CCFA Microbial–Host Interactions Workshop: Highlights and key observations

Abstract

This article provides a summary of the proceedings of the CCFA Microbial–Host Interactions Workshop that was held in St. Petersburg, Florida, on March 16–19, 2006. Approximately 75 senior and junior investigators from around the world shared their most current research findings through oral presentations, poster sessions, and active discussion. Because intestinal microbiota are significant contributors in the development of inflammatory bowel diseases (IBD), understanding the body's responses to and interactions with microbes, especially in the colon and distal small intestine, is critical to elucidating the etiology and pathogenesis of IBD and developing effective therapeutic interventions. Major advances have occurred recently in molecular detection of luminal bacterial species, identifying dominant microbial antigens that drive intestinal inflammation, the mechanisms of innate immune and epithelial responses to bacteria, and regulation of inflammation by innate and acquired immune cells.

(Inflamm Bowel Dis 2007)

We coexist with an extremely complex intestinal microbiota that outnumber our mammalian cells 10-fold (Fig. 1). These commensal bacterial and fungal species contain vast quantities of adjuvants and antigens that induce both protective and pathogenic innate and acquired immune responses that determine mucosal homeostasis versus inflammation. Moreover, these metabolically active microbial constituents metabolize ingested foods, drugs, and secreted epithelial components to produce short chain fatty acids, carcinogens, and other diverse products that regulate the expression of mammalian mucosal genes and provide substrates for epithelial cell energy production. The intestinal “microbiome” remains poorly characterized with respect to the identity of its components, since the majority of anaerobic bacterial species cannot be cultivated. As a consequence, the genome, dominant antigens, metabolic enzymes, and products of the enteric microbiota remain relatively unknown. These bacterial and fungal species interact not only with each other but also with the intestinal mucosa with which they are in intimate contact (Fig. 2). Thus, the commensal microbiota are dominant forces in the development and regulation of epithelial cell functions and innate and acquired mucosal immune responses.

Figure 1. Human gastrointestinal bacteria: the interface between the host and the environment. The luminal concentrations of the gastrointestinal microbiota are provided with the dominant identified species in each region.

Figure 2. A scanning electron micrograph of mucosally adherent bacteria at the mucosal surface of the intestine that shows the intimate association of the epithelial cells (upper left) and the physical interaction of bacteria in the biofilm.

Dramatic advances in the basic understanding of mucosal microbial ecology, pathogenesis, dominant antigens, bacterial–epithelial interactions, innate immune and immune regulatory responses to commensal microbes have occurred in the past few years. A dominant stimulus to this expanded research was a funding initiative to identify microbial antigens and adjuvants that induce innate and adaptive intestinal immune responses sponsored by the Crohn's & Colitis Foundation of America (CCFA) in May 2003. To further stimulate research in this area, the CCFA organized a Microbial–Host Interactions Workshop that was held March 16–19, 2006, in St. Petersburg, Florida. The goals of the workshop are outlined in Table 1. The meeting was designed to facilitate interaction and discussion between the 33 invited faculty, 15 discussants, and 29 poster presenters. The present workshop report is designed to disseminate the scientific highlights of this highly successful symposium to the research and clinical communities in a succinct manner. We hope that this synopsis of an extremely productive meeting will further stimulate research and understanding of a rapidly advancing scientific area that is vital to understanding the pathogenesis and development of therapies of inflammatory bowel diseases (IBD).

Table 1. Workshop Objectives

Review and objectively evaluate current information regarding enteric microbial antigens and mucosal host responses Facilitate interactions among investigators and between trainees and established investigators Bring new investigators with novel concepts and techniques into the field Provide updates on CCFA-funded research projects Identify priorities for future research funding

MICROBIAL ECOLOGY AND HOST INTERACTIONS

The introductory session provided a conceptual framework for the workshop by addressing the mechanisms by which microbial populations interact with each other and with the mucosal surface as well as some of the modern molecular tools that can be used to identify individual components of complex microenvironments. Speakers brought valuable insights from other disciplines that are more advanced in detecting microbial diversity within complex ecosystems. Molecular detection methods are required because of the disparity between cultivatable and total observed bacteria and fungi in the contents of the distal intestine in both health and disease. Emphasis was placed on understanding the composition of commensal bacteria of healthy hosts and their ability to form biofilms as a prelude to identifying specific alterations during disease.

Microbial Biofilms

Microbes associated with man—both in health and disease—often exist as biofilms. Michael S. Gilmore, PhD (Harvard Medical School and The Schepens Eye Research Institute, Boston, MA) explained that the definition of a biofilm has evolved with the discovery of new mechanisms and strategies employed by microbes to persist and flourish in diverse environmental and mammalian ecologies. Traditionally, biofilms were viewed as microbes encased in a secreted carbohydrate capsular material expressed by the microbe that mediates surface attachment. Classic examples of biofilms are Streptococcus mutans adherent to the coronal surface of a tooth or the attachment of Pseudomonas aeruginosa to the interior surface of a water line. This restrictive definition would not, however, apply to gastrointestinal microbes that are largely nonadherent. Therefore, biofilms are more accurately defined by the complex multicellular coordinated behaviors that result in the formation and persistence of microbial populations.

The life cycle of a bacterial biofilm can be roughly divided into three steps: 1) attachment, 2) growth, and 3) detachment. Bacteria with high affinity for a substratum produce a biofilm by initially adhering to a surface, followed by growth. Alternatively, those bacteria with high affinity for other bacteria will first aggregate and then adhere to a surface. Aggregation facilitates communication between bacteria and accelerates biofilm formation. Depending on the species, bacteria utilize a variety of tools (e.g., flagella, pili, or surface proteins) to form a microcolony. As the microcolony grows, a quorum is reached that responds to accumulated signaling peptides (e.g., acyl-homoserine lactones). The mature microcolony becomes heterogeneous with microniches that differ in oxygenation, pH, nutrient accessibility, growth patterns, and gene expression. Upon microcolony maturation, the center of the biofilm begins to liquefy due to apoptosis of the constituent bacteria, with emergence of a sustainable architecture community. As the biofilm structure degrades, fragmentation occurs, with potential seeding of distant sites.

Selection pressure for microbes to behave as coordinated communities have led to the evolution of very complex behaviors that enhance survival in harsh ecosystems. For example, some species of bacteria, such as Streptococcus pneumoniae, are hyperautolytic, without a clear explanation for why apparent mass suicide within a population is selectively advantageous. Some bacterial species possess mechanisms for the predatory induction of death and lysis of siblings to provide nutrients for survivors as well as intracellularly derived materials for matrix formation, including DNA, that presumably contribute to the protection and attachment of survivors. Understanding of the differentiated behaviors exhibited by microbial communities has nurtured the view that a complex consortium, such as the gastrointestinal microbial community, consists of extensive communications and interdependencies. Moreover, since microbial ecologies exist in niches within higher organisms, communications and interdependencies extend to mutualistic interactions between microbes and hosts.

Periodontal Microecology: Lessons for the Gut

Currently, the optimal method for exploring microbial diversity is based on isolating and analyzing DNA from the target environment. One of the best, and certainly the most easily accessible, target environments in humans is the oral cavity. Using periodontal disease as an example, Bruce J. Paster, PhD (Forsythe Institute and Harvard School of Dental Medicine, Boston, MA) described the development of the Human Oral Microbe Identification Microarray slide system that allows detection of hundreds of bacterial species in the oral cavity. Identification of microbial species depends on isolating DNA, amplifying of 16S rRNA by polymerase chain reaction (PCR), cloning of the amplicons, and sequencing the cloned rRNA insert. Molecular analysis of 16S rRNA from the oral cavity has identified over 700 oral bacterial species, many of which cannot be cultivated (“uncultivables”). Screening of multiple patients has identified bacterial species that are subject-specific and detected in most oral sites, while other bacterial species are site-specific, i.e., found only in subgingival plaque and not in other oral sites. For example, the microbial diversity of the tonsils, with several dozen phylotypes, differs vastly from the spectrum of organisms observed in the labial gingival, where 15 phylotypes were identified. Most importantly, several specific bacterial species, including some uncultivables, have been implicated in health and oral diseases. In a study of patients with periodontitis, in addition to more frequent detection of the three classic bacterial species usually associated with inflammation, i.e., Tannerella forsythia, Porphyromonas gingivalis, and Treponema denticola, there was increased frequency of 2 additional novel clones, Porphyromonas clone AU126 and Synergistes clone 84. Dr. Paster suggested that a similar microarray system could be developed for the lower intestinal tract with the goal of identifying bacterial species that dominate different regions of the normal intestine and those which may contribute to IBD and other intestinal disorders.

Molecular Detection of Commensals

The microbiota that inhabits the human gut appears to be an important source of antigens that drive the chronic immunological processes characteristic of IBD. Gerald W. Tannock, PhD (University of Otago, Dunedin, New Zealand) addressed the question of whether a subset of enteric microbial inhabitants serves as opportunistic IBD pathogens or as the source of dominant antigens that drive pathogenic immune responses, or whether the specific phylogeny of the microbial community is irrelevant. He noted that although most members of the intestinal microbiota have not yet been cultured, nucleic acid-based methods of detection and enumeration can provide important information about the composition of the enteric bacterial community. His presentation focused on detection methods targeting the nucleotide base sequence of the small 16S ribosomal subunit RNA or the gene that encodes it.

