In Vitro Generation of Regulatory CD8⁺ T Cells Similar to those Found in Mice with Anterior Chamber–Associated Immune Deviation

Abstract

purpose. When injected intravenously into naïve mice, peritoneal exudate cells (PECs) incubated with ovalbumin (OVA) in the presence of transforming growth factor (TGF)-β2 induce immune deviation similar to that evoked by injection of OVA into the anterior chamber of the eye. Intraocular antigen injection elicits two distinct populations of regulatory T cells that impair delayed hypersensitivity (DH) by two different mechanisms: a CD4⁺ T cell that suppresses the induction of DH (afferent) and a CD8⁺ T cell that inhibits DH expression. In an effort to understand the origin and mechanism of action of these regulatory cells, CD8⁺ T cells from OVA-specific T cell receptor (Tcr) transgenic mice (OT-1) were used.

methods. CD8⁺ T cells were harvested from Tcr transgenic OT-I mice whose Tcr recognize an OVA peptide in the context of the class I major histocompatibility complex molecule K^b. These cells were stimulated in vitro with OVA-pulsed PECs exposed (or not) to TGF-β2, then analyzed for their capacity to proliferate, to secrete various cytokines, to lyse OVA-expressing target cells, and to regulate bystander T cells in vitro and in vivo.

results. When OVA-pulsed PECs were used in vitro as stimulators, responding OT-I T cells proliferated and preferentially secreted interferon (IFN)-γ, interleukin (IL)-2, and tumor necrosis factor (TNF)-α, rather than IL-4 and IL-10. When the stimulator PECs were pretreated with TGF-β2 and then pulsed with OVA, responding OT-I T cells proliferated even more swiftly, but they secreted significantly less IFN-γ, IL-2, and TNF-α, and no IL-4 or IL-10. OT-I T cells, which constitutively display cytotoxicity toward OVA-expressing target cells, lost this activity when stimulated with OVA-pulsed, TGF-β2–pretreated PECs. Moreover, OT-I T cells stimulated in this manner displayed the capacity to inhibit proliferation of OVA-primed T cells exposed to OVA in vitro and to suppress in vivo the expression of OVA-triggered DH.

conclusions. OVA-pulsed PECs, pretreated with TGF-β2, coerce naïve OVA-specific CD8⁺ T cells to become efferent regulators of DH similar to the regulatory T cells evoked by intraocular injection of OVA.

Intraocular injection of exogenous antigens induces an Ag-specific impairment of systemic delayed hypersensitivity (DH). A wide range of antigens (including soluble proteins, tumor-associated, viral, and minor and major histocompatibility complex Ag) can induce this response, termed anterior chamber–associated immune deviation (ACAID). ^{1 2 3 4} In mice in whom ACAID has been induced, two functionally distinct populations of regulatory cells have been described that suppress DH. ⁵ One population of T cells is CD4⁺, which prevents in normal mice the induction of immunity leading to the generation of DH effectors (i.e.,. afferent regulators). Another population of regulatory T cells is CD8⁺, which inhibits the expression of DH in vivo (i.e., efferent regulators). To understand the cellular and molecular bases of ACAID, we have given the highest priority to generating, isolating, and studying these afferent and efferent regulatory cells. 

To the present, attempts to harvest such cells in sufficient purity and quantity from spleens of mice with ACAID have failed. As an alternative approach, we have attempted to generate candidate regulatory cells in vitro by stimulating antigen-specific Tcr transgenic T cells with antigen-presenting cells pretreated with transforming growth factor (TGF)-β2. It is now well established that peritoneal exudate cells (PECs) treated in vitro with TGF-β2 acquire ACAID-inducing properties. ⁶ We have recently determined that naïve CD4⁺ ovalbumin (OVA)-specific T cells from DO11.10 Tcr transgenic mice that were exposed to OVA-pulsed, TGF-β2–pretreated PECs acquired the ability to suppress both induction and expression of DH in normal BALB/c mice. ⁷  

In the present study, we examined the functional properties of CD8⁺, OVA-specific T cells from OT-I mice after in vitro stimulation with OVA-pulsed, TGF-β2–treated PECs. Our results indicate that OT-I T cells activated in this manner underwent proliferation but lost their cytotoxic capabilities. Moreover, the cells acquired the capacity to inhibit OVA-specific T cell proliferation in vitro and to suppress the expression of OVA-specific DH in vivo. 

Methods

Mice

Female C57BL/6 (B6) mice, between 6 and 8 weeks of age, were purchased from Taconic Farms (Germantown, NY). These mice were used as a source of PECs. OT-I Tcr transgenic mice (C57BL/6 background) were maintained in our colony (original parents were a kind gift of Michael Bevan, University of Washington, Seattle, WA). All animals were treated according to the ARVO Statement on the Use of Animals in Ophthalmic and Vision Research. T cells from these mice recognize a peptide (residues 257–264) derived from OVA presented in the context of K^b. ⁸ OT-I mice were used as the source of T cells, and as experimental subjects for in vivo studies of the expression of DH. OVA-specific Tcr transgenic T cells in OT-I mice were identified by surface flow cytometry for the expression of CD8 and V_β5. In general, 17% to 40% of CD3⁺ T cells in lymphoid cell suspensions expressed CD8 and V_β5. 

