Interleukin-15 redirects the outcome of a tolerizing T-cell stimulus from apoptosis to anergy

Interleukin (IL)-15 is a cytokine that shares many biologic activities with IL-2. IL-15 induces proliferation of activated T lymphocytes and natural killer (NK) cells and release of interferon (IFN)-γ, stimulates cytolytic activity of cytotoxic T lymphocytes (CTL) and NK cells, and costimulates immunoglobulin synthesis by B cells.1-4 These redundant functions of IL-2 and IL-15 are an obvious consequence of their common usage of the IL-2/IL-15 β and γc receptor subunits for binding and initiation of intracellular signaling.5 However, both cytokines use a specific high-affinity binding α chain to compose their heterotrimeric receptor complex in T and NK cells.6 Furthermore, IL-2 and IL-15 also differ in their cellular sources of production and in the way their expression and secretion are controlled. IL-2 production occurs exclusively in activated T cells and expression is mainly controlled by transcriptional regulation and messenger RNA (mRNA) stabilization. IL-15 mRNA is widely distributed in many different cell types and tissues, and its expression is tightly regulated at the level of transcription, translation, and secretion.7 Due to these differences, IL-15 exerts several unique functions in lymphoid and nonlymphoid tissues, not shared by IL-2.

In T lymphocytes, IL-15 and IL-2 have different roles in the regulation of apoptosis. Although both cytokines protect resting T cells from growth factor deprivation-induced cell death,8 they exert opposite activities during T-cell receptor (TCR)-induced cell death. Engagement of TCR during an antigenic response results in activation of the T cell, progression into the cell cycle, and production of IL-2, promoting further expansion. Re-engagement of TCR during clonal expansion activates the Fas/Fas ligand (FasL) apoptotic pathway and results in deletion of the activated T lymphocyte.9 This TCR-induced cell death not only restores cellular homeostasis at the conclusion of an immune response but also preserves peripheral tolerance. Lenardo10 originally demonstrated in vitro that the presence of IL-2 is essential for this apoptotic response of activated T cells to TCR re-engagement. Experiments with IL-2, IL-2Rα, and IL-2Rβ knockout mice further confirmed the essential role of IL-2 in peripheral T-cell homeostasis and tolerance; mice lacking the IL-2,11,IL-2Rα,12or IL-2Rβ13 gene suffered from severe lymphoproliferative disease and autoimmune disorders, explained by impaired deletion of excessive and potentially autoreactive T lymphocytes, respectively. We and others demonstrated that IL-15 protects against TCR-induced death signaling, in contrast to IL-2. In vitro, CD4⁺ T lymphocytes treated with IL-15 are resistant to apoptosis induced by antigen, resulting in enhanced and extended antigen-specific proliferation.14 In vivo, Bulfone-Paus and coworkers15 reported that cells from the liver, spleen, and thymus of mice challenged with an anti-Fas antibody were rescued from apoptosis by an IL-15 IgG2b fusion protein.15 However, under certain conditions rescue from TCR-induced cell death may represent a threat to the maintenance of peripheral tolerance in the animal, because the rescued cells may generate autoimmune activity or become anergic. Anergized T cells remain viable but lose their capacity for IL-2 production and proliferation in response to an otherwise full stimulatory signal consisting of TCR-engagement (signal 1) and costimulation (signal 2).16 Initially, anergy was described as the consequence of stimulating a T cell with signal 1 in the absence of signal 2 either by activating T cells either with antigen presented by chemically fixed antigen-presenting cells (APC), or with immobilized anti-CD3 mAb or with ConA.17 More recently, it has been documented that anergy can also be induced in the presence of appropriate costimulation by T-T antigen presentation or by inhibiting correctly stimulated T cells to divide with anti-IL2R antibody or rapamycin.18-20 Although these different conditions of anergy induction result in the same basic characteristics of proliferative unresponsiveness and blocked IL-2 production, anergized T cells show discrepancies in other features such as responsiveness to exogenous IL-2, reversibility of the anergic condition, and suppressive activity on other T cells. It is, however, still unknown how the 2 main mechanisms of tolerance induction—apoptosis and anergy—are related. Unclear is whether a particular T cell will preferentially die in response to a tolerizing stimulus, whereas others will become anergic, or whether both pathways can be induced in the same T cell, the outcome depending on environmental factors such as the presence or absence of certain cytokines. Considering the opposite effects of IL-2 and IL-15 on TCR-induced cell death, we investigated whether both cytokines differentially affect the apoptotic viz. anergic response of a CD4⁺ T cell clone to partial (signal 1) stimulation. The results demonstrate that the antiapoptotic cytokine IL-15 skews the T cell to anergy, whereas the proapoptotic cytokine IL-2 generates apoptosis in the same T-cell clone after stimulation with signal 1 in the absence of signal 2. Furthermore, to determine the possible impact on a later immune response, we followed the functionality of the anergized T cells in terms of their responsiveness to environmental IL-2 and IL-15 in the presence or absence of a normal antigenic stimulus.

