Rescuing Melanoma Epitope-Specific Cytolytic T Lymphocytes from Activation-Induced Cell Death, by SP600125, an Inhibitor of JNK: Implications in Cancer Immunotherapy¹

Programmed cell death (PCD)³ and activation-induced cell death (AICD) in T cells are important physiologic processes that prevent untoward side effects of a continued and uncontrolled T cell-mediated effector response as well as maintain homeostasis (1, 2, 3). Both processes involve apoptotic deletion of T cells. PCD entails the deletion of the expanded T cell population usually during the contraction phase of the response. AICD, in contrast, involves the apoptotic deletion of a significant fraction of the activated population after an effector response. In AICD, the effector function and the death are, paradoxically, triggered by TCR-driven signaling (i.e., activation induced). In any event, both processes (i.e., PCD and AICD) are designed to serve useful purposes by limiting the expansion of activated T cells. AICD in activated T cells in tumor immunotherapy, however, can be counterproductive particularly if the activated T cells undergo apoptotic death after the very first encounter of the specific epitope.

Lately, much interest has been generated in cancer vaccine therapy with specific peptides, protein Ags, DNA, etc. Most of these immunogens are “self” Ag (4), yet many cancer patients as well as normal healthy hosts harbor precursor CTLs for such “self” epitopes. Ex vivo stimulation of T cells as well as in vivo immunization with such self peptides or tumor-associated Ags (TAA) lead to the activation and expansion of the Ag-specific CTLs (5, 6). These TAA-specific T cells are also susceptible to AICD. To our knowledge, neither the extent of AICD in these self-but-TAA reactive primary CTLs nor the feasibility of rescuing them from AICD has been carefully examined. We have recently shown that Melan-A/MART-1_27–35 epitope-specific CTLs expanded in an in vitro dendritic cell (DC)-based stimulation protocol undergo apoptotic death on repetitive stimulation by immature as well as by fully activated DCs (7). Using the Melan-A/MART-1_27–35 epitope (8, 9) as a prototype self but melanoma-associated Ag, we studied the extent of AICD in MART-1_27–35 epitope-specific primary CTLs and examined whether it could be prevented. In this study, we show that a large fraction of MART-1_27–35 epitope-specific CTLs indeed undergo AICD upon the very first secondary encounter of the cognate epitope. The AICD in these CTLs is neither caspase dependent nor is it triggered by the engagement of extrinsic death receptors. We also show that the JNK inhibitor, SP600125, can rescue a significant fraction of them from death. In the process of rescuing, the JNK inhibitor interferes with their capacity to produce IFN-γ but does not interfere with their cytolytic function. Of further interest, the rescued MART-1_27–35 epitope-specific CTLs, when continued in culture in IL-15 and without the inhibitor, regain capacity to synthesize IFN-γ. These observations, therefore, have implications in cancer immunotherapeutic strategies and in further studies of AICD in CTLs.

The study population consisted of HLA-A2-positive melanoma patients or healthy donors. The participants were included in this study with informed consent.

The MART-1_27–35 peptide (AAGIGILTV) and MAGE-3_271–279 (FLWGPRALV) were purchased from Multiple Peptide Systems (San Diego, CA) while β₂-microglobulin was purchased from Sigma-Aldrich (St. Louis, MO). Culture medium consisted of IMDM (Invitrogen Life Technologies, Grand Island, NY) supplemented with 10% FBS (Gemini Bioproducts, Calabasas, CA), 0.55 mM l-arginine, 0.24 mM l-asparagine (both from Invitrogen Life Technologies), 1.5 mM l-glutamine (Sigma-Aldrich), 50 U/ml penicillin, and 50 μg/ml streptomycin (both from Abbott Laboratories, North Chicago, IL). This will be referred to as complete media (CM). The TAP-deficient line, T2, was a gift of P. Cresswell (Yale University, New Haven, CT). Recombinant human GM-CSF was purchased from Immunex (Seattle, WA). Recombinant human IL rhIL-4, rhIL-2, rhIFN-γ was purchased from R&D Systems (Minneapolis, MN). LPS from Escherichia coli 055:B5 was purchased from Sigma-Aldrich. An Annexin V kit to track the early apoptotic cells for exposure of phosphatidylserine was purchased from BD Pharmingen (San Jose, CA). MART-1_27–35 (EAGIGILTV) tetramer labeled with PE with and without FITC-labeled anti-CD8 was purchased from Beckman Coulter (Fullerton, CA). Fluorochrome-labeled mAbs to CD25, CD27, CD28, CD95, CD95L, 4-1BB, 4-1BBL, OX-40 were purchased from BD Biosciences (San Jose, CA). Inhibitors for various kinase pathways, as SB203580 for p38 kinase, SP600125 for JNK, PD98059 for ERK were purchased from BIOMOL (Plymouth Meeting, PA). Pan caspase inhibitor, human Fas/Fc chimera, human TNF- RI/Fc chimera, human TRAIL-RI/Fc chimera, human TRAIL-RII/Fc chimera, and human IFN-γ RI/Fc chimeric proteins were purchased from R&D Systems.

