a highly antigen specific sIgA that is capable of binding its targets with high affinity. T cells are multifaceted effector cells of the adaptive immune system that have the ability to modulate their responses to external cues. Notably, T cell development is dependent on the microbiota (10-12). While classically studied for their dependence on antigen presenting cells (APCs) for activation, the identification that T cells express Toll like receptors (TLRs), which recognize conserved microbial patterns, suggests that direct sensing of microbial products could provide a mechanism to influence T cell responses (13). This would be particularly relevant within the gut where an abundant source of commensal ligands exists. Previous studies support this idea through the discovery that a single commensal species, Bacteroides fragilis, utilizes TLR2 signaling to 4 promote its own colonization while enhancing T cell induced tolerance to the microbiota (14). Most TLRs signal through a common, downstream adaptor protein called Myeloid differentiation primary response gene 88 (MyD88) (15). Therefore, in order to interrogate the impact of direct recognition of microbial ligands on T cell function within the gut, we created a mouse that contains a specific deletion of MyD88 within T cells. We hypothesized that direct recognition of microbial products directs T cell development within the gut in order to shape the microbiota architecture and induce tolerant immune responses towards commensals. Indeed, we have discovered that mice lacking T cell specific MyD88 have a defective T cell-dependent IgA response and reduced ability to shape healthy microbiota communities (16). The resulting dysbiosis that occurs within these animals leads to a greater susceptibility to both colitis and the onset of obesity and insulin resistance. Providing these animals with a fecal microbial transplant (FMT) or recolonization with beneficial microbial communities is capable of rescuing disease. Thus, this dissertation demonstrates that direct recognition of microbial ligands on T cells is vital to proper IgA targeting within the gut to maintain beneficial microbiota structure and promote health. References 1. M. McFall-Ngai, Adaptive immunity: Care for the community. Nature 445, 153 (2007). 2. F. A. Carvalho et al., Transient inability to manage proteobacteria promotes chronic gut inflammation in TLR5-deficient mice. Cell Host Microbe 12, 139-152 (2012). 5 3. A. Couturier-Maillard et al., NOD2-mediated dysbiosis predisposes mice to transmissible colitis and colorectal cancer. J. Clin. Invest. 123, 700-711 (2013). 4. N. Kamada, S. U. Seo, G. Y. Chen, G. Nunez, Role of the gut microbiota in immunity and inflammatory disease. Nature Rev. Immunol. 13, 321-335 (2013). 5. V. K. Ridaura et al., Gut microbiota from twins discordant for obesity modulate metabolism in mice. Science 341, 1241214 (2013). 6. C. Petersen, J. L. Round, Defining dysbiosis and its influence on host immunity and disease. Cell. Microbiol. 16, 1024-1033 (2014). 7. N. J. Mantis, N. Rol, B. Corthesy, Secretory IgA's complex roles in immunity and mucosal homeostasis in the gut. Mucosal Immunol. 4, 603611 (2011). 8. J. L. Kubinak, J. L. Round, Do antibodies select a healthy microbiota? Nature Rev. Immunol. 16, 767-774 (2016). 9. O. Pabst, New concepts in the generation and functions of IgA. Nature Rev. Immunol. 12, 821-832 (2012). 10. Ivanov, II et al., Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell 139, 485-498 (2009). 11. K. Atarashi et al., Induction of colonic regulatory T cells by indigenous Clostridium species. Science 331, 337-341 (2011). 12. J. L. Round, S. K. Mazmanian, Inducible FoxP3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc. Natl. Acad. Sci. U. S. A. 107, 12204-12209 (2010). 13. D. Kabelitz, Expression and function of Toll-like receptors in T lymphocytes. Curr. Opin. Immunol. 19, 39-45 (2007). 14. J. L. Round et al., The Toll-like receptor 2 pathway establishes colonization by a commensal of the human microbiota. Science 332, 974977 (2011). 15. S. Rakoff-Nahoum, J. Paglino, F. Eslami-Varzaneh, S. Edberg, R. Medzhitov, Recognition of commensal microflora by toll-like receptors is required for intestinal homeostasis. Cell 118, 229-241 (2004). 6 16. J. L. Kubinak et al., MyD88 signaling in T cells directs IgA-mediated control of the microbiota to promote health. Cell Host Microbe 17, 153-163 (2015). CHAPTER 2 DEFINING DYSBIOSIS AND ITS INFLUENCE ON HOST IMMUNITY AND DISEASE Reprinted Petersen, C. and Round, J. L. (2014), Defining dysbiosis and its influence on host immunity and disease. Cell. Microbiol., 16: 1024–1033 8 Cellular Microbiology (2014) 16(7), 1024–1033 doi:10.1111/cmi.12308 First published online 2 June 2014 Microreview Defining dysbiosis and its influence on host immunity and disease Charisse Petersen and June L. Round* Department of Pathology, Division of Microbiology and Immunology, University of Utah School of Medicine, Salt Lake City, UT 84112, USA Summary Mammalian immune system development depends on instruction from resident commensal microorganisms. Diseases associated with abnormal immune responses towards environmental and self antigens have been rapidly increasing over the last 50 years. These diseases include inflammatory bowel disease (IBD), multiple sclerosis (MS), type I diabetes (T1D), allergies and asthma. The observation that people with immune mediated diseases house a different microbial community when compared to healthy individuals suggests that pathogenesis arises from improper training of the immune system by the microbiota. However, with hundreds of different microorganisms on our bodies it is hard to know which of these contribute to health and more importantly how? Microbiologists studying pathogenic organisms have long adhered to Koch’s postulates to directly relate a certain disease to a specific microbe, raising the question of whether this might be true of commensal–host relationships as well. Emerging evidence supports that rather than one or two dominant organisms inducing host health, the composition of the entire community of microbial residents influences a balanced immune response. Thus, perturbations to the structure of complex commensal communities (referred to as dysbiosis) can lead to deficient education of the host immune system and subsequent development of immune mediated diseases. Here we will overview the literature that describes the causes of dysbiosis and Received 21 March, 2014; revised 29 April, 2014; accepted 30 April, 2014. *For correspondence. E-mail [e-mail]; Tel. [phone] (office) or ([phone] (lab). the mechanisms evolved by the host to prevent these changes to community structure. Building off these studies, we will categorize the different types of dysbiosis and define how collections of microorganisms can influence the host response. This research has broad implications for future therapies that go beyond the introduction of a single organism to induce health. We propose that identifying mechanisms to re-establish a healthy complex microbiota after dysbiosis has occurred, a process we will refer to as rebiosis, will be fundamental to treating complex immune diseases. What is dysbiosis? Our current knowledge of the architecture of a healthy microbiota comes from multiple studies in individuals with no overt signs of disease (Huttenhower et al., 2012). This structure includes Bacteroidetes and Firmicutes as the dominant bacterial phyla present in stool samples and Proteobacteria and Actinobacteria being a small but consistent presence in most people. Broadly defined, dysbiosis is any change to the composition of resident commensal communities relative to the community found in healthy individuals. In the last decade, a number of studies have documented significant changes in the structure of microbial communities in patients and mouse models of inflammatory bowel diseases (IBD) such as Crohn’s and ulcerative colitis (UC) (Frank et al., 2007), diabetes (Karlsson et al., 2013), asthma (Abrahamsson et al., 2013), allergies and even autism (Parracho et al., 2005). Given the emerging importance of the microbiota to host development, it is speculated that these observed changes in microbial composition are contributing factors to the initiation and/or persistence of many of these diseases. There are multiple ways that the structure of the microbial community can be influenced. This includes the genetics of the host, diet, infection, or medical interventions (such as antibiotics). The hygiene hypothesis originally proposed that antibiotic usage and lifestyle alterations that limit microbial exposure were predisposing populations of people in developed countries to © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. cellular microbiology 9 How changes in microbiota structure influence health autoimmune disease (Strachan, 2000). Therefore it was suggested that, due to the inability to differentiate between pathogen and commensal, antibiotics disturb microbiota structure and, subsequently, the coevolutionary relationship between our immune system and the symbionts we host. Since then a significant amount of research has gone into exploring the functional importance of the microbiota and the influence antibiotic treatment has on its architecture (Ubeda et al., 2010; Buffie et al., 2012; Cho and Blaser, 2012). Interestingly, antibiotics commonly used in the clinic have significant impacts on our microbial residents (Table 1). Importantly, many antibiotics have long lasting effects on the microbiota, leading to the permanent loss of some organisms, while others outgrow and persist. While these studies have documented dysbiosis in the context of disease or antibiotic use, recent investigations have begun to identify causality between dysbiosis and disease progression. Collectively, these studies now allow us to better understand what dysbiosis is and begin to categorize it into three types that will be discussed in depth here. These include (i) loss of beneficial microbial organisms, (ii) expansion of pathobionts or potentially harmful microorganisms and (iii) loss of overall microbial diversity (Fig. 1). These three types of dysbiosis are not mutually exclusively and may all occur concurrently. This review is intended to help synthesize the current data to allow a more defined understanding of what dysbiosis is and how it influences health, immune system development and disease. Loss of beneficial microbial organisms It has become well established that the microbiota is important for the maturation and development of appropriate intestinal immune responses (Round and Mazmanian, 2009; Cho and Blaser, 2012; Hooper et al., 2012). Recent studies have begun to identify the organisms and their mechanisms responsible for eliciting immune development within the host. The immune response must be carefully balanced between the inflammation required for pathogen eradication and tolerant reactions that prevent unwanted immune responses towards self tissue and commensals. Tolerant mechanisms evolved by the host include production of mucus and antimicrobial peptides to create a barrier between host tissue and microbes (Johansson et al., 2008; 2011; Vaishnava et al., 2011), modification of bacterial products to make them less immunogenic (Bates et al., 2007), production of antibodies that can directly bind to and influence microbial function (Fagarasan and Honjo, 2003; Peterson et al., 2007; Slack et al., 2012) and anti-inflammatory T cells. Interestingly, specific members from the microbiota have been identified 1025 that can directly co-ordinate many aspects of the host immune response. Tolerance to self tissue and resident commensals is, in part, governed by a specialized subset of T lymphocytes termed T regulatory cells (Tregs) (Josefowicz et al., 2012). These cells are marked by the expression of a transcription factor, Foxp3 (Fontenot et al., 2003). Mutations in the Foxp3 gene result in a loss of Treg development in both mouse and humans and subsequent inflammation and autoimmunity at multiple organ sites, demonstrating the importance of Treg populations to host health. Several recent investigations have identified resident microorganisms that are capable of inducing these cell types within the intestine of the host (Round and Mazmanian, 2010; Atarashi et al., 2011). Initial experiments demonstrated that Treg function is compromised in germfree mice and can be restored by either monoassociation with the human commensal Bacteriodes fragilis or with a defined mix of Clostridium strains from groups IV and XIVa. Colonization of animals with either of these organisms protects animals from colitis in a Tregdependent manner through distinct mechanisms. The Clostridium strains induce TGF-β expression from intestinal epithelial cells (IECs) to enhance the differentiation of inducible colonic Tregs (Atarashi et al., 2013); while B. fragilis utilizes a capsular polysaccharide, PSA, that binds to TLR2 on both dendritic cells and T cells to induce IL-10 and enhance the suppressive capability of Tregs (Round et al., 2011; Shen et al., 2012). Other Bacteriodes strains have also been reported to increase the total numbers of Tregs within the colon, including B. caccae, B. thetaiotaomicron, and B. vulgaris, although this is not seen in all systems (Faith et al., 2014). Several other strains of commensal bacteria have evolved the capacity to induce Tregs and confer protection from inflammatory disease. These include multiple strains of Lactobacillus such as L. acidophilus and several strains of Bifidobacterium, including B. breve, likely making induction of Tregs a common mechanism employed by bacteria to induce tolerance in the gut. The mechanisms by which these many of these organisms influence Treg biology remain untested but likely include bacterial derived products such as short-chain fatty acids (SCFAs) (Schmidt et al., 2010). SCFAs are microbial fermentation products from dietary fibre and include propionate, acetate, butyrate, and formate. These products have been shown by multiple groups to be responsible for regulating the Treg pool within the colon and provide protection from experimental colitis (Arpaia et al., 2013; Furusawa et al., 2013; Smith et al., 2013). Butyrate and propionate, but not acetate, seem to be the most prominent SCFAs capable of driving Treg responses. Butyrate, in particular, is capable of blocking histone deacetylases that promote DNA condensation (HDACs) and can thereby regulate © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 • Quinoline: Inhibits DNA replication • Broad Spectrum • β-Lactam antibiotic: inhibits peptidoglycan synthesis • Narrow Spectrum • Aminoglycoside: Inhibit peptide synthesis • Narrow Spectrum • Widespread reductionsC,g • FirmicutesC,f • Widespread reductionsM&NM,d • Actinobacteria (Slackia and Bifidobacterium), Betaproteobacteria, Streptococcus spp., Roseburia, EubacteriumC,f • Lactobacillus spp, Enterococcus spp., Group D StreptococcusM,c • Lactobacillus spp., Enterococcus spp., Group D StreptococcusM,c • Bacteroidales, Ruminobcoccaceae, Lachnospiraceae, ClostridialesM&NM,d • Tenericutes, TuricibacteriaNM,d • Lactobacillus plantarum, Faecalibacterium prausnitzii, Eubacterium hallii, Clostridium cluster IV and XIVa, Escherichia coli, Haemophilus spp., SerratiaC,e • Clostridiales, Fecaelibacterium spp.C,g • Lactobacillus spp., Enterococcus spp., Enterobacteriaceae spp.NR,b • EnterobacteriaceaeM,c • EnterobacteriaceaeM,c • Lactobacillaceae, VerrucomicrobiaceaeNM,d • Paenibacillaceae, Firmicutes, AnaeroplasmataceaeM,d Community expansion Antibiotics commonly used in the clinic lead to alterations in the composition of the microbiota. This table describes how various antibiotics influence specific bacterial communities. M, mouse; NM, neonatal mouse; NR, neonatal rat; C, clinical. a. Mangin et al. (2010) Anaerobe 16: 433–438. b. Schumann et al. (2005) Physiol Genomics. c. Sekirov et al. (2008) Infect Immun. d. Russell et al. (2012) EMBO Rep. e. Vreize et al. (2013) J Hepatol. f. Perez-Cobas et al. (2013) Gut. g. Dethlefsen et al. (2008) PLoS Biol. Streptomycin Cephalosporin (Cephazolin) Ciprofloxacin Vancomycin • Bifidobacterium • β-Lactam antibiotic: inhibits peptidoglycan synthesis • Extended spectrum • Glycopeptide: Inhibits peptidoglycan cross-linkage Amoxicillin Community loss Reduced diversity Activity/spectrum Name C,a Alterations to the microbiota Antibiotic 1026 Table 1. Antibiotic induced alterations to the microbiota. 10 C. Petersen and J. L. Round © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 11 How changes in microbiota structure influence health Homeostasis Dysbiosis Pathobiont expansion Reduced diversity 1027 Fig. 1. A loss of beneficial microbes, expansion of pathobionts, and loss of diversity are events that encompass dysbiosis. During healthy, homeostatic conditions the microbiota is composed of a diversity organisms that are known to benefit host development and health. However, environmental insults, such as antibiotic use or diet can lead to disruptions in the structure of the microbial community. These disruptions can lead to a loss of organisms that are beneficial to the host and a subsequent overgrowth of commensals that have the potential to cause harm, termed pathobionts. Domination of the microbiota by pathobionts can lead to inflammation and pathology. Additionally, multiple studies have described the diversity of contributions made by the various members of the microbiota. Oftentimes, these are non-redundant influences on host health, thus a total loss of diversity in the microbiota can also influence disease progression or severity and thus also represents a dysbiosis event. Loss of beneficial microbes gene expression. Interestingly, if T cells are treated with butyrate under Treg inducing conditions in vitro, there is a marked increase in acetylation of the Foxp3 promoter region as well as enhancer elements thereby allowing Foxp3 to be expressed. SCFAs are sensed by the host through a variety of receptors, including G-protein coupled receptors GPCR43 and GPR109a (Smith et al., 2013; Singh et al., 2014). Not surprisingly, GPCR43 expression is increased on Tregs specifically within intestinal tissues (Smith et al., 2013). Consequently, loss of these receptors leads to susceptibility of animals to colonic inflammation and colon cancer. While Tregs are a key component of tolerance induction within the intestine, commensal organisms have evolved other mechanisms to suppress inflammation. Patients with IBD have elevated levels of inflammatory cytokines such as TNF-α and IP-10. As discussed above, the levels of these cytokines are often opposed by the induction of Tregs, and this is the mechanism utilized by the host to maintain homeostasis within the gut. Recently, however, a group uncovered a mechanism by which commensal bacteria can directly reduce inflammation by targeting the cytokines themselves. Lactobacillus paracasei and L. casei both encode a protease called lactoceptin that specifically degrades the inflammatory cytokine IP-10 (von Schillde et al., 2012). Degradation of secreted and cellassociated IP-10 lead to reduced lymphocyte recruitment during an animal model of ileitis. Thus, members of the microbiota have evolved ways to directly antagonize the effects of the host inflammatory response and therefore directly aid in the maintenance of intestinal homeostasis. Unlike T cells that recognize protein antigens, invariant NKT cells (iNKT) recognize lipid antigens and play important roles in innate and adaptive inflammation. Interestingly, the development of these cell types is dynamically regulated by the microbiota such that GF mice have elevated numbers of these cells within the gut (Olszak et al., 2012). Animals containing increased numbers of iNKT cells are more susceptible to colitis, thus the identification of how the microbiota functions to suppress iNKT cell numbers in the gut could lead to an important therapeutic intervention. A recent report demonstrated that mono-association of germfree mice with B. fragilis is sufficient to suppress the development of this population (An et al., 2014). B. fragilis contains a gene with high homology to enzymes involved in sphinoglipid biogenesis. Deletion of this gene in B. fragilis prevents the suppression of iNKT cell development, indicating that B. fragilis utilizes these lipids to influence iNKT cell biology. Indeed, these investigators purified sphingolipids from B. fragilis and showed that they are capable of reducing iNKT cell activation and protecting animals from disease. Importantly, early exposure to B. fragilis is required to inhibit iNKT cell development, and monocolonization of adult germfree mice is insufficient to protect against colitis. As discussed earlier, B. fragilis is also able to modulate Treg responses (Ochoa-Reparaz et al., 2010; Round and Mazmanian, 2010). These studies reveal that a single commensal organism can possess several mechanisms to positively influence host biology and highlights the complexity and intimacy of host-commensal relationships. © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 12 1028 C. Petersen and J. L. Round Clinical investigations provide support for these animal studies. Culture-independent analyses of patients with CD and UC have revealed a distinct loss of symbionts residing within the colonic mucosa as compared to healthy individuals. Specifically, loss of Clostridium groups XIVa and IV is observed within faecal and mucosal samples in IBD patients (Gophna et al., 2006; Frank et al., 2007). These observed reductions were seen consistently whether patients were actively experiencing clinical symptoms or within remission and were not correlated with recent antibiotics. This argues that the loss of beneficial microbes is an underlying issue and not simply a reflection of increased inflammation or treatment. Conflicting results have been reported for the phyla Bacteroidetes. Using mucosal tissues from small intestines and colons from 190 patients that included equal numbers of CD, UC, and healthy controls, Frank et al. identified approximately 800–1600 OTUs and observed a significant decrease in the OTUs and relative abundance of Bacteroidetes within IBD patients. Other studies including microbial identification from faecal samples have observed increases in Bacteroidetes within IBD patients (Mangin et al., 2004; Gophna et al., 2006). This, however, may reflect differences in culture-dependent versus -independent techniques, the depth of microbial sequence sampling, or tissue versus faecal sampling differences. Collectively, these studies identify important ways in which commensal bacteria positively influence mammalian biology, thus loss of these organisms represents an important aspect of dysbiosis. Expansion of pathobionts The microbiota also contains members that have the capacity to cause harm to the host. These organisms have been termed pathobionts to indicate their potential to cause pathology (Chow and Mazmanian, 2010). Pathobionts are typically kept at low levels within a healthy gut and do not cause problems in immunecompetent hosts; however, there are several examples that outgrowth of these organisms can contribute to disease. Thus, expansion of pathobionts represents a second category of dysbiosis. The most often reported incidence of pathobiont expansion is that of Proteobacteria and in particular the family Enterobacteriaceae, which contains members such as Escherichia coli, Shigella and Klebsiella (Lupp et al., 2007). This is true in multiple mouse models of colitis, including animal models carrying mutations within genes that are associated with IBD (Ayres et al., 2012). In particular, the deletion of TLR5, a Toll-like receptor (TLR) responsible for recognizing flagellar proteins, results in low grade inflammation, colitis and metabolic syndrome with incomplete penetrance (Vijay-Kumar et al., 2010; Carvalho et al., 2012; Cullender et al., 2013). These disease phenotypes are lost when animals are re-derived with a microbiota from Jackson labs, suggesting a role for the composition of the microbiota in driving disease. To identify microbial communities associated with colitis in TLR5 deficient animals, the microbiota was compared between colitic and non-colitic TLR5−/− animals. Interestingly, animals that developed spontaneous colitis also had a threefold increase in Proteobacteria. More importantly, members of Enterobacteriaceae, such as E. coli, were found to be preferentially associated with the mucosa and able to penetrate the tissue. To show the requirement for E. coli for colitis induction in TLR5−/− animals, they antibiotic treated TLR5−/− and WT animals and then treated the mice with an adherent invasive E. coli (AIEC) strain. TLR5−/− animals were unable to manage levels of these flagellated bacteria and harboured a 10-fold larger bacterial load. Moreover, all TLR5−/− animals receiving this strain succumbed to colitis, demonstrating that E. coli were sufficient to cause colitis in the absence of TLR5. Klebsiella pneumonia and Proteus mirabilis, both members of the Enterobacteriaceae family, also expanded to induce colitis within T-bet−/− X Rag2−/− mice (Garrett et al., 2007). Rag deficient animals lack any T or B cells and therefore crossing these mice causes a lack of Tbet in cells of the innate immune system. Interestingly, these species required an intact microbiota for disease induction and were not sufficient to induce colitis on their own. As a final example, when animals are treated with antibiotics and subsequently induced for colitis using DSS, expansion of a multi-drug-resistant E. coli strain occurs that is able to penetrate the intestinal mucosa and cause sepsis (Ayres et al., 2012). Multiple studies have demonstrated how these organisms are able to exploit the inflamed environment and expand their numbers, and these mechanisms have been reviewed elsewhere (Spees et al., 2013). Importantly, the expansion of Enterobacteriaceae is also seen in patients suffering from CD and UC in both tissue and faecal samples (Frank et al., 2007), consistent with what is seen in mouse models; however, much work needs to be done to better understand whether these organisms truly initiate disease or simply exacerbate disease progression. Other types of organisms have been reported to expand in other animal models of immune deficiency. NLRP6 is a member of the nucleotide-binding oligomerization domain protein like receptors (NLRs), a class of cytoplasmic receptors that recognize bacterial products. NLRP6−/− mice have an increased abundance of Prevotella spp. and TM7 and are highly susceptible to colitis induced by dextran sodium sulfate (DSS) (Elinav et al., 2011). The colitis is transmissible as co-housing NLRP6−/− and WT mice results in increased susceptibility © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 13 How changes in microbiota structure influence health to disease in the WT animals. Thus, the NLRP6 deficiency is not required for pathology, but rather allows for the expansion of a subset of organisms that promote inflammation. Finally, while the vast majority of our microbial inhabitants are bacteria, mammals are also colonized by fungal and viral members. In general, yeast are not susceptible to antibiotic treatment, and one common sideeffect of antibiotic use is the outgrowth of yeast strains such as Candida albicans. A recent study highlighted the importance of controlling fungal overgrowth to intestinal health by investigating the role of Dectin-1 within the gut (Iliev et al., 2012). Dectin-1 is a receptor that recognizes alpha-mannans from the cell wall of yeast, and mice deficient in this receptor are more susceptible to DSS colitis that can be treated with fluconazole, an antifungal drug that does not target bacteria. These data indicate that yeast outgrowth can worsen disease and our immune system has evolved elegant mechanisms to regulate yeast populations in the gut. Similarly, a study showed that antibiotic treatment of animals allowed overgrowth of C. albicans, which increased plasma concentrations of prostaglandin E2 (PGE2) (Kim et al., 2014). The increase in PGE2 caused susceptibility to allergies in these animals that could be suppressed when PGE2 synthesis was blocked. Thus, antibiotic treatment can allow for outgrowth of fungal species in the gut that can influence extra-intestinal disease within the host. Viral members of our microbiota are only just beginning to be identified, presenting new players in the complex relationship between genetic and environmental factors that lead to disease. Polymorphisms within the gene ATG16L1 confer a twofold higher risk to the development of IBD. When mice contain a hypomorphic mutation within ATG16L1, they experience histological abnormalities and increased colitis similar to Crohn’s disease patients carrying polymorphisms within the same gene (Cadwell et al., 2010). Interestingly, chronic infection with a specific strain of mouse norovirus is required for the observed phenotype within mice. Although this is not an example of expansion, it highlights the likelihood of unknown viral pathobionts playing a role in the development of dysbiosis associated diseases. It is unlikely that pathobiont expansion alone would cause disease, as decades of research have been dedicated to identifying a single organism capable of inducing IBD and none have been found. Supporting this, subclinical twins and immediate family members posses a microbiota that resembles that of a related IBD patient; however, none of these infectious agents had the ability to induce IBD within a healthy individual. Therefore, it is possible that concomitant reductions in beneficial microbes or a loss of diversity must also occur to promote disease. Collectively, these studies investigate both well- 1029 characterized and newly discovered pathobionts, capable of inciting harmful effects on the host when given the opportunity to expand. Loss of diversity: more is better There is evidence that members of the microbiota have diverse and non-redundant contributions to host health. For instance, some organisms promote the development of anti-inflammatory networks while others induce protective inflammatory responses. Additionally, while some host–microbe interactions can be initiated by multiple species, others involve a more unique relationship with exclusive community members (Faith et al., 2014).Therefore, gaining the maximum health benefits from the microbiota may require a more complex and diverse collection of organisms, and, indeed, several recent experiments support this notion. Thus, an important aspect of dysbiosis can also include a loss of total microbial diversity. As discussed above, Treg function is compromised in germfree animals. In order to identify organisms that complement this deficiency, diluted human stool samples enriched for chloroform-resistant bacteria were used to colonize germfree animals. Interestingly, these Clostridia species, containing over 30 different strains, induced a threefold expansion of Tregs within the gut over un-colonized controls. Importantly, Treg induction within the host was diminished when mice were colonized with only a single strain from this Clostridia collection, as compared to animals colonized with over 15 different Clostridia strains. These data suggest that a greater diversity of organisms, even within the same family, can maximally induce host cellular development (Atarashi et al., 2011; 2013). Whether these strains amplify Treg activity through the same pathway or work through distinct mechanisms remains unclear, but will be important to our understanding of how these organisms influence host health. Early exposure to these diverse communities of microbes appears to be an additional facet to maintaining a healthy host–microbiota relationship (Cho and Blaser, 2012; Cho et al., 2012). The antibody isotype, IgE, is associated with allergic responses, and animals with increased concentrations of IgE develop worsened allergic disease. Despite a reduction in all other isotypes, IgE is greatly elevated in the serum of germ-free mice, indicating a specific requirement for the microbiota in suppressing this antibody (Cahenzli et al., 2013). Colonization of germfree mice early in life with a complex microbiota reduces IgE levels and prevents allergy; however, if germfree animals are colonized during adult stages, suppression of IgE levels does not occur. To better understand the minimal microbial diversity required for © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 14 1030 C. Petersen and J. L. Round IgE suppression, germfree mice were first colonized by a single species, E. coli, which was incapable of suppressing IgE levels. Bi-colonization with two species, Lactobacillus murinus and Parabacteriodes distasonis, also failed to suppress IgE levels. When germfree animals were colonized with eight different organisms that included L. murinus and P. distasonis half of the animals still displayed hyper-IgE production. Suppression of IgE was only evident when the animals were colonized with up to 40 different strains of bacteria. Taken together, these studies suggest that a greater diversity of microbial organisms is required for maximal benefits to the host. While many studies have found correlations with changes in the microbiota abundance and allergy, few studies have been able to look at the microbiota early in life and then follow the same children out to see if they develop disease. However, a recent study was able to collect microbiota samples from children during there first week of life and again at 1 month and 12 months after birth (Abrahamsson et al., 2013). These same children were then assessed for allergic disease and asthma at age seven. Interestingly, the children that