Escherichia coli heat-labile enterotoxin B subunit prevents autoimmune arthritis through induction of regulatory CD4+ T cells.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 46, No. 6, June 2002, pp 1671–1682 DOI 10.1002/art.10328 © 2002, American College of Rheumatology Escherichia coli Heat-Labile Enterotoxin B Subunit Prevents Autoimmune Arthritis Through Induction of Regulatory CD4⫹ T Cells Jeffrey A. Luross, Tricia Heaton, Timothy R. Hirst, Michael J. Day, and Neil A. Williams Objective. The receptor-binding B subunit of Escherichia coli heat-labile enterotoxin (EtxB) is a highly stable, nontoxic protein that is capable of modulating immune responses. This study was conducted to determine whether mucosal administration of EtxB can block collagen-induced arthritis (CIA) and to investigate the mechanisms involved. Methods. Clinical arthritis in DBA/1 mice was monitored following mucosal administration of EtxB on 4 occasions. The dependence of disease prevention on receptor binding by EtxB and the associated alterations to the immune response to type II collagen (CII) were assessed. Adoptive transfer experiments and lymph node cell cocultures were used to investigate the underlying mechanisms. Results. Both intranasal and intragastric delivery of EtxB were effective in preventing CIA; a 1-g dose of EtxB was protective after intranasal administration. A non–receptor-binding mutant of EtxB failed to prevent disease. Intranasal EtxB lowered both the incidence and severity of arthritis when given either at the time of disease induction or 25 days later. EtxB markedly reduced levels of anti-CII IgG2a antibodies and interferon-␥ (IFN␥) production while not affecting levels of IgG1, interleukin-4 (IL-4), or IL-10. Disease protection could be transferred by CD4ⴙ T cells from treated mice, an effect that was abrogated upon depletion of the CD25ⴙ population. In addition, CD4ⴙCD25ⴙ T cells from treated mice were able to suppress anti-CII IFN␥ production by CII-primed lymph node cells. Conclusion. Mucosal administration of EtxB can be used to prevent or treat CIA. Modulation of the anti-CII immune response by EtxB is associated with a reduction in Th1 cell reactivity without a concomitant shift toward Th2. Instead, EtxB mediates its effects through enhancing the activity of a population of CD4ⴙ regulatory T cells. Rheumatoid arthritis (RA) is a chronic, systemic inflammatory disease that manifests predominantly in synovial tissues of articular joints. Disease progression is frequently linked with extraarticular complications that are associated with increased patient morbidity and mortality (1). The current approach to treatment of RA tends to focus on managing chronic inflammation and pain. These methods have little effect on the autoimmune processes that are thought to underlie disease and are of limited use in altering disease progression. Therefore, developing novel treatments that target the autoimmune component of the disease is highly desirable. Progress in this area has been hampered by uncertainty concerning the specific antigens that trigger the autoimmune response. Consequently, use of antigenindependent therapies that invoke natural regulatory processes within the immune system is an attractive option. Collagen-induced arthritis (CIA) is a welldefined animal model of RA that has provided considerable insights into disease pathogenesis and has been used extensively to test the potential of novel therapies (2). Like RA, CIA is characterized by the presence of fibrin deposition, hyperplasia of synovial cells, periosteal Supported by grants from the Arthritis Research Campaign and The Wellcome Trust. Dr. Luross’s work was supported by a joint Arthritis Research Campaign/Glaxo SmithKline research grant. Dr. Williams is a Wellcome Trust Research Leave Fellow. Jeffrey A. Luross, PhD, Tricia Heaton, PhD (current address: TVW Telethon Institute for Child Health Research, Subiaco, Western Australia, Australia); Timothy R. Hirst, PhD, Michael J. Day, PhD, MRCPath, Neil A. Williams, PhD: University of Bristol School of Medical Sciences, Bristol, UK. Address correspondence and reprint requests to Neil A. Williams, PhD, University of Bristol, Department of Pathology and Microbiology, School of Medical Sciences, University Walk, Bristol BS8 1TD, UK. E-mail: firstname.lastname@example.org. Submitted for publication August 27, 2001; accepted in revised form February 21, 2002. 1671 1672 bone formation, mononuclear infiltrates, pannus formation, and eventual ankylosis of 1 or more articular joints (3,4). Moreover, susceptibility to both CIA and RA is strongly associated with expression of specific major histocompatibility complex (MHC) class II molecules (5,6), and potential roles for non-MHC loci have been described (7,8). Generation of interferon-␥ (IFN␥)– producing CD4⫹ Th1 cells is thought to be critical for the induction of CIA (9). Indeed, IFN ␥ and interleukin-12 (IL-12) dominate the anti–type II collagen (CII) immune response in the acute disease phase, while remission coincides with reduced production of IFN␥ and increased production of IL-10 (10,11). IFN␥ promotes generation of complement-fixing IgG2a anti-CII antibodies, the deposition of which is thought to trigger early inflammatory events, allowing immune cell entry into the joint (12). In addition, the activity of IFN␥ in stimulating myeloid cells to produce IL-1 and tumor necrosis factor ␣ likely contributes to the localized tissue inflammation and subsequent joint destruction characteristic of CIA (13). Importantly, administration of exogenous Th1-associated cytokines has been shown to exacerbate CIA (14), while counterregulation of the Th1 response by administration of Th2 cytokines has been shown to protect against disease (15). The nontoxic B subunit of Escherichia coli heatlabile enterotoxin (EtxB) and its structural homolog from cholera toxin (CtxB) are highly stable proteins that mediate binding of their respective enterotoxins to membrane receptors on mammalian cells. The principal receptor to which both EtxB and CtxB bind is the ganglioside GM1. In addition, both molecules bind to GD1b, and EtxB has some affinity for asialo-GM1, lactosylceramide, and certain galactoproteins (16). Recent reports have indicated that the nontoxic B subunits may be useful agents for prevention and treatment of Th1-mediated autoimmune diseases (16). Mucosal administration of CtxB-autoantigen conjugates has been shown to prevent experimental allergic encephalomyelitis (17), diabetes in NOD mice (18), and CIA (19). The observation that this effect requires direct coupling of autoantigen with CtxB led to the suggestion that the B subunit acts as a carrier molecule, directing antigen into natural mucosal tolerance-inducing pathways. However, there is clear evidence indicating a more active immune modulatory role for the B subunits. Intravenous or intraperitoneal delivery of CtxB alone can delay the onset of diabetes in the NOD mouse (20), and