Autoreactive CD4 T Cells Protect from Autoimmune ... - Cell Press

2 downloads 102 Views 296KB Size Report
Bryan Coon,* Tom Wolfe,* Jacob Petersen,† ... T cells (Lider et al., 1989; Nussenblatt et al., 1990; Miller ... documented production of TGF (Miller et al., 1993),.
Immunity, Vol. 11, 463–472, October, 1999, Copyright 1999 by Cell Press

Autoreactive CD41 T Cells Protect from Autoimmune Diabetes via Bystander Suppression Using the IL-4/Stat6 Pathway Dirk Homann,* Andreas Holz,* Adrian Bot,* Bryan Coon,* Tom Wolfe,* Jacob Petersen,† Thomas P. Dyrberg,† Michael J. Grusby,‡ and Matthias G. von Herrath*§ * Department of Neuropharmacology Division of Virology The Scripps Research Institute 10550 North Torrey Pines Road La Jolla, California 92037 † Novo Nordisk A/S Krogshojvej 53a Bagsvaerd DK-2880 Denmark ‡ Department of Immunology and Infectious Diseases and Harvard School of Public Health Department of Medicine Harvard Medical School Boston, Massachusetts 02115-6017

Summary Targeted immune regulation can be achieved by use of tissue-specific T cells and offers the potential for organ-specific suppression of destructive autoimmune processes. Here, we report the generation and characterization of insulin B chain–specific “autoreactive” CD41 regulatory T cells that locally suppress diabetogenic T cell responses against an unrelated selfantigen (viral transgene) in a virus-induced model for type 1 diabetes. Interleukin 4 (IL-4) is essential for prevention of diabetes since regulatory T cells cannot be induced in the absence of IL-4 or stat6 (IL-4 signaling pathway). Our observations demonstrate that autoreactive regulatory T cells can suppress autoreactive destructive T cell activity of differential antigenic specificity locally in the pancreatic draining lymph node, probably via cytokine-mediated modulation of antigen-presenting cells. Introduction Autoreactive lymphocytes are usually thought of as destructive mediators of autoimmune disease. However, autoreactive lymphocytes capable of amelioration or prevention of autoimmune disease were also described and termed regulatory cells. In type 1 diabetes, a T cell–mediated autoimmune disease characterized by the specific destruction of insulin-producing b cells in the pancreatic islets of Langerhans, splenic b cell– reactive CD41 T cells as well as thymic regulatory cells were shown to prevent diabetes after adoptive transfer into prediabetic nonobese diabetic (NOD) mice (Sempe et al., 1994; Akhtar et al., 1995; Herbelin et al., 1998). A promising strategy to therapeutically induce and expand such regulatory cells in vivo has been collectively called § To whom correspondence should be addressed (e-mail: matthias@ scripps.edu).

“oral tolerance” and refers to prevention and treatment of autoimmune diseases by oral administration of antigens related to autoimmune target tissue-specific antigens. While high-dose regimens were shown to result in deletion or “anergy” of autoreactive lymphocytes specific for the administered antigen, low to medium dosages represent a form of mucosal immunization since their success relies on the induction of antigen-specific regulatory cells (Weiner et al., 1994; Weiner, 1997). In a variety of animal models for autoimmune disease, including type 1 diabetes, the generation of such regulatory cells by oral immunization (oral tolerance) has been described: initially, these cells were reported to be CD8 T cells (Lider et al., 1989; Nussenblatt et al., 1990; Miller et al., 1991); subsequent reports documented disease suppression by CD8 and CD4 T cells in experimental autoimmune encephalitis (EAE) (Chen et al., 1994, 1995) as well as regulatory CD4 T cells in collagen-induced arthritis (Tada et al., 1996), diabetes (Bergerot et al., 1994), and tolerance to ovalbumin (Garside et al., 1995a, 1995b) or bovine alpha s1-casein in splenocyte-reconstituted SCID mice (Yoshida et al., 1998). Recently, TCRgd T cells (reviewed in Kapp and Ke, 1997) have been demonstrated to mediate oral tolerance in experimental autoimmune uveitis (EAU) (Wildner and Thurau, 1995) and to ovalbumin (McMenamin et al., 1994, 1995; Mengel et al., 1995; Ke et al., 1997). CD8/TCRgd T cells capable of diabetes prevention in NOD mice have also been reported after aerosol insulin (Harrison et al., 1996). However, not much is known about their precise antigenic specificities and effector mechanism(s). The special immune environment for lymphocyte priming at large mucosal surfaces may be critical in the generation of regulatory cells, which are characterized by their ability to secrete cytokines such as interleukin 4 (IL-4), IL-10, and TGFb and can act as suppressive regulatory cells if they are able to home to an autoimmune targeted site and prevent disease. This mechanism has been termed “bystander suppression” if the specificity of regulatory cells differs from that of the autoreactive destructive lymphocytes. Bystander suppressor cells have been suggested in EAE (Miller et al., 1991, 1993; al-Sabbagh et al., 1994; Chen et al., 1994) and arthritis (Zhang et al., 1990; Yoshino et al., 1995), and their existence was strongly suggested in our virusinduced diabetes model (von Herrath et al., 1996; von Herrath and Homann, 1997). In some cases, phenotypic analyses were performed to characterize bystander suppressor cells (Miller et al., 1991), and other studies documented production of TGFb (Miller et al., 1993), IL-4, and IL-10 (von Herrath et al., 1996; Weiner, 1997). To our knowledge, only one study has demonstrated phenotypic and functional analysis of bystander suppressor T cells induced by oral antigen administration: EAE was prevented by CD41 regulatory cells that secreted TGFb, IL-4, IL-10, and IFNg (Chen et al., 1994). However, these regulatory clones were structurally identical to Th1-type encephalitogenic CD41 clones in T cell receptor usage, MHC restriction, and epitope recognition. This and further observations in a transgenic T cell

Immunity 464

Table 1. Adoptive Transfer of Insulin-Specific, IL-4-Producing Regulatory Cells Protects Prediabetic RIP-NP Recipients

Amount and Source of Transferred Cells

Cells Stimulated in Presence of

No transfer 2–2.5 3 105 CD41 T cells (PDLN, protected donor) 5 3 106 splenocytes (protected donor) 5 3 106 splenocytes (diabetic donor)

