CTLA4Ig-Induced Linked Regulation of Allogeneic T Cell Responses¹

CTLA4Ig is a fusion protein that can induce allospecific hyporesponsiveness in vitro (1) and in vivo (2) by preventing CD28 activation by B7.1 and B7.2. The powerful immunosuppressive effects exerted by CTLA4Ig and the availability of a humanized CTLA4Ig variant (3) combine to make this an important agent in therapeutic strategies aimed at inducing tolerance in human allo- and xenograft recipients (4). Although a variety of mechanisms have been put forth to explain the immunosuppressive effects of CTLA4Ig (2, 5, 6, 7), including T cell anergy (8, 9, 10), the exact cellular mechanisms underlying the action of CTLA4Ig are unclear. It is known that the suppressive effects of CTLA4Ig can be observed in euthymic recipients in the absence of circulating levels of the agent (11). This observation suggests that the mechanism of action of CTLA4Ig is not simply T cell anergy, as maturing T cells would be expected to retain responsiveness to donor Ag in the absence of CTLA4Ig.

T cell anergy has traditionally been defined as the absence of a T cell response, when the TCR is engaged with an MHC receptor but a second costimulatory signal is not delivered (12). However, the conventional understanding of T cell anergy has been challenged by the recent observations by Lechler and others that anergic T cells can suppress the responses of other T cells (13, 14, 15, 16). Thus, instead of simply not responding, certain anergic T cells can actively inhibit the alloresponses of other naive T cells. For instance, T cell clones rendered anergic either by soluble peptide cultures or immobilized anti-CD3 mAb have been shown to actively suppress the response of other T cell clones to Ags in an MHC-linked fashion (13, 14, 15). The ability of some anergic T cells to actively inhibit the response of other naive T cells represents a novel form of immune regulation and represents a potentially important strategy for achieving immune tolerance. However, the precise nature of this immune regulation is unknown.

The effect of CTLA4Ig in large animals is not well studied (2). We have recently investigated the in vitro effects of CTLA4Ig in functional porcine T cell assays and demonstrated that CTLA4Ig was able to induce Ag-specific T cell hyporesponsiveness in secondary allogeneic pig mixed lymphocyte cultures (3). We refer to these cells as hyporesponsive T cells as opposed to anergic T cells, as there is controversy concerning whether anergy requires signaling through CTLA4 (17). In this report, we extend previous findings by showing that Ag-specific hyporesponsive T (ASHT)³⁴ (3) cells induced by CTLA4Ig in mixed lymphocyte cultures were able to actively inhibit the allogeneic responses of naive, but not primed, porcine T cells. This inhibition occurred in an MHC-linked manner, required cell-cell contact, and appeared to be IL-10 dependent. These results suggest that the profound immunosuppressive effects of CTLA4Ig on T cell alloresponses is mediated in part through the generation of regulatory T cells using IL-10-dependent mechanisms.

The CTLA4Ig used in these studies was human CTLA4IgG4 purified from plasmid-transformed NSO cells. CTLA4IgG4 is a fusion protein that combines the human T cell surface receptor CTLA4 with the constant region of human (h) IgG4. The extracellular domain of hCTLA4 was cloned as a fusion protein to hinge, CH2, and CH3 domains of a mutant hIgG4 deficient in Fc receptor binding (data not shown). This hIgG4 sequence contains L235G and G237A missense mutations. CTLA4IgG4 cross-reacts with the porcine B7 molecule as previously described (3). Isotype control Abs included h60.1, a humanized anti-CD11b IgG4 that binds human, but not porcine, CD11b.

The inbred miniature swine used in this study have been described in detail previously (18, 19). Presently, inbred swine of three homozygous MHC (swine leukocyte Ag (SLA) in swine) haplotypes are maintained: SLA^a (I^aII^a), SLA^c (I^cII^c), and SLA^d (I^dII^d). In addition, four intra-MHC recombinants of the SLA^g (I^cII^d), SLA^h (I^aII^d), SLA^j (I^aII^c), and SLA^k (I^dII^c) haplotypes have been derived by spontaneous recombination events during the breeding of heterozygotes as part of the breeding program. All animal care and procedures were in compliance with the Principles of Laboratory Animal Care formulated by the National Society for Medical Research and the Guide for the Care and Use of Laboratory Animals prepared by the Institute of Laboratory Animals Resources and published by the National Institutes of Health.

For preparation of pig PBMC from freshly collected, heparinized whole blood, the blood was diluted 2/3 with HBSS (Life Technologies, Grand Island, NY), and mononuclear cells were obtained by gradient centrifugation using lymphocyte separation medium (Organon Teknika, Durham, NC). The mononuclear cells were washed once with HBSS, and contaminating RBC were lysed with ACK lysing buffer (BioWhittaker, Walkersville, MD). Cells were washed again with HBSS and resuspended in complete tissue culture medium.