subcutaneous administration of EtxB in Freund’s incomplete adjuvant (IFA) can prevent CIA (21). The fact that these effects can be achieved follow- LUROSS ET AL ing parenteral delivery shows that B subunit–mediated protection is not dependent on features of the mucosal immune system. Moreover, the observation that CtxB and EtxB are active in the absence of conjugated autoantigen shows that they are not acting simply as delivery vehicles. The precise mechanisms of immune modulation by the B subunits remain to be established. Prevention of CIA following subcutaneous administration of EtxB/IFA was associated with altered T cell reactivity to CII, as demonstrated by a reduction in IFN␥ production concomitant with increased release of IL-4 (21). Although these data suggest that EtxB can mediate a change from Th1 to Th2 responsiveness, measurement of serum anti-CII antibodies revealed an overall suppression of the response, indicating that other immunomodulatory mechanisms may be involved. The purpose of the present study was to investigate the basis of the protective effect of EtxB in CIA. We report that disease prevention can be mediated by mucosal administration of EtxB alone, and that this effect involves the enhanced activity of a population of splenic CD4⫹ regulatory T cells. MATERIALS AND METHODS Mice. Age-matched male DBA/1 mice were purchased from Harlan Olac (Bicester, UK) and were used for experimentation at 7 to 13 weeks of age. Animals were used in accordance with institutional and Home office guidelines. Antigens, immunizations, and induction of arthritis. Recombinant EtxB, EtxB(G33D) (a non–receptor-binding mutant of EtxB), and CtxB were synthesized and purified as previously reported (22). Levels of endotoxin were ⬍30 units/mg in all samples, as determined using a Kinetic-QCL chromogenic limulus amebocyte lysate assay (BioWhittaker, Walkersville, MD). Groups of mice were given various doses of EtxB, EtxB(G33D), CtxB, or chicken CII (Sigma-Aldrich, Poole, UK) by intranasal (20-l volume) or intragastric (200-l volume) routes at 24-hour intervals for 4 days, unless otherwise stated. On the final day of treatment (day 0), mice were challenged with 100 g of CII in Freund’s complete adjuvant (CII/CFA) or 100 g of keyhole limpet hemocyanin (KLH; Calbiochem, La Jolla, CA) in CFA, given subcutaneously at the base of the tail. Disease assessment. Mice were examined regularly for clinical evidence of arthritis, characterized by a combination of erythema and swelling of the tarsal joints, digits, and foot pads. Each paw was assigned a score of 0 (normal appearance), 1 (mild erythema and/or swelling), or 2 (significant erythema and/or swelling), and the cumulative score (maximum of 8) was then calculated. Data are presented as the mean ⫾ SEM clinical score for each group of mice. Histologic sections of stifle joints were examined for pathology using an established 4-point scoring system (21). All scoring of disease was performed under blinded conditions. PREVENTION OF AUTOIMMUNE ARTHRITIS BY ExtB Antibody responses. Fifteen days after CII immunization, serum samples were obtained and analyzed by enzymelinked immunosorbent assay (ELISA) for the presence of anti-CII antibodies. The distribution of CII-specific IgG1 and IgG2a antibody subclasses was determined using specific detecting antibodies for mouse IgG1 and IgG2a (Serotec, Oxford, UK). Antibody concentrations were calculated against defined concentrations of myeloma-derived IgG1 and IgG2a (Serotec) by weighted probit analysis (23). Adoptive transfer. Spleens from mice that were treated with intranasal EtxB or phosphate buffered saline (PBS) and challenged with CII/CFA were harvested 24 days after immunization. To enrich for CD4⫹ cells, splenocytes were separated using CD8, CD11b, and CD45R magnetic-activated cell sorting (MACS) beads (Miltenyi Biotec, Bisley, UK). In experiments in which CD4⫹ or CD25⫹ cells were depleted, splenocytes were labeled with anti-CD4 MACS beads or biotinconjugated anti-CD25 (7D4; BD PharMingen, San Diego, CA), followed by streptavidin-conjugated MACS beads (Miltenyi Biotec). Resulting purities were a minimum of 82% for CD4⫹ cells, 97% for CD4⫹ depleted cells, and 98% for CD25⫹ depleted cells, as assessed by flow cytometry. Purified cell populations were suspended in Hanks’ balanced salt solution for intravenous adoptive transfer of 1 ⫻ 107 cells into naive recipients. Groups of control mice received 200 l of vehicle. T cell proliferative and cytokine responses. Mice were killed 25 days after immunization, and spleen and/or inguinal lymph node (ILN) cell populations were derived for culture (2.5 ⫻ 106 and 2–3 ⫻ 106 viable cells/ml, respectively, in 2-ml culture). Fluorescently labeled antibodies to CD4 (H129.19; Sigma) and CD25 (PC61; BD PharMingen) were used for fluorescence-activated cell sorting (FACS) of spleen cells for coculture experiments. Cell purities were a minimum of 94%. Cells were cultured under previously described conditions (24) in the presence or absence of either 50 g/ml of CII or 10 g/ml of KLH. At select time points, duplicate or triplicate 100-l samples were transferred into 96-well Costar plates (Corning, Corning, NY) for assessment of proliferation and cytokine production using 3H-thymidine (3H-TdR) incorporation and cell-based ELISA (25) techniques, respectively. Concentrations of IFN␥, IL-4, and IL-10 were calculated against standard curves produced with recombinant cytokine (BD PharMingen). Limits of detection for each cytokine assayed were 20 pg/ml, 10 pg/ml, and 40 pg/ml, respectively. Statistical analysis. For normally distributed data, statistical comparisons were performed using Student’s unpaired t-test. For nonparametric data, the Mann-Whitney U test or Kruskal-Wallis analysis of variance was used. Differences in disease incidence were tested by chi-square analysis. RESULTS Prophylactic and therapeutic effect of mucosally delivered EtxB on CIA. We previously reported that subcutaneous injection of EtxB in IFA can prevent CIA (21). Although injectable therapies may be appropriate for the treatment of RA, it would be advantageous if a noninvasive method of delivery could be used. There- 1673 Figure 1. Effects of mucosal administration of Escherichia coli heatlabile enterotoxin B subunit (EtxB) on collagen-induced arthritis (CIA). A, Scores for clinical disease in mice receiving 100 g of EtxB intranasally (i.n.) (䡲; n ⫽ 8) or intragastrically (i.g.) (䡺; n ⫽ 5) or 1 g of EtxB i.n. (Œ; n ⫽ 8) or i.g. (‚; n ⫽ 7), or left untreated (✖; n ⫽ 10) prior to induction of CIA. Values are the mean ⫾ SEM disease score for each group. Similar observations were made in 3 separate experiments. B, Histopathologic scores in mice receiving 100 g or 1 g of EtxB i.n. or i.g. Values are the mean and SEM disease score for each group. C and D, The incidence of arthritis and the severity of clinical disease, respectively, in affected animals. The data are pooled from animals assessed in 3 separate experiments