The procedure for generating myeloid DCs from peripheral blood monocyte was followed. Briefly, circulating monocyte were isolated by 2 h adherence of Ficoll-Hypaque density gradient-cut PBMC as previously described (7). The adherent cells were cultured in CM with 1000 U/ml GM-CSF and 500 U/ml IL-4 for 3–5 days to obtain a population of immature DCs. Maturation of immature DCs was done by first priming in IFN-γ (1000 U/ml) for 2 h and then priming in the presence of 100 ng/ml LPS.

IFN-γ response assay for the effector cells has been described previously (10). Briefly, the effector cells were cocultured with the peptide-pulsed (1 μg/ml) T2 cells. After 4–16 h, culture supernatants were harvested and IFN-γ was measured by ELISA as per manufacturer’s protocol (R&D Systems).

The procedure for phenotypic analyses and for determining the number of epitope-specific cells with tetramer staining by flow cytometry has been described (7). To determine tetramer and annexin V-positive cells, the effector cells were stained with CD8, MART-1_27–35/HLA-A2 tetramer, and annexin V. The stained cells were then analyzed for single-positive vs double-positive populations in flow cytometry using a FACSCalibur and CellQuest software (BD Biosciences).

The chromium release microcytotoxicity assay has been previously described (11).

The basic procedure for peptide-loaded DC-based in vitro activation and expansion of epitope-specific CD8⁺ T cells has been described (7). Briefly, Ficoll-Hypaque gradient separated blood mononuclear cells were purified for CD8⁺ T cells (routinely exceeding 90%) by a Dynal magnetic bead isolation kit (Dynal, Oslo, Norway) and cocultured with autologous DCs pulsed with relevant peptides (100 μg/ml) and 5 μg/ml β₂-microglobulin at a CD8⁺ T cell to DC ratio of 100. For our current purpose, because IL-2 has been known to facilitate apoptosis in activated T cells (12), the cocultures were conducted in the presence of rhIL-15 (10 ng/ml). Before setting up cocultures, the DCs were irradiated to 3000 rad. The activated CTLs were also maintained in culture in IL-15.

To test whether or not the epitope-specific primary CTLs undergo AICD, the activated CTLs were exposed to peptide (1 μg/ml) loaded T2 cells (E:T = 100) at different time points of the cultures. Thereafter, the evidence of apoptosis was determined by flow cytometry with triple color staining (CD8, MART-1_27–35/HLA-A2 tetramer, and annexin V) at different time points (4–18 h). Experiments were conducted in triplicate wells and significance was calculated by one-way ANOVA using Sigma Stat statistical software (Chicago, IL).

To evaluate the effect of various agents in modulating AICD, the CTLs were preincubated with various compounds at optimal concentration for 45 min at 37°C and then exposed to T2 cells alone or loaded with peptide. The optimal dose used in the experiments shown in this paper was determined by using these compounds at different concentrations in the preliminary experiments (data not shown).

We selected donors (healthy donors as well as melanoma patients) who harbor MART-1_27–35 epitope-specific CTLs in relatively high frequencies and whose CTL precursors could be easily activated (7). Because IL-2 has been implicated in apoptosis in activated T cells (12), we looked for an alternative cytokine support in the in vitro CTL generation protocol. We found that rIL-15 supports the activation and expansion of the epitope-specific CTLs in the mature DC-based epitope presentation system. IL-15 not only supports the primary activation process, but also supports survival of the activated T cells in an Ag-independent manner (13). IL-15Rα is not down-regulated after T cell activation (14) and IL-15 has an important role in T cell memory generation, as such, in T cell survival (15). Therefore, we conducted all our experiments in the presence of IL-15. The MART-1_27–35 epitope-specific CTLs were readily activated and expanded when they were stimulated by peptide-pulsed matured DCs in the presence of IL-15, in vitro. Fig. 1,A shows an example of the expansion of the MART-1_27–35 epitope-specific CTLs–derived from a normal healthy individual and a melanoma patient. The expanded MART-1_27–35 epitope-specific cells also exhibited IFN-γ response in an epitope-specific manner (Fig. 1 B).

