Development of CD4+ T cell lines that suppress an antigen-specific immune response in vivo

Authors

  • L. VIEIRA DE MORAES,

    1. Department of Immunology, Instituto de Ciências Biomédicas, University of São Paulo,
    Search for more papers by this author
  • B. SUN,

    1. Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences, Shanghai, China
    Search for more papers by this author
  • L. V. RIZZO

    Corresponding author
    1. Department of Immunology, Instituto de Ciências Biomédicas, University of São Paulo,
    2. Laboratory of Medical Investigation-60, Division of Allergy and Clinical Immunology, University of São Paulo Medical School and
    3. Fundação Zerbini, São Paulo, Brazil and
    Search for more papers by this author

Prof. Luiz Vicente Rizzo, Clinical Immunology Laboratory, Department of Immunology, Instituto de Ciências Biomédicas, University of São Paulo, Avenue Prof Lineu Prestes 1730, São Paulo, SP, Brazil – CEP 005508–900.
 E-mail: lvrizzo@icb.usp.br

SUMMARY

It has been suggested for many years that the regulation of the immune system for the maintenance of peripheral tolerance may involve regulatory/supressor T cells. In the past few years, several investigators have demonstrated that these cells can be generated in vitro. It has also been shown that they can inhibit the progression of various autoimmune disease models when infused into susceptible mice. We have generated two murine T cell lines in the presence of KLH-specific T cell clones from BALB/c or DBA2 mice. The lines are characterized by a low proliferative response to mitogens, the capacity to secrete high amounts of IL-10 and TGF-β, and small amounts of IFN-γ. Interestingly, these cells are unable to produce IL-2, IL-4 or IL-5. The study of the surface phenotype of both lines revealed CD4+, CD25high, CD44low and CTLA-4 cells. When injected intravenously in (CBy.D2) F1 mice, these cells were able to inhibit 50–100% of the TNP-specific antibody production, when the hapten was coupled to KLH. In the present study we offer another evidence for the existence of regulatory T cells in the T lymphocyte repertoire, suggesting that they can also regulate immune responses to foreign antigens. Furthermore, we demonstrate an alternative pathway to generate these cells different from approaches used thus far.

INTRODUCTION

Self-regulation of the immune system involves many mechanisms not yet well understood. Repeated exposure to antigens (non-pathogenic or pathogenic) can generate chronic cell activation, inflammation and or exacerbated antibody production, all of which require down regulatory mechanisms to conclude responses to maintain the homeostasis when it becomes convenient.

The basic ideas underlying the concept of immunological tolerance have concentrated on processes of clonal elimination or functional inactivation (anergy) of autoreactive T cells. However, in the last decade evidence from various autoimmune disease models have pointed to the involvement of a CD4+ T cell population with immunoregulatory properties [1–4]. The concept of suppressor cells, although old was almost banned from immunology after the mid 1980s because of the lack of evidence to support the intricate network of soluble factors, some antigen-specific, some non-antigen specific initially described as taking part in the process. Furthermore, the suggestion that many of the T cell suppressor clones available at the time did not express a functional T cell receptor and the improbability that the I-J region could contain a full MHC gene to restrict such type of response [5–8] doomed the concept for many years. The advent of the Th1/Th2 paradigm [9] further faded the idea of suppressor cells, since the need for regulation of peripheral responses should be resolved by the cross regulation between Th1 and Th2 cells [10,11]. However, evidence has been mounting recently that suggest that suppressor cells may indeed exist, although they have been renamed regulatory cells [12–15]. It has been shown that the elimination of CD25+ peripheral T cells in normal naive mice, a small population of CD4+ or even 1% of CD8+ T cells, results in various autoimmune diseases which may be prevented with the reconstitution of the depleted population [14,16–18]. Further evidence to the autoimmune-preventive properties of these populations are the studies of Itoh and colleagues [19,20] showing that these cells are generated in the thymus and that they have an important role in maintaining immunological self tolerance.

Studies on transplantation tolerance have documented the role of regulatory T cells in active suppression of effector T cells from rejecting grafts and inducing a tolerant state [21,22]. Other investigators have shown the in vitro generation of regulatory T cells capable of preventing T cell mediated inflammatory disease in vivo[23] and T cell responses in vitro[24]. We have first characterized CD4+ regulatory T cells in 1992 (premlimincery data presented at the 8th International Congress of Immunology) suggesting that they could play an important role in the regulation of conventional immune responses.