in which mice were either untreated (✖; n ⫽ 26) or treated with EtxB intranasally (■; n ⫽ 33) prior to type II collagen/Freund’s complete adjuvant challenge. ⴱ ⫽ P ⬍ 0.005 compared with untreated controls. fore, we investigated whether mucosal delivery of EtxB could prevent CIA. Groups of mice were either left untreated or were treated using inhalation or gastric lavage with 100 g or 1 g of EtxB for 4 days before disease induction. Untreated mice developed a severe arthritis (Figure 1A), which was evident by day 25 and became progressively worse during the following 20 days. In contrast, mice that received 100 g of EtxB, either intranasally or intragastrically, had markedly lower clinical disease scores (P ⬍ 0.05 versus controls on 1674 days 30–45). Strong disease protection following intranasal administration of EtxB was also observed when the dose was reduced to 1 g (P ⬍ 0.05 versus controls on days 30 and 35; P ⬍ 0.001 versus controls on days 40 and 45). However, intragastric EtxB treatment at this dose was not effective. Mice that received four 10-g doses of CII intranasally before CII/CFA challenge (a regimen similar to that previously reported to induce mucosal tolerance) (26) exhibited an intermediate level of disease protection (data not shown). Histopathologic analysis of the stifle joints of mice on day 45 correlated well with clinical disease scores (Figure 1B). Clinically effective doses of EtxB reduced the severity of joint pathology. However, differences in cellular infiltration and joint destruction between treated mice and untreated mice were statistically significant only when the 100-g dose of EtxB was given intranasally (P ⬍ 0.005). To further investigate the process of protection against CIA following mucosal administration of EtxB, data from several similar experiments in which EtxB was given intranasally were pooled, and the incidence and severity of disease in affected mice were analyzed separately. This analysis (Figures 1C and D) clearly showed that intranasal EtxB treatment had a dual effect on disease progression. A marked difference in the incidence of CIA between treated and untreated mice was evident on day 25. Furthermore, during the following 3 weeks, increasing numbers of untreated animals developed signs of arthritis, while only a slight increase in the proportion of affected mice was observed in the EtxBtreated group. The difference in the incidence of arthritis between the PBS-treated and EtxB-treated groups was significant at each time point tested (P ⬍ 0.005). Although the severity of clinical disease in affected mice in the untreated group increased over time, progression of arthritis was markedly attenuated in EtxB-treated animals (P ⬍ 0.005 at every time point studied after day 25). To test whether prevention of CIA by mucosally delivered EtxB was dependent on receptor interaction, groups of mice were treated intranasally with an intermediate, 10-g dose of either EtxB or EtxB(G33D) before disease induction. The mean clinical score for arthritis in the EtxB-treated group remained low throughout the course of the experiment (Figure 2A). This effect differed from that in EtxB(G33D)-treated mice, which developed severe arthritis (P ⬍ 0.02 on days 28 and 32; P ⬍ 0.001 on days 35–46). Further experiments tested whether 10 g of intranasally administered CtxB could prevent CIA. At this dose, CtxB failed to LUROSS ET AL Figure 2. A, Data showing that receptor binding by EtxB is critical for modulation of CIA. Mice were given four 10-g intranasal doses of EtxB (■; n ⫽ 10) or EtxB(G33D) (E; n ⫽ 10), and CIA was induced. Data are the mean ⫾ SEM clinical score of arthritis. ⴱ ⫽ P ⬍ 0.02; ⴱⴱ ⫽ P ⬍ 0.001. B, Lack of modulation of CIA with intranasal administration of the structural homolog from cholera toxin (CtxB). Mice (n ⫽ 7) were left untreated (E) or were given 10 g of CtxB (■; n ⫽ 10) before disease induction. Values are the mean ⫾ SEM clinical score of arthritis (P ⬎ 0.7 at all time points studied). C and D, Effects of EtxB on recent-onset arthritis. Mice were immunized with type II collagen/Freund’s complete adjuvant on day 0 to induce arthritis. On days 25, 27, 30, and 32, mice were treated intranasally with 10 g of EtxB (■; n ⫽ 14) or EtxB(G33D) (E; n ⫽ 13). Arrows indicate days of treatment. C, Incidence of clinical arthritis. D, Severity of disease in affected animals. Values are the mean ⫾ SEM. ⴱ ⫽ P ⬍ 0.03. See Figure 1 for other definitions. alter the clinical arthritis score in treated mice compared with untreated mice (Figure 2B). In RA, it is not possible to predict the onset of disease. Therefore, effective therapies must be able to limit progression of established arthritis. Accordingly, we investigated the effects of administering intranasal EtxB after CIA was established. Mice challenged with CII/CFA to induce arthritis were randomized on day 25 to receive treatment with four 10-g doses of either EtxB or EtxB(G33D) over 7 days. During the treatment period, the incidence of CIA in the 2 groups was nearly PREVENTION OF AUTOIMMUNE ARTHRITIS BY ExtB identical (Figure 2C). However, disease incidence then declined in the EtxB-treated group, from a peak at day 32, and remained lower than that in the EtxB(G33D)treated group for the remainder of the experiment (P ⬍ 0.03 on days 35, 42, and 46). Among affected mice, the severity of CIA was also consistently lower in mice given EtxB (Figure 2D), although the difference did not reach statistical significance. Taken together, these data show that receptor binding following mucosal delivery of EtxB can prevent the onset of CIA and actively limit progression of established disease. Alteration of the anti-CII immune response by EtxB treatment. Production of complement-fixing IgG2a anti-CII autoantibodies is an important component of the pathologic immune response in CIA (9). Thus, serum was obtained from mice that were treated with EtxB intranasally or intragastrically before disease induction, in order to determine whether protection against CIA correlated with changes in the amount and isotype of anti-CII antibodies. Sera from mice that received CII intranasally were included for comparison. As expected, IgG2a was the most abundant isotype detected in sera from untreated mice, which is indicative of a Th1-dominated response (Figure 3). Importantly, the amounts of anti-CII IgG2a antibodies were markedly suppressed in all treatment groups (P ⬍ 0.01 compared with the untreated group), with the lowest levels present in mice that had received either 1 g or 100 g EtxB intranasally (P ⬍ 0.001). However, mice that received a disease-protecting dose (100 g) of EtxB intragastrically had amounts of CII-specific IgG2a similar to those of mice that received an ineffective dose of 1 g. The amounts of anti-CII IgG1 antibodies within treatment groups were comparable with those detected in untreated mice, with the exception of the group that received 100 g of EtxB intranasally, in which an ⬃2.3-fold increase was