Cholera Toxin Prevents Th1-Mediated Autoimmune Disease by Inducing Immune Deviation

Cholera toxin (CT), ² a major enterotoxin produced by Vibrio cholerae, is composed of a monomeric A subunit (CT-A) that is covalently linked to a pentameric B subunit (CT-B). The former causes the diarrhea that typifies the effect of CT on gastrointestinal mucosa, and the latter is responsible for the binding of the toxin to cell surface monosialogangliosides (GM1) that facilitates toxin entry into the cells (1). CT is used as a potent mucosal immunogen as well as potent mucosal adjuvant that stimulates strong responses to mucosally codelivered Ags (1). Its adjuvant effect is characterized by the induction of Ag-specific serum IgG1 and IgG2b, serum and mucosal IgA, and enhancement of CD4⁺ and to a lesser extent CD8⁺ T cell responses (1, 2). A number of studies reported that mucosally delivered CT stimulates a mixed Th1/Th2 response, but with a bias toward Th2 (3, 4, 5). Recent studies showed that oral administration of CT-B-autoantigen conjugates prevented autoimmune diseases such as experimental autoimmune encephalomyelitis and chondritis by inducing functional tolerance (6, 7, 8, 9). Feeding of CT-B-insulin conjugates also abrogated spontaneous autoimmune diabetes in NOD mice (10). In vitro experiments indicated that CT causes the maturation of dendritic cells, up-regulation of the costimulatory molecules B7.1 (CD80) and B7.2 (CD86) on dendritic cells and macrophages, suppression of IL-12 production and IL-12R expression, and enhanced IL-10 production (11, 12, 13).

Although its properties as a mucosal adjuvant are well established, the ability of CT to act as a systemic Th2 adjuvant has not been studied extensively. Experimental autoimmune uveitis (EAU) can be induced in mice by immunization with the retinal autoantigen interphotoreceptor retinoid-binding protein (IRBP) and serves as a model of autoimmune uveitis in humans. EAU pathology is dependent on a Th1 response. Our previous work showed that skewing of the immune response away from Th1 by coadministration of IL-4 and IL-10 concurrently with the uveitogenic immunization can protect from disease development. In this study, we explore the ability of CT to act as a systemic Th2 adjuvant to channel the normally pathogenic response to IRBP toward a nonpathogenic phenotype. Our data indicate that CT treatment concurrently with a uveitogenic challenge elicits a brisk innate IL-4 response and prevents EAU development by skewing the adaptive immune response to the immunizing self Ag away from the Th1 pathway. Our data indicate that promoting a more balanced immune response to the pathogenic autoantigen can have a therapeutic effect and confirm the utility of immune deviation as a therapeutic approach.

Six- to 8-wk-old B10.RIII mice were supplied by The Jackson Laboratory (Bar Harbor, ME). Animals were kept in a specific pathogen-free facility and given water and standard laboratory chow ad libitum. Animal care and use were in compliance with institutional guidelines and with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research.

Recombinant CT, CT-B, and purified pertussis toxin (PT) were purchased from Sigma-Aldrich (St. Louis, MO). CFA was from Difco (Detroit, MI). Rat anti-mouse IL-4 mAb (clone 11B11, IgG1) was from National Cancer Institute Biological Resources Branch. Isotype control mAb Y13 was purified from hybridoma ascites fluid by Harlan Bioproducts (Indianapolis, IN). IRBP was isolated from bovine retinas, as described previously, using Con A-Sepharose affinity chromatography and fast performance liquid chromatography (14). IRBP preparations were aliquoted and stored at −70°C.

EAU in B10.RIII mice was induced by active immunization with 15 μg of IRBP in PBS emulsified 1:1 v/v in CFA that had been supplemented with Mycobacterium tuberculosis strain H37RA (Sigma-Aldrich) to 2.5 mg/ml. A total of 200 μl of emulsion was injected s.c., divided among three sites: base of tail and both thighs. The indicated doses of CT, CT-B, or vehicle (PBS containing 2% normal mouse serum) were given by i.p. injection concurrently with immunization. Clinical EAU was evaluated by fundoscopy under a binocular microscope after dilation of the pupil and was graded on a scale of 0–4 using criteria described in detail elsewhere (15). Eyes harvested 21 days after immunization were prefixed in 4% phosphate-buffered glutaraldehyde for 1 h (to prevent artifactual detachment of the retina), and then transferred to 10% phosphate-buffered formaldehyde until processing. Fixed and dehydrated tissue was embedded in methacrylate, and 4- to 6-μm sections were stained with standard H&E. Eye sections cut through pupillary optic nerve planes were scored in a masked fashion. Severity of EAU were graded on a scale of 0–4 in half-point increments using the criteria described previously, based on the type, number, and size of lesions present (15, 16). Severity of disease was the average score of eyes from those animals in which disease developed (if disease was unilateral, both eyes were averaged).