Culture Medium

Serum-free medium was used for all cell cultures. ⁹ This medium was composed of RPMI 1640, 10 mM HEPES, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin (all from Biowhitaker, Walksville, MD), 1 × 10⁻⁵ M 2-ME (Sigma Chemical, St. Louis, MO), supplemented with 0.1% bovine serum albumin (Sigma Chemical), ITS⁺ culture supplement [1 μg/ml iron-free transferrin, 10 ng/ml linoleic acid, 0.3 ng/ml Na₂Se, and 0.2 μg/ml Fe(NO₃)₃; Collaborative Biochemical Products, Bedford, MA]. 

Reagents

Porcine TGF-β2, anti–TGF-β2 neutralizing antibody, and nonspecific goat antibody were purchased from R&D Systems (Minneapolis, MN). OVA and human serum albumin (HSA) were obtained from Sigma. 

Preparation of Pure OT-I T Cells

Spleens were removed from OT-I mice and pressed through nylon mesh to produce a single-cell suspension. Red blood cells were lysed with Tris–NH₄Cl. The remaining cells were washed three times with RPMI 1640 and passed through T cell enrichment columns (R&D Systems). The resultant cell suspension contained >98% CD3⁺ cells. 

Preparation of PECs Pretreated with TGF-β2

PECs were harvested from normal C57BL/6 mice that received 2.5 ml of thioglycolate (Sigma) intraperitoneally 3 days earlier. As described previously, ¹⁰ the cells were washed and resuspended, placed in 24- (1 × 10⁶/well) or 96-well (1 × 10⁵/well) culture plates, and treated with or without 5 ng/ml of porcine TGF-β2 in serum-free medium at 37°C in an atmosphere of 5% CO₂. After overnight culture, plates were washed three times with culture medium to remove TGF-β2 and nonadherent cells. Adherent cells were retained in the wells for use in all subsequent experiments. More than 90% of these adherent cells were F4/80⁺ macrophages. 

Cytokine Assays

To assay for content of interferon (IFN)-γ, interleukin (IL)-2, IL-4, IL-10, tumor necrosis factor (TNF)-α, and TGF-β2, OT-I T cells (2 × 10⁴/well) were added to 96-well plates containing TGF-β2–pretreated (or not) PECs (1 × 10⁵/well) and cultured with or without various concentrations of naive OVA in serum-free medium for 24 hours. In some experiments, irradiated (2000 R) OT-I T cells (2 × 10⁴/well) first cultured with OVA-pulsed, TGF-β2–pretreated (or not) PECs were then cocultured in 96-well plates for 24, 48, and 72 hours with or without naive OVA- or HSA-pulsed PECs, and with T cells from C57BL/6 mice primed in vivo to OVA or HSA. Primed responder T cells were obtained as splenocytes from normal B6 mice primed 7 days previously with OVA or HSA plus complete Freund’s adjuvant. At each time point, supernatants were collected and analyzed by quantitative capture enzyme-linked immunosorbent assay (ELISA), according to the manufacturer’s instructions (PharMingen, San Diego, CA). Rat monoclonal antibodies (mAbs) to mouse cytokine IFN-γ (R4-6A2), IL-2 (JES6-1A12), IL-4 (BVD4-1D11), IL-10 (JES5-2A5), and TNF-α (G281-2626) were purchased from PharMingen and used as coating Abs. Biotinylated rat mAbs to mouse cytokines IFN-γ (XMG1.2), IL-2 (JES6-5H4), IL-4 (BVD6-24G2), IL-10 (SXC-1), and TNF-α (MP6-XT3; PharMingen) were used as detecting Abs. TGF-β2 was quantified by ELISA kit (Promega, Madison, WI). For this ELISA to measure total TGF-β2, supernatants were first treated with 1N HCl at a dilution of 1/10 for 1 hour at room temperature, then neutralized with a mixture of 1N NaOH/1N HEPES at a dilution of 1/5. 

Proliferation Assay of T Cells Activated In Vitro

OT-I T cells were added (2 × 10⁴/well) to 96-well plates containing PECs (1 × 10⁵/well) pretreated (or not) with 5 ng/ml of TGF-β2, and cultured with or without various concentrations of naive OVA in serum-free medium for 24 to 96 hours at 37°C in an atmosphere of 5% CO_2. In some experiments, X-irradiated (2000 R) OT-I T cells (2 × 10⁴/well), first cultured with OVA-pulsed, TGF-β2–pretreated (or not) PECs, were then cocultured in 96-well plates with or without OVA- or HSA-pulsed PECs and T cells from C57BL/6 mice primed in vivo to OVA or HSA. The cultures were sustained for 24, 48, 72, or 96 hours, pulsed with 0.5μ Ci [³H]-thymidine 8 hours before termination, and then harvested onto glass filters using an automated cell harvester (Tomtec, Orange, CT). Radioactivity was assessed by liquid scintillation spectrometry, and the amount expressed as counts per minute. 

