Cytotoxic-T-Lymphocyte-Associated Antigen 4 Blockade Abrogates Protection by Regulatory T Cells in a Mouse Model of Microbially Induced Innate Immune-Driven Colitis

Abstract

Cytotoxic-T-lymphocyte-associated antigen 4 (CTLA-4) expressed at high levels by CD4⁺ CD25⁺ CD45RB^low regulatory T cells (Treg) is essential to their homeostatic and immunoregulatory functions. However, its relevance to anti-inflammatory roles of Treg in the context of colitogenic innate immune response during pathogenic bacterial infections has not been examined. We showed earlier in Rag2-deficient 129/SvEv mice that Treg cells are capable of suppressing colitis and colon cancer triggered by Helicobacter hepaticus, a widespread murine enterohepatic pathogen. Using this model, we now examined the effects of antibody blockade of CTLA-4 on Treg function during innate immune inflammatory response. Consistent with our previous findings, we found that a single adoptive transfer of Treg cells prior to infection prevented colitis development despite persistent H. hepaticus infection in recipient mice. However, when infected mice were injected with anti-CTLA-4 antibody along with Treg cell transfer, they developed a severe acute colitis with poor body condition that was not observed in Rag2^−/− mice without Treg cell transfer. Despite high numbers of Foxp3⁺ Treg cells, evident by immunohistochemical analyses in situ, the CTLA-4 antibody-treated mice had severely inflamed colonic mucosa and increased rather than decreased expression levels of cytokines gamma interferon and interleukin-2. These findings indicate that antibody blockade of CTLA-4 clearly abrogates Treg cell ability to suppress innate immune-driven colitis and suggest that Treg cell CTLA-4 cognate interactions may be necessary to maintain homeostasis among cells of innate immunity.

Inflammatory bowel disease (IBD), ulcerative colitis, and Crohn's disease are characterized by chronic and persistent colitis with a relapsing and remitting clinical course. Patients with extended and early-onset colonic inflammation have a heightened risk of developing epithelial dysplasia and intestinal cancer (19, 26, 27). Efforts on modulation of intestinal microflora by antibiotics (10) and probiotics (25) have led to the recognition that enteric bacteria are involved in pathogenesis of IBD. We and others have shown that an enteric pathogen, Helicobacter hepaticus, induces colitis and subsequent colon cancer in 129/SvEv Rag2-deficient mice lacking functional T and B lymphocytes (2, 3, 28). However, immunocompetent wild-type mice of the same genetic background are generally resistant to the proinflammatory effects of H. hepaticus infection (2). These findings suggest that a subset of endogenous lymphocytes have a crucial role in regulating the host innate responses to pathogenic enteric bacteria. Indeed, adoptive-transfer experiments have revealed that a subset of CD4⁺ lymphocytes with anti-inflammatory roles (CD45RB^low, CD25⁺, Treg cells) can prevent and/or regulate the colitogenic innate immune responses against H. hepaticus infection (2, 3).

Treg cells express constitutively on their surface high levels of cytotoxic-T-lymphocyte-associated antigen 4 (CTLA-4; CD152) in addition to the costimulatory molecule CD28 (23). Although both CTLA-4 and CD28 can bind to B7-1 (CD80) and B7-2 (CD86) on antigen-presenting cells (APCs) with varying affinity, the balance of dual signals delivered to T cells regulates the extent of their activation and subsequent immune response. In comparison, CTLA-4 shows higher affinity for both B7 molecules over CD28 and plays a dominant inhibitory role in limiting T-cell proliferation and interleukin-2 (IL-2) production. Moreover, CTLA-4-mediated suppression in part is thought to be accomplished by competing for stimulatory signals of CD28 (12, 14, 24). Read et al. have shown that in vivo administration of anti-CTLA-4 antibody blocks the suppressor function of CD4⁺CD25⁺ Treg cells on colitis induced in wild-type mice by the adoptive transfer of CD45RB^high effector T cells from syngeneic or B7-1/B7-2/CTLA-4 triple-deficient mice (22). These data suggested that anti-CTLA-4 antibody interfered with CTLA-4/B7 costimulatory signals between Treg cells and APCs but not those between effector T cells and APCs. Despite blocking of CTLA-4 on Treg cells, anti-CTLA-4 antibody treatment did not eliminate Treg cells or their peripheral development, accumulation, or suppressor function (22). In other studies, modification of CTLA-4 signaling failed to alter Foxp3 expression in vitro (11). However, the relevance of CTLA-4 to the Treg anti-inflammatory function or their ability to regulate innate immune response against bacterial infection has not been examined before. We hypothesized that blockade of CTLA-4 on Treg cells affects their ability to suppress innate immune inflammation in Rag2-deficient mice. To examine this possibility, we administered CTLA-4 antibody to H. hepaticus-infected 129/SvEv Rag2-deficient mice with or without addition of Treg cells, monitored their persistence in vivo, and assessed the outcome of CTLA-4 blockade on the gut innate immune inflammatory response and dysplasia.