N/A B chain/IL-4 B chaina B chain LCMV-NP118-126 B chain LCMV-NP118-126

Cytokines Produced by Transferred Cells IFNg

IL4

Diabetes Incidence

N/A 0.12 0.12 0.33 3.21 1.40 2.80

N/A 0.10 0.00 0.90 0.16 0.11 0.13

100% 0% 100% 50% 100% 83% 100%

Regulatory cell lines derived from pancreas draining lymph node (PDLN) of RIP-NP mice receiving oral insulin treatment were expanded in the presence of porcine insulin B chain, IL-2, and IL-4. After 2 months, IL-4 was withdrawn and cells subsequently tested in vitro (cytokine profiles) and in vivo (adoptive transfers). In some experiments, IL-4 was withdrawn for more than 2 months, resulting in a cytokine profile switch as indicated. Additional lines were derived from spleens of orally treated mice and propagated on porcine B chain or the MHC class I–restricted LCMV-NP118-126 peptide (RPQASGVYM) as a control. Recipient RIP-NP mice were infected with LCMV i.p. and 5 days post infection received 2 to 2.5 3 105 PDLN-derived regulatory cells i.v. or 5 3 106 cultured splenocytes i.p. Blood glucose levels were monitored weekly, and diabetic mice were euthanized after two consecutive blood glucose measurements of .300 mg/dl. IL-4 but not IFNg production by PDLNderived regulatory cells correlated with protection. Differential stimulation of spleen cell lines also showed that increased IL-4 production was associated with protection. Increase of IFNg secretion by stimulation with LCMV-NP118-126 completely abrogated protective capacity of transferred spleen regulatory cells. Cytokine production is expressed in ng/100 ml tissue culture supernatant after stimulation cultures of cell numbers as indicated for transfers. All ELISAs were performed three times, and SE was ,15%. a.2 months IL-4 withdrawal.

receptor (TCR) system (Chen et al., 1996) argued that antigen-specific autoreactive T cells can differentiate in vivo into autoaggressive or regulatory cells depending on the site and context of the initial priming event. However, no model has so far clearly demonstrated bystander suppression by using two distinct, defined selfantigens. To address the potential mechanism of bystander suppression in type 1 diabetes, we took advantage of the well-defined primary self (LCMV)-antigen in the RIPLCMV transgenic system of virus-induced autoimmune diabetes (von Herrath et al., 1996). The transgenic RIPNP line used in this report expresses the nucleoprotein (NP) of lymphocytic choriomeningitis virus (LCMV) under the control of the rat insulin promoter (RIP) in b cells of the pancreatic islets of Langerhans (Oldstone et al., 1991; von Herrath et al., 1994a, 1994b). Only after infection with LCMV, progressive destruction of b cells that express the viral NP eventually leads to overt type 1 diabetes following approximately 1 week after viral clearance. Development of disease is dependent on NPspecific CD81 T cells, CD41 T help, as well as IFNg (von Herrath and Oldstone, 1997). We have shown previously that oral insulin can prevent diabetes in these mice (von Herrath et al., 1996). Protection is associated with a “mixed” “TH1/TH2” type cytokine profile in the pancreas and a local (pancreas) but not systemic abrogation of NP-specific (“anti-self”) CTL activity. While these results strongly suggested bystander suppression as the mechanism for oral tolerance induction, a specific regulatory T cell population had not been directly demonstrated, and it is not precisely understood how and where TH2 cytokines could suppress autoreactive CTL. In this report, we derived cell lines from pancreatic draining lymph nodes of RIP-NP mice protected from type 1 diabetes by porcine insulin feedings. Lymphocytes were cultured in the presence of the immunodominant insulin B chain (Daniel and Wegmann, 1996; Polanski et al., 1997; Homann et al., 1999) or B chain peptides. Analysis of established cell lines revealed a cytokine

profile similar to that reported for other regulatory cells, namely production of IL-4, IL-10, and some IFNg (Weiner et al., 1994; von Herrath et al., 1996; Weiner, 1997). When transferred i.v. into prediabetic RIP-NP mice, regulatory cells proliferated only in the pancreatic draining node of recipients and completely protected from autoimmune diabetes by locally suppressing the virus (self)-specific, “diabetogenic” T cell responses. IL-4 and the signal transducer and activator of transcription 6 (stat6), a critical component of the IL-4 signaling pathway (Kaplan et al., 1996), were essential for induction of regulatory cells as demonstrated by the failure to efficiently protect RIPNP 3 IL-42/2 or RIP-NP 3 stat62/2 mice from diabetes by oral insulin treatment. Our results show that diabetes prevention likely occurred through IL-4/stat6-mediated modulation of antigen-presenting cells (APCs) resulting in decreased expansion and cytokine secretion of autoreactive CD81 and CD41 T cells. These findings provide direct evidence for bystander suppression in a dual selfantigen-specific model. Results Insulin B Chain–Specific Regulatory T Cells Are Generated by Oral Insulin Immunization and Prevent Diabetes after Adoptive Transfer into Prediabetic Recipients Based on our previous studies using oral administrations of porcine insulin to prevent diabetes in RIP-NP mice (von Herrath et al., 1996; Homann et al., 1999), we hypothesized that this procedure generated regulatory T cells with likely specificity for the immunodominant insulin B chain. Given the different T cell phenotypes among regulatory cells, crude lymphocyte cultures were established from pancreas draining lymph nodes (PDLN) of protected or diabetic mice treated with oral insulin as well as nonfed mice. Cultures were maintained in the presence of insulin B chain and IL-2, as well as IL-4. Only cultures from protected mice survived long enough to allow for sufficient expansion to test the cell lines in

Bystander Regulatory T Cells Act via IL-4/Stat6 465

Figure 1. Pancreatic Immunohistochemistry of Diabetic and Protected Mice/Trafficking of Regulatory T Cells (A and C) CD4 and CD8 lymphocytes found in islets of nonprotected diabetic RIP-NP mice. The histology shown is representative for mice receiving CD41 “regulatory” cell transfers from diabetic donors or no cell transfers. Pancreata were harvested on day 21 post LCMV. (B and D) CD4 and CD8 lymphocytes found in islets of protected nondiabetic RIP-NP mice that received regulatory cell transfers (day 5 post LCMV); pancreata were harvested on day 21 post LCMV. (E and F) Pancreatic draining node (PDLN [E]) or pancreas (F) from RIP-NP transgenic mouse receiving transfer of BrdU-labeled lymphocytes from a RIP-NP donor treated with oral insulin and protected from diabetes. Splenocytes from oral insulin–protected or unprotected mice were cultivated for 3 days in the presence of IL-4 insulin B chain and labeled overnight with BrdU (prior to transfer). Cells (5 3 106) were transferred into prediabetic (day 5 post LCMV) RIP-NP recipients. Organs were harvested 36 hr later and stained for BrdU. Significant amounts of labeled cells were only found in the PDLN (E), very few cells in the spleen (data not shown), and none in the pancreas (F). The experiment was performed three times.

vitro and in vivo. Phenotypic analysis of these cell lines demonstrated a homogenous population of TCRab bearing CD4hi T cells. No expression of CD8 or TCRgd was detected (data not shown). Adoptive transfers of 2 to 2.5 3 105 regulatory CD41 T cells into prediabetic RIP-NP mice (5 days after infection with LCMV) led to complete protection from type 1 diabetes (Table 1). Interestingly, long-term withdrawal of IL-4 from cultures, while neither affecting regulatory cell proliferation in the presence of insulin B chain nor their surface phenotype, abolished this protective effect (Table 1). Histologically, diabetes prevention was associated with a mild perinsulitis consisting mainly of CD41 T cells. In contrast, nonprotected mice showed pronounced CD81 T cell infiltration and destruction of islets (Figures 1A and 1C versus 1B and 1D). In control experiments, splenocytes from insulin-fed protected and unprotected RIP-NP mice were also stimulated in the presence of insulin B chain (and in some instances IL-4) or, alternatively, in the presence of the immunodominant, MHC class I–restricted LCMVNP118-126 peptide. Only splenocyte cultures from protected mice stimulated in the presence of B chain were able to efficiently protect prediabetic RIP-NP recipients, although to a lower degree (50%) than the PDLN-derived cell lines. In contrast, stimulation in the presence of the