Female C57Bl/6 (H-2^b) mice were purchased from the Broekman Instituut (Eindhoven, The Netherlands). For spleen cell preparations, mice were used at the age of 9 to 14 weeks.

Purified anti-CD3 mAb (145-2C11; kindly provided by Dr G. Leclercq, Ghent, Belgium) was used at a concentration of 10 μg/mL in phosphate-buffered saline (PBS) to coat flat-bottom microwells (30 μL/well) for 2 hours at 37°C. Unbound antibody was removed by 4 washing steps before adding cells. For immunofluorescence, binding of anti-CD3 mAb was detected with a fluorescein isothiocyanate (FITC)-conjugated antihamster IgG (clone G94-56; Pharmingen, San Diego, CA). Antimouse Fas (clone Jo2), antimouse FasL (clone Kay-10), and anticommon γ chain (clone TUG2m) were also purchased from Pharmingen. Anti-IL-2Rα (clone PC61) and anti-IL-2/15Rβ (clone TM-β1) mAbs were produced in our laboratory. Human recombinant IL-15 (rIL-15) (Peprotech, London, UK) had a specific activity of 2 × 10⁶ U/mg. Human recombinant IL-2 (rIL-2) was produced in our laboratory and had a specific activity of 1.3 × 10⁷ IU/mg as determined in a CTLL-2 assay (1 IU corresponds to 77 pg). Murine IL-2 present in cell culture supernatants was quantified with a Quantikine M mouse IL-2 Immunoassay kit (R&D Systems, Minneapolis, MN) following the manufacturer's instructions. IFN-γ was assayed by a cytopathic effect inhibition assay on murine L929 cells, using encephalomyocarditis virus as a challenger. Granulocyte/macrophage colony-stimulating factor (GM-CSF) was detected by measuring growth of factor-dependent FDCp1 cells. For quantification, murine IFN-γ (10³ U/mL; NIAID Research Resources Section, Bethesda, MD) and mGM-CSF (1 × 10⁵ U/μg; National Institute for Biological Standards and Control, Potters Bar, UK) were used as standards.

The influenza A/H3 HA-specific and H-2^b-restricted CD4⁺ murine T-cell clone T-HA was developed and maintained as previously described.14 Briefly, T-HA cells were cultured in vitro by biweekly restimulation with 10 ng/mL bromelain-cleaved hemagglutinin (BHA) and 5 × 10⁷syngenic spleen cells from C57 BL/6 mice (3000 rad gamma irradiated). On day 2, 30 IU/mL of human IL-2 was added, after which T cells were further expanded by renewing culture medium and IL-2 every 4 days. Culture medium consisted of RPMI 1640 buffered with 12.5 mmol/L HEPES (Life Technologies, Paisley, Scotland) and supplemented with 10% fetal calf serum (FCS), 2 mmol/L GlutaMAX-1 (Life Technologies), 100 U/mL penicillin, 100 μg/mL streptomycin, 1 mmol/L sodium pyruvate, and 5 × 10⁻⁵ mol/L 2-mercaptoethanol.