developed asthma all had lower microbial diversities early in life when compared to the non-asthmatic children, suggesting that early life microbial diversity might protect from disease later in development. Rigorous studies still need to be performed; however, these studies suggest that increased microbial diversity is better. While we have discussed three distinct types of dysbiosis, it is likely that dysbiosis encompasses all three of these manifestations concomitantly to influence disease. Rebiosis: establishing a microbial community back to a healthy state A handful of studies have now demonstrated that replacement of beneficial bacterial species can protect and even cure diseases in animal model systems and human patients (Mazmanian et al., 2008; Sokol et al., 2008; Ochoa-Reparaz et al., 2010; Round and Mazmanian, 2010). In humans, this has been best established for Clostridium difficile infections (Aas et al., 2003). C. difficile infection is a major cause of antibiotic-associated diarrhoea and becoming an increasing healthcare burden (Kachrimanidou and Malisiovas, 2011). C. difficile infections can cause a wide range of diseases that include severe diarrhoea, pseudomembranous colitis, and toxic megacolon. Infection is usually treated with vancomycin or metronidazole; however, in 25–30% of cases a recurrent disease follows, with each relapse increasing the likelihood of another recurrence. Interestingly, when a more narrow spectrum antibiotic is used that does less damage to the microbiota, there are lower recurrence rates of C. difficile, suggesting that the microbiota may play a protective role during infection. It has now been shown in both mouse models and, more importantly, humans that a faecal microbiota transplant (FMT) from a healthy donor can help displace C. difficile infection and prevent recurrences. After a large number of uncontrolled trials demonstrated high success rates for microbiota transplantation in resolving C. difficile infections, a randomized study recently revealed an 81% success rate following a single transplant with a 98% success rate after a second (van Nood et al., 2013; Vrieze et al., 2013; Pamer, 2014). These studies highlight the potential therapeutic value of identifying ways to restore a ‘healthy’ microbiota, a process we will refer to as rebiosis. There are several potential candidate approaches for rebiosis aside from FMT described above. Multiple investigations have sought to identify functional members of the microbiota that can be cultured and therefore used as a probiotic either in the form of a pill or embedded in food items. Given the importance of microbial diversity it is likely that no one single organism will work most effectively, but rather a complex assortment of organisms will provide the maximum benefit. Indeed, yogurts and probiotics are beginning to incorporate multiple strains of organisms. The downside to this therapy is that it relies on the ability to culture these organisms within the lab, yet it is estimated that only 20–30% of these organisms are culturable, meaning that a large, likely very important constituent of the microbiota will not be sampled. Another method to maintain or restore a healthy microbiota could potentially be the use of pre-biotics, which are nondigestible ingredients that stimulate the growth and activity of certain bacterial groups. There are a few known substances that meet these criteria, such as inulin, but much work is needed to identify the best nutritional sources for the microbiota. A better understanding of how host immune pathways influence the structure of the microbial community might also open up new avenues for improved therapies for manipulating microbial communities. Mammalian immune systems have acquired elegant mechanisms for mounting antigen-specific immunity against pathogens and therefore might also use similar mechanisms to control the microbiota. If true, then engineered therapies that target specific microbes while keeping the rest of the microbiota intact would be possible. The microbiota represents an important aspect of human health that we are only beginning to understand; therefore, much work lies ahead to tap into the therapeutic value of these microbial symbionts. Acknowledgements These authors have no conflict of interest. We would like to thank members of the Round and O’Connell labs for their critical review of the manuscript. C.P. is supported © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 15 How changes in microbiota structure influence health by a T32 fellowship in microbial pathogenesis (AI-055434). Work in the lab is supported by the Edward Mallinckrodt Jr. Foundation, Pew Scholars Program, NSF CAREER award (IOS-1253278), Packard Fellowship in Science and Engineering, NIAID (AI95375), NIAID (AI107090), and NIAID (AI95375) to J.L.R. References Aas, J., Gessert, C.E., and Bakken, J.S. (2003) Recurrent Clostridium difficile colitis: case series involving 18 patients treated with donor stool administered via a nasogastric tube. Clin Infect Dis 36: 580–585. Abrahamsson, T.R., Jakobsson, H.E., Andersson, A.F., Bjorksten, B., Engstrand, L., and Jenmalm, M.C. (2013) Low gut microbiota diversity in early infancy precedes asthma at school age. Clin Exp Allergy doi:10.1111/ cea.12253 [Epub ahead of print]. An, D., Oh, S.F., Olszak, T., Neves, J.F., Avci, F.Y., Erturk-Hasdemir, D., et al. (2014) Sphingolipids from a symbiotic microbe regulate homeostasis of host intestinal natural killer T cells. Cell 156: 123–133. Arpaia, N., Campbell, C., Fan, X., Dikiy, S., van der Veeken, J., DeRoos, P., et al. (2013) Metabolites produced by commensal bacteria promote peripheral regulatory T-cell generation. Nature 504: 451–455. Atarashi, K., Tanoue, T., Shima, T., Imaoka, A., Kuwahara, T., Momose, Y., et al. (2011) Induction of colonic regulatory T cells by indigenous Clostridium species. Science 331: 337–341. Atarashi, K., Tanoue, T., Oshima, K., Suda, W., Nagano, Y., Nishikawa, H., et al. (2013) Treg induction by a rationally selected mixture of Clostridia strains from the human microbiota. Nature 500: 232–236. Ayres, J.S., Trinidad, N.J., and Vance, R.E. (2012) Lethal inflammasome activation by a multidrug-resistant pathobiont upon antibiotic disruption of the microbiota. Nat Med 18: 799–806. Bates, J.M., Akerlund, J., Mittge, E., and Guillemin, K. (2007) Intestinal alkaline phosphatase detoxifies lipopolysaccharide and prevents inflammation in zebrafish in response to the gut microbiota. Cell Host Microbe 2: 371–382. Buffie, C.G., Jarchum, I., Equinda, M., Lipuma, L., Gobourne, A., Viale, A., et al. (2012) Profound alterations of intestinal microbiota following a single dose of clindamycin results in sustained susceptibility to Clostridium difficile-induced colitis. Infect Immun 80: 62–73. Cadwell, K., Patel, K.K., Maloney, N.S., Liu, T.C., Ng, A.C., Storer, C.E., et al. (2010) Virus-plus-susceptibility gene interaction determines Crohn’s disease gene Atg16L1 phenotypes in intestine. Cell 141: 1135–1145. Cahenzli, J., Koller, Y., Wyss, M., Geuking, M.B., and McCoy, K.D. (2013) Intestinal microbial diversity during early-life colonization shapes long-term IgE levels. Cell Host Microbe 14: 559–570. Carvalho, F.A., Koren, O., Goodrich, J.K., Johansson, M.E., Nalbantoglu, I., Aitken, J.D., et al. (2012) Transient inability to manage proteobacteria promotes chronic gut inflammation in TLR5-deficient mice. Cell Host Microbe 12: 139–152. Cho, I., and Blaser, M.J. (2012) The human microbiome: at the interface of health and disease. Nat Rev Genet 13: 260–270. 1031 Cho, I., Yamanishi, S., Cox, L., Methe, B.A., Zavadil, J., Li, K., et al. (2012) Antibiotics in early life alter the murine colonic microbiome and adiposity. Nature 488: 621–626. Chow, J., and Mazmanian, S.K. (2010) A pathobiont of the microbiota balances host colonization and intestinal inflammation. Cell Host Microbe 7: 265–276. Cullender, T.C., Chassaing, B., Janzon, A., Kumar, K., Muller, C.E., Werner, J.J., et al. (2013) Innate and adaptive immunity interact to quench microbiome flagellar motility in the gut. Cell Host Microbe 14: 571–581. Dethlefsen, L., Huse, S., Sogin, M.L., and Relman, D.A. (2008) The pervasive effects of an antibiotic on the human gut microbiota, as revealed by deep 16S rRNA sequencing. PLoS Biol 6: e280. Elinav, E., Strowig, T., Kau, A.L., Henao-Mejia, J., Thaiss, C.A., Booth, C.J., et al. (2011) NLRP6 inflammasome regulates colonic microbial ecology and risk for colitis. Cell 145: 745–757. Fagarasan, S., and Honjo, T. (2003) Intestinal IgA synthesis: regulation of front-line body defences. Nat Rev Immunol 3: 63–72. Faith, J.J., Ahern, P.P., Ridaura, V.K., Cheng, J., and Gordon, J.I. (2014) Identifying gut microbe-host phenotype relationships using combinatorial communities in gnotobiotic mice. Sci Transl Med 6: 220ra211. Fontenot, J.D., Gavin, M.A., and Rudensky, A.Y. (2003) Foxp3 programs the development and function of CD4+CD25+ regulatory T cells. Nat Immunol 4: 330–336. Frank, D.N., St Amand, A.L., Feldman, R.A., Boedeker, E.C., Harpaz, N., and Pace, N.R. (2007) Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proc Natl Acad Sci USA 104: 13780–13785. Furusawa, Y., Obata, Y., Fukuda, S., Endo, T.A., Nakato, G., Takahashi, D., et al. (2013) Commensal microbe-derived butyrate induces the differentiation of colonic regulatory T cells. Nature 504: 446–450. Garrett, W.S., Lord, G.M., Punit, S., Lugo-Villarino, G., Mazmanian, S.K., Ito, S., et al. (2007) Communicable ulcerative colitis induced by T-bet deficiency in the innate immune system. Cell 131: 33–45. Gophna, U., Sommerfeld, K., Gophna, S., Doolittle, W.F., and Veldhuyzen van Zanten, S.J. (2006) Differences between tissue-associated intestinal microfloras of patients with Crohn’s disease and ulcerative colitis. J Clin Microbiol 44: 4136–4141. Hooper, L.V., Littman, D.R., and Macpherson, A.J. (2012) Interactions between the microbiota and the immune system. Science 336: 1268–1273. Huttenhower, C., Gevers, D., Knight, R., Abubucker, S., Badger, J., Chinwalla, A., et al. (2012) Structure, function and diversity of the healthy human microbiome. Nature 486: 207–214. Iliev, I.D., Funari, V.A., Taylor, K.D., Nguyen, Q., Reyes, C.N., Strom, S.P., et al. (2012) Interactions between commensal fungi and the C-type lectin receptor Dectin-1 influence colitis. Science 336: 1314–1317. Johansson, M.E., Phillipson, M., Petersson, J., Velcich, A., Holm, L., and Hansson, G.C. (2008) The inner of the two Muc2 mucin-dependent mucus layers in colon is devoid of bacteria. Proc Natl Acad Sci USA 105: 15064–15069. © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 16 1032 C. Petersen and J. L. Round Johansson, M.E., Larsson, J.M., and Hansson, G.C. (2011) The two mucus layers of colon are organized by the MUC2 mucin, whereas the outer layer is a legislator of host– microbial interactions. Proc Natl Acad Sci USA 108 (Suppl. 1): 4659–4665. Josefowicz, S.Z., Lu, L.F., and Rudensky, A.Y. (2012) Regulatory T cells: mechanisms of differentiation and function. Annu Rev Immunol 30: 531–564. Kachrimanidou, M., and Malisiovas, N. (2011) Clostridium difficile infection: a comprehensive review. Crit Rev Microbiol 37: 178–187. Karlsson, F.H., Tremaroli, V., Nookaew, I., Bergstrom, G., Behre, C.J., Fagerberg, B., et al. (2013) Gut metagenome in European women with normal, impaired and diabetic glucose control. Nature 498: 99–103. Kim, Y.G., Udayanga, K.G., Totsuka, N., Weinberg, J.B., Nunez, G., and Shibuya, A. (2014) Gut dysbiosis promotes M2 macrophage polarization and allergic airway inflammation via fungi-induced PGE(2). Cell Host Microbe 15: 95–102. Lupp, C., Robertson, M.L., Wickham, M.E., Sekirov, I., Champion, O.L., Gaynor, E.C., and Finlay, B.B. (2007) Host-mediated inflammation disrupts the intestinal microbiota and promotes the overgrowth of Enterobacteriaceae. Cell Host Microbe 2: 204. Mangin, I., Bonnet, R., Seksik, P., Rigottier-Gois, L., Sutren, M., Bouhnik, Y., et al. (2004) Molecular inventory of faecal microflora in patients with Crohn’s disease. FEMS Microbiol Ecol 50: 25–36. Mangin, I., Suau, A., Gotteland, M., Brunser, O., and Pochart, P. (2010) Amoxicillin treatment modifies the composition of Bifidobacterium species in infant intestinal microbiota. Anaerobe 16: 433–438. Mazmanian, S.K., Round, J.L., and Kasper, D.L. (2008) A microbial symbiosis factor prevents intestinal inflammatory disease. Nature 453: 620–625. van Nood, E., Dijkgraaf, M.G., and Keller, J.J. (2013) Duodenal infusion of feces for recurrent Clostridium difficile. N Engl J Med 368: 2145. Ochoa-Reparaz, J., Mielcarz, D.W., Wang, Y., Begum-Haque, S., Dasgupta, S., Kasper, D.L., and Kasper, L.H. (2010) A polysaccharide from the human commensal Bacteroides fragilis protects against CNS demyelinating disease. Mucosal Immunol 3: 487–495. Olszak, T., An, D., Zeissig, S., Vera, M.P., Richter, J., Franke, A., et al. (2012) Microbial exposure during early life has persistent effects on natural killer T cell function. Science 336: 489–493. Pamer, E.G. (2014) Fecal microbiota transplantation: effectiveness, complexities, and lingering concerns. Mucosal Immunol 7: 210–214. Parracho, H.M., Bingham, M.O., Gibson, G.R., and McCartney, A.L. (2005) Differences between the gut microflora of children with autistic spectrum disorders and that of healthy children. J Med Microbiol 54: 987– 991. Perez-Cobas, A.E., Artacho, A., Knecht, H., Ferrus, M.L., Friedrichs, A., Ott, S.J., Moya, A., Latorre, A., and Gosalbes, M.J. (2013) Differential effects of antibiotic therapy on the structure and function of human gut microbiota. PLoS One 8: e80201. Peterson, D.A., McNulty, N.P., Guruge, J.L., and Gordon, J.I. (2007) IgA response to symbiotic bacteria as a mediator of gut homeostasis. Cell Host Microbe 2: 328–339. Round, J.L., and Mazmanian, S.K. (2009) The gut microbiota shapes intestinal immune responses during health and disease. Nat Rev Immunol 9: 313–323. Round, J.L., and Mazmanian, S.K. (2010) Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc Natl Acad Sci USA 107: 12204–12209. Round, J.L., Lee, S.M., Li, J., Tran, G., Jabri, B., Chatila, T.A., and Mazmanian, S.K. (2011) The Toll-like receptor 2 pathway establishes colonization by a commensal of the human microbiota. Science 332: 974–977. Russell, S.L., Gold, M.J., Hartmann, M., Willing, B.P., Thorson, L., Wlodarska, M., Gill, N., Blanchet, M.R., Mohn, W.W., McNagny, K.M., and Finlay, B.B. (2012) Early life antibiotic-driven changes in microbiota enhance susceptibility to allergic asthma. EMBO Rep 13: 440–447. von Schillde, M.A., Hormannsperger, G., Weiher, M., Alpert, C.A., Hahne, H., Bauerl, C., et al. (2012) Lactocepin secreted by Lactobacillus exerts anti-inflammatory effects by selectively degrading proinflammatory chemokines. Cell Host Microbe 11: 387–396. Schumann, A., Nutten, S., Donnicola, D., Comelli, E.M., Mansourian, R., Cherbut, C., Corthesy-Theulaz, I., and Garcia-Rodenas, C. (2005) Neonatal antibiotic treatment alters gastrointestinal tract developmental gene expression and intestinal barrier transcriptome. Physiol Genomics 23: 235–245. Schmidt, E.G., Claesson, M.H., Jensen, S.S., Ravn, P., and Kristensen, N.N. (2010) Antigen-presenting cells exposed to Lactobacillus acidophilus NCFM, Bifidobacterium bifidum BI-98, and BI-504 reduce regulatory T cell activity. Inflamm Bowel Dis 16: 390–400. Sekirov, I., Tam, N.M., Jogova, M., Robertson, M.L., Li, Y., Lupp, C., and Finlay, B.B. (2008) Antibiotic-induced perturbations of the intestinal microbiota alter host susceptibility to enteric infection. Infect Immun 76: 4726–4736. Shen, Y., Giardino Torchia, M.L., Lawson, G.W., Karp, C.L., Ashwell, J.D., and Mazmanian, S.K. (2012) Outer membrane vesicles of a human commensal mediate immune regulation and disease protection. Cell Host Microbe 12: 509–520. Singh, N., Gurav, A., Sivaprakasam, S., Brady, E., Padia, R., Shi, H., et al. (2014) Activation of Gpr109a, receptor for niacin and the commensal metabolite butyrate, suppresses colonic inflammation and carcinogenesis. Immunity 40: 128–139. Slack, E., Balmer, M.L., Fritz, J.H., and Hapfelmeier, S. (2012) Functional flexibility of intestinal IgA – broadening the fine line. Front Immunol 3: 100. Smith, P.M., Howitt, M.R., Panikov, N., Michaud, M., Gallini, C.A., Bohlooly, Y.M., et al. (2013) The microbial metabolites, short-chain fatty acids, regulate colonic Treg cell homeostasis. Science 341: 569–573. Sokol, H., Pigneur, B., Watterlot, L., Lakhdari, O., Bermudez-Humaran, L.G., Gratadoux, J.J., et al. (2008) Faecalibacterium prausnitzii is an anti-inflammatory commensal bacterium identified by gut microbiota analysis © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 17 How changes in microbiota structure influence health of Crohn disease patients. Proc Natl Acad Sci USA 105: 16731–16736. Spees, A.M., Lopez, C.A., Kingsbury, D.D., Winter, S.E., and Baumler, A.J. (2013) Colonization resistance: battle of the bugs or Menage a Trois with the host? PLoS Pathog 9: e1003730. Strachan, D.P. (2000) Family size, infection and atopy: the first decade of the ‘hygiene hypothesis’. Thorax 55 (Suppl. 1): S2–S10. Ubeda, C., Taur, Y., Jenq, R.R., Equinda, M.J., Son, T., Samstein, M., et al. (2010) Vancomycin-resistant Enterococcus domination of intestinal microbiota is enabled by antibiotic treatment in mice and precedes bloodstream invasion in humans. J Clin Invest 120: 4332–4341. 1033 Vaishnava, S., Yamamoto, M., Severson, K.M., Ruhn, K.A., Yu, X., Koren, O., et al. (2011) The antibacterial lectin RegIIIgamma promotes the spatial segregation of microbiota and host in the intestine. Science 334: 255– 258. Vijay-Kumar, M., Aitken, J.D., Carvalho, F.A., Cullender, T.C., Mwangi, S., Srinivasan, S., et al. (2010) Metabolic syndrome and altered gut microbiota in mice lacking Toll-like receptor 5. Science 328: 228–231. Vrieze, A., de Groot, P.F., Kootte, R.S., Knaapen, M., van Nood, E., and Nieuwdorp, M. (2013) Fecal transplant: a safe and sustainable clinical therapy for restoring intestinal microbial balance in human disease? Best Pract Res Clin Gastroenterol 27: 127–137. © 2014 The Authors. Cellular Microbiology published by John Wiley & Sons Ltd, Cellular Microbiology, 16, 1024–1033 CHAPTER 3 MYD88 SIGNALING IN T CELLS DIRECTS IGAMEDIATED CONTROL OF THE MICROBIOTA TO PROMOTE HEALTH Reprinted with permission from Elsevier Limited Kubinak, J.L., Petersen, C., Stephens, W.Z., Soto, R., Bake, E., O’Connell, R.M., and Round, J.L. (2015). Cell Host Microbe 17, 153–163. 19 Cell Host & Microbe Article MyD88 Signaling in T Cells Directs IgA-Mediated Control of the Microbiota to Promote Health Jason L. Kubinak,1,2 Charisse Petersen,1,2 W. Zac Stephens,1 Ray Soto,1 Erin Bake,1 Ryan M. O’Connell,1 and June L. Round1,* 1Department of Pathology, Division of Microbiology and Immunology, University of Utah School of Medicine, Salt Lake City, UT 84112, USA 2Co-first author *Correspondence: [e-mail] http://dx.doi.org/10.1016/j.chom.2014.12.009 SUMMARY Altered commensal communities are associated with human disease. IgA mediates intestinal homeostasis and regulates microbiota composition. Intestinal IgA is produced at high levels as a result of T follicular helper cell (TFH) and B cell interactions in germinal centers. However, the pathways directing host IgA responses toward the microbiota remain unknown. Here, we report that signaling through the innate adaptor MyD88 in gut T cells coordinates germinal center responses, including TFH and IgA+ B cell development. TFH development is deficient in germfree mice and can be restored by feeding TLR2 agonists that activate T cell-intrinsic MyD88 signaling. Loss of this pathway diminishes high-affinity IgA targeting of the microbiota and fails to control the bacterial community, leading to worsened disease. Our findings identify that T cells converge innate and adaptive immune signals to coordinate IgA against the microbiota, constraining microbial community membership to promote symbiosis. INTRODUCTION The development and function of the mammalian immune system is dependent upon signals conveyed by the microbiota (Belkaid and Hand, 2014; Hooper et al., 2012; Kamada et al., 2013). In particular, the abundance and type of T lymphocytes in the gut is severely reduced in germ-free (GF) mice (Atarashi et al., 2011; Ivanov et al., 2008; Mazmanian et al., 2005; Round and Mazmanian, 2010). While T cell activation is governed by ligation of the T cell receptor (TCR), the quality and nature of the response is dependent on secondary signals such as the cytokine milieu. The identification that T cells express receptors associated with innate signaling such as Toll-like receptors (TLRs) and the IL-1R suggests that T cells could directly utilize these signals as an additional mechanism to control responses (Caramalho et al., 2003; Kubinak and Round, 2012). This would be particularly relevant within the gut, where a constant and abundant source of commensal ligands exists. Supporting this, a single commensal species utilizes TLR2 to promote its own colonization (Round et al., 2011). Recent studies have identified that MyD88 functions within splenic T cells to overcome Treg suppression during immunization (Schenten et al., 2014), identifying the relevance of this pathway to immunity. However, it remains unknown whether these signals provided by the microbiota act directly on T cells in the gut to influence mutualism. The synthesis of IgA has been shown to promote intestinal health (Berry et al., 2012; Brandtzaeg, 2013; Fagarasan et al., 2002; Kawamoto et al., 2012; Lindner et al., 2012; Slack et al., 2009). IgA is the most abundantly produced antibody in mammals, with most being secreted into the intestine. Because of this, IgA represents a key host mechanism for regulating commensal microbial communities. A recent study has shown that IgA binds colitogenic members of the microbiota (Palm et al., 2014), which highlights the role of IgA as an important mediator of microbiota-induced inflammatory disease and a potential diagnostic biomarker. T cell help is required for the generation of high-affinity antibody production. In particular, TFH cells directly interact with B cells in the germinal center (GC) to induce somatic hypermutation and class switching (Crotty, 2011). Our understanding of the molecular pathways that influence GC formation in the gut and how the microbiota influences these pathways remains incomplete. In this present study, we identify that a classic innate immune molecule, MyD88, can function within the T cell compartment in the gut. Loss of MyD88 signaling in T cells leads to reductions in TFH cells and IgA-producing B cells, demonstrating a key role for molecular pathways that converge on this adaptor molecule leading to appropriate GC formation. Moreover, GC formation in the gut is orchestrated by signals provided by the microbiota in a T cell-intrinsic MyD88-dependent manner. Loss of GC formation leads to reduced IgA production and disrupted targeting of commensal bacterial populations. Animals lacking MyD88 within the T cell compartment fail to control mucosally associated communities of bacteria, resulting in dysbiosis. Finally, we demonstrate that animals lacking T cell-intrinsic MyD88 develop worsened disease that can be rescued by a microbial transplant from a healthy donor. Thus, we have identified a host molecular pathway that can integrate signals from the microbiota to promote GC formation and IgA production against intestinal bacteria to control the composition of these communities to ensure a benign symbiotic interaction. Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 153 20 A WT B T-MyD88-/16% C 20 7.9% *** 4 104 GC-TFH (#) GC-TFH (%) 15 3 104 0 PD-1+ 41.5% 80 28.9% Fas+ H WT T-MyD88-/- 6 104 4 104 20 60 **** 1 105 8 104 40 IgA+ plasmablast (%) in siLP 41.1% T-MyD88-/20.5% * 60 0 WT 0 WT T-MyD88-/- F GL-7+ G 1 104 E GC B cell (%) WT T-MyD88-/- 2 104 5 GC B cell (#) CXCR5+ 10 D *** 5 104 2 104 WT T-MyD88-/- 0 WT T-MyD88-/- * 40 CD138+ 20 IgA+ WT 50% T-MyD88-/28.7% 0 J 80 IgA+ plasmablast (%) in cLP I ** CD138+ 40 20 0 (A–C) GC-TFH cells from PPs, defined by CD3+CD4+B220!CXCR5hiPD-1hi, were measured by flow cytometry. (A) Representative plots were previously gated on CD3+CD4+B220! cells. (B) Frequency and (C) absolute numbers of GC-TFH cells are shown (n = 21 WT, n = 18 T-MyD88!/!). (D–F) GC B cells, defined by B220+IgDlowFas+GL7+, from PPs were measured by flow cytometry. (D) Representative plots were previously gated on B220+IgDlow cells. (E) Frequency and (F) absolute numbers of GC B cells are shown (n = 21 WT, n = 18 T-MyD88!/!). (G–J) IgA+ plasmablasts, defined as B220! CD138+IgA+, within the siLP and cLP were measured by flow cytometry (G). Representative plots were previously gated on CD138+B220! cells from siLP. (H) Frequencies of IgA+ plasmablasts in siLP are shown (n = 6 WT, n = 7 T-MyD88!/!). (I) Representative plots were previously gated on CD138+B220! cells from cLP. (J) Frequencies of IgA+ plasmablasts in cLP are shown (n = 8 WT, n = 5 T-MyD88!/!). Unpaired two-tailed Student’s t tests were used for all comparisons. p < 0.05 (*); p < 0.01 (**); p < 0.001 (***). Lines in scatterplots represent means. See also Figures S1–S4. WT T-MyD88-/- 60 IgA+ Figure 1. T Cells Utilize MyD88 Signaling to Coordinate Germinal Center Responses WT T-MyD88-/- RESULTS MyD88-Dependent Signaling in T Cells Influences GC Responses in the Gut Whether innate signaling by T cells influences the establishment of beneficial bacterial communities and host health remains to be elucidated. As MyD88 is a key molecule that governs signaling through multiple innate receptors, we crossed a MyD88-floxed animal with a T cell-specific Cre-driver to produce an animal model where MyD88 is specifically knocked out within T cells but retained in other cell types (the T-MyD88!/! mouse) (Figure S1) (Chang et al., 2013; Schenten et al., 2014). This allowed us to test whether innate and adaptive immune pathways converge to promote host-microbiota symbiosis within the gut. Recent studies have identified that T cell-intrinsic MyD88 signaling influences systemic induction of TH1 and TH17 cells during immunization (Chang et al., 2013; Raetz et al., 2013; Schenten et al., 2014). Thus, we first broadly examined CD4+ T cell development during steady-state conditions. We did not observe differences in TH17, TH1, or T regulatory cells (Tregs) within the spleen, mesenteric lymph nodes (MLNs), or colonic lamina propria (cLP) (Figures S2 and S3). Consistent with previous reports, we observed slight defects in the abundance of TH1 and TREGS within the small intestinal lamina propria (siLP) and TH17 cells within the Peyer’s patch (PP) (Figure S3) (Reynolds et al., 2010; Schenten et al., 2014). However, the most striking difference observed was within the TFH cell compartment in both the PP and MLNs in T-MyD88!/! animals (Figure 1A and Figures S4A–S4K). TFH cells are one of the most abundant subsets of CD4+ helper T cells within the PPs and play a fundamental role in the generation of antibody-mediated responses against the microbiota (Crotty, 2011; Linterman et al., 2012). Therefore, we focused our analysis on GC formation in the gut. TFH cells are broadly characterized by the expression of two or more of the following surface markers: ICOS, CXCR5, and PD-1. Subtypes of TFH cells can be delineated using these markers. These include GC-resident TFH (GC-TFH) cells (CXCR5hi PD-1hi) whose function is to promote class switching and somatic hypermutation of naive B cells to produce highaffinity IgA, and non-GC-resident TFH (non-GC-TFH) cells (CXCR5intPD-1int) that will ultimately migrate into the GC to promote B cell activity (Shulman et al., 2013). Any combination of these markers identifies a defect in any TFH subset in the PPs of T-MyD88!/! animals that cannot be accounted for by changes in PP cellularity (Figures 1A–1C and Figures S4A–S4I). These data suggest that T cell-intrinsic MyD88 signaling impacts multiple stages of TFH development. Importantly, these effects were gut specific because they were not observed in the spleen 154 Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 21 A SPF GF 10.7% B C *** 20 2.3% GC-TFH (#) GC-TFH (%) 15 4000 CXCR5+ 10 14% 0 GF SPF ABX F GC-TFH (%) 4000 GF SPF * 80000 GC-TFH (#) SPF J ABX WT CXCR5+ or T-MyD88-/- PD-1+ 10.9% Bone Marrow Transfer ABX Germfree T-MyD88-/4.5% TLR2 ligand ad libitum GF Rag-/BM recipients 0 TLR ligand SPF WT 2000 20000 5 Donor mice 4000 40000 10 I ** *** 15 0 PD-1+ 6000 60000 0 H K 2 weeks L 20 15 10 5 0 * 4000 GC-TFH (#) G GC-TFH (%) 2000 GF SPF 20 7.3% CXCR5+ GC-TFH (#) SPF 6000 0 GC-TFH (#) E ** 8000 2000 5 0 PD-1+ D Figure 2. Microbiota-Dependent TFH Development Relies on T Cell-Intrinsic MyD88 Signaling ** 6000 WT T-MyD88-/GF +TLR ligand (Figures S4L and S4M), indicating that T cell-intrinsic MyD88 signaling influences steady-state development of TFH cells in the intestine. TFH-dependent instruction of GC B cells induces IgA+ plasmablast migration to the lamina propria of the small and large intestine, leading to secretion of high-affinity IgA into the intestinal lumen. Based on the observed deficiencies in TFH development, intestinal B cell function was examined. Percentages and absolute numbers of PP GC B cells were significantly reduced in T-MyD88!/! mice (Figures 1D–1F), indicating a defect in the GC response. Consequently, IgA+ plasmablasts in the siLP and cLP were also diminished (Figures 1G–1J). These results demonstrate that MyD88 signaling within T cells regulates appropriate GC formation within the gut. Microbiota-Derived Signals Influence the Presence of TFH Cells in the Gut B cell responses within the gut are sensitive to the presence of the microbiota. Indeed, GF mice have fewer B cells and lower 3000 2000 1000 0 WT T-MyD88-/GF +TLR ligand (A–C) GC-TFH cells from PPs, defined by CD3+CD4+B220!CXCR5hiPD-1hi, were compared between SPF and GF Balb/c mice with flow cytometry. (A) Representative plots were previously gated on CD3+CD4+B220! cells. (B) Frequency and (C) absolute numbers of GC-TFH cells are shown (n = 5 SPF, n = 7 GF). (D) GC-TFH cells from PPs were compared between SPF and GF C57BL/6 mice with flow cytometry (representative plots not shown) to obtain absolute numbers of GC-TFH cells (n = 5 for each group). (E–G) GC-TFH cells from PPs were compared between SPF mice treated with oral antibiotics (ABX) and non-treated SPF animals with flow cytometry. (E) Representative plots were previously gated on CD3+CD4+B220! cells. (F) Frequency and (G) absolute numbers of TFH cells are shown (n = 8 SPF, n = 9 ABX). (H) Pam3CSK was provided in the drinking water of GF animals for 2 weeks to calculate absolute numbers of TFH cells in animals given TLR ligand (+) (n = 6) or not (!) (n = 8). (I) Experimental schematic showing GF Rag!/! mice reconstituted with either WT or T-MyD88!/! bone marrow and subsequently treated with TLR2 ligand, Pam3CSK, in the drinking water. (J–L) GC-TFH cells from MLNs were compared between GF Rag!/! mice reconstituted with WT or T-MyD88!/! bone marrow and fed the TLR2 ligand Pam3CSK. (J) Representative plots were previously gated on CD3+CD4+B220! cells. (K) Frequency and (L) absolute numbers of GC-TFH cells are shown (dotted line represents the number of cells within these populations when GF-Rag mice are given bone marrow for either genotype and not stimulated) (n = 4 WT and n = 3 T-MyD88!/!). p < 0.05 (*); p < 0.01 (**); p < 0.001 (***). Lines in scatterplots and bar graphs represent means. IgA levels. MyD88 functions downstream of TLRs, which are involved in responses to bacterial products. Thus, our observations suggest that TFH development within the gut is influenced by signals from the commensal microbiota. To test this we isolated PPs from either GF or specific pathogen free (SPF) mice from two different backgrounds of animals (Balb/c and C57BL/ 6) and examined the TFH population. Both C57BL/6 and Balb/c GF mice had significantly reduced numbers of GC-TFH cells within the PPs when compared to SPF mice (Figures 2A–2D). Moreover, GC-TFH cell populations were severely reduced when C57BL/6 SPF animals were treated with an antibiotic cocktail that depletes the microbiota (Figures 2E–2G). This demonstrates that development and maintenance of GC-TFH cells within the gut is promoted by the presence of commensal organisms. To determine whether live bacterial colonization is required for induction of GC responses or whether bacterial products alone are sufficient, we placed the TLR2 ligand, Pam3CysK, in the drinking water of GF mice. Oral treatment of GF mice with Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 155 22 150 WT 100 or T-MyD88-/- 50 0 Figure 3. IgA Production against Commensal Antigens Is Perturbed in the Absence of MyD88 Signaling in T Cells Donor mice WT T-MyD88-/- 1010 109 108 107 106 105 104 103 102 101 100 D WT T-MyD88-/Bone Marrow Donor 0.5 Colonize w/ OVA expressing commensal Bone Marrow Transfer E *** 0.4 0.3 0.2 0.1 10 5 0 0.0 WT T-MyD88-/Bone Marrow Donor WT T-MyD88-/Bone Marrow Donor FSC T-MyD88-/0.7% % IgA-coated Bacteria G 3.3% ns 15 F WT 8 weeks GF Rag-/BM recipients Total IgA ( g/mL) OVA (per 50ng DNA) C B ** 200 OVA specific IgA (Abs) Luminal IgA (mg/mL) A 10 IgA+ a purified TLR2 agonist alone was capable of significantly increasing GC-TFH abundance within PPs (Figure 2H). To definitively test the relevance of T cell-intrinsic MyD88 signaling during microbiota dependent TFH development in vivo, we reconstituted GF Rag!/! mice with bone marrow from either WT or T-MyD88!/! animals (Figure 2I). As GF Rag!/! animals do not develop PPs and a bone marrow reconstitution does not correct this, we analyzed the defect in TFH development within the MLNs of these mice. Eight weeks after transfer, animals were orally treated with a purified TLR2 agonist. Strikingly, TFH development occurred in GF WT animals in response to purified TLR2 ligands, while TFH development was not induced in TLR2treated GF animals reconstituted with T-MyD88!