immunodominant LCMV NP118-126 peptide abrogated the protective capacity of such cells (Table 1). The minimal protection (17%) observed after adoptive transfer of insulin B chain stimulated splenocytes from oral insulin– treated yet diabetic mice suggested that oral insulin led to the generation of regulatory cells, however presumably at frequencies too low to prevent diabetes in such donors. CD41 Regulatory T Cells Produce IL-4, IL-10, and IFNg Cytokine profiles of regulatory T cell cultures were determined by ELISA and flow cytometry. As shown in Figure 2A, protective regulatory CD41 T cells produced IL-4, IL-10, as well as IFNg in response to stimulation with insulin B chain. The loss of protective function following IL-4 withdrawal from cultures was accompanied by a switch in cytokine profile: retention of IFNg production but dramatically reduced IL-4 and IL-10 secretion (Figure 2A; Table 1). Similar results were obtained after polyclonal stimulation of protective and nonprotective CD41 T cell lines and enumeration of cytokine-producing cells by intracellular cytokine staining (Figure 2B). In addition, TNFa, which is largely membrane bound and not detectable under the ELISA conditions employed, was found in approximately 2.5-fold more nonprotective than protective CD41 T cells by intracellular cytokine staining

Immunity 466

Figure 2. Cytokine Profiles of CD41 Regulatory T Cells Regulatory T cells were recovered from the pancreas draining lymph node of RIP-NP mice treated with oral procine insulin and cultured under conditions detailed in the Experimental Procedures. Cultured cells were exclusively CD41, CD82, TCRab1, and TCRgd2. Protective but not nonprotective regulatory cells produced IL-4 and IL-10. (A) Regulatory cell cultures were stimulated for 3 days with procine B chain, and cytokines secreted into supernatant were analyzed by ELISA. After polyclonal stimulation with plate-bound antiCD3, nonprotective regulatory cells demonstrated z800-fold increased IFNg production (37.7 6 9.3 ng/50 ml) as well as some IL-4 (10.0 6 2.5 pg/50 ml) and IL-10 (7.7 6 4.2 pg/50 ml) production (data not shown). (B) To determine cytokine profiles by flow cytometry, cells were stimulated for 5 hr with PMA/ionomycin in the presence of Brefeldin A prior to surface and intracellular staining.

(Figure 2B). No IL-2 could be detected by either method. A comparable shift in cytokine production was also observed in splenocyte cultures differentially stimulated with insulin B chain or the LCMV-NP118-126 peptide (Table 1). Such pronounced differences in cytokine production were not observed in splenocyte cultures derived from insulin-fed but diabetic mice. These data demonstrate that oral insulin immunization can generate specific CD41 T regulatory cells that can be isolated and expanded in vitro and completely protect prediabetic RIPNP recipients from type 1 diabetes after adoptive transfer, if they are secreting IL-4 and IL-10 (Figure 2). Regulatory T Cells Are Not Generated in the Absence of IL-4 or Stat6 Earlier studies using RIP-NP mice that overexpressed IL-4 or IL-10 in the pancreatic islets have shown that IL-4 had a clear protective effect whereas IL-10 slightly enhanced diabetes following LCMV infection (Lee et al., 1994; R. Muller, personal communication). Furthermore, insulin-immunized, protected RIP-NP mice demonstrated

a more pronounced increase of IL-4 than IL-10 production in the pancreas (von Herrath et al., 1996). Based on these findings, we therefore hypothesized that IL-4 might be the key regulatory cytokine for suppression of diabetes onset and tested this notion using oral insulin immunization in IL-4-deficient mice that were crossed to the RIP-NP background. As shown in Figures 3A and 3B, protection could be readily induced in transgenic mice heterozygous for IL-4 but did not occur in IL-4deficient RIP-NP mice treated with porcine insulin. Further studies were initiated in order to understand whether the highly specific stat6-dependent IL-4 signaling pathway (Kaplan et al., 1996) was required for effective generation of regulatory T cells in RIP-NP mice. Insulin feeding did not prevent diabetes in stat6-deficient RIP-NP mice, whereas it protected in stat6-competent littermates (Figures 3C and 3D). Interestingly, a gene dose effect of the stat6 gene appears possible, since protection occurred to a somewhat lesser extent in oral insulin–fed stat6 heterozygous mice (Figure 3D). Absence of IL-4 or stat6 did not affect the kinetics of virus

Figure 3. RIP-NP Mice Develop Type 1 Diabetes in the Absence of IL-4 or Stat6 but Are Refractory to Induction of Regulatory T Cells by Oral Insulin Immunization Mice were bred and genotyped as detailed in the Experimental Procedures. Diabetes was defined as two consecutive blood glucose measurements of .300 mg/dl. (A and B) Incidence of diabetes in untreated (vehicle) and oral porcine insulin–treated (0.5 mg twice per week) RIP-NP mice that were IL-41/1, IL-41/2, or IL-42/2. There were no significant differences between IL-41/1 and IL-41/2 groups; thus, only the data for IL-41/2 mice are displayed. (C and D) Incidence of diabetes in untreated and oral insulin–treated RIP-NP mice that were stat61/1, stat61/2, or stat62/2. Mice were followed up to 6 months post infection, and no further cases of diabetes were observed over this timespan. The size of the experimental groups is noted in parentheses. Significant (p , 0.05, log-rank test) differences were found between oral insulin– treated IL-41/- and IL-42/2 mice and between oral insulin–treated stat61/1 and stat62/2 mice.

Bystander Regulatory T Cells Act via IL-4/Stat6 467

clearance (data not shown) nor the generation of LCMVspecific cytotoxic T lymphocytes (CTL) since comparable systemic CTL activity was detected in all mice tested (Figure 4A). However, in contrast to diabetic mice, RIPNP mice protected by oral insulin treatment did not have detectable virus-specific, i.e., autoreactive destructive, T cells in the pancreas. These results demonstrate the presence of local immune regulation without apparent effect on the systemic immune response. The Pancreas Draining Lymph Node Is a Critical Compartment for Regulatory T Cell Function C. Kurts and his colleagues (Kurts et al., 1997, 1998, 1999) had identified the PDLN as an essential checkpoint in the development of autoimmune diabetes. To analyze the distribution and localized activity of transferred regulatory T cells in our prediabetic RIP-NP recipients, tracking experiments with bromodeoxyuridine (BrdU)- and CFSE-labeled regulatory cells were performed. Spleen, lymph nodes (axillary, mesenteric, inguinal, and PDLN), and pancreas were harvested 2, 4, 24, 36, and 48 hr after transfer and analyzed for the presence of labeled cells. While transferred cells were readily detected in all lymphatic organs, (data not shown), no cells labeled by either method could be detected in the pancreas (Figure 1F). However, when analyzed for proliferation, transferred CD41 T cells only divided in the PDLN but not other lymphoid organs, as early as 4 hr post transfer into prediabetic RIP-NP mice (Figure 4C), and were detectable there at significant numbers for a prolonged period of time (36 hr, Figure 1E, and up to 3 days, data not shown). In accordance with the selective local proliferation of regulatory cells, suppression of LCMV-specific T cell responses was found only in the PDLN in vivo (Figure 4B, intracellular cytokine staining), demonstrating that the PDLN is the essential compartment for negative regulation of destructive, diabetogenic T cell responses by autoreactive regulatory lymphocytes. IL-4 Differentially Modulates the Function of Stat6-Competent and -Deficient APCs While the local suppression of LCMV-specific T cell responses in the PDLN results from regulatory T cell activation by cross-presented insulin B chain, it remains unclear how insulin B chain–specific CD4 T cells can