Variable regions within 16S rRNA contain the taxonomic signatures of bacterial groups or species, while consensus flanking sequences permit expansion of the entire region by PCR using universal primers. Nucleic acid-based profiling methods utilize either a chemical or temperature gradient (DGGE and TTGE, respectively) of 16S or rRNA PCR products that provide a snapshot of the composition of bacterial communities. Results from studies of these methods were presented, along with fluorescent in situ hybridization (FISH) and flow cytometry (FC) with respect to pouchitis in humans, real-time quantitative PCR and biopsy-associated bacteria, and analysis of 16S rRNA gene sequence libraries in relation to Crohn's disease (CD) and ulcerative colitis (UC).

DGGE studies in experimental colitis in IL-10-deficient mice have led to identification and isolation of a bacterial species, Bifidobacterium animalis, which was subsequently shown to cause duodenal and colonic inflammation in monoassociated IL-10^−/− mice. TTGE studies of stool profiles from UC patients with chronic or recurrent pouchitis as well as studies of patients with familial adenomatous polyposis have provided profiles that were specific for each disease. FISH/FC have identified a phylogenetic gap of about 25% of fecal bacterial species in the general community, and alteration of the gap following antibiotic therapy mirrors clinical response among these patients. Dr. Tannock speculated that these observations will permit targeting of specific bacteriological abnormalities in IBD and thereby provide targets for therapeutic interventions.

PATHOGENS

The Mycobacterium avium paratuberculosis (MAP) Story

An important focus of the workshop was to update the status of research on M. avium paratuberculosis (MAP) as a potential etiologic agent in CD. Saleh A. Naser, PhD (University of Central Florida, Orlando, FL) discussed detection of this organism and the anti-MAP response in CD. A high frequency of CD patients with anti-MAP serum antibodies has been documented using a number of MAP antigens. Various techniques have been used to detect MAP in intestinal tissue and other sites, providing a wide range of outcomes. The preferred methodology for detection has been PCR, with a readout of Southern hybridization, nested PCR amplication, or real-time quantification. Dr. Naser reported that Southern analysis revealed 0%–14% positive in control specimens (UC or non-IBD) and 20%–85% positive in CD specimens. Higher positive rates were associated with surgical resection versus biopsy specimens, indicating that sampling is a limitation to successful detection either due to the sporadic distribution of organisms or its enrichment in deeper compartments of intestinal tissue not sampled by biopsy (i.e., submucosa, muscularis, serosa). Additional unpublished work on full-thickness samples revealed a similar detection rate using nested PCR or quantitative PCR as readouts, suggesting that these alternate methods have similar sensitivities.

A second approach involved MAP culture (particularly using 12 weeks under an optimized Mycobacterial Growth Indicator Tube [MGIT] technology) with nested PCR readout. Addressing about 25 patients each in CD and non-CD categories, approximately 80% and 50% positive rates were found for intestinal and blood specimens in CD, respectively, and 0%–22% positive rates in non-CD specimens. In-lab contamination was an unlikely factor in this study, as sequencing indicated that the isolates were independent. However, there was a loss of sensitivity with respect to detection, presumably due to factors affecting recovery and successful culture of viable organisms from these biologic specimens. Morphologic analysis of MGIT cultures suggested that the early outgrowth involved spheroplast MAP, which might be particularly sensitive to recovery factors. Examples of FISH analysis were also presented, which if validated for sensitivity, might provide a quantitative method to determine MAP levels in a fashion less affected by tissue handling (Fig. 3).

Figure 3. FISH detection of bacteria microbiota. Fluorescent in situ hybridization of a mixture of bacterial species using probes recognizing specific 16s ribosomal DNA sequences. Yellow: Mycobacterium avium complex; Green: Mycobacterium avium paratuberculosis; Red: Clostridium difficile; Purple: Escherichia coli; White: Listeria monocytogenes.

Dr. Naser reported that a 4-center, blinded study is underway to compare detection of blood MAP by MGIT culture and nested PCR, and seropositivity in the same set of 50 cases stratified for disease category, caspase activating and recruiting domain 15 (CARD15), and MAP seropositivity. Edgar C. Boedeker, MD (University of New Mexico, Albuquerque, NM) reported the results of a blinded, NIH-sponsored, multicenter study (Carol Tacket, Enteric Pathogenesis Research Unit, NIAID [National Institute of Allergy and Infectious Diseases, NIH, Bethesda, MD]; EPRU 99-01) testing the association of MAP with CD. The study included Noam Harpaz, MD (Mount Sinai School of Medicine, New York, NY), Saleh Naser, PhD (University of Central Florida), Carol Tacket, MD (University of Maryland, College Park, MD), Norman Pace, PhD (University of Colorado, Boulder, CO), and Dennis Lang, PhD/Claire Schmidt, PhD (NIAID). A common set of serial surgical specimens were procured and validated (by pathology and chart review) at Mount Sinai School of Medicine. These specimens were sent in parallel for culture at 2 sites (University of Central Florida and University of Maryland) and for 16S PCR assessment (University of Colorado). Culture technologies included both BACTEC 12B (conventional) or MGIT conditions. Power analysis indicated that 40 specimens would be sufficient for discrimination (80% power) of CD versus non-CD, even at the relatively unfavorable conditions of 5% and 40% MAP+ rate in normal and CD, respectively. The results were surprising and important. First, culturable MAP was detected at substantial rates regardless of diagnosis: CD (35%–58%); UC (45%–55%); and non-IBD (22%–40%). Second, there was no concordance between laboratories regarding MAP+ samples (less than a quarter of samples positive in 1 site were detected as positive in the other site). Third, MAP was below the level of detection by standard (not nested) 16S PCR.

These findings indicate that methodologic limitations currently prevent reliable ascertainment of culturable MAP organisms in individual specimens. These data also indicate that on a population basis, culturable MAP is present at substantial and similar levels in both normal and IBD intestinal tissues. Accordingly, these results refute the idea that culturable MAP is sufficient for IBD. Clearly, new methodologies for reliable MAP detection are needed to readdress these issues.

Thomas J. Borody, MD, PhD (Center for Digestive Diseases, Sydney, Australia) discussed the status of mycobacterial antibiotic therapy in CD. The previously reported experience included about 15 studies with chemotherapy designed to treat conventional Mycobacterium tuberculosis, and three reports using single or combination antibiotic therapy for atypical mycobacterial species. These studies reported >65% response rates, but were not prospective or control-compared. Dr. Borody presented an unpublished Phase III multicenter, randomized, prospective, placebo-controlled, double-blind trial in active CD patients (CDAI >200, mean 290; >100 patients/treatment arm, with similar patient characteristics). Patients were treated with tapering prednisolone for 16 weeks and either combination antibiotic therapy (rifabutin 450 mg/d, clarithromycin 750 mg/d, clofazamine 50 mg/d) or placebo for 104 weeks and followed to Week 156. Relapse rates were significantly lower for active antibiotic therapy versus placebo (≈0.5 odds ratio) at monitored points through 104 weeks, and then became nonsignificant after a period of nontreatment (156 weeks). Relapse rates after 2 years of therapy were ≈65% with antibiotics and 78% with placebo and increased to ≈80% in both groups with an additional year of follow-up after therapy ended.

These findings indicate that such antibiotic treatment produces partial but significant improvement in CD. Unlike in Helicobacter pylori therapy, however, remission is not preserved after cessation of antibiotic therapy. Dr. Borody mentioned that the study was limited by possible antibiotic underdosing, including issues with clofazamine administration. Observational uncontrolled experience with 54 CD patients using revised antibiotic dosing, which included substantial and persistent reduction of CDAI and examples of resolved clinical complications (such as perianal fistula), was also presented. A prospective controlled trial is currently being organized to test this strategy.

Discussion at the session emphasized how antibiotic therapy could address the MAP theory of CD. The difficulties surrounding reliable MAP detection presently limit the ability to demonstrate an effect of antibiotic therapy on MAP levels. Furthermore, it is uncertain whether current antibiotic combinations effectively treat intracellular MAP. Accordingly, it is uncertain how to design studies to relate antibiotic therapy and clinical state to reduction or eradication of MAP. In the absence of such information, effective antibiotic therapy may target commensal and/or pathogenic enteric organisms other than MAP that contribute to disease activity. However, regardless of the mechanism, the prospect of highly effective antibiotic therapy in CD is both clinically and mechanistically important, particularly if remission can be maintained following cessation of treatment.