To assess DTH, 10 μg of IRBP in 10 μl of PBS or 10 μl of PBS alone as control was injected into the ear pinna. Ear thickness increment was measured 48 h later using a spring-loaded micrometer. The specific response was calculated as the difference between ear thickness of the IRBP-injected ear minus PBS control ear.

Draining (inguinal and iliac) lymph nodes were collected after 21 days, and were pooled within the group. Triplicate 0.2-ml cultures containing 5 × 10⁵ cells were seeded in round-bottom 96-well microtiter plates. The RPMI 1640 medium (BioWhittaker, Walkersville, MD) was supplemented with mouse serum, 2-ME, antibiotics, glutamine, and nonessential amino acids, as described (17), and contained 30 μg/ml IRBP as stimulant. The cultures were incubated for a total of 60 h. Tritiated thymidine (1 μCi/well) was added during the last 18 h. The data are shown as cpm.

For tissue culture supernatants, inguinal and iliac lymph node cells from immunized mice were cultured in 24-well flat-bottom plates (5 × 10⁶ cells/1 ml of culture medium per well) either alone or with IRBP at the concentration mentioned above. Supernatants were collected after 48 h and were kept frozen in small aliquots at −70°C. Cytokine concentration in the supernatants was measured by ELISA using Ab pairs from Endogen (Boston, MA) for IL-2, IL-4, IL-5, IL-10, and IFN-γ, or from R&D Systems (Minneapolis, MN) for IL-12 p40, as described previously (18). TGF-β1 was assayed with a kit from Promega (Madison, WI), according to manufacturer’s instructions. For serum levels, B10.RIII mice were injected with 5 μg of CT or 1 μg of PT i.v. Blood was collected by cardiac puncture 2, 6, 12, 24, and 48 h after toxin challenge. Cytokines were measured by multiplex ELISA using the Pierce SearchLight Technology (Pierce Boston Technology, Woburn, MA) (19) (www.searchlightonline.com).

Serum levels of anti-IRBP IgG2a and IgG1 subclasses in IRBP-immunized mice treated with or without CT were determined by ELISA, as previously described (20). Briefly, 96-well microtiter plates (Costar, Cambridge, MA) were coated with IRBP (1 μg/ml). After blocking the plates with BSA and overnight incubation with samples of the tested sera, the plates were developed using HRP-conjugated goat anti-mouse IgG1 or goat anti-mouse IgG2a Abs (Southern Biotechnology Associates, Birmingham, AL). The amount of each isotype bound to the IRBP-coated wells was estimated from standard curves constructed by coating wells with the same goat anti-mouse IgG1 or goat anti-mouse IgG2a Abs and adding dilutions of Ig standards of the pertinent isotype. The IgE determination was performed by a sandwich method using a coating anti-IgE Ab (clone R35-72; BD Pharmingen, San Diego, CA) and a developing rat anti-mouse HRP Ab (Southern Biotechnology Associates). The serum IgE concentration was calculated from a serially diluted standard IgE curve.

IRBP-immunized B10.RIII mice treated or not with CT were injected with 1 mg/mouse anti-IL-4 Ab (11B.11) or control isotype IgG1 (Y13) in 0.1 ml of PBS i.p. on days −1, 1, 3, 5, 7, and 9. Clinical EAU was evaluated by fundoscopy on day 14, and the eyes were harvested on day 21 and processed for histopathology, as described above.

Splenocytes from naive mice (control) or mice immunized with IRBP with or without CT treatment were collected 14 days after immunization and washed in sterile PBS. The cells were resuspended in a final volume of 500 μl of PBS. Cells from naive or from immunized donors at 1:1 donor:recipient ratio were injected i.p. into naive syngeneic recipients irradiated 1 day previously with 150 rad of γ radiation from a cesium source. EAU was induced 3 days after adoptive transfer by immunization with 15 μg of IRBP in CFA. EAU scores in recipients were evaluated 21 days later by histopathology.