Assay for Cytotoxic Activity

To generate cytotoxic T lymphocytes (CTL), T cells (5 × 10⁵/well) taken from OT-I mice were stimulated with TGF-β2 (5 ng/ml) pretreated (or not), OVA (400 μg/ml)-pulsed (or not) PECs (1 × 10⁶) for 0, 24, 48, or 96 hours in 24-well plates in serum-free medium. These cells were then washed three times and prepared as effector cells. EG.7 cells (EL4 cells transfected with the OVA gene, ¹¹ a kind gift of Michael Bevan, University of Washington, Seattle, WA) were labeled with[ ⁵¹Cr]sodium chromate in serum-free medium for 1.5 hour at 37°C. The cells were washed three times, and 10⁴ cells were transferred to each well of a 96-well plate. Effector cells were added to these wells at E:T cell ratios ranging from 50:1 to 3:1 and incubated at 37°C in CO₂ in serum-free medium. After 4 hours, the plate was centrifuged, and 25 μl of supernatant was removed and counted. Results are expressed as the percent lysis as a function of the E:T cell ratio. The percent lysis = (test ⁵¹Cr released − spontaneous ⁵¹Cr released)/(maximum ⁵¹Cr released − spontaneous ⁵¹Cr released) × 100. Spontaneous ⁵¹Cr release was determined using ⁵¹Cr-labeled targets in the absence of effectors and maximum ⁵¹Cr release after treatment of ⁵¹Cr-labeled targets with 5N HCl. 

Local Adoptive Transfer Assay of DH

This assay, as described previously, was used to detect the capacity of in vitro–activated OT-I T cells to suppress the expression of DH in vivo. ¹² Briefly, OT-I T cells (5 × 10⁵) were cultured in 24-well plates containing TGF-β2–treated, or –untreated, PECs and 400 μg/ml OVA. After 72 hours, nonadherent OT-I cells were harvested, washed three times, and exposed to X-irradiation (2000 R). These cells (as regulators) were added (1 × 10⁵/inoculum) to suspensions of OVA-pulsed PECs (as stimulators, 5 × 10⁵/inoculum) and responder T cells (5 × 10⁵/inoculum). Responder T cells were obtained as splenocytes from normal C57BL/6 mice primed 7 days previously with OVA plus complete Freund’s adjuvant (CFA). The mixture of responders, stimulators, and regulators was injected (10 μl/injection) into the ear pinnae of naive B6 mice. Ear swelling responses were assessed with an engineer’s micrometer (Mitsutoyo; MTI Corporation, Paramus, NJ) at 24 and 48 hours. In some experiments, neutralizing anti–TGF-β2 antibodies or nonspecific goat antibodies (100 μg/mouse) were mixed with the cells and injected into ear pinnae. 

Statistical Analyses

Results of experiments were analyzed by ANOVA and Scheffé test. Mean values were considered to be significantly different when P < 0.05. 

Results

Capacity of OVA-Pulsed, TGF-β2–Treated Antigen-Presenting Cells to Activate CD8⁺ T Cells from Tcr Transgenic OT-I Mice

Our first goal was to determine whether OT-I T cells, which are OVA-specific and restricted to H-2K^b, were capable of being activated in vitro after exposure to PECs that were treated with TGF-β2 and pulsed with OVA. Accordingly, PECs were obtained from C57BL/6 donors and cultured with or without 5 ng/ml TGF-β2 and various concentrations of OVA. After overnight culture, the cells were washed to remove TGF-β2, OVA, and nonadherent cells, and then cocultured with naïve OT-I T cells. Two types of experimental assays were then performed: proliferation and cytokine content of supernatants. For assessment of T-cell proliferation,[ ³H]-thymidine was added and individual cultures were halted at 24, 48, 72, and 96 hours. Radioisotope incorporation was then determined. The results of a representative experiment are presented in Figure 1 . As expected, OT-I T cells that had been stimulated with OVA-pulsed, TGF-β–untreated PECs began to proliferate at 72 hours, peaking at 96 hours. Similarly, OT-I T cells stimulated with OVA-pulsed, TGF-β–treated PECs underwent proliferation; however, OT-I cells stimulated in this manner initiated incorporation of radioisotope within 48 hours (earlier than OT-I T cells stimulated by untreated PECs). Moreover, OT-I T cells exposed to TGF-β2–treated PECs rapidly lost their proliferative capacity when assessed at 96 hours. These results reveal that OT-I T cells can be activated by both TGF-β–treated and TGF-β–untreated PECs but that the kinetics of the response are different. T cells responding to TGF-β–treated PECs showed earlier proliferation followed by loss of this function, whereas conventionally stimulated T cells gradually displayed a sustained, proliferation phenotype. 

