We have previously shown that systemic staphylococcal enterotoxin A (SEA) injections cause CD4 T cells in TCR-transgenic mice to become tolerant to subsequent ex vivo restimulation. An active IFN-γ-dependent mechanism of suppression was responsible for the apparent unresponsiveness of the CD4 T cells. In this study, we analyze the response of CD4 T cells isolated throughout the first 10 days of the in vivo response to injected SEA. We show that CD4 T cells isolated at the peak of the in vivo response undergo very little activation-induced cell death after sterile FACS sorting or restimulation in the presence of neutralizing Abs to IFN-γ. We also show that the IFN-γ-dependent tolerance develops soon after SEA injection in the spleens of both normal and TCR-transgenic mice. This suppression is dependent upon myeloid cells from the SEA-treated mice and is optimal when inducible NO synthase activity and reactive oxygen intermediates are both present. The data indicate that IFN-γ, myeloid cells, and a combination of NO and reactive oxygen intermediates all contribute to a common pathway of T cell death that targets activated or responding CD4 T cells. Sorted Gr-1+ cells from SEA-treated mice also directly suppress the response of naive CD4 T cells in mixed cultures, indicating that this tolerance mechanism may play a role in down-regulating other vigorous immune responses.
Superantigens (SAg)3 stimulate strong proliferative responses by both CD4 and CD8 T cells through specific interactions with the Vβ-chains of their TCRs (1). Although these proteins are potent T cell mitogens, they often cause shock and immune dysfunction in vivo, which directs the immune system away from protective immunity (2).
Injected SAg stimulate a brief period of immune hyperactivity in vivo, which lasts about 3 or 4 days (3, 4, 5). During this time, the numbers of responding T cells increase substantially in the spleens and lymph nodes (LN) of the staphylococcal enterotoxin A (SEA)-treated animals, and cytokines are released in high concentrations (3, 4, 5). However, the response peaks about 3 days after the SAg is injected and then declines abruptly, leaving diminished populations of SAg-reactive CD4 T cells in the periphery of the SAg-treated animals. In some studies, the residual SAg-reactive T cells were hyporesponsive to restimulation (6). Secondary T cell responses to injected SAg are also very transient and even more rapidly aborted than the primary response (7). Although T cell responses to injected SAg have been extensively studied, the mechanisms that are responsible for the acute loss of responding T cells in vivo and the subsequent immune unresponsiveness in vitro remain controversial.
Several mechanisms that can induce CD4 T cell death or unresponsiveness have been identified. These mechanisms include T cell anergy due to a lack of costimulation at the time of activation (8) and Fas-mediated activation-induced cell death (AICD) (9). Cytokines such as IFN-γ (10, 11), TNF-α (12), and IL-10 (13) can also mediate suppression of T cells. However, the detailed mechanisms involved in the induction and execution of these cytokine-regulated death pathways have not been clarified.
Bacterial SAg are an accepted model to analyze peripheral T cell tolerance. Because these proteins bind to class II MHC molecules and stimulate T cells through their Ag receptors, it is believed that they stimulate T cells in much the same way as conventional peptide Ags. In a previous study (11), we showed that injected SEA induced an active mechanism of T cell suppression to develop in the spleens of AND TCR-transgenic (Tg) mice (14). This suppression inhibited the proliferative responses and cytokine production of Vβ3+ CD4 T cells after in vitro restimulation. Although increased numbers of dead Vβ3+ CD4 T cells were found in the suppressed cultures, Fas expression was not required. The T cell death was prevented when CD4 T cells from the SEA-treated mice were highly purified or IFN-γ was neutralized in the cultures, indicating that exogenous factors were involved in the suppression. Since IFN-γ was not sufficient to suppress the response of the CD4 T cells by itself, the data suggested that other cells or factors from the SEA-treated mice were also required. This study demonstrated that an extrinsic mechanism of suppression was responsible for inhibiting the responses of the SEA-reactive CD4 T cells in vitro, but the precise mechanism of T cell death was not further analyzed.
Other evidence suggests that IFN-γ may play a significant role in peripheral T cell tolerance. This evidence comes from both in vitro (10) and in vivo studies. Some studies suggest that IFN-γ may play a protective role during the active phase of experimental autoimmune encephalomyelitis (15) and prevent the accumulation of activated T cells (16) by a mechanism that may involve neutrophils (17). IFN-γ and NO production also prevent activated CD4 T cells from accumulating in mice infected with bacillus Calmette-Guérin (BCG) bacteria (18) or lymphocytic choriomeningitis virus (19). In other studies, reactive oxygen intermediates (ROI) were suggested to play a role in regulating T cell survival in SAg-treated mice (20, 21). Together, these reports suggest that reactive molecules, which can be produced by myeloid cells, may play an important role in down-regulating T cell responses in vivo. However, the cells and molecules responsible for inducing T cell death have not been completely elucidated in these studies.
In this study, we have analyzed the response of Vβ3+ CD4 T cells harvested throughout the first 10 days of the in vivo response to injected SEA. We show that Vβ3+ CD4 T cells isolated at the peak of the in vivo response undergo very little AICD after purification by sterile FACS sorting or restimulation in the presence of neutralizing Abs to IFN-γ. We found that myeloid cells, IFN-γ, NO, and ROI all contribute to a single novel pathway of T cell suppression. This suppressive pathway is responsible for the apparent unresponsiveness of the SAg-reactive CD4 T cells in partially purified cultures. We show that this suppressive mechanism specifically targets activated or responding CD4 T cells, and we present evidence that this pathway is a physiological mechanism of T cell regulation. We discuss the potential role of this death mechanism in down-regulating other T cell responses in immunized animals.
Materials and Methods
Normal B10.BR mice were obtained from The Jackson Laboratory (Bar Harbor, ME). C57BL/6 (B6) × SJL founder Tg AND mice (14) were backcrossed to B10.BR mice for at least 10 generations. Mice between 3 and 7 mo of age received purified SEA (25 μg for AND mice and 150 μg for B10.BR mice) in HBSS by i.v. injection in the tail.