Michael Collins, DVM, PhD (University of Wisconsin, Madison, WI) presented an environmental overview of MAP (Fig. 4). This organism is a soil resident that causes endemic disease in wild and domestic ruminants with probably neonatal fecal–oral transmission from infected mothers. The onset of clinical disease in dairy cows occurs following an average incubation period of 5 years. Studies in Australia and the United States indicate that MAP+ cattle comprise 11% of dairy herds and 0.5% of beef cattle herds. Although MAP+ cattle infrequently have clinically recognized disease, current regulations and practices do not exclude MAP+ cattle from dairy or beef production. Notably, culled dairy cattle account for ≈35% of hamburger beef production. Dr. Collins explained that in most cases MAP is restricted to the intestine and associated lymph nodes, so milk and beef products would be expected to have limited MAP levels. However, culturable MAP is detected in about 2% of retail milk in the United States and Europe. This may in part reflect the organism's intracellular location within macrophages and its relative heat resistance to pasteurization. The levels of MAP in retail beef before or after cooking were not presented.

Figure 4. International dispersal of Mycobacterium avium paratuberculosis in agricultural herds. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Environmental exposure of humans to MAP was discussed, noting its detectable levels in soil and water, and the possible impact of domestic and wild ruminants on MAP levels on these environmental reservoirs. Going forward, the factual questions to establish are at what level are humans colonized with MAP, and when colonization occurs, to determine by molecular fingerprinting methodologies those reservoirs that contribute to such colonization. Finally, the foregoing presentations indicated that MAP intestinal colonization may be common in the human population (≈50% of normal individuals). A lively discussion weighed epidemiologic and/or Crohn's mechanism studies to resolve the questions of if and how MAP impacts human health. Also, MAP abatement measures are being considered in Europe and the United States. This policy debate, while focusing on the possible role of MAP in human health, might be more immediately guided by the defined impact on herd health. Finally, the session's discussion reflected interest in the challenge of identifying animal models in which important issues related to MAP pathogenesis and treatment might be addressed: minimum infectious units; the relative colonization and inflammatory responses elicited by divergent MAP strains; the genetics of MAP virulence; and optimal antibiotic regimens to cure Johne's disease.

Virulence Traits of Commensal Bacterial Species in CD

The session turned to other microbial species also distinguished for biologic traits relevant to CD. Arlette Darfeuille-Michaud, PhD (Université d'Auvergne, Clermont-Ferrand, France) reviewed adherent-invasive Escherichia coli (AIEC) in CD. Among mucosa-isolated organisms, E. coli is distinguished by an increased abundance of isolates from CD showing adherence to human intestinal cell lines (85% from CD tissue versus 40% from normal tissue). These isolates are prevalent in ileal but not colonic tissues. Conversely, AIEC strains more efficiently adhere to native enterocytes from ileum versus colon, and CD versus normal ileum. Adhesion correlated with CEACAM6 expression (a molecule upregulated by inflammation) in these different enterocyte populations and was shown to be required for AEIC-enterocyte interaction. Like various enteropathogenic E. coli, Dr. Darfeuille-Michaud explained that AEIC are invasive in intestinal and macrophage cell lines, but are distinguished by their prolonged intracellular survival and growth due to their capacity to escape endocytic residence and enter a cytoplasmic compartment. Like Listeria, this process involves microtubule recruitment and actin polymerization. Genetic analysis of a model strain, LF82, has uncovered a series of virulence traits required for the adherent-invasive phenotype (Fig. 5).

Figure 5. Virulence traits of an adherent-invasive E. coli strain recovered from the ileum of a patient with postoperative recurrence of Crohn's disease.

Enterocyte or macrophage interaction with AIEC triggers a molecular inflammatory response that selectively involves TNF activation. E. coli is strongly associated with microdissected granulomata in CD and is the etiologic agent for canine granulomatous colitis and leptomeningitis and avian Hjarre's disease. This area of inquiry in E. coli pathogenesis highlights the many traits of strains like AIEC that may cause its pathogenic effects in CD-susceptible individuals.

Cynthia L. Sears, MD (Johns Hopkins University School of Medicine, Baltimore, MD) presented information relating enterotoxigenic Bacteroides fragilis (ETBF) to chronic colonic inflammation and colorectal neoplasia. B. fragilis is among the most abundant species of colonic microbiota and is detectable in more than 40% of adults. Prevalence of ETBF increases from about 1% in infants to about 12% at age 20 years; its prevalence increases to about 40% in individuals with active IBD or colorectal cancer. The only known toxin associated with these strains is B. fragilis toxin (BFT), a zinc-dependent metalloprotease that stimulates enterocyte proliferation and IL-8 production via a mechanism independent of toll-like receptors (TLRs). In mice colonized with ETBF, chronic colitis and colonic epithelial hyperplasia developed over weeks to months.

Dr. Sears explained that strain background contributed to disease heterogeneity, with greater responses in C57BL/6 versus 129/Sv, synergy with DSS or the APC^min mutation, and a severe acute colitis in germ-free mice. Genetic analysis demonstrated that these responses were all dependent on expression of an active BFT protein. In view of these findings, it is conceivable that a human CD phenotype could be elicited among individuals colonized with ETBF and possessing host traits that facilitate chronic inflammatory responses to this organism and its virulence factor.

In summary, this session provided several important insights and challenges ahead for establishing the etiologic agents in CD. First, pathogens implicated in CD pathogenesis seem to commonly colonize the human intestine, even in non-IBD (<10% of healthy individuals). Since the prevalence of IBD in North America and Western Europe is ≈0.5%, no more than 5% of colonized individuals express a disease phenotype. This finding emphasizes the contribution of other environmental or genetic host factors in addition to candidate etiologic bacteria in the disease biology. Second, the specific virulence traits pertinent to the CD phenotype in the candidate microbial species are largely unknown. Progress on this issue will clearly advance when tissue culture and animal models for enterocyte or immunologic virulence are established. Third, on a population basis, CD patients are more likely to harbor each of the candidate organisms, with biologic and epidemiologic claims supporting their disease association. Thus, dissecting the responsible organism(s) for disease phenotype is challenging on a correlation basis. Fourth, the methodologic difficulties for detecting viable organisms, particularly MAP, are barriers to interpreting the relationship between specific antimicrobial efficacy and IBD therapy. However, such targeted antimicrobial therapy is promising, and optimization of the type of treatment and duration will greatly benefit from quantitative readout of target bacteria levels during therapy.

IDENTIFICATION OF DOMINANT ANTIGENS / COMMENSAL BACTERIA SPECIES

Studies of humoral (immunoglobulins) and T-cell (T-cell receptor repertoire) responses in IBD have suggested that there are a limited number of bacterial antigens that drive colitis. Specifically, both the B-cell and T-cell repertoires in a variety of animal models of colitis and humans with IBD display an oligoclonal response. This suggests that a limited number of bacterial species and, as a corollary, bacterial antigens, may be responsible for the development of disease. A significant amount of effort has gone into defining both the bacteria and their components that are responsible for the development of colitis.

Charles O. Elson, MD (University of Alabama at Birmingham, AL) began the workshop session with a review of work conducted in collaboration with Robert Hershberg, MD, PhD (Dendrion, Seattle, WA) and Stephan R. Targan, MD (Cedars-Sinai Medical Center, Los Angeles, CA) that led to the identification of flagellin as a dominant antigen in both mouse models and humans with CD (Fig. 6).

Figure 6. Flagellin. Left side: flagellated bacteria on macrophage; Right side: 3D structure of flagellin.

Specifically, utilizing a lambda phage DNA library from intestinal microbiota from C3H/HeJ Bir mice screened with sera from a colitic mouse, this group has identified flagellin as a major antigen recognized by the sera. Flagellin is a unique antigen because it binds to TLR-5 and therefore activates innate immune cells in addition to its ability to drive adaptive immune responses. Anti-flagellin responses have been described not only in C3H/HeJ Bir mice but also in MDR1a and IL-10-deficient animal models. Adoptive transfer of a flagellin-specific T-cell line to SCID mice induces colitis. Flagellins from Clostridia seem to be most antigenic in comparison to flagellins from other organisms. This group has now mapped the T-cell epitope from amino acids 406 to 415 in C3H mice and 452 to 468 in C57BL/6 mice.

The relevance of this strong serologic and T-cell-specific response to flagellins has now been extended to humans in work presented by Dr. Targan. Serologic reactivity to flagellin has been observed by this group in ≈50% of CD patients, but UC patients have no difference from normal controls. Together with previous responses to other bacterial antigens, such as the OmpC protein of E. coli, Pseudomonas fluorescens I2, and oligomannin, CD patients can be stratified into distinct clusters. The levels of reactivity to these various antigens can be correlated with overall severity of disease, prognosis, and specific subtypes of CD. For example, 25% of CD patients have reactivity to all antigens described and tend to have a progressive form of disease. In comparison, ≈25% of CD patients have no or limited serologic response to these antigens and have a much less severe course. Budesonide responsiveness correlates with OmpC positivity. Furthermore, the trait of having serologic activity to OmpC clusters in particular families, including unaffected relatives (Fig. 7). Serologic activity also correlates with the NOD2 genotype. These studies support a limited group of bacterial antigens as drivers of CD, and that these responses can potentially be utilized to predict the clinical course of IBD patients and their response to therapy.

Figure 7. Differential expression of anti-OmpC antibodies in serum of IBD families.

