Induction or Protection from Experimental Autoimmune Encephalomyelitis Depends on the Cytokine Secretion Profile of TCR Peptide-Specific Regulatory CD4 T Cells¹

The T cell response to self-Ags is associated with the pathogenesis of a number of autoimmune diseases in both animal models and humans. Whether an organ-specific autoimmune disease is induced depends upon a crucial balance between Ag-specific type 1 (Th1) and type 2 (Th2) CD4 T cells. In general, autoantigen-specific, pro-inflammatory type 1 T cells mediate organ-specific autoimmune disease, whereas type 2 responses are potentially protective (1, 2, 3). Knowledge of how a healthy balance among self-reactive type 1/type 2 T cells is maintained is important in understanding peripheral tolerance as well as in designing therapies for autoimmune diseases.

Experimental autoimmune encephalomyelitis (EAE)³ is a prototypic CD4 T cell-mediated autoimmune disease and is an instructive model for the human demyelinating disease, multiple sclerosis, because it shares many of its pathological and immune dysfunctions (4). It is characterized by inflammation and demyelination in the central nervous system accompanied by paralysis following immunization with myelin Ags, for example, myelin basic protein (MBP). A majority of the MBP-primed effector CD4 T cells that mediate EAE in H-2^u mice recognize the N-terminal peptide MBP Ac1-9/Ac1-20 and predominantly use the TCR Vβ8.2 gene segment (5, 6). The CD4 T cell lines and clones that adoptively transfer EAE invariably secrete proinflammatory cytokines, such as IFN-γ and/or lymphotoxin (7, 8, 9, 10). However, T cells with the same specificity producing Th2-like cytokines generally are nonencephalitogenic and in some cases prevent EAE in a bystander fashion (11, 12, 13). Similarly, in several experimental systems treatment of mice either with Abs to neutralize Th1 cytokines or with administration of Th2 cytokines have been shown to be protective from disease. Furthermore, endogenous levels of IL-4/IL-10 appear to correlate with the recovery from EAE (14). Although these observations collectively suggest that myelin Ag-specific Th1 cells are encephalitogenic and Th2 are not, recent experiments indicate that the regulation and function of individual cytokines is more complex; for example, T cells that express a transgenic TCR specific for MBP can induce EAE in immunodeficient recipients after culture under either Th1- or Th2-polarizing conditions (15). Furthermore, systemic treatment with cytokines can have an unexpected outcome as a result of complex effects on other lymphoid populations, including regulatory T cell populations (see Discussion).

Generally, Ag-induced EAE is transitory or monophasic in B10.PL mice, and spontaneous recovery from disease is associated with the physiological induction/expansion of regulatory CD4 T cells that recognize a framework 3 region determinant within TCR Vβ8.2 (peptide “B5”, amino acids 76–101/A^u, also referred to as the TCR Fr3 peptide) (16, 17, 18, 19). These regulatory CD4 T cells together with recruited CD8 T cells, specific for a different determinant within the TCR Vβ8.2 chain, appear to be involved in recovery from Ag-induced EAE. Thus, inactivation or absence of either the regulatory CD4 or CD8 T cells results in increased severity of disease and poor or delayed recovery (19, 20, 21). Furthermore, vaccination with disease-related T cells or their TCRs has been demonstrated to prevent or ameliorate experimental autoimmune diseases (17, 18, 22, 23, 24, 25, 26, 27). These observations have led to clinical trials of TCR-based vaccination in humans (28, 29, 30, 31). Although in different experimental systems, TCR peptide-reactive regulatory T cells have been shown to control disease, their mechanism of action or the role of cytokines in this regulation is largely unknown. In an attempt to directly investigate the mechanism of regulation, we asked whether a type 1 or type 2 cytokine profile within the well-defined TCR peptide-specific regulatory CD4 population is required for the regulation of EAE. Furthermore, does the action of regulatory CD4 T cells result in influencing the cytokine profile of the MBP-reactive T cell population and its capacity to mediate disease?

B10.PL and SJL/J mice were purchased from The Jackson Laboratory (Bar Harbor, ME). (SJL × B10.PL)F₁ mice were bred under specific pathogen-free conditions in our own colony. Female mice were used at 8–16 wk of age and were maintained on standard laboratory diet and water ad libitum.