Supernatants from similar cultures were collected and assayed for cytokine production (IFN-γ, IL-2, IL-4, IL-10, and TNF-α) at 24, 48, and 72 hours, using a quantitative capture ELISA. Results of a representative experiment conducted with 24-hour supernatants are displayed in Figure 2 . When maintained in culture for 24 hours, OT-I T cells stimulated with OVA-pulsed, TGF-β–untreated PECs secreted predominantly IFN-γ, IL-2, and TNF-α, rather than IL-4 and IL-10 (data not shown). The levels of cytokine secretion correlated positively with the antigen dose used to pulse the antigen-presenting cells. By contrast, OT-I T cells that had been stimulated with OVA-pulsed, TGF-β2–pretreated PECs secreted much less IFN-γ, IL-2, and TNF-α. Despite the poor production of these cytokines, OT-I T cells stimulated with OVA and TGF-β–treated PECs failed to generate detectable amounts of IL-4 or -10 (data not shown). Similar results were obtained with supernatants harvested from cultures of 48 and 72 hours’ duration (data not shown). Several conclusions can be drawn from these experiments. OVA-pulsed PECs, whether treated with TGF-β or not, were capable of activating OT-I T cells. However, the type and consequences of activation differed: OT-I T cells activated by TGF-β–treated PECs underwent mitosis very quickly but failed to secrete any of the assayed cytokines. This precocious mitotic activity proved to be short-lived, with little or no proliferation occurring beyond 48 hours of culture. By contrast, OT-I T cells stimulated with untreated PECs proliferated and secreted IFN-γ, IL-2, and TNF-α, but not IL-4 or IL-10, implying that these cells were differentiating toward the T1, rather than the T2, phenotype. 

Influence of OVA-Pulsed, TGF-β2–Treated Antigen-Presenting Cells on the Cytotoxic Activity of CD8⁺ OT-I T Cells

Because CD8⁺ T cells have been found classically to be cytotoxic effectors, we next determined whether naïve OT-I T cells stimulated in vitro by OVA-pulsed, TGF-β2–treated PECs acquired cytotoxic function. To prepare for this inquiry, it was first necessary to determine the cytotoxic activity of naïve T cells obtained from normal OT-I mice. As target cells for the cytotoxicity assay, we used EG.7 cells; these cells are EL-4 tumor cells (derived from H-2^b mice) that have been transfected with the OVA gene and express peptides derived from the OVA molecule in the context of H-2^b class I major histocompatibility complex (MHC) molecules. ¹¹ To our surprise, OT-I T cells harvested directly from normal OT-I mice and placed in a 4-hour ⁵¹Cr release assay with EG.7 target cells induced significant levels of radioisotope release, indicating that naïve OT-I T cells are constitutively cytotoxic (data not shown). Next, we cultured naïve OT-I T cells with OVA-pulsed PECs that had either been pretreated with TGF-β2 or not. These cultures were halted at 24, 48, or 96 hours, and the T cells were removed and assayed on EG.7 targets for cytotoxic activity. Results of one such experiment are presented in Figure 3 . Fresh OT-I T cells lysed EG.7 target cells directly. Moreover, OT-I T cells harvested from cultures containing OVA plus PECs that had not been pretreated with TGF-β2 continued to display high levels of cytotoxic activity at each time point examined (24, 48, and 96 hours). When OVA was omitted from these cultures, cytotoxic activity diminished progressively at each sequential time period. Alternatively, T cells harvested from cultures containing OVA plus TGF-β–treated PECs rapidly lost their cytotoxic functions, displaying little or no such activity at 24 hours and thereafter. Similarly, OT-I T cells lost their cytotoxic function when OVA was omitted from cultures containing TGF-β–treated PECs. Naïve T cells from normal C57BL/6 mice displayed no cytotoxic activity at any time point. These findings enable us to conclude that the constitutive cytotoxic capacity displayed by fresh, naïve OT-I T cells can be maintained if the cells are exposed continually in vitro to OVA-pulsed PECs. If, however, the PECs are pretreated with TGF-β, the OT-I T cells in the culture rapidly lose their cytotoxicity, whether OVA is present or not. 

In Vitro Regulatory Properties of OT-I T Cells Activated by Exposure to OVA-Pulsed, TGF-β2–Pretreated PECs

Based on the preceding results, which indicated that OT-I T cells exposed to OVA-pulsed, TGF-β–pretreated PECs had been activated but failed to secrete T1-type cytokines and lost their cytotoxic capacity, we next examined the possibility that these in vitro–activated cells might have acquired alternative, regulatory properties. Specifically, we examined whether in vitro–activated OT-I T cells were able to regulate the responses of in vivo–primed OVA-specific T cells. In the first of these experiments, we examined the effects of in vitro–activated OT-I T cells on the capacity of OVA-primed T cells to proliferate in vitro in response to OVA. OT-I T cells were stimulated with OVA-pulsed, TGF-β2–pretreated PECs for 72 hours, then washed and exposed to X-irradiation (2000 R). In separate culture dishes, freshly prepared PECs were cultured overnight with OVA, then washed and plated as “stimulators” in a 96-well culture plate. T cells obtained from C57BL/6 mice immunized 1 week previously with OVA plus CFA were added to these cultures as “responders,” and then X-irradiated OT-I T cells were added as “regulators.” In positive control cultures, X-irradiated OT-I T cells that had not been pretreated with TGF-β, or X-irradiated naïve C57BL/6 T cells were added as regulators. In negative controls, naïve C57BL/6 T cells were used as responders. Proliferation responses were measured after 72 hours of culture. The results of a representative experiment are displayed in Figure 4A . In positive controls, T-cell proliferation was readily apparent when OVA was present in the culture, whereas in cultures containing X-irradiated OT-I T cells activated in vitro with TGF-β–treated PECs proliferation was sharply curtailed. Similar results were obtained in cultures of 48 and 96 hours’ duration. 