Apoptotic cells were measured by addition of 30 μmol/L propidium iodide (PI) (ICN Pharmaceuticals, Costa Mesa, CA) to harvested T-HA cells. The percentage of PI-positive cells was measured with a FACScalibur flow-cytometer (Becton Dickinson, Sunnyvale, CA) at 610 to 630 nm. To discriminate T-HA cells from irradiated spleen cells in coculture experiments, T-HA cells were first irreversibly labeled with the membrane dye PKH2-GL (Sigma Chemical, St Louis, MO), as described.14 Percentages of apoptotic T-HA cells were then obtained by flow-cytometric analysis of PI-positive cells emitting green fluorescence (525 nm).

The T-HA cells, cultured with the indicated concentrations of IL-2 or IL-15, were harvested and washed 3 times to remove cytokines. T-HA cells (1 × 10⁴) were seeded in triplicate in flat-bottom 96-well plates for 72 or 96 hours with cytokine, immobilized anti-CD3 or antigen/APC; proliferation was measured by addition of [³H]thymidine (0.5 μCi/well) for the last 12 hours of incubation. Cells were harvested on glass fiber filters and incorporated [³H]thymidine was measured by a Topcount scintillation counter (Packard Instrument, Meriden, CT). In case of antigen-induced proliferation, 200 ng/mL BHA was used as antigen and 2 × 10⁵ irradiated C57BL/6 spleen cells as a source of APC.

We have previously compared the activities of IL-2 and IL-15 on CD4⁺ T lymphocytes that were stimulated with antigen presented by professional APC (signal 1 plus 2) or that were left unstimulated.14 In the absence of TCR engagement, both IL-2 and IL-15 delivered survival signals to CD4⁺ T lymphocytes, protecting the cells against growth factor deprivation-induced cell death. However, although IL-2, at its optimal growth factor activity (10 ng/mL), induced survival accompanied by proliferation, unstimulated CD4⁺ T lymphocytes did not proliferate with IL-15 but acquired a quiescent phenotype. This effect was observed using a wide dose range of IL-15 (0.1-200 ng/mL). The conditioned T lymphocytes also showed a differential responsiveness on stimulation with antigen/APC; IL-15-conditioned cells proliferated much stronger to antigenic stimulation than IL-2-conditioned cells.14 This difference in response was due to the fact that IL-15 protected antigen-stimulated CD4⁺ T lymphocytes against TCR-induced cell death, whereas IL-2 sensitized for this form of cell death. In the current study, we compared the response of IL-2- versus IL-15-conditioned CD4⁺ T cells on stimulation with signal 1 alone. In an introductory experiment, we wanted to ensure that IL-2 and IL-15 exerted their activity on the CD4⁺ T cell clone T-HA by usage of the same receptor components. Therefore, survival of unstimulated T-HA cells induced by IL-2 and IL-15 was measured in the presence of blocking antibodies against the different receptor chains. Table 1 shows that blocking the β and γ_c chain of the IL-2/15 receptor inhibited IL-2- and IL-15-induced survival signals in T-HA cells. As expected, mAb against the specific IL-2Rα chain only blocked IL-2-induced but not IL-15-induced survival signals. Thus, both IL-15 and IL-2 exerted their effects on T-HA cells via the βγ_c chains of the IL-2/15 receptor complex.

