Final Progress Report
Proposal No. IBD-0021
Principal Investigator: Hudson Freeze, Ph.D.
Applicant Organization: The Burnham Institute (La Jolla, California, U.S.A.)
Project Title: An anti-glycan antibody (mAbGB3.1) as a novel therapy for Crohn's disease and ulcerative colitis
Period of Award: August 1, 2002 - March 31, 2005
A. SUMMARY OF PROJECT AIMS
Crohn's disease (CD) and ulcerative colitis (UC) are chronic, debilitating, and multi-factorial inflammatory bowel diseases (IBD). It is now widely accepted that these diseases result from abnormal mucosal immune responses to enteric flora. However, little information is known about the initial signaling events in the pathogenesis as well as mechanisms that sustain inflammation in IBD.
We earlier identified N-linked glycans carrying novel carboxylated glycans on their outer branches. The glycans are found on endothelial cells and antigen presenting cells (APC), and are recognized by inflammatory proteins HMGB-1, S100A8/S100A9, S100A12 and annexin-1. S100A12 and HMGB-1 bind to RAGE (Receptor for Advanced Glycation End Products), a cell surface signaling receptor implicated in many inflammatory disorders. A subpopulation of RAGE molecules carries the glycans, and this population has especially potent ligand binding ability. Deglycosylation of RAGE reduces S100A12 and HMGB-1 binding, suggesting that ligand binding is glycan dependent. RAGE-ligand interactions result in an activation of NF-κB, increased expression of cytokines and adhesion molecules, and promote colitis. We hypothesized that blocking this glycan-dependent interaction would alleviate colitis.
Our aims during the first and second BMRP funding periods were to:
1. Test the hypothesis that the antibody mAbGB3.1 against carboxylated oligosaccharides may be a potential therapeutic for IBD by testing it in three different mouse models of colitis.
2. Determine whether mAbGB3.1 can rescue mice that have already developed colitis.
3. Determine whether any protective effects of mAbGB3.1 depend on RAGE or another molecule that carries these glycans.
4. Determine whether any protective effects of mAbGB3.1 are T cell dependent by testing the antibody in DSS colitis induced in Rag-/- mice.
Changes in Research Plan
During the funding period, we focused mainly on the adoptive transfer model of colitis. Due to the complexity and time requirements of establishing and maintaining this model, we requested elimination of testing the antibody on the other proposed models (DSS colitis and IL-10 knockout mice). This strategy was considered meritorious by the BMRP Director and the grant reviewers for the following reasons: 1) Focusing on one model to gather extensive preliminary data for supporting a clear hypothesis so that the work becomes competitive for national funding and 2) to clearly document the success of the antibody in one model and identify players and targets so that it becomes available for clinical trials and practice. In fact, we made substantial progress in understanding the role of the glycans in this model of colitis.
B. ACCOMPLISHMENTS (AIMS 1 - 3)
Studies on tissues from patients with Crohn's disease
In tissues from Crohn's disease patients, we found that the anti-glycan antibody mAbGB3.1 epitope is found in inflammatory infiltrates and serosal and submucosal aggregates of macrophages. The infiltrating macrophage lesions in Crohn's tissues also stain positive for S100A12 and S100A9, while the S100 proteins are not detected in tissues from healthy controls, in agreement with other studies. Because S100A12 and S100A9 bind to carboxylated glycans recognized by mAbGB3.1, this suggested upregulation and co-localization of the ligand-receptor pair in inflamed regions of tissue in the patients.
Studies in transfer model of colitis
Expression of glycan antibody epitope, RAGE and RAGE ligands in normal mice
Using immunohistochemistry and flow cytometric analysis, we detected strong anti-glycan antibody reactivity in colonic lamina propria cells of non-reconstituted Rag-/- mice on CD11c+ and CD205+ dendritic cells (DC) and CD11b+ macrophages. RAGE was also expressed in colonic lamina propria cells and co-localized with anti-glycan antibody reactivity. Immunoblotting showed the presence of S100A9 in cells of colonic lamina propria, and S100A8 and S100A9 in spleens of non-reconstituted Rag-/- mice. We also found HMGB-1, another RAGE ligand, in spleen, MLN and colonic lamina propria of these mice.