In the next set of experiments, the effects of in vitro–activated OT-I T cells on cytokine production by OVA-primed T cells stimulated in vitro were examined. Cultures similar to those just described were established. At 24, 48, and 72 hours, supernatants were collected and assayed for content of IFN-γ and IL-4 by ELISA. As the results displayed in Figure 4B indicate, IFN-γ production was suppressed at 24 hours when OVA-primed T cells were cocultured with OT-I T cells stimulated with OVA and TGF-β2 PECs, whereas IFN-γ production was robust in cultures containing the positive controls. In none of these cultures was IL-4 detected (data not shown). Essentially identical results were obtained with supernatants harvested at 48 and 72 hours of culture (data not shown). These data indicate that OT-I T cells that had been activated in vitro by OVA-pulsed, TGF-β2–pretreated PECs acquired the capacity to downregulate proliferation and IFN-γ secretion by OVA-primed T cells stimulated with OVA-pulsed antigen-presenting cells (APCs) in vitro. 

To determine whether OT-I T cells stimulated with OVA and TGF-β2–pretreated PECs suppressed responder cell function in an antigen-specific manner, C57BL/6 mice were immunized with HSA plus CFA. One week later, T cells were obtained from these mice and used as responders in cultures similar in design to those described above. In particular, the experimental cultures contained HSA-pulsed PECs plus X-irradiated OT-I T cells that had been activated previously by in vitro exposure to OVA-pulsed, TGF-β2–pretreated PECs. The results of this experiment are displayed in Figure 5 . “Regulator” OT-I T cells failed to inhibit the proliferation of HSA-primed T cells (Fig. 5A) , and they failed to prevent these T cells from producing IFN-γ when stimulated with OVA-pulsed APCs in vitro (Fig. 5B) . Once again, no IL-4 was detected in any culture supernatants (data not shown). We concluded that CD8⁺ OT-I T cells activated in vitro by TGF-β2–pretreated PECs acquire regulatory properties that equip them to modulate the behavior of bystander T cells in an antigen-specific manner. 

In Vivo Regulatory Properties of OT-I T Cells Activated by Exposure to OVA-Pulsed, TGF-β2–Pretreated PECs

Our ultimate goal is to understand the process by which antigen-specific CD8⁺ T cells in ACAID suppress the expression of DH in vivo. In vitro activated OT-I T cells represent candidates for this property, and the next experiments tested this possibility directly. The local adoptive transfer assay examines the capacity of CD4⁺ effector T cells to mediate DH after their injection with antigen into the ear pinnae. We generated effector T cells by immunizing normal C57BL/6 mice with OVA plus CFA. T cells were harvested from these mice 1 week later and used as responders in a cell suspension containing (a) OVA-pulsed PECs as stimulators and (b) OT-I T cells that had been exposed overnight for 72 hours in vitro to OVA-pulsed, TGF-β–pretreated PECs as regulators. The latter cells were X-irradiated with 2000 R immediately before injection. This cell mixture, containing 5 × 10⁵ responders, 5 × 10⁵ stimulators, and 1 × 10⁵ regulators, was injected subcutaneously into the ear pinnae of naïve C57BL/6 mice. Positive control injections contained similar responders and stimulators, but regulators that were either naïve C57BL/6 T cells or OT-I T cells that had been exposed in vitro to OVA-pulsed, TGF-β–untreated PECs. Negative controls contained naïve C57BL/6 T cells as “responders.” Ear swelling responses were assessed 24 and 48 hours later. The results of a representative experiment are presented in Figure 6 . When the injection mixture contained naive B6 T cells or OT-I T cells exposed in vitro to PECs that were not treated with TGF-β2 (positive control), ear swelling responses were intense. However, cell mixtures that contained regulatory OT-I T cells exposed in vitro to PECs treated with TGF-β2 displayed only feeble ear swelling responses, not significantly different from the negative controls. These results indicate that OT-I T cells activated in vitro by OVA-pulsed, TGF-β2–pretreated PECs acquired the ability to suppress the expression of DH in vivo. 

Possible Role of TGF-β in the Regulatory Activity of In Vitro–Activated OT-I T Cells

We have recently found that CD4⁺ DO11.10 T cells activated in vitro by OVA-pulsed, TGF-β2–pretreated PECs secrete TGF-β. ⁷ Moreover, the ability of these cells to suppress the expression of DH in vivo is mediated by the TGF-β2 they elaborate. We next wished to determine whether in vitro–activated OT-I T cells also used TGF-β2 to mediate their regulatory functions. Accordingly, regulatory T cells were generated by exposing OT-I T cells in vitro for 72 hours to OVA-pulsed, TGF-β2–pretreated PECs. The cells were then harvested and placed in culture for 24 hours under serum-free conditions. Supernatants were harvested and assayed for content of TGF-β2. Our results indicate that OT-I T cells spontaneously secrete both mature TGF-β2 and latent TGF-β2 but that they produced no more TGF-β2 when stimulated with TGF-β2–treated PECs than did OT-I T cells stimulated with PECs in the absence of TGF-β2 (data not shown). 

