Final Progress Report
Proposal No. NICE-03
Principal Investigator: R. Balfour Sartor, M.D.
Applicant Organization: University of North Carolina at Chapel Hill (U.S.A.)
Project Title: Identification of dominant bacterial antigens from IL-10 knockout mice and HLA-527 transgenic rats with colitis
Period of Award: January 1, 2003 - December 31, 2004
I. Summary of Project Aims
A. Overall Goal. The overall goal of this project was to identify the dominant bacterial antigens and adjuvants that induce pathogenic cell mediated immune responses leading to experimental colitis in genetically susceptible hosts and protective (tolerogenic) immune responses in resistant (normal) hosts. These studies were designed to be completed in two genetically engineered, well defined rodent models, interleukin (IL)-10 deficient mice and HLA-B27/human β2 microglobulin transgenic (HLA-B27 TG) rats.
B. Specific Aims (Year 2)
1. Determine whether recombinant bacterial proteins induce pathogenic and protective T cell responses in IL-10 knockout and wildtype (WT) mice, respectively.
2. Determine the ability of recombinant bacterial proteins to activate pathogenic and protective innate immune responses.
II. Accomplishments toward meeting these aims.
The second year original application focused on the recombinant flagellin discovered by Drs. Robert Hershberg and Charles Elson as a dominant antigen inducing colitis in the C3H/HeJ Bir colitis model in the first year of funding. Because Dr. Hershberg left Corixa and the company retained ownership of this protein and due to restrictions on the material transfer agreement between the University of Alabama, Birmingham and Corixa, we were unable to obtain this recombinant flagellin. Therefore, we pursued these goals with the recombinant Enterococcus faecalis antigens that we cloned in the first year of BMRP support in collaboration with Dr. Hershberg as well as a Bacteroides vulgatus dominant antigen that we independently cloned at the University of North Carolina. We have sent serum from SPF IL-10-/- and WT mice to Dr. Elson at the University of Alabama, Birmingham, to measure IgG responses to flagellin to pursue one of our original aims.
A. Identification of dominant antigens
1. Enterococcus faecalis. We have shown that commensal enteric bacteria are required for the development of TH1-mediated chronic colitis in IL-10-/-, since germ-free (sterile) IL-10-/- mice failed to develop colitis, but disease developed in IL-10-/- but not wildtype (WT) mice within one week of colonizing germ-free mice with specific pathogen-free colonic bacteria (Sellon, Infection and Immunity, 1998). We subsequently demonstrated that selective colonization (monoassociation) of gnotobiotic mice with E. faecalis, a common gram positive aerobic enteric bacterial species, caused colitis that predominantly localized to the distal colon (Kim, Gastroenterology 2005). CD4+ T cells isolated from the mesenteric lymph nodes of E. faecalis-monoassociated IL-10-/- mice produced large amounts of interferon γ (IFNγ) when co-cultured with antigen-presenting cells (APC) that were pulsed with E. faecalis lysate but not with E. coli, Bacteroides vulgatus or an irrelevant antigen, KLH (Kim, Gastroenterology 2005). We showed in collaboration with Dr. Terry Barrett’s group at Northwestern University that E. faecalis elicited an oligoclonal response in T cells from the colonic lamina propria, mesenteric lymph nodes and spleen of monoassociated IL-10-/- mice. In the first year of BMRP funding of this project, Dr. Robert Hershberg and Mike Lodes at Corixa created an expression library of our E. faecalis strain, which was screened with serum from our E. faecalis-monoassociated IL-10-/- mice with advanced colitis. This screening yielded 21 dominant antigens that were sequenced and identified. Three of these clones were expressed in E. coli and purified for further testing. These particular clones were chosen because the proteins were predicted on the basis of their structure to have both antigenic and adjuvant properties and were three of the four most commonly isolated clones identified by our serologic detection assay.
2. Bacteroides vulgatus. We used a different strategy to identify, clone and purify a dominant antigen and adjuvant from Bacteroides vulgatus. B. vulgatus selectively induces T cell-mediated colitis in monoassociated HLA-B27 TG rats (Rath, JCI 1996, Rath, Infection Immunity 1999). We performed a Western blot of B. vulgatus lysate with serum from HLA-B27 TG rats with active colitis after being monoassociated with B. vulgatus. Several dominant bands were present, but one 45kDa band was uniquely present in lysates of all Bacteroides species (B. vulgatus, B. thetaiotamicron and B. ovatus) that caused colitis in monoassociated HLA-B27 TG rats, but not in the B. distasonis lane. B distasonis does not induced colitis in monoassociated B27TG rats (Mann, Gastroenterology Abstract 2004). We then used a proteomic approach (GLC – mass spectroscopy) to identify this band as heat shock protein 60 (HSP60). We cloned HSP60 from B. vulgatus, expressed this gene in an E. coli vector and purified recombinant B. vulgatus HSP60.