MATERIALS AND METHODS

Mice and bacteria.

Rag2-deficient and wild-type mice on a 129/SvEv strain background were obtained from Taconic Farms. All mice were housed in Association for Assessment and Accreditation of Laboratory Animal Care-approved facilities. The mice were free of known murine viruses, Salmonella spp., Citrobacter rodentium, ecto- and endoparasites, and murine Helicobacter spp. Experimental mice dosed with H. hepaticus were housed separately in a biocontainment area of the animal facility. H. hepaticus (strain 3B1; ATCC 51449) (6) was grown under microaerobic conditions, prepared, and confirmed pure as described previously (2, 6).

Adoptive transfer of Treg cells and antibody treatment.

CD4⁺ lymphocytes were isolated from spleens of wild-type littermates by using magnetic beads (Dynal Biotech USA, Oslo, Norway) and then sorted by high-speed flow cytometry (MoFlow2) to obtain purified populations of CD4⁺ CD25⁺ CD45RB^low lymphocytes (∼96% pure) as described previously (2). The purified Treg calls from Helicobacter-free 129/SvEv donors were injected into syngeneic Rag2-deficient mice intravenously in the retro-orbital sinus with 3.0 × 10⁵ cells/animal suspended in 0.2 ml of media. Three days later, mice were dosed with either H. hepaticus (2.0 × 10⁷ bacteria/animal) or sham media every other day for a total of three doses. Treg recipient mice underwent a 2-week administration with either purified hamster anti-mouse CTLA-4 monoclonal antibody (UC10-4F10-11) or hamster control immunoglobulin at a dose of 100 μg/animal/day intraperitoneally. The administration of antibody started at 1 day before or 4 weeks after Treg transfer. The mice were euthanized 6 weeks after the last dose of anti-CTLA-4 antibody, or otherwise euthanized when the mice developed severe diarrhea and lost up to 20% of their initial body weights. A cohort of aging 129Sv/Ev Rag2-deficient mice remained untreated and served as controls. At the end of these experiments, samples of colon, cecum, ileum, duodenum, stomach, liver, and spleen were harvested at necropsy. Experimental infection of mice was confirmed in cecum samples by using H. hepaticus-specific primers (8). Helicobacter-free status was also confirmed in controls.

Histological evaluation.

Histological findings were evaluated as described previously (2). Briefly, samples of the colon, cecum, ileum, duodenum, stomach, and liver were fixed in formalin, embedded in paraffin, and stained with hematoxylin and eosin. Lesions were scored by a pathologist blinded to sample identity. The colonic and cecal lesions were scored on the basis of size and frequency of inflammatory lesions on a scale of 0 to 4 with ascending severity (grade 0, none; 1, minimal; 2, mild; 3, moderate; and 4, severe). Epithelial dysplasia and neoplasia were graded using a scale of 0 to 4 based on a recently described scheme (2) (grade 0, normal; 1, mild dysplastic changes; 2, low-grade adenoma/dysplasia; 3, high-grade adenoma/dysplasia, carcinoma in situ, or intramucosal carcinoma; and 4, invasive carcinoma). The data were compiled from two replicate experiments.

Immunohistochemistry.

After deparaffinization, formalin-fixed sections were antigen retrieved with pepsin (Zymed, San Francisco, CA) for 10 min at 37°C and labeled with rat monoclonal antibody recognizing mouse antigen (Foxp3; BD Pharmingen, San Diego, CA). Primary antibody binding was detected with species-appropriated biotinylated secondary antibodies (Sigma Chemical Company), streptavidin peroxidase, and 3,3-diaminobenzidine (Vector Laboratories, Burlingame, CA). Immunohistochemical assays were performed on an automated immunostainer (i6000; Biogenex, San Ramon, CA).