Here we describe the isolation of two T cell lines that apparently recognize other T cells specifically reactive against keyhole limpet haemocyanin (KLH). These two cell lines were shown to have inhibitory properties in vivo, by suppressing the response against the hapten TNP when KLH was used as a carrier protein. Moreover, we have shown that these anti-T cell lines have an important and specific suppressive activity and may constitute one of the physiological mechanisms by which immune responses are regulated.

MATERIALS AND METHODS

Animals

Six to 12 weeks old BALB/cByJ, DBA/2 J and (CBy.D2) F1 male mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA) or breed in our own animal facilities. Animals were kept in microisolator cages under specific-pathogen free conditions and were handled following the guidelines for animal use approved by NIH Animal Care and Users Committee, and the animals in research welfare committee in each of the participating institutions.

Antigens and mitogens

Keyhole limpet haemocyanin (KLH) was purchased from Calbiochem-Boehring Corp. (La Jolla, CA, USA). TNP-KLH was prepared as described [25]. Briefly, KLH was dissolved in borate-buffered saline (BBS), pH 8·0, at 20 mg/ml and reacted with trinitrobenzene sulphonic acid (TNBS) (ICN, Cleveland, OH, USA). Coupling of TNP to conalbumin and bovine serum albumin was similar. Bovine serum albumin (BSA), α-metil-manoside, concanavalin A (Con A), rat γ-globulin (RtG) and conalbumin were purchased from Sigma Chemicals Co. (St. Louis, MO, USA); phytohemaglutinin A (PHA) was from DIFCO; horseradish peroxidase (HPRO)-streptavidin from Southern Biotechnologies Associates (Birminghan, AL, USA).

Antibodies

The anti-CD4 GK1·5, the anti-murine IFN-γ R4–6A2 (HB 170) and XMG1·2 (12) and the anti-Thy 1·2 H0-22–1 (TIB 99) were obtained from the American Type Culture Collection and were purified by ammonium sulphate precipitation of ascites fluid. TRFK-4 and TRFK-5 (14) hybridomas were a generous gift of Dr Paulo Vieira (DNAX Research Institute). The anti-IL-4 antibody 11B11 was generously provided by Dr William Paul (NIH). The S4B6 (16) and anti-IL-2 antibody was a gift from Tim Mosmann, University of Alberta. The anti-CD-8 was obtained from Becton and Dickinson (Mountain View, CA, USA).

Recombinant cytokines

Recombinant IL-1 was from Cistron (Pine Brook, NJ, USA), recombinant IL-2 was obtained from Amgem, Inc. (Thousand Oaks, CA, USA), recombinant murine IL-4, IL-5 and IFN-γ were purchased from Genzyme Corporation (Boston, MA, USA), recombinant murine IL-10 for the ELISA was a gift from Dr Robert Coffman (DNAX research Institute, Palo Alto, CA, USA) and TGF-β was purchased from R & D Systems (Minneapolis, MN, USA).

Cell culture media

T cells were cultured in DMEM supplemented with 10% FCS (Hyclone, Logan, UT), 10−5M 2-ME (Sigma), 2 mM l-glutamine, 0,1 mM nonessential amino-acids and vitamins (Gibco, Grand Island, NY, USA). To maintain the T cell clones and lines, 2% of conditioned media containing IL-2 was added. The last step was omitted when lymphokine synthesis was to be assayed.

T cell clones

All T cell clones used were derived from BALB/cBy, DBA/2 or CBy.D2 F1 mice and are H-2d restricted. Clones D3, R3, DC10, DC12, D6, DD6, E6, B5, BE5 and DE5 respond to KLH. Clones F1, F2, F3, E4 and LE6 respond to rat γ-globulin (RtG). Clone B1 is autoreactive. The clones were generated by limited dilution and maintained by repeated stimulation with antigen and irradiated syngeneic APC in the presence of conditioned medium containing IL-2, as previously described [26]. The derivation of clones D3, DC10, DC12, D6, E6 and DD6 [10,25,27] and clone B1 [28] was described. The other clones were derived similarly. Table 1 shows the cytokine profile of each clone after antigen stimulation.

Table 1.  Cytokine profile of the T cell clones
CloneIL-2IL-4IL-5IL-10IFN-γ
  • *

    2·5 × 106 cells from either lineage were stimulated with syngeneic APC and 2·5 µg/ml of ConA. Supernatants were harvested 24 and 48 h later and their cytokines were evaluated as described in Materials and methods. The higher level of each measure is presented.