Figure 4. Local Suppression of Virus-Specific CD81 T Cell Response in Pancreas and PDLN-Selective Proliferation of CD41 Regulatory T Cells in PDLN (A) Splenic and pancreas infiltrating lymphocytes were prepared 45 days after LCMV infection of IL-4 or stat6-deficient RIP-NP mice and their IL-4 or stat6 wild-type littermates. Protection was achieved by oral insulin administration. Lymphocytes were restimulated and assayed for CTL activity as detailed in the Experimental Procedures. No significant differences in CTL activity were observed in spleens of IL-4 or stat6-competent or -deficient mice. However, protected mice exhibited markedly reduced CTL activity in the pancreas. ND, not determined. (B) After adoptive transfer of splenic CD41 regulatory T cells, protected and diabetic RIP-NP mice were analyzed by flow cytometry for presence of LCMV-NP-specific, IFNg-producing CD81 T cells

in different lymphocyte compartments. Protected mice exhibit a selective suppression of NP-specific CD8 T cells in the PDLN. (C) CD41 regulatory T cell cultures were established from spleens of protected and diabetic mice and labeled with CFSE prior to adoptive transfer into prediabetic RIP-NP recipients as detailed in the Experimental Procedures. A left shift in fluorescence intensity indicates proliferation of labeled cells. Panels are gated on CFSE1 CD41 T cells. CD4 regulatory cells from protected donors proliferate selectively in the PDLN of recipients 4 hr post transfer (upper panel). No proliferation was observed in other lymphatic organs or after transfer of similarly treated cells from diabetic donors (lower panel). Further proliferation was observed at later time points post transfer (data not shown), indicating that regulatory cells (although reaching multiple lymphoid organs) selectively proliferated in the PDLN, and only there were they found in detectable numbers at later time points (compare Figure 1E, BrdU labeling).

Immunity 468

Table 2. IL-4 Suppresses Virus-Specific T Cell Effector Function via Modulation of APCs

APC Genotype 1 /1

Effector T Cell Genotype 1/1

stat6

stat6

stat62/2

stat62/2

stat61/1

stat62/2

Cytokines Added to Culture

LCMV-Specific Cytotoxicity (% 51Cr Release)

none IL-4 IL-12 none IL-4 IL-12 none IL-4

10 0 38 15 60 20 13 0

CD41 T Cells

CD81 T Cells 1

% of Total

% IFNg

% of Total

% IFNg1

15 36 20 10 14 20 18 20

6 2 23 11 10 16 7 5

65 10 68 50 59 40 70 8

23 15 64 51 71 43 28 12

Splenocytes were harvested from LCMV-infected stat6-deficient or -competent mice and cultured in the presence of IL-4 (8 ng/ml), IL-12 (4 ng/ml), or no cytokines. APCs were LCMV-infected/uninfected or NP118-126-pulsed/nonpulsed, irradiated (2000 rads) stat6-deficient or -competent PECs. 2 3 105 APC/well were added to all cultures kept in 24-well plates. After 6 days, supernatants were harvested for ELISAs, and cells were harvested for CTL assays and FACS analysis. Staining for intracellular IFNg was preceded by a 5 hr restimulation with anti-CD3 and anti-CD28. In some experiments, IL-4 (4 ng/ml) was directly added to the chromium release assay. No suppression of CTL activity was observed; however, there was a slight tendency toward increased bulk CTL activity (10%–15% above regular killing). Results were obtained in two (FACS) to four (CTL assay) independent experiments; SE was , 6 15% for all values shown.

lead to suppression of CD8 T cells specific for an unrelated antigen (LCMV-NP). We hypothesized that rather than, or in addition to, acting directly on T cell effectors, regulatory T cells might induce a modulation/“deactivation” of APCs that subsequently fail to promote or actively suppress T cell activation to other presented antigens. To test this hypothesis, we analyzed the effect of IL-4 and IL-12 (control) on expansion and activation of LCMV-specific T cell effectors restimulated by LCMVinfected APCs or APCs pulsed with the immunodominant LCMV NP118-126 peptide (Table 2). APCs were peritoneal exudate cells (PEC) expressing uniformly CD11b while only a small fraction also coexpressed CD11c. In accordance with our in vivo findings (Figure 4B), addition of IL-4 but not IL-12 to restimulation cultures reduced expansion and LCMV-specific cytolytic activity as well as the numbers of IFNg producing CD81 and CD41 virusspecific effector T cells. While IL-4 induced slight changes in expression of costimulatory molecules (such as 25% CD40 upregulation on PECs), no other pronounced differences were observed. In contrast, when the same experiment was performed with stat6-deficient T cells and APCs, there was no decrease in anti-LCMV cytotoxicity, which correlated with the inability of oral insulin to provide protection of stat6-deficient RIPLCMV mice in vivo (Figure 3). Importantly, the differential effect of IL-4 on stat6-deficient cultures allowed us to test whether IL-4 predominantly acted on APCs or T cells themselves. When stat6-deficient effector T cells were cultured in the presence of wild-type APCs, the suppressive effect of IL-4 was completely restored. A similar but lesser effect was observed, when supernatant from regulatory lymphocyte cultures was directly added to secondary LCMV T cell cultures (data not shown). Thus, IL-4 acts via the stat6-signaling pathway in APCs resulting in the lack of expansion of LCMVspecific CD41 and CD81 T cell effectors. The precise changes induced in APCs are not yet known and are the subject of current studies. The in vivo relevance of this finding for diabetes pathogenesis is shown by the selective suppression of autoreactive (anti-LCMV-NP) CD81 responses that we documented in the pancreatic draining node of protected mice (Figure 4B).