Cell lines and reagents
Murine fibroblasts transfected with I-Ek class II MHC molecules and ICAM-1 (DCEK-ICAM) (22) and expressing high levels of B7.1 (23) were treated with mitomycin C (50 μg/ml) and used as APCs for in vitro experiments. NG-monomethyl l-arginine (l-NMMA), NG-monomethyl d-arginine (d-NMMA), N-acetylcysteine (NAC), reduced glutathione (GSH), and propidium iodide (PI) were all purchased from Sigma (St. Louis, MO). RB6-8C5, XMG1.2, and the other depletion Abs (11) were purified from murine ascites or cell culture supernatants. 5-(and -6)-Carboxyfluorescein diacetate succinimidyl ester (CFSE) was purchased from Molecular Probes (Eugene, OR). SEA was purchased from Toxin Technologies (Sarasota, FL).
Enrichment of CD4 T cells and in vitro restimulation
CD4 T cells were enriched with Abs and complement (C′), using anti-HSA (J11D), anti-class II MHC (M5114 and CA4), and anti-CD8 (3.155) Abs as previously described (11). Cell debris was removed with Lympholyte separation gradients, spun at 1,900 × g for 20 min at room temperature. These enriched CD4 T cell populations were cultured in RPMI 1640 medium supplemented with 200 μg/ml penicillin, 200 μg/ml streptomycin, 4 mM l-glutamine, 50 μM 2-ME, and 10% FCS as previously described (24). Tg CD4 T cells were cultured at 3 × 105/ml with 1 × 105 APC/ml, 5 μM pigeon cytochrome c fragment (PCCF) peptides, and 20 U/ml rIL-2. Non-Tg T cells were cultured at 106/ml with SEA (500 ng/ml), mitomycin C-treated APC (105/ml), and rIL-2 (20 U/ml). Purified anti-IFN-γ Abs (XMG1.2) and l-NMMA (50 μM) were added to the cultures at the time of restimulation. NAC and GSH were made up fresh before each experiment and added 16 h after in vitro restimulation.
Measuring CD4 T cell expansion
For proliferation studies, CD4 T cells were purified from the spleens of HBSS- or SEA-treated mice by Abs and C′ depletion and stimulated as described above. Three days after in vitro restimulation, the numbers of live Vβ3+ CD4 T cells were calculated by the counting trypan blue-excluding cells and correcting for the percentage of Vβ3+ CD4 T cells using FACS analysis. PI was added to stained CD4 T cells immediately before FACS analysis to quantify dead Vβ3+ CD4 T cells.
CFSE analysis to identify proliferating and anergic CD4 T cells
Enriched CD4 T cells were labeled with 1 μM CFSE in PBS at 37°C for 15 min (25). The cells were then washed with cold PBS and restimulated in vitro as described above. After in vitro restimulation the intensity of the CFSE dye was analyzed by three-color FACS analysis, using CyChrome-conjugated Abs to CD4 (BD PharMingen, San Diego, CA) and PE-conjugated Abs to Vβ3 (BD PharMingen). Rates of cell division were assessed by the intensity of the CFSE fluorescence, analyzed at 491–518 mM.
An IFN-γ-dependent mechanism of T cell suppression develops after SEA injection
We have previously shown that an active mechanism of T cell suppression prevented Vβ3+ CD4 T cells from SEA-treated Tg TCR mice from proliferating or producing cytokines in response to antigenic restimulation in vitro (11). Here, we have used ex vivo analyses to further investigate the mechanism of this SEA-induced T cell suppression, using both Tg TCR and normal B10.BR mice. In these studies, we analyze the proliferative responses of Vβ3+ CD4 T cells as a representative population of SEA-reactive CD4 T cells. The cultures are restimulated in vitro with SEA or PCCF peptides and subsequently examined for live and dead Vβ3+ CD4 T cells. Neutralizing Abs are also used to confirm the role of IFN-γ in the suppression.
B10.BR mice were used to determine when the IFN-γ-dependent suppression developed in vivo. The mice were given 150 μg of SEA (or HBSS alone) by i.v. injection. Duplicate animals were euthanized at 2, 3, 4, and 7 days after immunization, and the spleens were analyzed for live Vβ3+ CD4 T cells, Fig. 1 a (line graph). This study shows the kinetics of the in vivo response to injected SEA. As in other studies, the numbers of responding Vβ3+ CD4 T cells increased dramatically between 2 and 3 days after SEA injection. However, on the fourth day the numbers of T cells began to decline substantially, leaving relatively few live Vβ3+ CD4 T cells in the spleens of the SEA-treated animals by day 7. In other studies, we have used an adoptive transfer system to show that when naive Vβ3+ CD4 T cells from AND TCR Tg mice (9) are introduced into B10.BR hosts, virtually all of the transferred CD4 T cells undergo rapid cell division after systemic SEA injection (data not shown).
The remaining spleen tissue from the SEA-treated animals was used to analyze the IFN-γ-dependent T cell suppression. For this analysis, CD4 T cells were enriched with Abs and C′ to deplete CD8 T cells and B cells as previously described (11). The recovered cells were between 69 and 77% CD4 T cells. The others were a mixed population of residual B cells and some nonlymphoid cells. These enriched cell populations were then stimulated in vitro with I-Ek-bearing murine fibroblasts and SEA. Because naive CD4 T cells die if cultured in the absence of added cytokines (25, 26), we also used rIL-2 to keep the SEA-nonreactive CD4 T cells alive during the 3-day culture period. Half the cultures also received the neutralizing Abs to IFN-γ (XMG1.2). Three days later, the cultures were analyzed for live Vβ3+CD4 T cells by counting the trypan blue-excluding cells and FACS analysis (Fig. 1,a, bar graph). The results are expressed as the fold increase in the numbers of live Vβ3+ CD4 T cells during the 3-day culture period. The cultures were also analyzed for the number of dead Vβ3+ CD4 T cells by PI staining and FACS analysis (Fig. 1 b).