Vβ8.2 TCR peptides used were the same as reported previously (17): B1, amino acids 1–30L (TCR Fr1 peptide); B4, amino acids 61–90; B5, amino acids 76–101 (TCR Fr3 peptide).

Lymph nodes or spleens of mice were removed 10 or 30 days after s.c. immunization with peptides derived from the MBP or the Vβ8.2 chain, and single cell suspensions were prepared. To monitor nasal priming of regulatory CD4 T cells, spleens were removed 5–7 days after nasal instillation with TCR peptides. Lymph node cells (4 × 10⁵ cells per well) and splenocytes (8 × 10⁵ cells per well) were cultured in 96-well microtiter plates in 200 μl of serum-free medium (HL-1, Ventrex, Portland, ME, or X-vivo 10, BioWhittaker, Walkersville, MD) supplemented with 2 mM glutamine; peptides were added at concentrations ranging from 0.1 to 7 μM final concentration. Proliferation was assayed by the addition of 1 μCi [³H]thymidine (International Chemical and Nuclear, Irvine, CA) for the last 18 h of a 5-day culture, and incorporation of label was measured by liquid scintillation counting.

Mice were immunized s.c. with 100 μg of guinea pig MBP or Ac1-9 emulsified in CFA; 0.15 μg pertussis toxin (List Biological, Campbell, CA) was injected in 200 μl saline i.v. 48 h later. Mice were observed daily for signs of EAE until 50–60 days after immunization. The average disease score for each group was calculated by averaging the maximum severity of all of the affected animals in the group. Disease severity was scored on a 5-point scale, as described earlier (17): 1, flaccid tail; 2, hind limb weakness; 3, hind limb paralysis; 4, whole body paralysis; 5, death.

Lymphokine secretion by established T cell clones was measured in the supernatants using a standard sandwich ELISA technique, as described earlier (32). For fresh cultures, mice were challenged with the TCR Fr3 peptide, B5, emulsified in CFA or IFA. Ten days later, the frequency of Ag-induced IFN-γ- or IL-4-producing T cells was determined using the sensitive, single-cell ELISA spot assay, as described earlier (32). Briefly, after culture of lymph node or splenic cells with Ag for 48 h, live cells were recovered, washed, and transferred by serial dilution (from 10⁴ to 5 × 10⁵ cells per well) to 96-well microtiter plates (Millipore, Bedford, MA) that had been precoated with the capturing mAbs (anti-IFN-γ or anti-IL-4) at 2 mg/ml. After 24 h, cells were removed, and spots were visualized using biotinylated detecting mAbs and avidin d-peroxidase in conjunction with 3-amino-9-ethylcarbazole (Sigma, St. Louis, MO) substrate. Spots were counted under a dissecting microscope, and the frequency of Ag-specific cells was determined from the difference between number of spots seen with and without Ag. All capturing and detecting Abs pairs were purchased from PharMingen (San Diego, CA).

Following anesthesia with halothane, mice were nasally instilled with peptides in PBS in a total volume of 20 μl. The effectiveness of NI-mediated deviation of TCR Fr3 peptide responses toward Th2 is dependent upon the dose and age of mice used (Melo et al., manuscript in preparation). NI with 5–10 μg of B5 in 11- to 15-wk-old mice resulted in almost complete deviation to Th2. Under these conditions, IL-4 secretion predominates with insignificant levels of IFN-γ. However, NI with a similar dose of B5 in 4-wk-old mice resulted in simultaneous secretion of significant levels of both IFN-γ and IL-4. Accordingly, in one experiment, 4-wk-old mice appeared to be protected from disease.