Responder PBMC (2–4 × 10⁵) and irradiated (2500 cGy) PBMC (2–4 × 10⁵) were added to 200 μl of complete tissue culture medium in U-bottom wells in triplicate. Complete tissue culture medium for MLR assays consisted of RPMI 1640 (Life Technologies) supplemented with 6% fetal porcine sera (Sigma, St. Louis, MO), 100 U/ml penicillin, 135 μg/ml streptomycin (Life Technologies), 50 μg/ml gentamicin (Life Technologies), 10 mM HEPES (Cellgro, Agawam, MA), 2 mM l-glutamine (Life Technologies), 1 mM sodium pyruvate (BioWhittaker), nonessential amino acids (BioWhittaker), and 5 × 10⁻⁵ M 2-ME (Sigma). The cultures were incubated at 37°C in humidified air containing 7% CO₂ for 1–5 days. [³H]Thymidine (1 μCi/well; New England Nuclear, Boston, MA) was added for a 5- to 6-h period at the end of the culture on appropriate days. The samples were harvested onto glass-fiber filter mats, and [³H]thymidine incorporation was measured by beta scintillation counting on a liquid scintillation counter. Results are expressed as the mean counts per minute with SE bars.

Secondary MLRs were performed in two culture phases, termed primary and secondary MLR cultures. In the primary MLR culture (priming phase, days 0–7), 4 × 10⁶ responder PBMC and 4 × 10⁶ irradiated (2500 cGy) stimulator PBMC were preincubated in 2 ml of MLR medium in 24-well plates (Costar, Cambridge, MA) with either CTLA4IgG4 (50 μg/ml) or control IgG4 (h60.1; 50 μg/ml) at 4°C for 30 min. Plates were then incubated for 7 days at 37°C in 7% CO₂. On day 7, cells were harvested, washed, and reconstituted in fresh MLR medium. Cells were then rested for 3 days at 4 × 10⁶ cells/ml in 25-cm² flasks at 37°C in 7% CO₂. The secondary MLR culture (days 10–14) was initiated on day 10 when the responder cells were collected, washed, and reconstituted in fresh MLR medium at appropriate concentrations. Responder cells (2–4 × 10⁵) were replated in 96-well U-bottom plates with the appropriate irradiated (2500 cGy) stimulators (2–4 × 10⁶) in the absence of Ab unless indicated in the figure. The secondary MLR cultures were incubated at 37°C in humidified air containing 7% CO₂ for 5 days, then assayed for [³H]thymidine incorporation as indicated above. In some assays mouse anti-swine IL-10 mAb (IgG1κ; 5–20 μg/ml; 945A 4C4 37B1, BioSource International, Camarillo, CA), rabbit polyclonal anti-swine IL-4 mAb (10 μg/ml; BioSource), swine IFN-γ (1000 U/ml; BioSource International), control mouse IgG1 (MOPC-1; Sigma), or recombinant human IL-2 (20 IU/ml; Cetus, Emeryville, CA) were added to secondary MLR cultures.

For coculture assays, unirradiated or irradiated Ag-specific hyporesponsive T cells induced by CTLA4Ig (2 × 10⁵) were added to naive or primed responder T cells (2 × 10⁵) in the same well with stimulator cells (2–4 × 10⁵) in 96-well U-bottom plates. For Transwell assays, naive or primed responder T cells (2 × 10⁵) and the appropriate stimulator (2 × 10⁵) were cocultured with ASHT cells (2 × 10⁵) and the appropriate stimulator (2 × 10⁵) was separated by a semipermeable membrane using Anopore eight-well Transwell inserts (0.2 μm; Nalge Nunc, Rochester, NY). Recombinant human IL-2 (100 IU/ml) alone was added to the top well for a positive control, while medium alone was used as a negative control. The cultures were incubated at 37°C in humidified air containing 7% CO₂ for 5 days. [³H]Thymidine incorporation was measured as described above. Results are expressed as the mean counts per minute with SE bars.

ELISA kits specific for swine IFN-γ and IL-10 were purchased from BioSource International. Supernatants harvested on day 4 of incubation were tested for IFN-γ and IL-10 according to the manufacturer’s instructions. Appropriate standard controls were tested, and linear regression analysis was performed. Supernatants and standards were tested in duplicate.

One peptide, 25 aa in length, spanning the polymorphic region of the α₁ domain (aa 3–27) of the pig class I gene PC1 was purchased from Quality Controlled Biochemicals (Hopkinton, MA). Peptides were synthesized based on previously published swine class I sequences (20). The amino acid sequence of the PC1 peptide used in these experiments (aa 3–27) was HSLRYFDTAVSRPDRRKPRFISVGY. Peptide purity was >90% as verified by HPLC and mass spectrometry.