Experiments were repeated at least twice. Results were highly reproducible. Figures show pooled data from repeat experiments, or representative experiments, as indicated. Each point is one mouse (average of both eyes). Statistical significance of differences in disease scores was calculated using Snedecor and Cochran’s test (21) for linear trend in proportions, with each mouse (average of both eyes) as one statistical event. This is a nonparametric test that generates its p values by frequency analysis of the number of individuals at each possible score, thus taking into account both severity and incidence of disease. Delayed hypersensitivity and lymphocyte proliferation data were analyzed using an independent t test. Probability values of p ≤ 0.05 were considered to be significant.

To investigate the effects of CT on Th1 CD4⁺ T cell-mediated autoimmune disease, B10.RIII mice immunized with retinal Ag IRBP for EAU induction were given concurrently a single i.p. injection of 2 μg of CT or vehicle. Evaluation of eyes collected for histopathology on day 21 showed that whereas vehicle-treated mice developed full-blown disease, the mice treated with CT remained disease free (Fig. 1,A). Long-term follow-up by weekly fundoscopic examinations revealed that mice were still protected when the experiment was terminated 8 wk after IRBP immunization (data not shown). This inhibitory effect of CT on disease development is opposite of the effect of PT, which enhances EAU when administered at the time of active immunization (22, 23), and parallels the opposite effect of these two structurally and functionally related toxins on cellular cAMP levels (1, 24, 25, 26). The protective effect of CT was dose dependent, with maximal inhibitory effect achieved already at 2 μg/mouse (Fig. 1,B). The LD₅₀ of CT administered systemically is 20 μg (data not shown); however, the animals had no obvious systemic symptoms even at a dose of 10 μg/mouse CT. The protective effect was still present when CT was administered as late as 6 days after uveitogenic immunization, when priming of uveitogenic effector T cells is already well underway (Fig. 1,C). In contrast, administration of CT 6 or 3 days before IRBP challenge was not protective (Fig. 1 C).

Histopathology of eyes harvested 21 days after immunization showed fully developed EAU in eyes of IRBP-immunized mice, with mononuclear and polymorphonuclear cell infiltration, vitritis, choroiditis, and various degrees of photoreceptor cell damage (Fig. 2,A). In contrast, the histopathology of eyes from CT-treated mice was completely normal (Fig. 2 B).

IRBP-immunized and CT-treated B10.RIII mice were challenged for IRBP-specific DTH by ear assay. Animals treated with CT exhibited a dose-dependent inhibition of DTH. At supraoptimal doses of 5 and 10 μg of CT, the DTH was lower than that of vehicle-treated controls (Fig. 3,A). However, inhibited DTH was not a prerequisite for protection from EAU, as abrogation of disease was achieved at 2 μg of CT, a dose that did not inhibit DTH. Furthermore, Ag-specific proliferation of lymph node cells from mice treated with a protective dose of CT was enhanced, showing that immune suppression was not the mechanism of protection (Fig. 3 B).

Similarly to many other bacterial toxins (e.g., PT and Escherichia coli heat labile toxin), CT is composed of an enzymatically active subunit A and a membrane-binding subunit B that interacts with GM1 proteins. CT-A catalyzes an increase in cellular cAMP production, and its targets are stimulatory subunits of G proteins (Gs). To address the question as to whether it is the enzymatic activity of CT or the engagement of surface receptors by the B subunit that underlies its protective effect, we examined the ability of purified CT-B to inhibit EAU. Fig. 4 showed that substituting CT-B for whole CT did not prevent EAU even when its dose was as high as 10 μg/mouse. These data suggest that the protective effect of CT depends on the ADP-ribosyltransferase activity of the A subunit.

To assess the phenotype of the adaptive immune response to IRBP in CT-protected mice, draining lymph node cells were collected 21 days after immunization and were stimulated in culture with IRBP. Supernatants collected after 48 h were assayed for content of IL-2, IL-4, IL-5, IL-10, IL-12 p40, IFN-γ, and TGF-β1 by ELISA. Only IL-4 was increased significantly (p < 0.05) in the CT-treated group as compared with the vehicle-treated group (Fig. 5). Other cytokines showed only minor differences that did not attain statistical significance. We interpret this increase in IL-4 without a parallel decrease in IFN-γ as indicative of a moderate form of immune deviation.