  • Maximal amount of cytokine measured in cell culture media, presented in ng/ml.

D3*2·3<0·02<0·04<0·0381·335
B50·9<0·02<0·04<0·0381·83
E6*1·2<0·02<0·04<0·0380·74
DE5*1·7<0·02<0·04<0·0380·56
DC10<0·023·90·70·25<0·015
DC12<0·023·87<0·040·43<0·015
D6<0·023·00·850·56<0·015
DD6<0·021·70·0090·18<0·015
B11·40·025<0·04<0·041·126
R32·1<0·02<0·04<0·0380·46
E41·8<0·02<0·04<0·041·02
LE61·3<0·02<0·04<0·040·61
F1<0·022·10·1370·49<0·015
F2<0·024·90·36<0·038<0·015
F3<0·020·56<0·040·61<0·015
BE5<0·020·920·90·14<0·015

Regulatory T cell lines

(CBy.D2)F1 mice were immunized with 50 µg of KLH emulsified in CFA (1:1, v/v) by intraperitoneal and subcutaneous injection. Fifteen to 21 days later, the animals were sacrificed and their spleens and lymph nodes were extracted and teased into a single cell suspension. The cell suspension was then cultured in the presence of a mixture of the T cell clones cited above irradiated (3000 rads), at a ratio of 1 responder to 1 stimulator cell. Two days prior to their use the T cell clones were stimulated with KLH or RtG and APC, as described above. The cells were maintained in 24 well plates (Linbro) at 5 × 106 cell/1·5 ml of media and restimulated with the same mixture of irradiated T cell clones and irradiated syngeneic spleen cells every other week for a period of 6 weeks before they were first assayed. Cells were maintained in conditioned media containing IL-2.

Lymphokines assay

IL-2 and IL-4.  To measure IL-2 and IL-4 synthesys, supernatants from 24 or 48 h culture of the respective T cell clones and lines, stimulated as described for each experiment were collected. Cytokine content was measured using ELISA kit from R & D systems.

Interferon-γ, IL-5 and IL-10.  To examine IFN-γ production by the T cell lineage we used an ELISA assay that has been described [27]. Briefly, 96 well plates were coated with R4–6A2 (anti-IFN-γ monoclonal antibody (MoAb)); wells were blocked with a buffer containing 1% BSA for two hours. The samples were obtained in the same fashion as described for HT-2 assay and plated in serial dilutions at cold temperature. One day after the biotinilated XMG1·2 (MoAb anti-IFN-γ) was added and the plates were incubated for two hours, on ice under agitation. Wells were developed using streptavidin HRPO-conjugated for 1 h under agitation on ice. Samples concentrations were calculated against a standard curve for IFN-γ, using the recombinant. The ELISA for IL-5 was performed in a similar manner using the anti-IL-5 TRFK-5 MoAb for coating and the biotinilated anti-IL-5 MoAb TRFK4 and HRPO-streptavidin. The assay for murine IL-10 was kindly performed by Dr Rodrigo Corrêa de Oliveira and was described [25].

TGF-β assay.  In order to determine the production of TGF-β by the T cell lines, we used a bioassay based on the ability of TGF-β to inhibit IL-1 driven thymocyte proliferation as described elsewhere [29]. Briefly, after a 36-h stimulation period with ConA, the T cell lines culture supernatants were harvested and treated with α-metyl-manopyranoside. To activate the TGF-β, supernatants were heated to 80°C for 5 min Thymocytes were obtained from BALB/c mice and plated in flat bottom 96 well plates (Falcon) at the final concentration of 1·5 × 106 cell/well in the presence of 1 µg/ml of PHA and 5 U/ml of rIL-1 (Cistron). Samples were tested in serial two-fold dilutions. A standard curve for the effect of TGF-β on the IL-1-induced thymocyte proliferation was obtained by incubating the same amounts of thymocytes and PHA in the presence of purified TGF-β (R & D systems) from 0·4 ng/ml decreasing on two-fold dilution. Alternatively, TGF-β presence was measure by ELISA using a kit from Promega (Madison, WI, USA).