The fate of the MART-1_27–35 epitope-specific CTLs upon secondary encounter of the cognate epitope (i.e., whether following effector function, they survive or die?) was then examined. The CTLs and the peptide-pulsed T2 cells were cocultured in CM without any cytokine and 4 h later the evidence of early death was examined by annexin V staining. As shown in Fig. 2, although the CTLs were fully functional in IFN-γ assay (Fig. 2,A), a large fraction of these activated CTLs became annexin V positive at 4 h (Fig. 2,B). When the cocultures were continued in IL-15 for 5 days and the number of epitope-specific CTLs was counted, only 20–30% of the starting population could be recovered from the coculture of the CTLs with the cognate target. An example for loss of the epitope-specific CTLs following effector function is shown in Fig. 2,C. Thus, ∼50% of the Ag-specific population showed early evidence of death at 4 h (Fig. 2 B) and a much larger fraction eventually died following secondary encounter of the cognate epitope for the very first time.

We then examined the mechanism of AICD on these CTLs and also examined whether these CTLs could be rescued from AICD by interfering with the death signaling pathway(s). We examined the effect of interfering with the external death signal receptors (such as Fas, TNFR, TRAIL, etc.) on AICD. We also examined whether any of the MAPK inhibitors could block the AICD in these CTLs because MAPKs have been implicated in the negative selection of developing thymocytes (16, 17). As shown in Fig. 3,A, the death of these CTLs was not caspase dependent, as the pan caspase inhibitor, z-VAD-fmk did not prevent apoptosis. The apoptosis was not also affected by the blockade of the common extrinsic death signal receptors such as Fas, TNFR, or TRAIL. However, we found that while the p38 and ERK inhibitors had no effect, the JNK inhibitor, SP600125, rescued a significant fraction of the CTLs from AICD. Interestingly, SP600125 also inhibited the IFN-γ response by these CTLs (Fig. 3,B) but did not affect their cytotoxic function (Fig. 3 C) suggesting that the cytolytic machinery and the IFN-γ response pathways are differently regulated in these CTLs. Induction of AICD upon secondary exposure to the cognate epitope in the primary CTLs and the protective effect of SP600125 from AICD were observed in CTLs generated from two normal donors and two melanoma patients (collective data not shown).

The caspase-independent death in these primary CTLs upon TCR engagement was confirmed as poly(ADP-ribose) polymerase (PARP) was found to be uncleaved (Fig. 4,A). Of note, PARP was cleaved in staurosporine-treated Jurkat cells and z-VAD-fmk blocked the PARP cleavage. We also confirmed that SP600125 abrogates c-jun activation (Fig. 4 B).

We examined the status of the mitochondrial membrane potential in the CTLs during AICD. As shown in Fig. 5 A, the CTLs surprisingly exhibited hyperpolarization of the mitochondrial transmembrane potential (Δψ_m) (18, 19) while the apoptotic Jurkat cells exhibited hypopolarization of the mitochondrial membrane. Hyperpolarization has been associated with the apoptotic process in peripheral blood lymphocytes and has also been shown to occur independently from activation of caspases (20, 21). Further, hyperpolarization of mitochondrial membrane potential has been shown to be a reversible stage in the mitochondrial death decision process while hypopolarization usually represents a point of irreversible commitment to death (18, 19). Interestingly, SP600125 (JNK inhibitor) as well as SB203580 (p38 inhibitor) decreased the level of hyperpolarization induced by the restimulation with the cognate peptide (data not shown) suggesting that SP600125-mediated rescue is not specifically mediated through the modulation of mitochondrial membrane potential.