Because IFN-γ promotes isotype switching from IgM to IgG2a, and IL-4 promotes switching to IgG1 and IgE, the relative amounts of these Ab isotypes reflect the type of effector response (Th1 or Th2) that has developed to the immunizing Ag in vivo. We measured IgG2a and IgG1 Abs to IRBP and total IgE levels in B10.RIII mice treated, or not, with CT, in sera collected 21 days after immunization. Fig. 6,A showed that IgG2a was decreased, whereas IgG1 was increased in CT-treated mice in comparison with vehicle-treated mice. IgE was also increased in CT-treated group as compared with untreated group (Fig. 6 B). These results are compatible with the interpretation that CT treatment skewed the effector response toward Th2.

Bacterial products tend to have strong effects on cells of the innate immune system, which, in turn, help determine the effector choices of the adaptive response that follows. To assess the innate response to CT, we injected naive B10.RIII mice with a dose of CT within the protective range (5 μg). A parallel group of mice was injected with PT at a dose (1 μg) that promotes the Th1 response, enhances EAU scores in B10.RIII mice under suboptimal conditions of immunization, and permits EAU induction in less susceptible strains (22, 27). Serum samples were collected sequentially and were assayed for IFN-γ and IL-4 content by ELISA. PT treatment caused a detectable increase in serum IFN-γ, but IL-4 remained in the underdetection level. In contrast, CT-treated B10.RIII mice showed a brisk increase in serum IL-4 titers that peaked at 6 h after treatment and returned to normal by 24 h. No increase in IFN-γ was detected after CT treatment (Fig. 7).

To further address the mechanism and examine whether CT-induced regulatory cells could transfer protection, we tested whether or not the inhibition of EAU was induced by regulatory T cells that were probably generated by CT treatment. We transferred cells from donors protected by CT or from non-CT-treated controls into lightly irradiated naive mice, who were challenged for EAU 3 days later. Histological examination of eyes on day 21 after challenge showed that recipients of cells from protected donors were not protected from a uveitogenic EAU challenge. Their EAU scores were comparable to a control group that received naive splenocytes (data not shown). This is consistent with the interpretation that, in this system, CT acts by redirecting the priming of effector T cells to a nonpathogenic phenotype, rather than by induction of regulatory cells.

As was demonstrated in Figs. 5 and 7, the ability of CT to promote an innate and adaptive IL-4 response appears to cause protection from EAU. To put this hypothesis to the test, we examined whether endogenous IL-4 was in fact necessary for CT-induced immune deviation and protective effect. Mice immunized for EAU and given CT were treated with neutralizing anti-IL-4 mAbs until day 9 after immunization, which is the expected day of EAU onset. Eyes collected on day 21, when disease is expected to be maximal, were examined for EAU by histopathology. As shown in Fig. 8,A, infusion of mAb against IL-4 completely abrogated the protective effect by CT, and all mice thus treated developed full-blown EAU. Isotype analysis of Abs to IRBP in sera of anti-IL-4-treated mice revealed that the isotype shift toward IgG1 induced by CT had been abrogated. The ratios of IgG2a to IgG1 were similar to that of vehicle-treated mice (Fig. 8,B). In keeping with this, their lymph node cells secreted less IL-4 in response to the immunizing Ag as compared with the control Ig-treated group. In contrast, the production of IFN-γ by the CT/anti-IL-4-treated group was not significantly changed from the vehicle-treated group (Fig. 8 C). These data directly support the conclusion that endogenous IL-4 induced by CT treatment precipitated immune deviation and prevented development of EAU.

EAU in the genetically unmanipulated individual is a Th1-dependent disease. Our previous data indicated that strains susceptible to EAU are dominant Th1 responders, and conversely, coadministration of IL-4 and IL-10 skews the immune response away from the Th1 pathway and prevents development of EAU (22, 28). Nevertheless, an uncontrolled Th2-like response, such as that which develops in IFN-γ-deficient mice, is also pathogenic, and Th2-polarized effector cells to a retinal neo-Ag can drive an EAU-like retinal pathology (29, 30). This Th2-driven pathology is mediated by a deviant effector cell population dominated by eosinophils and neutrophils, and results in tissue destruction that is as severe as that caused by the classical Th1-driven EAU. This raises a note of caution when developing immune deviation strategies for Th1-driven diseases, because too strong a deviation could backfire and could itself induce pathology.