Detection of cytokine mRNA and bacterial DNA.

Segments (5 ml) of distal and proximal colon, cecum, terminal ileum, and spleen were collected and snap-frozen in liquid nitrogen. According to the manufacturer's instructions, RNA was extracted by using TRI-Reagent RNA isolation reagent (Sigma-Aldrich). Quantitative PCR was performed using TaqMan gene expression assay kits (Applied Biosystems) to analyze mRNA expression of gamma interferon (IFN-γ), IL-2, IL-10, Foxp3, and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) in a ABI Prism sequence detection system 7700 (Applied Biosystems) (4). The expression levels of target mRNA were normalized to micrograms of mouse chromosomal DNA, whose quantities in the samples were measured with 18S rRNA gene based primers and probe mixture (Applied Biosystems), as described previously (4, 8).

Statistical analyses.

Analyses of histological scores and bacterial DNA and expression levels of mRNA were performed by using a Mann-Whitney U nonparametric test and unpaired t test. Group differences of body weight changes were analyzed by analysis of variance.

RESULTS

CTLA-4 blockade abolished the Treg suppression of H. hepaticus-induced colitis.

Rag2-deficient 129/SvEv control mice, free of H. hepaticus, showed no evidence of colitis (Fig. 1A and 2) during the course of the study. H. hepaticus infection in a cohort of mice, housed in a separate location, led to moderate to severe typhlocolitis and epithelial dysplasia by 8 weeks postinfection (wpi), as shown in Fig. 1B. Furthermore, H. hepaticus infection initially caused a delay in body weight gain (50% [6 of 12 mice]) immediately after bacterial inoculation, but thereafter the animal's body weight recovered steadily (data not shown). Inflammation was not evident in the ilea, duodena, stomachs, or livers of the infected mice (data not shown). A group of H. hepaticus-infected mice (n = 11) that subsequently received CD4⁺ CD25⁺ CD45RB^low cells by adoptive transfer showed no signs of disease and had very little inflammation in their colons, a finding consistent with our prior data on Treg-mediated amelioration and colonic lesions (Fig. 1C). Also, the Treg-recipient mice showed no loss in their body weights unlike their infected counterparts that did not receive Treg cells. To examine the effect of CTLA-4 blockade on the colitis-protective function of Treg cells, anti-CTLA-4 antibody was administered to a group of infected mice (n = 7) that had earlier received Treg cells. Surprisingly, despite their circulating Treg cells, mice injected with CTLA-4 antibody developed a rapidly progressive wasting condition with bloody diarrhea starting as early as 1 wpi. The CTLA-4 antibody regimen could not be continued beyond 2 weeks due to the rapidly deteriorating clinical condition of the mice. The body weight loss continued even after the end of anti-CTLA-4 administration. Although 7 of the 11 mice injected with anti-CTLA-4 regained their body weight to their initial body weight by 4 weeks after the last injection, these mice showed signs of profound colitis, 6 weeks after the last dosing with anti-CTLA-4. The inflammation was characterized by discontinuous and patchy infiltration of mononuclear cells, neutrophils, and eosinophils with severe epithelial dysplasia (Fig. 1D). The remaining four anti-CTLA-4-treated mice (4 of 11) had to be euthanized due to poor body condition by 5 wpi. These mice, too, exhibited severe inflammation and transmural ulcers in the colon (Fig. 1E). Interestingly, despite severe colitis, anti-CTLA-4-administered mice showed no significant levels of inflammation in the ileum, duodenum, stomach, or liver (data not shown). There was no significant difference in the number of H. hepaticus organisms/μg of mouse bowel DNA between the infected Treg recipients that were treated with control immunoglobulin G (IgG) (5.8 × 10⁵ [5.3 × 10² to 1.2 × 10⁸]; median [range]) versus those treated with anti-CTLA-4 (3.1 × 10⁴ [3.6 × 10² to 8.6 × 10⁹]).

FIG. 1.