After in vitro restimulation, reduced numbers of live Vβ3+ CD4 T cells were detected in all the cultures from the SEA-treated mice compared with the numbers in the cultures from HBSS-treated control animals (Fig. 1,a, bar graph). PI analysis also revealed large numbers of dead Vβ3+ CD4 T cells in the suppressed cultures (Fig. 1,b). This demonstrated that the injected SEA had induced a form of T cell suppression in all the SAg-treated animals. In every case, the suppression was blocked when neutralizing Abs to IFN-γ were used (Fig. 1,a), and the numbers of dead Vβ3+ CD4 T cells were substantially reduced (Fig. 1 b). The control cultures (day 0) were not suppressed after HBSS injection and were unaffected by the neutralizing Abs to IFN-γ.
In this experiment, the numbers of Vβ3+ CD4 T cells increased 4- to 5-fold in the SEA-treated mice and then returned to numbers approximating the starting population by day 7. However, when CD4 T cells were enriched and restimulated in vitro, all of the Vβ3+ CD4 T cells exhibited similar proliferative capacity throughout the response. Even the cells that were harvested at the peak of the in vivo response (day 3) underwent very little AICD in the presence of the neutralizing Abs. This suggests that the mechanism that regulates normal T cell deletion in vivo was blocked when the Vβ3+ CD4 T cells were removed from the animals or by neutralizing Abs to IFN-γ and further suggests that fraternal Fas-Fas ligand interactions were not sufficient to induce a majority of the T cell death. The enriched CD4 T cells from SEA-treated B10.BR mice were also able to suppress the responses of fresh naive CD4 T cells from untreated animals (data not shown), indicating that the mechanism of suppression was similar to that originally described in TCR Tg mice (11).
A role for NO in SAg-induced in vitro suppression
NO has been shown to inhibit T cell responses in several different models (27). Since inducible NO synthase (iNOS) activity is regulated by IFN-γ (28), we used a chemical inhibitor of iNOS (l-NMMA) (29) to investigate whether reactive nitrogen intermediates (RNI) played a role in the IFN-γ-dependent T cell suppression (Fig. 1 c).
Enriched CD4 T cells from the same SEA-treated mice that are shown in Fig. 1, a and b, were restimulated in vitro with SEA and APC as before. In this case, replicate cultures received the iNOS inhibitor l-NMMA (50 μM) or its inactive enantomer d-NMMA (50 μM). Three days later, the cultures were analyzed for live (Fig. 1 c) and dead (data not shown) Vβ3+ CD4 T cells as previously described.
The suppression was unaffected by the inactive enantomer d-NMMA. However, l-NMMA significantly increased the numbers of live Vβ3+CD4 T cells in many of the cultures from the SEA-treated mice (Fig. 1,c) and reduced the numbers of dead cells (data not shown). This effect was particularly pronounced in the cultures that were isolated on days 4 and 7 after SEA treatment, indicating that NO was contributing to the suppression. However, l-NMMA had relatively little effect in cultures isolated on days 2 and 3. We have consistently found that CD4 T cells from SEA-treated mice do not proliferate to the same extent in the presence of l-NMMA (Fig. 1,c) as identical cells cultured in the presence of neutralizing Abs to IFN-γ (Fig. 1 a). This disparity was particularly pronounced in the cultures that were harvested on days 2 and 3 after SEA injection. Other experiments, in which T cells were restimulated at lower concentrations, have given similar results (data not shown), indicating that the activity of the iNOS inhibitor was not saturated. Since we have not found 50 μM l-NMMA to be toxic to proliferating T cells, we considered the possibility that additional factors were contributing to the IFN-γ-dependent suppression.
Gr-1+ cells are required for the SEA-induced in vitro suppression
Our previous study indicated that nonlymphoid cells from SEA-treated mice were contributing to the SEA-induced suppression (11). To identify these nonlymphoid cells, we used a panel of mAbs to deplete different subsets of cells from the restimulated cultures. In separate experiments, Abs were used to deplete NK cells (DX5), γδ T cells (GL3), and Gr-1+ cells (RB6-8C5) from the enriched cultures. The efficiencies of the depletions were confirmed by FACS staining. In these studies, only the anti-Gr-1 Abs reversed the suppression (Fig. 2), and the other Abs had no effect (data not shown).
In this experiment, CD4 T cells were harvested from SEA-treated AND mice at the peak of the in vivo response (day 3). The CD4 T cells were enriched with Abs and C′ to depleted CD8 T cells and B cells as before. However, during depletion each sample was divided into two equal aliquots, and one half received an additional Ab (RB6-8C5) to deplete Gr-1+ cells. After purification, the enriched CD4 T cells were restimulated in triplicate 1-ml cultures using 5 μM PCCF, APC, and rIL-2. Replicate cultures received 1) no further additions, 2) Abs to IFN-γ, 3) 50 μM l-NMMA, or 4) 50 μM d-NMMA. Three days later, the cultures were analyzed for live (Fig. 2) and dead (data not shown) Vβ3+ CD4 T cells as before.
When only CD8 T cells and B cells were depleted from the cultures, the SEA-induced T cell suppression was detected as before. The numbers of live Vβ3+ CD4 T cells in the SEA-treated cultures were also substantially increased when neutralizing Abs to IFN-γ were used (Fig. 3,b) and were partially restored in the cultures that received l-NMMA (Fig. 3,c). Strikingly, the suppression was completely eliminated when RB6-8C5 Abs were used to deplete Gr-1+ cells from the cultures before restimulation, and there was no added benefit from adding Abs to IFN-γ or iNOS inhibitors to the cultures (Fig. 2, b and c). As before d-NMMA had no effect on the suppression in any of the cultures (Fig. 3,d). The control CD4 T cells that were purified from HBSS-treated mice by the same protocols were not suppressed after enrichment and were largely unaffected by the presence of neutralizing Abs to IFN-γ, l-NMMA, or the RB6-8C5 purification protocol (Fig. 2, left).