Immunization protocol and T cell lines infusion

Six to 12 weeks (CBy.D2) F1 male mice were immunized with 100 µg of KLH in PBS divided in i.p. and s.c. in the four footpads. One week later 5 × 106 cells from the T cell lines were injected i.v. in the tail vein. Following the injection, animals were immunized with 100 µg/mouse of TNP-KLH or TNP-Conalbumin in CFA intraperitoneally. Their blood was collected every other week through the retro orbital plexus and assayed for anti-TNP specific antibodies using the ELISA technique. Control animals received the same number of syngeneic naive spleen lymphocytes.

Assay for antibody production

TNP-specific antibodies was measured using a modified antigen-specific ELISA [25]. Briefly, 96 well microtiter plates (Costar) were coated with TNP-BSA (2 µg/ml). After blocking the plates with BSA and an overnight incubation with serum samples, the assay was developed using horseradish peroxidase-conjugated goat anti-IgG or anti-IgM antibodies was purchased from Southern Biotechnology Associates. In these assays the concentration of anti-TNP antibodies was estimated using standard curves constructed by coating wells with anti-Ig antibodies against the appropriate isotype and adding polyclonal Ig standards of the pertinent isotype.

Flow cytometry analysis

Cells were washed 3 times in PBS 2·5% FCS, ressuspended in cold RPMI 1640 containing 2·5% FCS and 0·1% sodium azide. Samples were incubated with anti-CD4 (GK 1·5), anti-CD8, anti-CD25, anti-CD44, anti-CTLA-4 or anti-CD45RA/RO monoclonal antibodies (all from Pharmingen, San Diego, CA, USA) for 45 min on ice at the concentration of 1 µg/106 cells. Cells were washed again and incubated with FITC-conjugated goat anti-mouse antibody (Tago Immunologicals, Burlingame, CA, USA) or PE-conjugated goat anti-rat (Southern Biotechnologies Associates).

Statistical analysis

Statistical analysis was performed using the unpaired Mann–Whitney test a P <  0·05 was considered significant. The software StatView 4·5 for Macintosh computers was used as to perform the statistical analysis.

RESULTS

Cell surface markers

The characterization of the two cell lines regarding their surface markers revealed that both cell lines are CD3+ CD4+ CD25high (data not shown). These cells expressed low levels of CD44 and CD45RA, high levels of CD45RO and were CTLA-4 negative (data not shown). It is important to note that the expression of CD3, CD4 and CD25 was measured after a rest period of 7 days and that activation did not seem to change the expression of the other markers either.

Cytokine profile of the two T cell lines generated from KLH-immunized (CBy.D2) F1 mice

T cell lines were generated by culturing spleen and lymph node cells from KLH-immunized (CBy.D2) F1 mice with irradiated KLH-specific histocompatible T cell clones. The line denominated T Supressor 1 (TS1) was generated against a mixture of equal proportions of clones D3, DE5, B5 and E6 (all Th1-type clones); in a similar fashion, the cell line called TS2 was derived against a mixture of clones DC10, DD6, D6 and DC12 (all Th2-type clones). Table 2 shows the cytokines secreted by these T cell lines in the presence of either 2·5 µg/ml of Con A or the T cell clones used to raise them. Both cell lines secreted very high amounts of IL-10 and TGF-β in the presence of the T cell clones. IFN-γ was secreted at a considerable level by the TS1 cell line. Interestingly, these T cell lines were unable to secrete detectable levels of the known T cell growth factors IL-2 or IL-4 and also failed to secrete IL-5. Cytokine production against individual clones was measured and as often close to the limit of the detection of the assay. For instance, the line TS1 in response to D3 produced 0.089 ng/ml of IL-10, 0·076 ng/ml of TGF-β and 0.067 ng/ml of IFN-γ. Line TS2, in response to DC10 secreted 0·054 ng/ml of IL-10, 0·041 ng/ml of TGF-β and 0·039 of IFN-γ.

Table 2.  Cytokine profile of the supressor T cell lines
 Line TS1Line TS2
ConA*T cell clonesConA*T cell clones
  1. *2·5 × 106 cells from either lineage were stimulated with syngeneic APC and 2·5 µg/ml of ConA. Supernatants were harvested 24 and 48 h later and their cytokines were evaluated as described in Materials and methods. The higher level of each measure is presented. †A mixture of equal proportions of T cell KLH-specific irradiated clones (2·5 × 106 cells) was irradiated (3300 rads) and added to the same amount of cells from either lineage in 24 well plate. Two days prior to the irradiation, T cell clones were stimulated with the pertinent antigen plus irradiated (2500 rads) syngeneic APC. Supernatants were collected and assayed for the shown cytokines as described. ‡Maximal cytokine measured in cell culture media, presented in ng/ml. ND, Not done.