The Treg-mediated suppression of H. hepaticus-induced colitis in Rag2-deficient 129Sv/Ev mice is abolished by CTLA-4 blockade. Representative sections of mouse colon from various experimental groups are shown. (A) Unremarkably normal section from untreated control mouse free of H. hepaticus; (B) moderate to severe mononuclear and polymorphonuclear cell infiltration with epithelial dysplasia in H. hepaticus-infected mouse; (C) mild inflammation in infected Treg recipient injected with sham antibody; (D) infected Treg recipient injected with anti-CTLA-4 showing severe inflammation and moderate dysplasia; (E) severe inflammation with transmural ulcer from infected Treg recipient mouse that received anti-CTLA-4 and was euthanized at 5 wpi due to wasting and poor body condition; (F) infected Treg recipient injected with sham antibody showing no remarkable mucosal damage; (G) moderate inflamed mucosa from infected mouse injected with anti-CTLA-4 in the presence of Treg cells. n = 6 to 12 mice per group. Scale bars: 100 μm, A, B, C, D, F, and G; 250 μm, E. Hematoxylin and eosin staining was used.

Histologically, the inflammatory response characterized by mononuclear and polymorphonuclear cells was observed in the colon and cecum. To quantify the severity of inflammation among different groups of mice, the extent of inflammatory cell infiltration and epithelial dysplasia was evaluated in the distal, middle, and proximal colons and ceca of mice by two veterinary pathologists blinded to sample identity, according to the method described previously (2). Clearly, as shown in Fig. 3, the mice infected with H. hepaticus had higher levels of inflammation and epithelial dysplasia in their colons and ceca. Mice that received adoptive transfer of Treg cells showed significantly lower (P < 0.05) overall inflammation scores in colon, despite small foci of inflammatory cells seen occasionally in few sections of the infected mice that were transferred with Treg cells. Suppression of inflammation mediated by Treg cells was more complete in colon than in cecal tissue. Finally, the extent of inflammation in both colon and cecum was significantly higher after anti-CTLA-4 treatment in mice infected with H. hepaticus-infected group that had also received Treg cells compared to infected control mice that did not receive Treg cells. A longer observation period is needed to determine whether mice undergoing treatment with anti-CTLA-4 antibody have a similar trend toward increased epithelial dysplasia.

FIG. 3.

Foxp3 staining and expression in the colon. (A) Control mouse; (B) H. hepaticus-infected mouse; (C) H. hepaticus-infected Treg recipient injected with sham antibody. (D to F) Infected Treg recipients injected with anti-CTLA-4 (D); relative Foxp3 expression in distal colon (E) and proximal colon (F) is also shown (n = 6 to 12 mice per group as described in Results). The results in panels E and F are presented as means + the standard errors of the mean (SEM). *, P < 0.05; **, P < 0.01; ***, P < 0.001. Original magnification, ×100.

CTLA-4 antibody required Treg cells to promote inflammation in H. hepaticus-infected mice.

The findings of increased colon pathology and rapid onset of clinical disease in H. hepaticus-infected Treg recipient mice administered with anti-CTLA-4 antibody were unexpected. These signs or lesions observed only in the infected Treg recipients administered anti-CTLA-4 antibody but not control IgG suggest that CTLA-4 blockade of Treg cells had clearly mediated enhancement in disease pathology. It is possible that CTLA-4 antibody could induce alteration in the phenotypic and/or functional profile of Treg cells in vivo in the context of inflammation, as shown by Poutahidis et al. (20). To test the requirement for bacterial challenge, Treg recipient mice maintained under H. hepaticus-free conditions were administered the antibody. Administration of anti-CTLA-4 to the uninfected mice did not induce lower bowel lesions (Fig. 1F and 2), nor did it alter their body weight (data not shown). To determine whether effects of anti-CTLA-4 required Treg cells during the course of infection, we injected anti-CTLA-4 into H. hepaticus-infected Rag2-deficient mice without the addition of Treg cells. Anti-CTLA-4 alone did not significantly accelerate the inflammation (Fig. 1G and 2) or affect body weight in H. hepaticus-infected Rag2-deficient mice. These findings confirmed that H. hepaticus infection, Treg cells, and CTLA-4 blockade were all essential to exacerbate clinical disease. These findings also further excluded the possible role for antibody contaminants such as endotoxin in this process.

FIG. 2.