Anti-glycan antibody treatment blocks onset of colitis in the adoptive cell transfer model
To further explore the role of the glycans in IBD, we tested the anti-carboxylate glycan antibody in the CD4+CD45RBhigh T cell transfer model of colitis. Animals were injected with a non-blocking control antibody or anti-glycan antibody once every ten days and monitored for approximately seven weeks following cell transfer (half-life of the antibodies in circulation was approximately seven - eight days). Reconstituted mice untreated or treated with the control antibody lost an average of 20-25% of their initial body weight and suffered from severe diarrhea. Mice treated with the anti-glycan antibody showed minimal weight loss, did not have any diarrhea, and remained healthy throughout the treatment period. Histopathological examination of the colons revealed marked mucosal hyperplasia, goblet cell depletion, distortion of crypt architecture, and extensive inflammatory cell infiltration with occasional crypt abscesses in untreated or control antibody treated mice, while in two-thirds of anti-glycan antibody treated mice colonic architecture was intact with minimal or no inflammatory changes.
Anti-glycan antibody treatment initiated after cell transfer blocks progress of disease
In the transfer model, CD4+ T cell colonization in lamina propria is apparent eight -11 days after reconstitution, along with an early inflammatory infiltration. Severe colitis is established by 21 days post transplantation. We therefore evaluated whether antibody administration after initiation of colitis would block or reverse the disease. Mice received the antibodies weekly starting either ten days or 21 days after T cell transfer and sacrificed seven weeks post transfer. Anti-glycan antibody treatment started 10 days after cell transfer significantly impaired progression of disease. This recovery was accompanied by reduced or no weight loss, good health, and reduced inflammation scores. Splenomegaly observed in reconstituted, untreated or control antibody treated mice was also absent in all reconstituted mice treated with anti-glycan antibody. However, administration of the anti-glycan antibody from 21 days after cell transfer did not prevent weight loss or symptoms of disease (not shown). This suggested that the antibody blocked an early step in colitis pathogenesis.
In the transfer model, lymphocyte reconstitution is associated with an initial homeostatic as well as antigen driven expansion of CD4+ T cells in secondary lymphoid organs. This is followed by generation of organ-tropic T cells with a Th1 cytokine profile. Since APC play an important role in T cell maturation and activation, the anti-glycan antibody may block one or more of APC mediated pathways early in T cell pathology. To understand the mechanisms underlying the protective effects of the anti-glycan antibody, we therefore carried out systematic examination of tissues of mice (seven weeks or three weeks after reconstitution) for cellular accumulation and expression of cytokines and other molecules.
Leukocyte accumulation is reduced in the colon of antibody treated mice
We first examined accumulation of T cells in different compartments of reconstituted mice. At seven weeks post transfer, CD4+ T cell infiltration was found throughout the intestine and colon in recipient mice, and this was unaffected by control antibody treatment. In mice treated with the anti-glycan antibody, there was a marked reduction in the accumulation of CD4+ T cells in the lamina propria of the colon. This effect was site specific, because accumulation of T cells in the lung, and small intestine, was unaffected by anti-glycan antibody treatment. This suggested that the colonic microenvironment, whose critical feature is the presence of a rich bacterial flora, contributes to the antibody-mediated effects.
We also examined CD4+ T cell accumulation three weeks after T cell reconstitution. Compared to mice treated with the control antibody, the anti-glycan antibody significantly reduced accumulation of CD4+ T cells in the colonic lamina propria, but it did not affect accumulation in the spleen and mesenteric lymph nodes (MLN). This indicated that the anti-glycan antibody treatment affected early as well as late accumulation of T cells in the colon.
Next, we examined colon tissues of reconstituted mice for accumulation of inflammatory cells seven weeks post transfer. In non-reconstituted mice, CD11b+ and F4/80+ myeloid cells were distributed throughout lamina propria of both the proximal and distal colon. The expanded lamina propria of reconstituted mice untreated or treated with control antibody contained significant inflammatory infiltrates consisting predominantly of CD11b+ and F4/80+ macrophages and a smaller number of neutrophils in crypt abscesses. Clusters of activated macrophages were confined to lamina propria, but in severe disease they also extended to the submucosa. In reconstituted mice treated with anti-glycan antibody, accumulation of CD11b+ and F4/80+ cells was markedly reduced.
Anti-glycan antibody reduces expression of mucosal addressin cell adhesion molecule 1 (MAdCAM-1)
MAdCAM-1 and the gut-homing integrin α4β7 are important for recruitment of T cells to the inflamed colonic tissues. We therefore examined if reduced T cell accumulation in colons of anti-glycan antibody treated mice was due to reduced frequency of gut tropic T cells or reduced expression of MAdCAM-1. Analysis of T cells from MLN of anti-glycan antibody treated mice at three weeks post transfer showed no reduction in the frequency of α4β7+ cells. However, markedly increased expression of MAdCAM-1 was seen in inflamed tissues from untreated or control antibody treated mice, while the addressin expression on venules in colon of anti-glycan antibody treated mice was similar to constitutive expression seen in non-reconstituted mice. This suggested that the anti glycan antibody could block the infiltration of gut tropic T cells to the colon by decreasing MAdCAM-1 expression.