First, we tested the cytokine secretion profile of two TCR Fr3 peptide-specific regulatory CD4 T cell clones (B5.1 and B5.2) capable of inducing protection from MBP-induced EAE when adoptively transferred into B10.PL mice (17). Interestingly, both of these clones secrete large amounts of IFN-γ, as well as IL-2, IL-4, IL-5, and a borderline but detectable amount of IL-10 (for example, clone B5.2 secreted IFN-γ, 641-1383 pg/ml; IL-4, 74–125 pg/ml; IL-5, 52–63 pg/ml; IL-2, 14–25 U/ml; and IL-10, 35–47 U/ml). Fresh cultures of T cells isolated from mice challenged with TCR peptide B5 also secreted IFN-γ and IL-4. T cell clones as well as freshly cultured lines, specific for TCR Fr3 peptide, did not produce detectable amounts of TGF-β. Therefore, TCR peptide-specific regulatory CD4 clones do not resemble Th3 (11) or TR1 (33) cells, which exhibit heightened secretion of TGF-β or IL-10, respectively.

Anti-inflammatory type 2 cytokines, such as IL-4, IL-10, or TGF-β, have been shown to be involved in a bystander fashion in control of pathogenic autoreactive T cells (1, 34). To determine whether Th2 cytokine secretion by TCR peptide-specific CD4 T cells could control EAE, we used a NI technique known to deviate responses in a Th2 direction (35). As shown in Fig. 1, NI of B10.PL mice with an appropriate level (5–10 μg) of TCR Fr3 peptide, B5, results in priming of regulatory T cells that predominantly secrete IL-4 and very little IFN-γ. There was no response to a control TCR Vβ8.2 peptide (B1, amino acids 1–31, referred to as TCR Fr1 peptide). Five days following NI with a control Ag, such as HEL or another TCR Vβ8.2 peptide (B4, amino acids 61–90), splenic cells did not show any proliferative response or cytokine production in in vitro recall assays (data not shown). Collectively, these data suggest that TCR Fr3 peptide-reactive T cells can be readily activated upon nasal exposure. Having established the nasal instillation conditions for Th2 priming of regulatory CD4 T cells, groups of B10.PL or (SJL × B10.PL)F₁ mice (both of these strains behave almost identically in response to the TCR peptide B5/A^u, as well as in the anti-MBP encephalitogenic response in our experiments) were nasally instilled with the TCR Fr3 peptide, or as a control the TCR Fr1 peptide, PBS, or another immunogenic HEL peptide, 30-53, and were s.c. challenged 5–7 days later with MBP Ac1-9/CFA followed by i.v. pertussis toxin to induce EAE. A typical disease course in these mice is shown in Fig. 2, and a summary of five independent experiments is presented in Table I. It is clear that mice nasally instilled with the TCR Fr3 peptide contract a distinctly more chronic and severe disease (mean disease score, 4.6) than mice in control groups (mean disease score, 2.1). In the combined TCR Fr3 peptide groups, 20 of 29 mice did not show the typical monophasic disease course but became moribund or died from severe paralysis. This disease profile is similar to that induced in the absence of regulatory CD4 or CD8 T cells (19, 20, 21). In contrast, mice nasally instilled with PBS, the TCR Fr1 peptide, or HEL 30-53 before Ac1-9 challenge contracted milder, typical EAE with over 90% of the mice showing spontaneous recovery (Table I). Notably, in the group nasally instilled with the TCR Fr3 peptide B5, mice contract disease about 3 days earlier than in the control group (see Table I).

To determine whether NI per se was disruptive for regulation, Th1 priming via the mucosal route was attempted. The presence of IL-12 during an initial antigenic challenge has been shown to result in the development of Th1 responses and enhanced secretion of IFN-γ by differentiated Th1 cells (36). Therefore, in an attempt to reinstate Th1 induction of regulatory T cells, mice were nasally instilled with IL-12 along with the peptide. NI with the TCR Fr3 peptide (10 μg B5 per mouse) and IL-12 (0.2–0.3 μg per mouse) resulted in priming of B5-reactive T cells secreting significant amounts of IFN-γ. Groups of mice were nasally instilled with the TCR Fr3 peptide or the control TCR Fr1 peptide in combination with IL-12 or IL-4 and were subsequently challenged with Ac1-9/CFA/pertussis toxin to induce EAE. The data in Table II clearly demonstrate that the groups of mice nasally instilled with TCR Fr3 peptide and IL-12 were significantly protected from Ag-induced EAE. In experiment 1, only 2 of 8 mice contracted mild disease, while in experiment 2, the severity of disease was strongly down-modulated. In contrast, mice in other groups, including those receiving IL-12 with the irrelevant TCR Fr1 peptide, contracted typical EAE. These results suggest that Th1-like cytokine secretion by regulatory TCR peptide-specific CD4 T cells is required for efficient control of encephalitogenic MBP-reactive T cells.