The status of Bcl family pro- and antiapoptotic proteins and costimulation through certain TCRs (CD28, 4-1BB, OX-40, etc.) influence the survival of activated T cells (22). As such, we examined the expression levels of the anti- and proapoptotic Bcl family proteins and these receptors in the CTLs in a condition that induces AICD. As shown in Fig. 5,B, while Bcl-2 and Bcl-x_L were nonselectively up-regulated by SB203580 and SP600125, only phosphorylated Bcl-2 and Mcl-1 were selectively up-regulated by the JNK inhibitor SP600125. Bim–a pro-apoptotic protein–was markedly up-regulated (or released from sequestered sites) by the CTLs undergoing AICD. Of interest, the CTLs up-regulated 4-1BB (10- to 20-fold increase of the geometric mean intensity of fluorescence) and CD25 (3- to 6-fold increase of the geometric mean intensity of fluorescence) upon secondary encounter of the Ag (Fig. 5 C).

Finally, we examined the functional status of SP600125-rescued CTLs upon continuous culture. After overnight exposure of the CTLs to the cognate epitope in the presence or absence of SP600125, the CTLs were washed and then maintained in continuous culture in IL-15 containing CM without the inhibitor. Five days after, the number of viable MART-1_27–35 tetramer-positive cells were determined and their function was tested in the IFN-γ response assay. Fig. 6,A shows that only a small fraction of the starting population could be recovered from the cognate epitope-exposed culture in the absence of the JNK inhibitor. A larger fraction of the starting epitope-specific population was recovered from the coculture that was started in the presence of the JNK inhibitor, SP600125. Remarkably, the rescued CTLs regained their capacity to synthesize IFN-γ (Fig. 6 B).

Ashwell et al. (1) first noted growth arrest in T cell hybridoma upon TCR-mediated stimulation. Shi et al. (2) thereafter showed that TCR- driven signaling leads to death of the cells and coined the term AICD. In physiology, T cell responses follow three phases: amplification, contraction, and memory generation. During the contraction phase of a primary T cell response, the majority of the Ag-specific T cells undergo apoptotic death (23) as a form of PCD primarily to maintain homeostasis (24). It has been thought for some time that the apoptotic signal is initiated through the Fas and TNF receptors and that the cell death is mediated by the eventual activation of an executioner caspase (25, 26). A universal role of the Fas and TNF family receptor-driven signaling for T cell apoptosis has, however, been controversial (27) and both caspase-dependent and caspase-independent deaths in T cells have been described (28). Further, it is now increasingly apparent that an intrinsic cell death pathway triggers AICD in CTLs without involving the extrinsic death receptors such as Fas and/or TNF family receptors (29, 30). Although further studies will be needed to establish the mechanism underlying apoptosis in the epitope-specific primary CTLs in our system, our data lend support to the latter view.

The results are also noteworthy for a number of other reasons. First, the data show that a large fraction of the epitope-specific primary CTLs generated, in vitro, are indeed susceptible to AICD from their very first re-encounter of the Ag even during the amplification phase (Fig. 2). It should be mentioned that the MART-1_27–35 epitope-specific populations expanded more than 1000-fold in IL-15 without a second round of stimulation and then senesced around day 50. These CTLs, interestingly, exhibited propensity to undergo AICD as early as on day 9 (collective data not shown). This, therefore, suggests that the apoptotic program in this in vitro CTL activation system is turned on quite early and lends support to the recent observations that the apoptotic program in viral Ag-specific CTLs gets activated substantially before the contraction phase, in vivo (31). Of interest, stimulation of the CTLs with the cognate peptide that induced apoptosis paradoxically up-regulated the expressions of IL-2Rα (CD25) and 4-1BB (CD137) substantially (Fig. 5,C). Although these molecules are associated with T cell growth and survival, 4-1BB has lately been implicated in apoptosis and has also been shown to interfere with IFN-γ response but not with cytotoxicity in CTLs (32, 33). A connection between 4-1BB-mediated signaling and activation of the JNK pathway has also been suggested (34). Second, the MART-1_27–35 epitope-specific CTLs generated in the in vitro DC-based epitope presentation protocol in the presence of IL-15 and maintained in IL-15 were just as susceptible to AICD as the CTLs that were generated and maintained in IL-2 (collective data not shown). Third, the results clearly reveal that SP600125 is capable of protecting these CTLs from AICD (Fig. 3). Because SP600125 functions as an inhibitor of the JNK, the data suggest a role for JNK in the apoptotic process. Fourth, while SP600125 negatively regulates the IFN-γ responsiveness of the CTLs, it does not affect their cytotoxic function, suggesting that the two functional pathways in CTLs are differently regulated. Finally, and most importantly, the rescued CTLs regained IFN-γ responsiveness in continued culture after the inhibitor being washed off.