detected (P ⬍ 0.01). No anti-CII antibodies were measured in the sera of mice that were not challenged with CII/CFA. In further investigations, we determined whether mucosal administration of EtxB altered the anti-CII T cell response. Spleen cells from mice treated with EtxB or PBS intranasally before being immunized with CII/ CFA were cultured in the presence or absence of CII. The proliferative and cytokine responses in the cultures were monitored over a 5-day period. Figure 4 shows data from 1 of 5 identical experiments on day 3 of culture (the time at which responses were maximal). Spleen cells from both EtxB-treated and PBS-treated mice proliferated strongly in response to CII. In the experiment shown, the level of proliferation was lower in the culture 1675 Figure 3. Changes in the anti–type II collagen (anti-CII) antibody response following mucosal delivery of EtxB. Mice were left untreated (⫹) or were treated intranasally or intragastrically with 100 g or 1 g of EtxB or with 10 g of CII intranasally, for 4 days prior to induction of CIA (day 0). Each animal was bled on day 14 to determine the presence of anti-CII–specific IgG1 and IgG2a serum immunoglobulin. Sera from unchallenged mice were used as a negative control (⫺). Data for individual mice are shown. Bars show the median antibody concentration. ⴱ ⫽ P ⬍ 0.01; ⴱⴱ ⫽ P ⬍ 0.001. NS ⫽ not significant. See Figure 1 for other definitions. from EtxB-treated mice. However, when the results of 5 identical experiments were analyzed, the difference in levels of spleen cell proliferation between cultures from EtxB-treated and untreated mice was not significant (mean ⫾ SEM proliferative response 7,753 ⫾ 1,467 counts per minute in PBS-treated mice and 7,712 ⫾ 1,791 cpm in EtxB-treated mice; P ⬍ 0.98). Analysis of the cytokines produced in the spleen cell cultures revealed that EtxB treatment dramatically altered the nature of the T cell response. Spleen cells from mice receiving EtxB produced lower levels of IFN␥. The mean ⫾ SEM reduction was 42 ⫾ 11% in 5 1676 Figure 4. Altered splenic T cell responses to type II collagen (CII) after Escherichia coli heat-labile enterotoxin B subunit (EtxB) treatment. Mice (n ⫽ 10) were given 20 g of EtxB or phosphate buffered saline (PBS) intranasally and were then immunized with type II collagen (CII)/Freund’s complete adjuvant. Splenocytes, harvested 24 days later, were cultured for 5 days in the presence (䊐) or absence (■) of 50 g/ml of heat-denatured CII for the analysis of cellular proliferation (3H-thymidine [3H-TdR] incorporation [cpm]) and the production of the cytokines interferon-␥ (IFN␥), interleukin-4 (IL-4), and IL-10. Data are the mean and SEM value of triplicate samples from the day at which maximum proliferation was recorded (day 3). The results shown are from 1 of 5 identical experiments. identical experiments. The decrease in IFN␥ production following administration of EtxB was not associated with increased levels of IL-4 or IL-10 (the higher background release of IL-10 in cultures from EtxB-treated animals was not a consistent feature of the 5 experiments). Protection against CIA following adoptive transfer of CD4ⴙ T cells from EtxB-treated mice. The observation that mucosal administration of EtxB modulated the differentiation of anti-CII–responsive T cells led us to hypothesize that the B subunit may enhance the activity of a regulatory T cell population. To investigate this theory, we administered 1 ⫻ 107 splenic CD4⫹ cells (isolated from either EtxB-treated or PBS-treated mice 24 days after CII/CFA challenge) intravenously to naive recipients, 3 days before induction of CIA. Additional groups were either given intravenous vehicle only or were left untreated before CII/CFA challenge. Clinical signs of arthritis were monitored, and the results of 1 of 3 identical experiments are shown in Figure 5A. CIA developed normally in mice that were not manipulated and in those that had been given vehicle only. In contrast, animals that received CD4⫹ cells from EtxB-treated mice developed an attenuated arthritis. The mean clinical score in this group was significantly lower than that in any of the control groups from day 26 onward (P ⬍ 0.01 on days 26 and 29; P ⬍ 0.05 on the remaining days studied). The ability of transferred CD4⫹ cells to prevent CIA was dependent on whether donors had been treated with EtxB; recipients of CD4⫹ cells from PBS-treated mice developed an arthritis that LUROSS ET AL was similar to that seen in control animals. The decreased mean clinical score in mice that received CD4⫹ cells from EtxB-treated donors was reflected in both a lower incidence of arthritis and decreased disease severity in affected animals. Data from these transfer experiments clearly demonstrate that mucosal administration of EtxB leads to the presence of CD4⫹ cells in the spleen that are capable of suppressing CIA. Further experiments were performed to determine whether CD4⫹ cells were the only population of spleen cells from EtxB-treated mice capable of conferring protection against CIA. As in previous adoptive Figure 5. A, Attenuation of CIA development by adoptive transfer of CD4⫹ cells from mice given EtxB intranasally. Mice (n ⫽ 10) were given 20 g of EtxB intranasally or phosphate buffered saline (PBS) and immunized with type II collagen/Freund’s complete adjuvant (CII/CFA). Twenty-four days later, 1 ⫻ 107 CD4⫹ enriched splenocytes from EtxB-treated mice (■; n ⫽ 8) or PBS-treated mice (Œ; n ⫽ 9), or an equal volume of the vehicle (Hanks’ balanced salt solution; 䉬; n ⫽ 10) was inoculated intravenously into naive recipients. Three days after adoptive transfer, recipient mice were immunized with CII/CFA to induce arthritis (day 0). An additional group of untreated mice received CII/CFA immunization at this time (E; n ⫽ 9). Values are the mean ⫾ SEM clinical score of arthritis, the incidence, and the severity of disease in affected animals. Similar results were observed in 2 additional identical experiments. ⴱ ⫽ P ⬍ 0.05, ⴱⴱ ⫽ P ⬍ 0.01 versus other groups. B, Spleen cells were isolated 25 days following administration of 4 doses of 10 g EtxB intranasally on consecutive days and were either left unselected (■; n ⫽ 10) or were depleted of CD4⫹ cells (Œ; n ⫽ 9) before transfer into recipient mice. Disease progression was compared with that seen in untreated controls (E; n ⫽ 10). Values are the mean ⫾ SEM clinical score of arthritis, the incidence, and the severity of disease in affected animals. ⴱ ⫽ P ⬍ 0.05, ⴱⴱ ⫽ P ⬍ 0.01 versus untreated controls. See Figure 1 for other definitions. PREVENTION OF AUTOIMMUNE ARTHRITIS BY ExtB transfer experiments, spleen cells were isolated following intranasal administration of EtxB and were then left unselected or were depleted of CD4⫹ cells before being injected intravenously into naive recipients. Arthritis was induced in the recipients, as well as in an additional group of mice that was left untreated. The results (Figure 5B) demonstrate that although unselected