Five hundred micrograms of the PC1 peptide (aa 3–27) in 0.25 ml of PBS was injected s.c. in the neck of an anesthetized SLA^dd pig in CFA (Sigma; 1/1, v/v).

Approximately 2 wk after immunization with the PC1 allogeneic peptide, PBMCs from the peptide-immunized pig were tested against the same PC1 allogeneic peptide (aa 3–27) in 96-well U-bottom plates at 37°C in 7% CO₂. Naive syngeneic SLA^dd (I^dII^d) or (SLA^cd)F₁ (I^cII^c × I^dII^d) nylon wool-adherent PBMCs, used as APCs (2–4 × 10⁵), were preincubated with 50 μg of the class I allopeptide for 2 h at 37°C in 7% CO₂. After incubation, peptide-loaded APCs were washed in fresh medium and added to the appropriate wells (2–4 × 10⁵). After 5 days of incubation, [³H]thymidine incorporation was measured as the mean counts per minute as described above.

Flow cytometry was analyzed using a Becton Dickinson FACScan microfluorometer (San Jose, CA). Swine IgG was used to block FcR binding for porcine cells. Apoptotic cells were detected using TdT- and FITC-labeled nucleotides (ApopTag; Intergen, Purchase, NY) according to the manufacturer’s instructions. Biotinylated mouse anti-class I^d (2.12.3) (21) was used to stain for SLA class I^d expression. PE-avidin (Becton Dickinson) was used as the secondary staining reagent. 74-12-4-PE or CyChrome Ab was used to stain swine CD4 cells, and 76-2-11-PE or FITC Ab was used to stain swine CD8 cells (22). Naive SLA^dd PBMCs, irradiated (600 cGy) SLA^dd PBMCs (6 h after irradiation), MLCs containing naive SLA^dd PBMC (4 × 10⁶) and irradiated (2500 cGy) stimulator SLA^cc PBMCs (4 × 10⁶), and cocultures containing CTLA4Ig-induced hyporesponsive SLA^dd PBMCs (4 × 10⁶) and irradiated (2500 cGy) stimulator SLA^cc PBMCs (4 × 10⁶) were analyzed for apoptosis. Cells taken from culture were analyzed 60 and 80 h after incubation at 37°C in 7% CO₂. For staining, 5–10 × 10⁵ cells/tube of porcine or human cells were resuspended in 100 μl of HBSS (Life Technologies) containing 0.1% BSA and 0.1% NaN₃ (FACS medium). Ten microliters of primary or isotype control Ab at ∼1 μg/1 × 10⁶ cells was added to the appropriate tubes for 30 min at 4°C. After two washes, a saturating concentration of secondary Ab was added and incubated for 30 min at 4°C. Cells were washed with FACS medium twice and then were analyzed by double-color flow cytometry.

We have previously shown that miniature swine T cells cultured with complete SLA-mismatched stimulators in the presence of CTLA4Ig were hyporesponsive to specific stimulators upon restimulation in the absence of CTLA4Ig, but maintained primary responses to third-party allogeneic stimulators (3). For clarity, we have labeled these CTLA4Ig-modulated T cells ASHT cells (3).

To further characterize ASHT cells that were generated after exposure to CTLA4Ig, secondary MLRs were performed (Fig. 1,A). SLA^dd (I^dII^d) T cells were incubated with fully allogeneic SLA^cc (I^cII^c) stimulators in the presence of CTLA4Ig or isotype control IgG4 during the primary MLR culture. After 7 days of incubation, responder cells were collected, washed, and rested in fresh medium. Subsequently, the responders were replated in secondary MLR cultures in the absence of any Ab. SLA^dd T cells incubated with allogeneic SLA^cc stimulators in the presence of CTLA4Ig were hyporesponsive to specific SLA^cc stimulators upon restimulation in the absence of CTLA4Ig compared with responders incubated originally with control IgG4 (Fig. 1 B). This inhibition was significant compared with the control IgG4 group; however, a residual proliferative response could be detected on days 2 and 3 of the secondary MLR. This residual response could not be eliminated in multiple assays, even with higher titrations of CTLA4Ig, suggesting that either CD28-negative populations of T cells were responding or alternative pathways of costimulation were being used.

When the ASHT cells were restimulated with (SLA^ac)F₁ (I^aII^a ×I^cII^c) stimulators, no primary proliferative response was observed on days 4 and 5 despite a vigorous primary response to SLA^aa (I^aII^a) and (SLA^ad)F₁ (I^aII^a × I^dII^d) stimulators on days 4 and 5 (Fig. 1,A). In fact, the kinetics and magnitude of proliferation to (SLA^ac)F₁ stimulators mirrored that observed against SLA^cc stimulators, suggesting that the presence of the SLA^c Ag on the APC prevented a primary response from T cells that would otherwise react to the SLA^a Ag (linked Ag). This inhibition was not observed when a 1/1 mixture of SLA^aa and SLA^cc stimulators (unlinked Ag) was used (Fig. 1 A). The proliferative curve against the 1/1 mixture occurred earlier than that observed with SLA^aa stimulators. Although the reason for this change in kinetics was unclear, the result was observed consistently in four independent assays. Perhaps, the residual proliferative response of the ASHT cells to SLA^cc stimulators generated early production of IL-2 and facilitated an accelerated response of naive cells to SLA^aa stimulators through the action of secreted cytokines.