The JNK is a member of the MAPK family of signaling proteins. JNK has been associated with a variety of biological processes–most notably in inflammation and transformation (35). All three MAPKs (i.e., p38, JNK, and ERK) have been found to have different roles in T cell biology such as in proliferation, effector function, survival as well as death (16, 17, 36, 37, 38, 39, 40). Different isoforms of JNK have been found to have divergent roles in CD4⁺ and CD8⁺ T cells (17). For example, JNK1 is needed for the expression of IL-2Rα (CD25) upon activation and for proliferation. As such, jnk-1-null mice exhibit marked reduction in the expansion of Ag-driven CD8⁺ T cells (39). JNK2 down-regulates IL-2 production in CD8⁺ T cells (39, 40). JNK has also been implicated in negative selection of thymocytes (17, 40). Both JNK1 and JNK2 have been found to have stage-dependent roles in T cell development. For example, while JNK1 has been associated with anti-CD3-induced apoptosis of double-positive thymocytes, jnk2-null double-positive thymocytes are resistant to anti-CD3-induced apoptosis although they are sensitive to apoptosis induced by dexamethasone, anti-Fas Ab, or UV radiation (38, 39). Thus, immature thymocytes seem to need one form of JNK or another for undergoing receptor-driven apoptosis but mature T cells need JNK1 for proliferation and IL-2 synthesis, and they do not seem to need JNK2 for activation-induced apoptosis (39, 40). Although these observations suggest a role for JNK in apoptosis in immature thymocytes but not in mature T cells, the results of our study reveal that JNK may be an important player in apoptotic death in CTLs in the periphery. A careful examination of this type of AICD in CTLs with specificity for a “self” vs a “dangerous” Ag in the same donor will be informative.

JNK is activated by dual phosphorylation of the Thr and Tyr residues within the Thr-Pro-Tyr motif and it can be activated by cytokines and environmental stress. Its activation leads to the up-regulation of c-jun and eventual phosphorylation of transcription factors such as AP-1 and other proteins some of which are associated with apoptosis. Given that death in CD8⁺ T cells is turning out to be mostly driven by the internal pathway of apoptosis and results from mitochondrial dysfunction, the release of reactive oxygen intermediates, and stress (29, 30), and as JNK can play a role in stress-induced activation of the cytochrome c-mediated death pathway (41), a role for JNK in AICD in Ag-specific CD8⁺ CTLs can be envisioned.

As mentioned earlier, Bcl family pro- and antiapoptotic proteins play an important role in the survival or death of cells (22). Our observation on the modulation of some of these proteins, therefore, is of interest. The JNK inhibitor, SP600125, increased the level of pBcl-2 and Mcl-1. The effect of phosphorylated Bcl-2 on apoptosis of T cells remains an unsettled issue (42, 43). However, Mcl-1 is an important Bcl family member with a positive effect on T cell survival (44). It is therefore tempting to suggest that the JNK inhibitor SP600125 prevents AICD in these CTLs by up-regulating Mcl-1. Admittedly, more work will be needed to establish this issue.

A role for JNK in AICD in CTLs in an Ag-specific manner in our study has been based from the observation of the effect of SP600125 on AICD in these cells (Fig. 3). SP600125 has been extensively used as a JNK inhibitor in numerous studies although it is not absolutely specific for JNK (35). The possibility exists that the rescuing effect of SP600125 could have resulted from its effect on some other molecule(s). More work will be needed to firmly establish a role for JNK in AICD in Ag-specific CTLs and to “connect” JNK with the upstream and downstream events in the apoptotic processes in primary CTLs. Meanwhile, considering that the antitumor effects of the CTLs–induced by active vaccination or adoptively transferred–can be vastly enhanced by preventing their “premature” death and as JNK inhibitors have generated much interest in inflammatory diseases and have also entered into clinical trials (35), our observations have implications in active as well as adoptive immunotherapeutic strategies for cancer and in further studies of the regulation of survival and death in CTLs.

We thankfully acknowledge Drs. Leo Lefrancois for his careful reading of the manuscript, Meenal Mehrotra for help in some of the initial experiments, and S. K. Suneja for help with densitometric analysis.