Consistent with the fact that multiple bacterial antigens are likely to be responsible for driving chronic T-cell-mediated intestinal inflammation, Sandra C. Kim, MD (University of North Carolina at Chapel Hill, NC) has shown that the T-cell responses in germ-free IL-10-deficient mice monocolonized with Enterococcus faecalis are oligoclonal, and that the humoral response is also likely to be oligoclonal. This is evident from the fact that expression of an E. faecalis library with serum from a monocolonized IL-10-deficient animal reveals a limited number of bacterial gene products as being responsible for the immune response. Several of these recombinant gene products, when processed by antigen-presenting cells, drive T-cell reactivity. This T-cell reactivity is consistent with a Th1 response and shows how antigens detected by a humoral response are also able to drive T-cell responses that may be associated with the development of colitis.

Work from R. Balfour Sartor, MD (University of North Carolina at Chapel Hill, NC) and colleagues has shown that a preferential expansion of certain commensal enteric bacteria occurs during the course of chronic immune-mediated colitis, as modeled in germ-free IL-10-deficient mice colonized with specific pathogen-free bacteria. Several of these expanded bacterial species, including E. coli, Klebsiella pneumoniae, and Bifidobacterium animalis, induce colitis and bacterial-specific TH1 immune responses in selectively colonized (monoassociated) IL-10 deficient mice. In other studies, this group has shown that monocolonization of HLA B27 transgenic rats with Bacteroides vulgatus induces colitis that is associated with several dominant antigens (e.g., heat-shock protein 60 [HSP-60] and novel antigens discovered by probing a metagenomic array of luminal bacterial DNA) that may be responsible for driving experimental colitis.

Finally, Ingo B. Autenrieth, MD (University of Tübingen, Tübingen, Germany) and colleagues reported that the immune responsiveness and development of colitis by particular strains of bacteria are dependent on the combination of immune and microbiologic factors and how they interact in a host. In gnotobiotic IL-2-deficient mice, Dr. Autenrieth and colleagues have shown that monocolonization with E. coli, but not B. vulgatus, leads to the development of colitis. However, colonization with both organisms results in the absence of colitis. This finding is contrasted with HLA-B27 transgenic rats in which B. vulgatus is colitogenic and E. coli is noncolitogenic. The biologic basis for these differences may be the way in which genetic factors regulate immune responses by dendritic cells (DCs), i.e., E. coli drives DCs from IL-2-deficient mice to secrete TNF and IL-6, whereas B. vulgatus blocks this effect.

MECHANISMS OF BACTERIAL/EPITHELIAL INTERACTIONS

In the intestinal lumen, immunoglobulins play key roles in microbial inactivation and clearance, and in the formation of complexes that are efficiently sampled by receptor-bearing epithelium or DCs for local mucosal or immune responses. IgA is well recognized as the prototypic biologically active immunoglobulin (Ig) secreted into the intestinal lumen by the epithelial polymeric Ig receptor. However, as discussed by Richard Blumberg, MD (Brigham and Women's Hospital, Harvard Medical School, Boston, MA), the neonatal Fc receptor (FcRn) has greatly broadened and refined the understanding of immunoglobulins as a surveillance and effector system. FcRn has an MHC class 1-like structure that binds IgG at pH 6, but negligibly at physiologic pH. Because of its intracellularly trafficking itinerary, FcRn can function as a shuttle molecule, i.e., it can deliver basolateral IgG from local plasma cells to the lumen. Reciprocally, FcRn samples IgG from the lumen, transcytoses via intracellular endosomes (acidic), bypassing lysosomes, and releases IgG and its bound antigens at the basolateral plasma membrane, thus delivering antigens to lamina propria cells (Fig. 8). IgG is a major component of luminal immunoglobulins in the human, and studies using FcRn^−/− mice transgenically engineered for constitutive small intestinal epithelial FcRn expression show dependence on the FcRn transcytosis mechanism.

Figure 8. Immune surveillance by FcRn-mediated transport of IgG and IgG/antigen complexes.

To demonstrate the contribution of this IgG sampling system to mucosal immune function, Dr. Blumberg presented studies on Citrobacter rodentium, an attaching and effacing pathogen of mice that causes diarrheal disease. Bacterial clearance and disease resolution requires an IgG immunoglobulin response. An essential role of FcRn in such protective responses was established by the increased disease severity and delayed clearance of FcRn^−/− mice, with reversal of susceptibility when FcRn transcytosis was restored in genetically engineered mice.

The mucosal FcRn/IgG system might also elicit pathogenic antimicrobial immune responses. To model this idea, Dr. Blumberg demonstrated that anti-E. coli IgG promoted mucosal uptake of nonpathogenic E. coli by FcRn-mediated transcytosis. Such uptake resulted in an anti-ovalbumin CD4⁺ T-cell response in ovalbumin-transgenic E. coli. Moreover, FcRn^−/− mice were resistant to dextran sulfate-induced colitis. These findings indicate that microbial commensal sampling by FcRn augments immune responses to microbial enteric residents. The FcRn system may provide a mechanism of aggressive immune responses to commensal bacteria.

Lora Hooper, PhD (University of Texas Southwestern Medical Center Dallas, TX) addressed microbial sensing of antimicrobial innate defenses. The Paneth cell is a key small intestinal cell type specialized to produce antimicrobial peptides that defend the crypt, and perhaps the intestinal lumen, from bacterial invasion. Comparisons of germ-free and colonized (Bacteroides thetaiotaomicron) mice demonstrated that this common enteric bacterial resident induces a broad program of gene expression in Paneth cells that includes a new antimicrobial peptide, termed RegIIIγ (Fig. 9). This peptide and its human counterpart, HIP/PAP, are frequently upregulated in the mucosa of experimental and human intestinal inflammation. RegIIIγ bears a carbohydrate recognition domain and structural studies demonstrated that the peptide specifically binds to moieties in the peptidoglycan cell wall of Gram-positive bacteria. Moreover, RegIIIγ rapidly and specifically killed Gram-positive bacteria, with ultramicroscopic evidence of cell wall damage as the mechanism. Further analysis demonstrated that RegIIIγ expression occurred after weaning in wildtype and RAG1^−/− mice. These findings suggest that this antimicrobial peptide is induced by enteric microbial bacteria.

Figure 9. RegIIIγ is a C-type lectin induced in Paneth cells by commensal resident bacteria. Expression in germ-free and bacterial-colonized mouse intestine measured by Western blot analysis. From Cash HL, et al. Symbiotic bacteria direct expression of an intestinal bactericidal lectin. Science. 2006;313:1126–1130. Reprinted with permission from AAAS.

Charles Bevins, MD, PhD (University of California Davis School of Medicine, Davis, CA) discussed innate microbial clearance mechanisms represented by Paneth cell defensins. These are an evolutionarily old family of effector molecules (present in horseshoe crab, birds, and widely distributed in mammalian tissues) with much variation between species. Most human defensin genes cluster at chromosome 8p23, with substantial locus polymorphism. Defensins are membrane-active, broad-spectrum antimicrobials with activity for Gram-positive and Gram-negative bacteria, fungi, protists, and viral agents. Paneth cells express α-defensins, as well as other antimicrobials (e.g., lysozyme and soluble phospholipase A2) that are released from their granules in response to luminal bacterial products and to cholinergic agonists. Paneth cells constitutively express CARD15/NOD2, a susceptibility gene for CD that serves as a microbial sensing molecule. Through such sensing and effector molecule production, Paneth cells are postulated to play an important role in small intestinal homeostasis (Table 2).

Table 2. Putative In Vivo Roles for Paneth Cell Antimicrobials

Protect intestinal stem cells Decrease mucosally associated bacteria, especially in the crypts Regulate composition of intestinal commensal bacteria Protect from food- and water-borne pathogens

The predominant antimicrobial peptide in Paneth cells is the α-defensin HD-5, which is stored as an inactive propeptide that colocalizes with trypsin (also present in an inactive zymogen form). Thus, processing is key to HD-5 function. Since defensins are part of a redundant antimicrobial system of the intestine, characterization of their exact physiologic role is an experimental challenge. Mice transgenic for human HD-5 display elevated resistance to a bacterial pathogen (Salmonella) and exhibit a shift in the composition of their enteric microbial resident bacteria from small morphologic taxa to fusiform bacteria. In human CD, gene expression of HD-5 and HD-6 by Paneth cells is selectively reduced compared to normals (3-fold and 2-fold, respectively), with no significant change in other antimicrobial molecules and a large panel of other Paneth cell products (Fig. 10). Moreover, the deficiency was not related to inflammatory activity. One mechanism for this deficiency is illuminated by reduction of Paneth cell defensins in mice bearing null or missense NOD2 mutations. CD patients with NOD2 polymorphisms have an intrinsic deficiency of Paneth cell defensin production. In this setting, impaired innate control of enteric resident microbiota may drive the formation of destructive acute and adaptive immunity to these organisms and their products, which comprises the ileal CD phenotype.

Figure 10. Selective deficiency of Paneth cell defensin (HD-5 and HD-6) formation in Crohn's disease. From Wehkamp J, et al. Reduced Paneth cell alpha-defensins in ileal Crohn's disease. Proc Natl Acad Sci USA 2005;102:18129–18134. Reprinted with permission. Copyright 2005 National Academy of Sciences, USA.