IL-2<0·02<0·02<0·02<0·02
IL-4<0·02<0·02<0·02<0·02
IL-5<0·1<0·1<0·1<0·1
IL-10ND3·87ND 2·58
TGF-β2·44·2 4·1 1·1
IFN-γND3ND 0·42

To evaluate the specificity of the lines, cells were cultured in the presence of the T cell clones used to raise them or in the presence of non-KLH specific cells. Although cell proliferation could be detected, it was minimal (data not shown). As an alternative, specificity was measured by the ability of the T cell lines to secrete cytokines in the presence of the various T cell clones. The results shown on Table 3 suggest that TS1 cells produce high levels of IL-10 and small amounts of IFN-γ in the presence of the clones used to derive it (D3, DE5, B5 and E6) but fail to secrete the same cytokines in the presence of eight non-KLH specific histocompatible T cell clones. Similar results were obtained with line TS2. In the presence of the clones which were used to generate this T cell line (DC10, DD6, D6 and DC12) detectable levels of IFN-γ and important amounts of IL-10 were secreted whereas when these cells were cultured with non-KLH specific clones cytokines were not produced at detectable levels. Both lines secreted TGF-β in response to the KLH-specific T cell clones.

Table 3.  Cytokine production pattern of the supressor T cell lines in response to antigen and KLH-specific and non-specific T cell clones
ClonesLine TS1Line TS2
IFN-γIL-10TGF-βIFN-γIL-10TGF-β
  • *

    Stimulation with clones was performed as described in footnote

  • on Table 1. Cytokine levels were evaluated 24 or 48 h later. The higher value of each measure is presented. BDL, Below detection limit.

D3, DE5, B5, E6 *100 pg/ml1080 pg/ml1230 pg/ml100 pg/ml250 pg/ml1050 pg/ml
DC10, DD6, D6, DC12BDLBDL823 pg/ml6900 pg/ml927 pg/ml1723 pg/ml
B1, R3, E4, LE6BDLBDLBDLBDLBDLBDL
F1, F2, F3, BE5BDLBDLBDLBDLBDLBDL

Both lines also failed to proliferate or produce cytokines in the presence of KLH and syngeneic APC. These results suggest that the lines respond specifically to the T cell clones against which they were raised.

After irradiation the T cell clones used to stimulate the T cell lines produced significant amounts of cytokines for up to 36 h. However, the only cytokines that could be measured at the time they were used to stimulate the regulatory cell lines were IL-2 (0·3 ng/ml for the Th1 clones) and IL-4 (0·54 ng/ml). Since these cytokines were not detected in the assay after coculture with the regulatory clones we assume their production had ceased, was suppressed or the cytokines produced were consumed by the regulatory cell lines.

Evaluation of TS1 and TS2 cell lines in vivo

Since the T cell lines were shown to secrete a combination of regulatory cytokines in vitro, we decided to investigate the in vivo effects of these cells. To this end (CBy.D2) F1 mice were immunized with KLH in PBS to elicit KLH-specific T cells. Seven days later the animals were immunized with TNP-KLH with or without the injection of cells from either one of the lines (105, 106 or 5 × 106 per mouse). Data presented on Figs 1 and 2 show the mean of anti-TNP IgG levels in the sera of the experimental animals 14–21 days after immunization with TNP-KLH. According to the results, mice that received TS1 or TS2 cells produce considerably less anti-TNP antibody than the control animals. Furthermore, data presented on Fig. 1 suggests that 5 × 106 cells seems to be the minimal number of cells necessary to induce significant and constant inhibition of antibody production by both cell lines.

Figure 1.

Titration of the number of TS1 and TS2 cells necessary to inhibit antibody synthesis in vivo. (CBy.D2) F1 mice were immunized with TNP-KLH with or without injection of 105, 106 or 5 × 106 cells from either one of the lines, as a control some mice received 5 × 106 ConA-activated naïve lymph node cells. Data represent the mean of anti-TNP IgG levels in the sera of the experimental animals 14 days after challenge with TNP-KLH. Values are mean ± SEM. †P= 0·05; *P < 0·01 versus animals that received no cells.