/! bone marrow (Figures 2J–2L). This was not a result of differences in reconstitution of GF Rag!/! animals, as baseline TFH abundance is the same in Rag!/! animals receiving WT or T-MyD88!/! bone marrow in the absence of TLR2 treatment (Figure 2L). Collectively, these data argue strongly that cues from the microbiota can act through MyD88 within T cells to coordinate TFH biology, thus identifying a pathway by which commensal bacteria can directly influence GC responses within the intestine. ** 8 6 (A) Concentrations of soluble IgA within gut lumenal contents measured by ELISA and normalized to fecal weight (n = 20 per group). (B) Experimental schematic showing GF Rag!/! mice reconstituted with either WT or T-MyD88!/! bone marrow and subsequently mono-colonized with B. fragilis-OVA. (C) Raw abundances of DNA specific to the OVA gene were detected via qRT-PCR to quantify the loads of B. fragilis-OVA per 50 ng of fecal DNA (n = 5 per group). (D) ELISA was used to measure the concentration of IgA within the lumen of these animals, normalized to fecal weight (n = 5 per group). (E) Abundance of OVA-specific IgA within gut lumenal contents measured by ELISA and normalized to fecal weight. Data are shown as absorbance at 450 nm (n = 5 per group). (F) IgA-bound bacteria in intestinal lumen was measured by flow cytometry. Representative plots were previously gated on SYBR Green+ events ( = bacteria). (G) Frequencies of IgA-bound bacteria (n = 8 per group) with dashed line showing mean nonspecific binding in RAG!/! controls. Unpaired two-tailed Student’s t tests were used for all comparisons. p < 0.05 (*); p < 0.01 (**); p < 0.001 (***). Lines in scatterplots represent means. Bar graphs represent means ± SD. 4 2 0 T Cell-Intrinsic MyD88 Signaling Coordinates IgA Responses against the Microbiota WT GC responses function within the intestine to produce high-affinity IgA against food, self, and microbial antigens (Bemark et al., 2012). As animals that lack MyD88 within the T cell compartment have reduced TFH and B cell responses within the gut, we wished to test whether these deficiencies resulted in impaired IgA production. Intestinal contents were isolated from WT and T-MyD88!/! and analyzed for the production of secretory IgA (SIgA) by ELISA. We found that total SIgA was reduced in T-MyD88!/! mice (Figure 3A), illustrating the importance of this pathway in T cells for the generation of productive GC responses. To determine whether this pathway regulates the generation of antigen-specific IgA against the microbiota, we mono-colonized GF Rag!/! mice previously reconstituted with WT or T-MyD88!/! bone marrow with a commensal bacteria strain that was engineered to express the model antigen ovalbumin (Bacteroides fragilis-OVA). This allowed us to track a commensal-specific antibody response (Figure 3B). GF Rag!/! animals have no detectable OVA-specific IgA within the intestine (data not shown). While B. fragilis-OVA colonization in the lumen was similar between cohorts (Figure 3C), there was a complete absence of OVA-specific SIgA production in mono-associated T-MyD88!/! animals (Figure 3D), despite similar levels of total SIgA being produced (Figure 3E). These data indicate that MyD88 signaling within T cells directs antigen-specific SIgA against the microbiota. T-MyD88-/- 156 Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 23 A B Figure 4. Innate Recognition by T Cells Influences the Composition of Mucosally Associated Microbial Communities (A) Principal coordinates plot of unweighted UniFrac distances shows separation of microbial communities by genotype and sample type within separately housed animals. See also Table S1. (B) Mucosal communities are more similar than fecal communities (non-parametric t test; **p < 0.001, ***p = 0.0001). Boxplot whiskers = interquartile range. (C) Differences in mucosal communities between genotypes are maintained in cohoused animals (PERMANOVA; p = 0.0111). See also Table S1. C Reports have indicated that IgA targeting of the microbiota is modulated during intestinal disease in humans (van der Waaij et al., 2004) and may target colitogenic members of the microbiota (Palm et al., 2014). SIgA can be produced in the intestine as a result of either T cell-dependent (TD) or -independent (TiD) mechanisms (Mantis et al., 2011; Stephens and Round, 2014). TiD often results in low-affinity, cross-reactive antibody. TD responses, mediated by TFH cells, result in commensal targeting via specific, high-affinity IgA binding (Lycke and Bemark, 2012). To better characterize the TD antibody defect in TMyD88!/! animals, we utilized a flow cytometry-based assay to look at host-generated SIgA against the microbiota. T celldeficient animals have undetectable IgA-bound bacteria with this method, so this assay quantifies high-affinity TD antibody production (Slack et al., 2009, 2012). The microbiota was isolated from animals and immunostained with antibody against IgA and treated with the DNA dye SYBR Green. Consistent with other reports (Palm et al., 2014), 1%–10% of SYBR+ microbiota was bound by IgA in WT mice, while a 10-fold reduction in IgA targeting of the microbiota was observed in T-MyD88!/! animals (Figures 3F and 3G). These results reveal that innate recognition of commensal products by T cells influences the abundance and quality of the TD SIgA response toward commensal bacteria. MyD88 within T cells Prevents Dysbiosis of TissueAssociated Microbial Communities Defects in commensal-directed SIgA might have important effects on the microbial ecology of the gut (Cerutti et al., 2011; Cullender et al., 2013; Fagarasan et al., 2002; Peterson et al., 2007). Therefore, we sought to determine via high-throughput 16S rRNA sequencing whether the loss of T cell-intrinsic MyD88 signaling could influence microbial composition. To control for differences in bacterial communities based on housing conditions, we performed analyses on separately housed animals whose breeding lines were originally derived from heterozy- gote crosses as well as cohoused animals. Regardless of housing conditions, fecal and mucosal communities were distinct in both genotypes (Figure 4A and Table S1). Interestingly, greater variation exists between fecal communities among individuals when compared to mucosa-associated communities (Figure 4B). Since mucosal communities reside in closer proximity to host tissue, this result suggests that the immune system exerts stronger selection on community membership at this site. Consistent with this, the differences in community composition between genotypes are only maintained in mucosa-associated communities (Figure 4C), but not fecal communities (Table S1), during cohousing. Among these differences, we observed increases in multiple taxa containing mucolytic members in TMyD88!/! animals, including Desulfovibrionaceae and Mucispirillum (Figures S5A and S5B), as well as significant increases in Ruminococcus (Figures S5A and S5B). Collectively, these data indicate that loss of T cell-intrinsic TLR signaling results in significant shifts in microbial composition in the gut. T Cell-Intrinsic MyD88 Signaling Governs IgA Targeting of Mucosa-Associated Microbiota We next investigated how innate signaling in T cells would influence SIgA targeting of microbial communities. To do this, we developed an assay to purify IgA-bound bacteria with 94% purity (Supplemental Information) and characterized SIgA-targeted bacterial communities via high-throughput sequencing. IgA binds a distinct group of bacteria when compared to either the total fecal or mucosal community in either genotype, indicative of specific targeting of commensal species by SIgA (Figure 5A and Table S1). Additionally, the assemblage of bacterial species targeted by the host via SIgA is altered in T-MyD88!/! mice compared to WT animals (Figure 5A). The SIgA-bound fraction is more similar to the mucosa-associated community than the fecal community in WT animals, indicating that SIgA tends to target mucosa-associated organisms more so than fecal community members (Figure 5B). Remarkably, this IgA bias toward mucosal communities is lost in T-MyD88!/! animals as IgA-bound species are equally representative of mucosal and fecal communities (Figures 5B). This is not just a result of blending of the mucosa-associated and fecal communities, because these communities are as dissimilar in T-MyD88!/! animals as they are in WT animals Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 157 24 A B D F C E G Figure 5. T Cell-Intrinsic MyD88 Signaling Regulates IgA Selection of the Microbiota (A) PCoA plot of unweighted UniFrac distances comparing IgA-bound and mucosa-associated bacterial communities between genotypes (PERMANOVA; p < 0.01 for all pairwise comparisons). See also Table S1. (B) Community dissimilarity (weighted UniFrac distance) between IgA-bound versus fecal and IgA-bound versus mucosal communities for each genotype. Non-parametric t test. p = 0.0001 (****). Increasing values along the y axis represent increasing community dissimilarity among individual samples within a given cohort. Boxplot whiskers = interquartile range. (C) UniFrac distances between fecal and mucosal communities are not significantly different between genotypes (non-parametric t test). Boxplot whiskers = interquartile range. (D) Plot of the mean relative abundance of individual OTUs in all mucosal and IgA-bound samples colored by class of the OTU. A linear model fit to the data is shown and is plotted on log scale. (legend continued on next page) 158 Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 25 B WT 24 TMyD88-/- R2 =0.400 P = 0.020 30 IR Dissimilarity Phylogenetic Diversity A 22 20 20 TMyD88-TMyD88 WT-TMyD88 WT-WT Immune Response (IR) TFH cell (%) 10 18 0 16 30 40 50 GCB cells (%) 60 0.4 0.5 Community Mantel r P IgA Bound 0.36 0.003 0.008 TFH cell (%) Mucosal 0.44 GC B cell (%) IgA Bound 0.28 0.024 GC B cell (%) Mucosal 0.09 0.358 0.6 Community Dissimilarity Figure 6. Phenotypic Variation in Immune Response Correlates with Microbial Community Similarity (A) Phylogenetic diversity in mucosal communities is correlated with GC B cell abundance (%). (B) Left: A representative plot defines the axes and pairwise comparisons used to derive results of a Mantel’s test. Right: Correlations and p values for the association between the differences in immune response (IR) and community dissimilarity among individuals. IR dissimilarity and community dissimilarity are estimates calculated from distance matrices built from immune phenotype and microbial community phylogenetic (UniFrac) data, respectively. (Figure 5C). These results indicate that the host IgA response is primarily directed toward specific bacteria within the mucosa and that this bias is disrupted in T-MyD88!/! animals. Given these results, we focused on how a loss of IgA mucosal bias, as seen in T-MyD88!/! animals, would influence the microbial community. There is a significant positive correlation between bacterial species (OTU) abundance in the total mucosal community and in the IgA-bound fraction, which implies that in general IgA binds the most abundant organisms (Figure 5D). Additionally, several taxonomic groups were targeted differently between WT and T-MyD88!/! animals (Figure 5E). These include reduced binding of the classes Clostridia and Actinobacteria and more robust targeting of many OTUs in the Bacteroidia (primarily within the uncultured S24-7 family) in T-MyD88!/! animals (Figure 5E). Resolution of microbial communities at finer taxonomic depth reveals a reduction of IgA targeting of Bifidobacterium pseudolongum and increased IgA binding of Ruminococcus gnavus and unclassified Prevotella species in T-MyD88!/! animals (Figure 5E and Figure S5A). Additionally, OTUs of the genus Lachnospiraceae were the most differentially targeted by IgA (including cohoused animals) (Table S2), as has been observed in a mouse model of colitis (Berry et al., 2012). Differences in IgA targeting in T-MyD88!/! animals did not simply reflect significant differences in OTU abundance in the mucosa-associated community (Figure 5E), suggesting that abundance alone does not dictate the specificity of IgA targeting. Importantly, we observed significantly increased dissimilarity among individual T-MyD88!/! animals in both their IgA-bound and mucosaassociated communities compared to WT animals (Figure 5F), suggesting that there is greater variability in which species persist in the gut of a T-MyD88!/! animal. We also observed an increased total bacterial load at the mucosa in T-MyD88!/! animals (Figure 5G). Collectively, these data demonstrate that innate signaling by T cells dictates IgA specificity to constrain the composition of the microbial community, while also limiting mucosal association of commensal microbes. We next sought to determine whether any of the observed defects in the host humoral response were associated with changes in microbial community composition. Consistent with previous reports, we found a significant positive correlation between the relative abundance of GC B cells and mucosal community diversity (Figure 6A). This implies that a stronger GC response promotes diversity, perhaps by opening niches for rare species by limiting the densities of abundant members. Similarly, we utilized a Mantel’s test to demonstrate that the magnitude of difference in immune phenotype among individuals was positively associated with dissimilarity between microbial communities (Figure 6B). This effect was observed for multiple immune parameters and was driven by differences between host genotypes (Figure 6B). Thus, differences in microbial composition between WT and T-MyD88!/! animals are directly associated with differences in their immune response. Collectively, these data demonstrate that host humoral responses sculpt microbial communities in the gut through IgA-mediated selection and that innate sensing of microbial products by T cells regulates this process. Alterations to the Microbiota in T-MyD88–/– Animals Results in Increased Intestinal Disease Changes to microbial communities have been shown to promote inflammatory disease in the gut. Our observed shifts in Lachnospiraceae are similar to those seen in colitogenic animals, and R. gnavus has been associated with intestinal disease in humans (Hansen et al., 2013; Png et al., 2010). Moreover, defects in antibody targeting of bacteria have been seen in patients with IBD (Harmsen et al., 2012; van der Waaij et al., 2004), suggesting that a reduced ability to control the microbiota within TMyD88!/! animals could lead to harmful consequences to the host. To determine whether changes in microbial community composition observed in T-MyD88!/! mice led to worsened disease, we examined colitis susceptibility in this model. Typical of the TNBS colitis model in C57BL/6 animals, WT animals lost (E) Class, genus, and species abundances detected as significantly different in the IgA-bound fraction are shown along with their abundances in mucosaassociated communities. Welch’s t test. p < 0.05 (*); p < 0.001(***). Bar graphs represent mean taxa abundance ± SD. See also Table S2. (F) Comparison of community dissimilarity between genotypes for IgA-bound and mucosa-associated communities. Non-parametric t test. p < 0.05 (*); p < 0.001 (***); p = 0.0001 (****). Boxplot whiskers = interquartile range. (G) Total bacterial load in mucosa of WT and T-MyD88!/! animals as measured by qPCR. Student’s t test. p < 0.05 (*). Bars represent means. Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 159 26 A * ** *** * % Initial Weight 100 95 90 85 80 B C EtOH WT EtOH T-MyD88-/TNBS WT TNBS T-MyD88-/- 2 3 4 1 5 Days post challenge WT microbiota T-MyD88-/- T-MyD88-/- microbiota *** 15 Histology Score WT *** 10 5 0 WT T-MyD88-/- WT T-MyD88-/microbiota microbiota T-MyD88-/- Figure 7. Increased Intestinal Disease Observed in T-MyD88–/– Animals Is Dependent on the Composition of the Microbiota (A) TNBS-induced weight loss (represented as the percentage of initial weight) was compared among WT and T-MyD88"/" animals. EtOH vehicle controls are shown for comparison. Points represent means ± SEM (representative of two independent experiments). (B) Representative H&E-stained colon sections from WT and T-MyD88"/" mice following TNBS-induced colitis. (C) Histology scores reflecting disease severity for TNBS-treated animals (WT animals, n = 10; T-MyD88"/" animals, n = 11; T-MyD88"/" animals with WT microbiota transplant, n = 8; T-MyD88"/" animals with T-MyD88"/" microbiota transplant, n = 7). Dotted line represents scores for mice treated with EtOH only. A Student’s t test was used for all comparisons. p < 0.05 (*); p < 0.01 (**); p < 0.001 (***). !10% of their body weight and quickly regained their original mass. However, T-MyD88"/" animals lost more weight and did not recover to WT levels by the end of the experiment (Figure 7A). Histological analysis of these colons revealed that T-MyD88"/" mice had increased inflammatory infiltrate and greater crypt loss than WT animals (Figures 7B and 7C). Increased bacterial loads at the mucosa or differences in the assemblage of species living at this site could explain this. To determine whether microbial composition in the gut influenced disease progression, we tested whether susceptible T-MyD88"/" animals could be rescued with a WT microbiota. To this end, cohorts of TMyD88"/" animals were treated with antibiotics to clear endogenous commensals and subsequently given a transplant with either WT or T-MyD88"/" microbiota. Animals that received a microbial transplant from a T-MyD88"/" animal developed severe crypt loss (Figures 7B and 7C). Remarkably, the extent of disease observed in T-MyD88"/" animals provided a microbiota from WT mice was significantly reduced and indistinguishable from that observed in WT animals (Figure 7C). Notably, WT GF mice given a T-MyD88"/" microbiota had no difference in disease severity (data not shown), indicating that the microbial composition formed in T-MyD88"/" animals is not sufficient to cause disease in an immune-competent host. Thus, this model closely mirrors the complexity seen in human IBD whereby genetic and environmental factors interact to promote disease. These data also highlight the potential curative value of fecal transplantations from healthy donors. DISCUSSION While originally thought to play a role only within the innate immune compartment, emerging data support that MyD88 functions also in cells of the adaptive immune system such as T lymphocytes (Chang et al., 2013; Schenten et al., 2014). How this pathway could function to influence resident commensal communities has remained unexplored. MyD88 functions downstream of multiple receptors including the cytokine receptors, IL-1, IL-18, and IL-33, as well as all the TLRs with the exception of TLR3. The microbiota is an abundant source of TLR ligands and can induce the basal expression of these cytokines, making MyD88 an attractive molecule to integrate multiple signals induced by the microbiota to directly control host T cell function. Indeed, here we identify that MyD88 signaling in T cells governs the development of functional GCs within the gut. It is unclear at this stage what exact signals are triggering activation of MyD88 within T cells to control GC responses. However, it is likely that multiple signals act directly on T cells and converge on MyD88 as T cells express several of these TLRs and cytokine receptors (Reynolds et al., 2010). The use and development of new conditional animal models to study these individual signals will be required to fully understand how innate signals elicited by the microbiota can directly influence T cell development and function within the gut. One of the primary mechanisms to promote homeostasis within the gut is through the production of antigen-specific IgA. Understanding the mechanisms by which this pathway governs antigen-specific TFH responses will be an important future endeavor. By employing a newly developed technology, we were able to determine that antigen-specific IgA toward the microbiota is influenced by MyD88 in T cells (Cullender et al., 2013; Kawamoto et al., 2012; Palm et al., 2014). There is evidence that there are niche-specific microbial communities within the 160 Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 27 gut, including organisms that are in close proximity to the host (tissue- or mucosa-associated communities) and those that are found to reside away from the host within the intestinal lumen. We identify that host IgA tends to target tissue-associated microbial communities in a healthy animal. However, in the absence of MyD88 signaling within the T cells, the mucosal bias of IgA is lost, indicating that this pathway serves to dictate IgA specificity within the gut. More importantly, loss of this signaling pathway in T cells leads to defective control over microbial community composition, resulting in worsened intestinal disease. Immune systems evolved within a microbe-dominated world under strong selection to protect host tissues from pathogenic invasion. However, the dependence of host health on the microbiota strongly suggests that the immune system evolved to serve an additional purpose: to promote colonization by microbial species whose presence can be exploited to benefit host health (McFall-Ngai et al., 2013). Thus, coevolutionary forces between resident microbes and hosts have resulted in mechanisms to respond to one another, creating a flexible dialog to ensure stability of symbiosis. IgA represents one host tool to directly influence bacterial communities; however, it has been unclear whether IgA binds to specific members of the community and, more importantly, what host molecular machinery governs this specificity. Our data support a model whereby signals from commensal microbes are perceived directly by host T cells to promote the production of TFH and GC B cell responses, leading to the generation of high-affinity IgA toward commensals. This represents an important mechanism through which hosts can maintain a benign assemblage of microbial species in the gut. Segregation of innate and adaptive arms of the immune system would be appropriate in sterile environments (like the systemic compartment) as this would provide a binary switch to respond to microbial invasion. However, in tissues with persistent bacterial exposure, cellular convergence of innate and adaptive immune pathways could provide more flexibility to fine-tune antigen-specific signaling and respond to environmental changes properly. Thus, modulation of T cell responses through microbiota-derived signals might provide a therapeutic target to prevent disease associated with human autoimmunity. EXPERIMENTAL PROCEDURES Animal Models MyD88fl/fl (Jackson Labs) mice were crossed to CD4-cre+ mice (Taconic) (both on C57BL/6 background) to produce the T-MyD88!/! (MyD88fl/fl-CD4cre+) mouse model. MyD88+/+-CD4-cre+ mice were used as wild-type (WT) controls for all comparisons. GF animals were on a C57BL/6 background except when noted in the figure legend. Animal use adhered strictly to federal guidelines and those set forth by the University of Utah’s Institutional Animal Care and Use Committee. In Vitro Activation Experiments Purified cells were plated in 250 ml of supplemented RPMI containing 20 ng/ml IL-2 in 96-well plates (2 3 105 cells per plate) that had been previously coated with 5 mg/ml purified anti-CD3. For TFH skewing conditions, 1 mg/ml of the costimulatory antibody anti-CD28, 5 mg/ml anti-CD3e, 10 mg/ml anti-IFNg, 50 ng/ml IL-21, and 50 ng/ml IL-6 were added. For TLR ligand stimulation, 1 mg/ml of either Pam2CSK (Invivogen), Pam3CSK (Invivogen), flagellin from S. Typhimurium (Invivogen), or LPS (Invivogen) was added. Cells were cultured for 4 days at 37" C in these conditions before being analyzed by flow cytometry. RNA Isolation and RNA-Sequencing of Purified TFH Cells mRNA was collected from #50,000 TFH cells with the QIAGEN miRNeasy Mini Kit (QIAGEN) following kit instructions. mRNA was prepared following QC via an Illumina TruSeq Stranded RNA Sample Prep with RiboZero treatment (human, mouse, rat, etc.) and analyzed with Illumina HiSeq Sequencing. Disease Model-TNBS Colitis One hundred microliters of a 50/50 (v/v) mixture of EtOH and TNBS was administered intra-rectally on days 0 and 5. Weights were collected daily for the first 5 days, and tissue was collected on day 7. Animals were scored as described in detail in the Supplemental Information and as reported in Round and Mazmanian, 2010. GF Model GF mice were maintained in sterile isolators and assayed monthly for GF status by plating and PCR. Balb/c GF mice were compared to SPF mice to measure TFH cells. GF Rag1!/! mice from the C57BL/6 background were given bone marrow from WT or T-MyD88!/! mice and allowed to reconstitute for 8 weeks. Mice were either given Pam3CSK in their drinking water or were gavaged with B. fragilis-expressing OVA. Quantification of IgA+ Fecal Bacteria Fecal microbiota were stained with SYBR Green dye and a PE-conjugated IgA antibody to enumerate IgA-coated bacteria via flow cytometry. Please see detailed Supplemental Experimental Procedures. Sequencing of Microbiota Communities DNA was extracted with bead beating methods from fecal, mucosal, and IgAbound enriched samples from each animal. The V3 and V4 regions of the bacterial 16S rRNA gene were amplified and sequenced on an Illumina MiSeq with 300 cycles from each end. Sequences were processed and analyzed with mothur 1.32 (Schloss et al., 2009) and Qiime 1.8.0 (Caporaso et al., 2010). Statistical Analysis Figure creation and statistical analysis were performed with the Prism6.0 statistical software, R 3.1.0, STAMP 2.0.5 (Parks and Beiko, 2010), and utilities within the Qiime software. Unpaired two-tailed Student’s t tests were used for statistical comparisons unless otherwise noted (see Supplemental Information). Non-parametric t tests were performed with 9,999 Monte Carlo simulations. All statistical tests, and estimates of dispersion, are referenced in the figure legends. Animal Housing/Cohousing Experiments SPF mice in our facility are housed under identical conditions and fed the same diet, which minimizes the effect of environmental variability on host phenotypes and microbial communities. In an effort to standardize any potential effect of isolation (across cages and between genotypes) on microbiota communities, multiple independent homozygous WT and T-MyD88!/! breeder pairs were set up at the same time from initial heterozygous crosses. Clustering of genotypes in 16S microbiota analysis confirms that host genotype and not isolation is the primary driver of differences among individuals in their communities. For cohoused experiments, pregnant homozygous WT or T-MyD88!/! animals gave birth to litters in the same cage (births occurred within 1 week of each other). Animals remained in the same cage until weaning (4 weeks of age) and were subsequently housed separately until analysis (8 weeks of age). Lamina Propria Lymphocyte Isolation Lamina propria lymphocytes (LPLs) were isolated with a combination of two previously described methods (Atarashi et al., 2011; Round et al., 2011). Prior to LPL isolations of small intestines (SI), all visible Peyer’s patches were removed. Colons and SI tissue were opened longitudinally and mucus was removed by gently scraping and then rinsing in sterile 1X HBSS. The tissue was cut into small pieces and incubated in sterile 1X HBSS (without Ca2+ and Mg2+) containing 5 mM EDTA (Cellgro) and 1 mM DL-Dithiothreitol (DTT) (Sigma) for 45 min at 37" C on a shaker. Supernatant was removed by filtering through a 100 mM filter, and remaining tissue was incubated in a solution containing sterile 1X HBSS containing 5% (v/v) fetal bovine serum (GIBCO BRL), 50 U/ml Dispase (Roche), 0.5 mg/ml Collagenase D (Roche), and 0.5 mg/ml Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 161 28 DNaseI (Sigma) for 45 min at 37! C on a shaker. The supernatant was filtered over a 40 mm cell strainer into ice-cold sterile 1X HBSS. Cells were passed through a Percoll (GE Healthcare) gradient (40%/80% [v/v] gradient) and spun at 620 3 g for 20 min with no brake. Cells at the 40/80 interface were collected and washed twice with supplemented HBSS (10 mM HEPES [Cellgro], 2 mM EDTA [Cellgro], and 0.5% [v/v] fetal bovine serum [GIBCO BRL]) and prepared for flow cytometry analysis. While this isolation strategy is widely used as an approach for isolating lamina propria tissue-resident lymphocytes, it does not prohibit contamination by lymphocytes residing within the microenvironment of isolated lymphoid follicles (ILFs), which are organized lymphoid structures found throughout the lamina propria; therefore we cannot completely exclude the possibility that some of the cells we are assaying reside with ILFs. Flow Cytometry Staining of Isolated Lymphocytes Lymphocytes were isolated from the colonic and small intestinal lamina propria as mentioned above. Spleen, MLN, or Peyer’s patch tissues were gently pushed through a 40 mM filter to obtain white blood cells. Spleen cells were additionally treated with 1X RBC Lysis Buffer (Biolegend) to lyse and remove red blood cells. Surface staining for lymphocytes was done in sterile 1X HBSS (Corning) supplemented with 10 mM HEPES (Cellgro), 2 mM EDTA (Cellgro), and 0.5% (v/v) fetal bovine serum (GIBCO BRL) for 20 min at 4! C. Cells were then washed twice in supplemented 1X HBSS and enumerated via flow cytometry. The following antibodies were used: anti-CD4 (Biolegend: clone GK1.5 APC/Pacific Blue, clone RM4-5 FITC; eBioscience clone RM4-5 PerCP-Cy5.5), anti-CD3 (Biolegend: 145-2C11 Pacific Blue), anti-B220 (Biolegend: clone RA3-6B2 PerCP-Cy5.5), anti-CD138 (Biolegend: clone 281-2 PE), anti-CXCR5 (eBioscience: clone SPRCL5 PE), anti-PD-1 (Biolegend: RMP1-30 Pe/Cy7), anti-ICOS (eBioscience: clone 7E.17G9 FITC), anti-GL-7 (eBioscience: clone GL-7 Alexa Fluor 488), anti-Fas (BD Biosciences: clone Jo2 PE-Cy7), anti-IgD (Biolegend: clone 11-26c.2a Alexa Fluor 647), or antiIgA (Southern Biotech goat anti-mouse IgA FITC). For intracellular staining, cells were first stimulated with ionomycin (500 ng/ml), PMA (5 ng/ml), and Brefeldin A (5 mg/ml) (Biolegend) for 4 hr at 37! C. Cells were surface stained, washed, and then permeabilized and fixed in 100 ml Perm/Fix buffer (eBiosciences) at 4! C overnight. Cells were washed twice in Perm/Wash buffer (eBioscience) and then stained for intracellular cytokines with the following antibodies: anti-Foxp3 (eBioscience: clone FJK-16 s APC/PerCP-Cy5.5), anti-IL-17A (eBioscience: clone eBio17B7 eFluor 660), or anti-IFNg (Biolegend: clone XMG1.2 PE). Cells were again washed twice in Perm/Wash buffer and then placed with supplemented HBSS (10 mM HEPES [Cellgro], 2 mM EDTA [Cellgro], and 0.5% [v/v] fetal bovine serum [GIBCO BRL]) and enumerated via flow cytometry. Gating and analysis of positive cell populations was done utilizing respective isotype controls for antibodies against ICOS, CXCR5, GL-7, and PD-1. Other cell populations were identified with single stain controls. These data were collected with a BD LSR Fortessa and analyzed with FlowJo software. ACCESSION NUMBERS Illumina MiSeq 16S rRNA gene sequences have been deposited in the NCBI sequence read archive (SRA) under the accession number SRP050978. ACKNOWLEDGMENTS We would like to thank members of the Round and O’Connell labs for their critical review of the manuscript. The B. fragilis-OVA strain was created by Dr. Yue Shen and Dr. Sarkis Mazmanian (California Institute of Technology). We thank James Marvin for assistance with flow sorting. The flow cytometry core is supported by the National Center for Research Resources of the National Institutes of Health under Award Number 1S10RR026802-01. Some of the GF mice used in this publication were provided by UNC’s Gnotobiotic Facility, which is supported by grants 5-P39-DK034987 and 5-P40-OD010995. J.L.K. and C.P. are supported by a T32 fellowship in microbial pathogenesis (AI-055434). R.S. was supported by a T32 genetics training grant (GM007464). R.M.O. is supported by the NIH New Innovator Award DP2GM111099-01, the NHLBI R00HL102228-05, an American Cancer Society Research Grant, and a Kimmel Scholar Award. Support for this project comes from the Edward Mallinckrodt Jr. Foundation, Pew Scholars Program, NSF CAREER award (IOS-1253278), Packard Fellowship in Science and Engineering, NIAID K22 (AI95375), NIAID (AI107090, AI109122), and an NIH innovator award (DP2AT008746-01) to J.L.R. Received: November 24, 2014 Revised: December 10, 2014 Accepted: December 19, 2014 Published: January 22, 2015 REFERENCES Atarashi, K., Tanoue, T., Shima, T., Imaoka, A., Kuwahara, T., Momose, Y., Cheng, G., Yamasaki, S., Saito, T., Ohba, Y., et al. (2011). Induction of colonic regulatory T cells by indigenous Clostridium species. Science 331, 337–341. Belkaid, Y., and Hand, T.W. (2014). Role of the microbiota in immunity and inflammation. Cell 157, 121–141. Bemark, M., Boysen, P., and Lycke, N.Y. (2012). Induction of gut IgA production through T cell-dependent and T cell-independent pathways. Ann. N Y Acad. Sci. 1247, 97–116. Berry, D., Schwab, C., Milinovich, G., Reichert, J., Ben Mahfoudh, K., Decker, T., Engel, M., Hai, B., Hainzl, E., Heider, S., et al. (2012). Phylotype-level 16S rRNA analysis reveals new bacterial indicators of health state in acute murine colitis. ISME J. 6, 2091–2106. Brandtzaeg, P. (2013). Secretory IgA: Designed for Anti-Microbial Defense. Front Immunol 4, 222. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Peña, A.G., Goodrich, J.K., Gordon, J.I., et al. (2010). QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336. Caramalho, I., Lopes-Carvalho, T., Ostler, D., Zelenay, S., Haury, M., and Demengeot, J. (2003). Regulatory T cells selectively express toll-like receptors and are activated by lipopolysaccharide. J. Exp. Med. 197, 403–411. SUPPLEMENTAL INFORMATION Cerutti, A., Cols, M., Gentile, M., Cassis, L., Barra, C.M., He, B., Puga, I., and Chen, K. (2011). Regulation of mucosal IgA responses: lessons from primary immunodeficiencies. Ann. N Y Acad. Sci. 1238, 132–144. Supplemental Information includes Supplemental Experimental Procedures, five figures, and two tables and can be found with this article online at http:// dx.doi.org/10.1016/j.chom.2014.12.009. Chang, J., Burkett, P.R., Borges, C.M., Kuchroo, V.K., Turka, L.A., and Chang, C.H. (2013). MyD88 is essential to sustain mTOR activation necessary to promote T helper 17 cell proliferation by linking IL-1 and IL-23 signaling. Proc. Natl. Acad. Sci. USA 110, 2270–2275. AUTHOR CONTRIBUTIONS Crotty, S. (2011). Follicular helper CD4 T cells (TFH). Annu. Rev. Immunol. 29, 621–663. J.L.R., J.L.K., and C.P. designed the study. C.P. performed most of the immunological experiments, including microbiota transfers, disease induction, and GF mouse studies. J.L.K. performed the initial immunology experiments, developed the IgA bacterial purification protocol, and helped analyze the microbiota data. W.Z.S. performed the 16S sequencing and W.Z.S. and J.L.K. analyzed the data. R.S. assisted during mouse harvests and provided comments during the study. J.L.R., J.L.K., C.P., and W.Z.S. wrote the manuscript. Cullender, T.C., Chassaing, B., Janzon, A., Kumar, K., Muller, C.E., Werner, J.J., Angenent, L.T., Bell, M.E., Hay, A.G., Peterson, D.A., et al. (2013). Innate and adaptive immunity interact to quench microbiome flagellar motility in the gut. Cell Host Microbe 14, 571–581. Fagarasan, S., Muramatsu, M., Suzuki, K., Nagaoka, H., Hiai, H., and Honjo, T. (2002). Critical roles of activation-induced cytidine deaminase in the homeostasis of gut flora. Science 298, 1424–1427. 162 Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 29 Hansen, S.G., Skov, M.N., and Justesen, U.S. (2013). Two cases of Ruminococcus gnavus bacteremia associated with diverticulitis. J. Clin. Microbiol. 51, 1334–1336. Peterson, D.A., McNulty, N.P., Guruge, J.L., and Gordon, J.I. (2007). IgA response to symbiotic bacteria as a mediator of gut homeostasis. Cell Host Microbe 2, 328–339. Harmsen, H.J., Pouwels, S.D., Funke, A., Bos, N.A., and Dijkstra, G. (2012). Crohn’s disease patients have more IgG-binding fecal bacteria than controls. Clin. Vaccine Immunol. 19, 515–521. Png, C.W., Lindén, S.K., Gilshenan, K.S., Zoetendal, E.G., McSweeney, C.S., Sly, L.I., McGuckin, M.A., and Florin, T.H. (2010). Mucolytic bacteria with increased prevalence in IBD mucosa augment in vitro utilization of mucin by other bacteria. Am. J. Gastroenterol. 105, 2420–2428. Hooper, L.V., Littman, D.R., and Macpherson, A.J. (2012). Interactions between the microbiota and the immune system. Science 336, 1268–1273. Ivanov, I.I., Frutos, Rde.L., Manel, N., Yoshinaga, K., Rifkin, D.B., Sartor, R.B., Finlay, B.B., and Littman, D.R. (2008). Specific microbiota direct the differentiation of IL-17-producing T-helper cells in the mucosa of the small intestine. Cell Host Microbe 4, 337–349. Kamada, N., Seo, S.U., Chen, G.Y., and Núñez, G. (2013). Role of the gut microbiota in immunity and inflammatory disease. Nat. Rev. Immunol. 13, 321–335. Kawamoto, S., Tran, T.H., Maruya, M., Suzuki, K., Doi, Y., Tsutsui, Y., Kato, L.M., and Fagarasan, S. (2012). The inhibitory receptor PD-1 regulates IgA selection and bacterial composition in the gut. Science 336, 485–489. Kubinak, J.L., and Round, J.L. (2012). Toll-like receptors promote mutually beneficial commensal-host interactions. PLoS Pathog. 8, e1002785. Lindner, C., Wahl, B., Föhse, L., Suerbaum, S., Macpherson, A.J., Prinz, I., and Pabst, O. (2012). Age, microbiota, and T cells shape diverse individual IgA repertoires in the intestine. J. Exp. Med. 209, 365–377. Linterman, M.A., Liston, A., and Vinuesa, C.G. (2012). T-follicular helper cell differentiation and the co-option of this pathway by non-helper cells. Immunol. Rev. 247, 143–159. Lycke, N.Y., and Bemark, M. (2012). The role of Peyer’s patches in synchronizing gut IgA responses. Front Immunol 3, 329. Mantis, N.J., Rol, N., and Corthésy, B. (2011). Secretory IgA’s complex roles in immunity and mucosal homeostasis in the gut. Mucosal Immunol. 4, 603–611. Mazmanian, S.K., Liu, C.H., Tzianabos, A.O., and Kasper, D.L. (2005). An immunomodulatory molecule of symbiotic bacteria directs maturation of the host immune system. Cell 122, 107–118. McFall-Ngai, M., Hadfield, M.G., Bosch, T.C., Carey, H.V., Domazet-Lo!so, T., Douglas, A.E., Dubilier, N., Eberl, G., Fukami, T., Gilbert, S.F., et al. (2013). Animals in a bacterial world, a new imperative for the life sciences. Proc. Natl. Acad. Sci. USA 110, 3229–3236. Palm, N.W., de Zoete, M.R., Cullen, T.W., Barry, N.A., Stefanowski, J., Hao, L., Degnan, P.H., Hu, J., Peter, I., Zhang, W., et al. (2014). Immunoglobulin A coating identifies colitogenic bacteria in inflammatory bowel disease. Cell 158, 1000–1010. Parks, D.H., and Beiko, R.G. (2010). Identifying biologically relevant differences between metagenomic communities. Bioinformatics 26, 715–721. Raetz, M., Hwang, S.H., Wilhelm, C.L., Kirkland, D., Benson, A., Sturge, C.R., Mirpuri, J., Vaishnava, S., Hou, B., Defranco, A.L., et al. (2013). Parasiteinduced TH1 cells and intestinal dysbiosis cooperate in IFN-g-dependent elimination of Paneth cells. Nat. Immunol. 14, 136–142. Reynolds, J.M., Pappu, B.P., Peng, J., Martinez, G.J., Zhang, Y., Chung, Y., Ma, L., Yang, X.O., Nurieva, R.I., Tian, Q., and Dong, C. (2010). Toll-like receptor 2 signaling in CD4(+) T lymphocytes promotes T helper 17 responses and regulates the pathogenesis of autoimmune disease. Immunity 32, 692–702. Round, J.L., and Mazmanian, S.K. (2010). Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc. Natl. Acad. Sci. USA 107, 12204–12209. Round, J.L., Lee, S.M., Li, J., Tran, G., Jabri, B., Chatila, T.A., and Mazmanian, S.K. (2011). The Toll-like receptor 2 pathway establishes colonization by a commensal of the human microbiota. Science 332, 974–977. Schenten, D., Nish, S.A., Yu, S., Yan, X., Lee, H.K., Brodsky, I., Pasman, L., Yordy, B., Wunderlich, F.T., Brüning, J.C., et al. (2014). Signaling through the adaptor molecule MyD88 in CD4+ T cells is required to overcome suppression by regulatory T cells. Immunity 40, 78–90. Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., et al. (2009). Introducing mothur: open-source, platform-independent, communitysupported software for describing and comparing microbial communities. Appl. Environ. Microbiol. 75, 7537–7541. Shulman, Z., Gitlin, A.D., Targ, S., Jankovic, M., Pasqual, G., Nussenzweig, M.C., and Victora, G.D. (2013). T follicular helper cell dynamics in germinal centers. Science 341, 673–677. Slack, E., Hapfelmeier, S., Stecher, B., Velykoredko, Y., Stoel, M., Lawson, M.A., Geuking, M.B., Beutler, B., Tedder, T.F., Hardt, W.D., et al. (2009). Innate and adaptive immunity cooperate flexibly to maintain host-microbiota mutualism. Science 325, 617–620. Slack, E., Balmer, M.L., Fritz, J.H., and Hapfelmeier, S. (2012). Functional flexibility of intestinal IgA - broadening the fine line. Front Immunol 3, 100. Stephens, W.Z., and Round, J.L. (2014). IgA targets the troublemakers. Cell Host Microbe 16, 265–267. van der Waaij, L.A., Kroese, F.G., Visser, A., Nelis, G.F., Westerveld, B.D., Jansen, P.L., and Hunter, J.O. (2004). Immunoglobulin coating of faecal bacteria in inflammatory bowel disease. Eur. J. Gastroenterol. Hepatol. 16, 669–674. Cell Host & Microbe 17, 153–163, February 11, 2015 ª2015 Elsevier Inc. 163 30 Cell Host & Microbe, Volume 17 Supplemental Information MyD88 Signaling in T Cells Directs IgA-Mediated Control of the Microbiota to Promote Health Jason L. Kubinak, Charisse Petersen, W. Zac Stephens, Ray Soto, Erin Bake, Ryan M. O’Connell, and June L. Round 31 A B MyD88+/+ MyD88fl/fl MyD88fl/fl CD4-Cre+ MyD88+/+ CD4-Cre+ MyD88fl/fl! CD4-Cre ! C WT T-MyD88-/- MyD88 Fig. S1. 32 A WT T-MyD88-/- B 0.15 C ns 8 6 IL17A TH1 (%) 0.05 0.00 T-MyD88-/- 4 2 0 WT T-MyD88-/- WT T-MyD88-/- ns 3 TREG (%) WT E 4 Spleen TH17 (%) 0.10 IFNγ D ns CD4 2 1 FoxP3 F 0 WT T-MyD88-/- G 0.8 WT T-MyD88-/- ns h 0.6 TH1 (%) TH17 (%) IL17A I WT T-MyD88-/- J 16 WT T-MyD88-/- ns CD4 TREG (%) 12 FoxP3 8 4 0 WT T-MyD88-/- Fig. S2. 0.5 0.0 MLN 0.2 0.0 ns 1.0 0.4 IFNγ 1.5 WT T-MyD88-/- 33 A B WT C T-MyD88-/- ns 8 8 6 4 2 0 WT T-MyD88-/- E WT T-MyD88-/- 4 2 0 WT T-MyD88-/- ns 25 cLP IFNγ TH1 (%) IL17A TH17 (%) 6 D ns CD4 TREG (%) 20 WT 10 5 0 FoxP3 F 15 T-MyD88-/- G 25 WT T-MyD88-/H ns 6 ** TH1 (%) 15 IL17A 10 I 5 0 IFNγ WT T-MyD88-/- J 4 2 0 WT T-MyD88-/- WT T-MyD88-/- siLP TH17 (%) 20 * 30 TREG (%) 20 CD4 10 0 FoxP3 K WT T-MyD88-/- L WT T-MyD88-/** 2.5 M 3 ns 2 h 1.5 0.5 T-MyD88-/- WT T-MyD88-/- 0 WT T-MyD88-/- ns 15 TREG (%) WT 0.0 O 1 PP IL17A 1.0 IFNγ N g TH1 (%) TH17 (%) 2.0 CD4 10 FoxP3 j 5 0 WT T-MyD88-/- Fig. S3. 34 A WT B T-MyD88-/- * 50 TFH (%) 40 ICOS 30 20 10 PD-1 C 0 WT D T-MyD88-/- 50 WT T-MyD88-/*** 30 TFH (%) CXCR5 PP 40 20 10 PD-1 D WT T-MyD88-/- 0 F 50 WT T-MyD88-/** CXCR5 20 10 0 WT T-MyD88-/- H 50 non-GC TFH (%) G 30 WT T-MyD88-/* 40 20 10 PD-1 WT T-MyD88-/- K ICOS CXCR5 WT ICOS 3 T-MyD88-/- WT T-MyD88-/p=0.05 2 1 0 L Spleen 0 TFH (%) CXCR5 MLN J WT T-MyD88-/- M 6 4 TFH (%) CXCR5 30 PP Total Cells ICOS TFH (%) 40 2 0 WT T-MyD88-/- I 1!107 8!106 6!106 4!106 2!106 0 WT T-MyD88-/- 35 IgA Bound A Mucosal Fecal Adlercreutzia * *** Anaeroplasma Bifidobacterium * *** Clostridium Coprococcus Dorea Halomonas Lactobacillus * Mucispirillum Nesterenkoria Odoribacter Oscillospira Prevotella Ruminococcus * *** * *** *** SMB53 * * Turicibacter UC Christensenellaceae * UC Costridiaceae UC Clostridiales (Order) UC Desulfovibrionaceae UC Enterobacteriaceae UC Lachnospiraceae UC Mogibacteriaceae * UC Peptostreptococcaceae UC Planococcaceae * UC Ruminococcaceae UC S24-7 0 * 61.2 0 50.1 0 56.6 Mean Proportion % Desulfovibrionaceae Enterobacteriaceae Mucispirillum Ruminococcus *** Relative Abundance (%) B WT TMyD88-/- WT TMyD88-/- Fig. S5. WT TMyD88-/- WT TMyD88-/- 36 Sample Subset Cohoused Genotypes Separately Housed Genotypes Test Unweighted UniFrac Weighted UniFrac Bray-Curtis Unweighted UniFrac Weighted UniFrac Bray-Curtis Table S1. Mucosa IgA Bound Fecal WT vs. TMyD88 WT vs. TMyD88 WT vs. TMyD88 WT Only TMyD88-/- Only Mucosal vs. IgA Mucosal vs. IgA Bound Bound WT Only Fecal vs. Mucosal TMyD88-/- Only Fecal vs. Mucosal 0.0005 0.0013 0.0005 0.0023 0.0004 0.0016 0.0123 0.0014 0.0087 0.1245 0.0143 0.0087 0.0094 0.0004 0.0007 0.0011 0.0005 0.0021 0.0004 0.0024 0.0022 0.0111 0.1267 0.1000 0.0002 0.0001 0.0006 0.0001 0.4578 0.4100 0.6234 0.0007 0.0020 0.0053 0.0020 0.0681 0.0519 0.0827 0.0002 0.0001 0.0138 0.0002 37 WT Mean † T-MyD88-/Mean † P-value P-value (FDR) 263907 0.90 4.45 0.0005 0.0233 Clostridiales Lachnospiraceae UNCLASSIFIED 274021 2.90 11.55 0.0005 0.0233 Clostridiales Lachnospiraceae UNCLASSIFIED 262104 4.10 0.82 0.0034 0.1054 Clostridiales Lachnospiraceae UNCLASSIFIED 327900 0.50 4.09 0.0118 0.2434 Clostridiales Ruminococcus UNCLASSIFIED 214116 13.40 2.45 0.0143 0.2434 Clostridiales UNCLASSIFIED UNCLASSIFIED 276312 4.80 16.91 0.0157 0.2434 Clostridiales UNCLASSIFIED UNCLASSIFIED 311695 18.70 56.91 0.0207 0.2581 Clostridiales UNCLASSIFIED UNCLASSIFIED de novo 2.40 8.18 0.0222 0.2581 Clostridiales Lachnospiraceae UNCLASSIFIED 209030 8.20 14.09 0.0459 0.4650 Bacteroidales UNCLASSIFIED UNCLASSIFIED Reference OTU ID * Order * Greengenes 13_8 Reference OTU IDs, except as noted for single de novo OTU. † Mean abundance per 1500 sequences per sample Table S2. Family Genus 38 METHODS Mice Conventionally colonized (SPF) mice. C57Bl/6 MyD88 LoxP/LoxP mice (Jackson Laboratories) were crossed to C57Bl/6 CD4-Cre animals (Taconic) to produce MyD88WT/WT CD4-Cre+ mice (WT) and MyD88LoxP/LoxPCD4-Cre+ (T-MyD88-/-) animals. Age matched male and female animals were used to compare phenotypic differences between genotypes as well as severity of TNBS induced colitis. For the antibiotics experiment, age matched FoxP3-GFP+ mice (Jackson Laboratories) were compared for phenotypic differences in TFH cells within gut associated lymphoid tissues. Aged matched female C57Bl/6 WT and IL1R-/- mice (Jackson Laboratories) were used to compare TFH development and germinal center responses in this absence of IL1 signaling. The use of animals in all experiments was in strict adherence to federal regulations as well as the guidelines for animal use set forth by the University of Utah Institutional Animal Care and Use Committee. Germ-free (GF) mice. GF mice were maintained in sterile isolators and verified monthly for GF status by plating and PCR of feces. GF BALB/c, GF C57Bl/6, and GF RAG1-/animals were used in this study. Age matched GF BALB/c mice were compared to SPF BALB/c mice to compare phenotypic differences in TFH cells within gut associated lymphoid tissues. The TLR-ligand experiment: GF C57Bl/6 mice were born GF and then given ad libitum access to drinking water containing 10µg/mL of Pam3CSK4 (Invivogen) for two weeks. Bone marrow (BM) reconstitutions of the T and B cell compartments: GF Rag1-/- mice were reconstituted with bone marrow (~2.5x106 cells) from WT and TMyD88-/- mice via retro-orbital injection. Mono-association experiments,: BM reconstituted GF Rag1-/- mice were colonized with an erythromycin/gentamicin resistant strain of Bacteroides fragilis that was engineered to express OVA (B.fragilis-OVA; kindly provided by Dr. Sarkis Mazmanian (California Institute of Technology)). Animals were maintained for two months with 1mg/mL of both erythromycin and gentamicin in drinking water under SPF housing conditions and were validated to be correctly colonized by aerobic & anaerobic plating as well as PCR of feces. BM reconstituted GF Rag1-/- mice were analyzed two-months post-colonization. TLR-ligand feeding experiment: reconstituted GF Rag1-/- mice were maintained on 1mg/mL of ampicillin (Fisher Scientific), neomycin (Fisher Scientific), erythromycin (Fisher Scientific), and gentamicin (GoldBio) for two months. During the last two weeks, 10µg/mL of Pam3CSK4 (Invivogen) was added to antibiotic cocktails. For IL1β blocking experiments, GF Rag1-/mice were reconstituted with WT bone marrow (~2.5x106 cells). Reconstituted mice were maintained on 1mg/mL of ampicillin (Fisher Scientific), neomycin (Fisher Scientific), erythromycin (Fisher Scientific) and gentamicin (GoldBio) for two months. 10µg/mL of Pam3CSK4 (Invivogen) was added to antibiotic cocktails during the final two weeks of the experiment. During this two-week period, 50µg of anti-IL1β neutralizing antibody (Thermo Scientific Cat#MM425B) or 50µg of IgG control antibody (Thermo Scientific Cat#31903) were administered to mice every other day via retro-orbital injection1 . Mice were sacrificed less than 24 hours after the final injection of antibody. 39 Enzyme-linked immunosorbant assay (ELISA) To quantify luminal IgA, colons were cut open longitudinally and feces and mucus were scraped out and placed into 1.5mL Eppendorf tubes. Luminal contents were resuspended in 500µL of sterile 1X HBSS and spun at 400 x g for 5 minutes to remove course materials. Supernatants were then placed in a new 1.5mL Eppendorf tube and spun at 8000 x g for 5 minutes to pellet bacteria. This step was repeated until samples were clear of bacterial pellets. Supernatants (containing IgA) were then placed in a new 1.5mL Eppendorf tube and used as samples (1/10 and 1/100 (v/v) dilutions) for an IgAspecific ELISA kit (eBioscience; performed per kit instructions). Absorbance was read at 450nm and concentrations of IgA were calculated with a standard curve. Concentrations were normalized to fecal weight. For quantification of OVA specific IgA, supernatants containing IgA were collected as above and IgA quantification was performed with the same IgA-specific ELISA kit with slight modification. Instead of coating plates with capture antibody specific for IgA, plates were coated overnight at 4°C with 4µg/mL ovalbumin in 1X PBS. Absorbance was read at 450nm and normalized to fecal weight. T cell isolation Lymphocytes were isolated from spleens and the resulting cells were sorted through MACS columns with either positive (CD4 microbeads (Miltenyi)) or negative selection (CD4+ T Cell Isolation Kit II (Miltenyi)). Lymphocyte enrichment was performed following kit instructions. Isolated T cells were further purified via FACS with a BD FACSAria Cell Sorter. For Naïve T cells, CD3+CD4+CD62L+CD25- cells were collected into RPMI media (Corning) supplemented with 10% fetal bovine serum (v/v) (Gibco BRL), 50 U/mL penicillin, 50 µg/mL streptomycin (Cellgro), 5µM 2-Mercaptoethanol (CalBiochem), 1µM sodium pyruvate (Cellgro), 1X MEM Nonessential Amino Acids (Cellgro), and 2.05mM L-glutamine (Cellgro). Purified TFH cells (CD3+CD4+B220-CXCR5+PD1+) and TFHdepleted T cells (CD3+CD4+B220-CXCR5-PD1-) were collected in supplemented RPMI media. TNBS-Induced Colitis and Microbial Rescues Age matched WT and T-MyD88-/- mice were challenged with TNBS as previously described4 with a few modifications. Briefly, 100µL per mouse of 2.5% (v/v) Picrylsulfonic acid (Sigma) in 100% EtOH was administered intra-rectally with a silicone catheter (Solomon Scientific). Weights were collected just prior to challenge and for five days following challenge to quantify weight loss. On the fifth day after challenge a second dose of TNBS was administered and tissues were collected for histology two days later. Histology was performed blindly according to previously published criteria (9). Briefly, both crypt loss and inflammation was scored according to the following guidelines. The entire length of the colon from just under the cecum to the rectum was analyzed and percent of crypt loss and percent of colon affected by inflammation was also taken into consideration. Therefore each animal received a score for crypt loss and inflammation severity as well as the percent of the colon affected. For crypt loss severity 40 a score of 0-3 was given (0=no crypt loss; 1=mild crypt loss, most crypts still visible with a few areas effected; 2=medium severity, greater crypt loss, fewer crypts visible in large areas; 3=very large areas of total crypt loss, places where crypts are completely gone). For inflammation a score was given from 0-3 (0=no evidence of inflammatory infiltrate; 1=very low level of cells infiltrating into the tissue; 2=thickening of lamina propria, and clear infiltrating lymphocytes into epithelial tissue; 3=thickening of lamina propria and large boluses of inflammatory infiltrates that correspond with areas a crypt loss). The percentage scoring was as follows: 0=no area affected; 0.5=1-5%; 1=5-20%; 1.5=2030%; 2=30-45%; 2.5=45-60%; 3:=60-70%; 3.5=70-80%; 4=>80%. This was the same scoring system used for crypt loss and inflammation. For microbiota-rescue experiments, T-MyD88-/- mice were treated with 0.5mg/mL of ampicillin (Fisher Scientific), neomycin (Fisher Scientific), erythromycin (Fisher Scientific), and gentamicin (GoldBio) for one week. Antibiotic treatment was then terminated and mice were orally gavaged daily for one week with 100uL of microbiota stock derived from either WT or T-MyD88-/- mice. Microbiota stocks were prepared from luminal contents (fecal material and mucus scrapings) that were suspended in sterile 1X HBSS (1mL/gram materials) and then gently spun down at 400 x g for 5 minutes to remove course materials. Mice were administered TNBS as described above one week after the last gavage. qPCR measurement of total bacterial load in feces and mucosa. Animals were sacrificed and a single fecal pellet as well as a 1cm length of colon (immediately distal to the cecum) were collected, placed in 1X cell lysis solution (sterile water containing 10mM TRIS HCL-pH8.0, 25mM sterile EDTA, and 1% (v/v) sterile SDS), and stored at -20oC until DNA extraction. Prior to storage, colon sections were gently flushed of residual feces with sterile 1X HBSS buffer. A two-day DNA isolation protocol (with a 24 hour proteinase K lysis step) was followed. Briefly, one mouse fecal pellet (or a 1cm snippet of colon) was placed in 500µl 1X cell lysis. 3µl of proteinase K (20mg/mL (Thermo Scientific)) was added and then samples were placed in a thermal block for 24 hours at 55oC. Samples were then cooled to RT, then 3µl of RNase A (10mg/mL (Thermo Scientific)) were added and samples were incubated at 37oC for 30 minutes. Samples were then cooled to RT and 200µl of 5M ammonium acetate was. To precipitate out proteins, samples were subsequently vortexed for 10 seconds, placed in an ice bath for 10 minutes, and then centrifuged at 20000 x g for 6 minutes. Supernatants (excluding pelleted proteins) were then placed in a new 1.5mL Eppendorf tube and mixed with 600µl of 100% isopropanol containing glycogen (20mg/mL (Thermo Scientific)). Samples were gently inverted 50 times to mix and then left to sit at RT for 10 minutes. Samples were spun at 20000 x g for 6 minutes. Two washes of the DNA pellet were then performed. For each wash, DNA pellets were re-suspended in 600µl of 70% (v/v) EtOH and spun at 20,000 x g for 6 minutes. After the second wash, tubes were inverted onto a sterile tissue and allowed to air-dry for 15 minutes. DNA pellets were then re-suspended in 100µl sterile H2O and stored at -20oC. Sample DNA concentrations were quantified with a Nanodrop spectrophotometer (Thermo Scientific). The total amounts of bacteria within the feces and mucosa was then quantified with Eubacteria-specific 16S primers (see Supplementary Table 1). All qPCR reactions were conducted in 12.5µl volumes with the GoTaq qPCR Master Mix (Promega). qPCR experiments were conducted on a Lightcycler LC480 instrument (Roche) with the parameters described above. Flow cytometric analysis of IgA-bound Bacteria. 41 Fecal pellets were collected from animals and homogenized in 500µl of sterile 1X HBSS buffer. Samples were spun once at 400 x g to remove course materials from the fecal suspension. Supernatants (containing bacteria) were collected and placed in a new 1.5mL tube. Samples were spun at 8000 x g for 5 minutes to pellet bacteria. Bacterial pellets were washed twice by re-suspending pellet in 500µl of sterile 1X HBSS and spinning for 5 minutes at 8000 x g. Bacterial suspensions were then blocked on ice for 15 minutes in 500µl sterile 1X HBSS containing 1% (v/v) BSA. Samples were spun at 8000 x g for 5 minutes. Bacterial pellets were then stained for 20 minutes at 4oC in the dark in 500µl of sterile 1X HBSS containing a 1/250 (v/v) dilution of a rat anti-mouse IgA antibody conjugated to a PE fluorochrome (Southern Biotech, cat#1165-09L). Pellets were washed twice as above in sterile 1X HBSS containing 1% (v/v) BSA. Pellets were then stained for 20 minutes at 4oC in 500µl of sterile 1X HBSS containing SYBR green I (1/10000 (v/v) dilution, Life Technologies). After SYBR staining samples were analyzed on an LSRFortessa flow cytometer (BD Biosciences). A representative flow cytometry plot is provided. Isolation and 16S rRNA sequencing of IgA-bound, fecal and mucosal bacteria. Animals were sacrificed and their entire lower digestive tract (from duodenum to rectum) was removed and longitudinally sectioned. One fecal pellet and a 1cm section of the distal colon were collected from each animal to characterize the fecal and mucosal microbiota communities, respectively. The colon section was scraped and rinsed in sterile 1X HBSS during collection. Fecal and mucosa samples were immediately frozen at -80oC in 2mL screw cap tubes containing ~250 mg of 0.15 mm garnet beads (MoBio, cat# 13122-500) for down-stream DNA extraction. For IgA-bound sample collection the luminal contents, including mucosal scrapings, from distal ileum to rectum were collected with forceps, placed in a 15mL conical tube, and spun briefly to sediment materials. 2mL of sterile 1X HBSS (without Ca2+) was then added to each sample. Samples were vortexed at medium speed for 20 seconds and then spun for 5 minutes at 60 x g at 4oC to collect coarse material. 1.5 mL of the supernatant was collected in a 1.5 mL Eppendorf tube and kept on ice while 2mL of new sterile 1X HBSS was added to the sample. This process was repeated 4 times to collect 4-1.5ml Eppendorf tubes per animal sample containing bacteria separated from the most coarse fecal material. Tubes were then spun at 5000 x g for 5 minutes and supernatant was discarded to remove unbound Ig. Pellets were washed with 600µl sterile 1X HBSS and 2 sets of 21.5mL Eppendorf tubes from each sample were combined. Tubes for each sample were then spun again at 5000 x g for 5 minutes. Supernatants were discarded and the 42 pellets were re-suspended in 500µl of sterile 1X HBSS with 0.1% (v/v) BSA. Pairs of tubes for each sample were then combined into a single 1.5mL Eppendorf tube to obtain ~1 mL of bacterial suspension depleted of unbound Ig. This suspension contained Ig-bound and unbound bacteria. 125µl of streptavidin-coated magnetic beads (CELLection Biotin Binder Kit, Life Technologies cat. # 11533D) per sample were washed and pre-incubated with 10µg of biotinylated anti-mouse IgA (Biolegends, cat. # 407004) for 30 minutes at room-temperature (with gentle rocking) according to the manufacturer’s specifications. The pre-incubated bead-antibody mixture was then added to the 1mL of bacterial sample and incubated for 30 minutes at 4oC with gentle shaking. After incubation, 1mL of sterile 1X HBSS supplemented with 0.1% (v/v) BSA and 2 mM EDTA was added. Samples were transferred to a sterile 5mL Falcon tube (BD Biosciences) and placed on a magnet. The solution contained unbound bacteria was removed by pipette and the tube was then removed from the magnet. Magnetic beads (with IgA-bound bacteria) were re-suspended in 1X HBSS supplemented with 0.1% (v/v) BSA to wash. This was done three times to thoroughly wash beads containing IgA-bound bacteria. Buffer was removed and samples were washed three then washed three times with HBSS with 0.1% (v/v) BSA. After the final wash was removed, samples were removed from the magnet and 200 µl of sterile TE (pH 8.0) with 0.1% (v/v) Tx-100 (filter sterilized through 0.22 µm filter) was added. The suspended magnetic bead and IgA-bound bacteria mixture was added to 2 mL screw cap tubes containing ~250 mg of 0.1 mm zirconia/silica beads (Biospec, cat. # 11079101z) and immediately frozen and stored at -80oC for down-stream DNA extraction. IgA-bound, fecal and mucosal samples were all processed similarly for DNA extraction and sequencing. First, each sample that already contained beads and 200µl TE with 0.1% (v/v) Tx-100 was thawed at 75oC for 5 minutes and then chilled on ice. 200µl lysis buffer AL was added (Qiagen, cat. # 19075) and bead-beating for 1 minutes with a Mini-Beadbeater-16 (Biospec, cat. # 607) was performed. Mucosal tissue samples were bead-beat for an additional minute to complete homogenization, with 5 minutes on ice in between beatings to prevent sample heating. Samples were then spun down for 1 minute at 8,000 x g to avoid foaming prior to adding 200µl of 100% ethanol and vortexing. Samples were spun down again to avoid transfer of any remaining coarse material and supernatant was added to a Qiagen DNeasy (Qiagen, cat. # 69504) (fecal and mucosal samples) or Qiagen DNA micro (Qiagen, cat. # 56304)(IgA-bound) column. IgA-bound samples were first placed against a magnet to remove the magnetic beads prior to loading on columns. All samples were then further processed according to Qiagen’s specifications after loading on columns. DNA was eluted in 65µl of Qiagen’s buffer AE. In order to amplify the bacterial 16S rRNA and obtain high-quality, long reads, we targeted the hypervariable regions 3 and 4, initially primer sequences based off the S-DBact-0341-b-S-17/S-D-Bact-0785-a-A-21 pair identified by Klindworth, et al 20135. However, empirically, we determined these primers to have strong amplification efficiency of host genomic and mitochondrial DNA in addition to the bacterial 16S rRNA gene. We modified them to avoid human and mouse amplification while having similar in silico predicted coverage of bacterial taxa. Then, in a strategy based off that developed by Kozich, et al . 20136, we added 2 nucleotide linkers non-complementary to most bacteria sequences followed by Illumina adapter sequences with different 8 43 nucleotide barcodes on forward and reverse primers. We differ from Kozich’s strategy in that we used Illumina’s sequencing primers to initiate reads and subsequently trimmed our primer sequences from the raw reads, instead of using the 16S primers themselves to initiate Illumina sequencing reads. The modified primers (Supplemental Table 1) resulted in less than 0.06% of our quality-filtered reads from either end aligning to the host (mouse) genome (data not shown). PCR cycling conditions were as follows: 98oC initial denaturation for 2 minutes; 26 cycles of 98oC for 30 sec seconds, 53.6oC anneal for 20 seconds, 72oC extension for 30 seconds; final single extension at 72oC for 2 minutes. Each PCR was done in triplicate with Phusion HotStart II (Thermo Scientific, cat. # F-549L) with the supplied GC buffer and 200nM of each primer in 25µl reaction volumes, then combined and cleaned up with ZR-96 DNA cleanup-kit (Zymogen, cat. #D4017). Cleaned, barcoded PCR amplicons from each sample were multiplexed then mixed with PhiX control (5% of final library)(Illumina, cat. #FC-110-3001) to increase base diversity, and sequenced on an Illumina MiSeq at the University of Utah’s highthroughput sequencing core with paired-end 300 cycle sequencing. De-multiplexed sequences were processed with mothur7, Qiime 8, and custom perl scripts. Briefly, mothur’s make.contigs was first used to trim primer and linker sequences and combine each set of paired-end reads into a single long contig, requiring a minimum of 20 nucleotides overlap (thereby discarding contigs greater than 536 nucleotides after primer and linker trimming) and a maximum quality score difference of 6 between overlapping bases. We then discarded sequences that had any ‘N’ nucleotides (initially uncalled by MiSeq or introduced by mothur due to difference in pairs of overlapping reads), homopolymer stretches greater than 15, or those that did not align to the region targeted by our primers. Finally, quality-filtered long contigs were used in Qiime 1.8.0 to pick open-reference 97% OTUs with uclust 9 and make taxonomic calls against the Greengenes 13_8 10reference set and taxonomy, requiring a minimum OTU cluster size containing 10 sequences. Chimeric sequences were subsequently screened out with ChimeraSlayer 11 and a phylogeny made with the program FastTree 212. All analyses involving fecal and mucosal samples were based off an OTU table rarefied to 1,500 sequences per sample allowing us to maximize retention of biological replicates, whereas analyses involving comparisons within only IgA-bound samples were rarefied to 5,100 sequences per sample due to a more even sequencing depth among these samples. qPCR validation of conditional MyD88 expression knockout. CD4+, CD8+, and non T cells were sort purified via FACS on a FACS ARIA instrument (BD) from the spleens of a WT and T-MyD88-/- animal and qPCR was used to determine the fold reduction in MyD88 expression. Briefly, cellular RNA was extracted via phenolchloroform precipitation and cDNA was created with the qScript cDNA synthesis kit (Quanta Biosciences). qPCR was conducted as described above with the GoTaq qPCR Master Mix (Promeg ! ! Primers'used'in'this'study.' Primer Name Use Sequence (5' - 3') * Reference 44 ill_S%D%Bact%0346%a%S%171†1 Microbiota Sequencing AATGATACGGCGACCACCGAGATCTACACXXXXXXXXACACTCTTT CCCTACACGACGCTCTTCCGATCTTAGGGRGGCWGCAGTRRGG ill_S%D%Bact%0781%b%A%231†1 Microbiota Sequencing CAAGCAGAAGACGGCATACGAGATXXXXXXXXGTGACTGGAGTTCA GACGTGTGCTCTTCCGATCTTTCTACHVGGGTATCTAATCCTGTT Total Bacteria qPCR UniF340 Total Bacteria qPCR UniR514 ACTCCTACGGGAGGCAGCAGT ATTACCGCGGCTGCTGGC OVAqPCR F B. fragilis OVA qPCR AGAAATGTCCTTCAGCCAAGCTC OVAqPCR R MyD88-F B. fragilis OVA qPCR GCCCATAGCCATTAAGACAGATGTG MyD88 qPCR KO validation MyD88-R MyD88 qPCR KO validation CCCACTCGCAGTTTGTTG Barman, et al. 13 2008 Barman, et al. 13 2008 Ling, et al. 2013 14 14 TGCCTCCCAGTTCCTTTG Ling, et al. 2013 * Bold nucleotides indicate linker sequence, underlined nucleotides indicate the 16S rRNA gene targeting sequences, the 8 nucleotide barcode position is indicated by 'X' 15 † Naming after "ill_" by convention in (Alm., 1996. ) ! Statistics Pair-wise comparison of experimental groups in Figures 1, 2, 3, and 5, as well as their respective Extended Data Figures, were performed with an unpaired two-tailed Student’ t-test. A Welch’s correction was used for data sets with unequal variance (Figure 1b, 1c, 1f, 2g, 3a, 3g). For microbial community similarity comparisons, pairwise comparisons of phylogenetic similarity (Figure 4b, 4d, 5b, 5e) was performed with an unpaired twotailed non-parametric t test with 9999 Monte Carlo simulations. Significance testing of genotype and sample type effects in 4a, 4c and 5a were performed with a PERMANOVA incorporating 9,999 permutations on relevant distance matrices as noted. A Mantel’s test was used to test for significant correlations between distance matrices summarized in Figure 6b. Estimates of dispersion in all figures represent standard deviation around the mean with the exception of Figure 7a which represents standard error. REFERENCES 1 2 3 4 5 Cecic, I. & Korbelik, M. Mediators of peripheral blood neutrophilia induced by photodynamic therapy of solid tumors. Cancer Lett 183, 43-51 (2002). Atarashi, K. et al. Induction of colonic regulatory T cells by indigenous Clostridium species. Science 331, 337-341, doi:10.1126/science.1198469 (2011). Round, J. L. et al. The Toll-like receptor 2 pathway establishes colonization by a commensal of the human microbiota. Science 332, 974-977, doi:10.1126/science.1206095 (2011). Round, J. L. & Mazmanian, S. K. Inducible Foxp3+ regulatory T-cell development by a commensal bacterium of the intestinal microbiota. Proc Natl Acad Sci U S A 107, 12204-12209, doi:10.1073/pnas.0909122107 (2010). Klindworth, A. et al. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res 41, e1, doi:10.1093/nar/gks808 (2013). 45 6 7 8 9 10 11 12 13 14 15 Kozich, J. J., Westcott, S. L., Baxter, N. T., Highlander, S. K. & Schloss, P. D. Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl Environ Microbiol 79, 5112-5120, doi:10.1128/AEM.01043-13 (2013). Schloss, P. D. et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol 75, 7537-7541, doi:10.1128/AEM.01541-09 (2009). Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods 7, 335-336, doi:10.1038/nmeth.f.303 (2010). Edgar, R. C. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460-2461, doi:10.1093/bioinformatics/btq461 (2010). McDonald, D. et al. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. Isme J 6, 610-618, doi:10.1038/ismej.2011.139 (2012). Haas, B. J. et al. Chimeric 16S rRNA sequence formation and detection in Sanger and 454-pyrosequenced PCR amplicons. Genome Res 21, 494-504, doi:10.1101/gr.112730.110 (2011). Price, M. N., Dehal, P. S. & Arkin, A. P. FastTree 2--approximately maximumlikelihood trees for large alignments. PLoS One 5, e9490, doi:10.1371/journal.pone.0009490 (2010). Barman, M. et al. Enteric salmonellosis disrupts the microbial ecology of the murine gastrointestinal tract. Infect Immun 76, 907-915, doi:10.1128/IAI.0143207 (2008). Ling, H. P. et al. Expression of intestinal myeloid differentiation primary response protein 88 (Myd88) following experimental traumatic brain injury in a mouse model. J Surg Res 179, e227-234, doi:10.1016/j.jss.2012.03.030 (2013). Alm, E. W., Oerther, D. B., Larsen, N., Stahl, D. A. & Raskin, L. The oligonucleotide probe database. Appl Environ Microbiol 62, 3557-3559 (1996). Supplementary Figure Legends -/- Fig. S1. T-MyD88 mice lack expression of MyD88 specifically within T cells (CD4+ and CD8+), Related to Figure 1. (A) A schematic representing the insertion of loxP sites flanking fl/fl exon 3 in MyD88 animals. Primer locations (shown in blue) and expected PCR product lengths shown on the representative gel. DNA bands from CD4-Cre specific primers are also shown on a representative gel. (B) MyD88 mRNA expression was measured by qPCR as +/+ + described in the methods section. WT, defined by MyD88 CD4-Cre were compared to T-/fl/fl + MyD88 , defined as MyD88 CD4-Cre (C) RNA-sequencing of samples from genotypes -/revealed a specific loss of RNA from the third exon of the MyD88 gene within T-MyD88 T cells. Fig. S2. Analysis of TH1, TH17, and TREG population frequencies within spleens and -/- mesenteric lymph nodes (MLNs) of WT and T-MyD88 mice, Related to Figure 1. (A-E) T cell subsets within the spleen were measured by flow cytometry. Representative plots were 46 + + + + + initially gated on CD3 CD4 cells; (A) TH1 cells are defined by CD3 CD4 IFNγ , and TH17 cells + + +- are defined by CD3 CD4 IL-17 ; Data are compiled for (B) TH17 frequencies, (C) TH1 + + + frequencies (n=7 for each group). (D) TREG cells are defined by CD3 CD4 FoxP3 ; data are compiled for (E) TREG frequencies (n=7 for each group). (F-J) T cell subsets within the MLNs + + were measured by flow cytometry. Representative plots were initially gated on CD3 CD4 cells; + + + + + + (F) TH1 cells are defined by CD3 CD4 IFNγ and TH17 cells are defined by CD3 CD4 IL-17 ; Data are compiled for (G) TH17 frequencies, (H) TH1 frequencies (n=11 for each group). (I) TREG + + + cells are defined by CD3 CD4 FoxP3 ; data are compiled for (J) TREG frequencies (n=8 for each group). P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using an unpaired, two-tailed t test. Related to Figure 1 Fig. S3. Analysis of TH1, TH17, and TREG population frequencies within cLP, siLP, and PPs -/- of WT and T-MyD88 mice, Related to Figure 1. (A-E) T cell subsets within the cLP were + + measured by flow cytometry. Representative plots were initially gated on CD3 CD4 cells; (A) + + + + + + TH1 cells are defined by CD3 CD4 IFNγ and TH17 cells are defined by CD3 CD4 IL-17 ; Data -/- are compiled for B TH17 frequencies, c, TH1 frequencies (n=9 WT and n=14 T-MyD88 ). (D) TREG + + + cells are defined by CD3 CD4 FoxP3 ; data are compiled for (E) TREG frequencies (n=10 for each group). f-j, T cell subset within the siLP were measured by flow cytometry. Representative + + + + + plots were initially gated on CD3 CD4 cells; (F) TH1 cells are defined by CD3 CD4 IFNγ and + + + TH17 cells are defined by CD3 CD4 IL-17 ; Data are compiled for (G) TH17 frequencies, (H) TH1 -/- + + + frequencies (n=9 WT and n=7 T-MyD88 ). i, TREG cells are defined by CD3 CD4 FoxP3 ; data -/- are compiled for j, TREG frequencies (n= 10 WT and n=8 T-MyD88 ). (K-O) T cell subset within + PPs were measured by flow cytometry. Representative plots were initially gated on CD3 CD4 + + + + + cells; (K) TH1 cells are defined by CD3 CD4 IFNγ and TH1 cells are defined by CD3 CD4 IL- + + 17 ; Data are compiled for (L) TH17 frequencies, (M) TH1 frequencies (n=10-11). (N) TREG cells + + + are defined by CD3 CD4 FoxP3 ; data are compiled for o, TREG frequencies (n=10-11). Pvalue<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using an two-tailed, unpaired t test. Related to Figure 1 Fig. S4. Analysis of PP cellularity and TFH populations within PPs, spleen, and MLNs, Related to Figure 1. (A-F) TFH populations within PPs were measured by flow cytometry. + + - Representative plots were initially gated on CD3 CD4 B220 cells (A) TFH populations were + + - + + defined as CD3 CD4 B220 PD-1 ICOS . Data are compiled for (B) TFH frequencies (n=15 WT -/- + + - + + and n=13 T-MyD88 ). (C)TFH populations were defined as CD3 CD4 B220 PD-1 CXCR5 . Data -/- are compiled for d, TFH frequencies (n=21 WT and n=18 T-MyD88 ). (E) TFH populations were + + - + + defined as CD3 CD4 B220 ICOS CXCR5 . Data are compiled for f, TFH frequencies (n=15 WT -/- and n=13 T-MyD88 ). (G,H) Gating strategy for non-GC TFH cells as defined by 47 int int. CD3+CD4+B220-CXCR5 PD-1 representative facs plot and h, data compiled (n=15 WT and -/n=13 T-MyD88 ). i, Data are compiled from the quantification of Trypan Blue (HyClone) resistant cells from single cell suspensions of PPs. (J,K) TFH populations within MLNs were + + - measured by flow cytometry. Representative plots were initially gated on CD3 CD4 B220 cells + + + + (G) TFH populations were defined as CD3 CD4 B220 ICOS CXCR5 . Data are compiled for (H) TFH frequencies (n=7 for each group). (L,M) TFH populations within the spleen were measured + + - by flow cytometry. Representative plots were initially gated on CD3 CD4 B220 cells l, TFH + + - + + populations were defined as CD3 CD4 B220 ICOS CXCR5 . Data are compiled for m, TFH frequencies (n=7 for each group). P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using an unpaired, two-tailed t test. Related to Figure 1. Fig. S5. Most significant taxonomic differences in mucosa, IgA bound and fecal -/communities between WT and T-MyD88 . Related to Figure 4 and 5. (A) Genera mean abundances (± SD) detected in IgA bound, mucosal and fecal fractions of intestinal microbial communities. Only genera detected as > 0.1% average relative abundance within a sample type group are shown. UC, Unclassified. A large number of OTUs can only be classified to the order level within the Clostridiales. (B) Boxplots showing the abundance within the mucosal fraction of four taxonomic groups containing known mucolytic bacteria are shown. Despite the trend towards increased mucolytic bacteria in the mucosal fraction of T-MyD88 animals, only the genus Ruminococcus attains significance due to large variability between animals. (* Pvalue<0.05, *** P-value<0.001. Welch’s t-test). Related to Figure 4 and 5. Table S1. Table of P-values for microbial community dissimilarities, related to Figures 4 and 5. Values are based on PERMANOVA with 9999 permutations shows that most observed patterns are robust to weighted, unweighted UniFrac as well as the non-phylogenetic BrayCurtis. Related to Figure 4 and 5 Table S2. 9 OTUs identified as significantly different in separately housed animals as well as cohoused animals, related to Figure 5D. P-values (non-parametric t-test with 9,999 Monte Carlo simulations) are shown for differences between genotypes in cohoused animals, illustrating the conservation of patterns among some OTUs between different housing conditions, as well as different direction of shifts in abundance of OTUs identified as belonging to the Lachnospiraceae genus. Related to Figure 4 and 5. ' CHAPTER 4 DIRECT RECOGNITION OF TLR2 LIGANDS INFLUENCES TFH DIFFERENTIATION AND GERMINAL CENTER DEVELOPMENT 49 Introduction A key component of the host immune response is T cell-dependent B cell activation that results in antibody responses. These interactions occur within lymphoid follicles and result in rapid B cell mutation and cellular turnover whose end product is a GC B cell capable of producing high affinity, antigen-specific immunoglobulins (1). Of the many subsets of T helper cells, TFH cells are tasked with driving and maintaining germinal center responses. While this immune response is particularly important during a pathogenic invasion, it occurs at steady-state conditions within gut-associated lymphoid tissue (GALT) in order to direct sIgA targeting towards colonizing commensal microbes (2). The decision point for a naïve T cell to differentiate into a TFH cell occurs upon its initial interaction with an antigen presenting cell (APC), through expression and ligation of Inducible T cell Costimulator (ICOS) (1, 3). Costimulation of ICOS supports the expression of BCL6, a master transcription factor required for the TFH lineage (4-6). Indeed, ICOS deficient mice lack TFH cells, and blocking the interaction between ICOS and its ligand (ICOSL) leads to a reduction of TFH populations (3). Commitment to the TFH cell fate requires a concerted network of cellular signaling that result in the secretion and colocalization of surface proteins required to form cell-cell interactions with B cells. These include CXCR5, a chemokine receptor found on both T cells and B cells that promotes co-localization at the follicle; PD1, CD40L, and SLAM, which interact and ligate with B cell surface proteins in order to promote activation and class switching within GC B cells; and cytokines IL-6 and IL-21, whose 50 extracellular secretion provides necessary autocrine activation to TFH cells as well as maintenance of GC B cell responses (1). The resulting T cell- B cell interactions lead to downstream cell signaling activation of the PI3K/AKT/mTOR pathway that provide metabolic shifts necessary to support TFH cell populations and GC response. Seminal studies have provided careful mechanistic insight into the antigen-specific interactions that occur during pathogen infections within the systemic compartment. However, how TFH cell responses were regulated within mucosal sites to commensal antigens has remained ill-defined. While immunoglobulin G (IgG) is the major antibody produced within the systemic compartment, GC responses within the GALT primarily result in the secretion of highly specific IgA (2). T cells and other host immune cells experience a constant inundation of microbial cues that shift in relationship with changes within the microbiota. While T cells are classically considered to be dependent on APC activation and costimulation before committing to a T helper cell fate, whether or not T cells could directly recognize microbial cues and dynamically respond to the microbiota had been previously unknown. One such way this might be occurring is through innate signals recognized by Toll-like receptors (TLRs) that signal through a common signaling adaptor, MyD88 (7). Indeed, we have recently uncovered a mechanism by which MyD88 dependent signaling in T cells is required for germinal center development within the gut and IgA secretion into the mucosa to provide commensal specific targeting of the microbiota (8). Mice lacking T cell 51 specific MyD88 signaling experience less IgA secretion and defective targeting of the microbiota that results in dysbiosis and susceptibility to colitis. This study provided a novel pathway by which innate signals could alter TFH populations and germinal center responses towards the microbiota. In our current study, we aimed to identify a mechanism by which MyD88 signaling could influence TFH differentiation and interrogate whether T cells could directly respond to microbial cues independent of APC signals. We demonstrate that T cells express TLR1, TLR2, and TLR6, which respond to TLR2 ligands, including Pam3CSK, in order to promote the expression of ICOS. In agreement with this, we uncovered defects in ICOS expression within T-MyD88/- T cells in peyer’s patches (PPs) throughout the gut as well as reduced proliferation. However, further experimentation revealed the likelihood that multiple pathways rely on MyD88 signaling to promote TFH differentiation and germinal center responses. Finally, we uncovered a potential mechanism by which commensals may utilize TLR2 signaling to instruct extra-intestinal T cells to prime antigen-specific responses within the systemic compartment. Results ICOS expression in gut T cells is deficient in the absence of MyD88 signaling MyD88 is a common signaling adaptor that propagates TLR ligand and cytokine detection to downstream transcriptional and metabolic responses. TFH populations are heavily dependent on concerted transcriptional and metabolic 52 pathways that influence differentiation, function, proliferation, and survival. We previously demonstrated that mice lacking MyD88 signaling within T cells were defective in both frequency and number of TFH populations within the gut. We sought to understand which pathways downstream of MyD88 were responsible for promoting TFH differentiation and maintenance. Using flow cytometry, we identified that PPs within T-MyD88-/- mice contain significantly fewer ICOS+ T cells (Figure 4.1a). Expression of ICOS was also reduced in ICOS+ T cells within T-MyD88-/- PPs, suggesting that, even when the protein is expressed, MyD88 is required to maintain proper expression levels within T cells (Figure 4.1b). Expression and ligation of the ICOS receptor is required at numerous stages of TFH differentiation. This initially begins when non-TFH cells interacting with dendritic cells, but continues through migration into the follicle, and maintenance of the GC-TFH population through B cell interactions, making ICOS induction a key event in coordinating TFH biology. We compared ICOS expression throughout TFH differentiation by gating on nonTFH cells, cells that had differentiated to TFH phenotype but were not residing within the germinal center, and germinal center TFH cells (Figure 4.2a). As cells differentiate from non-TFH cells to GC-TFH cells, there was a stepwise decrease in the percentage of ICOS+ cells within T-MyD88-/- mice, as well as a decrease in ICOS expression on a per cell basis (Figure 4.2bc). Notably, this was specific for ICOS. We compared the expression of CXCR5, PD1, and CD40L within TFH cells of WT and T-MyD88-/- mice and found no significant differences in protein expression, suggesting that once these proteins are made MyD88 is not required 53 B CD4 31.6% ICOS 23.5% 60 ICOS+ T cells (%) T-MyD88 WT -/- 40 20 0 ** 3000 ICOS MFI A WT T-MyD88-/- * 2000 1000 0 WT T-MyD88-/- Figure 4.1 T-MyD88-/- PPs have less ICOS+ T cells. A-B) ICOS expression on T cells within PPs was measured by flow cytometry. A) Representative plots were previously gated on CD3+CD4+ cells to measure frequency of ICOS+ T cells. B) Mean fluorescence intensity (MFI) of ICOS was measured on CD3+CD4+ICOS+ cells. 54 A B T-MyD88-/- 80 60 10 40 5 20 non-GC-TFH 0 0 C WT T-MyD88-/4000 GC-TFH ns 200 100 0 WT T-MyD88-/- 2000 ns non-GCTFH GC-TFH 4000 1500 1000 500 0 non-TFH 5000 PD-1 (MFI) 300 φ 1000 ns 2500 CXCR5 (MFI) CD40L (MFI) 400 **** ** 2000 0 WT T-MyD88-/- WT T-MyD88-/- 3000 PD-1 D * 100 15 ICOS+ (%) non-TFH CXCR5 ns 20 ICOS (MFI) WT WT T-MyD88-/- 3000 2000 1000 0 WT T-MyD88-/- Figure 4.2 ICOS expression becomes increasingly defective on T-MyD88-/- T cells as they differentiate to TFH cells. ICOS expression on non-TFH cells, nonGC-TFH cells, and GC-TFH cells within PPs was measured by flow cytometry. A) Representative plots were previously gated on CD3+CD4+. Non-TFH cells were defined as CD3+CD4+CXCR5-PD1-. Non-GC-TFH cells were defined as CD3+CD4+CXCR5lowPD1low.GC-TFH cells were defined as CD3+CD4+CXCR5highPD1high. B) Frequency of ICOS+ non-TFH cells, non-GCTFH cells, and GC-TFH cells. C) MFI of ICOS on non-TFH cells, non-GC-TFH cells, and GC-TFH cells. D) MFI of CD40L, CXCR5, PD1, and ICOS on TFH cells, defined as CD3+CD4+CXCR5+PD1+. 55 for their proper expression (Figure 4.2d). Collectively, these data strongly suggest that MyD88 promotes expression of the ICOS receptor in order to promote TFH differentiation. Direct detection of TLR2 ligands by T cells promotes ICOS expression in gut T cells We previously created WT and T-MyD88-/- germfree mice and demonstrated that the presence of TLR2 ligands alone in drinking water is sufficient to promote TFH differentiation in a T cell intrinsic MyD88 dependent manner. In addition to having fewer TFH cells within germfree T-MyD88-/- mice that were fed TLR2 ligands, these animals also had significantly less ICOS expression on their surface as well (Figure 4.3). These data demonstrate that the presence of TLR2 ligands alone are able to promote ICOS expression and TFH differentiation, however, it was not clear whether T cells are directly capable of detecting those microbial cues themselves. MyD88 is a downstream adaptor for TLR ligands and host cytokines, such as IL1b and IL18. To understand whether T cells can directly recognize TLR2 ligands or are dependent on secondary host cytokines to differentiate and maintain TFH cell populations, we performed RNA-seq on T cells within Peyer’s patches and looked for the expression of TLRs in non-TFH cells and TFH cells (Figure 4.4a). We found that the receptors involved in the TLR2 pathways, TLR2, TLR1, and TLR6 were all expressed on both non- TFH and TFH cells. In contrast, other TLRs responsible for bacterial detection such as TLR4 and TLR5 were not 56 T-MyD88-/- % of Max 33.9% 10.5% 2700 ICOS (MFI) WT 1800 900 0 ICOS+ * WT T-MyD88-/GF+ TLR2 ligand Figure 4.3 TLR2 ligands alone promote ICOS expression in vivo in T cells with MyD88 signaling. ICOS expression on T cells within PPs was measured by flow cytometry. Representative plots were previously gated on CD3+CD4+ cells to measure MFI of ICOS on CD3+CD4+ cells. 57 Figure 4.4 T cells directly recognize TLR2 ligands to increase ICOS expression in vitro. A) RNA-seq expression in non-TFH cells and TFH cells within PPs. B-E) ICOS expression on sort purified naïve T cells, defined as CD3+CD4+CD62L+CD69-, cultured for 3 days in the presence of anti-CD3, with or without TLR or cytokine activation. B) Representative plots were previously gated on CD3+CD4+ to measure percent increase in ICOS MFI of naïve T cells cultured with anti-CD3 and Pam3CSK (P3CSK), Pam2CSK (P2CSK), Lypopolysaccharide (LPS), or Flagellin compared to naïve T cells cultured with anti-CD3 along. C) Representative plots were previously gated on CD3+CD4+ to measure ICOS MFI of naïve T cells cultured with anti-CD3 alone (Media) or in combination with Pam3CSK (P3CSK), Interleukin-18 (IL-18), or Interleukin-1beta (IL-1beta) for 1, 2, and 3 days. D) ICOS MFI was calculated on naïve T cells cultured with anti-CD3 alone (Media) or in combination with Pam3CSK (P3CSK), Interleukin-18 (IL-18), or Interleukin-1beta (IL-1beta) in the presence of DMSO or and NFkB inhibitor. E) Representative plots were previously gated on CD3+CD4+ to measure ICOS MFI of naïve WT or T-MyD88-/- T cells cultured with anti-CD3 alone (Media) or in combination with Pam3CSK (P3CSK). 58 A 40 20 10 Tlr5 % of Max LPS ICOS C IL-1β IL-18 P3CSK 2000 ** 60 ** 40 20 0 P3CSK IL-18 IL-1β Media 1500 * % of Max 1000 500 0 ICOS+ 1 2000 * 2700 DMSO NFkB inhibitor T-MyD88-/- WT * 1800 1000 500 Media P3CSK IL-1β IL-18 % of Max 1500 0 ICOS (MFI) 2500 3 2 Days E D ICOS (MFI) % increase in ICOS MFI P3CSK LPS Flagellin Myd88 Tlr4 B P2CSK n.d. n.d. Tlr9 n.d. Tlr7 n.d. Tlr6 n.d. Tlr3 Tlr2 Tlr1 IL1r1 0 IL18r1 n.d. Tlr8 FPKM 30 33.9% ICOS+ 10.5% 900 0 WT T-MyD88-/GF+ TLR2 ligand 59 detected. To test a role for T cell intrinsic TLR signaling during the induction of ICOS, fluorescence activated cell sorting (FACS)-purified T cells were stimulated with various TLR ligands in vitro. In agreement with our sequencing data, TLR2 ligands, in concert with TCR stimulation, robustly induced ICOS, while TLR4 or TLR5 ligands did not (Figure 4.4b). Promotion of MyD88 pathway is specific to TLR2 and does not require IL1b or IL18, neither of which were able to increase ICOS expression over media during a three-day period (Figure 4.4c) Importantly, the upregulation of ICOS requires the downstream transcription factor, NFkB (Figure 4.4d). Inhibition of this transcription factor reduced ICOS expression to that of media alone. Furthermore, T cells lacking MyD88 were unable to upregulate ICOS in the presence of TLR2 ligands (Figure 4.4e). These data suggest that direct recognition of TLR2 ligands by T cells leads to MyD88 signaling to NFkB to upregulate ICOS expression and TFH differentiation. MyD88 signaling is required for systemic TFH responses to immunization Peyer’s patches are an important site for T cell dependent IgA targeting of microbial populations with the gut. TFH populations at this site along the gut are required to respond to shifts within microbial cues in order to shape the microbiota and maintain healthy interactions between the microbiota and the host immune system. We previously demonstrated that disruption of MyD88 signaling in T cells led to a reduction in T cell dependent IgA responses at steady-state conditions. We next sought to understand whether MyD88 signaling was required 60 for systemic TFH development as well. At steady-state conditions, there is a very low frequency of TFH populations within spleens of healthy animals, however, the frequency of these cells is increased during an immune challenge to provide antigen-specific direction to germinal center B cells, leading to the production of highly specific antibodies. To understand how MyD88 promotes TFH development during an antigen-specific immune response, we immunized animals with ovalbumin (OVA) in the presence of Pam3CSK to provide a TLR2 activated immune response. Compared to WT mice, T-MyD88-/- mice were defective in their ability to promote systemic TFH responses to ovalbumin (Figure 4.5a). Reduced TFH differentiation within T-MyD88-/- mice resulted in fewer GC B cells (Figure 4.5b). To identify the impact that these reductions in systemic germinal center responses had in antigen-specific antibodies, we performed ELISAs to measure both total IgG as well as OVA specific IgG. While we detected no defects in total IgG antibodies within the T-MyD88-/- mouse serum, these animals contain less OVA specific IgG within the serum (Figure 4.5c). This demonstrates that, while MyD88 is not necessary for steady-state responses within the systemic compartment, it is required to upregulate TFH responses during an immune response. MyD88 signaling promotes TFH differentiation through a cell intrinsic mechanism The T-MyD88-/- mouse is lacking MyD88 within all CD4+ and CD8+ T cells. While we have so far seen a specific defect within TFH cells, it was unclear 61 WT .155% Fas 2.13% GL-7 4 2 2 1 0 Total IgG (µg/mL) C TFH cells (#) 6 3 T-MyD88-/- GC B cells (%) B 8 0 PD-1 8×105 * p=.05 2×105 0 60 40 20 WT T-MyD88-/- WT T-MyD88-/- * 2×104 1×104 0 WT T-MyD88-/- 80 0 4×105 3×104 100 p=.08 6×105 WT T-MyD88-/- GC B cells (#) CXCR5 8.2% 10 T-MyD88-/3.57% TFH cells (%) WT OVA specific IgG (ng/mL) A WT T-MyD88-/- 15 10 5 0 WT T-MyD88-/- Figure 4.5 OVA immunization of T-MyD88-/- mice results in significant defects in TFH and GC responses. A) Representative plots were previously gated on CD3+CD4+ cells to measure frequency and total numbers of TFH cells, defined as CD3+CD4+CXCR5+PD1+ within spleens. B) Representative plots were previously gated on B220+IgDlow cells to measure frequency and total numbers of GC B cells, defined as B220+IgDlowFas+GL7+ within spleens. C) Enzymelinked Immunosorbant Assay (ELISA) was used to quantify Total IgG and OVA specific IgG within serum. 62 whether this is a cell intrinsic defect that occurs during TFH differentiation or was due to T cell-wide defects that result in an extracellular environment that reduces TFH differentiation and germinal center responses. We created mice that have OVA specific CD4+ T cells (OTII) that either contained or lacked MyD88. We then transferred these cells into WT B6 mice and immunized them with OVA along with Pam3CSK. Despite the presence of WT endogenous T cells, OTII cells lacking MyD88 were less able to differentiate to TFH cells than WT OTII cells, arguing that it is disruption of a cell intrinsic pathway required for TFH differentiation that reduces the presence of TFH cells within T-MyD88-/- mice (Figure 4.6a). However, the GC response was the same in mice that had received either WT OTII cells or T-MyD88-/- OTII cells, indicating that there is likely a second, extracellular component that MyD88 signaling stimulates to produce an effective germinal center response (Figure 4.6b). Defects within splenic TFH responses are independent of ICOS signaling The TFH and germinal center defect observed within PPs is recapitulated in the systemic compartment during OVA immunization. We measured ICOS expression within spleens during the OVA immune response, however, and did not detect any significant difference between ICOS expression within WT T cells or T-MyD88-/- T cells (Figure 4.7a). This was true for both endogenous T cell experiments as well as OTII experiments (Figure 4.7b). Additionally, efforts to rescue ICOS expression within T-MyD88-/- T cells during OVA immunization in 63 TMYD OTII 10 4.11% CXCR5 GC-TFH cells (%) 8.79% PD1 6 4 2 B Fas GL7 20000 0 WT TMYD OTII OTII p=0.15 40000 ns TMYD OTII WT TMYD OTII OTII ns 1.5 80000 1.11% GC B cell (%) 1.5% 60000 8 0 WT OTII * GC-TFH cells (#) WT OTII GC B cell (%) A 1.0 0.5 0.0 WT TMYD OTII OTII 60000 40000 20000 0 WT TMYD OTII OTII Figure 4.6 T cell intrinsic MyD88 signaling promotes GC responses through a combination of cell intrinsic and extrinsic mechanisms. A) Representative plots were previously gated on CD3+CD4+ cells to measure frequency and total numbers of TFH cells, defined as CD3+CD4+CXCR5+PD1+ within spleens. B) Representative plots were previously gated on B220+IgDlow cells to measure frequency and total numbers of GC B cells, defined as B220+IgDlowFas+GL7+ within spleens 64 A TMYD WT 14.8% CD4 ICOS+ (%) 16.2% 40 TMYD OTII CD4 ICOS % of Max 40 20 0 293T cell MIGR1-ICOS D ICOS MFI 293T cell MIGR1 TMYD 60 37.3% ICOS+ (%) 29.5% ICOS 10 WT WT OTII C 20 0 ICOS B 30 WT TMYD OTII OTII 300 200 100 0 WT TMYD TMYD MIGR MIGR ICOS Figure 4.7 MyD88 promotes TFH differentiation in an ICOS dependent manner. A) Representative plots were previously gated on CD3+CD4+ cells to measure frequency and total numbers of TFH cells, defined as CD3+CD4+CXCR5+PD1+ within spleens of WT or T-TLR2-/- mice. B) Representative plots were previously gated on CD3+CD4+ cells to measure frequency and total numbers of TFH cells, defined as CD3+CD4+CXCR5+PD1+ within spleens of mice given WT or TTLR2-/- OTII T cells. C) Representative plots were previously gated on GFP+ MIGR plasmid transfected 293T cells to measure ICOS+ 293T cells. B) ICOS MFI was calculated on CD3+CD4+ T cells transfected with the MIGR-control or MIGR-ICOS plasmid. 65 order to promote TFH differentiation were unsuccessful. We created a plasmid capable of overexpressing ICOS within transfected 293T cells, however, T cells transfected with this plasmid showed no increased expression of ICOS, preventing us from testing whether ICOS overexpression could rescue the defects seen in TFH differentiation (Figure 4.7c,d). Regardless, the fact that splenic T-MyD88-/- T cells show no defect in ICOS expression, but experience reduced TFH populations, suggests that MyD88 might influence multiple pathways important for TFH populations. MyD88 signaling promotes TFH proliferation to maintain TFH subsets We expanded our search to look at defects within cell death or proliferation that might contribute to the reductions in TFH populations. There is no difference in cell death of TFH cells within PPs between WT and TMyD88-/- mice, as measured by cells that are annexin-V positive and 7-AAD positive (data not shown). To interrogate the requirement of MyD88 for T cell proliferation, we analyzed the percentage of Ki-67+ non-TFH and TFH cells within peyer’s patches of WT and T-MyD88-/- mice. Ki-67 is absent in resting cells, but becomes expressed once cells have entered into the cell cycle. While MyD88 does not influence proliferation of Ki-67 in non-TFH cells (Figure 4.8a), T-MyD88-/- mice had significantly fewer Ki-67+ TFH cells within their PPs (Figure 4.8b), suggesting that defects in proliferation might contribute to the observed reductions of TFH cell populations within the guts of these mice. 66 A WT 33.4% T-MyD88-/28.6% TFH+ Ki-67+(%) CXCR5 30 PD-1 17.3% 25 20 15 10 WT T-MyD88-/- CD4 19.8% * B Ki-67 T-MyD88-/- WT Ki-67+(%) CXCR5 15 PD-1 10% 10 5 0 WT T-MyD88-/- CD4 11.3% ns Ki-67 Figure 4.8 MyD88 signaling promotes proliferation to maintain TFH populations. A) Representative plots were previously gated on CD3+CD4+CXCR5+PD1+ cells to measure frequency of Ki67+ TFH cells. B) Representative plots were previously gated on CD3+CD4+CXCR5-PD1- cells to measure frequency of Ki67+ non-TFH cells. 67 TLR2 signaling on T cells promotes GC responses within the systemic compartment Our data so far have suggested that T cells directly detect TLR2 ligands in order to alter TFH differentiation and function. In order to test this in vivo, we created a T cell specific knock out of TLR2 by crossing TLR2 fl/fl mice with mice containing the cre-recombinase driven by the CD4 promoter (Figure 4.9a). These mice have severe reductions in the presence of TLR2 on T cells, while maintaining it on non-T cell subsets (Figure 4.9b). At steady-state conditions, these mice exhibited no defects in TFH cell and GC B cell populations within the intestines (Figure 4.10a,). However, during OVA immunization, T-TLR2-/- mice experience significant defects in GC B cell responses, despite the fact that they have similar frequencies of TFH populations (Figure 4.10b). Importantly, this defect is not due to non-T cell specific defects within these mice. Naïve T-TLR2-/T cells transferred to TCRb-/- mice also demonstrate a significant defect in their ability to induce germinal center responses during OVA immunization (Figure 4.10c). While this does not completely recapitulate the TFH defects seen in TMyD88-/- mice, it demonstrates that direct recognition of TLR2 ligands by T cells is required for the proper function of TFH cells. OVA immunization of WT B6 mice containing WT or T-MyD88-/- OTII cells demonstrated that effects on GC responses are likely cell extrinsic. To determine whether this was the case for defects seen in T-TLR2-/- mice, we crossed these animals to OTII mice and injected CD4+ OTII cells in WT B6 mice before immunizing them with OVA. In agreement with our previous findings, OVA immunized B6 mice given T-TLR2-/- 68 A TLR2fl/fl TLR2 fl allele WT allele CD4-Cre B T-TLR2-/- TLR2-/- 300 TLR2 MFI % of Max TLR2fl/fl 100 0 TLR2 C T-TLR2-/- 2500 TLR2-/TLR2 MFI TLR2fl/fl % of Max 200 2000 1500 1000 500 0 TLR2 Figure 4.9 Specific deletion of TLR2 within T cells was created by crossing a TLR2fl/fl mouse to a CD4-cre+ mouse. A) The TLR2 fl/fl mouse contains loxp sites surrounding Exon 3 of TLR2. This was crossed to mice containing crerecombinase driven by the CD4 promoter. Genotyping identified successful offspring that contain TLR2fl/fl allele and the presence of CD4-cre. B-C) TLR2 expression was measured on T cells, defined as CD3+CD4+, and non-T cells, defined as CD3-, using flow cytometry. B) Representative plots were previously gated on CD3+CD4+ cells to measure MFI of TLR2. C) Representative plots were previously gated on CD3- cells to measure MFI of TLR2. 69 Figure 4.10 TLR2 signaling on T cells promotes GC responses within the systemic compartment. A) Flow cytometry was used to measure GC-TFH cells and GC B cells within PPs. B) Flow cytometry was used to measure TFH cells and GC B cells within spleens following OVA immunization. C) Flow cytometry was used to measure TFH cells and GC B cells within spleens of TCRb-/- mice given WT or T-TLR2-/- T cells following OVA immunization. D) Flow cytometry was used to measure TFH cells and GC B cells within spleens of WT B6 mice given WT OTII or T-TLR2-/- OTII T cells following OVA immunization. 