In contrast, SLA^dd T cells incubated with specific SLA^cc stimulators in the presence of control IgG4 during the primary MLC showed brisk and vigorous secondary responses on days 2 and 3 to SLA^cc, (SLA^ac)F₁, and a 1/1 mixture of SLA^aa and SLA^cc stimulators (Fig. 1,B). The primary response of the control group to third-party SLA^aa stimulators peaked on day 4 (Fig. 1 B). Of note, the lack of a primary response to SLA^ac cells was not due to the immunodominance of SLA^c Ags on the (SLA^ac)F₁ cells, as SLA^dd responders primed with (SLA^ac)F₁ cells in vitro responded to both SLA^aa cells and SLA^cc cells upon restimulation (data not shown).

One trivial interpretation of these data is that the response of the bulk culture to third-party SLA^aa stimulators was not due to the proliferation of naive anti-SLA^aa T cell clones present in the bulk culture, but, rather, was due to cross-reactive responses of the ASHT cells to SLA^aa stimulators. This interpretation could explain why the response to SLA^ac resembled that to SLA^cc. To address this concern, we cocultured SLA^dd ASHT cells with naive SLA^dd T cells and tested the allogeneic responses in a primary MLR. Fig. 2,A shows that the addition of SLA^dd ASHT cells (A-DD, Fig. 2,A) in a 1/1 mixture with naive SLA^dd T cells led to inhibition of the anti-SLA^cc response of naive T cells, as no proliferation was observed on days 4 and 5, whereas a primary response was observed with naive T cells cultured without ASHT cells. SLA^dd ASHT cells also inhibited naive SLA^dd T cell responses to (SLA^ac)F₁ stimulators, which link the specific SLA^c Ags to third-party SLA^a Ags on the same cell (Fig. 2,B). In contrast, naive responses to a 1/1 mixture of SLA^aa and SLA^cc stimulators were not inhibited in coculture assays with SLA^dd ASHT cells (Fig. 2,B), although the presence of ASHT cells did accelerate the kinetics of response to the 1/1 mixture of SLA^aa and SLA^cc stimulators as shown in Fig. 1,A. The precise reason for this acceleration was not apparent; perhaps, Ag-experienced cells from the original culture secreted low levels of IL-2 early in the culture period. Allogeneic inhibition was observed even when the number of specific stimulators was doubled during the coculture assays (CC-4, Fig. 2,A), arguing against Ag competition as a mechanism of inhibition. Naive T cells cultured without ASHT cells responded normally to (SLA^ac)F₁, stimulators with peak proliferation on day 5 (Fig. 2,B). Naive T cells cultured with or without ASHT cells did not respond to third-party SLA^aa stimulators alone (Fig. 2,B). The early proliferation in response to SLA^cc stimulators in cocultures of ASHT cells with naive T cells (Fig. 2,A) observed on days 2 and 3 was probably due to the residual response of the ASHT cells, consistent with the data in Fig. 1 A. Thus, ASHT cells were able to regulate the response of naive T cells to both specific and third-party MHC Ags provided that the third-party Ags were expressed on the same cell as the original Ag, i.e., linked regulation.

We analyzed the phenotype of the ASHT cells after 7 days of incubation with fully allogeneic stimulators and CTLA4Ig. Double staining for CD4 and CD8 (22) revealed that 44.1% of the population were CD8⁺ T cells, 3.3% were CD4⁺ CD8^low T cells, and 49.7% were double-positive T cells (Fig. 3,A). The simultaneous expression of CD4 and CD8 Ags by a large population of resting peripheral T lymphocytes has been well documented in swine (6–60% in swine vs 0.5–8% in humans) (23, 24). The porcine CD8⁺CD4⁺ T cell subset contains both mature resting T cells (25) and Ag-activated memory Th cells (26). Furthermore, after Ag exposure, single-positive CD4⁺ T cells in swine may acquire a double-positive CD4⁺CD8⁺ phenotype (27). Consistent with the literature, we found a 6-fold increase (6.8–38.4%) in the double-positive population after Ag exposure (Fig. 3, C and B). The presence of a significant population of CD8⁺CD4⁺ T cells in swine and the dynamic nature of the expression of CD4 and CD8 after Ag exposure made definitive T cell subset phenotyping of ASHT cells difficult to interpret.