To further examine the role of TCR peptide-specific type 1 CD4 T cells in the regulation of EAE, short-term CD4 T cell lines specific for the TCR Fr3 peptide, B5, were established from mice following NI with B5 peptide as well as IL-4 or IL-12. Splenic cells isolated from mice nasally instilled 5 days earlier with the TCR peptide and cytokines were cultured in vitro and tested for the secretion of IFN-γ or IL-4. Regulatory CD4 T cell lines derived from mice nasally instilled with B5 peptide plus IL-4 secreted more IL-4 than IFN-γ, whereas T cell lines from mice nasally instilled with B5 peptide plus IL-12 secreted much more IFN-γ than IL-4 following in vitro recall with the TCR peptide (Table III). These lines were then adoptively transferred into naive (SJL × B10.PL)F₁ animals to examine their ability to protect recipients from subsequent induction of EAE. As shown in Table III, only groups of mice receiving type 1 regulatory CD4 T cell lines were significantly protected from Ac1-9-induced EAE. In contrast, mice in groups immunized with PBS or Th2-like regulatory T cell populations contracted typical EAE.

Next, we determined how regulatory T cells influenced the MBP Ac1-9/Ac1-20-specific encephalitogenic T population. Our earlier experiments in B10.PL mice with adoptive transfer of cloned regulatory CD4 T cells, the administration of TCR Fr3 peptide, a recombinant single-chain TCR (scTCR) protein containing a Vβ8.2 domain, or plasmid DNA encoding the Vβ8.2 chain all indicate that the heterogeneous population of Vβ8.2 T cells are neither deleted nor inactivated (17, 18, 37, 38, and V. Kumar, unpublished observations). Recently, we and others have shown that in vivo priming/expansion of B5-reactive T cells using the Vβ8.2 scTCR protein or plasmid DNA-encoding TCR Vβ8.2 results in Ag-specific immune deviation and significant protection from EAE (17, 32, 37–39, and V. Kumar, unpublished observations).

To directly address the fate of the MBP Ac1-20-specific effector T cell population and the predominance of Th1- or Th2-like cells within it following the induction of TCR peptide-reactive regulatory (type 1) vs nonregulatory (type 2) CD4 T cells, groups of mice were nasally instilled with TCR Fr3 peptide alone or together with IL-12 or IL-4 and then s.c. challenged with MBP Ac1-20. Ten days later, the frequency of Ac1-20-specific and TCR peptide-reactive T cells, secreting either IFN-γ (type 1) or IL-4 (type 2), was determined using an ELISA-spot assay. There were no significant differences in the proliferative responses to Ac1-20 in the draining lymph node cells from mice in these groups, and the responses ranged from 34,571 ± 4,806 to 46,641 ± 1,150. However, as shown in Fig. 3, the frequency of TCR Fr3 peptide-specific T cells secreting IFN-γ was much higher and the frequency of cells secreting IL-4 was lower in mice nasally instilled with the TCR Fr3 peptide and IL-12. The opposite was true for mice nasally instilled either with the TCR Fr3 peptide alone or together with IL-4. Most importantly, the frequency of Ac1-20-specific T cells secreting IL-4 increased while those secreting IFN-γ decreased in mice nasally instilled with TCR Fr3 peptide plus IL-12, indicative of deviation toward a type 2 response among the Ac1-20-reactive effector T cells. In contrast, the frequency of Ac1-20-reactive T cells secreting IFN-γ was higher in mice nasally instilled with the TCR Fr3 peptide alone or along with IL-4, in accord with the observed exacerbation of pathogenicity.