Figure 2.

TS1 and TS2 cell lines inhibit antibody synthesis in vivo. (CBy.D2) F1 mice were immunized with TNP-KLH with or without injection of 5 × 106 cells from either one of the lines. Data represent the mean of anti-TNP IgG levels in the sera of the experimental animals 14 days after immunization with TNP-KLH. Values are mean ± SEM. *P < 0·01 versus controls.

The suppressive effect shown by these cells in vivo could have been deduced from the cytokine profile they secrete. However, its magnitude was far beyond the expected suggesting that these cells were actively interacting inside the animals and may have even proliferated.

The inhibitory effect of TS1 and TS2 in vivo is directed against KLH-specific cells

We have shown that in vitro both T cell lines only secreted TGF-β, IFN-γ and IL-10 in the presence of the clones used to derive them. To determine whether such specificity was maintained in vivo, groups of 4 animals were immunized with either TNP-KLH or TNP-Conalbumin, as described in the Materials and methods section. Animals were further divided in three groups. Control animals received 5 × 106 naive splenocytes and the other groups received five million cells from either the TS1 or the TS2 line. The anti-TNP response was measured 14 days after immunization and administration of T cell lines. Reinforcing the results presented above, there is a significant decrease in the amount of anti-TNP IgG secreted by mice that received either one of the cell lines as compared to the control animals (Fig. 3). However, it is clear that the cell lines were much more effective in reducing the anti-TNP response in those animals immunized with TNP-KLH (P < 0·01). Although there was some degree of suppression when the cell lines were infused into mice immunized with TNP-Conalbumin, it decrease was not statistically significant (P < 0·06). These results suggest that the inhibition observed in these experiments is specific for the carrier protein KLH. Moreover, it suggests that the suppression is directed against KLH-responsive cells and capable to inhibit their helper activity.

Figure 3.

The inhibitory effect of both TS1 and TS2 cell lines in vivo is directed against KLH-specific cells. Animals were immunized with TNP-KLH or TNP-Conalbumin and received or not injection of 5 × 106 cells from TS1 or TS2 cell lines. The anti-TNP IgG response 14 days after immunization is shown. Values are mean ± SEM. *P < 0·01 versus controls, †P < 0·06.

DISCUSSION

In this study we describe the generation and characterization of two CD4+ T cell lines that display suppressor activity over antibody production in vivo. These cell lines (TS1 and TS2) can specifically suppress an anti-TNP antibody response when TNP is coupled to KLH, probably through its ability to suppress the helper activity of KLH-specific T cells (Figs 1 and 2). We have also demonstrated that the ability to regulate the immune response is specific since these cells are capable to inhibit the synthesis of anti-TNP antibodies much more effectively in mice immunized with TNP-KLH than in mice immunized with TNP-Conalbumin (Fig. 3).

The importance of cytokines as regulators of the immune response has been extensively discussed, with special regards to their function in autoimmunity and allergy [30–32]. Their role and the role of the cells that produce them have not been nearly as dissected for responses against ‘conventional’ antigens such as KLH. Here we describe T cell lines with a similar cytokine production pattern that are capable of exerting a suppressive function in vivo. Both T cell lines secrete large amounts of IL-10 and TGF-β, small amounts of IFN-γ and undetectable levels of IL-2, IL-4 and IL-5 (Table 2). The deficient secretion of growth factors may explain the difficulties in isolating and maintaining these cells in culture, and consequently cloning them. It is important to note that although neither TS1 nor TS2 secrete IL-2 or IL-4, they respond to both cytokines with significant proliferation and increase in cell viability (data not shown). This ability to respond to T cell growth factor but not to produce them is of particular interest, in spite of the simplicity of the data. One can suggest that it is part of their self-regulatory process. If indeed the function of these cells is merely to suppress an immune response, it would keep this activity restricted to an ongoing immune response since the cells do not secrete their own growth factors they would have to use those produced by the cells proliferating in response to stimuli. Once the responder cells ceased to produce growth factors the suppressor cells would rapidly stop their own proliferation.