B. Aim 1. Determine whether recombinant bacterial proteins induce pathogenic and protective T cell responses in IL-10-/- and wild type mice, respectively.
1. Enterococcus faecalis proteins
We first performed dose response studies with the three expressed E. faecalis proteins using unfractionated MLN cells from E. faecalis monoassociated IL-10-/- and WT mice. One of the clones, EF-60, stimulated large amounts of IFNγ in a dose-responsive manner and another recombinant protein, EF-33, looked promising with evidence of toxicity to the MLN at higher concentrations (50 μg/ml). EF-20 showed no activity in this assay. Of considerable importance, MLN cells from WT mice had no response to any the recombinant proteins.
We then determined if antigen-pulsed APC (T cell-depleted splenocytes) from SPF IL-10-/- mice would stimulate TH1 immune responses. APC were pulsed with 10 or 50 μg/ml of the three E. faecalis recombinant proteins, KLH as an irrelevant antigen and E. faecalis lysate (positive control). The APCs were then washed to remove non-processed antigen and co-cultured for 72 hours with purified CD4+ T cells from E. faecalis-monoassociated IL-10-/- mice with active colitis. Supernatants were assayed for IFNγ by ELISA. EF-60 again stimulated 10-55 fold higher IFNγ responses than KLH at 50 μg/ml, EF-33 stimulated minor IFNg responses without a dose response and EF-20 had no response.
These results indicate that at least one recombinant E. faecalis protein is a dominant antigen for TH1-mediated colitis in IL-10-/-monoassociated IL-10-/- mice. We are measuring IL-10 in the supernatants of identically stimulated monoassociated WT MLN cells and CD4+ T cells to see if the same antigen can elicit protective regulatory responses in the normal hosts.
2. B. vulgatus HSP 60
In preliminary studies, we have investigated the ability of purified recombinant B. vulgatus HSP60 to stimulate IFNγ in cells from B. vulgatus- monoassociated HLA-B27TG rats with active colitis. To date, B. vulgatus HSP60 has not induced IFNγ responses by either unfractionated MLN cells or CD4+ T cells co-cultured with HSP60-pulsed APC. This is consistent with our inability to demonstrate consistent IFNγ responses to B. vulgatus lysates, despite a clear T cell dependence of colitis in B. vulgatus-monoassociated B27TG rats. We implicated T cell in this model by showing that nude B. vulgatus-monoassociated HLA-B27TG rats do not develop colitis unless they are reconstituted with CD4+ T cells from B. vulgatus–monoassociated HLA-B27TG rats (Hoentjen, submitted manuscript). Of note, we have very recently shown that T cells from HLA-B27 TG nude rats that develop colitis after CD4+ T cell transfer produce IFNg in response to in vitro stimulation with B. vulgatus lysates. We will use these cells to study their response to B. vulgatus HSP60. In addition, we will assay our previously collected supernatants to see if IL-17 is being produced as an alternative TH1 cytokine.
C. Aim 2. Determine the ability of recombinant bacterial antigens to activate pathogenic and protective innate immune responses.
1. E. faecalis proteins
We derived dendritic cells (DC) from bone marrow (BM) of IL-10-/- mice by in vitro exposure to GM-CSF and IL-4 by standard techniques. In addition to inducing IFNγ, these same recombinant E. faecalis proteins stimulated IL-12p40, the inducible common subunit of IL-12 and IL-23 in BMDC. Both of these cytokines that are produced by DC, monocytes and macrophages stimulate TH1 immune responses. BMDC secreted abundant IL-12 p40 upon exposure to EF-60. The ability of each recombinant E. faecalis protein to induce IL-12 p40 was directly related to its ability to stimulate IFNγ. As noted above, these three clones were chosen because of their structural potential to serve as adjuvants.
Because these recombinant proteins were produced in E. coli, it was necessary to exclude contamination by LPS, which can itself activate innate immune responses through toll-like receptor (TLR)4. We stimulated unfractionated splenocytes from C3H/HeJ (TLR4 defective spontaneous mutant) and control C3H/HeOuJ mice. No difference in IL-12p40 secretion was noted in BMDC from TLR4 defective and normal mice, indicating that there was no contaminating LPS in our preps and that these recombinant proteins did not activate innate immune responses through TLR4.
2. B. vulgatus HSP-60
Bacterial HSP-60 is reported to be a legend for TLR4 and thus have adjuvant activities. We tested these adjuvant activities on splenocytes isolated from normal Fischer F344 rats, the genetic background of our HLA-B27 rat strain. Purified recombinant HSP 60 from B. vulgatus stimulated TNF, IL-1β and IL-12 secretion by unfractionated rat splenocytes in a dose and time-dependent fashion, with optimal responses at 1 μg/ml at 24 hr (Holt, Gastroenterology abstract 2005). A control reverse HSP60 had no effect. TNF secretion was only minimally inhibited by polymyxin B and was partially inactivated by heat and protease (trypsin) treatment, suggesting that recombinant HSP60 is a protein and not contaminated by LPS. We indirectly implicated TLR 4 signaling by showing that purified recombinant BV HSP 60 and B. vulgatus lysates induced TNF by cultured spleen cells from C3H/HeOuJ mice but not C3H/HeJ mice that have defective TLR 4 binding and signaling.