To define the specificity of ASHT cells, SLA^dd responders were cultured with either class I-disparate SLA^gg (I^cII^d) or class II-disparate SLA^kk (I^dII^c) stimulators in the presence of CTLA4Ig during the primary MLR culture. After 3 days of rest, the responder cells were cocultured with naive SLA^dd cells and restimulated in a secondary MLR culture without Ab using allogeneic stimulators that were linked with the original Ag or a third-party Ag. Fig. 4 A demonstrates that ASHT cells were generated after incubation of SLA^dd cells with class I-disparate SLA^gg stimulators in the presence of CTLA4Ig during the primary culture. The proliferative response of cocultured naive SLA^dd cells against (SLA^ag)F₁, (I^aII^a) × (I^cII^d), stimulators, which linked the original class I^c Ag to third-party SLA^aa Ags, was markedly inhibited. However, the response to (SLA^ak)F₁, (I^aII^a) × (I^dII^c) stimulators, which presented a new class II^c Ag, was comparable to that observed with third-party SLA^aa and SLA^kk cells.

Similar results were observed when SLA^dd cells were incubated with class II-disparate SLA^kk (I^dII^c) stimulators in the presence of CTLA4Ig during the primary MLR culture (Fig. 4 B). The primary response of cocultured naive SLA^dd cells against (SLA^ak)F₁ and SLA^kk stimulators, but not (SLA^ag)F₁ stimulators, was inhibited by the SLA^dd ASHT cells. The primary response to third-party SLA^aa cells remained intact. As the specificity of CD8⁺ and CD4⁺ cells is primarily restricted to class I and class II Ags, respectively, these data suggest that CTLA4Ig can generate CD8⁺ and CD4⁺ (or CD4⁺CD8⁺) ASHT cells that mediate the linked regulation of naive T cell alloresponses. Furthermore, the finding that the regulation was MHC class I or class II specific (a specific characteristic of T cells) supports our hypothesis that T cells were mediating suppression. Flow cytometric analysis revealed that the majority of ASHT cells were CD25 negative (>90%) and CD154 negative (>99%; data not shown), suggesting that CTLA4Ig did prevent Ag activation.

We next cocultured naive T cells with irradiated (250 cGy) SLA^dd ASHT cells and tested them against specific SLA^cc stimulators. Fig. 2 A demonstrates that irradiation of the SLA^dd ASHT cells before coculture with naive SLA^dd T cells completely eliminated the inhibition of naive T cell responses to specific stimulators. Of note, irradiation of the ASHT cells also eliminated the early proliferation on days 2 and 3, confirming the idea that the residual response of the ASHT cells to SLA^cc was responsible for the early proliferation in the cocultures. The inhibition of naive allogeneic responses was probably not due to overcrowding, as the cell counts of naive cells cocultured with ASHT and stimulator cells on day 4 were comparable with those of naive cells cultured with stimulators alone (data not shown).

To determine whether this regulatory phenomenon was contingent on cell-cell contact, coculture assays with Transwells were performed. SLA^dd ASHT cells and SLA^cc stimulators were cultured in the upper wells, while naive SLA^dd T cells were cultured with SLA^cc stimulators in the lower wells. Fig. 5 demonstrates that naive SLA^dd T cells were able to respond to SLA^cc stimulators when the SLA^dd ASHT cells were separated from responders by a semipermeable membrane. In contrast, when the ASHT cells were cocultured with naive SLA^dd T cells in the absence of a membrane, full inhibition of the primary allogeneic MLR was observed (Fig. 5). In addition, no inhibition was observed when naive SLA^dd T cells were tested against third-party SLA^aa stimulators in the lower wells. For a positive control, exogenous IL-2 (100 IU/ml) was added to the upper well. This led to maximal proliferation to SLA^cc stimulators on day 5 (Fig. 5). Thus, the regulation of naive T cells by ASHT cells required cell-cell contact.

Next, we investigated whether T cells rendered hyporesponsive by CTLA4Ig could inhibit the response of in vitro and in vivo primed T cells. When SLA^dd T cells that were primed in vitro against SLA^cc stimulators were cocultured with ASHT cells in a secondary MLR, the response against specific SLA^cc stimulators at the same or double the responder concentration (CC-2 or CC-4) was not inhibited (Fig. 6,A). Instead, the Ag-activated T cells displayed primed MLR responses against specific allogeneic stimulators, with brisk and robust responses on days 2 and 3 after restimulation (Fig. 6 A). These results suggested that, unlike naive T cells, directly primed T cells were resistant to the regulation mediated by CTLA4Ig-induced ASHT cells.