Although production of IFN-γ was observed in all experiments, the line TS2 secreted higher amounts of this cytokine on the experiments depicted on Table 3 than those shown on Table 2. This discrepancy in the levels of IFN-γ produced was also observed for the line TS1, which secreted more IFN-γ in the experiment depicted on Table 2 than on that shown on Table 3. The differences in the amount of interferon secreted cannot be ascribed to the activation status of the cells since in all experiments the surface markers for activation were evaluated by flow cytometry and no differences were detected. It is important to stress that whatever the cause for the variation in the cytokine production, IFN-γ was always detected in the supernatants suggesting it is part of the portfolio of cytokines secreted by these regulatory cell lines.

The high levels of IL-10 and TGF-β secreted by TS1 and TS2 cells strongly suggest an inhibitory function. The immunoregulatory properties of these cytokines have been well documented [33–35]. Studies have shown that T cell clones derived in vitro predominantly producing TGF-β and IL-10 can prevent T cell mediated inflammatory bowl disease, experimental autoimmune encephalomyelitis (EAE) and diabetes [1,13,23,36–38] when injected into susceptible recipients. Studies on oral tolerance have been particularly bountiful in revealing the immunoregulatory properties of these cytokines. These studies have suggested that either alone or in combination, IL-10 and TGF-β play an essential role in the development of the tolerant state [37,39]. Other reports have also shown that mRNA for IL-4, IL-10 and TGF-β is expressed in Peyer's patches and in the lamina propria of the gut as a response to the fed antigen [23]. Another evidence for the participation of IL-10 in oral tolerance is the observation that IL-10 receptors are expressed in the epithelial cells from murine small and large intestine [40]. The involvement of TGF-β and IL-10 in the maintenance of tolerance has also been studied using a different approach. TGF-β deficient mice or TGF-receptor inactivated mice have also been observed [41] and IL-10 knockout mice kept at non-sterile conditions develop inflammatory bowel disease spontaneously [42]. Numerous data have shown the immunosuppressive effects of IL-10 in mouse and human in vitro assays, by preventing antigen-specific T cell proliferation as a consequence of down-regulation of class II MHC [43–45] expression and costimulatory molecules such as B7·1, B7·2 and ICAM-1 [46,47]. In addition, IL-10 inhibits cytokine synthesis by monocytes and activated macrophages [41,43,48]. Because both T cell lines were very efficient in inhibiting antigen-specific antibody synthesis in vivo(Figs 1–3), it is possible that these cells either act through an amplification mechanism such as inhibiting antigen-presentation or they proliferate. The later hypothesis is the least probable based on the proliferation profile exhibited by these cells in vitro, where although optimum conditions were provided an increase in cell numbers were hardly ever achieved and never surpassed 30% for each cell cycle (data not shown). Which also explain the difficulties in cloning these cells.

Suppressor/regulatory T cells have also been shown to perform their activity through the CTLA-4 molecule [49–51]. Recent studies have suggested that regulatory CD4+ T cells activated through CTLA-4, a constitutively expressed molecule on these lymphocytes, might suppress other T cells by secreting TGF-β[37,52]. It has also been shown that IL-4, IL-10 and TGF-β can induce the differentiation of T cells that produce these same cytokines [23,53] suggesting that they influence the developmental milieu of the regulatory T cells. Since we have not detected CTLA-4 in the surface of either T cell lines, we suggest that this molecule may not be essential for the expression of regulatory activity in the population of cells we described here. This data supports previous reports that suggest that some regulatory cell populations did not require the expression of CTLA-4 to function [54].

We choose to raise the regulatory cell lines in the presence of a mixture of T cell clones, instead of a single clone, because we suspected that in order to measure a significant change in the in vivo response to a given antigen one would have to regulate many different stimulating cells. This hypothesis was supported by the fact that when T cell lines were raised against only one of the T cell clones (D3 or DC10) it did not affect antibody production when introduced into mice (data not show). However, it is possible that regulatory cell lines raised against other individual T cell clones would have a similar effect than that of the lines TS1 and TS2. Furthermore, although the regulatory T cell lines would respond to a single clone with the production of the same cytokines synthesized when the mixture of clones was used, the magnitude of the response was much lower, often time on the borderline of the limit of detection, as described in the text.