Toll-like receptors have received detailed study as a mode of microbial sensing in the regulation of microbial clearance and immunoregulation of the mucosal surface. One receptor of interest is TLR-9, since genetic and pharmacologic studies indicate that it plays a role in local and systemic control of mucosal inflammation. Eyal Raz, MD (University of California San Diego, CA) discussed the emerging understanding of how this receptor regulates microbial sensing by the intestinal epithelium. TLR-9 is expressed on colonic epithelial cells and hemopoietic cells such as B cells and DCs. While TLR-9 in hemopoietic cells exclusively resides in an intracellular compartment, it is deployed on both apical and basolateral cell surfaces of polarized intestinal epithelial cells. TLR-9 in both cell surface compartments is active for biochemical signaling since they both generate similar levels and timing of Jun kinase activation, IκB phosphorylation, and IκB ubiquinylation. However, apical TLR-9 signaling failed to proceed to nuclear factor-κB (NFκB) activation. Further biochemical dissection demonstrated that apical signaling blocked proteosomal degradation of ubiquinated IκB, a penultimate step in formation of NFκB (Fig. 11). Apical TLR-9 signaling induced an inhibited state that intensified over 24 hours, such that basolateral signaling to NFκB activation was impaired for a variety of TLR isoforms and other NFκB-activating receptors. This suggested that apical TLR-9 signaling failed to permit activation or localization of an E3 ubiquitin ligase pertinent to IκB processing and that it also induced an expression change of this E3 ligase or its regulatory partners. One appealing target to explain the signaling differences in apical versus basolateral compartments is the Skp1-Cullin1-F-box protein (SCF) (betaTrCP), an E3 ubiquitin ligase that mediates proteosomal capture and degradation of phosphorylated IκB. This line of investigation suggests that epithelial cells are vectorally programmed for microbial sensing and indicates the differentiation fate regarding subsequent microbial interaction. As an example, Dr. Raz presented evidence that TLR-9^−/− mice were impaired for constitutive levels of antimicrobial peptides, but were protected against DSS-induced colitis. This information frames the Janus-like issues confronting biologists and therapists in linking TLR-9 sensing, through its chemical ligands and bacterial probiotics, to amelioration of colitis.

Figure 11. Expression of TLR-9 (green) in native intestinal mucosa, and effect of apical versus basolateral TLR-9 ligation (ISS) on NFκB activation. Reprinted by permission from Macmillan Publishers: Lee J, et al. Maintenance of colonic homeostasis by distinctive apical TLR9 signaling in intestinal epithelial cells. Nature Cell Biology. Copyright 2006.

Mechanisms of probiotic action were discussed by D. Brent Polk, MD (Vanderbilt University School of Medicine, Nashville, TN). Lactobacillus rhamnosus GG (LGG) is one of the best-studied probiotic bacteria in clinical trials. LGG prevent and/or treat diarrhea, Clostridium difficile and Rotavirus infections, CD, and atopic dermatitis. A major challenge has been to understand the mechanisms of such protective action and thereby refine therapeutic strategies. Dr. Polk opened this issue by framing the signaling pathways that control proliferation, survival, and death of intestinal epithelium after receptor or inflammatory challenge. TNF is a major cytokine produced during intestinal inflammation including CD. TNF receptors can mediate either epithelial death or survival, with the latter favored by activating the PI3-kinase/AKT pathway. Work by Dr. Polk and colleagues has shown that one major nexus of these pathways is activation of the host protein kinase suppressor of Ras1 (KSR1).

LGG or its soluble products induced survival to TNF-induced intestinal epithelial cell death by activating the AKT pathway. This group isolated 2 microbial glycoproteins, i.e., p40 and p75, that promoted cell growth, differentiation, and tight junction resistance to H₂O₂ in YAMC and HT29 epithelial cells. In addition, these proteins mediated cell survival to TNF and other NFκB- or p38-activating cytokines in association with AKT activation, through a PI3 kinase-dependent, KSR1-independent process. While the receptor and signaling pathway of these molecules is uncertain, blocking experiments indicate that the epidermal growth factor (EGF) receptor is required for their protective action. In mouse intestinal organ explant culture, p40 and p75 protected colonic epithelium from TNF-induced apoptosis. Immunodepletion and transwell experiments showed that p40 and p75 largely accounted for the protective bioactivity of native LGG. Moreover, the protection among different Lactobacillus species and strains correlated with production of p40 and p75. Thus, secreted bacterial glycoproteins serve as central mediators of probiotic protection by activating epithelial cell survival, growth, and differentiation (Fig. 12).

Figure 12. Potential mechanism of Lactobacillus GG regulation of intestinal epithelial cells.

INNATE (NONEPITHELIAL) RESPONSES

The initial bias that IBD represented an imbalance in adaptive T-cell immunity in the gut has given way to the appreciation that multiple cell types including the innate immune system can play a significant role in mucosal homeostasis and disease pathogenesis. This realization was facilitated by the recognition that pure innate immune defects can lead to mucosal inflammation in both murine models (e.g., STAT3 DN mutation in macrophages) and human disease (IBD-like disease in patients with intrinsic neutrophil or macrophage defects, e.g., chronic granulomatous disease).

While the epithelial cell has been increasingly identified as a regulator of barrier function as well as of innate and adaptive immunity, neutrophils, macrophages, dendritic, and mast cells now are recognized to mediate many mucosal inflammatory processes. In fact, the first gene identified as a contributor in a complex multigenic disease like CD, NOD2, is highly expressed in macrophages, Paneth cells, and inflamed epithelial cells. Mutations in this gene promote intracellular bacterial persistence, altered defensin production, and defective regulatory responses to TLR-2 ligands. The normally functioning mucosal immune system is recognized for its ability to control inflammatory processes despite an exposure level to luminal bacteria and food antigens that dwarfs any antigen exposure systemically. If IBD is an immune dysregulated state, where tolerance to luminal contents is lost, we need to understand the responses of cells and pathways regulating inflammation.

NOD2 is part of a larger family of proteins involved in bacterial recognition and inflammatory responses. This family, termed the CATERPILLAR or NLR family, has been implicated in a number of autoinflammatory, granulomatous, immunodeficiency, and periodic fever syndromes. Their common features include leucine-rich repeats (LRR), nuclear binding domains (NBD), and either a CARD or a pyrin domain (Fig. 13).

Figure 13. CATERPILLER family of genes. There are 20–30+ CATERPILLER genes per mammalian species. CARD = caspase activating and recruiting domain; NBD = nuclear binding domain; LRR = leucine-rich repeats.

There are 25 human genes in this family, whose importance is underscored by the existence of 600 defense-related family members in plants. The interaction of these proteins with members of the TLR family is being increasingly recognized, since some of these molecules are able to regulate or modulate the inflammatory signals mediated through TLRs. Monarch, for example, as described by Jenny P. Ting, PhD (University of North Carolina at Chapel Hill, NC) can inhibit inflammatory responses in macrophages, neutrophils, and eosinophils. Activation of Monarch reduces TLR-mediated increases in NF-κB and IL-6 by inhibiting IRAK1. Monarch blocks IRAK4-mediated phosphorylation of IRAK1 leading to the failure of IκB to dissociate from NF-κB. This event occurs as part of a larger complex with molecules such as ASC (apoptosis-associated speck-like protein containing a CARD). ASC contains both a CARD and pyrin domain and associates with other CATERPILLAR molecules, NALP3 and cryopyrin, forming an “inflamasome.” A mutation of NALP3 that leads to a constitutively active form results in spontaneous inflammatory diseases in man. ASC-deficient macrophages fail to specifically generate IL-1 or IL-18 (both caspase 1-dependent) in response to TLR signals but exhibit more global defects in response to bacteria such as Pseudomonas gingivalis (decreased IL-6, IL-8, and IL-10, all independent of the IL-1 secretion defect). Richard A. Flavell, PhD (Yale University School of Medicine, New Haven, CT) and his group have shown that ASC is a critical regulator of IL-1 production. NALP3-deficient mice demonstrate a more global deficit in IL-1, IL-18, and IL-10 without any effects on IL-12 in response to Toll signals and ASC and NALP3-deficient mice are resistant to LPS-induced death. Differences do exist in response to intact bacteria expressing multiple Toll ligands, and cell death meditated by intracellular bacteria is affected by other members of this family (not NALP3).

NOD2 is the CATERPILLAR family member that has received the greatest attention. Three common mutations in this gene predispose to the development of CD. NOD2 recognizes muramyl dipeptide (MDP), the smallest component of peptidoglycan from Gram-positive organisms. NOD2 is an intracellular pattern recognition receptor with an LRR that binds MDP, which in turn results in NFκB activation. No activation occurs in the presence of either of these three mutations. In the normal state NOD2 signals synergize with those mediated through Toll receptors, promoting the production of inflammatory cytokines (e.g., IL-12, IL-23). Kiochi Kobayashi, MD, PhD (Harvard Medical School, Boston, MA) reported on his work in collaboration with Dr. Flavell in which they generated an NOD2 knockout mouse to address the role of NOD2 in normal and disease states. This mouse was completely normal in appearance but did demonstrate defects in the synergistic TLR signaling pathway (for some Tolls, e.g., TLR-7, but not all). In addition, when these mice are infected with Listeria orally but not intravenously there are greater numbers of recoverable bacteria in the liver and spleen. These findings suggest that NOD2 expression in Paneth cells may play a key role in keeping luminal bacteria in check.