spleen cells from EtxB-treated mice are capable of potently suppressing CIA, depletion of the CD4⫹ subset abrogates this ability. Both the mean clinical arthritis score and the disease incidence were markedly lower in mice receiving unselected cells, and no statistically significant differences were observed between the groups receiving CD4⫹ cell–depleted splenocytes and those left untreated. Suppression of IFN␥ production by CD4ⴙCD25ⴙ T cells from EtxB-treated mice. Coexpression of CD25 on CD4⫹ T cells defines both activated cells and naturally occurring T regulatory cells (27,28). Activated effector cells may be capable of mediating immune deviation (29), and CD4⫹CD25⫹ cells from naive mice can suppress T cell responses in vitro (30) and prevent Th1-associated diseases following adoptive transfer (31,32). Two approaches were taken to test whether CD4⫹CD25⫹ cells were involved in immune regulation following mucosal EtxB treatment. First, spleen cells were isolated from either EtxB-treated or PBS-treated mice 24 days after CII/CFA challenge, and the CD4⫹CD25⫹ population was purified by FACS. CD4⫹CD25⫹ cells were added to cultures of CII/CFAprimed ILN cells from untreated mice at a ratio of 1:10. During 5 days of culture, cell proliferation and cytokine production were monitored in the presence or absence of CII. Results at the time of peak response (day 4) are shown in Figure 6A. In the presence of CII, ILN cells from CII/CFA-primed mice responded vigorously, producing high levels of IFN␥ but no IL-4 or IL-10 above background levels. Addition of CD4⫹CD25⫹ cells from PBS-treated mice had minimal effect on the response; there was a low-level increase in 3H-TdR incorporation and a slight reduction in IFN␥ secretion. However, in cultures of ILN cells to which CD4⫹CD25⫹ cells from EtxB-treated mice were added, IFN␥ was markedly suppressed. A similar level of suppression was seen in identical experiments (mean IFN␥ concentration ⫾ SEM [ng/ml]: ILN cells alone, 3.60 ⫾ 0.49; ILN cells with CD4⫹CD25⫹ cells from EtxB-treated mice, 1.00 ⫾ 0.17 [P ⬍ 0.01; n ⫽ 4]). The addition of CD4⫹CD25⫹ cells from EtxB-treated mice did not affect either levels 1677 Figure 6. Suppression of IFN␥ production in vitro by CD4⫹CD25⫹ cells from Escherichia coli heat-labile enterotoxin B subunit (EtxB)– treated mice. Purified CD4⫹CD25⫹ splenocytes from mice (n ⫽ 10 per group) treated with four 20-g intranasal doses of EtxB or PBS intranasally and immunized with CII (A) or keyhole limpet hemocyanin (KLH) (B) were harvested 24 days after challenge and added at a ratio of 1:10 into cultures of antigen-primed inguinal lymph node (ILN) cells derived from mice immunized with corresponding antigen/ Freund’s complete adjuvant on day 0. Cells were cultured in the presence or absence of either 50 g/ml heat-denatured CII or 10 g/ml KLH for 5 days. Cellular proliferation (3H-TdR [cpm]) and the production of IFN␥, IL-4, and IL-10 are shown for the day at which maximum proliferation was observed in cultures of antigen-primed ILN cells (white bars), or cocultures of antigen-primed ILN cells and CD4⫹CD25⫹ splenocytes from PBS-treated mice (grey bars) or EtxB-treated mice (black bars). All samples were assayed in duplicate. Horizontal lines represent the maximum background activity detected in each assay. Data are representative of 4 experiments. See Figure 4 for other definitions. of cellular proliferation or production of IL-4 or IL-10 in ILN cell cultures. The ability of EtxB to trigger the modulatory activity of CD4⫹CD25⫹ T cells in the spleen may be related to the fact that CII is an autoantigen. Thus, EtxB could be activating CII-reactive CD4⫹CD25⫹ regulatory T cells that are present as a result of normal tolerance-inducing pathways. To test this possibility, we investigated whether similar regulatory activity could be detected in experiments in which a foreign antigen was 1678 Figure 7. CIA in mice following adoptive transfer of EtxB-treated whole splenocytes or splenocytes depleted of CD25⫹ cells. Splenocytes from mice (n ⫽ 12) given four 20-g intranasal doses of EtxB were harvested 25 days after immunization with type II collagen/ Freund’s complete adjuvant (CII/CFA). Groups of naive mice were inoculated intravenously with 1 ⫻ 107 splenocytes (■; n ⫽ 10), 1 ⫻ 107 CD25⫹ depleted splenocytes (Œ; n ⫽ 14), or vehicle (Hanks’ balanced salt solution) (E; n ⫽ 9). Three days after adoptive transfer, mice were challenged with CII/CFA to induce CIA (day 0). Values are the mean ⫾ SEM clinical score of arthritis, the incidence, and the severity of disease in affected animals. ⴱ ⫽ P ⬍ 0.05, ⴱⴱ ⫽ P ⬍ 0.001 versus controls. See Figure 1 for other definitions. the target of the T cell response. Mice were treated intranasally with PBS or EtxB before challenge with KLH/CFA. On day 25, CD4⫹CD25⫹ cells were purified and added to cultures of ILN cells from untreated, KLH/CFA-primed animals (Figure 6B). The results showed that CD4⫹CD25⫹ cells from EtxB-treated mice were capable of reducing IFN␥ production in response to KLH. Once again, this effect was not associated with altered levels of proliferation or with IL-4 and IL-10 production. A marginal decrease in KLH-associated IFN ␥ production was seen in cocultures of CD4⫹CD25⫹ cells from PBS-treated mice. Similar results were obtained in 2 identical experiments using KLH. In a second approach to investigate the role of CD25⫹ cells in the modulation of CIA by EtxB, splenic cells from CII/CFA-challenged mice treated with intranasal EtxB were isolated and left unselected or were depleted of the CD25⫹ population for adoptive transfer into naive recipients. Three days after transfer, recipients were challenged with CII/CFA and subsequently monitored for signs of arthritis (Figure 7). CIA developed normally in mice that had received vehicle, and, as expected, transfer of unselected spleen cells from EtxBtreated mice inhibited development of CIA. Importantly, depletion of CD25⫹ cells from the transferred population significantly reduced their ability to suppress arthritis (P ⬍ 0.001 for mean clinical scores between recipients of unselected and depleted cells from day 25 to day 46). This difference was also manifested when LUROSS ET AL separate data regarding disease incidence and severity were analyzed. Interestingly, although depletion of the CD25⫹ population markedly diminished the capacity of the transferred cells to prevent arthritis, some protective effect was still observed in comparison with vehicle-only controls. The mean clinical scores in mice receiving cells depleted of the CD25⫹ population were lower than those in controls (P ⬍ 0.05 on days 21, 28, and 32), an effect that resulted primarily from a decreased severity of disease in this group. DISCUSSION The results of these experiments show that mucosal administration of EtxB can block the establishment and progression of CIA. Intranasal or intragastric treatment with EtxB at the time of disease induction reduced clinical signs of joint