Reciprocal results were obtained by PI analysis, with large numbers of dead Vβ3+ CD4 T cells in the suppressed cultures and relatively few dead cells when neutralizing Abs to IFN-γ or Gr-1+ cells were used. The dead cells accounted for only 5–10% of the recovered Vβ3+ CD4 T cells in these cultures (data not shown). These data indicated that IFN-γ and Gr-1+ cells were part of a common pathway of T cell suppression. Other experiments with non-Tg mice and CD4 T cells isolated up to 10 days after SEA injection have given similar results (data not shown).
In this experiment, the Gr-1+ cells were ∼10–12% of the restimulated cell populations from SEA-treated mice and were <3–4% of the control cultures. Because the CD4 T cells and Gr-1+ cells were purified together, they were present in the restimulated cultures at similar ratios to the numbers present in the original SEA-treated spleens. These numbers were sufficient to completely suppress the response of the CD4 T cells from all SEA-treated mice. As before, l-NMMA was less effective in reducing the rate of T cell death in the cultures from SEA-treated mice than were neutralizing Abs to IFN-γ, further suggesting that there may be another component to the suppression.
ROI play a role in the SAg-induced in vitro suppression
Blocking iNOS activity only partially prevented IFN-γ-dependent suppression in the cultures from SEA-treated mice (Fig. 1,c). This incomplete blocking was particularly apparent in the cultures that were isolated on days 2 and 3 after SEA treatment, suggesting that other factors were contributing to the suppression of CD4 T cells. Murine neutrophils and related bone marrow precursors express Gr-1 at high levels. Lower levels of Gr-1 can also be found on some monocytes and macrophages (30). All of these cells produce a combination of RNI (NO) and ROI (superoxide and H2O2) in response to IFN-γ (31, 32). Since RNI and ROI are often produced together and have both been implicated in causing the death of CD4 T cells in other models (21, 33, 34), we investigated whether the response of the SEA-treated CD4 T cells was further restored when ROI-scavenging antioxidants (NAC and GSH) were used in combination with l-NMMA (Fig. 4). For these experiments, CD4 T cells were again harvested at the peak of the in vivo response (3 days after SEA injection).
CD4 T cells were enriched from the spleens of B10.BR mice and restimulated in vitro as described in Fig. 1 (Fig. 3,a). Some of the cultures received 50 μM l-NMMA at the time of restimulation. Sixteen hours later, varying concentrations of NAC or GSH were also titrated into the cultures. The live Vβ3+ CD4 T cells were then counted 3 days after in vitro restimulation, and the results are expressed as the fold increase in the number of live Vβ3+ CD4 T cells during the 3-day culture period, as before (Fig. 3 a).
As in previous experiments, after 3 days of in vitro restimulation very reduced numbers of live Vβ3+ CD4 T cells were recovered from the cultures from SEA-treated mice compared with the control cultures. However, these numbers increased 5-fold when the cells were restimulated in the presence of 50 μM l-NMMA. Significant numbers of live Vβ3+ CD4 T cells were also detected in the cultures that were treated with NAC or GSH, resulting in a 6-fold expansion of the Vβ3+ CD4 T cell population at 5 mM GSH or NAC. The expansion was further enhanced when 50 μM l-NMMA was used in combination with one of the antioxidants, resulting in a maximum 9-fold expansion of the Vβ3+ CD4 T cell population at 5 mM antioxidant. Both antioxidants appeared to show toxicity at the 10-mM concentration for 3 days. These results demonstrated significant additive protection when l-NMMA was used in combination with NAC or GSH, suggesting that RNI and ROI were both contributing to the SEA-induced T cell suppression.
In a second set of experiments, CD4 T cells were harvested from SEA- and HBSS-treated AND mice. The CD4 T cells were enriched with Abs and C′ and restimulated in vitro with a specific peptide Ag (PCCF), APC, and rIL-2, as described in Fig. 2. At 16 h after restimulation, varying concentrations of NAC were added to some of the cultures. In this experiment, PI analysis was used to quantify the percentages of live and dead CD4 T cells in the cultures at 40 h after in vitro restimulation (Fig. 3 b). The results show the percentages of PI+ CD4 T cells averaged from three replicate animals.
The PI analysis revealed large numbers of dead (PI+) CD4 T cells in the cultures from SEA-treated mice, when no antioxidants were added. In these cultures the dead cells were ∼65% of the recovered CD4 T cells after 40 h of in vitro restimulation. However, when NAC was added to the cultures, the percentages of PI+ CD4 T cells were substantially reduced (down to 22%). The numbers of PI+ CD4 T cells decreased as the dose of NAC increased. This confirmed the role of ROI in SEA-induced T cell death in cultures isolated 3 days after SAg injection. NAC had only a slight enhancing effect on the percentage of live CD4 T cells in the control cultures and showed no evidence of toxicity during the 40-h incubation period.
These experiments showed additive effects from blocking NO and ROI production, suggesting that a combination of RNI and ROI was responsible for mediating the death of CD4 T cells in the suppressed cultures. We have found some experimental variability in the relative contributions of NO and ROI to the IFN-γ-dependent suppression between individual animals.
Purified Thy-1−Gr-1+ cells from SEA-treated mice kill responding Vβ3+ CD4 T cells in reconstituted cultures
To definitively establish the role of the Gr-1+ cells in SEA-induced T cell suppression (Fig. 2), we have used sterile FACS sorting to isolate Thy-1−Gr-1+ cells from SEA-treated AND mice. The purified Thy-1−Gr-1+ cells were then mixed with either highly purified CD4 T cells from SEA-treated mice or fresh naive CD4 T cells from untreated Tg TCR mice. Neutralizing Abs to IFN-γ were also added to some of the cultures to measure suppression. These experiments showed that the Thy-1−Gr-1+ cells from SEA-treated mice were able to suppress the proliferative response of the Vβ3+ CD4 T cells in reconstituted cultures when IFN-γ was present (Fig. 4).