Other components of the pattern recognition receptor pathway play a role in inflammation as well. MyD88, a critical regulator of TLR signaling, is involved in the inflammation seen in IL-10^−/− mice. MyD88/IL-10 double knockouts do not develop inflammation. In contrast, IRAK-M^−/− mice have intestinal inflammation and IL-10/IRAK-M double knockouts express more active disease.

The TLR signaling pathway is clearly important for intestinal homeostasis. Averil I. Ma, MD (University of California at San Francisco, CA) and colleagues have focused on a regulator of NF-κB called A20. This molecule appears to play a role in the integration of responses. A20 is a TNF-induced enzyme that is expressed in almost every tissue in the body. The A20^−/− mouse develops diffuse skin and intestinal inflammation within days to months. In A20-deficient cells activation by an inflammatory stimulus leads to persistent NF-κB activity. Thus, A20 is responsible for the termination of NF-κB activity. The inflammation in A20^−/− mice is TNF independent (double knockouts develops similar disease) and is not dependent on T cells. The response to all Toll ligands is comparable—an aggressive inflammatory uncontrolled response. This response is due to IKKα kinase activity that results in IκB phosphorylation (and degradation) with persistent NF-κB activation. These mice die from the persistent inflammation. Interestingly, A20/MyD88 double knockouts do not die. Mortality is mediated through cells of the innate immune system as A20/RAG knockout mice still die.

It appears that A20 can both ubiquitinate and de-ubiquitinate substrates. Furthermore, the site where ubiquitination occurs affects the functional outcome. Ubiquitination at lysine 48 (K48) targets molecules for degradation, while ubiquitination of K63 activates the molecule. A conserved DXXC motif in the N terminus of A20 regulates this function. The effect is mediated through TRAF6. TLR signaling results in K63 ubiquitination of TRAF6 turning off its activity.

All of the above-mentioned signaling pathways occur within cells of the innate immune system. Macrophages and DCs in the intestine are very different than their counterparts in the systemic immune system. Phillip D. Smith, MD (University of Alabama at Birmingham, AL) and his group have shown that jejunal macrophages exhibit limited expression of TLRs, MD2, CD14, CD8, CD16 (FcγR), CD80, CD86, and CD40. These cells express HLA-DR, CD13, IL-8 receptors, CXCR1 and CXR2, but do not express markers CD25 and TREM1, or the HIV-related chemokine receptors, CXCR4 and CCR5.

Consistent with this altered phenotype, intestinal macrophages fail to respond to LPS and do not signal intracellularly. These cells constitutively express the IRAK inhibitor discussed above, IRAK-M. In addition, they possess phagocytic capacity without inflammatory potential. This altered functional and phenotypic state appears to be mediated by factors elaborated by intestinal stromal cells, e.g., most likely TGF-β. Responses to this cytokine are enhanced by the absence of SMAD7 and mediated by the activation of SMAD4. This phenotype has been defined as “inflammation anergic”; however, while this is true in the proximal small bowel, it may not reflect what occurs in the colon.

Intestinal DCs, like intestinal macrophages, differ from their systemic counterparts. Some of these differences depend on the location of the DC, i.e., in the Peyer's patch (PP), in the subepithelial space, or in the mesenteric lymph node (MLN). Hans-Christian Reinecker, MD (Harvard Medical School, Boston, MA) and his colleagues used a novel marking approach knocking out the fractalkine receptor and inserting GFP in its place. These cells exist in high numbers in the lamina propria. They are immature by phenotype and are capable of active phagocytosis. Their presence appears to be regulated by luminal flora, as the number of these cells are reduced in germ-free mice. Furthermore, they extend dendrites across the tight junctions of the epithelium of the small intestine. In the absence of luminal microbial stimulation these dendrites do not form. Using labeled wheat germ agglutinin, this group showed that these dendrites actively take up antigen from the lumen. Similar findings were noted with labeled E. coli. Interestingly, bacterial uptake into PPs appears to be fractalkine receptor-independent. In contrast, DCs in the PP appear to express CCR6. Using a T-cell receptor (TCR) transgenic mouse (specific for SMI) this group showed that CCR6+ DCs are involved in PP and MLN activation of transgenic T cells. There is cross-talk between DCs potentially exchanging material and locally activating T cells. The fractalkine receptor-expressing DCs may be involved in intestinal homeostasis, whereas CCR6+ DCs may control active immunity. Clearly, differences in the antigen (protein versus bacteria) as well as the location of the DC may dictate their physiologic function.

The question then remains as to how these findings relate to IBD. As alluded to above, defects in innate immunity can result in IBD. Alterations in how cells sense bacteria and their products can also predispose the host to uncontrolled inflammation. Knowing these pathways and the cellular contributors will allow us to develop novel therapeutic targets in IBD.

MECHANISMS OF PROTECTIVE/REGULATORY RESPONSES TO COMMENSAL MICROBES

Multiple lymphocyte subsets contain cells that regulate immune function. The term “regulatory cells” refers to cells that inhibit immune responses, including responses that result in inflammation. Many cell types may serve this function ancillary to their major purpose; these include B cells and natural killer (NK) T cells. Other cells are highly specialized to serve only regulatory function, with CD4⁺ T cells being the major subset. CD4⁺ T regulatory cells (Tregs) have at least 2 important subtypes: 1) “natural Tregs” that are selected in the thymus, require the foxp3 transcription factor for their generation, and constitutively express CD25 on their surface; and 2) “adaptive Tregs” that are induced in the periphery and appear to be heterogeneous. Thus, some adaptive Tregs express foxp3 or express high amounts of IL-10 but others do not.

CD4⁺ Tregs are present in the lamina propria of the intestine. Adoptive transfer of CD4⁺ Tregs can prevent and treat colitis in experimental models. Thus, these Tregs represent an attractive mode of potential therapy for human IBD if we are able to manipulate their function and numbers in patients. Intestinal Tregs appear to be induced in response to the microbiota, but little is known about this induction. Intestinal helminths appear to have evolved the ability to stimulate Tregs as a mechanism to prevent deleterious immune reactions to themselves. Together these observations provide support for the notion that intestinal Tregs can be selectively modulated as a mode of therapy for intestinal inflammation. This session included research related to multiple types of regulatory cells as well as studies on the molecular mechanisms of immune regulation in the gut.

Despite intensive interest in the function of regulatory T cells in mucosal tissues, there remains a relatively poor understanding of the number of T regulatory lineages, their anatomic localization, and their in situ function. To address this question, Casey T. Weaver, MD (University of Alabama at Birmingham, AL) and colleagues have generated an IL-10 reporter mouse in which T cells that express IL-10 can be identified by coexpression of Thy1.1 on the cell surface. These IL-10 reporter mice have been used to characterize patterns of IL-10 and foxp3 expression in the intraepithelial (IEL) and lamina propria (LP) compartments of the normal mouse small and large intestine.

IEL and LP compartments develop effector T cells in response to the enteric microbiota. Regulatory T cells also are presumed to exist in these compartments, but there has been little direct analysis of these cell populations in either normal or diseased states. Dr. Weaver's presentation focused on 2 regulatory T-cell populations—the natural Tregs marked by foxp3 and adaptive Tregs producing IL-10. IL-10 deficiency is associated with spontaneous intestinal inflammation, yet foxp3⁺ T cells are unchanged in such mice, suggesting an important role for IL-10 producing Tregs in intestinal homeostasis. Yet the cells that produce IL-10 have yet to be characterized. Some of the difficulty in identifying these latter cells is due to poor intracellular staining for IL-10 and the resulting underestimation of their number. Using this new mouse reporter model, Dr. Weaver presented data showing that Thy1.1⁺-IL-10 competent T cells are present in all peripheral lymphoid compartments of normal B6 mice and are enriched in the intestinal lymphoid tissues in both LP and IEL compartments. The majority of the IL-10⁺ T cells in the intestine were foxp3-negative, but analysis of the different intestinal compartments revealed a surprising degree of complexity in these Treg populations.

Jonathan Braun, MD, PhD (University of California at Los Angeles, CA) addressed the role of regulatory B cells that protect against colitis in Gαi2 knockout mice. This unexpected role involves the capacity of B cells to induce the expansion and activation of intestinal CD8⁺CD4⁺ and NK T cells, which may be the proximate cell types mediating immunoregulation for this Th1-mediated colitis. The B-cell subset proficient in immunoregulation is localized to mesenteric nodes, is CD19^hi after activation, and is genetically dependent on IL-10, CD40, CD80/CD86, MHC class I, and CD1d for its protective function. Surprisingly, MHC class II is dispensable for immunoregulatory competence, suggesting that B cells do not regulate this activity through direct CD4 T-cell interaction, but instead through CD1d-dependent NK T cells and perhaps MHC class I-dependent CD8 T cells.