Discussion In the present report, we document that autoreactive, regulatory T cells specific for one self-antigen (insulin B chain) can locally decrease activity of destructive autoreactive lymphocytes specific for a different self-antigen (LCMV-NP) in a model of type 1 diabetes and therefore act as “true” bystander suppressors. The effect is highly dependent on the ability of these regulatory CD41 T cells to produce IL-4, which decreases the activity of autoreactive destructive T cells depending on the stat6 pathway in APCs, since it neither occurs after IL-4 deprivation of isolated B chain–specific regulatory cells nor in the absence of stat6-competent APCs (see also Figure 3). Thus, autoreactive regulatory cells can be a potent tool to treat IDDM, even if the primary eliciting selfantigen is not precisely known. In contrast to oral antigen treatment, which protected 50%–60% of treated mice (von Herrath et al., 1996; Homann et al., 1999), adoptive transfer of in vitro expanded B chain–specific regulatory cells resulted in complete protection. This high degree of protection after transfer of B chain–specific CD41 regulatory cells observed in our model is unprecedented, indicating that there is a limiting factor for induction of regulatory cells by oral antigens that are activating these lymphocytes in the gut. This could be due to insufficiency of local antigen presentation in the gut, or, alternatively and more likely, to limited numbers (availability) of these regulatory cells among the peripheral autoreactive T cell repertoire. This conclusion has important potential consequences for the treatment of human autoimmune diabetes. It means that increasing the numbers of regulatory lymphocytes in vitro and transferring them back could be an effective “second-generation” immunotherapeutic approach. Immunodominance of the insulin B chain was first reported for diabetes prevention studies using subcutaneous insulin administration (Muir et al., 1995; Daniel and Wegmann, 1996). While tolerance induction after parenteral insulin administration is likely to have a mechanism different from mucosal tolerance induction, oral B chain has also been associated with a shift from TH1 to TH2 cytokines (Maron et al., 1998) and diabetes protection in the NOD cotransfer model (Polanski et al., 1997).

Bystander Regulatory T Cells Act via IL-4/Stat6 469

Since TGFb production did not strictly correlate with protection (data not shown), although it might still be a necessary immunosuppressive factor downstream or upstream of IL-4 function, we further investigated the role of IL-4 in bystander suppression. IL-4 is produced by many regulatory cells and has also been shown to exhibit a synergistic effect when given orally in conjunction with low-dose oral antigen treatment (Inobe et al., 1998). Furthermore, IL-4 is also a differentiation factor for TGFb secreting TH3 cells (Inobe et al., 1998). Although impaired mucosal immune responses were reported in IL-4-deficient mice (Vajdy et al., 1995), oral tolerance induction was successful with a high-dose regimen (Garside, 1995), suggesting that IL-4-independent deletion/anergy of autoreactive T cells rather than the generation of IL-4-deficient regulatory cells was the mechanism for protection. In our RIP-NP model, diabetes induction was not affected by the absence of IL-4, but IL-4-deficient RIP-NP mice were refractory to regulatory T cell induction with oral insulin. To further corroborate these findings, protection studies were also conducted with stat6-deficient RIP-NP mice. The signal transducer and activator of transcription 6 is required for mediating responses to IL-4 and development of TH2 cells (Kaplan et al., 1996) but likely not for their survival (Vella et al., 1997). Similarly to the IL-4-deficient mice, STAT6-deficient RIP-NP mice mounted an effective CTL response (Figure 4A) (Ihle et al., 1998) to eliminate the virus and developed type 1 diabetes with comparable kinetics (Figure 3). Again, as in IL-4-deficient mice, protection from type 1 diabetes by oral insulin administration was not achieved (Figure 3). These findings are in good agreement with previous studies by Nora Sarvetnick and her colleagues (Gallichan et al., 1999) showing isletreactive TH2 regulatory cells in NOD mice over-expressing IL-4 in their pancreatic islets. However, their experiments did not provide any information about antigen specificity or the mechanism of action of such lymphocytes. The PDLN has recently been shown to be a critical checkpoint in the development of diabetes (Kurts et al., 1997). Our observations of regulatory T cell proliferation (Figure 4C) and suppression of diabetogenic T cell activity (Figure 4B) in the PDLN suggest that it also serves as a checkpoint for immune regulation. Such targeted immune regulation could circumvent problems associated with systemic suppression of immune responses and is thus likely to have fewer undesirable effects. The differential specificity of regulatory and diabetogenic T cells allows in principle for immune regulation of destructive, autoreactive T cell activity without a precise knowledge of their specificity. The “tertium datur” in this scenario are likely APCs, which were shown to suppress or support virus-specific T cell activity in the presence of IL-4 or IL-12, respectively (Table 2). While the precise changes in APC function are currently not understood, our findings suggest a powerful role for the stat6 signaling pathway: while IL-4 did not inhibit the lytic function of already differentiated effector virus-specific effector T cells, T cell expansion and IFNg production were significantly decreased (Table 2). This occurred in the presence of levels of externally added IL-4 that were not higher than those found in supernatants of regulatory lymphocyte cultures. Importantly, the differential effect

of IL-4 on T cell effector function in the presence of stat6competent compared to stat6-deficient APCs allows the conclusion that a significant portion of the suppressive effect of IL-4 can be mediated by stat6-competent APCs. The following scenario emerges for the employment of bystander suppression to treat diabetes as well as possibly other autoimmune diseases. Regulatory cells primed against a target organ-specific autoantigen via exposure to this antigen on mucosal surfaces, or potentially by other routes (Coon et al., 1999) traffic through the body and will reach the draining node of the diseased organ. Self-antigens (such as insulin) released by an ongoing autoimmune process are cross-presented by resident APCs selectively in this lymphocyte compartment and lead to further activation of the regulatory T cells. Resulting IL-4 secretion by the regulatory T cells in turn modifies APC activity leading to local silencing of autoaggressive lymphocytes specific for the diseaseinitiating autoantigen presumably cross-presented on the same APC. Thus, such regulatory cells are only efficient if they are activated and present during organ destruction. In conclusion, the induction of regulatory cells is of particular interest, since it may overcome two fundamental problems that have hampered antigenspecific immune interventions that directly target autoreactive destructive lymphocytes. First, the specificity of initiating destructive, autoreactive cells is not known in most T cell–mediated autoimmune diseases, and second, epitope spreading may lead to a broadening of T cell autoreactivity in the target organ. Knowledge of all self-antigens involved in an autoimmune process is not required for induction of regulatory cells, which have to be directed to only one tissue-specific autoantigen involved in an ongoing autoimmune process. Experimental Procedures Mice Breeding Scheme and Origin Generation of H-2d RIP-LCMV-NP transgenic mice has been described previously (von Herrath et al., 1994). These mice were crossed to the H-2d (BALB/c) background for 10 generations and express the nucleoprotein of LCMV in the pancreatic b cells and thymus but not in any other organs. Such mice were intercrossed with stat6-deficient (Kaplan et al., 1996) or IL-4-deficient (Jackson Laboratories) H-2d (BALB/c) mice, and the resulting stat6 or IL-4 heterozygous NP-expressing or nonexpressing offspring were intercrossed for one generation. Offspring (F2) were used in all experiments and were stat61/1, stat61/2, or stat62/2 or IL-41/1, IL-41/2, or IL-42/2 and did or did not express LCMV-NP yielding six experimental groups each. Mouse Phenotyping The presence of the RIP-NP constructs was determined by slotblot hybridization or PCR using gene-specific probes or primer sets, respectively, as described (Oldstone et al., 1991; von Herrath et al., 1994a, 1994b). Stat6 animals were genotyped either by Southern blotting as described (Kaplan et al., 1996) or by PCR. Two independent standard PCR reactions were performed to test for presence of stat6 or stat6 gene copies disrupted by neomycin (neo) gene integration. The stat6 wild-type gene was detected by primer combination stat6-sense (59-GGCCGAGGCTTCACATTTTGGC-39) and STAT6-antisense (59-CCCGGATGACGTGTGCAATGG-39). A second reaction determined the presence of the neo gene using the primer pair neo-sense and neo-antisense (59-CCGGCCACAGTCGATGA ATCC-39). Similarly, PCR reactions were performed for IL-4. The reaction determined the presence of the IL-4 (59-GTGAGCAGATGA