CT is a well-characterized mucosal adjuvant that promotes responses to oral immunization. The mucosal adjuvant effects of CT are due in large measure to its ability to augment T cell activation and differentiation in vivo. Orally administered CT augments Th2 cytokine responses, including IL-4, IL-5, and IL-10, to coadministered Ags, or can promote a mixed Th1 and Th2 cytokine response showing enhanced production of IFN-γ as well as IL-4 (4, 31, 32). In the present study, we show that when administered systemically, CT can counteract generation of pathogenic Th1 effector cells by inducing immune deviation. Immune suppression was excluded as a mechanism because DTH at the optimal protective dose was undiminished, and IRBP-specific proliferation was even increased. This last effect could be attributed to enhancement of CD4⁺ T cell responses, as reported by others (2). Importantly, the protective effects of CT were achieved at systemically nontoxic doses, and even supraoptimal doses of CT did not result in induction of Th2-driven pathology. This is in line with the IRBP-specific cytokine profile of the CT-treated mice, which showed an enhanced IL-4 response, but an undiminished IFN-γ response. Thus, CT treatment resulted in a more balanced cytokine response to the autoantigen rather than inducing an extreme Th2 polarization. Finally, a role for suppressor/regulatory T cells generated in vivo by CT administration is not supported because splenocytes from CT-treated mice did not protect recipients from EAU, consistent with the interpretation that CT redirected the priming of effector T cells to a nonpathogenic phenotype, rather than induced regulatory cells.

Deviation of the adaptive response appeared to be dependent on the ability of CT to induce an innate IL-4 response, which was easily detectable in the serum within hours of CT administration. The effect of anti-IL-4 Abs to reverse the protective effect of CT is consistent with the interpretation that this innate IL-4 production is a prerequisite for the deviating and protective effects of CT. Interestingly, administration of CT even as late as 6 days after Ag challenge was still able to prevent EAU. This is well into the priming phase of the disease, as uveitogenic effector T cells have been demonstrated in IRBP-immunized mice already at 7 days after immunization (33). The ability to counteract generation of pathogenic effector T cells at this relatively advanced phase of disease induction may suggest that newly primed cells can still be deviated, and attests to the potency of the CT effect.

Our data reveal several differences from observations reported by others. First, in our hands, only the IL-4 response to Ag was significantly augmented. Although there was a minor increase in IL-10, it did not achieve statistical significance, and Ag-specific IL-5 production did not differ from controls. This discrepancy from cytokine responses reported by others, who administered CT mucosally (4), may be attributed to the different administration route, or to strain-specific quantitative differences. Second, in contrast to the reports of Yamamoto et al. (34, 35), showing that mucosal adjuvanticity was retained by an enzymatically inactive mutant of CT, and the observations of Sobel et al. (36), who were able to inhibit diabetes in NOD mice by CT-B, in our hands the immunoregulatory effects were strictly dependent on the enzymatic A subunit. In this regard, our findings are in line with the observations of Bagley et al. (26), who reported inhibition of IL-12 and TNF-α production by CT through a cAMP-dependent pathway.

Although, because of its high immunogenicity and potential toxicity, it is unlikely that CT can be used directly as an immunomodulator for Th1-driven pathologies. Understanding its mode of action might help to devise deviation therapies that successfully deviate the autopathogenic response to a nonpathogenic pathway, while avoiding excessive Th2 polarization that can potentially result in a whole new set of pathologies. CT-B binds to GM1 gangliosides on the cell membrane and delivers A protomer into cells (11). In the cytosol, A subunit catalyzes the transfer of an ADP-ribose from NAD to Gs. After ADP-ribosylation, Gs binds to adenylate cyclase and constitutively activates it, leading to a sustained increase in intracellular cAMP concentration (37). Direct stimulation of cAMP pathways with forskolin, or dibutyryl-cAMP, can inhibit the expression of IL-12 (26, 38). Furthermore, the combination of forskolin and phosphodiesterase inhibitors, or direct stimulation of cAMP pathways with dibutyryl-cAMP, results in the enhanced production of IL-4 and IL-5, with concurrent suppression of IL-2 production (39). However, data showing that enzymatically inactive CT and LT can retain Th2 adjuvant activity exceeding that of their recombinant B subunits raise the possibility that the A-B structure itself, independent of ADP-ribosyltransferase activity, might also contribute to the immunological effects of these toxins (35, 40, 41).

In summary, our results demonstrate that systemic administration of CT can prevent EAU induced by retinal Ag in B10.RIII mice. This prevention depends on its ability to inhibit Th1-associated immune responses without excessive polarization of the adaptive response toward the Th2 pathway. Our findings may suggest that the use of substances mimicking some of the biological effects of CT should be explored as treatment modalities for Th1-mediated autoimmune diseases.