Evidence that regulatory B cells are stimulated in response to the microbiota was provided by a second experimental system, C57Bl/6 mice with a restricted flora (RF mice) that are maintained in isolators at UCLA. The microbiota of these RF mice are enriched in microbes of the Firmicutes phylum and depleted of Bacteroides microbes compared to conventional B6 mice. RF mice were functionally and phenotypically deficient in protective CD21⁺, CD23⁻ B cells, CD5⁺ B cells, plasmacytoid DCs, naïve CD4 and CD8 T cells, and iNK T cells. Dr. Braun speculated that pDC and protective B cells reflect an overlapping or identical population of immunoregulatory cells, whose formation is programmed by resident enteric bacteria.

Enteric bacteria are a critical component in the initiation and perpetuation of chronic intestinal inflammation. Dirk Haller, PhD (Technical University of Munich, Germany) presented an emerging new paradigm whereby intestinal epithelium responds to the microbiota by upregulating inflammatory genes, but that host mediators serve to dampen this proinflammatory epithelial response in normal hosts. Haller and co-workers have used a variety of in vitro and in vivo systems to study this question, including the 129.IL-10^−/− mouse monoassociated with E. faecalis, which is a harmless commensal in normal mice but an inducer of colitis in this IL-10^−/− host. They have identified 15-deoxy-prostaglandin J2, TGF-β, and IL-10 as mediators providing negative feedback in bacterial-stimulated intestinal epithelial cells (IECs).

Enterococcus faecalis induced phosphorylation of p38 MAPK in normal 129 intestinal IECs in vivo, but not in 129.IL-10^−/− epithelial cells. In the latter, E. faecalis monoassociated 129.IL-10^−/− epithelial cells demonstrate persistent phosphorylation of NF-κB p65 (RelA), rather than p38 MAPK. The persistent RelA expression in IL-10^−/− mouse epithelium might be related to persistent expression of TLR-2, which was found on IL-10-deficient epithelial cells, but not on epithelial cells of control mice. In vitro studies have demonstrated that TGF-β plus E. faecalis stimulation together, but neither one alone, induced degradation of epithelial TLR-2 and its removal from the cell surface. These TGF-β effects were shown to be associated with SMAD2 signaling.

To identify regulatory disease mechanisms in experimental colitis, Haller and colleagues used functional epithelial cell proteomics in E. faecalis-monoassociated wildtype (WT) and 129.IL-10^−/− mice. Persistent induction of NF-κB RelA phosphorylation in IL-10^−/− IEC was associated with the increased expression of grp-78 (glucose regulated protein-78), and recruitment of grp-78 to the IKK complex was required for TNF-induced NF-κB RelA phosphorylation, suggesting proinflammatory mechanisms for grp-78 expression in IEC. Of note, IL-10-stimulated p38 MAPK signaling inhibited grp-78 expression, and blocked TNF-induced RelA phosphorylation in IL-10 receptor-reconstituted IEC in vitro. This may relate to the p38 MAPK signaling seen in WT but not IL-10^−/− IECs in vivo. Dr. Haller suggested that TGF-β and IL-10 signaling to epithelial cells might critically contribute to immune homeostasis toward the microbiota (Fig. 14).

Figure 14. Intestinal epithelial cell signaling and host-derived negative regulators of chronic inflammation: to be or not to be activated determines the balance of inflammation versus protection toward commensal bacteria.

A fundamental question related to the pathogenesis of IBD is whether intestinal inflammation results from an immune response directed against bacteria, or whether bacteria simply provide costimulation for an autoimmune response. Marika C. Kullberg, PhD (University of York and Hull York Medical School, UK) presented evidence that components of the gut microbiota induce CD4⁺ T cells with the potential to induce colitis, but that this immunopathologic response is controlled by bacterial antigen-specific Treg cells in normal individuals. Knowing that SPF-C57BL.IL-10-deficient (IL-10 KO), but not WT mice, develop colitis after experimental infection with Helicobacter hepaticus, Kullberg and colleagues examined both the pathogenic and disease-protective arm of the immune system to H. hepatitis using an RAG-deficient transfer murine model.

In this RAG-deficient transfer model, colitis was induced by transfer of CD4+ T-cell clones specific for H. hepaticus, among them 1 clone recognizing a 15-mer peptide from the Helicobacter flagellar hook protein. This response suggests that an effector T-cell response directed against a gut bacterial antigen is sufficient to trigger colitis. Further, they found that the bacterial infection triggers the development of disease-protective CD4⁺ CD45RB^low Treg cells in WT mice. These cells, which can be either CD25⁺ or CD25⁻, block colitis development in vivo through their production of IL-10, secrete IL-10 after bacterial antigen stimulation in vitro, and are not present in mice not colonized with H. hepaticus, suggesting antigen-specificity. The disease-protective role of IL-10⁺ production by Treg cells in infected WT hosts was confirmed by anti-IL-10R treatment that reversed the protective effect of the Treg cells in vivo.

TLR activation, following recognition of specific microbe-associated molecular patterns, triggers diverse transcriptional responses that impact early innate immune responses, which in turn influence adaptive immune outcome. Key transcription factors responsive to TLR signaling include NF-κB, MAP kinases, and interferon (IFN) regulatory factor complexes, all of which regulate the transcription of genes encoding proteins involved in inflammation and immunity. Specificity of TLR activity is crucial for efficient bacterial/viral clearance, and to some extent is dictated by a network of adaptor proteins that are differentially recruited toward the activated TLR. Of equal importance is the downregulation of TLR signaling.

Several endogenous modulators control initiation, duration, and repression of the TLR response, including nuclear receptors. Denise Kelly, PhD (Rowett Research Institute, Aberdeen, Scotland) reviewed data showing the existence of several distinct but potentially combinatorial mechanisms of nuclear receptor regulation of TLR signaling. She noted that peroxisome proliferator activated receptors (PPARs) repress inflammatory signaling cascades by interfering with NF-κB, including those mediated by TLRs. More recently, 5-aminosalicylic acid (5-ASA) has been shown to mediate its antiinflammatory effects through PPARγ. Addition of 5-ASA to IECs induced nucleocytoplasmic shuttling of the PPARγ protein. Dr. Kelly noted that similar shuttling occurs in IECs in response to the presence of certain commensal bacteria, such as B. thetaiotaomicron and Roseburia hominus that mediate potent antiinflammatory effects in response to TLR-5-induced NF-κB activation in IECs.

The mechanism(s) by which PPARs regulate NF-κB has not been fully elucidated; however, Dr. Kelly speculated that the cellular effects elicited by PPARγ and its natural and synthetic ligands involve complex and diverse mechanisms and that their eventual elucidation will identify novel therapeutic cellular targets or PPARγ ligands with greater biological efficacy.

Joel V. Weinstock, MD (Tufts-New England Medical Center, Boston, MA) presented data showing that IBD and other immune-mediated illnesses, while common in industrialized countries, are rare in tropical, less-developed countries. Dr. Weinstock noted that although genetic predisposition plays an important role in autoimmune diseases, dramatic increases in disease frequency within 1 to 2 generations must be related to environmental changes. Among the many changes that come with socioeconomic improvement is loss of exposure to helminths, which induce regulatory T cells, an effect that likely allows them to persist in their hosts.

Dr. Weinstock explained that helminths alter host immune responses in several ways that might inhibit IBD. Lamina propria mononuclear cells (LPMC) isolated from mice with helminths produce more IL-10 and TGF-β and less IL-12p40 and IFN-γ than LPMC from worm-free mice. Colonization with Heligmosomoides polygyrus increases foxp3 expression by MLN and LP T cells. Transfer of MLN T cells from H. polygyrus-colonized mice to colitic mice ameliorates inflammation. Helminth colonization induces an LP TLR4⁺ T-cell population that produces TGF-β in response to LPS. Colonization also induces an LP CD8⁺ regulatory T cell that can inhibit antigen-induced T-cell proliferation and prevent colitis in an IBD animal model.

Dr. Weinstock and colleagues have found that mice harboring helminths are protected from developing colonic inflammation, and that previously established colonic inflammation resolves in mice that are exposed to helminths. In addition, he noted that clinical trials have shown that patients with IBD may improve when given viable embryonated Trichuris suis ova.

Cox Terhorst, PhD (Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA) closed out the final session of the workshop by discussing the Signaling Lymphocyte Activation Molecule (SLAM) gene family of cell surface receptors and its potential role in controlling immune response to bacteria. Dr. Terhorst explained that the SLAM family of receptors is encoded by 9 genes that map in close proximity of each other on human or mouse chromosome 1. This family is emerging as a set of regulators of adaptive and innate immunity. In addition, SLAM-family-specific adapters, i.e., EAT-2A, EAT-2B, and SAP, regulate SLAM-family-receptor functions during innate and adaptive immune responses.

SLAMF1 is a self-ligand receptor located at the interface between T cells and professional antigen-presenting cells. SLAMF1-deficient mice have abnormal responses to bacteria and Leishmania major and have abnormal airway hyperresponsiveness. This is the result of defects in the final stages of CD4+ T-cell differentiation and/or because of defective killing of bacteria by SLAMF1-deficient macrophages, tissue DCs, and neutrophils. In contrast, only defects in CD4+ T-cell differentiation and in neutrophil killing of bacteria are found in SLAMF6-deficient mice. Abnormal macrophage functions have also been observed in SLAMF2- and in SLAMF8-deficient mice.

