Agonist Anti-GITR Antibody Enhances Vaccine-Induced CD8⁺ T-Cell Responses and Tumor Immunity

Immunization of mice with plasmids encoding xenogeneic orthologues of tumor differentiation antigens can break immune ignorance and tolerance to self and induce protective tumor immunity. We sought to improve on this strategy by combining xenogeneic DNA vaccination with an agonist anti–glucocorticoid-induced tumor necrosis factor receptor family–related gene (GITR) monoclonal antibody (mAb), DTA-1, which has been shown previously both to costimulate activated effector CD4⁺ and CD8⁺ T cells and to inhibit the suppressive activity of CD4⁺CD25⁺ regulatory T cells. We found that ligation of GITR with DTA-1 just before the second, but not the first, of 3 weekly DNA immunizations enhanced primary CD8⁺ T-cell responses against the melanoma differentiation antigens gp100 and tyrosinase-related protein 2/dopachrome tautomerase and increased protection from a lethal challenge with B16 melanoma. This improved tumor immunity was associated with a modest increase in focal autoimmunity, manifested as autoimmune hypopigmentation. DTA-1 administration on this schedule also led to prolonged persistence of the antigen-specific CD8⁺ T cells as well as to an enhanced recall CD8⁺ T-cell response to a booster vaccination given 4 weeks after the primary immunization series. Giving the anti-GITR mAb both during primary immunization and at the time of booster vaccination increased the recall response even further. Finally, this effect on vaccine-induced CD8⁺ T-cell responses was partially independent of CD4⁺ T cells (both helper and regulatory), consistent with a direct costimulatory effect on the effector CD8⁺ cells themselves. (Cancer Res 2006; 66(9); 4904-12)

Over the past 2 decades, it has become clear that patients with cancer have detectable antibodies and T cells specific for antigens expressed by autologous tumor cells (1–4). Unlike infection with foreign pathogens, cancers arise from normal host tissues, reflected by the fact that most human tumor antigens identified to date are nonmutated self-antigens (5). T cells with potential to respond to self-antigens typically have low avidity and recognition efficiency and are often maintained in a tolerized state. Inhibition of self-reactivity is also maintained through active suppression by Foxp3⁺CD4⁺CD25⁺ regulatory T cells (Treg; refs. 6–9). Overcoming tolerance or ignorance to self-tumor antigens while minimizing serious autoimmunity is a central challenge in developing cancer immunotherapy.

Glucocorticoid-induced tumor necrosis factor (TNF) receptor family–related gene (GITR) or TNF receptor superfamily member 18 (TNFRSF18) is a type I transmembrane protein with homology to TNF receptor family members (10, 11). GITR is expressed at low levels on resting CD4⁺ and CD8⁺ T cells and up-regulated following T-cell activation. Ligation of GITR provides a costimulatory signal that enhances both CD4⁺ and CD8⁺ T-cell proliferation and effector functions, particularly in the setting of suboptimal T-cell receptor (TCR) stimulation (12–16). In addition, GITR is expressed constitutively at high levels on Tregs and has been explored as a potential target for overcoming Treg suppression. Signaling through GITR, using either agonist anti-GITR antibodies or GITR ligand, abrogates the suppressive effects of Tregs, enhances autoreactive and alloreactive T-cell responses, and exacerbates autoimmunity and graft-versus-host disease (GVHD; refs. 12, 17–21). Whether these effects are due to loss of suppressive activity by Tregs, increased resistance of effector T cells to suppression, or both is currently debated, but the net effect of GITR signaling is the potential for enhanced ability of effector T cells to recognize and respond to self.

We have explored GITR ligation as a strategy to enhance active immunization against cancer. In previous experiments, we showed that treating mice with the agonist anti-GITR mAb DTA-1 at the time of inoculation with a poorly immunogenic tumor led to the rejection of a secondary challenge with the same tumor, a phenomenon called concomitant immunity (22). In the present report, we have combined DTA-1 treatment with active immunization against defined cancer self-antigens to overcome immune tolerance or ignorance and generate more robust antitumor immunity through inhibition of Tregs and/or costimulation of antigen-specific effector T cells. For these studies, we used the clinically relevant melanoma differentiation antigens, gp100 and tyrosinase-related protein 2 (TRP2), also called dopachrome tautomerase, as tumor antigens. For active immunization, plasmids encoding the human orthologues of mouse gp100 and TRP2 were used, as we (23–25) and others (26, 27) have shown that xenogeneic DNA vaccination can induce antibody and T-cell responses against self-antigens and rejection of B16 melanoma, an aggressive, poorly immunogenic tumor. Because protective immunity following gp100 and TRP2 vaccination is primarily dependent on CD8⁺ T cells (24, 25), we sought to characterize the effect of agonist GITR signaling on antigen-specific effector CD8⁺ T-cell responses. We report here that GITR activation during immunization leads to enhanced primary and recall CD8⁺ T-cell responses with associated increases in tumor immunity and autoimmune hypopigmentation. Furthermore, this enhancement is at least partially independent of any effects of anti-GITR antibody on CD4⁺CD25⁺ Tregs.

C57BL/6 mice (6- to 8-week-old females) were from The Jackson Laboratory (Bar Harbor, ME), and Abb−/− C57BL/6 mice (MHC class II deficient) were from Taconic Farms (White Plains, NY). Thy1.1⁺ pmel-1 TCR transgenic mice have been reported (28). For adoptive transfer experiments, 30 × 10⁶ pmel-1 splenocytes were injected by tail vein into naïve Thy1.2⁺ C57BL/6 recipients. Mice were maintained in a pathogen-free vivarium according to NIH Animal Care guidelines. Experiments were done under the governance of an institutional protocol approved by the Memorial Sloan-Kettering Cancer Center (MSKCC) Institutional Animal Care and Use Committee.