As before, CD4 T cells were enriched from the spleens of SEA-treated AND mice at 3 days after injection. In addition, highly purified populations of Thy-1−Gr-1+ cells and Vβ3+ CD4 T cells were isolated from the same mouse spleens by sterile FACS sorting. Abs to Thy-1 were used to exclude cross-reactive Ly6C+ T cells from the sorted cell populations, since they do not contribute to the suppression (data not shown). A third population of naive CD4 T cells was isolated from untreated control AND mice with Abs and C′. Each of the purified CD4 T cell populations was restimulated in vitro with APC, rIL-2, and PCCF. The purified Thy-1−Gr-1+ cells from SEA-treated mice were then titrated into some of the cultures. Finally, neutralizing Abs to IFN-γ were used to measure suppression. After 3 days of in vitro restimulation the live (Fig. 4,a) and dead (Fig. 4 b) CD4 T cells were analyzed as before.
As expected, the enriched CD4 T cells from SEA-treated mice did not proliferate unless neutralizing Abs to IFN-γ were used (Fig. 4,a). However, when the CD4 T cells were isolated by sterile FACS sorting, they responded vigorously to antigenic restimulation in vitro (Fig. 4 a) and were not affected by the neutralizing Abs to IFN-γ. The suppression reappeared when sorted Gr-1+ cells were titrated back into the cultures and was inhibited by the neutralizing Abs to IFN-γ. Naive CD4 T cells from treated mice were also suppressed in the presence of the sorted Gr-1+ cells.
These experiments showed that a ratio of 1:8 Gr-1+ cells from the SEA-treated mice was sufficient to completely inhibit the proliferative response of the sorted CD4 T cells from SEA-treated mice. Slightly larger numbers of Gr-1+ cells were required to suppress the response of the naive CD4 T cells, suggesting that recently activated CD4 T may be slightly less sensitive to suppression. This may be because naive CD4 T cells make very little IFN-γ in response to a primary stimulation. Small numbers of Thy-1−Gr-1+ cells were also purified from the spleens of the HBSS-treated control mice. However, titration experiments showed that much larger numbers of these cells (a 1:2 ratio) were required to suppress the proliferative response of the Vβ3+ CD4 T cells in mixed cultures (data not shown).
Again, PI analysis revealed a reciprocal pattern of results. In general, large populations of dead Vβ3+ CD4 T cells were recovered from all of the suppressed cultures from the SEA-treated mice (Fig. 4 b). The numbers of dead cells were substantially reduced when the CD4 T cells were highly purified by sterile FACS sorting or when neutralizing Abs to IFN-γ were used. In these rescued cultures, only 5–10% of the Vβ3+ CD4 T cells underwent AICD after in vitro restimulation.
The Thy-1−Gr-1+ cells were further analyzed by three-color FACS analysis. This analysis showed two populations of Thy-1−Gr-1+ cells in the spleens of the SEA-treated mice (Fig. 5,a), which expressed high and low levels of Gr-1. Both populations of Gr-1+ cells also expressed high levels of Mac-1 and LFA-1 (Fig. 5,b). Hematoxylin- and eosin-stained cytospins were used to further analyze the sorted Gr-1+ cells by standard differential analysis. This analysis showed a mixed population of mature neutrophils with highly segmented nuclei (70–80%) and smaller numbers of macrophage/monocytes (20–30%). Although a very small number of contaminating lymphocytes and some other unidentified cells were also found (<5%), no eosinophils or basophils were detected. The Gr-1+ cells from control mice included some Mac-1-negative cells (Fig. 5 b) and some immature neutrophils with banded nuclei (data not shown). This suggests that some of the Gr-1+ cells from control animals may have a less activated phenotype than the Gr-1+ cells from SEA-treated animals.
IFN-γ-dependent suppression develops in the spleens, but not the peripheral LN, of SEA-treated animals
Two groups of four AND mice were given 25 μg of SEA or HBSS alone by i.v. injection. Five days later, the spleens from each group of four animals were pooled and enriched for CD4 T cells with Abs and C′ as described in Fig. 2. The peripheral LN from each group of four animals were also pooled and enriched for CD4 T cells by the same protocol. After purification the CD4 T cells were restimulated in vitro with 5 μM PCCF, APC, and IL-2 as before. Neutralizing Ab to IFN-γ were also added to half the cultures. Four days later, the cultures were analyzed for live CD4 T cells by counting and FACS analysis (Fig. 6). This analysis showed the IFN-γ-dependent suppression only in the spleen cultures from the SEA-treated mice. The LN cultures and cultures from control animals were not suppressed.
The mechanism of CD4 T cell suppression: only responding CD4 T cells are targets
We further investigated the mechanism of SEA-induced in vitro suppression by analyzing the kinetics of the proliferative response of the SEA-reactive Vβ3+ CD4 T cells to in vitro restimulation. In this case, the live CD4 T cells were counted on consecutive days after restimulation to measure the size of the proliferating T cell population (Fig. 7,a). CFSE was used to follow the kinetics of T cell division in the cultures (Fig. 7, b–e). B10.BR mice were used for this analysis to compare the response of the SEA-reactive Vβ3+ CD4 T cells with the response of CD4 T cells that express SEA-nonreactive Vβ-chains in the same cultures. Recombinant IL-2 was used to keep the nonresponding T cells alive throughout the 4-day culture period.
CD4 T cells were enriched from the spleens of SEA-treated (150 μg) and HBSS-treated B10.BR mice 5 days after injection. Each sample was divided into two equal aliquots during enrichment, and one half received the additional Abs to Gr-1. After enrichment, the CD4 T cells were labeled with CFSE dye and restimulated in vitro with SEA, APC, and rIL-2. Neutralizing Abs to IFN-γ were added to half the cultures to produce suppressed and nonsuppressed cultures. On consecutive days after in vitro restimulation the cultures were analyzed for live Vβ3+ CD4 T cells (Fig. 7,a). The CD4 T cells were also analyzed for the intensity of their CFSE fluorescence, which was reduced by half with each cell division. The overlaid histograms show gated populations of Vβ3+ CD4 T cells from cultures stimulated in the presence (heavy lined open histograms) and absence (shaded histograms) of neutralizing Abs to IFN-γ (Fig. 7, b and c).