A valuable additional index of Th1 or Th2 cytokine skewing is the shift in Ab isotype between IgG2a and IgG1 (40). To confirm the skewing of Ag-specific effector populations toward type 1 or type 2, we analyzed the isotypes of Abs directed against the Ag Ac1-20 under the conditions mentioned above. As shown in Fig. 3 C, mice nasally instilled with TCR Fr3 peptide plus IL-12 showed higher levels of anti-Ac1-20 Abs of the IgG1 isotype (Th2) and lower levels of Abs of the IgG2a isotype (Th1). In comparison, the levels of anti-Ac1-20 Abs of the IgG2a isotype were higher, and of the IgG1 isotype lower, in mice nasally instilled with TCR Fr3 peptide alone or together with IL-4.

The studies reported in this paper collectively demonstrate that under conditions of efficient regulation, when the cytokine secretion profile of TCR peptide-reactive regulatory CD4 T cells is of the type 1 phenotype, deviation of the MBP-specific effectors is induced in a type 2 direction. Such deviated effectors are unable to induce EAE under these experimental conditions. Thus, regulation induced by Fr3 TCR peptide-reactive T cells raised in a Th1 milieu dominantly affects the pathogenicity of the response. In contrast, during circumstances of dysregulation when regulatory T cells are forced to deviate to the Th2 phenotype, the frequency of thereby unregulated Ac1-20-reactive type 1 T cells increases, resulting in even more severe and chronic paralysis. This indicates that TCR peptide-specific regulatory T cells play a dominant and persistent role in the control of encephalitogenic T cells in vivo, while under these conditions other potential regulatory mechanisms are without apparent effect.

Earlier studies in two experimental autoimmune disease models, EAE and collagen-induced arthritis (16, 25, 41), as well as findings described here suggest that spontaneously primed TCR peptide-specific regulatory T cells are crucially involved in controlling autoreactive T cells, and that immune regulation is one of the important mechanisms for maintaining peripheral tolerance to self-Ags. Furthermore, a functional balance among Ag-specific T cells and regulatory T cells is crucial for the maintenance of peripheral tolerance (42). Thus, under conditions of dysregulation where regulatory CD4 T cells are forced to deviate in a Th2 direction and are unable to control MBP-reactive T cells, mice not only contract severe and chronic EAE but also develop disease earlier than in control mice. The development of early, more severe and chronic EAE, as well as the very poor recovery from chronic disease, indicates that the induction of regulatory Th2 cells represents a condition of physiological dysregulation. Therefore, it is likely that physiologically primed/activated TCR Fr3 peptide-reactive T cells are Th1-like, which mediate spontaneous recovery from a single episode of EAE. Consistent with this idea, in preliminary experiments TCR peptide B5-reactive T cells, in mice recovering from EAE, appear to be predominantly Th1-like (V. Kumar, unpublished observations). Also, B10.PL mice challenged with B5 in IFA or CFA secrete much more IFN-γ than IL-4 (data not shown). Furthermore, our findings that regulatory T cells are readily primed within 3–5 days either following NI or following s.c. immunization with TCR Fr3 peptide suggest that TCR-reactive T cells pre-exist in healthy animals. Consistent with this, we have earlier suggested (16, 17) that TCR determinants could be potentially displayed in the context of both class I and/or class II molecules by professional APCs that are able to pick up TCR from apoptotic T cells during normal turnover in the periphery.

In a number of systems, the mucosal route has been shown to be an important site for priming Ag-specific responses (43, 44). Our data further suggest that NI with appropriate cytokines can be used to deviate responses in a type 1 or type 2 direction. Thus, NI with rIL-12 concomitant with the Ag leads to priming of a predominantly Th1-like response, whereas NI in the presence of IL-4 primes a predominant Th2-like response. It is noteworthy that significant priming (both proliferative response and cytokine secretion) to TCR Fr3 peptide B5 occurs 5 days following a single NI with 10 μg of this peptide in the absence of any adjuvant. This clearly indicates the effectiveness of priming via the nasal route. Because under identical conditions a significant proliferative or cytokine response is not seen this early with other immunogenic TCR or HEL peptides, it appears that TCR peptide-reactive T cells represent an already activated or memory population, as suggested above.