swelling. However, the intranasal route appeared to be more effective, because mice were protected from CIA when 1 g of EtxB was given intranasally but not when it was administered intragastrically. EtxB is highly pH- and protease-resistant and would not be expected to undergo degradation during transit into the intestine (33). Therefore, this difference in effective dose probably results from dilution of EtxB over the large surface area of the gut. The ability of EtxB to prevent clinical arthritis correlated with reductions in inflammatory changes in the stifle joints, indicating that EtxB is capable of reducing pathologic changes within the joints. Further analysis of the effects of intranasal EtxB treatment revealed that clinical disease was prevented in some mice (as revealed by the lowered incidence of CIA), and mean disease severity was reduced in those animals that did develop CIA. The incidence and severity of CIA were also reduced when intranasal EtxB treatment was delayed until the anti-CII response had been established. The observation that the incidence of clinical joint swelling decreased following EtxB treatment indicates that the B subunit can ameliorate established disease in some animals. An analysis of disease severity suggested that disease progression was attenuated in the remaining arthritic animals. Collectively, these findings indicate that EtxB can modify the nature of the immune response following CII challenge, blocking either the establishment or progression of joint inflammation. Investigations into the mechanisms underlying mucosal treatment of CIA by EtxB revealed a critical role played by B subunit–receptor interaction. The non– PREVENTION OF AUTOIMMUNE ARTHRITIS BY ExtB receptor-binding mutant, EtxB(G33D), failed to prevent CIA. Interestingly, as previously reported (19), intranasal administration of CtxB also failed to block CIA. The inability of CtxB, when used alone, to modulate CIA reveals critical differences between EtxB and CtxB, which probably relate to their disparate stabilities (i.e., CtxB is unstable as a pentamer below pH 3.9, and EtxB is stable at pH 2.0) or the slightly wider receptor specificity of EtxB. Although CtxB was unable to prevent CIA in the absence of conjugation to CII (19), it has been shown to prevent diabetes in NOD mice following parenteral delivery (20) and can block trinitrobenzene sulfonic acid–induced colitis when given orally in the presence of a pH buffer (34). Attenuation of CIA following mucosal delivery of EtxB was associated with a marked reduction in the proinflammatory anti-CII immune response. Intranasal and intragastric EtxB treatment decreased levels of CII-specific IgG2a antibodies, and intranasal delivery of EtxB decreased CII-dependent splenic T cell production of IFN␥. The reduction of the anti-CII IgG2a antibody response was analogous to that observed using an established intranasal CII tolerization protocol (35). Suppression of the anti-CII Th1 response following subcutaneous injection of EtxB in IFA was associated with enhanced IL-4 production, suggesting deviation toward Th2 (21). However, enhanced Th2 responsiveness was not the mechanism underlying EtxB-mediated protection against CIA following mucosal administration. Although intranasal delivery of 100 g of EtxB did increase levels of IgG1 anti-CII antibodies, this effect was not seen following treatment with 1 g of EtxB intranasally or 100 g of EtxB intragastrically, even though both of these doses effectively prevented clinical arthritis. Furthermore, no enhancement of IL-4 or IL-10 production was observed in cultures of splenic cells from mice treated with EtxB intranasally. Because suppression of the anti-CII Th1 response did not correlate with heightened Th2 reactivity, we hypothesized that EtxB may enhance the activity of a regulatory CD4⫹ T cell population. Such cells capable of suppressing immune reactivity by the production of soluble factors and/or via cell–cell contact have been described both in vitro and in vivo (36). Although characterization of these cell types is incomplete, they are thought to be critical for the maintenance of immune homeostasis, because their in vivo depletion or reconstitution, respectively, can induce or protect against Th1mediated inflammatory disease. Conclusive evidence that mucosal administration of EtxB enhanced the activity of CD4⫹ T regulatory cells came from our finding 1679 that adoptively transferred splenic CD4⫹ cells inhibited clinical arthritis, provided that they were derived from B subunit–treated mice. Moreover, the finding that depleting CD4⫹ cells ablated the capacity of spleen cells to transfer disease protection demonstrates the critical role played by this subset in the B subunit–mediated prevention of CIA. Our further observation that depletion of CD25⫹ cells from spleen cells before transfer abrogated their ability to prevent CIA suggests a role for a CD4⫹CD25⫹ subset in mediating EtxB-induced protection against CIA. Nevertheless, differences in the pattern of disease between whole spleen cell–protected mice and mice that had received splenocytes depleted of CD25⫹ cells indicate that other cell populations may contribute to immune regulation following EtxB treatment. Experiments directly testing the ability of sorted CD4⫹CD25⫹ cells from the spleens of EtxB-treated mice to prevent CIA may provide further insight into the role of this population. However, such studies would need to consider that this population may also contain activated CII-responsive effector cells. Additional evidence to associate EtxB treatment with enhanced regulatory activity of CD4⫹CD25⫹ T cells came from in vitro experiments in which purified populations of these cells were added to ILN cell cultures from CII/CFAprimed mice. These studies revealed that CD4⫹CD25⫹ cells from EtxB-treated mice are enhanced in their capacity to block CII-specific IFN␥ production. This reduction in IFN␥ was not associated with increased IL-4 or IL-10 production, which suggests that the regulatory CD4⫹CD25⫹ population was not mediating its effects due to the presence of Th2 cells (which might share this surface phenotype). Interestingly, the addition of CD4⫹CD25⫹ cells from EtxB-treated mice did not alter levels of proliferation in ILN cell cultures responding to CII. This finding differs from observations using naturally occurring CD4⫹CD25⫹ cells, in which both proliferation and cytokine production are suppressed (30). There are several possible reasons for this difference. First, because of technical constraints of cell isolation, we used a relatively low ratio of CD4⫹CD25⫹ cells to responder ILN cells. Although regulatory CD4⫹CD25⫹ cells suppressed anti-CD3–stimulated responses of CD4⫹CD25⫺ cells at this ratio, the extent of inhibition was suboptimal (30). Second, the CD4⫹CD25⫹ regulatory T cell population modulated by EtxB may have functional activity different from that in naive animals. This difference could result from altered activation of