As a model to assess the response of IL-2- versus IL-15-conditioned T-HA cells to signal 1 alone, T-HA cells were treated for 48 hours with IL-2 (10 ng/mL) or IL-15 (1 ng/mL) and subsequently stimulated with immobilized anti-CD3 mAb in the absence of APC, a stimulation condition that induces unresponsiveness in T cell clones.21 As expected, T-HA cells did not proliferate in response to immobilized anti-CD3 (Figure 1A), although amounts of IL-2 in the nanogram range were produced (5-15 ng/mL). Also, addition of exogenous IL-2 or IL-15 during CD3 cross-linking could not restore proliferation (data not shown). This state of proliferative unresponsiveness was not differentially influenced by conditioning with IL-2 or IL-15. As a control, the same cells stimulated with antigen/APC showed the expected differential response after IL-2 or IL-15 conditioning, in agreement with our previous results.14 To verify whether the proliferative unresponsiveness induced by anti-CD3 mAb was the consequence of cell death, apoptosis was measured after 24 and 48 hours by flow cytometric analysis of PI uptake. As shown in Figure 1B, T-HA lymphocytes conditioned by IL-2 massively underwent apoptosis, whereas IL-15 protected against anti-CD3-induced cell death. Although immobilized anti-CD3 mAb thus induces proliferative unresponsiveness in both IL-2- and IL-15-conditioned T-HA cells, the mechanism responsible for this unresponsiveness differs; IL-2 promotes elimination of the T cells by apoptosis, whereas IL-15 promotes survival accompanied by a proliferative block of T cells.

Although most authors support the view that anti-CD3, in combination with IL-2, activates mainly the Fas/FasL cell death pathway,22 other effector molecules such as tumor necrosis factor or reactive oxygen species can be used to induce apoptosis after T-cell activation.23,24 To assess the contribution of the Fas death pathway to apoptosis induced by anti-CD3 plus IL-2, antagonistic Abs against Fas and FasL were applied. The combined addition of both mAbs reduced the apoptotic response of the IL-2-treated T-HA cells against immobilized anti-CD3 by approximately 50% (Figure 1C). This suggests that Fas/FasL interaction is an important but not the sole death effector mechanism induced by anti-CD3 plus IL-2 in the T-HA clone. Moreover, it can be concluded that IL-15 protects against both these Fas-dependent and Fas-independent apoptotic pathways. Finally, the possibility existed that a differential expression of CD3 by the T-HA cells after culture in IL-2 or IL-15 caused the quantitatively different apoptotic signal induced by anti-CD3 cross-linking. Although unlikely considering the strong proliferative response of IL-15-treated T-HA cells against antigen/APC and the similar up-regulation of IL-2Rα induced by immobilized CD3, we nevertheless analyzed CD3 expression on IL-2- versus IL-15-conditioned T-HA cells. As shown in Figure 1D, no difference in CD3 levels was observed, thus ruling out this possibility. Therefore, we conclude that the antiapoptotic activity of IL-15, previously reported to act on appropriately stimulated CD4⁺ cells, also is effective on partially, signal 1-stimulated T cells, allowing the cells to remain viable but unresponsive under these tolerizing stimulation conditions.

During stimulation with antigen/APC, T-HA cells are responsive to the mitogenic activity of IL-15. However, in the absence of TCR engagement, T-HA cells become quiescent in response to IL-15 although they continue to proliferate with IL-2.14 To determine whether T-HA cells activated by signal 1 alone also became quiescent after removal of the stimulus, IL-15-conditioned cells were activated by immobilized anti-CD3 for 48 hours, harvested, and cultured with IL-15 or IL-2 for an additional 96 hours. In contrast to T-HA cells stimulated with signal 1 plus 2, the cells that received signal 1 alone proliferated not only strongly with IL-2 but also with IL-15 (Figure2A,B). This result was surprising because T-HA cells are unable to proliferate as long as their CD3 molecules are cross-linked by immobilized anti-CD3 Ab (Figure 1A), although they produce a mitogenic amount of IL-2 (14 ng/mL) under this condition. After release from the stimulus, however, they transgress to an activated, IL-2- and IL-15-responsive state. The persistence of this semiactivated state was evaluated by continued culture of T-HA cells with a mitogenic dose (10 ng/mL) of IL-2 or with IL-15 (1 ng/mL) for 12 days and subsequently measuring proliferation in response to the respective cytokines. Figure 2, panels C and D, show that the partially stimulated T-HA cells remained responsive to IL-2 but finally lost their responsiveness to the mitogenic activity of IL-15, thus behaving in a similar manner as appropriately stimulated T-HA cells. In conclusion, IL-15-treated T-HA cells stimulated with immobilized anti-CD3 remain viable, incapable of autocrine growth, but can still respond to the exogenous growth factors IL-2 and IL-15 for some days after removal of the TCR stimulus. Then they gradually lose the capacity to proliferate in response to IL-15, whereas IL-2 responsiveness remains unaffected.