When Gr-1+ cells and IFN-γ were both present in the cultures, the Vβ3+ CD4 T cell populations from SEA-treated mice did not expand (Fig. 7,a). However, many more live Vβ3+ CD4 T cells were recovered when Gr-1+ cells or IFN-γ were removed. In these cultures T cell response peaked between 2 and 4 days after restimulation, and there were no appreciable additive effects of combining the Gr-1 depletion protocol with blocking Abs to IFN-γ. The cultures from the control mice did not begin to expand until 3 days after stimulation (Fig. 7 a) and were not significantly affected by the presence of Gr-1 cells or IFN-γ. Since the Vβ3+CD4 T cells from SEA-treated mice expanded with slightly accelerated kinetics in the rescued cultures compared with the cultures from control animals, these data suggest that some cells had been preactivated by their in vivo exposure to the injected SEA.
The CFSE analysis showed dividing Vβ3+ CD4 T cells in the control cultures on days 2, 3, and 4 after stimulation (Fig. 7, b and c). This rate of cell division was not influenced by the Abs to IFN-γ or treatment to remove Gr-1+ cells. In contrast, very few dividing Vβ3+ CD4 T cells were recovered from the cultures from SEA-treated mice unless neutralizing Abs to IFN-γ were used (Fig. 7,d) or Gr-1+ cells were removed (Fig. 7,e). The Vβ3+ CD4 T cells proliferated extensively in these cultures and were not further enhanced when Abs to IFN-γ and Gr-1+ cell depletion were used together (Fig. 7 e).
The few live CD4 T cells that were recovered from the suppressed cultures appeared to initiate cell division on day 2, but no further divisions were detected on the subsequent days of the analysis (Fig. 7 d). These data strongly suggest that responding Vβ3+ CD4 T cells were deleted from suppressed cultures as they started dividing on day 2. Very few live Vβ3+ CD4 T cells remained in the suppressed cultures by day 4. Some residual populations of naive CD4 T cells that did not express Vβ3 also remained in all the cultures throughout the 3-day restimulation period. The size of this unstimulated cell population remained relatively unchanged throughout the analysis in both the SEA-treated and control cultures (data not shown). Together these data suggest that naive or resting CD4 T cells are resistant to the deletion and that only activated or responding CD4 T cells are targeted by the IFN-γ-dependent death mechanism. Additional experiments were used to further address this possibility.
Activated or responding CD4 T cells are the targets of the IFN-γ-dependent suppression
To determine when CD4 T cells became susceptible to the IFN-γ-dependent death mechanism, we analyzed the surviving CD4 T cells from the suppressed and rescued cultures for the expression of T cell activation Ags, using three-color FACS analysis. B10.BR mice were used for this analysis to compare the response of SEA-reactive Vβ3+ CD4 T cells with the response of CD4 T cells that express SEA-nonreactive Vβ-chains in the same cultures. Total populations of live CD4 T cells are shown (Figs. 8, a–d). The live Vβ3+ CD4 T cells from the cultures that were rescued with neutralizing Abs to IFN-γ are also shown as a separate analysis (Fig. 8, e and f). There were insufficient numbers of live Vβ3+ CD4 T cells in the suppressed cultures for extensive phenotypic analysis.
FACS analysis showed the presence of large numbers of live CD4 T cells with increased forward scatter in all cultures that received neutralizing Abs to IFN-γ, indicating the presence of enlarged CD4 T cell blasts (Fig. 8,a). In contrast, relatively few live CD4 T cells were recovered from the suppressed cultures, and residual cells were low in forward scatter, indicating that they were primarily unstimulated CD4 T cells. Further analysis showed that the enlarged cells in the rescued cultures were also primarily Vβ3+ CD4 T cells (Fig. 8,b), which expressed elevated levels of CD44 (Fig. 8,c) and reduced levels of CD62L (Fig. 8,d), indicating that they were Ag-experienced T cells. The Vβ3+ T cells were virtually all activated CD4 T cells with high levels of CD44 (99%) and reduced CD62L expression (81%), suggesting that they were also activated T cells (Fig. 8, e and f). The residual CD4 T cells in the suppressed cultures did not express Vβ3, indicating that they were primarily SEA-nonreactive T cells. These cells were naive in phenotype, with low CD44 and high CD62L expression. Small populations of nonlymphoid cells were also recovered from all the cultures, which are not shown in this analysis.
These data suggest that only CD4 T cells that did not express SEA-reactive TCRs, and were thus unable to interact with the SAg, survived in the suppressed cultures when IFN-γ was present, whereas the responding CD4 T cells, including the Vβ3+ cells, were deleted. In light of these data, we conclude that responding or activated CD4 T cells were specifically targeted by the death/deletion mechanism in the suppressed cultures.
We have further analyzed the mechanisms that cause CD4 T cells to become tolerant to antigenic restimulation after SEA injection. We find that when Tg TCR or normal B10.BR mice are given tolerizing doses of SEA, an active mechanism of T cell suppression develops in the spleens of the SAg-treated animals that induces CD4 T cell death. Gr-1+ myeloid cells, ROI, and RNI are all important mediators of this tolerance mechanism. Our data suggest that in the presence of IFN-γ the myeloid cells suppress the proliferative response of the SEA-reactive CD4 T cells by triggering cell death as the CD4 T cells initiate cell division. Purified Gr-1+ cells from the SEA-treated animals are also able to suppress the proliferative response of fresh naive CD4 T cells from untreated animals when stimulated with PCCF peptides. This suggests that once the conditions for the IFN-γ-dependent suppression have been established in vivo, SAg may not be directly required to induce T cell death in vitro. We therefore suggest that this mechanism of suppression may contribute to the down-regulation of T cell responses to other immune stimuli.
We have used cell purification and Ab-mediated depletion protocols to identify the cells responsible for the IFN-γ-dependent suppression. We show that as an alternative to removing IFN-γ, the SEA-induced suppression can be prevented by depleting Gr-1+ myeloid cells from the cultures before restimulation (Figs. 3 and 5). In contrast, depleting NK cells or γδ T cells did not prevent the suppression (data not shown). Since Abs to IFN-γ did not further enhance the survival of the CD4 T cells in the restimulated cultures once the Gr-1+ cells had been removed, our data suggest that IFN-γ induces suppression by regulating the activity of the Gr-1+ cells.