naturally occurring CD4⫹CD25⫹ cells or may be a 1680 consequence of EtxB-mediated induction of a novel cell population that shares this phenotype. Finally, in our culture system, CD4⫹CD25⫹ cells were used to modulate a Th1-dominated antigen-specific response. This contrasts with other systems in which CD4⫹CD25⫹ cells are assessed for their ability to modulate naive T cell responses or the responses of CD4⫹CD25⫺ cells to polyclonal stimulation (30,37). The reduction in IFN␥ release observed following EtxB treatment, and in cocultures of ILN cells and CD4⫹CD25⫹ cells from EtxB-treated mice, is consistent with the attenuation of arthritis. IFN␥ is a primary switch factor for production of IgG2a antibodies, which are critical for the establishment of joint inflammation in CIA. Nevertheless, although anti-CII IgG2a antibodies were reduced following EtxB treatment, the observed reduction was similar between mice given diseaseprotecting intranasal or high-dose intragastric treatment and mice receiving a non–disease-protecting low intragastric dose. Taken together with the observation that IgG2a anti-CII antibody levels were not reduced in recipients of CD4⫹ cells from EtxB-treated mice (data not shown), these findings highlight the importance of modulatory effects of EtxB that are independent of antibody production. Such effects may include a reduction in IFN␥-dependent macrophage and neutrophil activation/recruitment into joint tissue. However, alterations to the anti-CII response, additional to suppression of IFN␥ production, may also be critical for EtxBmediated protection against CIA. How might mucosal administration of EtxB enhance the activity of T cells that can regulate the response to CII? As an autoantigen, CII is probably involved in driving normal processes that maintain peripheral tolerance. Indeed, T cells from DBA/1 mice fail to proliferate in response to autologous CII challenge but secrete some IFN␥ (38). The enhanced regulatory activity of CD4⫹CD25⫹ cells following EtxB treatment was not unique to the response to CII. Similar regulatory activity was evident when CD4⫹CD25⫹ cells from EtxB-treated mice were added to ILN cell cultures responding to KLH. Although the extent of IFN␥ suppression was lower than that seen in cultures with CII, the higher magnitude of the ILN response to KLH may mean that increased numbers of CD4⫹CD25⫹ cells would be required to achieve a similar level of suppression. Although these findings suggest that EtxB does not protect against CIA by activating existing CII-specific regulatory T cells, they do not preclude the possibility that EtxB influences differentiation of recently activated T cells, causing them to acquire regulatory function. LUROSS ET AL Interestingly, we found that CD4⫹CD25⫹ cells from EtxB-treated mice not given CII/CFA were unable to suppress IFN␥ by CII-responsive ILN cells (data not shown). Altered T cell activation by EtxB could result from its trafficking to lymphoid tissues that drain sites of antigen/CFA challenge. This is unlikely, however, because relatively low doses of EtxB are protective, and high-affinity receptors for the B subunit are found throughout the body. Alternatively, recently activated CII-primed cells may enter sites draining EtxB and encounter a modified microenvironment during a critical period in their differentiation. Uncommitted activated T cells recirculate throughout peripheral lymphoid tissues (39), and EtxB can modulate the activation of several cell types (16). For example, EtxB triggers IL-10 secretion by monocytes and inhibits their release of IL-12 (40). If IL-10 secretion was stimulated in the lymphoid tissues draining mucosal EtxB, it might favor generation of regulatory T cells, as described in other systems (36). Such an effect would influence the immune response to any immunizing antigen. It is conceivable that the regulatory T cells identified in our experiments are triggered following recognition of EtxB itself. The activation and differentiation of EtxB-specific T cells would also occur within the modified local environment that is likely established following B subunit–receptor interaction. As a result, EtxB-specific regulatory T cells may be generated and enter distant sites of inflammation, where they suppress the ongoing immune response in an antigen-nonspecific manner. In support of this premise, endogenous CD4⫹CD25⫹ regulatory cells require activation through T cell receptor engagement to become suppressive, but once stimulated, their subsequent regulatory activity is antigen nonspecific (41). A final possible explanation of why enhanced regulatory activity is found within the CD4⫹CD25⫹ spleen cell population after EtxB administration and CII/CFA challenge is that either soluble factors (e.g., IL-10, transforming growth factor ␤, prostaglandins) or non–antigen-specific cells (e.g., natural killer T cells) are released at sites local to mucosal B subunit delivery. These may then influence the nature of T cell differentiation at distant inflammatory tissues. Further investigations will be required to determine the specificity of the regulatory T cell population triggered by EtxB and to define the exact processes by which EtxB enhances their activity. In conclusion, our study highlights the potential of EtxB as a therapeutic agent in RA. Furthermore, the finding that disease prevention by EtxB is linked to PREVENTION OF AUTOIMMUNE ARTHRITIS BY ExtB enhanced activity of a population of regulatory T cells indicates that EtxB may have wider applicability as a means of interfering in the processes driving related inflammatory disorders. Elucidation of the precise mechanisms by which EtxB–receptor interaction enhances regulatory cell activity will be required in order to maximize the chances of its successful application in human disease. ACKNOWLEDGMENTS We thank C. J. Elson and S. Parry for their helpful advice, S. Sreckovic for his technical expertise in cell sorting, J. Baker and A. J. Coad for assistance with histology, and M. Kenny for toxin preparation and purification. REFERENCES 1. Feldmann M, Brennan FM, Maini RN. Rheumatoid arthritis. Cell 1996;85:307–10. 2. Myers LK, Rosloniec EF, Cremer MA, Kang AH. Collageninduced arthritis, an animal model of autoimmunity. Life Sci 1997;61:1861–78. 3. Trentham DE, Townes AS, Kang AH. Autoimmunity to type II collagen: an experimental model of arthritis. J Exp Med 1977;146: 857–68. 4. Courtenay JS, Dallman MJ, Dayan AD, Martin A, Mosedale B. Immunisation against heterologous type II collagen induces arthritis in mice. Nature 1980;283:666–8. 5. Wooley PH, Luthra HS, Stewart J, David CS. Collagen arthritis in mice: an MHC linked disease. Federation Proc 1981;40:971. 6. Gergersen PK, Silver J, Winchester RJ. The shared epitope hypothesis: an approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum 1987;30: 1205–13. 7. Becker KG, Simon RM, Bailey-Wilson JE, Freidlin B, Biddison WE, McFarland HF, et al. Clustering of non-major histocompatibility complex susceptibility candidate loci in human autoimmune diseases. Proc Natl Acad Sci U S A 1998;95:9979–84. 