Immune deviation of Ag-specific T cells at the population level may explain how TCR-based regulation directed to a single Vβ-chain is able to control disease-inducing, MBP Ac1-9-specific T cells that use other TCR V-chains, for example Vβ13 or Vβ4 (45). Accordingly, such modulation of T cell responsiveness to the target antigenic determinant may provide a suppressive environment for responses to other antigenic determinants from the same protein, as well as from other myelin components that may arise as a result of determinant spreading during chronic demyelination (46). Consistent with this, it has been shown that the bulk T cell infiltrate disappears from the target organ and EAE is reversed in PL/J mice following deviation of Ag-specific T cells by an altered peptide ligand (12). Furthermore, it is not yet clear in this system whether physiologically primed T cells reactive to subdominant determinants of MBP detected in the spleen following determinant spreading are pathogenic or protective (47). Our recent observations suggest that priming of at least some of the T cells reactive to a subdominant determinant region (amino acids 121–150) of MBP results in significant protection from Ac1-9-induced EAE (V. Kumar et al., unpublished observations).

It is not yet clear how the action of type 1 regulatory T cells results in the eventual dominance by the type 2-deviated population of CD4 effectors. Because TCR Fr3-reactive regulatory CD4 T cells cannot recognize MBP-reactive effectors directly, it is possible that secretion of type 1 cytokines indirectly influences encephalitogenicity of effector CD4 T cells: inflammatory cytokines secreted by regulatory CD4 T cells are required for efficient recruitment/activation of regulatory CD8 T cells, either directly by interaction with CD8 cytotoxic effectors or via up-regulation of costimulatory molecules on APCs to optimize induction of this CD8 population (16, 17, 18). Regulatory CD8 T cells may induce apoptosis or anergy of the initially, rapidly expanding, high-avidity, MBP-specific Vβ8.2 Th1 cell population (48, 49, 50). This would enable a relatively slower reacting compartment of low-avidity, MBP-specific type 2 cells (which may or may not express Vβ8.2) to expand in the absence of cross-regulatory IFN-γ secreting cells, resulting in immune deviation (37). We are currently investigating these possibilities. Furthermore, which specific type 1 cytokines are involved in regulation by TCR-peptide-specific CD4 T cells remains to be investigated.

These data are consistent with recent observations in several experimental autoimmune disease systems where removal of a Th1 cytokine such as IFN-γ seems to exacerbate disease, or treatment at a particular time with Th1-inducing cytokines like IL-12, was shown to protect rather than potentiate autoimmune disease (51, 52, 53, 54, 55). Accordingly, systemic treatment of autoimmune disease with Th1- or Th2-promoting cytokines can have unexpected, deleterious effects as has been noted in some cases (51, 52, 53, 54, 55, 56). It is evident that the existence of a distinct regulatory CD4 population in the correct functional (proinflammatory) state plays a crucial and dominant role in the eventual deviation of the potentially pathogenic autoreactive T cell population in a type 2 direction, thus providing protection from autoimmune disease. Consistent with this view, a number of TCR-peptide-reactive T cells in humans have also recently been shown to produce IFN-γ (31).

Although it may seem counterintuitive, type 1 regulatory CD4 T cells apparently are a prerequisite for the efficient control of disease-causing myelin-Ag-specific T cells. In the inflammatory milieu arising from the activation of Vβ8.2 Ac1-9-reactive Th1 cells, the situation is established for the eventual down-regulation and deviation of this effector population. We have demonstrated that the TCR Fr3 peptide-reactive T cells are predominantly of the type 1 phenotype following priming of B10.PL mice with recombinant single-chain Vβ8.2 TCR or plasmid DNA-encoding Vβ8.2, which prevent EAE (37, 38, and V. Kumar, unpublished observations). A full understanding of the specific cytokine requirements of TCR peptide-specific regulatory CD4 T cells and their mechanism of action in regulation is crucial in exploiting therapeutic intervention of autoimmune disease in humans.

We thank Dr. Marco Melo for help with initial NI experiments and Andrew Koh for technical help. We thank Dr. J. Thorbecke (New York University, New York, NY) for the analysis of TGF-β. We also thank Drs. Douglas Green, Howard Grey, Mitchell Kronenberg, Alexander Miller, and Sally Ward for critical reading of the manuscript.