Pro-inflammatory cytokines are reduced in tissues and blood of anti-glycan antibody treated mice
TNFα and IFNγ are also important mediators of disease in the adoptive transfer model. Th1 cells produce IFNγ, while TNFα is a product of both Th1 cells and macrophages that are activated by them. To investigate the role of the anti-glycan antibody in blocking activation of T cells and effector functions of Th1 cells, we analyzed expression of pro-inflammatory cytokines in tissues and sera of mice.
At seven weeks post transfer in reconstituted and untreated or control antibody treated mice, increased IFNγ expression was associated with the presence of increased CD4+ lymphocytes in the lamina propria and submucosa, while in the colon of anti-glycan antibody treated mice, IFNγ expression was minimal. Anti-glycan antibody treatment also reduced the frequency of IFNg positive lymphocytes in spleen examined three weeks post transfer.
At seven weeks post transfer, TNFα was highly expressed in colonic lamina propria in reconstituted and untreated or control antibody treated mice. TNFα expression showed temporal correlation with macrophage (CD11b+/F4/80+) infiltration, suggesting that macrophages are the principal source of TNFα. Expression of TNFα in colonic tissue were greatly reduced in anti-glycan antibody treated mice. We examined the efficacy of antibody therapeutic modality on the production of TNFα in colitis. TNFα levels in blood were elevated as early as ten days post transfer and was maximum seven weeks post transfer. When anti-glycan antibody treatment was initiated ten days after T cell transfer, TNFα measured seven weeks post transfer was reduced to the levels seen in non-reconstituted mice.
Anti-glycan antibody blocks pro-inflammatory cytokine production by macrophages in vitro and induces their apoptosis
To further confirm that anti-glycan antibody treatment downregulates pro-inflammatory cytokine production in colon, we analyzed the effects of the antibody on LPS-elicited production of cytokines by macrophages in vitro. LPS interacts with CD14 and the Toll-like receptor-4 (TLR4) complex on macrophages to activate multiple signaling pathways and secretion of pro-inflammatory cytokines IL-12, TNFα and IL-1, as well as anti-inflammatory cytokines, including IL-10. Anti-glycan antibody reactive epitope is constitutively expressed on RAW264.7 macrophages (not shown). We therefore activated these macrophages in vitro with LPS in the presence or absence of anti-glycan antibody or the isotype control antibody. LPS increased TNFα mRNA production in untreated cells by three-fold. This was followed by an enhanced secretion of TNFα into the culture medium. The anti-glycan antibody significantly reduced LPS-stimulated TNFα. It also inhibited LPS-induced production of IL-23 mRNA, and secretion of IL-12, and NO, while it had no effect on the production of IL-10. In addition, treatment of LPS-activated cells with the anti-glycan led to increased apoptosis of the macrophages, as determined by cellular growth and annexin V staining. This was activation-dependent, because the antibody had no effect on the growth of unstimulated cells. These results suggested that the anti-glycan antibody might specifically block pro-inflammatory but not anti-inflammatory cytokine production and may also promote apoptosis of infiltrating macrophages in colitis.
RAGE and carboxylated glycan binding proteins S100A8, S100A9 and HMGB-1 are upregulated in colitis
Both IFNγ and TNFα stimulate macrophage expression of S100A8, and S100A9 expression is upregulated in activated spleen cells. S100A8 is chemotactic in mice and S100A9 promotes integrin-mediated adhesion of phagocytes. Together they may provide a strong stimulus for early infiltration of neutrophils and monocytes. We therefore examined if S100A8 and S100A9 are upregulated in inflamed tissues of reconstituted mice. Immunoblot analysis showed that S100A8 and S100A9 are strongly upregulated in spleen and colonic lamina propria of reconstituted and untreated or control antibody treated mice examined three weeks post-transfer, but not in anti-glycan antibody treated mice. We also found upregulation of RAGE in the spleen of untreated and control antibody treated mice, but not in anti-glycan antibody treated mice. The two forms of RAGE seen in spleen may represent proteins encoded by alternatively spliced mRNAs. In addition, we also detected HMGB-1 in the sera of mice with inflammation, while it was undetectable in unreconstituted mice.