8. McIndoe RA, Bohlman B, Chi E, Schuster E, Lindhardt M, Hood L. Localization of non-Mhc collagen-induced arthritis susceptibility loci in DBA/1j mice. Proc Natl Acad Sci U S A 1999;96:2210–4. 9. Luross JA, Williams NA. The genetic and immunopathological processes underlying collagen induced arthritis. Immunology 2001; 103:407–16. 10. Mauri C, Williams RO, Walmsley M, Feldmann M. Relationship between Th1/Th2 cytokine patterns and the arthritogenic response in collagen-induced arthritis. Eur J Immunol 1996;26:1511–8. 11. Okamoto Y, Gotoh Y, Tokui H, Mizuno A, Kobayashi Y, Nishida M. Characterization of the cytokine network at a single cell level in mice with collagen-induced arthritis using a dual color ELISPOT assay. J Interferon Cytokine Res 2000;20:55–61. 12. Watson WC, Townes AS. Genetic susceptibility to murine collagen II autoimmune arthritis: proposed relationship to the IgG2 autoantibody sub-class response, complement C5, major histocompatibility complex (MHC) and nonMHC loci. J Exp Med 1985; 162:1878–91. 13. Joosten LAB, Helsen MMA, Saxne T, van de Loo FAJ, Heinegård D, van den Berg WB. IL-1 alpha beta blockade prevents cartilage and bone destruction in murine type II collagen-induced arthritis, whereas TNF-alpha blockade only ameliorates joint inflammation. J Immunol 1999;163:5049–55. 1681 14. Mauritz NJ, Holmdahl R, Jonsson R, van der Meide PH, Scheynius A, Klareskog L. Treatment with gamma-interferon triggers the onset of collagen arthritis in mice. Arthritis Rheum 1988;31: 1297–304. 15. Bessis N, Boissier MC, Ferrara P, Blankenstein T, Fradelizi D, Fournier C. Attenuation of collagen-induced arthritis in mice by treatment with vector cells engineered to secrete interleukin-13. Eur J Immunol 1996;26:2399–403. 16. Williams NA, Hirst TR, Nashar TO. Immune modulation by the cholera-like enterotoxins: from adjuvant to therapeutic. Immunol Today 1999;20:95–101. 17. Sun J-B, Rask C, Olsson T, Holmgren J, Czerkinsky C. Treatment of experimental autoimmune encephalomyelitis by feeding myelin basic protein conjugated to cholera toxin B subunit. Proc Natl Acad Sci U S A 1996;93:7196–201. 18. Bergerot I, Ploix C, Petersen J, Moulin V, Rask C, Fabien N, et al. A cholera toxoid-insulin conjugate as an oral vaccine against spontaneous autoimmune diabetes. Proc Natl Acad Sci U S A 1997;94:4610–4. 19. Tarkowski A, Sun JB, Holmdahl R, Holmgren J, Czerkinsky C. Treatment of experimental autoimmune arthritis by nasal administration of a type II collagen-cholera toxoid conjugate vaccine. Arthritis Rheum 1999;42:1628–34. 20. Sobel DO, Yankelevich B, Goyal D, Nelson D, Mazumder A. The B-subunit of cholera toxin induces immunoregulatory cells and prevents diabetes in the NOD mouse. Diabetes 1998;47:186–91. 21. Williams NA, Stasiuk LM, Nashar TO, Richards CM, Lang AK, Day MJ, et al. Prevention of autoimmune disease due to lymphocyte modulation by the B-subunit of Escherichia coli heat-labile enterotoxin. Proc Natl Acad Sci U S A 1997;94:5290–2. 22. Richards CM, Aman AT, Hirst TR, Hill TJ, Williams NA. Protective mucosal immunity to ocular herpes simplex virus type 1 infection in mice by using Escherichia coli heat-labile enterotoxin B subunit as an adjuvant. J Virol 2001;75:1664–71. 23. Bailey M, Williams NA, Wilson AD, Stokes CR. PROBIT: weighted probit regression analysis for estimation of biological activity. J Immunol Methods 1992;153:261–2. 24. Williams NA, Hill TJ, Hooper DC. Murine epidermal antigenpresenting cells in primary and secondary T-cell proliferative responses to herpes simplex virus in vitro. Immunology 1991;72: 34–9. 25. Beech JT, Bainbridge T, Thompson SJ. Incorporation of cells into an ELISA system enhances antigen-driven lymphokine detection. J Immunol Methods 1997;205:163–8. 26. Al-Sabbagh A, Nelson PA, Akselband Y, Sobel RA, Weiner HL. Antigen-driven peripheral immune tolerance: suppression of experimental autoimmune encephalomyelitis and collagen-induced arthritis by aerosol administration of myelin basic protein or type II collagen. Cell Immunol 1996;171:111–9. 27. Ortega G, Robb RJ, Shevach EM, Malek TR. The murine IL 2 receptor. I. Monoclonal antibodies that define distinct functional epitopes on activated T cells and react with activated B cells. J Immunol 1984;133:1970–5. 28. Sakaguchi S, Sakaguchi N, Asano M, Itoh M, Toda M. Immunologic self-tolerance maintained by activated T cells expressing IL-2 receptor alpha-chains (CD25): breakdown of a single mechanism of self-tolerance causes various autoimmune diseases. J Immunol 1995;155:1151–64. 29. Dutton RW, Bradley LM, Swain SL. T cell memory. Annu Rev Immunol 1998;16:201–23. 30. Thornton AM, Shevach EM. CD4⫹CD25⫹ immunoregulatory T cells suppress polyclonal T cell activation in vitro by inhibiting interleukin 2 production. J Exp Med 1998;188:287–96. 31. Shevach EM. Regulatory T cells in autoimmunity. Annu Rev Immunol 2000;18:423–49. 32. Read S, Malmstrom V, Powrie F. Cytotoxic T lymphocyte-associated antigen 4 plays an essential role in the function of 1682 33. 34. 35. 36. 37. CD25⫹CD4⫹ regulatory cells that control intestinal inflammation. J Exp Med 2000;192:295–302. Ruddock LW, Ruston SP, Kelly SM, Price NC, Freedman RB, Hirst TR. Kinetics of acid-mediated disassembly of the B subunit pentamer of Escherichia coli heat-labile enterotoxin. Molecular basis of pH stability. J Biol Chem 1995;270:29953–8. Boirivant M, Fuss IJ, Ferroni L, De Pascale M, Strober W. Oral administration of recombinant cholera toxin subunit B inhibits IL-12-mediated murine experimental (trinitrobenzene sulfonic acid) colitis. J Immunol 2001;166:3522–32. Myers LK, Seyer JM, Stuart JM, Kang AH. Suppression of murine collagen-induced arthritis by nasal administration of collagen. Immunology 1997;90:161–4. Roncarolo MG, Levings MK. The role of different subsets of T regulatory cells in controlling autoimmunity. Curr Opin Immunol 2001;12:676–83. Suri-Payer E, Amar AZ, Thornton AM, Shevach EM. CD4⫹CD25⫹ T cells inhibit both the induction and effector LUROSS ET AL 38. 39. 40. 41. function of autoreactive T cells and represent a unique lineage of immunoregulatory cells. J Immunol 1998;160:1212–8. Malmstrom V, Michaelsson E, Burkhardt H, Mattsson R, Vuorio E, Holmdahl R. Systemic versus cartilage-specific expression of a type II collagen-specific T-cell epitope determines the level of tolerance and susceptibility to arthritis. Proc Natl Acad Sci U S A 1996;93:4480–5. Iezzi G, Scheidegger D, Lanzavecchia A. Migration and function of antigen-primed nonpolarized T lymphocytes in vivo. J Exp Med 2001;193:987–94. Turcanu V, Hirst TR, Williams NA. Modulation of human monocytes by Escherichia coli heat-labile enterotoxin B-subunit: altered cytokine production and its functional consequences. Immunology. In press. Thornton AM, Shevach EM. Suppressor effector function of CD4⫹CD25⫹ immunoregulatory T cells is antigen nonspecific. J Immunol 2000;164:183–90.