To examine whether ASHT cells could regulate the alloresponses of T cell primed in vivo through the indirect pathway of allorecognition, we isolated PBMCs from an SLA^dd pig immunized in vivo to an allogeneic SLA class I^c PC1 peptide (aa 3–27), which spanned the polymorphic region of the α₁ domain. These in vivo peptide-immunized SLA^dd T cells were cocultured in a peptide proliferation assay with SLA^dd ASHT cells (SLA^dd hyporesponsive to SLA^cc) and (SLA^cd)F₁ stimulators loaded with PC1 allospecific peptide. Peptide-loaded (SLA^cd)F₁ stimulators presented the specific class I^c Ag linked with the allogeneic peptide presented in the context of syngeneic class II^d molecules. Fig. 6,B shows that peptide-immunized SLA^dd responders demonstrated vigorous proliferative responses to the specific allopeptide presented by syngeneic SLA^dd cells, confirming sensitization to the allogeneic peptide (Fig. 6,B, left panel). When, SLA^dd ASHT cells were added to the coculture, the primary response to SLA^cc stimulators by the peptide-immunized SLA^dd responders was inhibited (Fig. 6,B, center panel). However, the response to (SLA^cd)F₁ stimulators loaded with specific class I peptide remained intact (Fig. 6,B, center panel). Unimmunized, naive SLA^dd responders failed to respond to peptide-loaded syngeneic APCs, as expected (Fig. 6 B, right panel). Interestingly, the response of the peptide-immunized SLA^dd cells to (SLA^cd)F₁ stimulators without loaded peptide was higher than the response directed against SLA^cc stimulators, but less than the response directed against peptide-loaded (SLA^cd)F₁ cells. This suggests that (SLA^cd)F₁ cells naturally present class I^c peptides in association with class II^d molecules, which is consistent with previous studies (28). Thus, ASHT cells were not able to inhibit the allogeneic responses of T cells primed in vivo through the indirect pathway of Ag presentation, although it is not known whether this failure of inhibition was due to the fact that the response was a primed one or an indirect one.

The absence of a secondary proliferative response to SLA^cc or SLA^ac stimulators by SLA^dd ASHT cells could be due to deletion of responding cells, which would not be expected to be sensitive to exogenous IL-2. To test this hypothesis, IL-2 was added to appropriate wells during the secondary MLR. The addition of exogenous IL-2 restored proliferative responses of SLA^dd ASHT cells to specific SLA^cc stimulators, with peak proliferation 4 days postrestimulation (Fig. 7). Exogenous IL-2 also restored the primary response of the bulk culture to (SLA^ac)F₁ stimulators, which presented a linked Ag. In the absence of IL-2, specific hyporesponsiveness to SLA^cc stimulators was observed (Fig. 7). The kinetics of the proliferative response appeared to be accelerated by 1 day compared with the normal naive response (cf Figs. 1,A and 7). The reason for this acceleration was not apparent. Possibly, the ASHT cells were more responsive to exogenous IL-2 because of previous exposure to allogeneic stimulators. In any case, the ability of IL-2 to restore the proliferative response to specific and linked stimulator argues against a deletional mechanism of T cell regulation. Furthermore, regulation mediated by the ASHT cells was not simply due to consumption of nutrients and IL-2 within the medium during the early response of the ASHT cells (Fig. 1 A), as supernatants from MLCs were harvested on day 4 and used to replate primary allogeneic MLRs. Vigorous primary allogeneic proliferative responses were observed after plating MLRs with the culture supernatants (data not shown).

The in vivo administration of CTLA4Ig has been shown to suppress Th₁ responses but spare Th₂ responses (7). To ascertain the phenotype of CTLA4Ig-induced, ASHT cells, we examined the cytokine profile of the ASHT T cells in thepresence of specific SLA^cc stimulators. Fig. 8,A demonstrates that ASHT T cells secreted sustained low levels of IL-10 (15–25 pg/ml) throughout the culture period, with peak production 2 days after restimulation. The level of IL-10 production was approximately one-tenth of the amount secreted by primed T cells in response to specific SLA^cc stimulator (Fig. 8,B). In contrast, ASHT cells produced only minimal amounts of IFN-γ throughout the culture period (Fig. 8,C), while control primed responders generated high levels of IFN-γ for the duration of the culture period (Fig. 8 D). This cytokine profile was consistent with a Th2 or possibly a Th3 phenotype (29). Further analysis of the cytokine profile of the ASHT cells was precluded by the lack of available swine-specific reagents.

Since ASHT cells secreted IL-10, but not IFN-γ, neutralization and supplementation MLRs were performed to further characterize the nature of the T cell regulation mediated by these cells. Fig. 9,A demonstrates that the addition of anti-swine IL-10 mAb (5 μg/ml) to cocultures containing ASHT cells and naive SLA^dd T cells restored the primary allogeneic response to SLA^ac stimulators. The primary response to third-party SLA^aa stimulators was also maintained. In contrast, anti-swine IL-4 Ab (10 μg/ml), exogenous swine IFN-γ (1000 U/ml), and control IgG had no effect on reversing the inhibition (Fig. 9, B–D). Thus, the immunoregulatory effects mediated by the ASHT cells appeared to be IL-10 dependent.