Experimental colitis does not develop when SLAMF1xRag-2-or SLAMF2xRag-2-deficient mice are recipients of adoptively transferred CD45RB^hi wildtype CD4⁺ T cells. Furthermore, monoclonal antibodies directed against SLAMF1, SLAMF2, or SLAMF6 ameliorate colitis in this model. Dr. Terhorst concluded that these results are consistent with a model that bacterial killing by macrophages or neutrophils is prerequisite for development of colitis.

CONCLUSIONS

These cutting-edge presentations and provocative discussions have advanced our understanding of the mechanisms by which epithelial cells, innate immune cells, and T cells respond to ubiquitous luminal microbial antigens and adjuvants (Fig. 15). Normal individuals have redundant protective mechanisms that include a resistant mucosal barrier that limits exposure of immune cells to microbial and dietary components, secretion of antimicrobial peptides by Paneth cells, production of inhibitory intracellular (A20, PPARγ, etc.) and secreted (IL-10, PGJ2, IFNα/β) proteins by innate cells, including epithelial cells, and activation of regulatory T-cell subsets that secrete inhibitory TGF-β and IL-10. Genetic defects in any of these processes can lead to enhanced antigen/adjuvant uptake or inappropriate immune activation that results in pathogenic T-cell responses to commensal microbial antigens and chronic intestinal inflammation. Individual cytokines and signal transduction can be either inflammatory or protective, depending on the responding cell type and the timing of activation. For example, TLR signaling in epithelial cells is generally protective, while stimulation of the same pathways in lamina propria immune cells causes inflammation.

Figure 15. Microbial/epithelial/immune interactions in the intestinal mucosa. Microbial adjuvants and antigens are mostly excluded by the intact mucosal barrier, but selectively sampled luminal adjuvants and antigens stimulate protective immune responses by epithelial cells, antigen presenting cells (APC), and T cells (TR1, Treg) in the normal host. In genetically susceptible hosts or following extensive mucosal damage, aggressive innate and acquired immune responses are generated that lead to tissue damage and chronic inflammation in the absence of appropriate downregulation.

Refinement of molecular detection techniques has rapidly expanded our understanding of the composition of mucosal microbiota and those species that are expanded in experimental colitis, CD, and UC. Although it is possible that a specific pathogen causes CD or UC, current data do not strongly support this hypothesis. Probing bacterial expression libraries have identified novel antigens, such as flagellin, that drive pathogenic immune responses in both experimental colitis and CD. It is highly likely that additional antigens will be discovered by metagenomic studies. Finally, understanding the mechanisms of biofilm formation and quorum sensing will lend insights into bacterial aggregation at the mucosal surface.

Although clear advances have been made in understanding microbial–host interactions at mucosal surfaces, many challenges remain before this understanding can be translated into effective therapies. We need much more detailed analysis of the dominant microbial species, antigens, and adjuvants that induce pathogenic immune responses in different subsets of patients with CD and UC so that efficient antibiotic therapies and perhaps immunization can be designed. Insights into key pathogenic and protective molecules and their signaling pathways that are activated by dominant antigens and adjuvants should suggest molecular and pharmacologic therapeutic strategies to induce and maintain lasting remission of disease. The most physiologic approaches to long-term remission are to restore the balance of aggressive versus protective commensal bacterial species and to stimulate endogenous protective molecules.

Acknowledgements

The authors thank the Crohn's & Colitis Foundation of America for sponsoring this conference and Marjorie Merrick, Vice President of Research and Scientific Programs, for guidance and support. We also thank Claire Klepner and Maestro Communications for outstanding logistical coordination and invaluable writing and editorial assistance with this article and UCB Pharma for its generous unrestricted educational grant. Most important, we thank all of the outstanding investigators who participated in this workshop: Kazumichi Abe, MD (University of California San Diego, USA); Ingo B. Autenrieth, MD (University of Tübingen, Germany); Sean L. Barnes, MD, PhD (University of Virginia, USA); Terrence A. Barrett, MD (Northwestern University, USA); Charles L. Bevins, MD, PhD (University of California Davis School of Medicine, USA); Edgar C. Boedeker, MD (University of New Mexico, USA); James Borneman, PhD (University of California Riverside, USA); Thomas J. Borody, MD, PhD (Centre for Digestive Disease, Australia); Elke Cario, MD (University Hospital of Essen, Germany); Steven M. Cohn, MD, PhD (University of Virginia, USA); Michael T. Collins, DVM, PhD (University of Wisconsin-Madison School of Veterinary Medicine, USA); Fabio Cominelli, MD, PhD (University of Virginia, USA); Arlette Darfeuille-Michaud, PhD (Université d'Auvergne, France); Peter B. Ernst, DVM, PhD (University of Virginia, USA); Sean D. Fine (University of California San Diego, USA); Richard A. Flavell, PhD (Yale University School of Medicine, USA); Julia-Stefanie Frick, MD (University of Tübingen, Germany); Michael S. Gilmore, PhD (Harvard Medical School, USA); Laura P. Hale, MD, PhD (Duke University Medical Center, USA); Dirk Haller, PhD (Technical University of Munich, Germany); Gail Hecht, MD (University of Illinois at Chicago, USA); Robert M. Hershberg, MD, PhD (Dendreon Corporation, USA); Stuart S. Hobbs, PhD (Vanderbilt University School of Medicine, USA); Lora V. Hooper, PhD (University of Texas Southwestern Medical Center at Dallas, USA); Onyinye I. Iweala (Massachusetts General Hospital, USA); Stephen P. James, MD (National Institute of Diabetes & Digestive & Kidney Diseases, USA); Christian Jobin, PhD (University of North Carolina at Chapel Hill, USA); Martin F. Kagnoff, MD (University of California San Diego, USA); John Y. Kao, MD (University of Michigan Medical School, USA); Dennis L. Kasper, MD (Harvard Medical School, USA); Denise Kelly, PhD (Rowett Research Institute, Scotland); Sandra C. Kim, MD (University of North Carolina at Chapel Hill, USA); Koichi Kobayashi, MD, PhD (Harvard Medical School, USA); Marika C. Kullberg, PhD (University of York, United Kingdom); Averil I. Ma, MD (University of California at San Francisco, USA); Richard P. MacDermott, MD (Albany Medical College, USA); Lillian Maggio-Price, VMD, PhD (University of Washington School of Medicine, USA); Linda S. Mansfield, VMD, PhD (Michigan State University, USA); Sarkis K. Mazmanian, PhD (Harvard Medical School, USA); Makoto Naganuma, MD, PhD (University of Virginia, USA); Cathryn Nagler, PhD (Harvard Medical School, USA); Saleh A. Naser, PhD (University of Central Florida, USA); Edward E.S. Nieuwenhuis, MD, PhD (Sophia Children's Hospital, The Netherlands); Bruce J. Paster, PhD (Harvard Medical School, USA); Sven Pettersson, MD, PhD (Karolinska Institute, Sweden); Barbara N. Phenix (UCB Pharma, USA); Daniel K. Podolsky, MD (Massachusetts General Hospital, USA); D. Brent Polk, MD (Vanderbilt University School of Medicine, USA); Guénolée Prioult, PhD (Harvard Medical School, USA); Eyal Raz, MD (University of California San Diego, USA); Hans-Christian Reinecker, MD (Harvard Medical School, USA); Maria Rescigno, PhD (European Institute of Oncology, Italy); Annette L. Rothermel, PhD (National Institute of Allergy & Infectious Diseases, USA); Pedro A. Ruiz (Technical Institute of Munich, Germany); Cynthia L. Sears, MD (Johns Hopkins University School of Medicine, USA); Jishu Shi, DVM, PhD (Auburn University, USA); Anna Shkoda (Technical University of Munich, Germany); Donald W. Smith (Massachusetts General Hospital, USA); Phillip D. Smith, MD (University of Alabama at Birmingham, USA); Scott B. Snapper, MD, PhD (Massachusetts General Hospital, USA); William F. Stenson, MD (Washington University, USA); Jun Sun, PhD (University of Chicago, USA); Hong Tang, MD (National Institute of Allergy & Infectious Diseases, USA); Gerald W. Tannock, PhD (University of Otago, New Zealand); Stephan R. Targan, MD (Cedars-Sinai Medical Center, USA); Cox Terhorst, PhD (Harvard Medical School, USA); Jenny P. Ting, PhD (University of North Carolina at Chapel Hill, USA); Shahid Umar, PhD (University of Texas Medical Branch, USA); John F. Valentine, MD (University of Florida, USA); James Versalovic, MD, PhD (Baylor College of Medicine, USA); Matam Vijay Kumar, PhD (Emory University, USA); Casey T. Weaver, MD (University of Alabama at Birmingham, USA); Joel V. Weinstock, MD (Tufts-New England Medical Center, USA); Heather D. Wood (Michigan State University, USA); Fang Yan, MD, PhD (Vanderbilt University School of Medicine, USA); and Vincent B. Young, MD, PhD (Michigan State University, USA).