20000 10 5 0 15000 10000 5000 0 WT T-TLR2-/- WT T-TLR2-/- 50000 40 40000 30 20 10 0 4 2×105 2 WT T-TLR2-/- C ns 25 WT T-TLR2-/- ns 1.5×105 15 TFH (#) TFH (%) 0 2.0×105 20 10 5 0 1×105 WT T-TLR2-/TCRb-/- 1.0×105 5.0×104 0.0 WT T-TLR2-/TCRb-/- 10 5 10000 0 WT T-TLR2-/- 2×104 1×104 0 p=.06 1 p=.10 3×104 WT T-TLR2-/- 2 0 20000 p=.05 15 0 30000 WT T-TLR2-/- 20 3 GC B cells (%) 0 25 GC B cells (%) 3×105 TFH (#) TFH (%) B 6 GC B cells (#) 15 50 GC B cells (#) 25000 8×104 GC B cells (#) 20 TFH (#) TFH (%) A GC B cells (%) 70 p=0.13 6×104 4×104 2×104 0 WT T-TLR2-/- WT T-TLR2-/- TCRb-/- WT T-TLR2-/TCRb-/- D 400000 20 0 300000 200000 100000 WT OTII T-TLR2-/OTII 0 WT OTII T-TLR2-/OTII GC B cells (#) TFH (#) TFH (%) 40 50000 2.0 GC B cells (%) 60 1.5 1.0 0.5 0.0 WT OTII 40000 30000 20000 10000 T-TLR2-/OTII 0 WT OTII T-TLR2-/OTII 71 OTII had no defects in GC formation, indicating that TLR2 ligands and MyD88 signaling promote GC responses through cell extrinsic signals (Figure 4.10d) TLR2 signaling promotes commensal specific immune responses in extra-intestinal sites In order to understand whether TLR2 signaling on T cells might influence TFH responses to commensal antigens, we created germfree WT and T-TLR2-/mice by reconstituting germfree Rag-/- mice with bone marrow from WT or TTLR2-/- mice. During reconstitution, we monocolonized these mice with the commensal, Bacteroides fragilis, that has been designed to express OVA. WT and T-TLR2-/- mice were monocolonized to similar levels as measured by the detection of OVA reads within fecal DNA (Figure 4.11a). Colonization with this single species expressing OVA, however, revealed a defect in the ability of T cells lacking TLR2 to produce an antigen-specific immune response. We did not detect any differences in activated T cells trafficking along the lamina propria between monocolonized WT or T-TLR2-/- mice (Figure 4.11b). In contrast to this, we detected dramatic defects within commensal antigen-specific responses within the spleen. Spleens of monocolonized T-TLR2-/- mice had significantly fewer activated T cells and TFH cells (Figure 4.11c,d). Furthermore, restimulation of splenocytes with OVA resulted in significantly less IFNg production from TTLR2-/- spleens (Figure 4.11,e). While these data do not provide a mechanism by which TLR2 might be influencing antigen-specific immune responses, it provides consistent argument that direct detection of TLR2 ligands influences 72 Figure 4.11 TLR2 signaling promotes commensal specific immune responses in extra-intestinal sites. A) quantitative PCR (qPCR) measured total presence of the OVA gene within fecal DNA of monocolonized WT and T-TLR2-/- mice. B) Representative plots were previously gated on CD3+CD4+ cells to measure frequencies activated T cells, defined as CD3+CD4+CD69+CD44+, within the lamina propria of the gut. C) Representative plots were previously gated on CD3+CD4+ cells to measure frequencies and total numbers of activated T cells, defined as CD3+CD4+CD69+CD62L-, within spleens. D) Representative plots were previously gated on CD3+CD4+ cells to measure frequencies and total numbers of TFH cells defined as CD3+CD4+CXCR5+PD1+ within spleens. E) ELISA was used to measure IFNg produced from splenic cells that were cultured with or without OVA for 72 hr. 73 B 62.4% DT-TLR2 -/- 10.5% 62.8% 23.3% 20.6% 22.1% CD69 6.01% 10.6% T-TLR2-/- WT 17.4% 0 15 10 0 500 400 300 200 100 20 WT T-TLR2-/- 10 0 OVA - + ns 80 60 40 20 0 WT T-TLR2-/- 2.5×105 p=.12 2.0×105 1.5×105 1.0×105 5.0×104 WT T-TLR2-/** 0.0 2.0×105 20 TFH (%) CXCR5 10 WT T-TLR2-/p=.12 1.5×105 5 ** IFNγ (pg/mL) ** 20 25 9.98% PD-1 E 30 Activated T cell (#) CD62L 44.1% CD44 TFH (#) WT T-TLR2-/WT D 68.9% Activated T cell (%) C T-TLR2-/- WT CD69 OVA (per 50ng DNA) 109 108 107 106 105 104 103 102 101 100 Activated T cells (%) A WT T-TLR2-/- 1.0×105 5.0×104 0.0 WT T-TLR2-/- 74 antigen-specific immune responses on T cells. Additionally, it reveals an interesting pathway by which antigen-specific responses of systemic T cells are influenced by commensal antigens within the gut. Discussion Collectively, these data demonstrate that T cells respond directly to microbial cues via TLR2 to promote TFH differentiation in an MyD88 dependent manner. MyD88 is a common signaling adaptor for both host cytokines and multiple TLRs that recognize microbial cues. MyD88 activation leads to downstream transcriptional and metabolic changes within the cell. T cells express TLR1, TLR2, and TLR6 as well as IL1b and IL18 receptors, however, only stimulation of TLR1/TLR2 or TLR6/2 is capable of inducing ICOS expression and TFH differentiation. T cells lacking MyD88 signaling maintain steady-state defects within PPs along the gut, resulting in reduced IgA targeting of the microbiota and a greater susceptibility to colitis. MyD88 is also required for antibody production via TFH differentiation and germinal center responses. Interestingly, while our findings within the gut suggest that ICOS expression is defective when MyD88 is not present within T cells, the results from our systemic immunizations imply that other factors are required for TFH differentiation. This could mean that ICOS is more strongly upregulated within TFH differentiation in PPs. Indeed, multiple subtypes of TFH cells have been identified that each maintain a unique ability to direct various antibody responses (9, 10). It is possible that the TFH cells within PPs depend more strongly on 75 ICOS than splenic TFH cells. Comparison of ICOS expression between splenic TFH cells and PP TFH cells supports such a possibility. It is also feasible that ICOS is acting concomitantly with another pathway during TFH differentiation. Recent studies have revealed that, due to the increased metabolic burden on TFH cells, ICOS signaling and the mTOR pathway cooperate to promote TFH differentiation and germinal center responses (11, 12). Mice lacking T cell intrinsic mTOR1 and mTOR2 signaling have fewer TFH cells within PPs, and are less able to promote BCL6 activation during ICOS ligation in vitro. During differentiation of other T cell subsets, MyD88 is required for proper mTOR activation, making this a suggestive candidate for an alternative mechanism necessary to maintain TFH populations in an MyD88 dependent manner (13). Indeed, in the absence of ICOS defects, we identified a cell intrinsic defect in TFH differentiation within the spleen. Furthermore, TFH cells within PPs are also defective in their proliferative capacity, suggesting that differentiation alone from ICOS signaling is not the only defective pathway in T-MyD88-/- germinal center responses. Future investigation into the metabolic capacity of T-MyD88-/- mice is necessary to characterize its contribution to antigen-specific germinal center responses. We have demonstrated that T cell recognition of TLR2 ligands are capable of promoting TFH differentiation and ICOS expression in vitro and in vivo. While mice specifically lacking TLR2 contain similar levels of TFH populations as WT within their PPs, OVA immunization experiments revealed an inability to promote GC B cell responses within the spleen. Additionally, TLR2 signaling on T cells 76 promotes commensal specific immune responses within the systemic compartment. Experiments such as these may reveal a pathway that is one part of a greater network. While no defects are revealed in steady-state conditions, the inundation of microbial signals might be enough to compensate for the loss of TLR2 on T cells. It is also possible that measurements of frequency and total numbers alone of germinal center populations might not be sensitive enough to identify defects within IgA targeting of microbes when T cells lack TLR2. Despite seeing no defects within PPs under steady-state conditions, we consistently observe fewer germinal center B cells in immunized T-TLR2-/- mice. These data, along with the identification that commensal antigens can drive systemic T cell responses in a TLR2 dependent manner reveal an intriguing method by which microbiota colonization can influence systemic immune responses to infections or vaccines. Recent studies have revealed antibody cross- reactivity between commensal antigens and the envelope protein gp41 found on HIV (14, 15). These results lead to speculation that some individuals may be colonized with a microbiota that could provide greater protection by an HIV or other vaccine response. Additionally, introduction of specific commensals concomitant with immunizations may have the potential to boost vaccine efficacy. Whether or not commensal microbes are acting as an adjuvant through other T cell expressed microbial pattern receptors or just through TLR2 remains unknown, however, TLR2 signaling on T cells may be an ideal target as a means to enhance TFH responses during vaccines to promote stronger germinal center and antigen-specific antibody responses. Greater delineation of these signaling 77 pathways is required to fully understand and harness T cell responses to the microbiota in order to promote healthy symbiotic interactions and protect against disease. Experimental Procedures Mice MyD88 LoxP/LoxP mice were purchased from Jackson Laboratories and subsequently crossed to CD4-Cre animals (Taconic) to produce homozygote offspring. MyD88WT/WT CD4-Cre+ mice are designated as WT and MyD88LoxP/LoxPCD4-Cre+ are designated as T-MyD88-/-. TLR2loxp/loxp were created using EUCOMM. Mice were crossed to CD4-cre+ mice to create TLR2loxp/loxp designated as WT and TLR2loxp/loxpCD4-Cre+ mice designated as T-TLR2-/- mice. Age matched animals were used to compare phenotypic differences between genotypes. TCRb-/- mice were purchased from Jackson and maintained as a colony within out mouse facility. MyD88LoxP/LoxPCD4Cre+ and TLR2loxp/loxpCD4-Cre+ were crossed to OT-II mice (Jackson) in order to obtain mice that were CD4-Cre+ OT-II mice designated as WT OTII mice, MyD88LoxP/LoxPCD4-Cre+ OT-II mice designated as T-MyD88-/- OTII mice, and TLR2loxp/loxpCD4-Cre+ OT-II mice designated as T-TLR2-/- OTII mice. Germ-free mice were maintained in sterile isolators and assayed monthly for germ-free status by plating and PCR. Age matched germfree Balb/c mice were compared to SPF Balb/c mice to compare phenotypic differences in TFH cells along gut associated lymphoid tissues. For the TLR-ligand experiment 78 Germ-free C57Bl/6 mice were born germ-free and then fed drinking water containing 10µg/mL of Pam3CSK4 (Invivogen) for two weeks. For the reconstituted mono-association and TLR ligand experiments, germfree Rag1-/mice were reconstituted with bone marrow (~2.5x106 cells) from WT and TMyD88-/- mice. For the mono-association study, reconstituted germfree Rag1-/mice were mono-associated with erythromycin/gentamicin resistant Bacteroides fragilis that was engineered to express OVA (B.fragilis-OVA: a gift from Sarkis Mazmanian (California Institute of Technology) via oral gavage. Animals were maintained for two months with 1milligram per milliliter of both erythromycin and gentamicin in drinking water in SPF animals housing conditions (not within germfree isolators) and were validated to be correctly colonized by fecal aerobic & anaerobic plating of feces as well as PCR analysis. The reconstituted Rag1-/mice were analyzed two-months postcolonization. For the TLR-ligand experiment, reconstituted Rag1-/- mice were maintained on 1milligram per milliliter of ampicillin (Fisher Scientific), neomycin (Fisher Scientific), erythromycin (Fisher Scientific), and gentamicin (GoldBio) for two months. During the last two weeks, 10µg/mL of Pam3CSK4 (Invivogen) was added to antibiotic cocktails. For ovalbumin immunization, mice were subcutaneously immunized with150µL of 100µg of OVA and 20µg of Pam3CSK mixed 1:1 in incomplete Freunds adjuvant (IFA). Animals were sacrificed 7-10 days later. For OTII and TCRb-/- experiments, one day before OVA immunization, 2.5X106 MACS CD4+ enriched cells or FACS purified Naïve T cells were retro-orbitally injected into WT B6 or TCRb-/- mice, respectively. 79 Enzyme-linked immunosorbant assay (ELISA) For quantification of serum IgG, blood was collected into 1.5mL eppendorf tubes and spun at 4°C for 15 min at 12,000g. Serum supernatant was then transferred to a new Eppendorf tube, flash frozen and stored at 70°C. 1:10 dilutions of serum samples were used to detect IgG and OVA specific IgG using the Mouse IgG total Ready-set-go ELISA kit (eBioscience), performed as per kit instructions. Absorbance was read at 450nm, and concentrations of IgG were calculated using the provided standard. For quantification of OVA specific IgG, IgG quantification was performed using the same IgG specific ELISA kit, with slight modification. Instead of coating plates with capture antibody specific for IgG, plates were coated overnight at 4°C using 4µg/mL ovalbumin in PBS. For in vitro supernatant detection of IFNg and IL-17, 1X105 splenic cells were incubated with or without 1µg/mL of OVA for 3 days. Supernatent was diluted 1:1000 for IFNg and 1:10 for IL17 and analyzed using Mouse IFNg ELISA MAX (Biolegend) and Mouse IL17A Ready-set-go (eBioscience), performed as per kit instructions. Absorbance was read at 450nm, and concentrations of were calculated using the provided standard. T cell isolation Lymphocytes were isolated from spleens as mentioned above and the resulting cells were sorted through MACS columns using negative selection (CD4+ T Cell Isolation Kit II (Miltenyi)). Sorts were performed as per kit instructions. Isolated T cells were further purified via FACS using a BD FACSAria 80 Cell Sorter For Naïve T cells, CD3+CD4+CD62L+CD25- cells were collected into RPMI media (Corning) supplemented with 10%FBS (v./v.) (Gibco BRL), 50 U/mL penicillin and 50 µg/mL streptomycin (Cellgro), 5µM 2-Mercaptoethanol (CalBiochem), 1µM sodium pyruvate (Cellgro), 1x MEM Nonessential Amino Acids (Cellgro) and 2.05mM L-glutamine (Cellgro). For TFH purified cells CD3+CD4+B220-CXCR5+PD1+ cells were collected into supplemented RPMI media. For TFH-depleted T cells, CD3+CD4+B220-CXCR5-PD1- cells were collected into supplemented RPMI media. Quantification of OVA gene within fecal DNA Fecal pellets were collected from monocolonized mice in 2mL screw cap tubes containing ~250 milligram of 0.15 mm garnet beads (MoBio) for DNA extraction. DNA was extracted using Power Fecal Kit (MoBio) per kit instructions. Concentrations of fecal DNA was quantified using Nanodrop, and 50ng of fecal dna was used to quantify the levels of OVA gene using LightCycler ® 480 SYBR Green I Master (Roche) and a Lightcycler 480. OVA forward primer: AGAAATGTCCTTCAGCCAAGCTC. reverse OVA primer: GCCCATAGCCATTAAGACAGATGTG Retroviral overexpression Full length ICOS was inserted into the MIGR1 plasmid (Addgene) for overexpression of ICOS. ICOS was amplified and the ICOS PCR product and MIGR1 plasmid were digested using restriction enzymes EcoRI and Bgl2 (New 81 England Biolabs) and ligated using T4 DNA Ligase (New England Biolabs). Plasmid was introduced into Mix & Go Competent E. coli (Zymo Research) and plated on LB + Ampicillin overnight. 7μg of MIGR1-EICOS or 7 μg of control MIGR1 was mixed with 7 μg of pCL-Eco in 1.5 mL serum-free DMEM containing 35 μL TransIT-293 Transfection Reagent (Mirus Bio). After 15 min of incubation, this mixture was added to 5 mL of DMEM in a tissue culture flask containing ~5,000,000 293T cells. After 3 days of incubation, supernatant from cells was collected and passed through a .45 μm syringe tip. 150-200 μL of retroviral containing media was then added to 96 well plates containing 200 μL of RPMI containing CD4+ splenic cells isolated from mice using CD4+ T cell Isolation Kit (Miltenyi) with anti-CD3 (5 μg/mL) and anti-CD28 (1 μg/mL) (eBioscience) and spun at 2500 RPM from 90 min at 37°C. 100 μL of RPMI was removed and replaced with complete RPMI containing anti-CD28 and plates were incubated for 3 days at 37°C. After incubation, cells were sort purified for GFP+ CD3+ using FACSAria. Confirmation of ICOS over expression was verified in 293T cells and primary T cells through flow cytometry analysis of ICOS on the cell surface. In vitro activation experiments Purified Cells were plated in 250µL of supplemented RPMI containing 20ng/mL IL-2 in 96 well plates (2X105 cells per plate) that had been previously coated with 5µg/mL purified α-CD3. For TFH skewing conditions, 1µg/mL of the co-stimulatory antibody α-CD28, 10µg/mL α-CD4, 10µg/mL α-IFNγ, 50ng/mL IL21, and 50ng/mL IL-6 were added. For TLR ligand stimulation, 1µg/mL of either 82 Pam2CSK (Invivogen), Pam3CSK (Invivogen), Flagellin from S. typhimurium (Invivogen) or LPS was added. For cytokine stimulation, 50ng/mL of IL1b or IL18 were added. For NFKB inhibition, 100ng/mL of BAY 11-7082 dissolved in DMSO was added and equal volume of DMSO was added in control wells. Cell were cultured for 3 days at 37°C in these conditions before being analyzed by flow cytometry. Flow cytometry staining of isolated lymphocytes Lymphocytes were isolated from PPs, spleens, and inguinal lymphnodes by gently pushing tissue through a 40µM filter to obtain white blood cells. Splenic cells were additionally treated with 1x RBC Lysis Buffer (Biolegend) to remove red blood cells. Surface staining for lymphocytes was done in HBSS (Corning) supplemented with 10mM Hepes (Cellgro), 2mM EDTA (Cellgro), and 0.5% FBS (Gibco BRL) using α-CD4 (Biolegend), α-CD3 (Biolegend), α-B220 (Biolegend), α-CXCR5 (eBioscience), α-PD-1 (Biolegend), α-ICOS (eBioscience), α-GL-7 (eBioscience), α-Fas (BD Biosciences), α-IgD (Biolegend), for 20 min at 4°C. Intra-Cellular Staining For α-ki-67 (eBioscience) intracellular staining, cells were first surface stained and then permeabilized and fixed in 100ml of Perm/Fix buffer (eBiosciences) at 4°C overnight. Cells were washed twice in Perm/Wash buffer (eBioscience). Gating and analysis of positive cell populations was done utilizing respective isotype controls for antibodies against ICOS, CXCR5, GL-7 and PD-1. Other cell populations were identified using single stain controls. Data were collected using a BD LSR Fortessa and analyzed using FLOWJO software. 83 References 1. H. Qi, T follicular helper cells in space-time. Nature Rev. Immunol. 16, 612-625 (2016). 2. O. Pabst, New concepts in the generation and functions of IgA. Nature Rev. Immunol. 12, 821-832 (2012). 3. Y. S. Choi et al., ICOS receptor instructs T follicular helper cell versus effector cell differentiation via induction of the transcriptional repressor Bcl6. Immunity 34, 932-946 (2011). 4. A. L. Dent, A. L. Shaffer, X. Yu, D. Allman, L. M. Staudt, control of inflammation, cytokine expression, and germinal center formation by BCL-6. Science 276, 589-592 (1997). 5. R. J. Johnston et al., Bcl6 and Blimp-1 are reciprocal and antagonistic regulators of t follicular helper cell differentiation. Science 325, 10061001 (2009). 6. E. L. Stone et al., ICOS coreceptor signaling inactivates the transcription factor FOXO1 to promote Tfh cell differentiation. Immunity 42, 239-251 (2015). 7. S. Akira, K. Takeda, Toll-like receptor signalling. Nature Rev. Immunol. 4, 499-511 (2004). 8. J. L. Kubinak et al., MyD88 signaling in T cells directs IgA-mediated control of the microbiota to promote health. Cell Host Microbe 17, 153-163 (2015). 9. R. Morita et al., Human blood CXCR5(+) CD4(+) T cells are counterparts of T follicular cells and contain specific subsets that differentially support antibody secretion. Immunity 34, 108-121 (2011). 10. Y. Che et al., Circulating memory T follicular helper subsets, Tfh2 and Tfh17, participate in the pathogenesis of Guillain-Barre syndrome. Sci. Rep. 6, 20963 (2016). 11. J. Yang et al., Critical roles of mTOR Complex 1 and 2 for T follicular helper cell differentiation and germinal center responses. Elife 5, 17936 (2016). 12. H. Zeng et al., mTORC1 and mTORC2 Kinase Signaling and Glucose Metabolism Drive Follicular Helper T Cell Differentiation. Immunity 45, 540-554 (2016). 84 13. J. H. Chang et al., MyD88 is essential to sustain mTOR activationnecessary to promote T helper 17 cell proliferation by linking IL-1 and IL-23 signaling. Proc. Natl. Acad. Sci. U. S. A. 110, 2270-2275 (2013). 14. A. M. Trama et al., HIV-1 envelope gp41 antibodies can originate from terminal ileum B cells that share cross-reactivity with commensal bacteria. Cell Host Microbe 16, 215-226 (2014). 15. W. B. Williams et al., Diversion of HIV-1 vaccine-induced immunity by gp41-microbiota cross-reactive antibodies. Science 349, aab1253 (2015). CHAPTER 5 GUT IMMUNITY REGULATES INTER-MICROBIOTA INTERACTIONS THAT PREVENT METABOLIC DISEASE 86 Introduction Over the past century, obesity has developed into a U.S. epidemic. Currently, less than one-third of the U.S. population is considered to have a healthy normal weight. Obesity, as measured by a body-mass-index (BMI) of 30 or greater, significantly increases mortality compared to a healthy BMI of less than 25 (1). Additionally, obesity is associated with a number of comorbidities such as type II diabetes, cardiovascular disease, and liver disease. Heart disease alone is the cause of 1 in 4 deaths within the US, highlighting the significant impact this disease has on the health and longevity of individuals (2). While it has been previously established that both diet and genetics play a role in the development of obesity, it has become more apparent over the past decade that an additional factor, the microbiota, also contributes. Obese individuals have a distinct microbiota composition as compared to lean individuals (3). This has functional consequences, as obese microbiota is capable of transferring weight gain to germfree mice, demonstrating its ability to induce obesity in addition to diet and genetics (4). While researchers have gained insight into which microbes are associated with lean or obese body mass, it has been difficult to identify the mechanisms by which these shifts within the microbiota occur. Diet has been demonstrated to have a significant influence on microbiota composition, however, host genetics also shape the microbiota architecture (5). One major biological pathway responsible for sculpting commensals within the gut is the secretion of immunoglobulin A (IgA) (6-9). IgA is the most abundant antibody produced within the body and is secreted into 87 mucosal surfaces to modulate community membership and antigen expression within the microbiota as well we maintain a healthy barrier between the commensals and host tissue (7, 10). IgA production is driven by both T cell-dependent and T cell–independent mechanisms. We recently revealed a requirement for MyD88 signaling within T cells to direct proper IgA targeting of the microbiota via follicular T helper (Tfh) cells development and germinal center responses along the gut (7). Animals lacking MyD88 within T cells developed dysbiosis, or harmful shifts within the microbiota, that predisposed them to increased susceptibility to colitis. We have since uncovered a requirement for this pathway in the prevention of dysbiosisinduced obesity. Disruptions to IgA targeting of commensals within mice lacking MyD88 signaling in T cells, lead to the development of obesity as mice age. Shifts within the microbiota prior to weight gain are capable of transferring the disease to otherwise healthy animals. These shifts include expansion of the pathobiont, Desulfovibrio, and widespread loss of species and functional diversity. Reintroduction of beneficial spore-forming microbes associated with a lean phenotype are able to rescue weight gain in predisposed animals. Finally, the initial defect in T cell-dependent IgA is responsible for initiating dysbiosis and weight gain, providing us with valuable insight into host dependent mechanisms responsible for an obese associated microbiota composition. 88 Results Loss of MyD88 signaling within T cells results in the spontaneous development of obesity The microbiota is a rich and constant source of microbial ligands that can be recognized by Toll-like receptors (TLRs). These microbial products are not only found within the gut, but have also been shown to be circulating within the blood. As T cells are known to express TLRs, this represents a mechanism by which microbial products can directly influence T cell responses. Most TLRs, with the exception of TLR3, signal through the adaptor molecule MyD88; therefore, we crossed an MyD88 floxed animal to a CD4-Cre driver to make T-MyD88-/mice in order to study how T cells might be influenced by MyD88 dependent signals (7). Deletion of MyD88 in the T cell compartment leads to local reductions in TFH development within the gut, and we have subsequently demonstrated that this pathway functions to control the microbiota by directing IgA binding (7, 11). While conducting this previous study, we noted that as animals aged they gained significantly more weight when compared to age and sex matched wild-type animals (Figure 5.1A). Despite being fed a standard diet, T-MyD88-/- mice exhibit significantly increased weight gain by 6 months of age that is characterized by increased fat accumulation (Figure 5.1A-C and Figure S5.1A and S5.1B). By one year of age, male T-MyD88-/- mice weigh as much as sixty grams and are composed of, on average, fifty percent fat based on NMR analysis (Figure 5.1D and 5.1E). In addition to weight gain, there are a number of comorbidities that are associated with obesity, including diabetes, heart disease, and liver disease. 89 B T-MyD88-/- C Weight Gained (%) WT 100 WT T-MyD88-/* **** * 50 15 5 0 2 3 4 5 6 7 8 0 9 10 40 20 0 F WT T-MYD H **** 60 40 20 0 WT T-MYD ** 8 6 4 2 0 I WT T-MyD88-/- 3 4 5 6 7 8 9 10 G WT T-MYD Liver 2 Age (months) Glucose (mg/dl) 60 E Insulin (ng/mL) **** Fat (%) Weight (g) 80 * 10 Age (months) D **** WT T-MyD88-/-** *** 20 ** Fat (g) A WT 200 WT T-MyD88-/- 150 100 **** 50 0 0 20 40 Minutes 60 Adipose T-MyD88-/- 10x 10x 20x 20x Figure 5.1: Mice lacking MyD88 signaling within T cells develop obesity. A) Side by side image of 6-month WT and T-MyD88-/- mice. B) Percentage of weight gained as mice age, starting at 2 months of age (WT, N=8; T-MyD88-/-, N=7). C) Grams of fat accumulation as mice, age starting at 2 months of age (WT, N=8; T-MyD88-/-, N=7). D) Total weight in grams of one year old WT and TMyD88-/- mice (N=6). E) Total fat percentage as measured by NMR of one year old WT and T-MyD88-/- mice (N=6). F) Fasting serum insulin (ng/mL) concentrations within 1 year old WT and T-MyD88-/- mice (WT, N=9; T-MyD88-/-, N=10). G) Blood levels of glucose (milligram per decaliter) measured over time following i.p. insulin (0.75U/kg) injection during insulin resistance test. H) Representative image of H&E histology of the liver from one year old WT and TMyD88-/- mice. I) Representative image of H&E histology of adipose tissue from one year old WT and T-MyD88-/- mice. Statistics: P-value<0.05 (*); P-value<0.01 (**); Pvalue<0.001 (***) using a two-tailed, unpaired t-test (B-F) and a repeated measures ANOVA (G). 90 While one-year-old T-MyD88-/- mice raised on a standard diet cleared glucose to similar levels as their WT counterparts (Figure S5.1C), these mice have significantly higher levels of circulating insulin (Figure 5.1F). Moreover, when animals are challenged with additional insulin, T-MyD88-/- fail to clear glucose with similar kinetics as WT animals, indicating that these animals also develop insulin resistance (Figure 5.1G). Histological analysis of T-Myd88-/- liver tissue by hematoxylin and eosin (H&E) staining revealed the presence of steatosis, or fat deposition, that is not seen in WT animals (Figure 5.1H). Moreover, while visceral adipocytes are uniform in size and shape in WT mice, adipocytes within T-MyD88-/- animals are much larger and irregular in shape. In addition to this, we detected macrophage infiltration and prominent formation of “crown-like” structures, a hallmark of obesity, in the visceral adipose tissue (VAT) of TMyD88-/- animals but not in WT animals. (Figure 5.1I). Compared to WT animals, T-MyD88-/- mice have a number of metabolic shifts that contribute to their agedependent obesity. These include reductions in oxygen and carbon dioxide respiration that result in reduced energy expenditure by two months of age, as well as increases in food intake that occur at the onset of spontaneous weight gain (Figure S5.1D-G). Other factors such as heat and movement, however, are not altered in these mice (Figure S5.1H-S5.1I). Collectively, these data demonstrate the requirement for intact innate signaling within T cells to prevent the onset of metabolic syndrome Obesity takes months to develop on a standard mouse diet in this model, making it difficult to tease apart the mechanisms for increased weight gain. Therefore, we tested whether placing animals on a high- 91 fat diet (HFD) would accelerate the phenotype. WT and T-MyD88-/- mice were fed a control diet containing 10% of the calories made from fat or a HFD containing 45% of calories made from fat for 4 months. As has been reported, mice receiving HFD gained more weight than the control fed animals, however, TMyD88-/- mice fed HFD gained significantly more weight than WT animals (Figure S5.1J and S5.1K). Thus, similar trends were observed in HFD and aged animals, and obesity in T-MyD88-/- mice is independent of the type of diet. Host immune responses are dysregulated within adipose tissue of T-MyD88-/- mice Given that we have observed deficiencies in the immune response within the intestine of animals lacking T cell intrinsic MyD88 signaling, and inflammation is a known contributor to obesity and metabolic syndrome, we interrogated the immune cell populations within adipose tissue. Under lean conditions, regulatory T cells (Tregs) are maintained in high frequency within adipose tissue (12, 13). However, as obesity develops, the immune system shifts towards an inflammatory response that can exacerbate fat accumulation. Consistent with a shift in adipose specific immune responses, T-MyD88-/- mice have significantly fewer Tregs within adipose tissue on either diet (Figure 5.2A-C and Figure S5.2A-B). During obesity, increased adipocyte cell death leads to the recruitment of macrophage populations that surround dead adipocytes and promote inflammatory immune responses (14). We also observe an increase in macrophage accumulation within the VAT in T-MyD88-/- mice when compared to 92 A WT C B T-MyD88-/- *** CD4 30 20 10 0 FoxP3 WT cd11b 4 ** 1 0 p=.07 3 p=.09 12 month Relative expression ** ** * ** 1 0.0 F 30 20 10 0 3×104 WT T-MYD p=.07 2×104 1×104 0 WT T-MYD 2 11 b 45 TMYD 5-ASA TMYD CTRL WT 5-ASA WT CTRL 40 35 ns ** 30 25 20 2 4 Age 6 8 D L1 H C L2 C XC 3 xP C C IL 4 Fo et IF N g G at a3 Tb O R gT 0 R 40 5.0×103 p=0.08 6 month 2 2 ** * 1.0×104 WT T-MYD WT T-MYD 3 WT T-MYD Weight (g) F4/80 7.16% G E T-MyD88-/17.7% Macrophages (%) D # Tregs per gram 8.11% 34.3% 1.5×104 # macrophages per gram Tregs (%) 40 Figure 5.2: Obese T-MyD88-/- mice have greater inflammation within adipose tissue. A) Representative flow cytometry plots were previously gated on CD3+CD4+ T cells within adipose tissue and measured for Tregs, defined as CD3+CD4+FoxP3+ cells. B) Percentage of Tregs within adipose tissue. C) Total number of Tregs within adipose tissue (WT, N=12; T-MyD88-/-, N=10). D) Representative flow cytometry plots were previously gated on CD45+ cells within adipose tissue and measured for macrophages, defined as CD45+CD11b+F4/80+ cells. E) Percentages of macrophages within adipose tissue. F) Total numbers of macrophages within adipose tissue (WT, N=12; TMyD88-/-, N=10). G) qPCR of mRNA expression of various T cell and myeloid cell immune parameters for 6-month and 12-month-old mice (6 month: WT, N=3; T-MyD88-/-, N=4; 12 months: WT, N=8; T-MyD88-/-, N=7). H) Percentage of weight gained by WT and T-MyD88-/- mice fed control diet or 5-ASA diet, starting at 2 months of age (WT CTRL, N=3; WT 5-ASA, N=4; T-MYD CTRL, N=3; TMYD 5-ASA, N=4). Statistics: P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using a two-tailed, unpaired t-test (A-G) and a repeated measures ANOVA (H). macrophage accumulation within the VAT in T-MyD88-/- mice when compared to 93 WT animals in both frequency and total number of macrophages on normal chow but not HFD (Figure 5.2D-F, Figure S5.2C-D). As a number of MyD88 dependent cytokines, including IL-33, have been implicated in the development of metabolic syndrome, we isolated mRNA from VAT to analyze a number of T cell and myeloid cell markers. T-MyD88-/- mice that were either aged or fed a HFD had significant shifts within a number of immune genes when compared to their respective controls. These included elevations in the TH1 genes, Tbet and IFN-g, and reductions in Foxp3 and Gata3 (Figure 5.2G and Figure S5.2E). Changes in gene expression within the adipose tissue were subtle in aged T-MyD88-/animals when compared to WT animals, however HFD appeared to exacerbate some of these differences (Figure S5.2E). As many of the shifts observed within immune parameters revealed enhanced inflammation, and chronic low grade inflammation is thought to influence weight gain, we placed WT or T-MyD88-/- mice on a normal chow diet containing 5-ASA, an anti-inflammatory derivative of salicylic acid that has been demonstrated to reduce complications of obesity in mice fed HFD (15). We then monitored age-dependent weight gain over 8 months. Treatment of T-MyD88-/mice with 5-ASA did not significantly ameliorate enhanced weight gain, suggesting that chronic low grade inflammation is likely not involved in the development of spontaneous obesity within T-MyD88-/- mice (Figure 5.2H). Collectively, these data suggest that shifts within the host immune system alone were not the main cause of the metabolic syndrome and obesity that develops in these mice. 94 T cell intrinsic MyD88 signaling supports microbiota diversity and colonization of beneficial Clostridia We have previously reported that defective IgA targeting of the microbiota in T-MyD88-/- animals lead to shifts in microbial composition that exacerbated chemically induced models of intestinal colitis (7). Worsened disease is rescued by a microbial transplant from a healthy donor, indicating that the absence of the T cell intrinsic signaling lead to shifts in the microbiota that might be associated with disease. To identify whether the composition of the microbiota in T-MyD88-/mice was driving obesity, we placed WT and T-MyD88-/- mice on a broad spectrum antibiotic cocktail while on HFD. While WT mice exhibit no difference in weight gain while on antibiotics, obesity is almost completely rescued by antibiotic treatment of T-MyD88-/- (Figure 5.3A). This was accompanied by a reduction in body fat percentage and VAT mass that was similar to the fat accumulation observed in WT animals (Figure 5.3B and Figure S5.3A). Notably, antibiotic treatment did not influence the Treg populations within T-MyD88-/adipose tissue (Figure S5.3B). Collectively these data demonstrate that alterations within the microbiota of T-MyD88-/- mice promote obesity independent of adipose Treg function. Sequencing of 16s rRNA from fecal samples isolated from one-year-old WT or T-MyD88-/- mice was performed to determine compositional differences in the microbiota between these two genotypes. In agreement with our previous study with young mice, microbial communities between WT and T-MyD88-/- mice are significantly different between each other (Figure S5.3C). In addition to this, 95 Figure 5.3: Dysbiosis within T-MyD88-/- mice is associated with spontaneous weight gain. A) Grams of weight gained by WT and T-MyD88-/- mice fed HFD with or without antibiotics. B) Total fat percentage as measured by NMR of WT and T-MyD88-/- mice fed HFD with or without antibiotics (WT CTRL, N=5; TMYD CTRL, N=4; WT ABX, N=5, TMYD ABX, N=5). C) a-diversity measured by observed species within ileal and fecal 16s sequencing samples from one year old WT and T-MyD88-/- mice (WT, N=8; T-MyD88-/-, N=7). D) Volcano plot of ratio of T-MyD88-/- bacterial transcripts to WT transcripts within Ileal samples from one year old WT and T-MyD88-/- mice. E) Volcano plot of ratio of T-MyD88/- bacterial transcripts to WT transcripts within fecal samples from one year old WT and T-MyD88-/- mice. F) Heat map of biological pathways within fecal bacterial between one year old WT and T-MyD88-/- mice (N=6 per genotype). G) Parts of whole pie chart of bacterial classes within fecal 16s sequencing samples of one year old WT and T-MyD88-/- mice. H) Genus level comparison of Turicibacter relative abundance within fecal 16s sequencing samples of one year old WT and T-MyD88-/- mice. I) Genus level comparison of SMB53 relative abundance within fecal 16s sequencing samples of one year old WT and TMyD88-/- mice. J) Genus level comparison of Dorea relative abundance within fecal 16s sequencing samples of one year old WT and T-MyD88-/- mice (WT, N=8; T-MyD88-/-, N=7). Statistics: P-value<0.05 (*); P-value<0.01 (**); Pvalue<0.001 (***) using a repeated measures ANOVA (A), two-tailed, unpaired ttest (B,C), Mann-Whitney U test (G-J). 