In most experiments, Gr-1+ cells accounted for about 10–15% of the spleen cells from SEA-treated mice and <3–4% in control animals. In several experiments, the suppressive cells and SEA-reactive Vβ3+ CD4 T cells were purified together ( Figs. 1–3). Since all of the cultures were profoundly suppressed after restimulation, these data demonstrate that there were sufficient numbers of Gr-1+ cells in the spleens of the SEA-treated mice to completely suppress the proliferative response of Vβ3+ CD4 T cells at physiological ratios. In other experiments, Thy-1−Gr-1+ cells were isolated from the spleens of SEA-treated mice by sterile FACS sorting and titrated into mixed cultures of CD4 T cells. These experiments confirmed that as little as 5–10% Gr-1+ cells (i.e., a ratio of 1 Gr-1+ cell to 10 T cells) were sufficient to suppress the proliferative response of fresh naive CD4 T cells from untreated animals as well as sorted Vβ3+ CD4 T cells isolated from SEA-treated mice (Fig. 4). Although small numbers of Gr-1+ cells were also isolated from untreated control animals, much higher ratios of these cells (i.e., 1:2) were required to induce equivalent suppression in the mixed cultures (data not shown).
Cytospin analysis shows that 70–80% of the sorted Gr-1+ cells from SEA-treated mice are mature neutrophils, with highly segmented nuclei (Fig. 5 a) and about 20–30% are monocyte/macrophages. In other experiments, we have found that purified neutrophils (98% pure) are able to suppress T cell responses in the presence of IFN-γ by a mechanism that can be inhibited with l-NMMA (L.S. Cauley and S L. Swain, unpublished observations). As in other studies, we also find that purified macrophages are able to suppress T cell responses; however, in the SAg model, subfractionation experiments indicate that neutrophils and macrophages both contribute to the suppression (data not shown). Both groups of Gr-1+ cells from the SEA-treated mice also expressed high levels of Mac-1 and LFA-1, which is characteristic of highly activated granulocytes (30). In contrast, the Gr-1+ cells recovered from control animals included some immature neutrophils with banded nuclei (data not shown) and some Mac-1-negative cells, suggesting that the granulocytes from the control animals were less activated.
Our studies indicate that LN cultures are not affected by the IFN-γ-dependent mechanism of T cell suppression after SEA injection. Others have shown that neutrophils rapidly down-regulate L-selectin (CD62L) upon activation (35), which may explain why much larger numbers of these cells are found in the spleens of the SEA-treated mice than in their peripheral LNs (data not shown). We also usually find slightly larger numbers of neutrophils in the spleens of the Tg AND mice than in the spleens of normal animals (data not shown), which may explain why smaller doses of SEA are required to induced the IFN-γ-dependent suppression in AND mice than in normal animals. Although neutrophils are short lived cells that turn over very rapidly in vivo, their half-life can be significantly prolonged by several cytokines, including IFN-γ released from T cells stimulated with SAg (36). Activated neutrophils also express a variety of chemokines that are known to be chemotactic to activated T cells, including the IFN-γ (MIG), IFN-inducible T cell α chemoattractant (I-TAC), and IFN-γ-inducible protein-10 (IP-10) (37), which are likely to bring the neutrophils into close proximity with activated T cells within the SAg-treated animals.
Neutrophils and monocytes release a variety of metabolic products during the oxidative burst that is stimulated by IFN-γ, including superoxide, H2O2, and NO (31, 32). We find that the iNOS inhibitor (l-NMMA) and the two ROI-scavenging antioxidants (NAC and GSH) substantially reduce the IFN-γ-dependent suppression ( Figs. 1–3) and show substantial additive effects in many of the cultures. These data indicate that NO and ROI both contribute to the SEA-induced suppression. However, the antioxidants were most effective in cultures isolated at the peak of the in vivo response (Fig. 3). Others have shown that superoxide is produced only transiently after exposure to receptor-dependent agents, such as IFN-γ (32), whereas iNOS will continue to produce NO until the substrate has been consumed or the enzyme itself is degraded. The different reaction kinetics of these two enzymes may explain why NO appears to play a greater role in SEA-induced T cell suppression in cultures isolated >4 days after SEA injection (Fig. 2).
Neither NO nor superoxide behaves as a strong oxidant toward most types of organic compounds. However, they react rapidly with one another to generate a variety of more potent oxidants, including peroxynitrite (ONOO), hydroxyl anions (·OH), and, in a reaction catalyzed by myeloperoxidase, hypochlorous acid (HOCl) (reviewed in Refs. 38 and 39). These potent oxidants can cause tissue damage through a variety of mechanisms, including interference with cellular respiration and direct DNA damage. DNA damage, in turn, stimulates repair activity by poly(ADP-ribose) synthase and depletes cellular energy supplies (40, 41). Peroxynitrite can also modify tyrosine residues, interfering with cell functions such as cell cycle progression (42) and inactivating enzymes such as superoxide dismutase (SOD) (43) and promoting apoptosis in activated T lymphocytes (34). NAC reacts slowly with superoxide and H2O2, but is an excellent scavenger of hydroxyl radicals and hypochlorous acid (44). Since we found significant additive effects of inhibiting both NO and ROI, it is possible that peroxynitrite contributes to the SEA-induced suppression.
Many previous studies have shown a dramatic decline in the numbers of responding T cells during the in vivo response to injected SAg (Fig. 1). Some studies suggested that Fas-mediated AICD may be largely responsible for the loss of responding T cells in mice primed with staphylococcal enterotoxin B (SEB) (45). However, others have found that the Fas mutation in lpr mice is not sufficient to prevent T cell deletion in SEB-treated mice (46, 47, 48) (L. S. Cauley and S. L. Swain, unpublished observations). One study went on to show that Tg Bcl-2 also had little effect on the T cell response to SEB in vivo. However, when the Tg Bcl-2 mice were cross-bred with lpr mice, significant accumulation of Vβ8+ CD4 T cells was detected 7 days after SEB injection (48). Although it is likely that injected SAg stimulates some Fas-mediated AICD in normal animals (45), recent evidence suggests that Fas-mediated AICD in lymphocytes is not sensitive to inhibition by Bcl-2 (49). Together, these data suggest that two or more independently regulated mechanisms with some redundant function may be responsible for removing activated CD4 T cells from the circulation after in vivo SAg treatment. It is likely that at least one of these mechanisms may be sensitive to inhibition by Bcl-2 (50). Some evidence suggests that NO-induced apoptosis in T cells may be sensitive to inhibition by Bcl-2 (51). Bcl-2 has also been shown to influence the levels of intracellular antioxidants, including SOD and GSH, and thus inhibit apoptosis in cells exposed to H2O2 (52, 53).