Since IL-10 leads to inactivation of APCs with a reduction in the expression of class II molecules and costimulatory ligands (CD80, CD86) (30), we examined the expression of both B7 and class II on APCs after coculture with ASHT and naive T cells. No significant differences in the level of B7 or class II expression were detected at 12, 24, and 36 h of coculture (data not shown). Thus, the IL-10-dependent immune regulation mediated by CTLA4Ig-induced ASHT cells did not appear to be related to the down-regulation of class II or CD80/CD86 by APCs.

The effects of anti-IL-10 mAb might be contributing to the direct reversal of hyporesponsiveness, which would, in effect, lead to the reversal of suppression indirectly. To distinguish whether anti-IL-10 mAb was directly reversing the suppression mediated by the regulatory cells or actually reversing the hyporesponsiveness of the ASHT cells, neutralization MLRs were performed using ASHT cells for responders and specific SLA^cc or linked SLA^ac stimulators. If anti-IL-10 mAb were reversing hyporesponsiveness directly, then a primary MLR should be detected against both specific and linked stimulators. If, however, anti-IL-10 mAb were reversing the suppression, then a primary MLR should be observed for only the linked stimulator. Fig. 10 demonstrates the addition of anti-IL-10 mAb restored the proliferative response to linked SLA^ac stimulators, but not to specific SLA^cc stimulators, while the response to third-party SLA^aa stimulators remained intact. Thus, anti-IL-10 was effecting the reversal of suppression and not hyporesponsiveness.

Another possible hypothesis to explain the mechanism of immune regulation mediated by ASHT cells is the deletion of naive cells by either the APC or the regulatory T cells themselves. This would be similar to a veto-like mechanism found in mice (31, 32). To address this hypothesis, we examined whether the level of apoptosis in responder class I^d+ cells after coculture with ASHT cells was increased relative to that in cultures with naive responders and stimulator cells alone. Two-color flow cytometric analysis was performed using a mouse mAb specific for class I^d (responder haplotype; 2.12.3) (21) and TdT- and FITC-labeled nucleotides, which indicate DNA fragmentation, the hallmark of apoptosis. Fig. 11 demonstrates that the percentage of responder class I^d+ cells that stained brightly for dNTP-FITC after coculture of naive SLA^dd cells with SLA^dd ASHT cells and irradiated SLA^cc stimulators was ∼3% at 60 h of culture (Fig. 10,D) and ∼10% at 84 h of culture (Fig. 11,F). These levels were actually lower than those observed after culturing naive SLA^dd cells with irradiated SLA^cc stimulators alone (8.5% at 60 h (Fig. 11,C) and 30.6% at 84 h (Fig. 11,E)). Irradiated (600 cGy) SLA^dd cells were analyzed after 6 h of incubation for a positive apoptotic control (Fig. 11,B), while naive SLA^dd cells served as a negative control (Fig. 11 A). The failure to observe a significant population of apoptotic class I^d-negative cells (i.e., SLA^cc irradiated stimulators) in the cultures was probably due to complete cell death and fragmentation by 60 and 84 h of incubation. Flow cytometric analysis revealed very few class I^c+ cells (<7%) remaining as early as 12 h postculture using irradiated (2500 cGy) SLA^cc cells (data not shown). Finally, SLA^dd ASHT cells did not lyse specific SLA^cc or syngeneic SLA^dd targets in chromium release assays (data not shown), arguing against a direct cytotoxic mechanism of regulation.

Our results demonstrate that blockade of the B7-CD28 pathway by CTLA4Ig gives rise to hyporesponsive T cells, which can actively inhibit the allogeneic responses of naive, but not primed, T cells to specific alloantigen or to third-party alloantigens expressed on the same APC as the specific Ag (i.e., linked regulation). We have referred to these CTLA4Ig-induced regulatory cells as hyporesponsive T cells as opposed to anergic T cells as there is controversy concerning whether anergy requires signaling through CTLA4 (17). The specificity of the CTLA4Ig-induced regulatory cells could be either class I or class II Ags, suggesting that both CD8⁺ and CD4⁺ cells functioned as regulatory cells. The regulation mediated by these cells was not due to mere consumption of IL-2 or medium nutrients, but did require cell-cell contact, as demonstrated by the Transwell assays. In addition, this regulation did not appear to involve a deletional or cytotoxic mechanism. Rather, the regulatory phenomenon was mediated by an IL-10-dependent pathway, since neutralization of IL-10 with anti-IL-10 mAb reversed the inhibition. Of note, the regulatory mechanism did not depend on the actions of the inhibitory cytokine IL-4, which is consistent with a previous report demonstrating that the immunoregulatory effects of CTLA4Ig are not IL-4 dependent in vivo (33).