It is interesting to note that both TS1 and TS2 seem to recognize a specific pattern on the T cell clones against which they were raised. This is clear both in vitro (Table 3) and in vivo(Figs 1–3). The response to clonotypic, idiotypic or ergotypic markers has been evoked as one of the main mechanisms through which T cell vaccination with autoreactive cells resulted in protection against the development of autoimmune diseases (for an updated review see Cohen [55]). This approach to the treatment of autoimmune diseases has been used successfully in many different animal models, including Lupus, autoimmune hepatitis, arthritis and EAE [56–61]. Although initial reports strongly suggested that regulatory cells are antigen specific [37,60], some debate has been raised since other reports have suggested that these cells suppress the immune response in a polyclonal fashion [62,63]. Both TS1 and TS2 cells specifically respond in vitro to the T cell clones against which they were derived, secreting cytokines and with enhanced viability (Tables 2 and 3). The data suggest that these cells may recognize the T cell receptor on these clones. In vivo, both TS1 and TS2 cells significantly suppress the anti-TNP response when triggered by TNP-KLH (Figs 1–3). Interestingly, data from Fig. 3 reveal that the anti-TNP response in TNP-Conalbumin immunized mice is suppressed by 50% when TS1 or TS2 cells are infused into immunized animals. Although not statistically significant (P < 0·06), the data is intriguing and may suggest that under certain circumstances the suppressive activity of these T cell lines may be non antigen-specific and directed simply towards activated cells.

Numerous studies on autoimmunity have presented evidences for the importance of the thymus in generating CD4+CD8 thymocytes with the capacity to prevent autoimmune activity [19,63,64]. Similar studies have shown it can also generate transplantation antigen-specific regulatory CD4+ T cells when allogeneic thymic epithelial cells are transplanted to nude mice [14,16]. These phenomena points out for the nature of the antigens involved in the generation of these cells as being intrathymic. However, the development of functional peripheral T regulatory lymphocytes depends also on the presence of the self antigen in the periphery, as shown in data from a variety of systems, reviewed by Seddon & Mason [65].

The data presented here, together with the studies on inflammatory bowl disease [23,66], suggest that T cell subpopulations with regulatory properties may respond to a variety of antigens in the periphery and not only to autoantigens or particular epitopes on the MHC molecules.

In all the in vitro experiments shown we used irradiated, recently stimulated KLH-specific T cell clones to activate the regulatory cell lines. This proved crucial since in all of the experiments using non-activated T cell clones we were unable to obtain significant cytokine production from either TS1 or TS2. It is possible that the requirement for activated T cell clones to stimulate a response from regulatory cells is solely linked to the ability of such clones to produce small amounts of stimulatory cytokines (IL-2 and IL-4) which the regulatory cells lack. However, a more complex explanation may be linked to the anti-ergotypic response of such cells. One could speculate that the in vivo function of regulatory cells is to recognize and modulate on going immune responses, thus its ability to distinguish between activated and resting target cells is pivotal for execution of their function [67].

One final point has to be made regarding the quality of the cell population used in the studies we presented here. It can be argued that by virtue of using T cell lines and not clones it is very difficult to correlate their regulatory activity with either the cytokine profile or the surface markers. As for the surface markers it is important to point out that over 95 percent of the cells used for the in vivo experiments expressed the same markers and we would be hard pressed to explain the results obtained based on the five percent that did not. Regarding the cytokines, their function is well known and they are the only cytokines these cells secrete. To further study the role of each cytokine in the suppression of antibody synthesis, cytokine-specific neutralizing antibodies could be used in vivo. However, since our read out system evokes the measurement of antibodies and these cytokines are involved in antibody class switch and antibody production its administration would likely interfere with the background synthesis of immunoglobulins. In fact, preliminary results from our laboratory so indicate. Furthermore, it is likely that other factors, such as cognate interaction between the regulatory cells and their targets are also involved in the suppressor activity exercised by these cell.

Regarding the origin of these cells some questions still remain. First, are these cells originated in the thymus, and if so do they differ from the intrathymic generated T regulatory cells once they were able to mediate an important degree of inhibition of the antibody response specific to a foreign antigen? Second, is antigen-specificity an intrinsic characteristic of these cells or a consequence of the activation profile of the cells that they target for suppression? and finally, is there a difference between the population of suppressor/regulatory cells involved in the control of autoimmunity and these cells that seem to be involved in the regulation of responses against nominal antigens?

ACKNOWLEDGEMENTS

This work was supported in part by grants from FAPESP. LVR is a recipient of a personal scientific achievement grant from CNPq. LV de Moraes is a recipient of a doctorate grant from FAPESP.

This work was developed in part at the Department of Paediatrics, Stanford University, California and at the Laboratory of Immunology, National Eye Institute, NIH, Maryland.

Ancillary