Further studies are needed to determine whether the IFN-γ-dependent mechanism of T cell suppression described in this study contributes to the deletion of responding CD4 T cells during the in vivo response to injected SAg. However, circumstantial evidence suggests that related mechanisms may be involved. In particular, CD4 T cells harvested from the SEA-treated animals at the peak of the in vivo response (day 3) undergo very little AICD in the presence of neutralizing Abs to IFN-γ (Figs. 1 and 4) or after sterile FACS sorting (Fig. 4). This indicates that the mechanism that would normally have deleted many of these SEA-reactive CD4 T cells in vivo had they remained in the SAg-treated animals was blocked because the cells were removed from the animal or by the Abs to IFN-γ. Either alternative suggests that there is an active component to the mechanism that regulates T cell deletion in vivo. The suppressed cultures also contain large numbers of dead Vβ3+ CD4 T cells, indicating that T cell death is the cause of the in vitro suppression.
Other studies also suggest that cytokine-regulated mechanisms may contribute to the loss of responding T cells in SAg-treated mice. In particular, LPS treatment has been shown to inhibit the loss of responding CD4 T cells in mice given relatively low doses of injected SEA (54). The mechanism of this LPS-mediated rescue is not yet clear; however, some evidence suggests that TNF-α may play a role (55). More recently, a mimetic of SOD was used to show that low doses of injected SEA can induce a superoxide-mediated mechanism of cell death in T cells from SAg-treated mice (21). This study did not show in vivo data; however, an earlier study showed that NAC slightly enhanced T cell survival in mice stimulated with a mouse mammary tumor virus-derived SAg (20). The origin of the ROI was not investigated in either of these studies. A potential link between the ROI-mediated cell death and the enhancing effects of LPS on T cell survival in SAg-treated mice may lie in the ability of TNF-α to up-regulate expression of the SOD enzyme (56). Alternatively, TNF-α may induce apoptosis in neutrophils (57), preventing them from releasing ROI and NO in response to IFN-γ. Priming with LPS has also been shown to inhibit IFN-γ production in SEB-treated mice (58), which could reduce NO and ROI production by both macrophages and neutrophils.
Superantigens are produced by a wide variety of bacteria and viruses (59). In many respects, T cell responses to SAg are similar to the response to conventional Ags, and it is likely that both responses are regulated by similar mechanisms. IFN-γ, Gr-1+ cells, and RNI have all been suggested to a play role in the effector phase of an experimental autoimmune encephalomyelitis model (15, 16, 17). Large numbers of activated CD4 T cells have also been shown to accumulate in IFN-γ knockout mice after BCG infection, and CD4 T cells from BCG-infected wild-type mice underwent apoptosis by a mechanism that required IFN-γ and NO (18). Another study has reported T cell hyperproliferation in IFN-γ knockout mice infected with lymphocytic choriomeningitis virus and partial resistance to cell death (19). These studies support the suggestion that IFN-γ may play an important role in controlling T cell responses in vivo through mechanisms similar to those described here and suggest that these mechanisms may play a role in preventing autoimmunity.
A model that may explain our results is that in addition to activating large numbers T cells in vivo and stimulating the release of large quantities of cytokines (1, 2, 3, 4, 5, 6), injected SEA may also promote activated neutrophils and macrophages to colocalize with the responding T cells in an inflammatory site in the spleens of the SAg-treated animals. We postulate that once the conditions for suppression have been established in vivo, subsequent antigenic restimulation in vitro promotes further IFN-γ production, stimulating small numbers of Gr-1+ cells in the enriched T cell cultures to produce a combination of RNI and ROI. Our data suggest that RNI and ROI directly cause activated or responding CD4 T cells to undergo apoptosis, as has been suggested in other models (20, 21, 33, 34). This possibility will be further investigated in future studies. IFN-γ is produced by a variety of cells in vivo, including CD8 T cells and NK cells. It is likely that macrophages and/or responding CD4 T cells produce IFN-γ in the restimulated cultures.
In conclusion, we find that the CD4 T cells that persist in vivo after systemic SEA injection are not intrinsically unresponsive to antigenic stimulation, but become targets of an active mechanism of suppression upon in vitro restimulation. This suppression is IFN-γ dependent, requires a mixed population of Gr-1+ myeloid cells, and targets activated or responding CD4 T cells for death in the presence of NO and ROI. We suggest that the mechanism described here may constitute a general form of peripheral down-regulation of intense immune responses that would otherwise lead to massive CD4 expansion.
We thank Dr. R. Dutton for suggesting experiments that contributed to the completion of this manuscript, and Simon Monard for help with sterile FACS experiments.
This work was supported by National Institutes of Health Grants AI26887 and AI4666.
Abbreviations used in this paper: SAg, superantigen; AICD, activation-induced cell death; BCG, bacillus Calmette-Guérin; CFSE, 5-(and -6)-carboxyfluorescein diacetate succinimidyl ester; iNOS, inducible NO synthase; LN, lymph node; NAC, N-acetylcysteine; l-NMMA, NG-monomethyl l-arginine; d-NMMA, NG-monomethyl d-arginine; PCCF, pigeon cytochrome c fragment; PI, propidium iodide; RNI, reactive nitrogen intermediate; ROI, reactive oxygen intermediate; SEA, staphylococcal enterotoxin A; SEB, staphylococcal enterotoxin B; SOD, superoxide dismutase; Tg, transgenic.