To explore the same issue in vivo, OT-I T cells were activated in vitro with OVA-pulsed, TGF-β2–pretreated PECs and then used as regulators in a local adoptive transfer assay as described above. Neutralizing anti–TGF-β2 antibodies were added to cell mixtures containing regulators, responders, and stimulators. Controls contained isotype control antibodies. The results of a representation experiment are presented in Figure 7 . Ear swelling responses were reduced when in vitro–activated OT-I T cells were present in the inoculum, whether anti–TGF-β2 antibodies were present or not. We concluded that the ability of in vitro–activated OT-I T cells to regulate bystander T cells in vivo is not dependent on the generation of TGF-β2. 

Discussion

ACAID has emerged as a key component of the multifactorial mechanisms responsible for ocular immune privilege. ^{1 2} This stereotypic systemic immune response to intraocular antigens helps to explain why antigenic material introduced into the anterior chamber fails to evoke a tissue-destructive cell-mediated immunity of the DH type. ACAID represents a form of peripheral tolerance in which regulatory T cells contribute to the antigen-specific unresponsive state. ^{13 14} First described more than 25 years ago and termed “suppressor T cells,” regulatory T cells have been difficult to isolate and study from animals rendered experimentally tolerant, irrespective of the strategy for tolerance induction. We know of the existence of regulatory T cells in ACAID, not by virtue of having isolated and studied purified cells but by using negative selection experiments. Thus, regulatory CD4⁺ T cells have been described in ACAID by harvesting CD8-depleted splenic T cells from donors with ACAID and using them in adoptive transfer experiments that result in impaired induction of DH in naïve recipients. ⁵ Similarly, regulatory CD8⁺ T cells of ACAID have been identified by virtue of the capacity of CD4-depleted spleen cells from donors with ACAID to suppress the expression of DH in vivo. ^{5 15 16}  

Unfortunately, the apparent rarity of CD4⁺ and CD8⁺ regulatory T cells in spleens of mice with ACAID has precluded their direct isolation and purification. Alternative strategies to study these cells “at a distance” have met with only limited success. Kosiewicz et al. ¹⁷ have reported that splenic T cells removed from mice that received intracameral injections of OVA 1 week previously failed to undergo detectable proliferation, or to secrete IL-2, IL-4, IL-10, or IFN-γ when stimulated in vitro with OVA. However, CD4⁺ T cells stimulated in this manner did secrete increased amounts of TGF-β compared with controls that received an injection of OVA by the intravenous or subcutaneous routes. ¹⁷ In related studies, Kosiewicz et al., ¹⁷ and D’Orazio et al., ^{18 19 20} have found OVA-specific T cells that resemble the Th2 phenotype in the spleens and lymph nodes of mice that first encounter OVA via the anterior chamber and then receive a subcutaneous immunization with OVA plus CFA. It is still unclear whether these putative Th2 cells are the direct result of the anterior chamber route of antigen injection, or whether this route of antigen injection primes T cells in a manner that promotes their differentiation into Th2 cells when subsequently stimulated with antigen plus adjuvant. In any event, it has not been possible to harvest responding cells of this type from mice with ACAID for further detailed analysis. 

Until or unless we succeed in isolating regulatory T cells from living mice with ACAID, we are forced to turn to in vitro models that in one way or another resemble the ACAID response. Such an in vitro model is created when PECs are treated with TGF-β in vitro and then pulsed with antigen. ^{21 22 23 24} Cells of this type induce OVA-specific immune deviation when injected intravenously in naïve mice, a response that resembles ACAID in several parameters. In the present studies, we stimulated OT-I T cells with PECs exposed to TGF-β2 and OVA in vitro and examined the T-cell responses thereafter. T cells of OT-I mice are relevant because (a) they recognize an OVA-derived peptide in the context of a class I molecule and (b) they are CD8⁺. ^{8 25} Thus, they have the potential to become the efferent regulatory T cells found in mice with ACAID. 

Our results confirm the value of this alternative approach to studying ACAID regulatory T cells. OT-I T cells were activated to proliferate promptly when exposed to OVA-pulsed, TGF-β2–pretreated PECs. However, the functional properties of the responding T cells differed substantially from OT-I T cells activated by OVA-pulsed PECs that had not been treated with TGF-β. The former cells secreted none of the cytokines we assayed (IFN-γ, IL-2, TNF-α, IL-4, IL-10). In addition, they lost their capacity to lyse OVA-expressing target cells, a property constitutively displayed by naïve OT-I T cells. However, OT-I T cells activated in vitro by TGF-β–treated PECs were not rendered nonfunctional. When added to cultures containing primed OVA-specific T cells plus OVA-pulsed APCs, they suppressed responder cell proliferation and cytokine production. When mixed with OVA-specific effector T cells and OVA and injected into ear pinnae of normal mice, they prevented the expression of DH. We concluded that by exposing naïve CD8⁺ T cells to antigen presented by TGF-β–treated APCs we can generate regulatory T cells that resemble functionally the efferent regulatory T cells typically found in the spleens of mice with ACAID. Whether these cells are identical to the efferent suppressor cells found in ACAID remains to be determined. 