Hargreaves and colleagues25 showed that T-cell clones stimulated by T-T antigen presentation died from Fas/FasL interaction. Rescue from cell death by an anti-Fas mAb rendered the cells anergic to appropriate restimulation with antigen/APC. Because IL-15 similarly rescued the cells from immobilized anti-CD3-induced cell death, we questioned whether T cells could still respond when appropriately restimulated with signal 1 plus 2. Therefore, IL-15-conditioned T-HA cells were stimulated for 48 hours on anti-CD3-coated plates, detached, and further cultured with IL-2 (10 ng/mL) or IL-15 (1 ng/mL) for 12 days. In parallel, control cultures of T cells stimulated with antigen/APC were maintained under the same conditions. On day 12, the cultures were harvested, labeled with a green fluorescent dye, and stimulated with antigen/APC. The proliferative response, IL-2 production, and occurrence of cell death were measured. As shown in Figure 3A, no proliferative response could be induced by antigen/APC in T-HA cells that had previously been stimulated with immobilized anti-CD3 mAb, irrespective of the cytokine (IL-2 or IL-15) added during the intermittent culture period. As opposed to these partially stimulated T-HA cells, T-HA cells that received signal 1 plus 2 as first stimulus proliferated in response to a rechallenge with antigen/APC, although proliferation after culture in 10 ng/mL IL-2 was limited for the reasons mentioned above. In addition, partially stimulated T-HA cells were unable to produce IL-2 after restimulation with antigen/APC (Figure 3A, inset). Thus, although IL-15 protects T-HA cells against immobilized anti-CD3-induced cell death, the surviving cells have become unresponsive to restimulation with antigen/APC. It should also be noted here that addition of a dose of IL-2 (10 ng/mL) that clearly induces cell division during the intermittent culture period (Figure 2A,C) apparently does not break the state of unresponsiveness in our T-cell clone. To resolve the question whether unresponsiveness was the consequence of cell death or anergy, that is, a state of proliferative unresponsiveness to antigen coupled to viability, we followed the emergence of dead T-HA cells during the antigenic challenge. Figure 3B shows that cell death occurred in the absence of any stimulus or growth factor but was prevented by signals derived from the APC. The presence of antigen did not further enhance nor reduce the number of living cells, in agreement with the anergic state of the cells. Hence the unresponsiveness to antigen/APC of T-HA cells rescued from immobilized anti-CD3-induced cell death by IL-15 is the result of anergy and not of TCR-induced cell death.

To verify whether anergic T cells may still contribute to immune responses despite their proliferative block, we investigated whether TCR signaling in anergic cells induced responsiveness to exogenous growth factors and production of endogenous cytokines. T-HA cells rendered anergic by IL-15 treatment, followed by 12 days of culture in IL-2 or IL-15, were restimulated with antigen/APC in the absence or presence of a suboptimal, nonmitogenic concentration of exogenous IL-2 (1 ng/mL) or a standard, likewise nonmitogenic, concentration of IL-15. Figure 4 shows that antigen/APC activation of anergic cells makes them responsive to the growth factor activity of exogenous IL-15, and especially of low-dose IL-2. This result suggests that availability of even low concentrations of bystander growth factors such as IL-2 and IL-15 in the microenvironment of antigen-activated, anergic cells may lead to the further expansion of these cells. Furthermore, in culture supernatants from antigen-activated, anergic cells we detected a significant amount of IFN-γ and marginal GM-CSF production as compared to appropriately stimulated T-HA cells (Table 2).