The recombinant E. coli containing the empty vector used to express B. vulgatus did not induce colitis in monoassociated HLA B27 TG rats. Dr. Edson Rocha, East Carolina University, has deleted HSP 60 (BV HSP 60 KO) in B. vulgatus by inserting an erythromycin/clindamycin resistance cassette into this gene. Lysates of this deletion mutant have minimal abilities to stimulate TNF by mouse splenocytes. This mutant will be used to monoassociate GF HLA B27 TG rats to investigate the ability of BV HSP 60 to cause colitis.
Fractionation studies showed that TNF was produced by a nonlymphoid population since depletion of T cells and B lymphocytes with antibody-coated Miltyni beads had very little effect on TNF secretion. Together, these results demonstrate that a dominant serologic antigen from B. vulgatus can activate proinflammatory innate immune responses.
III. Significant results (positive and negative)
A. We have identified at least two dominant antigens by two different strategies from two common commensal enteric bacterial species (E. faecalis and E. vulgatus) that cause colitis in two different genetically engineered rodent models of chronic, T cell-mediated colitis (IL-10-/- mice and HLA-B27TG rats).
B. We have shown that at least one E. faecalis protein has both antigenic and adjuvant activities in IL-10-/- mice CD4+ T cells and DC, respectively.
C. We have demonstrated that a dominant serologic antigen, HSP60, from B. vulgatus can elicit innate immune responses in rat cells through TLR4 ligation.
D. We have created an HSP60 deletion mutant of B. vulgatus and expressed B. vulgatus HSP60 in a non-colitogenic E. coli strain so that we can directly test the hypothesis that B. vulgatus HSP60 causes colitis in monoassociated HLA-B27 TG rats.
E. We have established an expression library of E. faecalis and have identified 19 additional clones that react with E. faecalis monoassociated IL-10-/- mouse serum that will be investigated for their abilities to stimulate TH1 and innate immune responses.
F. We have created a strategy to screen bacterial species by Western blot for dominant epitopes that react with serum from rodents with colitis, then identify reactive bands with combination GLC-mass spectrometry (proteomic approach). This can be applied to any type of enteric commensal species (bacterial or fungus).
G. On the negative side, we were unable to obtain adequate amounts of recombinant Flax and HeJ Bir1 flagellin to determine the role of this antigen/adjuvant in our IL-10-/- mice, as originally proposed. We hope to pursue this line of research in the near future once restrictions on use of these materials are lifted.
IV. Lay summary
In the past two years, we have made substantial progress toward identifying several dominant bacterial components that cause persistent immune activation that leads to chronic colitis in genetically susceptible hosts, but protective responses in normal hosts. Our previous work has demonstrated that normal bacteria are required for chronic intestinal inflammation to develop in genetically susceptible hosts and that different bacterial species uniquely cause colitis in different genetically engineered rats and mice. For example, Bacteroides vulgatus can induce colitis in rats that are engineered to express the human gene, HLA-B27, but other common gut bacteria, including E. coli and Enterococcus faecalis, do not cause disease in this model. In contrast, Enterococcus faecalis and E. coli cause T lymphocyte activation and colitis in mice deficient in the immunoprotective molecule, interleukin 10 (IL-10 knockout mice), but Bacteroides vulgatus has no detrimental effect. We have identified two proteins, one in Bacteroides vulgatus and the other in Enterococcus faecalis, that selectively stimulate immune responses in these transgenic and knockout rodents with colitis. The Enterococcus faecalis protein activates both innate and adaptive immune responses. Innate immune cells rapidly respond to certain bacterial signals and produce mediators that activate a whole host of inflammatory processes. Adaptive immune responses include production of antibody that circulate as well as T lymphocyte responses that cause injury by cell contact or secreting inflammatory proteins. Both adaptive and innate immune responses are necessary to cause chronic inflammation, including ulcerative colitis, Crohn’s disease and experimental colitis in IL-10 knockout mice and HLA B27 transgenic rats. The Bacteroides vulgatus protein (heat shock protein 60) activates innate immune cells to secrete proinflammatory mediators and stimulates antibody responses, but does not elicit detectible T lymphocyte responses.
These studies are important because these proteins could lead to better diagnostic tests for human Crohn’s disease and ulcerative colitis and may allow us to better identify subsets of IBD patients that may respond selectively to different treatments, such as antibiotics to eliminate the dominant bacterial species that stimulates inflammation. In addition, these studies help us understand the mechanisms by which susceptible hosts respond to discrete bacterial antigens and lay the foundation for inducing protective immune responses to bacterial components by immunizing IBD patients with novel bacterial antigens, while manipulating the immune system to induce protective rather than detrimental patterns of responses.