The implication of our results extends well beyond this important conclusion. First, the loss of cytotoxic function by OT-I T cells activated in vitro by TGF-β–treated PECs implies that the ability of these cells to suppress DH expression is not mediated by cytotoxicity directed at OVA-bearing APCs, a possibility we had originally considered. Second, the failure of OT-I T cells activated by TGF-β–treated PECs to secrete enhanced amounts of TGF-β, as well as the inability of anti–TGF-β antibodies to reverse suppression of DH expression by these cells, indicates that TGF-β is not directly responsible for the inhibition observed. This is somewhat surprising because we have recently found that DO11.10 T cells activated in vitro by OVA-pulsed, TGF-β–pretreated APCs differentiate into regulatory T cells that suppress the expression of DH through their secretion of TGF-β. ⁷ Third, OT-I T cells stimulated with OVA-pulsed, TGF-β–pretreated PECs failed to secrete detectable amounts of IL-4 and IL-10. This implies that under the conditions of in vitro activation, OT-I T cells do not differentiate in the direction of T2 cells. In this manner, OT-I T cells differ from DO11.10 T cells similarly activated. Based on reports of Takeuchi et al. ¹⁰ and D’Orazio et al., ¹⁹ DO11.10 T cells differentiate into Th2-type cells when activated in vitro by TGF-β–treated APCs. Taken together, these findings have enabled us to exclude several mechanisms postulated to account for the inhibitory activities of CD8⁺ OT-I T cell regulators. Unfortunately, positive identification of the operative mechanism remains to be achieved. 

The specificity of the suppression mediated by in vitro–activated OT-I T cells is worthy of comment. OT-I T cells first exposed to OVA-pulsed, TGF-β–pretreated PECs readily suppressed proliferation and IFN-γ secretion by T cells from mice primed with OVA in vivo. However, in vitro–activated T cells failed to suppress activation of T cells primed to HSA in vivo. The failure was even evident in cultures in which the APCs were pulsed with both HSA and OVA. It is pertinent that regulatory OT-I T cells neither secrete TGF-β nor use TGF-β in suppression of DH expression in vivo. The failure of in vitro–activated OT-I T cells to suppress T cells primed to irrelevant antigen is consistent with our interpretation that suppression by these cells is probably not dependent on the release of a soluble immunosuppressive factor. 

Much has been learned about the effects of TGF-β2 treatment on the antigen processing and presenting functions of APCs. Takeuchi et al. reported that PECs treated in this manner secrete reduced amounts of IL-12, upregulate CD40 expression poorly, and secrete increased amounts of mature TGF-β compared with untreated PECs. ²⁶ Circumstantial evidence has been presented to support the view that each of these effects contributes to the unusual phenotype of DO11.10 T cells that are exposed to OVA-pulsed, TGF-β–treated PECs. It remains to be determined whether the unusual functional properties displayed by OT-I T cells exposed to OVA-pulsed, TGF-β-treated PECs owe their origins to similar properties of the APC, or whether another set of TGF-β–dependent properties is involved. Experiments to examine this issue are under way. 

We are at a loss to explain the prompt, almost precocious, proliferation displayed by OT-I T cells exposed to OVA-pulsed, TGF-β2–pretreated PECs. Nor do we know whether early proliferation followed by inability to proliferate is related to the regulatory properties displayed by these cells. At the very least the fact that regulatory cells emerge from these cultures indicates that early proliferation does not lead directly to apoptotic cell death. 

 Supported by USPHS Grant EY 05678.

 Submitted for publication September 27, 1999; revised December 20, 1999; accepted December 30, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

Proliferative responses of OT-I T cells stimulated with TGF-β2–pretreated PECs and various doses of OVA. PECs were cultured overnight with or without 5 ng/ml of TGF-β2 and indicated concentrations of OVA, then washed, and cocultured with OT-I T cells. OT-I T cells were cultured with no PECs as negative control. T-cell proliferation responses were measured at 48, 72, and 96 hours.[ ³H]thymidine was added 8 hours before termination of culture. The experiment was repeated three times with similar results. Each data point represents the mean ± SD of triplicate cultures.* P < 0.05, **P < 0.01.

Figure 2.

 

Secretion of cytokines by OT-I T cells stimulated with OVA-pulsed, TGF-β2–treated PECs. PECs were cultured overnight with and without 5 ng/ml of TGF-β2 and indicated concentrations of OVA, then washed and cocultured with OT-I T cells. After 24 hours, the culture supernatants were harvested and assayed for IFN-γ, IL-2, and TNF-α by ELISA. The experiment was repeated four times with similar results. Each data point represents the mean ± SD of duplicate cultures.* P < 0.05, **P < 0.01.