The observation that anergized T cells still possess the capacity to produce cytokines in response to antigenic stimulation emphasizes the potential of anergic cells to play a regulatory role during immune reactions.

Apoptosis and anergy have both been proposed as important mechanisms for the establishment of tolerance in the periphery. It is, however, still unknown how both mechanisms are related. In the past, various T-cell clones have been examined for the nature of their response to a tolerizing stimulus. Studies using immobilized anti-CD3 mAb or other stimuli to tolerize clearly showed that some T-cell clones responded by undergoing apoptosis, whereas others became anergic.26-30Furthermore, in T cells that undergo apoptosis induced by T-T antigen presentation, blockade of apoptosis with an anti-Fas mAb resulted in anergy induction, thus showing that both pathways can be activated simultaneously.25 So far, however, no physiologic factors have been identified that can discriminate between anergy and apoptosis induction in a single T cell clone. So it remains an open question which inactivating mechanism is chosen by a particular, potentially autoreactive T-cell clone in response to a tolerizing stimulus and which factors govern this choice. In the present study, we used immobilized anti-CD3 mAb to induce unresponsiveness in the CD4⁺ T-cell clone T-HA and compared the influence of the cytokines IL-2 and IL-15 on the type of unresponsiveness induced in this clone. These cytokines were chosen because they exert opposing activities with respect to cell death initiated by TCR engagement; IL-2 promotes and IL-15 inhibits TCR-induced cell death.10,14,15In our T-HA clone, immobilized anti-CD3 mAb induced proliferative unresponsiveness. However, the cytokine added during the culture period preceding anti-CD3 treatment critically determined the nature of this unresponsiveness. Thus, T-HA cells that had been cultured with IL-2 underwent robust apoptosis after CD3 cross-linking, explaining the absence of proliferation. IL-15-treated cells, although also incapable of division, remained on the contrary viable. These surviving cells no longer responded by proliferation to a normally full stimulatory condition, namely, antigen presented by professional APC, demonstrating that they had acquired an anergic phenotype. These results indicate that cytokines can govern the mechanism by which a particular T-cell clone is tolerized. Delivery of signal 1 alone in the presence of IL-2 results in rapid apoptosis, whereas in the presence of IL-15 anergy is favored. Because anti-CD3-induced cell death of IL-2 treated T-HA cells is at least partially mediated by Fas/FasL interaction, the rescue from apoptosis by IL-15 correlates with the described protective activity of IL-15 against Fas-induced cell death. Apparently, IL-15 skews a signal 1 response from apoptosis to anergy by blocking the Fas death signaling pathway. Clearly, Fas/FasL-induced apoptosis is extremely important for maintenance of tolerance in the secondary lymphoid organs,31 and sensitization for this Fas death pathway by IL-2 is accordingly a key regulatory event to avoid autoimmunity. Inhibition of this tolerizing death pathway by IL-15 might result in the establishment of a less absolute form of tolerance, namely, anergy instead of depletion. Recently, also IL-12 has been attributed protective capacity against Fas-mediated cell death of antigen-specific T cells.32 Addition of an anti-IL-12 neutralizing antibody to mice tolerized for OVA enhanced peripheral tolerance, possibly by allowing apoptosis to prevail. Thus, whereas IL-2 clearly promotes a tolerizing response to signal 1, other cytokines such as IL-12 and, as we show here, IL-15 may abate such a response by preventing apoptosis but instead generating anergy.

Whereas our data clearly demonstrate a decisive, albeit opposite role of the γ_c-signaling cytokines IL-2 and IL-15 in induction of anergy, other data seem to contradict such a role. Thus, Boussiotis and associates33 clearly demonstrated that signaling via the γ_c chain, leading to Jak3 tyrosine-phosphorylation, prevented induction of unresponsiveness to a tolerizing stimulus in human T cell clones. However, these authors used antigen presented by APC with blocked B7 function to stimulate the T lymphocytes. Thus, alternative costimulatory molecules expressed by the APC could influence the response of the T cells to signal 1 and the γ_c-signaling cytokine. In our model, using anti-CD3 mAb as signal 1, this problem is avoided. Also, a weaker aggregation of TCRs by antigen presented via APC, as opposed to immobilized anti-CD3 mAb, could facilitate escape from unresponsiveness provided additional cytokine signaling is available.