Anti-glycan antibody reduces activation of NF-κB p65
Since HMGB-1, S100A8, S100A9 and S100A12 are all strong inducers of NF-κB, the above results suggested that NF-κB activation could play a role in carboxylated glycan mediated effects. RAGE promoter has two NF-κB binding sites and activation of NF-κB by RAGE ligation results in upregulation of the receptor, thus amplifying the signal and initiating a pathological cycle of cellular perturbation.
At three weeks post-transfer, NF-κB p65 was highly upregulated in lamina propria cells of recipients untreated or treated with control antibody, but was significantly reduced in cells from anti-glycan antibody treated mice. In vitro, LPS stimulated RAW264.7 macrophages showed significant activation of NF-κB, which was blocked by anti-glycan antibody treatment. We observed inhibition of activation of NF-κB p65, but not that of other NF-κB /Rel family members. This finding is interesting since NF-κB p65 is increased in colonic lamina propria of Crohn's disease patients, and administration of p65 antisense oligonucleotides reduces clinical and histological signs of TNBS or DSS- induced colitis. Thus, inhibition of NF-κB activation could represent a central mechanism by which the anti-glycan antibody blocks colitis.
Carboxylated glycans are expressed on a subpopulation of cells from secondary lymphoid organs of RAGE KO mice
To begin to assess the requirement of RAGE in the immune responses mediated by the carboxylated glycans, we obtained RAGE -/- mice from Dr. Angelika Bierhaus, University of Heidelberg. We bred our colony here and genotyped by PCR using specific probe. In preliminary experiments, we analyzed cells from spleen and MLN of null, heterozygous, and wild type littermates by staining with mAbGB3.1 or anti-RAGE. We found two populations of cells in wildtype organs, RAGE+/GB3.1+ and RAGE-/GB3.1+. Loss of RAGE in null mice led to loss of mAbGB3.1 reactivity in the first population of cells, but not the second. This suggests that other molecules besides RAGE are modified by the glycans and are expressed on immune cells.
CD4+CD45RBhi T cells from RAGE KO mice induce colitis
Lymphocytes express RAGE. To check whether naïve CD4+CD45RBhi lymphocytes from a RAGE-/- mouse can potentiate disease, in pilot experiments we tested them in Rag-/- mice. RAGE-/- lymphocytes were as effective as wildtype donor T cells in eliciting disease. This brings up the next question, whether RAGE expressed by the host is important in initiating and sustaining disease.
Summary
Our studies show that carboxylated glycans may play a critical role in early events of colitis pathology by blocking NF-κB activation. However, the specific events and mediators remain unknown. The antibody may block either a single step or multiple steps during the initiation or recurrence of disease. Likely early targets include DC maturation, DC-T cell interactions, proliferation and polarization of T cells, homing, and infiltration and survival of inflammatory cells into the colon. The sugar chains may mediate multiple interactions involving different cell types and ligand-receptor pairs, such as RAGE and its ligands, and a network, rather than a linear sequence in the activation of NF-κB. We hope to find answers to these questions in future studies.
C. SIGNIFICANT RESULTS
In summary, our findings indicate that carboxylated glycans expressed on APC and other cells may play critical roles in the initiation and progression of colitis. mAbGB3.1, a monoclonal antibody that recognizes the glycans blocks onset of colitis in an adaptive transfer model, and it reverses colitis in the emerging phase of disease by blocking NF-κB activation. Our initial studies have suggested reasonable cell types and targets. Our findings also suggest that other cell surface receptors besides RAGE could contribute to the signaling events. Further characterizing and targeting of the carboxylated glycan-dependent pathway(s) may be a promising new approach to the treatment of human IBD.
D. LAY SUMMARY OF REPORT
How ulcerative colitis and Crohn's disease arise is not completely understood, but it is probably a multi-step process. Treatments that block one or more steps along the way may serve as therapeutics. Our lab studies sugar chains on proteins. These sugar chains affect the way proteins behave, recognize and bind to each other and send signals inside the cells. We found a novel kind of sugar chain on cells and found that it promotes the binding of protein receptors and their partners. This interaction leads to inflammation in mice. We made an antibody that binds to the sugar chains, thus blocking the two proteins from binding to each other. Using funding from the Broad Medical Research Program, we tested the antibody in a mouse model of colitis. We obtained dramatic results. Not only was the antibody able to prevent the development of IBD-like disease in mice, but also reverse the symptoms in early stage of disease. We were able to identify reasonable cell types and molecules involved in the process. We believe that the sugar chains may mediate multiple interactions involving different cell types and protein-pairs. Ultimately, we want to understand how these sugar chains mediate the initiation and propagation of IBD pathology.