Immunity 470

CATGGGC39 and 59-CTTCAAGCATGGAGTTTTCCC-39) or neo genes (59-CTTGGGTGGAGAGGCTATTC-39 and 59-AGGTGAGATGACAG GAGATC-39). Virus Virus used was LCMV strain Armstrong (ARM), clone 53b. LCMV was plaque purified three times on Vero cells and stocks prepared by a single passage on BHK-21 cells. Mice were infected with a single intraperitoneal dose of 105 pfu LCMV ARM. Oral Antigens Porcine insulin was provided by Novo Nordisk, Bagsvaerd, Denmark, and obtained as dry crystals from the very last stage of the purification process, immediately before formulation into the injectable product used for patients. Porcine, human, and murine (I1II) insulin B chains were also obtained from Novo Nordisk. Insulin and B chains were solubilized in acid buffer, pH adjusted, and the solution stored at 2208C until used. Oral antigen was administered via a blunt-ended curved feeding tube inserted into the esophagus/stomach. RIP-NP mice were fed biweekly with 0.5 ml of an aqueous solution containing 1 mg porcine insulin. Feeding was started 1 week prior to infection with LCMV and discontinued 9 weeks later. Control groups received 1 mg of bovine serum albumin (BSA) at same intervals. Generation and Expansion of Insulin B Chain–Specific Regulatory Cells Single cell suspensions of spleen, pancreas draining, and mesenteric lymph nodes from protected and diabetic RIP-NP mice fed porcine insulin as well as nonfed mice were prepared by disrupting and pressing organs through a steel mesh. CD41 T cells from pancreata were obtained by fractionated digestion with collagenase P (Boehringer Mannheim) and separation with anti-CD4 coated magnetic beads (Dynal). Cells were cultured in FCS-rich medium on irradiated (2000 rads) BALB/c peritoneal exudate cells (PEC) pulsed with 30 mg/ml porcine insulin B chain or B chain peptides. For some cultures, recombinant murine IL-4 was added at 40 ng/ml. After 1 week, cells were transferred to fresh PEC-coated plates and resuspended in FCS-rich medium containing 5% ConA stimulated rat spleen supernatant containing “T cell growth factor” (TCGF) and B chain as well as irradiated syngeneic spleen feeder cells. Stimulation with porcine B chain or peptides was gradually lowered to 2 mg/ml, IL-4 to 8 ng/ml, and cells were expanded using TCGF medium and feeder cells. After z2 months, IL-4 was withdrawn from some cultures, and cells were tested in vitro and in vivo z6 weeks later. Under these culture conditions, only cells from pancreas draining lymph nodes of protected mice could be maintained and expanded for extended time periods. In some cases, these cells were then recultured in the presence of 8 ng/ml IL-4. Stimulation cultures for LCMV-specific lymphocytes. Splenocytes obtained from LCMV-infected stat6-competent or -deficient mice 45 to 60 days post infection were stimulated in vitro for 6 days in the presence of LCMV-infected, irradiated PECs or coated with LCMV NP H-2d (RPQASGVYM) peptide (1025 M final concentration). In some experiments, IL-4 (8 ng/ml) or IL-12 (4 ng/ml) was added to stimulation cultures. IFNg and IL-4 in culture supernatants were quantitated by ELISA and the number of cytokine-producing CD41 or CD81 T cells by intracellular staining for flow cytometry as described below. Cytotoxicity Assays LCMV-specific CTL activity in spleens harvested 45 days after LCMV infection was assessed in a standard 4 to 5 hr 51Cr release assay on LCMV-infected and uninfected, MHC-matched (BALB/c17[H-2d]) and mismatched (MC57[H-2b] target cells. For determination of secondary LCMV-specific CTL activity, single cell suspensions of spleens or ficoll purified lymphocytes from pancreata harvested 45 days postinfection were cultured for 8–15 days on LCMV-infected, irradiated PECs prior to testing CTL activity in a 51Cr release assay.

CD11b (M1/70), CD11c (HL3), CD40 (HM40-3), CD44 (IM7), Il-2 (JES61A12, JES6-5H4, and S4B6), IL-4 (11B11, BVD6-24G2, and BVD41D11), IL-6 (MP5-20F3 and MP%-32C11), IL-10 (JES5-16E3 and SXC-1), IFNg (R4-6A2 and XMG1.2), and TNFa (MP6-XT22). Cytokine ELISA Quantitation of cytokines in cell culture supernatants or after in vitro stimulation was performed in a sandwich ELISA as described (Homann et al., 1998). In brief, 96-well plates were precoated overnight at 48C with 2 mg/ml purified capture antibodies. After washing and blocking with 10% FCS in PBS, serial dilution of supernatants and recombinant cytokine standards (PharMingen) overnight at 48C. Washing and a 1 hr incubation with 1 mg/ml matched biotinylated detection were followed by a 30 min incubation with a streptavidin-peroxidase conjugate (1:1000) (Boehringer Mannheim). For color reaction, H2O2 activated ABTS (2,2’-Azino-di-[3-ethylbenzthiazolin-6-sulfonic acid], Sigma) solution in 0.1 M citric acid (pH 4.35) was added. Plates were read at 405 nm with an EL-800 microplate reader (Biotek) using DeltaSoft 3 (BioMetallics) software. For TGFb quantitation, the TGFb1 Emax ImmunoAssay kit (Promega) was used. Flow Cytometry Staining of cell surface antigen and intracellular antigens was performed as described (Homann et al., 1998). For staining of intracellular cytokines, cells were stimulated for 5–10 hr by either of the following methods. Polyclonal stimulation was provided by 5 ng/ml PMA and 500 ng/ml ionomycin (Sigma) or, alternatively, plate-bound anti-CD3 and soluble anti-CD28 (Homann et al., 1998). Virus-specific stimulation was provided by addition of 1026 M LCMV-NP118-126 peptide in the presence of 50 U/ml recombinant human IL-2. All stimulation cultures contained 1 mg/ml Brefeldin A (BFA, #B7651, Sigma) to block protein transport into post-Golgi compartments and allow cytokines to accumulate within cells. In some experiments, antigen-specific stimulation was provided by LCMV-infected, irradiated PECs, and BFA was added for 5–10 hr after transfer to fresh PECs. Negative controls were stained with cytokine-specific PE-conjugated antibodies preincubated for 30 min at 48C with an excess of recombinant cytokine. Cells were acquired and analyzed on a FACSort or FACScalibur flow cytometer (Beckton Dickinson) using Cell Quest software (Beckton Dickinson). Immunohistochemistry Tissues taken for histologic analysis were placed in Bouin’s fixative and then stained with hematoxylin and eosin. Immunochemical studies were carried out on 6–10 mm cryomicrotome sections as described (von Herrath et al., 1994a). CFSE Labeling for Cell Tracking Regulatory cell cultures from spleens of protected or diabetic RIP-NP mice were labeled with carboxy fluorescein succiminidylester (CFSE, Molecular Probes) prior to i.v. transfer into prediabetic RIP-NP mice (5 days post LCMV). Use of CFSE has been described (Homann et al., 1998). Acknowledgments M. J. G. is a scholar of the Leukemia Society of America and supported by National Institutes of Health AI40171, a grant from the Juvenile Diabetes Foundation, and a gift from the Mather’s Foundation. M. G. V. H. is supported by National Institutes of Health AI44451 and DK51091 and a Juvenile Diabetes International Foundation Career Development Award, JDFI 296120. D. H. is supported by a Juvenile Diabetes International Foundation Fellowship Award and A. H. by a National Multiple Sclerosis Fellowship. This is The Scripps Research Institute manuscript number 12235-NP. We thank Diana Frye for assistance with the manuscript. Received April 22, 1999; revised August 22, 1999.