Scores of inflammation and epithelial dysplasia. Inflammatory lesions at 8 wpi were scored on a scale of 0 to 4 with ascending severity (grade: 0, none; 1, minimal; 2, mild; 3, moderate; 4, severe). Epithelial dysplasia and neoplasia were graded by using a scale of 0 to 4 (grade: 0, normal; 1, mild dysplastic changes; 2, low-grade adenoma/dysplasia; 3, high-grade adenoma/dysplasia, carcinoma in situ, or intramucosal carcinoma; 4, invasive carcinoma). Horizontal lines show the median of each group. Four of eleven H. hepaticus-infected Treg recipient mice injected with anti-CTLA-4 were euthanized before completion of experimental schedule because of wasting disease. These mice were excluded from the analysis. *, P < 0.05; **, P < 0.01.

Previously, we have shown that Treg cells once transferred persist and retain their ability to suppress colitis for several weeks. However, it was unclear whether CTLA-4 blockade at 4 weeks after Treg cell transfer and infection still affected their suppressor function. Thus, we administered CTLA-4 antibody (n = 6) or control IgG (n = 6) separately to two groups of H. hepaticus-infected Treg recipients starting at 4 weeks after Treg transfer and monitored their body condition and development of clinical disease. The mice receiving anti-CTLA-4 later in the course of infection failed to gain body weight, and this late intervention was still effective at abolishing the Treg suppressive effect on inflammation in the colon for an additional 6 weeks after the final dose of anti-CTLA-4 (data not shown).

CTLA-4 blockade led to inflammation in colon despite accumulation of Foxp3⁺ cells.

The findings that CTLA-4 blockade abrogated the anti-inflammatory function by Treg cells suggested that their colitis-protective role depends on signaling via CTLA-4. To localize and confirm the presence of Treg cells, staining for Foxp3, a marker specifically expressed by CD25⁺ Treg cells, was undertaken. As shown in Fig. 3 and 4, Foxp3⁺ cells were evident throughout the gastrointestinal tissues in H. hepaticus-infected Treg recipient mice. As expected, Rag2-deficient mice, whether infected or uninfected, had no Foxp3⁺ cells. Notably, the extent of infiltration by Foxp3⁺ cells was higher in Treg recipient mice that were subjected to the CTLA-4 blockade regimen (Fig. 3). There was increased expression of Foxp3 evident in colons in situ; however, Foxp3⁺ cells in the mucosa of ceca, ilea, duodena, and stomachs of the mice injected with anti-CTLA-4 were comparable to those of mice treated with control immunoglobulin (Fig. 4). Foxp3⁺ cells in lymph nodes adjacent to colons and stomachs were also observed in anti-CTLA-4-administered mice (data not shown). The immunohistochemical findings matched the quantitative analysis of Foxp3 using real-time PCR assays, which indicated that colonic Foxp3 expression was increased in Treg cell recipient mice given anti-CTLA-4 (Fig. 3E and F). Foxp3 mRNA expression levels were low in the ceca, ilea, and spleens and unaffected by CTLA-4 administration (Fig. 4Q, R, and S).

FIG. 4.

Foxp3 staining in the cecum, terminal ileum, duodenum, and stomach and Foxp3 gene expression in the cecum, ileum, and spleen. Quantitative PCR expression assays were performed on 5-mm segments of tissue from each region of bowel by using a TaqMan gene expression assay kits. The expression levels of target mRNA were normalized to micrograms of mouse chromosomal DNA. (A, E, I, and M) Control mice; (B, F, J, and N) H. hepaticus-infected mice; (C, G, K, and O) H. hepaticus-infected Treg recipients injected with sham antibody; (D, H, L, and P) H. hepaticus-infected Treg recipients injected with anti-CTLA-4; (Q and R) relative Foxp3 expression to control mice in the cecum and ileum, respectively; (S) relative Foxp3 expression to infected Treg recipients injected with sham antibody. ND, not detectable. n = 6 mice per group. The results are presented as means + the SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Original magnification, ×100.

CTLA-4 blockade led to increased IFN-γ and IL-2 but not IL-10 expression in colon of H. hepaticus-infected Treg recipient mice.