To further characterize the state of (un)responsiveness of anergized T-HA cells, we analyzed the activity of IL-2 and IL-15 on anergized cells maintained for prolonged periods. T-HA cells released from immobilized anti-CD3 regained proliferative responsiveness to exogenous IL-2 as well as IL-15. IL-2 induced proliferation for indefinite periods of time (up to 15 days), whereas IL-15 only exerted growth factor activity for a few days, after which the cells proceeded to a resting phase. The phenomenon that T cells are not responsive to endogenous IL-2 when bound on immobilized CD3 mAb, but regain responsiveness after release from the stimulus, has been described before for cloned human T cells.34 Our data demonstrate that this regulation also applies to the growth factor activity of IL-15. Also the previously reported promotion by IL-15 of a quiescent state on disappearance of antigen still applies to the anergized cells with, however, the deviation that the cells once released from immobilized anti-CD3 remained sensitive to IL-15 growth factor activity longer than those cleared from antigen. Because residual anti-CD3 mAb stayed bound to the cell surface for some days (data not shown), the cells were kept in an activated condition for an additional period. Possibly, this explains the prolonged responsiveness of the cells to IL-15 growth factor activity. Eventually, however, these cells acquire a resting state when further cultured in IL-15, as concluded from the absence of proliferation and down-regulation of IL-2Rα expression on day 17 (data not shown). However, independently of their proliferating viz. resting state, T-HA cells maintained in IL-2 or IL-15, respectively, retained their anergic condition. This result is similar to that of other investigators who could not break anergy by the induction of proliferation with IL-2 in anergized cells35,36 and argues against the idea that anergy is maintained by factors in the cytosol that dilute after a few cell cycles, resulting in reversal of anergy.19,37,38 

Although blocked in their proliferative response to signal 1 plus 2, T cells nevertheless still exhibited responsiveness, apparent from their production of IFN-γ and GM-CSF, but not IL-2, and their increased sensitivity to the growth-factor activity of IL-2 and IL-15. The observation that delivery of signal 1 plus 2 to anergized cells induced sensitivity to IL-15 growth factor activity can have important consequences for certain in vivo situations. Thus, anergized cells stimulated by signal 1 plus 2, although failing to produce autocrine IL-2, nevertheless may start to divide due to the availability of IL-15 in the microenvironment. This IL-15 could even be produced by the dendritic cells presenting the cognate antigen to the anergic cells, for example when TRANCE/TRANCE-R interactions would occur under these circumstances.39 This activation condition could possibly represent a breach in the defense line against self-reactive lymphocytes.

In summary, the results described in this paper demonstrate that the outcome of a partial T-cell stimulation may be reoriented in a single T-cell clone from apoptosis to anergy, depending on the cytokine available to the cell. This result directly implicates that cell death can be considered the dominating response of a T cell to signal 1 and its inhibition by whatever means, IL-15 in our case or anti-Fas as reported by Hargreaves and coworkers, automatically leads to emergence of the back-up anergic pathway. Furthermore, our results emphasize the critical influence of the cytokine environment on the final outcome of a tolerizing stimulus. Availability of IL-2 will favor apoptosis to occur, whereas IL-15 avoids deletion of the cell while preserving anergy induction. Thus, in models of T-cell stimulation with signal 1 in the absence of signal 2, the cytokine could be considered as the “signal 2,” determining what tolerizing response is induced.

We are grateful to D. Ginneberge and W. Burm for technical assistance. T.V.B. is a fellow with the Vlaams Instituut voor de Bevordering van het Wetenschappelijk-technologisch Onderzoek in de Industrie.