Antibodies The following fluorescein isothiocyanate (FITC), phycoerythrin (PE), CyChrome, allophycocyanin (APC), biotinylated, and/or purified antibodies (PharMingen) were used for flow cytometry, ELISA, and ELISPOT, as well as immunohistochemistry: CD4 (RM4-5), CD8 (53-6.7),

References Akhtar, I., Gold, J.P., Pan, L.Y., Ferrara, J.L., Yang, X.D., Kim, J.I., and Tan, K.N. (1995). CD41 beta islet cell-reactive T cell clones that

Bystander Regulatory T Cells Act via IL-4/Stat6 471

suppress autoimmune diabetes in nonobese diabetic mice. J. Exp. Med. 182, 87–97. al-Sabbagh, A., Miller, A., Santos, L.M., and Weiner, H.L. (1994). Antigen-driven tissue-specific suppression following oral tolerance: orally administered myelin basic protein suppresses proteolipid proteininduced experimental autoimmune encephalomyelitis in the SJL mouse. Eur. J. Immunol. 24, 2104–2109. Bergerot, I., Fabien, N., Maguer, V., and Thivolet, C. (1994). Oral administration of human insulin to NOD mice generates CD41 T cells that suppress adoptive transfer of diabetes. J. Autoimmun. 7, 655–663. Chen, Y., Kuchroo, V.K., Inobe, J.I., Hafler, D.A., and Weiner, H.L. (1994). Regulatory T cell clones induced by oral tolerance: suppression of autoimmune encephalomyelitis. Science 265, 1237–1240.

Kurts, C., Kosaka, H., Carbone, F.R., Miller, J.F.A.P., and Heath, W.R. (1997). Class I-restricted cross-presentation of exogenous self-antigens leads to deletion of autoreactive CD81 T cells. J. Exp. Med. 186, 239–245. Kurts, C., Miller, J.F.A.P., Subramaniam, R.M., Carbone, F.R., and Heath, W.R. (1998). Major histocompatibility complex class I-restricted cross-presentation is biased towards high dose antigens and those released during cellular destruction. J. Exp. Med. 188, 409–414. Kurts, C., Carbone, F.R., Krummel, M.F., Koch, K.M., Miller, J.F., and Heath, W.R. (1999). Signalling through CD30 protects against autoimmune diabetes mediated by CD8 T cells. Nature 398, 341–344.

Chen, Y., Inobe, J., and Weiner, H.L. (1995). Induction of oral tolerance to myelin basic protein in CD8-depleted mice: both CD41 and CD81 cells mediate active suppression. J. Immunol. 155, 910–916.

Lee, M.S., Wogensen, L., Shizuru, J., Oldstone, M.B.A., and Sarvetnick, N. (1994). Pancreatic islet production of murine interleukin-10 does not inhibit immune-mediated tissue destruction. J. Clin. Invest. 93, 1332–1338.

Chen, Y., Inobe, J., Kuchroo, V.K., Baron, J.L., Janeway, Jr., C.A., and Weiner, H.L. (1996). Oral tolerance in myelin basic protein T-cell receptor transgenic mice: suppression of autoimmune encephalomyelitis and dose-dependent induction of regulatory cells. Proc. Natl. Acad. Sci. 93, 388–391.

Lider, O., Santos, L.M., Lee, C.S., Higgins, P.J., and Weiner, H.L. (1989). Suppression of experimental autoimmune encephalomyelitis by oral administration of myelin basic protein. II. Suppression of disease and in vitro immune responses is mediated by antigen-specific CD81 T lymphocytes. J. Immunol. 142, 748–752.

Coon, B., An, L.L., Whitton, J.L., and von Herrath, M.G. (1999). DNA immunization to prevent autoimmune diabetes. J. Clin. Invest. 104, 189–194.

Maron, R., Palanivel, V., Weiner, H.L., and Harn, D.A. (1998). Oral administration of schistosome egg antigen and insulin B-chain generates and enhances Th2-type responses in NOD mice. Clin. Immunol. Immunopathol. 87, 85–92.

Daniel, D., and Wegmann, D.R. (1996). Protection of nonobese diabetic mice from diabetes by intranasal or subcutaneous administration of insulin peptide B-(9–23). Proc. Natl. Acad. Sci. USA 93, 956–960. Gallichan, W.S., Balasa, B., Davies, J.D., and Sarvetnick, N. (1999). Pancreatic IL-4 expression results in islet-reactive Th2 cells that inhibit diabetogenic lymphocytes in the nonobese diabetic mouse. J. Immunol. 163, 1696–1703. Garside, P., Steel, M., Liew, F.Y., and Mowat, A.M. (1995a). CD41 but not CD81 T cells are required for the induction of oral tolerance. Int. Immunol. 7, 501–504. Garside, P., Steel, M., Worthey, E.A., Satoskar, A., Alexander, J., Bluethmann, H., Liew, F.Y., and Mowat, A.M. (1995b). T helper 2 cells are subject to high dose oral tolerance and are not essential for its induction. J. Immunol. 154, 5649–5655. Harrison, L.C., Dempsey-Collier, M., Kramer, D.R., and Takahashi, K. (1996). Aerosol insulin induces regulatory CD8 gamma delta T cells that prevent murine insulin-dependent diabetes. J. Exp. Med. 184, 2167–2174. Herbelin, A., Gombert, J.-M., Lepault, F., Bach, J.-F., and Chatenoud, L. (1998). Mature mainstream TCRab1CD41 thymocytes expressing L-selectin mediate “active tolerance” in the nonobese diabetic mouse. J. Immunol. 161, 2620–2628. Homann, D., Tishon, A., Berger, D.P., Weigle, W.O., von Herrath, M.G., and Oldstone, M.B.A. (1998). Evidence for an underlying CD4 helper and CD8 T-cell defect in B-cell-deficient mice: failure to clear persistent virus infection after adoptive immunotherapy with virus-specific memory cells from mMT/mMT mice. J. Virol. 72, 9208–9216. Homann, D., Dyrberg, T., Petersen, J., Apostolopoulos, V., Oldstone, M.B.A., and von Herrath, M.G. (1999). Insulin in oral immune “tolerance”: a one amino acid change in the B-chain makes the difference. J. Immunol. 163, 1833–1838. Ihle, J.N., Thierfelder, W., Teglund, S., Stravapodis, D., Wang, D., Feng, J., and Parganas, E. (1998). Signaling by the cytokine receptor superfamily. Ann. NY Acad. Sci. 865, 1–9. Inobe, J., Slavin, A.J., Komagata, Y., Chen, Y., Liu, L., and Weiner, H.L. (1998). IL-4 is a differentiation factor for transforming growth factorbeta secreting Th3 cells and oral administration of IL-4 enhances oral tolerance in experimental allergic encephalomyelitis. Eur. J. Immunol. 9, 2780–2790. Kaplan, M.H., Schindler, U., Smiley, S.T., and Grusby, M.J. (1996). Stat6 is required for mediating responses to IL-4 and for development of Th2 cells. Immunity 4, 313–319.