The immunoregulatory effects of IL-10 are well documented (30). IL-10 directly inhibits T cell proliferation and cytokine production in response to Ag (34) and indirectly inhibits T cell function through its effects on the APC (30). Regulatory CD4⁺ T cells cells are characterized by the production of high levels of IL-10, but no IL-4 (29), and IL-10 not only leads to the generation of regulatory T cells, but also serves as a mechanism through which regulatory T cells exert their effect (29). Recently, a surface-bound form of IL-10 was described on human PBMCs (35). The existence of a surface form of IL-10 might explain our seemingly paradoxical finding that anti-IL-10 mAb could reverse the inhibition mediated by CTLA4Ig-induced regulatory cells, yet cell-cell contact was required for the regulatory cells to effect inhibition. Perhaps, the regulatory T cells interacted with APCs and delivered a negative signal through surface-bound IL-10, which led to inactivation of the APC and prevention of a primary response by a naive T cell interacting with that APC. One would expect the effects of this regulation to be very local, since the surface-bound IL-10 would be required to interact with a surface receptor. An alternative interpretation is that the regulatory T cells inactivated naive T cells after direct cell-cell contact with the naive T cell through a veto-like mechanism. While this is formally possible, we think that this is unlikely, as a 1/1 mixture of SLA^aa and SLA^cc stimulators did not lead to suppression. In bulk culture it seems statistically unlikely that a T cell interacting with SLA^aa APC would not also be in contact with a T cell that is recognizing SLA^cc APC. Furthermore, no evidence for a deletional or cytotoxic mechanism was found.

To our knowledge, this represents the first report describing the ability of blocking of the B7-CD28 pathway alone with CTLA4Ig to mediate linked regulation of allogeneic T cell responses. Early models of anergic T cells with regulatory properties were based on the use of immobilized anti-CD3 Ab, soluble peptides, or T cell presentation of Ag and were not dependent upon inhibitory cytokines, such as IL-10 (13, 14, 15, 16). Although there are some similarities between these models of immune regulation and our own, our model appears to be distinct because 1) primed cells were resistant to regulation, 2) bulk populations of cells were used to generate regulatory cells and not Ag-specific T cell clones, and 3) linked regulation was dependent upon the actions of IL-10 (13, 14, 15, 16). Blockade of the CD40-CD154 pathway alone has been shown to induce linked suppression to MHC-matched, minor Ag-disparate skin grafts in mice, but anti-CD8 mAb was also required for graft prolongation (36). Very recently, blockade of both CD40 and CD86 pathways was shown to induce regulatory human T cells; however, the degree of inhibition was only partial (up to 60% maximal) compared with the level of inhibition in our studies (37). In the human study regulation required the use of both anti-CD40 and anti-CD86 mAbs (CTLA4Ig was not addressed), while our study showed complete suppression with the use of CTLA4Ig alone. IL-10 also appeared to play a role in the mechanism of regulation in the human study; however, neutralization of IL-10 only partially restored the allogeneic response in that study. These differences suggest that the characteristics of CTLA4Ig-induced regulatory cells and the nature of their regulation are probably distinct from those exhibited by the anergic regulatory T cells previously described.

Evidence for the in vivo generation of a regulatory T cells by CTLA4Ig has been suggested in several rodent models of allotransplantation. The adoptive transfer of CTLA4Ig-treated CD4⁺ cells along with naive T cells led to donor-specific tolerance of mouse allogeneic islets (38). Rat cardiac allograft recipients treated with CTLA4Ig and donor-specific transfusion led to indefinite graft survival in 50% of recipients (39). Furthermore, the transfer of cells from the CTLA4Ig-treated rats led to infectious tolerance in naive hosts (39). The ability of costimulatory blockade with CTLA4Ig to mediate immune regulation in a linked fashion is reminiscent of previous reports demonstrating linked suppression through the use of nondepleting CD4 Abs in vivo (40, 41). Indeed, linked suppression mediated by regulatory T cells may be a general mechanism of immune regulation that contributes to various models of peripheral tolerance (42).

Our model could also explain the observation that immune modulation by CTLA4Ig often does not require continued administration or persistent circulating Ab, since Ag-specific hyporesponsive T cells induced by CTLA4Ig in our system were able to regulate the responses of naive T cells even after cessation of therapy. In light of the recent phase I clinical trial of CTLA4Ig in psoriasis vulgaris patients (11) and the use of CTLA4Ig to prevent graft-vs-host disease in bone marrow transplant recipients (43), the ability of CTLA4Ig-induced regulatory cells to suppress the naive T cell response could have significant implications for clinical transplantation and therapies to treat autoimmunity (43).

We thank Drs. Hugh Auchincloss, Jr., Henry Winn, and Kathryn J. Wood for their suggestions and critical review of this manuscript and Laurie Niederer for preparation of this manuscript.