To further elucidate the basis for loss of disease suppression in mice given anti-CTLA-4, we measured IFN-γ, IL-2, and IL-10 expression levels in the colon, cecum, and spleen tissues. As shown in Fig. 5, H. hepaticus infection led to an increase in IFN-γ expression in the colon, cecum, and spleen. Suppression of IFN-γ and other cytokines was evident in corresponding tissues from mice that received Treg cells. In contrast, however, CTLA-4 blockade resulted in a remarkable increase of IFN-γ mRNA expression especially in the colon. In contrast, IL-2 expression in the colons, ceca, and spleens of Rag2-deficient mice was unaffected by H. hepaticus infection or the addition of Treg cells since Rag2-deficient mice lack functional lymphocytes that are a major source of IL-2, and the adoptively transferred Treg cells at best produce only small quantities of IL-2 (1, 5, 7). Administration of anti-CTLA-4 antibody resulted in higher levels of IL-2 expression in colon tissues compared to cecal tissues from H. hepaticus-infected Treg recipient mice. Expression levels of IL-10, a key immunoregulatory cytokine expressed by cells of both innate and acquired immunity, were elevated after H. hepaticus infection in the colon, cecum, ileum, and spleen. Neither the adoptive transfer of Treg cells nor CTLA-4 blockade affected IL-10 expression in these tissues.

FIG. 5.

Gene expression of IFN-γ, IL-2, and IL-10 relative to control mice in ascending and descending colon, cecum, and spleen. Gene expression was assayed by quantitative RT-PCR using TaqMan gene expression assay kits. The expression levels of target mRNA were normalized to micrograms of mouse chromosomal DNA. n = 6 mice per group. The results are presented as means + the SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

DISCUSSION

The profound influence of CTLA-4 on T-cell immunity has been known for over a decade; however, the precise roles played by this molecule continue to emerge. Evidence that CTLA-4-deficient mice die by 3 to 4 weeks of age from extensive lymphoproliferative disease, along with immune-mediated damage to multiple organs, indicates its essential role in negative regulation of the CD4⁺ helper T-cell compartment. That blockade of CTLA-4 on CD25⁺ Treg cells abrogates their homeostatic role on helper T cells, as demonstrated in the CD45RB^hi T-cell transfer mouse model, highlights its role in the suppressor function of regulatory T cells. However, the relevance of CTLA-4 to the suppressor function of regulatory T cells in isolation, when inflammation is generated by cells of innate immunity, has not been examined until now. Growing attention to roles of microbiota and chronic inflammation in initiating and sustaining cancer development led us to test the outcome of CTLA-4 blockade on the development of bacterially triggered colitogenic inflammatory response and dysplasia in a well-characterized Rag2^−/− murine model.

Our previously established model of innate immune-driven colitis triggered by H. hepaticus made it feasible to test the relevancy of CTLA-4 on Treg functions in the absence of other lymphocytes. We find that CTLA-4 blockade clearly disables the anti-inflammatory functions of Treg cells. Interestingly, CTLA-4 antibody administration in H. hepaticus-infected Treg recipients led to an acute colitis and clinical disease that was more severe than innate immune-driven colitis alone. One possible explanation is that adoptively transferred cells included incompletely differentiated CD4⁺ cells that assumed a proinflammatory phenotype in the absence of suppressive functions by Treg cells. Even small numbers of non-Treg CD4⁺ cells present during adoptive transfer may expand rapidly in a Rag2-deficient host. An alternative is that CTLA-4 blockade stimulates cells of innate immunity to increase production of proinflammatory cytokines. The observation that disruption of Treg cell function, in this case blockade of CTLA-4, induces a proinflammatory phenotype in mice raises the possibility that CTLA-4 blockade may predispose to the development of inflammation-associated cancer, as previously described in mice with Treg cells lacking IL-10 (20). When extrapolated to lymphopenic human subjects, these data suggest that CTLA-4 blockade or dysfunction of any remaining lymphocytes may exacerbate ongoing inflammatory responses and precipitate acute clinical disease.

In the present study, adoptive transfer of Treg cells into Rag2^−/− mice suppressed the induction of IFN-γ in response to H. hepaticus. In these Rag2^−/− mice, H. hepaticus infection induced expression of a key inflammatory cytokine, IFN-γ, that was correlated with a severe inflammatory host response. Although Treg cells showed strong suppression of IFN-γ, CTLA-4 blockade resulted in abolishment of Treg suppression on IFN-γ expression and the induction of colon-specific pathology. Histopathology resembled features of colitis induced in Rag2^−/− mice after transfer of naive CD4⁺ CD45RB^high T cells (15). The finding that Foxp3⁺ cells were evident only in inflammatory foci in the gastrointestinal tract and only after bacterial challenge matched findings of Poutahidis et al. (20). Localization of Treg cells in the gastrointestinal tract matched the expectation that Treg cells traffic and function in the periphery in active inflammatory foci.