McMenamin, C., Pimm, C., McKersey, M., and Holt, P.G. (1994). Regulation of IgE responses to inhaled antigen in mice by antigen-specific gamma delta T cells. Science 265, 1869–1871. McMenamin, C., McKersey, M., Kuhnlein, P., Hunig, T., and Holt, P.G. (1995). Gamma delta T cells down-regulate primary IgE responses in rats to inhaled soluble protein antigens. J. Immunol. 154, 4390–4394. Mengel, J., Cardillo, F., Aroeira, L.S., Williams, O., Russo, M., and Vaz, N.M. (1995). Anti-gamma delta T cell antibody blocks the induction and maintenance of oral tolerance to ovalbumin in mice. Immunol. Lett. 48, 97–102. Miller, A., Lider, O., and Weiner, H.L. (1991). Antigen-driven bystander suppression after oral administration of antigens. J. Exp. Med. 174, 791–798. Miller, A., al-Sabbagh, A., Santos, L.M., Das, M.P., and Weiner, H.L. (1993). Epitopes of myelin basic protein that trigger TGF-beta release after oral tolerization are distinct from encephalitogenic epitopes and mediate epitope-driven bystander suppression. J. Immunol. 151, 7307– 7315. Muir, A., Peck, A., Clare-Salzler, M., Song, Y.H., Cornelius, J., Luchetta, R., Krischer, J., and Maclaren, N. (1995). Insulin immunization of nonobese diabetic mice induces a protective insulitis characterized by diminished intraislet interferon-gamma transcription. J. Clin. Invest. 95, 628–634. Nussenblatt, R.B., Caspi, R.R., Mahdi, R., Chan, C.C., Roberge, F., Lider, O., and Weiner, H.L. (1990). Inhibition of S-antigen induced experimental autoimmune uveoretinitis by oral induction of tolerance with S-antigen. J. Immunol. 144, 1689–1695. Oldstone, M.B.A., Nerenberg, M., Southern, P., Price, J., and Lewicki, H. (1991). Virus infection triggers insulin-dependent diabetes mellitus in a transgenic model: role of anti-self (virus) immune response. Cell 65, 319–331. Polanski, M., Melican, N.S., Zhang, J., and Weiner, H.L. (1997). Oral administation of the immunodominant B-chain of insulin reduces diabetes in a co-transfer model of diabetes in the NOD mouse and is associated with a switch from Th1 to Th2 cytokines. J. Autoimmun. 10, 339–346. Sempe, P., Richard, M.F., Bach, J.F., and Boitard, C. (1994). Evidence of CD41 regulatory T cells in the non-obese diabetic male mouse. Diabetologia 37, 337–343.

Kapp, J.A., and Ke, Y. (1997). The role of gammadelta TCR-bearing T cells in oral tolerance. Res. Immunol. 148, 561–567.

Tada, Y., Ho, A., Koh, D.R., and Mak, T.W. (1996). Collagen-induced arthritis in CD4- or CD8-deficient mice: CD81 T cells play a role in initiation and regulate recovery phase of collagen-induced arthritis. J. Immunol. 156, 4520–4526.

Ke, Y., Pearce, K., Lake, J.P., Ziegler, H.K., and Kapp, J.A. (1997). Gamma delta T lymphocytes regulate the induction and maintenance of oral tolerance. J. Immunol. 158, 3610–3618.

Vajdy, M., Kosco-Vilbois, M.H., Kopf, M., Kohler, G., and Lycke, N. (1995). Impaired mucosal immune responses in interleukin 4-targeted mice. J. Exp. Med. 181, 41–53.

Immunity 472

Vella, A., Teague, T.K., Ihle, J., Kappler, J., and Marrack, P. (1997). Interleukin 4 (IL-4) or IL-7 prevents the death of resting T cells: stat6 is probably not required for the effect of IL-4. J. Exp. Med. 186, 325–330. von Herrath, M.G., and Homann, D. (1997). Treatment of virusinduced autoimmune diabetes by oral administration of insulin: study on the mechanism by which oral antigens can abrogate autoimmunity. Endocrinol. Diabetes 105, 24–25. von Herrath, M.G., and Oldstone, M.B.A. (1997). IFN-gamma is essential for b-cell destruction by CTL. J. Exp. Med. 185, 531–539. von Herrath, M.G., Dockter, J., and Oldstone, M.B.A. (1994a). How virus induces a rapid or slow onset insulin-dependent diabetes mellitus in a transgenic model. Immunity 1, 231–242. von Herrath, M.G., Dockter, J., Nerenberg, M., Gairin, J.E., and Oldstone, M.B.A. (1994b). Thymic selection and adaptability of cytotoxic T lymphocyte responses in transgenic mice expressing a viral protein in the thymus. J. Exp. Med. 180, 1901–1910. von Herrath, M.G., Dyrberg, T., and Oldstone, M.B.A. (1996). Oral insulin treatment suppresses virus-induced antigen-specific destruction of beta cells and prevents autoimmune diabetes in transgenic mice. J. Clin. Invest 98, 1324–1331. Weiner, H.L. (1997). Oral tolerance: immune mechanisms and treatment of autoimmune diseases. Immunol. Today 18, 335–343. Weiner, H.L., Friedman, A., Miller, A., Khoury, S.J., Al-Sabbagh, A., Santos, L., Sayegh, M., Nussenblatt, R.B., Trentham, D.E., and Hafler, D.A. (1994). Oral tolerance: immunologic mechanisms and treatment of animal and human organ-specific autoimmune diseases by oral administration of autoantigens. Annu. Rev. Immunol. 12, 809–837. Wildner, G., and Thurau, S.R. (1995). Orally induced bystander suppression in experimental autoimmune uveoretinitis occurs only in the periphery and not in the eye. Eur. J. Immunol. 25, 1292–1297. Yoshida, H., Hachimura, S., Hirahara, K., Hisatsune, T., Nishijima, K., Shiraishi, A., and Kaminogawa, S. (1998). Induction of oral tolerance in splenocyte-reconstituted SCID mice. Clin. Immunol. Immunopathol. 87, 282–291. Yoshino, S., Quattrocchi, E., and Weiner, H.L. (1995). Suppression of antigen-induced arthritis in Lewis rats by oral administration of type II collagen. Arthritis Rheum. 38, 1092–1096. Zhang, Z.Y., Lee, C.S., Lider, O., and Weiner, H.L. (1990). Suppression of adjuvant arthritis in Lewis rats by oral administration of type II collagen. J. Immunol. 145, 2489–2493.