96 B WT ABX TMYD ABX WT CTRL TMYD CTRL 35 30 50 ** ** 25 20 15 C *** ** 40 Fat (%) Weight (g) 40 Observed Species A 30 20 10 0 2 4 6 8 * 600 400 200 0 0 10 ns 800 Weeks on HFD ABX D E Fecal F Fecal -log10 (p-value) -log10 (p-value) Ileum Ileal log2 (ratio) log2 (ratio) Reduced Increased in T-MyD88-/- G WT Reduced Increased in T-MyD88-/- T-MyD88-/Actinobacteria Actinobacteria Coriobacteriia Actinobacteria Coriobacteriia Bacteroidia Coriobacteriia Bacteroidia Bacteroidia Deferribacteres Deferribacteres Deferribacteres Bacilli Bacilli BacilliClostridia Clostridia Clostridia Deltaproteobacteria Deltaproteobacteria * p=0.01 Actinobacteria Actinobacteria Coriobacteriia Actinobacteria Coriobacteriia Bacteroidia Coriobacteriia Bacteroidia Bacteroidia Deferribacteres Deferribacteres Deferribacteres BacilliBacilli Bacilli Clostridia Clostridia Clostridia Deltaproteobacteria Deltaproteobacteria Deltaproteobacteria Deltaproteobacteria WT TMYD 0.025 0.020 ** r= -0.84 p=0.0002 0.015 0.010 0.005 0.000 30 40 50 Weight (g) 60 J SMB53 0.006 ** r= -0.79 p=0.0008 0.004 0.002 0.000 30 40 50 Weight (g) 60 Dorea Rel. Abundance I Turicibacter Rel. Abundance Rel. Abundance H 0.0015 *** r= -0.77 p=0.001 0.0010 0.0005 0.0000 30 40 50 Weight (g) 60 97 we identified significant reductions in observed species within fecal, but not ileal populations in T-MyD88-/- mice, demonstrating that microbial diversity is lost in TMyD88-/- animals (Figure 5.3C). Reduced diversity is correlated to a number of western-lifestyle associated diseases, including inflammatory bowel disease, asthma, and obesity (16-19). Indeed, individuals that contain a microbiota with lower gene richness are more likely to be obese and more likely to gain weight over time (18). One mechanism that genetic diversity influences host metabolism is by altering the network of microbial derived metabolites that are absorbed within the small and large intestine. Therefore, to understand how loss of T cell intrinsic signaling might functionally impact the microbiota, we compared the transcriptomes of ileal and fecal microbiota populations between WT and TMyD88-/- animals. The vast majority of transcripts were found to be reduced within both niches of T-MyD88-/- mice (Figure 5.3D and 5.3E). These included wide spread reductions within biological pathways such as glucose metabolism, amino acid synthesis, and fatty acid synthesis (Figure 5.3F and Figure S5.3D) Loss of functional diversity in the fecal metatranscriptome was consistent with reduced diversity identified by 16s sequencing (Figure 5.3C and 5.3E). However, the reduced diversity in T-MyD88-/- ileal transcripts were independent of any reductions in species diversity detected by 16s sequencing at this site, potentially highlighting a key difference between measuring the composition and metabolic landscape of the microbiota (Figure 5.3C and 5.3D). These data indicate that there is a loss of species and functional diversity in the microbiota of a T-MyD88-/- animal. In order to identify organisms that could 98 explain the major differences between WT and T-MyD88-/- microbiota communities, we analyzed our 16s sequencing results using LEfSe, a method which factors both effect size and statistical significance (20). LEfSe revealed that this loss of diversity was mainly restricted to the Firmicutes phylum and, more specifically, reductions in Clostridia (Figure 5.3G and Figure S5.3E). In agreement with this, 16s sequencing detected significant reductions within the bacterial families, Turicibacteraceae, Clostridiaceae, Lachnospiraceae, and Peptostreptococcaceae within T-MyD88-/- fecal samples (Figure S5.3F-I). The reductions within the first three were driven primarily by the genera Turicibacter, SMB53, and Dorea, respectively (Figure 5.3H-J). When we correlated the abundance of these microbes with total weights of both WT and T-MyD88-/- mice, we found that each exhibited a highly significant negative correlation with weight gain, supporting that loss of these organisms might influence obesity. In addition to 16s sequencing, Transcriptomic analysis provided additional resolution to identify a number of Clostridia species that were reduced within mice lacking MyD88 signaling in T cells (Figure S5.3J-K). When combined, these two analyses uncovered loss of both species and functional diversity as well as specific reductions within beneficial microbes associated with leanness. T cell-dependent reductions in Clostridiaceae are associated with metabolic syndrome Our data indicate that the microbiota formed within the absence of T cell intrinsic MyD88 signaling leads to obesity and metabolic syndrome. To determine 99 whether the microbiota is sufficient to cause disease in WT animals we conducted fecal transplants by oral gavage from either WT or T-MyD88-/- animals into two different backgrounds (Swiss Webster and C57Bl/6) of germfree WT mice. Additionally, we performed a cross-fostering experiment in which either SPF WT or T-MyD88 -/- pregnant dams were cohoused with germfree WT pregnant dams and the resulting colonized pups were separated at weaning and measured for weight gain. However, none of these methods were able to transfer the obesity phenotype of T-MyD88-/- to WT animals (Figure S5.4A). Due to the resistance of germfree animals to this dysbiosis-induced obesity, we considered that intact T cell-dependent IgA within WT animals was capable of reshaping the microbiota to a healthy state. IgA sculpting of the microbiota is initiated by T celldependent and T cell-independent pathways, and we previously discovered a defect within the former pathway when T cells lacked MyD88 signaling (7). We sought to quantify how MyD88-dependent T cell shaping of the microbiota influenced the development of spontaneous weight gain. To achieve this, we purchased TCRb-/- mice from Jackson laboratories and reduced the colonizing microbiota with broad spectrum antibiotic treatment. Animals were subsequently colonized with a 1:1 mixture of WT and T-MyD88-/- microbiota prior to adoptive transfer of either WT or T-MyD88-/- CD4+ T cells (Figure 5.4A). Mice were separated into individually housed cages and monitored weekly for weight gain and compositional changes to the microbiota. Despite the fact that these mice were initially colonized with the same microbiota, TCRb-/- - mice given T-MyD88-/CD4+ T cells gained significantly more weight when compared to TCRb-/- mice 100 Figure 5.4: T cell shaping of the microbiota is associated with spontaneous weight gain. A) Schematic of T cell shaping of the microbiota experiment. B) Percent weight gained by TCRKO mice given WT or T-MyD88-/- cells and fed a normal chow for 8 weeks (N=6 per cohort). C) Representative flow cytometry plot was previously gaited on SyBRGreen+ cells in order to quantify the percentage of IgA bound bacteria within TCRKO mice given WT or T-MyD88-/- cells and fed a normal chow for 8 weeks. D) Representative flow cytometry plot was previously gaited on SyBRGreen+ cells in order to quantify the percentage of IgG1 bound bacteria within TCRKO mice given WT or T-MyD88-/- cells and fed a normal chow for 8 weeks. E) Box and whisker plot of Unweighted Unifrac Distance between TCRKO mice given WT and T-MYD88-/- over time. F) Box and whisker plot of Weighted Unifrac Distance between TCRKO mice given WT and TMYD88-/- over time. G) Relative abundance of Clostridiaceae within fecal samples from TCRKO mice given WT and T-MYD88 after 4 weeks. Statistics: Pvalue<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using a two-tailed, unpaired t-test (B-F) and a Mann-Whitney U test (G). 101 A B Inject WT CD4+ T cells 80 Weight gained (%) Gavage 1:1 WT:TMYD Fecal Transfer Individually house and measure weight gain and microbiota composition ABX treated TCRb-/- mice Inject TMYD CD4+ T cells * 20 0 15 40 20 0 IgGA 40 10 5 WT TMYD - TCRβ-/- 0.7 ns *** 0.6 0.5 0.4 0.8 Week 0 Week 1 Week 4 Weighted ns ** ** 0.6 0.4 0.2 0.0 0.3 TCRβ-/- Week 0 Week 1 Week 4 G Clostridiaceae Relative Abundance F Unweighted p=0.09 Unweighted WT vs. TMyD88-/Unifrac Distance WT vs. TMyD88-/Unifrac Distance E * 0 IgG1 WT TMYD - WT TMYD TCRβ-/- % Max % Max IgA bound (%) 60 60 D IgG1 bound (%) C * 0.05 * 0.04 0.03 0.02 0.01 0.00 WT TMYD TCRβ-/- 102 given WT CD4+ T cells (Figure 5.4B), demonstrating that defects in MyD88 signaling within T cells drives the metabolic defects in animals. Additionally, consistent with our previous study, IgA targeting of the microbiota was defective when analyzed by flow cytometry. Only 10% of IgA coated bacteria are detected within TCRb-/- mice. However, one week post-T cell transfer, mice given WT T cells have a 3-fold increase in IgA bound microbes (Figure S5.4B). IgG1 or IgG3 responses against the microbiota took a bit longer to develop but were detectable eight weeks post T cell transfer (Figure S5.4C-D). While TCRb-/- animals receiving WT or T-MyD88-/- T cells had similar levels of total IgA as measured by ELISA, TCRb-/- mice given T-MyD88-/- T cells bound significantly fewer bacteria with IgA (Figure 5.4C and Figure S5.4B and S5.4E). IgG1-bound bacteria was also significantly reduced, however, we did not detect differences of IgG3-bound bacteria between mice receiving WT or T-MyD88-/- T cells (Figure 5.4D and Figure S5.4D and S5.4F). We analyzed the contribution of T cell intrinsic MyD88 signaling during T cell-dependent shaping of the microbiota. Over time, the microbiota composition of TCRb-/- mice given WT or T-MyD88-/- T cells became significantly dissimilar, as measured by both weighted and unweighted unifrac analysis (Figure 5.4E-F), demonstrating the influence of T cell-directed antibody targeting on the microbiota. One of the most striking differences in microbial composition between one-year-old WT and T-MyD88-/- animals is loss of Clostridia species; and, despite the fact that the mice within this experiment started with the same microbiota composition, animals receiving T cells from T-MyD88-/- had 103 significantly fewer Clostridaceae after 4 weeks when compared to animals receiving WT T cells (Figure 5.4G). While both sets of animals begin with relatively low levels of Clostridiaceae at Day 0, the Clostridiaceae were only able to expand in animals given WT T cells (Figure S5.4G). Collectively, these data demonstrate that metabolic disease is a result of a failure in the immune system to maintain appropriate microbial composition. These data support that metabolic syndrome and obesity is a combination of host genetics and environment. Obesity and metabolic syndrome in the absence of innate signaling within T cells is transmissible Our results argue that loss of critical organisms within the Clostridaceae might play a role during disease. Therefore, we sought to determine whether we could rescue metabolic syndrome in T-MyD88-/- animals with a microbiota transplant from a healthy donor. Our adoptive transfer experiments suggest that the microbiota might be inappropriately pruned if we transplanted mircrobiota only once by oral gavage. Since mice are copraphagic, we reasoned that a cohousing experiment would allow efficient and frequent transfer of microbes between genotypes. To this end, WT or T-MyD88-/- animals were either housed together with animals of the same genotype or co-housed with animals of the opposite genotype upon weaning. After 1 week of cohousing animals were placed on a HFD and monitored for signs of fat accumulation (Figure 5.5A and Figure S5.5A). The results of these experiments were quite striking; however, they were opposite of what we hypothesized. Indeed, compared to separated 104 Figure 5.5: Dysbiosis within T-MyD88-/- mice transfers obesity to WT animals. A) Schematic of timeline and samples collection for cohousing experiment. B) Percent weight gained by separated or cohoused WT and T-MyD88-/- mice fed a HFD (N=4 per cohort). C) Blood levels of glucose (milligram per decaliter) measured over time following i.p. insulin (0.75U/kg) injection during insulin resistance test. D) Total fat percentage as measured by NMR of separated or cohoused WT and T-MyD88-/- mice fed a HFD. E) Grams of VAT from separated or cohoused WT and T-MyD88-/- mice fed a HFD. F) b-diversity measured by unweighted unifrac analysis of separated or cohoused WT and T-MyD88-/- mice fed a normal chow both prior to cohousing and one week following cohousing. G) b-diversity measured by unweighted unifrac analysis of separated or cohoused WT and T-MyD88-/- mice at final timepoint following HFD (N=4 for each cohort). Statistics: P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using a repeated measures ANOVA (B), two-tailed, unpaired t-test (C-E), and permanova (F, G). 105 A B Weight Increase (%) 4 wks old 1. Collect Fecal DNA 2. Begin HFD 3 wks old 1. Collect Fecal DNA 2. Cohouse animals 250 18 wks old 1. Sacrifice animals 2. Collect Fecal DNA 200 * 100 * 100 E ns 60 ** * 50 40 50 30 20 20 Minutes 40 0.1 0.0 -0.1 ***p=0.0001 -0.2 -0.3-0.2-0.1 0.0 0.1 0.2 0.3 PC1 (24.8%) 30 Cohoused G Cohoused 1 week 0.2 0.0 -0.2 -0.4 -0.2 0.0 0.2 Cohoused High Fat Diet Cohoused 14 weeks 0.3 PC2 (13.3%) PC2 (15.4%) 0.2 0.4 40 60 Normal Chow Before Cohousing ns ** *** 20 0 0 15 50 10 F PC2 (13.1%) 5 10 Weeks on HFD VAT (g) ** Fat (%) Glucose (mg/dl) * ** 150 0.3 0 D 200 0 * 50 Insulin Resistance Separated WT Cohoused WT Separated T-MYD Cohoused T-MYD ns 150 0 C 250 Separated WT Cohoused WT Separated T-MYD Cohoused T-MYD 0.4 PC1 (20.7%) Sep WT vs. Sep TMYD, * Sep WT vs. Coh WT, * Sep TMYD vs. Coh TMYD, * Coh WT vs. Coh TMYD, ns Separated WT Cohoused WT Separated T-MYD Cohoused T-MYD 0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.3 -0.2 -0.1 0.0 0.1 0.2 PC1 (18.0%) Sep WT vs. Sep TMYD, * Sep WT vs. Coh WT, * Sep TMYD vs. Coh TMYD, ns Coh WT vs. Coh TMYD, ns 106 WT mice, T-MyD88-/- mice and any animal cohoused with them gained significantly more weight and developed insulin resistance (Figure 5.5B-C). Following three months, T-MyD88-/- mice and their cohoused counterparts weighed significantly more, had greater body fat percentages, and contained significantly more VAT than separated WT animals (Figure 5.5D-E and Figure S5.5B). This was once again independent of Treg populations within adipose tissue, however, the transferred obesity phenotype was accompanied with an increase in a number of other inflammatory markers (Figure S5.5C-D). We also monitored the composition of the microbiota over the course of the cohousing experiment. Prior to cohousing, T-MyD88-/- mice had a distinct microbiota composition, which is consistent with our previous results (Figure 5.5F). Cohousing for just one week, caused homogenous mixing of the two WT and T-MyD88-/- communities (Figure 5.5F). Following three months, however, the microbiota from cohoused animals became significantly distinct from separately housed WT mice and looked similar to separately housed T-MyD88-/- microbiota (Figure 5.5G). These data demonstrate that the microbiota formed in a T-MyD88/- animal can transfer metabolic syndrom to an otherwise healthy WT animal and suggest that the immune environment within a T-MyD88-/- animal is conducive to the growth of aggressive and potentially harmful organisms. 107 Expansion of Desulfovibrio within T-MyD88-/- mice leads to a loss of beneficial microbes that promote leanness To identify organisms that were transferred from T-MyD88-/- animals into WT hosts, we compared the microbiota composition between separately and cohoused analyzed WT and T-MyD88-/- mice. Differences in weight gain of cohoused animals were detected within the first 3 weeks, therefore, we focused on candidates that were transferred at both the early and final time points (Figure 5.5B). After three months of cohousing and HFD, Desulphovibrio, Lactobacillales, and Bifidobacterium pseudolongum, were all present at greater abundances within T-MyD88-/- and cohoused mice (Figure 5.6A and Figure S5.6A-B). However, only Desulfovibrio was significantly more abundant in separately housed T-MyD88-/- animals and cohoused animals after just one week of cohousing (Figure 5.6B). Desulfovibrio are mucolytic d-proteobacteria that produces hydrogen sulfide as a byproduct of disulfide bond degradation within mucin (2124). Colonization of Desulfovibrio is increased within patients with IBD as well as animal models of obesity and diabetes (25-30). Interestingly, it is also elevated in human patients with type II diabetes, suggesting that Desulfovibrio is highly relevant to human metabolic disease (31). We previously demonstrated that defective IgA targeting of the microbiota within T-MyD88-/- mice was correlated with an increased abundance of a number of mucolytic microbes (7). Using a recently published IgA-index calculation, we reanalyzed our previously published IgA sequencing of bound and unbound microbes in WT and T-MyD88-/- animals using LEFsE (32). Consistent with our 108 Figure 5.6: Expansion of Desulfovibrio and reductions in beneficial spore-forming microbes are associated with obesity. A) Relative abundance of Desulfovibrio within fecal 16s sequencing samples from separated or cohoused WT and TMyD88-/- mice at the final timepoint. B) Relative abundance of Desulfovibrio within fecal 16s sequencing samples from separated or cohoused WT and TMyD88-/- mice following one week of cohousing (N=4 per cohort). C) LEFsE analysis of IgA-index reanalysis of IgA targeting between WT and T-MyD88-/mice (WT, N=6; T-MyD88-/- N=7). D) Relative abundance of Dorea within fecal 16s sequencing samples from separated or cohoused WT and T-MyD88-/- mice following one week of cohousing (N=4 per cohort). E) Relative abundance of Lachnospiraceae within fecal 16s sequencing samples from mice colonized with or without D. desulfuricans and fed a HFD (N=5 per cohort). F) Relative abundance of Dorea within fecal 16s sequencing samples from mice colonized with or without D. desulfuricans and fed a HFD. G) Relative abundance of SMB53 within fecal 16s sequencing samples from mice colonized with or without D. desulfuricans and fed a normal chow (N=7 per cohort). H) Representative parts of whole pie chart of bacterial families within fecal 16s sequencing samples from germfree mice colonized with 3% chloroform treated fecal microbiota. I) Grams of weight gained by control (CTRL) or spore-forming (SF) microbe gavaged T-MyD88-/- mice fed HFD (CTRL, N=4; SF, N=5). J) Total fat percentage as measured by NMR control (CTRL) or spore-forming (SF) microbe gavaged T-MyD88-/- mice fed HFD. K) Grams of VAT of control of spore-forming microbe gavaged T-MyD88-/- mice fed HFD. Statistics: P-value<0.05 (*); Pvalue<0.01 (**); P-value<0.001 (***) using a Mann-Whitney U test (A-G), repeated measures ANOVA (I), two-tailed, unpaired t-test (J, K). 109 B Desulfovibrio Rel. Abundance * Rel. Abundance * 0.06 WT T-MYD C Desulfovibrio 0.04 0.02 0.00 0.06 * * Prevotellaceae Prevotella Odoribacteraceae Odoribacter Deferribacterales Deferribacteres Mucispirillum Deferribacteraceae Deferribacteres M. shaedleri 0.04 0.02 0.00 -4 -2 R. gnavus Desulfovibrionaceae 0 2 4 D Dorea Rel. Abundance A 0.020 * * 0.015 0.010 0.005 0.000 LDA Cohoused 0.05 0.00 - + 0.008 0.006 0.004 0.002 0.000 - 40 35 30 25 20 15 * CTRL SF 0 5 10 Weeks on HFD 15 0.0010 0.0005 0.0000 - + K 40 * 35 30 25 CTRL SF H * 0.0015 D.d + J I Total Weight (g) * 0.010 D.d SMB53 Rel. Abundance 0.10 G Dorea Visceral Adipose Tissue * 0.15 D.d F Rel. Abundance Rel. Abundance Lachnospiraceae Fat (%) E Cohoused Cohoused 2.5 *** 2.0 1.5 1.0 0.5 0.0 CTRL SF 53.84% Unclassified Clostridiales 30.61% Ruminococcaceae 6.36% Clostridiaceae 5.36% Lachnospiraceae 2.17% Peptostreptococcaceae 0.96% Turicibacteraceae 0.43% Mogibacteriaceae 0.06% Anaeroplasmataceae 0.06% Christensenellaceae 0.04% S24-7 0.02% Peptococcaceae 0.02% Streptococcaceae 0.02% Dehalobacteriaceae 0.01% Enterobacteriaceae 0.01% Coriobacteriaceae 0.01% Erysipelotrichaceae 0.01% Alcaligenaceae 53.84% Unclassified Clostridiales 30.61% Ruminococcaceae 6.36% Clostridiaceae 5.36% Lachnospiraceae 2.17% Peptostreptococcaceae 0.96% Turicibacteraceae 0.43% Mogibacteriaceae 0.06% Anaeroplasmataceae 0.06% Christensenellaceae 0.04% S24-7 0.02% Peptococcaceae 0.02% Streptococcaceae 0.02% Dehalobacteriaceae 0.01% Enterobacteriaceae 0.01% Coriobacteriaceae 0.01% Erysipelotrichaceae 0.01% Alcaligenaceae 110 earlier findings, WT and T-MyD88-/- IgA showed differential targeting of Mucispirillium schaedleri and Ruminococcus gnavus (Figure 5.6C). However, Lefse analysis of IgA-coated microbes also identified members from the family Desulfovibrionaceae that exhibited significantly altered IgA targeting between WT and T-MyD88-/- mice, providing a potential mechanism for the expansion of this pathobiont (Figure 5.6C). Inappropriate expansion of organisms can have a negative impact on the niche of beneficial microbes. One of the most significant changes to the microbiota that we have identified throughout these experiments is a profound loss of abundance and function of organisms that are known to have beneficial consequences, including members of Clostridia (Figure 5.3G). Specifically, Lachnospiraceae such Dorea and SMB53, are significantly reduced in T-MyD88-/mice, and cohousing of WT mice with T-MyD88-/- animals leads to reduced Dorea colonization (Figure 5.3I and 5.3J and Figure 5.6D). Therefore, we hypothesized that an increased abundance of Desulfovibrio might cause concurrent deficiencies in the colonization of other organisms. To directly test this, we colonized mice with an ATCC isolate of Desulfovibrio desulfuricans that contained greater than 97% 16s sequence similarity to our commensal. Mice were fed a normal chow or HFD for one week with or without the addition of D. desulfuricans within their drinking water. Consistent with our hypothesis, WT SPF animals fed a HFD had significant reductions in Lachnospiraceae, and specifically, Dorea, when they were colonized with D. desulfuricans (Figure 5.6E and 5.6F). Similarly, animals fed normal chow had significant decreases in 111 SMB53 in the presence of D. desulfuricans (Figure 5.6G). Collectively, these data demonstrate that an expansion in Desulfovibrio species, as seen in TMyD88-/- mice, can impact the colonization of microbes associated with leaness. Obese T-MyD88-/- mice have pronounced reductions in the colonization of Turicibacter, Dorea, and SMB53 that have strong negative correlations with fat accumulation. We sought to determine whether replacement of organisms associated with leaness could reduce fat accumulation. Turicibacteraceae, Lachnospiraceae, Peptostreptococcaceae, and Clostrideaceae are known sporeformers which can be enriched from the fecal microbiota by chloroform treatment. To ensure that this treatment selects for organisms that are specifically depleted in the T-MyD88-/- mice, we gavaged germfree animals with chloroform treated fecal slurries isolated from WT SPF mice in our colony. Indeed, this purification enriched for Clostridiaceae and Lachnospiraceae (Figure 5.6H). Our cohousing and T cell transfer experiments suggest that continuous administration of organisms might be required to see an effect. Therefore, we performed twiceweekly gavages of T-MyD88-/- mice with sporeforming microbes and subsequently fed them a HFD. Strikingly, treatment of obesity prone T-MyD88-/animals with a cocktail of spore-forming bacteria significantly reduced weight gain and fat accumulation (Figure 5.6I). At the end of three months, T-MyD88-/mice treated with spore-forming microbes had a lower body fat percentage and a reduced VAT mass when compared to untreated T-MyD88-/- mice (Figure 5.6J and 5.6K). Together, these data argue that early expansion of Desulfovibrio within T-MyD88-/- mice reduces beneficial microbes required to maintain healthy 112 weight and prevent metabolic disease. Discussion Here we have demonstrated a mechanism by which dysbiosis within the microbiota, caused by defective T cell-dependent IgA targeting within the gut, causes obesity. T-Myd88-/- mice experience an expansion of Desulfovibrio early in life which results in reductions in beneficial microbes and widespread losses in both species and functional diversity. Weight gain and fat accumulation can be prevented in T-MyD88-/- mice through the reintroduction of spore-forming microbes. Over the last decade, investigators have identified shifts within the microbiota that are concomitant with the development of metabolic syndrome. Fecal transfers of obese microbiota to WT germfree mice induce increased weight gain when compared to germfree animals given lean microbiota (4). Seminal works such as these have demonstrated that the microbiota does indeed help to drive obesity, however, much has remained unknown regarding which microbes are responsible for this, as well as the mechanism by which the obesity-inducing microbiota develops. In recent studies, obesity and type 2 diabetes have been correlated with decreased diversity, expansion of Desulfovibrio, and losses of beneficial spore-forming microbes (18, 31). While there is a clear role that connects the microbiota to obesity there is little known about the organisms that cause or protect against increased weight gain and associated pathologies. Additionally, the impetus of dysbiosis development has been previously unknown and difficult to study. Our mouse model has allowed us 113 to define the contribution of host regulation of the microbiota and dysbiosis during early development that leads to metabolic syndrome later in life. Furthermore, it has highlighted subclinical shifts within the microbiota that may occur prior to obesity and can be used as therapeutic targets for the prevention of metabolic syndrome. The increase of many diseases, including obesity, has coincided with shifting social customs linked to the wester-lifestyle, including antibiotics use, increased ratio of simple to complex carbohydrate intake, and hyperhygienic practices (33). These alterations to our behavior and environment have lasting effects on our microbiota (34-36). Despite this rise, however, many individuals experience no clinical consequences to a westernized lifestyle. Given these discrepancies, it is accepted within the scientific community that environmental and genetic factors likely interact to promote or protect against these complex diseases. Directed IgA targeting of the microbiota may be one mechanism by which host immune responses predispose individuals to a greater susceptibility toward disease onset. In agreement with this notion, studies have revealed that roughly 1 in 600 people is IgA-deficient, making it the most common primary immunodeficiency (37). As many as 30-50% of these individuals experience recurrent mucosal infections or develop autoimmune diseases (38, 39). This is a dramatic increase in disease susceptibility over individuals with healthy levels of IgA. Despite these statistics, many individuals are never diagnosed with IgAdeficiency, and, therefore, additional screening for this disorder might reveal new links between IgA deficiency and westernized lifestyle-linked diseases. Studies 114 such as this provide unique opportunities to understand intricacies of immune development while gaining insight into the impact of specific commensals on the metabolic and compositional landscape of the microbiota. While the microbiota as a whole has evolved a vast network of species and redundant metabolic functions, identifying the underlying mechanisms of disease associated perturbations reveals keystone interactions that are necessary to push the balances in favor of health. Experimental Procedures Mice C57Bl/6 MyD88LoxP/LoxP mice (Jackson Laboratories) were crossed to C57Bl/6 CD4-Cre animals (Taconic) to produce MyD88+/+ CD4-Cre+ mice (WT) and MyD88LoxP/LoxPCD4-Cre+ (T-MyD88-/-) animals. C57Bl/6 MyD88LoxP/LoxP mice (Jackson Laboratories) were crossed to C57Bl/6 FoxP3-EGFP-cre+ (Jackson Laboratories) animals to produce MyD88+/+-FoxP3-cre+ (FoxP3-cre+) mice, MyD88-LoxP/LoxP (MyD88fl/fl) mice, and MyD88loxp/loxp FoxP3-cre+ (TregMyD88-/-) mice. Age-matched male mice were used to compare the spontaneous weight phenotype, including immune and microbiota response, on standard diet. Age-matched male and female mice were used to compare the weight phenotype, including immune and microbiota response, on high-fat diet (HFD). TCRb-/- To measure T cell-dependent shaping of the microbiota, 4-week old mice (Jackson Laboratories) were purchased. Desulfovibrio desulfuricans dependent shaping of the microbiota, 6-week old WT C57Bl/6 mice 115 (Jackson Laboratories) were purchased or age-matched CD4-Cre+ (WT) mice from our facility were used. GF mice were maintained in sterile isolators and verified monthly for GF status by plating and PCR of feces. GF C57Bl/6 animals were used in this study. The use of animals in all experiments was in strict adherence to federal regulations as well as the guidelines for animal use set forth by the University of Utah Institutional Animal Care and Use Committee. Colonization of mice with spore-forming microbes Fecal pellets were taken from WT mice and incubated in reduced PBS containing 3% chloroform (v/v) for 1 hr at 37°C in an anaerobic chamber. A control tube containing just reduced PBS and 3% chloroform was also incubated for 1 hr at 37°C in an anaerobic chamber. After incubation, tubes were gently mixed and fecal material was allowed to settle for 10 sec. Supernatant was transferred to a fresh tube and chloroform was removed by forcing CO2 gas into the tube. Mice within the SF cohort were orally gavaged with 100µL of spore forming fecal fraction. Mice within the CTRL cohort were orally gavaged with 100µL of PBS control that also had chloroform removed. Adipose immune cell isolation Reproductive fat pads were dissected from animals and weighed prior to immune cell isolations. Fat tissue was placed in ice-cold, sterile 1X HBSS and mechanically homogenized by repeated suction through a 1 mL syringe. Homogenized fat tissue was incubated in a solution containing sterile 1X HBSS 116 containing 5% (v/v) fetal bovine serum (GIBCO BRL), 50 U/ml Dispase (Roche), 0.5 milligram per milliliter Collagenase D (Roche), and 0.5 milligram per milliliter DNaseI (Sigma) for 30 min at 37°C on a shaker. The supernatant was filtered over a 40 mm cell strainer into ice-cold, sterile 1X HBSS. Cells were pelleted by spinning at 1000g at 4°C for 10 min. The pellet was treated with 1X RBC Lysis Buffer (Biolegend) to lyse and remove red blood cells. Remaining cells were washed and resuspended with HBSS (10 mM HEPES [Cellgro], 2 mM EDTA [Cellgro], and 0.5% [v/v] fetal bovine serum [GIBCO]) and prepared for flow cytometry analysis. Diet treatment Animals housed within the SPF facility were fed a standard chow of irradiated 2920x (Envigo). Mice were fed a high-fat diet of 45% kcal% fat DIO mouse feed (Research Diets) or a diet of 10% kcal% fat DIO mouse feed (Research Diets) as a control during HFD experiments. Mice were also fed a custom diet containing irradiated standard 2020 chow containing 1% 5-ASA (Envigo) or a control diet lacking the 5-ASA (Envigo) during 5-ASA inflammation experiments. Antibiotics treatment WT and T-MyD88-/- mice were maintained on 0.5 milligram per milliliter of ampicillin (Fisher Scientific), neomycin (Fisher Scientific), erythromycin (Fisher Scientific), and gentamicin (GoldBio) within their drinking water for 14 weeks 117 while being fed a HFD in order to determine the relative contribution of the microbiota to the weight gain phenotype. TCRb-/- mice were placed on 0.5 milligram per milliliter of ampicillin (Fisher Scientific), neomycin (Fisher Scientific), erythromycin (Fisher Scientific), and gentamicin (GoldBio) within their drinking water for 1 week to reduce the endogenous microbiota before being recolonized by fecal transfers. T cell shaping of the microbiota within tcrb-/- mice Three separate cages of 4 TCRb-/- mice were placed on an antibiotic cocktail within their drinking water for one week. Antibiotics was removed for twenty-four hr before any further treatment. One fecal pellet from a WT donor and 1 fecal pellet from a T-MyD88-/- donor was mashed in reduced PBS containing 0.1% cysteine and immediately orally gavaged into the TCRb-/- mice. This oral gavage was repeated every other day for one week. Forty-eight hr following the final gavage, mice were placed into individually housed cages and retro-orbitally injected with 5x106 CD4+ MACS-enriched WT or T-MyD88-/- cells. This was labeled as D0. Glucose tolerance test Mice were fasted for 6 hr prior to being challenged with glucose. Fasting levels of glucose were detected using a Contour Glucose Meter (Bayer) and Contour Glucose Strips (Bayer). One milligram per gram of body weight of glucose was injected intraperitoneally (i.p.) into animals at timepoint zero. Blood 118 levels of glucose were measured at 5,15,30,60, and 120-min time points using the glucose meter. Insulin resistance test Mice were fasted for 6 hr prior to being challenged with glucose. Fasting levels of glucose were detected using a Contour Glucose Meter (Bayer) and Contour Glucose Strips (Bayer). 