Recent studies have shown that CD4⁺ Treg cells are also increased in inflamed mucosa during IBD (16) and in H. pylori-infected gastritis patients (21), even though these cells express Foxp3 and retain their suppressor capacity ex vivo (16). Thus, uncontrolled inflammation in IBD may be unrelated to a failure of Treg to localize in affected tissue but rather due to a defect in the ability of Treg to function properly in the inflammatory microenvironment. Our data implied that the disruption of CTLA-4 costimulatory signal may be involved in the impaired Treg functions and may play a role in the pathogenesis of inflammatory diseases. It is possible that, as a consequence of CTLA-4 blockade, CD28-dominant signal stimulates innate immunity to produce proinflammatory cytokines. CD28 costimulation can deliver stimulatory signals not only to T cells to proliferate but also to dendritic cells to promote proinflammatory cytokine production, such as IFN-γ and IL-6 (17, 18, 29). Treg cells with insufficient CTLA-4 may stimulate innate immune cells to increase inflammation in order to promote bacterial clearance during an acute bacterial infection.

Recent studies have shown that APCs have an ability to produce IL-2 in response to bacterial stimulation (9). However, H. hepaticus infection alone was not sufficient to induce IL-2 expression in colon of Rag2-deficient mice. The addition of anti-CTLA-4 to H. hepaticus-infected Treg recipients resulted in a remarkable increase of IL-2 expression in colon, even though donor Treg cells are typically unable to produce IL-2 (1, 7). Interestingly, IL-2 production was increased after H. hepaticus infection only in the colon and not in the cecum or the spleen. Further studies of the effect of CTLA-4 signaling on IFN-γ and IL-2 regulation are required but will be challenging because ex vivo data may not always reliably parallel relevant pathogenic in vivo events.

An unexplained finding is the inconsistency between inflammatory cell infiltrates and gene expression observed in the ceca of H. hepaticus-infected Treg recipient mice at 8 wpi. Despite cellular infiltration in ceca of some Treg cell recipient mice, IFN-γ transcription in the ceca was consistently suppressed after Treg cell transfer. One possible explanation for the discrepancy between gene expression and pathology is that ceca exhibited discontinuous inflammation that was misrepresented in our assays. Discontinuous inflammation in the appendiceal orifice is often noted in IBD patients, but its presence does not correlate with clinical activity (13). Another explanation for the mismatch between IFN-γ and IBD is that eventual suppression of histopathology may temporally lag behind suppression of cytokine transcriptional levels (i.e., taking 12 weeks instead of 8 weeks after infection). Indeed, we previously reported that Treg cell transfer had statistically lowered the inflammation scores but did not completely eradicate IBD in H. hepaticus-infected mice at 12 wpi. In those studies, a subset of Treg recipient mice infected with H. hepaticus had localized inflammation (2). Further analysis of histological and immunological responses in the appendices of humans and ceca of mice in the progressions of IBD is warranted.

Finally, it remains to be determined how IL-10 relates to CTLA-4 in this setting. IL-10 is an anti-inflammatory cytokine that counter-regulates IFN-γ and other proinflammatory cytokines during the development of IBD. Treg cells collected from IL-10-deficient donors, such as Treg cells undergoing blockade of CTLA-4, are incapable of inhibiting colitis triggered by H. hepaticus (3) in Rag2^−/− mice, and similarly develop severe colitis only after bacterial infection (20). In the present study, however, IFN-γ expression was independent of IL-10 expression, suggesting that the CTLA-4-mediated Treg mechanism may be separate from IL-10-mediated functions of Treg cells in this setting.

In summary, these data provide evidence that blockade of CTLA-4 on Treg cells clearly abrogates their suppressive function on colitis induced by H. hepaticus in Rag2-deficient mice. A better understanding of immune regulation involving CTLA-4 and cells of innate immunity will be required as trials with human anti-CTLA-4 antibodies progress into a clinical setting.