Figure 3.

 

Cytotoxic activity of naive and OT-I T cells after in vitro stimulation. OT-I T cells (5 × 10⁵/well) were cocultured with or without OVA (400 μg/ml) and TGF-β2 (5 ng/ml)–pretreated PECs (1 × 10⁶/well) in 24-well plates for 0, 24, 48, and 96 hours. Effector-to–target cell ratio was 50:1. Viable cells were tested for their ability to lyse EG.7 cells during a 4-hour assay. The experiment was repeated four times with similar results. *P < 0.05.

Figure 4.

 

Impairment of proliferative responses of OVA-primed T cells cocultured with X-irradiated OT-I T cells stimulated with OVA and TGF-β2–pretreated PECs. OT-I T cells were stimulated with OVA-pulsed, TGF-β2–pretreated PECs for 72 hours, then washed and exposed to X-irradiation (2000 R). For positive controls, OT-I T cells were X-irradiated after stimulation with OVA-pulsed PECs in the absence of TGF-β2. PECs (1 × 10⁵/well) were cultured overnight with 100 μg/ml of OVA, then washed and prepared as“ stimulators” in a 96-well culture plate. In vitro–activated X-irradiated cells were added (2 × 10⁴/well) as“ regulators” to cell mixtures comprised of “responders” (T cells from B6 mice primed in vivo with OVA plus CFA, 3 × 10⁵/well). X-irradiated T cells from naive B6 mice were used as regulators in negative controls. (A) T-cell proliferation responses were measured at 72 hours.[ ³H]thymidine was added 8 hours before termination of culture. (B) The culture supernatants after 24 hours were harvested and assayed for IFN-γ by ELISA. The experiment was repeated twice with similar results. Each data point represents the mean ± SD of triplicate cultures.** P < 0.05.

Figure 5.

 

Specificity of impaired proliferative responses of HSA-primed T cells cocultured with X-irradiated OT-I T cells stimulated with OVA and TGF-β2–pretreated PECs. Responder T cells were obtained from normal C57BL/6 mice immunized with HSA plus CFA 1 week previously. Regulator OT-I T cells were prepared by stimulation with OVA-pulsed, TGF-β2–pretreated PECs for 72 hours, then washed and exposed to X-irradiation (2000 R). Regulator X-irradiated T cells for positive controls were stimulated with OVA-pulsed PECs in the absence of TGF-β2. Stimulator PECs (1 × 10⁵/well) were cultured overnight with 100 μg/ml of OVA and/or HSA, then washed and added to 96-well culture plates. In vitro–activated X-irradiated cells were added (2 × 10⁴/well) as regulators to cell mixtures comprised of responders (3 × 10⁵/well). X-irradiated T cells from naive B6 mice were used as regulators in negative controls. (A) T-cell proliferation responses were measured at 72 hours. [³H]thymidine was added 8 hours before termination of culture. *Indicates values not significantly different from each other or from negative control. (B) Culture supernatants were harvested at 24 hours and assayed for IFN-γ by ELISA. The experiment was repeated three times with similar results. Each data point represents the mean ± SD of triplicate cultures. Asterisks indicate activity not detected.

Figure 6.

 

Inhibition of DH expression by OT-I T cells stimulated with OVA and TGF-β2–pretreated PECs. Regulator cells, OVA-specific responder cells, and stimulator cells pulsed with OVA were prepared as described in the legend to Figure 5 . Mixtures containing regulators (1 × 10⁵/injection), responders (5 × 10⁵/injection), and stimulators (5 × 10⁵/injection) were injected (10 μl) into the ear pinnae of normal B6 mice. T cells from naive B6 mice were used as responders in negative controls. Ear swelling responses at 24 hours were assessed and expressed as mean ± SEM (n = 5) and compared with positive control. Bars represent the mean ± SEM (n = 5) of ear measurements for typical experiment. Asterisks indicate values significantly less than positive controls, P < 0.05.

Figure 7.

 

Effect of anti–TGF-β antibodies on inhibition of DH expression by OT-I T cells stimulated with OVA and TGF-β2–pretreated PECs. Regulator cells, OVA-specific responder cells, and stimulator cells pulsed with OVA were prepared as described in the legend to Figure 5 . Mixtures containing regulators (1 × 10⁵/injection), responders (5 × 10⁵/injection), and stimulators (5 × 10⁵/injection) were injected (10 μl) into the ear pinnae of normal B6 mice. Neutralizing anti–TGF-β2 antibodies or nonspecific goat antibodies (100 μg/mouse) were mixed with the cells and injected into ear pinnae. T cells from naive B6 mice were used as responders in negative controls. Ear swelling responses at 24 hours were assessed and expressed as mean ± SEM (n = 5), and compared with positive control. Bars represent the mean ± SEM (n = 5) of ear measurements for typical experiment. Asterisks indicate values significantly less than positive controls, P < 0.01.

We thank Bruce Ksander, MD, for helpful suggestions and Jacqueline Doherty, MD, and Marie Ortega for managerial assistance.