0.75U/kg of body weight of insulin was injected intraperitoneally into animals at timepoint zero. Blood levels of glucose were measured at 5, 10,15, 20, 25, 30, 40, and 60-min time points using the glucose meter and strips. Animals were removed from the experiment following an 150µL i.p. injection of 25% glucose if blood glucose levels dropped to 30milligram per decaliter. Flow cytometry of isolated immune cells Surface staining for immune cells was done in sterile 1XHBSS (Corning) supplemented with 10mMHEPES (Cellgro), 2mMEDTA (Cellgro), and 0.5% (v/v) fetal bovine serum (GIBCO BRL) for 20 min at 4°C. Cells were then washed twice in supplemented 1X HBSS and enumerated via flow cytometry. The following antibodies were used: anti-CD4 (eBioscience clone RM4-5 PerCPCy5.5), anti-CD3 (Tonbo bio: 145-2C11 Pacific Blue), anti-CD45 (eBioscience clone 30-F11 Fitc), anti-F4/80 (Biolegend clone BM8 PE-Cy7), anti-CD11b (Tonbo bio clone M1/70 PerCP-Cy5.5). For intracellular staining, permeabilized and fixed in 100 ml Perm/Fix buffer (eBiosciences) at 4°C overnight. Cells were 119 washed twice in Perm/Wash buffer (eBioscience) and then stained for intracellular cytokines with the following antibodies: anti-Foxp3 (eBioscience: clone FJK-16 s APC). Cells were again washed twice in Perm/Wash buffer and then placed with supplemented HBSS (10 mM HEPES [Cellgro], 2 mM EDTA [Cellgro], and 0.5% [v/v] fetal bovine serum [GIBCO BRL]) and enumerated via flow cytometry. These data were collected with a BD LSR Fortessa and analyzed with FlowJo software. RNA isolation and qPCR Sections of tissue 0.5 cm in length or 1X105 cells were stored at -70°C in 700µL of RiboZol (VWR). RNA was isolated using the Direct-zol RNA MiniPrep Kit (Zymoresearch). cDNA was synthesized using qScript cDNA synthesis kit (Quanta Biosciences). qPCR was conducted using LightCycler 480 SYBR Green I Master (Roche). qPCR experiments were conducted on a Lightcycler LC480 instrument (Roche). Quantification of immunoglobulins within feces To quantify luminal IgA, fecal pellets were collected in 1.5mL Eppendorf tubes and weighed. Luminal contents were resuspended in 10µL of sterile 1X HBSS per milligram of fecal weight and spun at 100 x g for 5 min to remove course materials. Supernatants were then placed in a new 1.5mL Eppendorf tube and spun at 8000 x g for 5 min to pellet bacteria. Supernatants (containing IgA) were then placed in a new 1.5mL 120 Eppendorf tube and used as samples (1/10 and 1/100 (v/v) dilutions) for an IgA specific ELISA kit (eBioscience; performed per kit instructions). Absorbance was read at 450nm and concentrations of IgA were calculated with a standard curve. Concentrations were normalized to fecal weight. Bacterial pellet was resuspended in 500µL of sterile PBS and washed twice by spinning at 8000g for 5 min. The washed bacterial pellet was then resuspended in 10µL of sterile PBS per milligram of feces. 5µL of each sample were plated on to a 96 well, round-bottom plate. Bacteria was blocked for 15 min at room temperature with 100µL of sterile HBSS containing 10% (v/v) FBS. Without washing cells, 100µL of anti-IgA (ebioscience clone mA-6E1 PE), antiIgG1(Santacruz CruzFluor555), or anti-IgG3 (Santacruz CruzFluor555) diluted at 1:500 sterile HBSS containing 10% (v/v) FBS was added to the wells. Wells were incubated at 4°C for 30 min. The plate was washed twice by spinning at 2500g for 5 min before flicking off supernatanant and resuspending cells in sterile HBSS. After final wash, bacterial wells were resuspended in 250µL of HBSS containing 5µL of 1x SYBRgreen stain (Invitrogen cat #S7563). Wells were incubated for 20 min at 4°C before immediate enumeration on a flow cytometer. Rag1-/- fecal pellets were included in all experiments as negative controls. FACS purified Treg and non-Treg for validation of Treg-MyD88-/- mouse Lymphocytes were isolated from MyD88+/+ FoxP3-EGFP-cre+ or MyD88loxp/loxp FoxP3-EGFP-cre+ mouse spleens and the resulting cells were 121 sorted through MACS columns by negative selection (CD4+ T Cell Isolation Kit II (Miltenyi)). Lymphocyte enrichment was performed per kit instructions. Isolated T cells were further purified via FACS with a BD FACSAria Cell Sorter. For Tregs, CD3+CD4+ FoxP3-EGFP-cre+ cells were gated on and sorted. For non-Treg T cells, CD3+CD4+ FoxP3-EGFP-cre were gated on and sorted. Cells collected into RPMI media (Corning) supplemented with 10% fetal bovine serum (v/v) (Gibco BRL), 50 U/mLpenicillin, 50 μg/mL streptomycin (Cellgro), 5μM 2Mercaptoethanol (CalBiochem), 1μMsodium pyruvate (Cellgro), 1X MEM Nonessential Amino Acids (Cellgro), and 2.05mM L-glutamine (Cellgro). 1X105 Tregs and non-Treg T cells were spun down and stored in Ribozol at 70°C. Growth of Desulfovibrio desulfuricans ATCC 27774 The bacterial species Desulfovibrio desulfuricans was purchased from ATCC (#27774). Vial was handled and opened per ATCC instructions for anaerobic bacteria and cells were grown in Desulfovibrio media described in Rey et al., PNAS 2013. Media was composed of NH4Cl (1g/L) (Fisher Chemical), Na2SO4 (2g/L) (Fisher Chemical), Na2S2O35H2O (1g/L) (Sigma), MgSO47H2O (1g/L) (Fisher Chemical), CaCl22H2O (0.1g/L) (Fisher Chemical), KH2PO4 (0.5g/L) (Fisher Bioreagents), Yeast Extract (1g/L) (Amresco), Resazurin (0.5mL/L) (Sigma), Cysteine (0.6g/L) (Sigma), DTT (0.6g/L) (Sigma), NaHCO3 (1g/L) (Fisher Chemical), Pyruvic Acid (3g/L) (Acros Organics), Malic Acid (3g/L) (Acros Organics), ATCC Trace Mineral Mix (10mL/L), ATCC Vitamin Mix (10mL/L). Adjusted to pH of 7.2 bacteria was grown for 48hr in an anaerobic chamber (Coy 122 Labs) and stored in growth media containing 25% glycerol at 70°C. 2.5X108 bacterial CFUs were added to 250µL of drinking water of mice for one week. Isolation and 16s sequencing of fecal and ileal DNA Animals were sacrificed and their entire lower digestive tract (from duodenum to rectum) was removed and longitudinally sectioned. One fecal pellet and luminal content from lower 10cm of small intestine were collected from each animal to characterize the fecal and ileal microbiota communities, respectively. Fecal and ileal samples were immediately frozen at -70°C in 2mL screw cap tubes containing ~250 milligram of 0.15 mm garnet beads (MoBio, cat# 13122500). DNA was extracted using the Power Fecal DNA Isolation Kit (MoBio), per kit instructions. 16s DNA amplification and sequencing was done as previously described in Kubinak, Petersen et al., Cell Host and Microbe 2015. Metabolic phenotyping Total body fat composition was measured on an NMR Bruker Minispec. CLAMS Metabolic Cages were used to measure indirect calorimetry. Both services were provided by the Metabolic Phenotyping Core, a part of the Health Sciences Cores at the University of Utah. References 1. K. M. Flegal, B. K. Kit, H. Orpana, B. I. Graubard, Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis. JAMA 309, 71-82 (2013). 123 2. D. Mozaffarian et al., Heart disease and stroke statistics—2015 update a report from the american heart association. Circulation (2015). 3. R. Ley, P. J. Turnbaugh, S. Klein, J. I. Gordon, Microbial ecology: Human gut microbes associated with obesity. Nature 444, 1022-1023 (2006). 4. V. K. Ridaura et al., Gut microbiota from twins discordant for obesity modulate metabolism in mice. Science 341, 1241214 (2013). 5. J. K. Goodrich et al., Human genetics shape the gut microbiome. Cell 159, 789-799 (2014). 6. J. L. Kubinak, J. L. Round, Do antibodies select a healthy microbiota? Nature Rev. Immunol. 16, 767-774 (2016). 7. J. L. Kubinak et al., MyD88 signaling in T cells directs IgA-mediated control of the microbiota to promote health. Cell Host Microbe 17, 153-163 (2015). 8. F. Fransen et al., BALB/c and C57BL/6 mice differ in polyreactive iga abundance, which impacts the generation of antigen-specific iga and microbiota diversity. Immunity 43, 527-540 (2015). 9. J. J. Bunker et al., Innate and adaptive humoral responses coat distinct commensal bacteria with immunoglobulin A. Immunity 43, 541-553 (2015). 10. T. C. Cullender et al., Innate and adaptive immunity interact to quench microbiome flagellar motility in the gut. Cell Host Microbe 14, 571-581 (2013). 11. S. Wang et al., MyD88 adaptor-dependent microbial sensing by regulatory T cells promotes mucosal tolerance and enforces commensalism. Immunity 43, 289-303 (2015). 12. M. Feuerer et al., Lean, but not obese, fat is enriched for a unique population of regulatory T cells that affect metabolic parameters. Nature Med. 15, 930-939 (2009). 13. M. A. Exley, L. Hand, D. O'Shea, L. Lynch, Interplay between the immune system and adipose tissue in obesity. J. Endocrinol. 223, R41-48 (2014). 14. S. P. Weisberg et al., Obesity is associated with macrophage accumulation in adipose tissue. J. Clin. Invest. 112, 796-1808 (2003). 15. H. Luck et al., Regulation of obesity-related insulin resistance with gut anti-inflammatory agents. Cell Metabolism 21, 527-542 (2015). 124 16. D. Gevers et al., The treatment-naive microbiome in new-onset Crohn's disease. Cell Host Microbe 15, 382-392 (2014). 17. H. Bisgaard et al., Reduced diversity of the intestinal microbiota during infancy is associated with increased risk of allergic disease at school age. J. Allergy Clin. Immunol. 128, 646-652 e641-645 (2011). 18. E. Le Chatelier et al., Richness of human gut microbiome correlates with metabolic markers. Nature 500, 541-546 (2013). 19. P. J. Turnbaugh, F. Backhed, L. Fulton, J. I. Gordon, Diet-induced obesity is linked to marked but reversible alterations in the mouse distal gut microbiome. Cell Host Microbe 3, 213-223 (2008). 20. N. Segata et al., Metagenomic biomarker discovery and explanation. Genome Biology 12 (2011). 21. G. Gibson, J. H. Cummings, G. T. Macfarlane, Use of a three-stage continuous culture system to study the effect of mucin on dissimilatory sulfate reduction and methanogenesis by mixed populations of human gut bacteria. Appl. Environ. Microbiol. 54, 2750-2755 (1988). 22. M. C. Pitcher, J. H. Cummings, Hydrogen sulphide: a bacterial toxin in ulcerative colitis? Gut 39, 1-4 (1996). 23. F. E. Rey et al., Metabolic niche of a prominent sulfate-reducing human gut bacterium. Proc. Natl. Acad. Sci. U. S. A. 110, 13582-13587 (2013). 24. M. S. Desai et al., A dietary fiber-deprived gut microbiota degrades the colonic mucus barrier and enhances pathogen susceptibility. Cell 167, 1339-1353 e1321 (2016). 25. D. Berry, W. Reinisch, Intestinal microbiota: a source of novel biomarkers in inflammatory bowel diseases? Best. Pract. Res. Clin. Gastroenterol. 27, 47-58 (2013). 26. J. Loubinoux, J. P. Bronowicki, I. A. Pereira, J. L. Mougenel, A. E. Faou, Sulfate-reducing bacteria in human feces and their association with inflammatory bowel diseases. FEMS Microbiol. Ecol. 40, 107-112 (2002). 27. C. L. Karlsson et al., The microbiota of the gut in preschool children with normal and excessive body weight. Obesity 20, 2257-2261 (2012). 28. S. Xiao et al., A gut microbiota-targeted dietary intervention for amelioration of chronic inflammation underlying metabolic syndrome. FEMS Microbiol. Ecol. 87, 357-367 (2014). 125 29. C. Zhang et al., Structural resilience of the gut microbiota in adult mice under high-fat dietary perturbations. ISME J. 6, 1848-1857 (2012). 30. Y. Lu et al., Short chain fatty acids prevent high-fat-diet-induced obesity in mice by regulating g protein-coupled receptors and gut microbiota. Sci. Rep. 6, 37589 (2016). 31. J. Qin et al., A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature 490, 55-60 (2012). 32. J. D. Planer et al., Development of the gut microbiota and mucosal IgA esponses in twins and gnotobiotic mice. Nature 534, 263-266 (2016). 33. J. F. Bach, The effect of infections on susceptibility to autoimmune allergic diseases. N. Engl. J. Med. 347, 911-920 (2002). 34. T. Vatanen et al., Variation in microbiome lps immunogenicity contributes to autoimmunity in humans. Cell 165, 842-853 (2016). 35. L. M. Cox et al., Altering the intestinal microbiota during a critical developmental window has lasting metabolic consequences. Cell 158, 705-721 (2014). 36. M. C. Arrieta et al., Early infancy microbial and metabolic alterations affect risk of childhood asthma. Sci. Transl. Med. 7, 1-14 (2015). 37. N. Wang et al., Selective IgA deficiency in autoimmune diseases. Mol. Med. 17, 1383-1396 (2011). 38. R. Diaz Heijtz et al., Normal gut microbiota modulates brain development and behavior. Proc. Natl. Acad. Sci. U. S. A. 108, 3047-3052 (2011). 39. P. G. Bronson et al., Common variants at PVT1, ATG13-AMBRA1, AHI1 and CLEC16A are associated with selective IgA deficiency. Nature Genet. 48, 1425-1429 (2016). and 126 Appendix A Supplemental figures Figure 5.S1: Mice lacking MyD88 signaling within T cells develop obesity. A) Grams of weight gained as mice age, starting at 2 months of age (WT, N=8; TMyD88-/-, N=7). B) Percentage of fat gained as mice age, starting at 2 months of age (WT, N=8; T-MyD88-/-, N=7). C) Blood levels of glucose (milligram per decaliter) measured over time following i.p. glucose (1milligram per gram) injection during glucose tolerance test of 1 year old WT and T-MyD88-/- mice. D) VO2 of 2 month and 12 month mice (2 month N=3 per group; 12 month N=5 per group) E) VCO2 of 2 month and 12 month mice (2 month N=3 per group; 12 month N=5 per group) F) Energry Expenditure of of 2 month and 12 month mice (2 month N=3 per group; 12 month N=5 per group) G)Grams of food intake per mouse while being fed normal chow (2 month N=3 per group; 6 months N=4 per group; 12 months N=5 per group) H)Heat of 2 month and 12 month mice (2 month N=3 per group; 12 month N=5 per group) I) Movement of 2 month and 12 month mice (2 month N=3 per group; 12 month N=5 per group) J)Grams of weight gained of WT and T-MyD88-/- mice fed a control or HFD (WT CTRL, N=8; WT HFD, N=15; T-MyD88-/- CTRL, N=9; T-MyD88-/- HFD, N=13). K) Percentage of weight gained of WT and T-MyD88-/- mice fed a control or HFD (WT CTRL, N=8; WT HFD, N=15; T-MyD88-/- CTRL, N=9; T-MyD88-/- HFD, N=13). Statistics: P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using a repeated measures ANOVA (A, C, J, K), two-tailed, unpaired t-test (B,D-I). 127 B ** 20 30 * 20 10 2 4 6 Energy Expenditure * 25000 * 20000 * 1000 1000 5000 0 0 0 Day Night Day Day 2 m.o. 12 m.o. Night 10000 Day 2000 Night 15000 H I 0 2 6 1.0 0.8 0.6 0.4 0.2 0.0 12 Age (Months) 500 0 Day 1 Xamb+Yamb+Ztot 2 1000 Night 3 1500 20 15 10 5 Day p=0.07 Night 4 WT TMYD * ** Movement Heat kcal/hr grams/day 5 2 m.o. 12 m.o. 2 m.o. 12 m.o. Food Intake 2 m.o. 12 m.o. 2 m.o. 12 m.o. J Weight Gained (%) K TMYD HFD WT HFD TMYD CTRL WT CTRL 35 Weight (g) * p=0.13 p=0.10 Day ml/kg/hr * 3000 Night ml/kg/hr 5000 * 4000 * G 30 30 25 * * 20 15 10 0 4 8 12 Weeks on HFD 16 175 150 125 100 75 50 25 0 TMYD HFD WT HFD TMYD CTRL WT CTRL 0 4 8 60 Minutes VCO2 2000 0 F VO2 3000 0 10 Age (months) E 5000 * 4000 8 Night D 0 9 Day 5 6 7 8 Age (months) 100 Night 4 200 Night 3 WT T-MyD88-/- 300 Day 2 400 Day 0 ** ** WT T-MYD Night WT T-MYD 40 C 40 Glucose (mg/dL) 60 Fat (%) Weight (g) A ** 12 Weeks on HFD 16 90 120 128 B HF Diet T-MyD88-/WT 3.95% 14.5% CD4 CTRL Diet WT T-MyD88-/5.86% 4.75% 40 30 20 10 IL 4 Macrophages (%) F4/80 G at a3 0 HFD ns 2 1 b 11 2 cd L1 3 L2 C XC C C xP Fo g N IF B T- R O R et 0 gT Relative Expression 10 T-MYD ns * T-MYD * 20 WT * 30 CTRL ns HFD 40 T-MYD ns WT T-MyD88-/- 9.58% CTRL WT ns * 8 6 4 3 T-MyD88-/- WT WT 7.74% cd11b E D HF Diet T-MYD CTRL Diet WT T-MyD88-/8.89% 10.3% WT 0 FoxP3 C * 50 TREGS (%) A Figure 5.S2: Obese T-MyD88-/- mice have greater inflammation within adipose tissue. A) Representative flow cytometry plots were previously gated on CD3+CD4+ T cells within adipose tissue and measured for Tregs, defined as CD3+CD4+FoxP3+ cells. B) Percentage of Tregs within adipose tissue of WT and T-MyD88-/- mice fed a control or HFD (WT CTRL, N=8; WT HFD, N=15; TMyD88-/- CTRL, N=9; T-MyD88-/- HFD, N=13). C) Representative flow cytometry plots were previously gated on CD45+ cells within adipose tissue and measured for macrophages, defined as CD45+CD11b+F4/80+ cells. D) Percentages of macrophages within adipose tissue of WT and T-MyD88-/- mice fed a control or HFD (WT CTRL, N=8; WT HFD, N=15; T-MyD88-/- CTRL, N=9; T-MyD88-/- HFD, N=13). E) qPCR of mRNA expression of various T cell and myeloid cell immune parameters from WT and T-MyD88-/- mice fed a HFD (WT, N=6; T-MyD88-/-, N=6). Statistics: P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) twotailed, unpaired t-test (B,D-E) 129 Figure 5.S3: Dysbiosis within T-MyD88-/- mice is associated with spontaneous weight gain A) Grams of VAT within WT and T-MyD88-/- mice fed HFD with or without antibiotics (WT CTRL, N=5; TMYD CTRL, N=4; WT ABX, N=5, TMYD ABX, N=5). B) Representative flow cytometry plots were previously gated on CD3+CD4+ T cells within adipose tissue and measured for percentage of adipose Tregs, defined as CD3+CD4+FoxP3+ cells, within WT and T-MyD88-/mice fed HFD with or without antibiotics. C) Beta-diversity of ileal and fecal 16s sequencing samples from 1 year old WT and T-MyD88-/- mice, measured by unweighted unifrac and weighted unifrac (WT, N=8; T-MyD88-/-, N=7). D) Heat Map of biological pathways within fecal bacterial between one year old WT and T-MyD88-/- mice (N=6 per genotype). E) Cladogram of LEFSe analysis comparing one year old WT and T-MyD88-/- fecal 16s sequencing samples. F) Family level comparison of Turicibacteraceae relative abundance within fecal 16s sequencing samples of one year old WT and T-MyD88-/- mice G) Family level comparison of Clostridiaceae relative abundance within fecal 16s sequencing samples of one year old WT and T-MyD88-/- mice. H) Family level comparison of Lachnospiraceae relative abundance within fecal 16s sequencing samples of one year old WT and T-MyD88-/- mice. I) Family level comparison of Peptostreptococcaceae relative abundance within fecal 16s sequencing samples of one year old WT and T-MyD88-/- mice. J) Mapped reads per million of significantly different species from WT and T-MyD88-/- fecal transcripts (N=6 per genotype). K) Mapped reads per million of significantly different species from WT and T-MyD88-/- ileal transcripts. Statistics: P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) using a two-tailed, unpaired t-test (A,B,J,K), permanova (C), Mann-Whitney U test (F-I). 130 2.5 * ns ABX WT TMYD WT 1.5 1.0 CD4 0.5 WT TMYD WT TMYD 0.0 CTRL ** 40 TMYD 30 20 10 0 FoxP3 ABX WT TMYD WT TMYD VAT (g) 2.0 CTRL B ** Tregs (%) A D CTRL ABX C Unweighted Fecal Weighted Fecal PC2 (23.3%) PC2 (16.95%) 0.2 0.0 -0.2 p=0.01 -0.2 -0.4 -0.6 -0.4 -0.2 0.0 0.2 0.4 PC1 (22.1%) PC1 (40.7%) Unweighted Ileal Weighted Ileal 0.6 PC2 (16.7%) PC2 (14.7%) 0.0 p=0.11 -0.4 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.2 0.2 0.0 -0.2 p=0.02 -0.4 -0.4 -0.2 0.0 0.2 0.4 0.2 0.0 -0.2 -0.4 -1.0 -0.5 0.4 PC1 (30.0%) p=0.03 0.0 0.5 1.0 PC1 (42.3%) E WT 0.005 Bacteroides 0.01 0.00 0.020 0.10 0.015 0.010 0.05 0.005 0.00 0.000 WT TMyD88 ** 4000 ** ** Clostridium sp * C. sordelli C. ramosum C. perfringens ** C. mangenotii * C. cocleatum C. bifermentans ** C. ochlearium *** Bacillus sp 0 ** *** 400 200 **** E gallinarum *** C sporogenes * 2000 *** 0.025 6000 A. bizertensis P. goldsteinii C. cocleatum * K 0.02 Peptostreptococcaceae * Erysipelotrichaceae sp Clostridia * C. cochlearium ** C. perfringens ** C. sporogenes ** TMyD88 * A. bizertensis * C. sordelli 20 15 10 5 0 WT Mapped Reads per Million 400 300 200 100 Clostridium sp Mapped Reads per Million 0.000 0.03 I ** T. glycolicus 0.010 0.04 Lachnospiraceae p=0.07 0.15 S. xylosus 0.015 Relative Abundance Relative Abundance 0.020 J H Relative Abundance G Turicibacteraceae Clostridiaceae ** ** 0.025 0.05 Relative Abundance F 131 Figure 5.S4: T cell shaping of the microbiota is associated with spontaneous weight gain. A) Weight gained in germfree mice given WT or T-MyD88-/microbiota through multiple methods of transfer (CF=cross fostered). B) Flow cytometry was used to quantify the percentage of IgA bound bacteria within TCRKO mice given WT or T-MyD88-/- cells at Day 0, Week 1, and Week 8 (N=6 per cohort). C) Flow cytometry was used to quantify the percentage of IgG1 bound bacteria within TCRKO mice given WT or T-MyD88-/- cells at Day 0, Week 1, and Week 8. D) Flow cytometry was used to quantify the percentage of IgG3 bound bacteria within TCRKO mice given WT or T-MyD88-/- cells at Day 0, Week 1, and Week 8. E) Concentration of luminal IgA (µg/mL) within was measured TCRKO mice given WT or T-MyD88-/- cells after 8 weeks using an ELISA. F) Representative flow cytometry plot was previously gaited on SyBRGreen+ cells in order to quantify the percentage of IgG3 bound bacteria within TCRKO mice given WT or T-MyD88-/- cells after 8 weeks. G) Relative abundance of Clostridiaceae within fecal samples from TCRKO mice given WT and T-MYD88 Day 0 and after 4 weeksStatistics: P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) twotailed, unpaired t-test (A-F) and a Mann-Whitney U test (G). 132 B 1.0 Young GF B6 Gavage 30 p=0.08 1 Weeks Relative Abundance G Clostridiaceae * 0.05 0.04 0.03 0.02 0.01 0.00 WTTMYDWTTMYD D0 Week4 1 Weeks 8 F 8 800 600 400 200 IgG3 0 * 6 4 2 0 0 1 Weeks F 1000 WT T-MyD88-/- 0 0 E 20 0 0 8 IgG1 WT T-MyD88-/- CF IgG3 10 20 IgA (µg/mL) IgG3 bound (%) D * 10 IgG3 bound (%) Young GF SW 0.0 WT T-MyD88-/- 40 Young GF B6 0.5 60 W TT M YD 88 -/- 1.5 IgA bound (%) WT TMYD Adult GF B6 Weight gain normalized to WT 2.0 C IgA IgG1 bound (%) A 40 30 20 10 0 ns 8 133 B A ns Separated T-MYD mice Cohoused mice WT mice Separated WT mice C ** *** 50 40 30 20 Cohoused 5 ns p=.07 Relative Expression 30 20 10 4 3 ** * nsns ns * 2 nsns *** ns ** ns ** ns * ns 11 b D 2 L1 XC C L2 C C C 3 xP Fo IL 4 a3 at G gT R O R IF N g 0 Cohoused ET WT T-MyD WT 1 T-MyD 0 ns Separated WT Cohoused WT Separated T-MYD Cohoused T-MYD TB 40 p=0.06 D ns Tregs (%) 60 Total Weight (g) T-MYD mice Figure 5.S5: Dysbiosis within T-MyD88-/- mice transfers obesity to WT animals. A) Schematic of cohousing experiment. B) Total grams of weight gained by separated or cohoused WT and T-MyD88-/- mice fed a HFD (N=4 per cohort). C) Percentage of adipose Tregs within WT and T-MyD88-/- mice fed a HFD D) qPCR of mRNA expression of various T cell and myeloid cell immune parameters from separated or cohoused WT and T-MyD88-/- mice fed a HFD. Statistics: P-value<0.05 (*); P-value<0.01 (**); P-value<0.001 (***) two-tailed, unpaired t-test (B-D) 134 A B Figure 5.S6: Expansion of Desulfovibrio and reductions in beneficial sporeforming microbes are associated with obesity. A) Relative abundance of Bifidobacterium pseudolongum within fecal 16s sequencing samples from separated or cohoused WT and T-MyD88-/- mice at the final timepoint (N=4 per cohort). B) Relative abundance of Lactobacillales within fecal 16s sequencing samples from separated or cohoused WT and T-MyD88-/- mice at the final timepoint (N=4 per cohort). Statistics: P-value<0.05 (*); P-value<0.01 (**); Pvalue<0.001 (***) using a Mann-Whitney U test (A, B) CHAPTER 6 DISCUSSION 136 Here we have discovered a mechanism by which recognition of microbial products by T cells directs IgA targeting of commensals to maintain healthy microbiota composition. T cells lacking MyD88 are defective in their ability to differentiate into Tfh cells to promote GC responses during steady-state conditions within the gut as well as immunogenic challenge of the systemic compartment. The resulting loss of GC responses leads to defective T celldependent IgA targeting, and animals containing a specific deletion of MyD88 within T cells develop dysbiosis that enhances susceptibility to colitis and drives the onset of obesity and insulin resistance. These diseases are rescued by either providing an FMT from a healthy WT donor or specifically reintroducing beneficial microbes that are capable of protect against obesity. We have thus uncovered a novel mechanism by which dysbiosis and disease develop, providing a new avenue in which potential preventative and therapeutic methods can be developed. While this dissertation makes great strides in understanding T celldependent MyD88 signaling with regards to its role in shaping healthy microbiota communities, much work needs to be done to appreciate the mechanism by which T cells use innate signals to promote TFH responses, as well as how dysbiotic communities arise in mice lacking MyD88 signaling in T cells. Additionally, indentifying the mechanism by which these communities promote both obesity and colitis will lead to novel therapeutic targets in order to treat these diseases. Future studies are required to identify the mechanism by which T cells 137 use MyD88 signaling to promote TFH differentiation preliminary evidence from this thesis has identified defects within proliferation, ICOS signaling, and maintenance of GC B cell responses; however, the extent by which these pathways are working in conjunction or independently of each other is unknown. Additionally, while much of the work within this thesis looked at shifts within gene and protein expression, MyD88 has been demonstrated to modulate metabolic pathways within T cells. It should therefore be considered whether or not TFH cells require MyD88 signaling to maintain proper metabolic function. Lastly, this thesis has demonstrated a common defect within both splenic and gut TFH development when T cells lack MyD88 signaling. The types of germinal center responses that occur within the splenic compartment and PPs within the gut differ within respects to antibody production as well as stimulus. PPs maintain a constant germinal center response that results in IgA, while spleens have very little TFH differentiation and germinal center responses in the absence of infection. Furthermore, during pathogen invasion, antibody production results primarily in an IgG response. Therefore, it is likely that the TFH stimuli and differentiation vary between these two sights. Future studies to characterize site specific immune responses using the model antigen, OVA, or other antigens may reveal alternate roles for MyD88 signaling that could explain some of the inconsistencies identified within the results of this thesis. Mice lacking MyD88 signaling in T cells develop dysbiosis that predisposes them to both colitis and obesity. We have identified a correlative loss of IgA and IgA targeting of specific microbes with expansion of the pathobiont, 138 Desulfovibrio, the loss of beneficial Clostridia, and the reduction of diversity. A large body of literature supports the requirement of secretory IgA within the gut, however, it is has remained difficult to definitively test whether defects within T cell-dependent IgA signaling within the gut are the sole impetus for the dysbiosis that we observe in T-MyD88-/- mice. Future experiments using IgA deficient mice as well as BCL6 deficient animals, will provide additional insight, however, compensation by other immune responses within these animals may provide additional confounders. Regardless of this, the dysbiosis observed within these animals resembles a number of microbiota communities found within obese patients and individuals with type II diabetes. The parallels between our mouse model and humans provides us with a unique opportunity to understand intermicrobe interactions and microbe-derived metabolites that contribute to both dysbiosis and metabolic syndrome. This can be achieved through two independent lines of experiments: (1) Utilizing the germfree facility, we can understand how co-colonization of microbes alters both the abundance and genetic expression of both commensals and pathobionts. Microbial communities are filled with overlapping niches and lead to both antagonistic and supportive interactions between commensals. The identification that dysbiosis reduces the abundance of Clostridia members points to a potential therapeutic target. if we can elucidate how Desulfovibrio promotes the loss of Clostridia, we can design therapies to restore the beneficial microbe; (2) Through metabolomics analysis, we can identify shifts within gut metabolites that promote weight gain. Commensals influence weight gain through both microbial-derived metabolites 139 and the alteration of host-derived metabolites. It is likely that one of these pathways is leading to the dyslipidemia identified in the T-MyD88-/- mice. Considering the parallels that the dysbiosis in these mice draws with clinical studies, it is likely that identifying which metabolic shifts within the microbiota drive disease will lead to a therapeutic target in the treatment of both obesity and diabetes. Harnessing the tools within our germfree facility, the metabolomics core, and meta-transcriptomics will provide valuable knowledge on the development and perpetuation of metabolic syndrome. The increase of allergic, autoimmune, and metabolic disorders in western society has coincided with shifting social practices involving antibiotics use, dietary behaviors, and hygiene (1). These alterations to our behavior and environment have lasting effects on our microbiota (2-4). Because of this, it has been postulated that a combination of these effects has led to a broad extinction of beneficial microbes within our society, resulting in reduced microbiota diversity, and a greater susceptibility to disease pathology (5). Despite this rise in disorders, however, many individuals experience no clinical consequences to a westernized lifestyle. Given these discrepancies, it is accepted within the scientific community that environmental and genetic factors likely interact to promote or protect against these complex diseases. Directed IgA targeting of the microbiota may be one mechanism by which host immune responses predispose individuals to a greater susceptibility toward disease onset. In agreement with this notion, studies have revealed that roughly 1 in 600 people is IgA-deficient, making it the most common primary immunodeficiency (6). As many as 30-50% of these individuals 140 experience recurrent mucosal infections or develop autoimmune diseases (6, 7). This is a dramatic increase in disease susceptibility over individuals with healthy levels of IgA. While many individuals are never diagnosed with IgA deficiency, future studies that screen for this disorder and compare their susceptibility to other diseases might reveal additional links between IgA deficiency and a westernized lifestyle. This dissertation has identified a novel component of a pathway that has co-evolved to maintain symbiosis between mammals and the microbiota. Studies such as this provide unique opportunities to understand intricacies of immune development while gaining insight into the impact of specific commensals on the metabolic and compositional landscape of the microbiota. While the microbiota as a whole has evolved a vast network of species and redundant metabolic functions, identifying the underlying mechanisms of disease associated perturbations reveals keystone interactions that are necessary to push the balances in favor of health. References 1. J. F. Bach, The effect of infections on susceptibility to autoimmune and allergic diseases. N. Engl. J. Med. 347, 911-920 (2002). 2. T. Vatanen et al., Variation in microbiome LPS immunogenicity contributes to autoimmunity in humans. Cell 165, 842-853 (2016). 3. L. M. Cox et al., Altering the intestinal microbiota during a critical developmental window has lasting metabolic consequences. Cell 158, 705-721 (2014). 4. M. C. Arrieta et al., Early infancy microbial and metabolic alterations affect risk of childhood asthma. Sci. Transl. Med. 7, 1-14 (2015). 141 5. P. Hunter, Where next for antibiotics? The immune system and the nature of pathogenicity are providing vital clues in the fight against antibioticresistant bacteria. EMBO Rep. 13, 680-683 (2012). 6. N. Wang et al., Selective IgA deficiency in autoimmune diseases. Mol. Med. 17, 1383-1396 (2011). 7. P. G. Bronson et al., Common variants at PVT1, ATG13-AMBRA1, AHI1 and CLEC16A are associated with selective IgA deficiency. Nature Genet. 48, 1425-1429 (2016).