The human gp100 (hgp100) expression vector contains full-length hgp100 cDNA cloned into the WRG/BEN vector with a cytomegalovirus (CMV) promoter and kanamycin-resistance gene as described (23, 29). The human TRP2 (hTRP2) or mouse TRP2 (mTRP2) expression vectors contain full-length hTRP2 or mTRP2 cDNA cloned into the pCR3 vector with a CMV promoter and ampicillin-resistance gene (24).

Mice were immunized by particle bombardment as reported (30). Briefly, plasmid DNA was purified and coated onto 1 μm gold particles (Alfa Aesar, Ward Hill, MA) and precipitated on bullets of Teflon tubing. Gold particles containing 1 μg DNA were delivered to each abdominal quadrant (total of 4 μg DNA/mouse) using a gene gun (Accell, PowderMed, Oxford, United Kingdom). Mice were immunized weekly for a total of three immunizations unless otherwise specified.

B16F10/LM3 (hereafter called B16) is a mouse melanoma cell line of C57BL/6 origin derived from the B16F10 line provided by I. Fidler (M.D. Anderson Cancer Center, Houston, TX). EL4, a C57BL/6 mouse lymphoma cell line, was used as an antigen-presenting cell (APC) in T-cell assays. Cell lines were cultured as described (31). For tumor challenge, 0.5 × 10⁵ B16 cells per mouse were injected i.d., and tumor diameters were measured by calipers every 2 to 3 days.

The DTA-1 hybridoma line, created by Shimon Sakaguchi (Kyoto University, Kyoto, Japan), was used to produce mAb by the MSKCC Monoclonal Antibody Core Facility. Affinity-purified DTA-1 (0.25 or 1 mg) in 500 μL sterile PBS was injected i.p. Purified rat IgG (Sigma-Aldrich, St. Louis, MO) was used as a control antibody. For depletion experiments, 250 μg GK1.5 (anti-CD4 mAb) or PC61 (anti-CD25 mAb) was injected i.p. at the indicated time points. Depletion of >98% target T-cell populations at the time of DTA-1 administration (2 days after GK1.5 and 11 days after PC61) was confirmed by flow cytometry (data not shown).

The following fluorochrome-labeled anti-mouse mAbs were from BD Biosciences (San Diego, CA): CD3 (145-2C11), CD8 (53-6.7), CD122 (TM-β1), CD62L (MEL-14), CD44 (IM7), CD107a (1D4B), Thy1.1 (OX-7), and IFN-γ. All stained cells were acquired on a FACSCalibur cytometer (Becton Dickinson, San Jose, CA), except for the Annexin V assay, which used a Cyan cytometer (Dako, Carpinteria, CA), and analyzed using FlowJo (TreeStar, San Carlos, CA). For all assays, spleens and/or draining lymph nodes were harvested from mice (three per group), pooled within groups, crushed, and filtered through 0.22 μm cell strainers. RBC were lysed using an ammonium chloride lysis buffer. Cells were washed twice in RPMI 1640 plus 7.5% FCS before assays. All samples were run in triplicate.

Tetramer assay. Phycoerythrin-conjugated hgp100_25-33-D^b-tetramer, containing the D^b epitope KVPRNQDWL (23, 25), was from Beckman Coulter (Fullerton, CA). Splenocytes and/or lymph node cells were washed twice in fluorescence-activated cell sorting (FACS) buffer (PBS plus 1% bovine serum albumin), and 10⁶ cells in 100 μL FACS buffer were incubated with tetramer and mAbs to T-cell markers for 30 minutes at room temperature in the dark.

Intracellular cytokine assay. Splenocytes and/or lymph node cells (5 × 10⁶) were incubated with 0.5 × 10⁶ irradiated EL4 cells and either with or without 1 μg/mL peptide or with 0.5 × 10⁶ irradiated B16 melanoma cells with 10 μg/mL Brefeldin A (Sigma-Aldrich) added after 1 hour. After 12 to 16 hours at 37°C, cells were stained for surface markers, fixed and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences), and stained for intracellular IFN-γ.

CD107a mobilization assay. The procedure was adapted from previously published reports (32, 33). Splenocytes and/or lymph node cells (10⁶) were incubated in 96-well plates with 0.1 × 10⁶ irradiated EL4 cells with or without 1 μg/mL peptide and 1.25 μg/mL FITC-conjugated anti-mouse CD107a antibody (total of 200 μL/well). After 1 hour at 37°C, 1 μL of 2 mmol/L monensin sodium solution (Sigma-Aldrich) was added per well. After 6 hours of incubation, cells were stained for surface markers and intracellular IFN-γ using the Cytofix/Cytoperm kit.

Annexin V assay for activation-induced cell death. Cells (10⁶) were added to 96-well plates with 0.1 × 10⁶ irradiated EL4 cells and stimulated either with or without 1 μg/mL peptide. After 16 hours at 37°C, cells were washed twice in FACS buffer and stained with mAbs to surface markers, washed in 1× Annexin V buffer (BD Biosciences), and stained with Annexin V allophycocyanin (BD Biosciences) and 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich). The apoptotic population was defined as Annexin V positive and DAPI negative.

Differences in tumor-free survival were evaluated by log-rank analysis of Kaplan-Meier survival curves (GraphPad Prism 4.0). Differences in number of depigmented mice in hgp100-immunized groups were analyzed by two-sided Fisher's exact test. For T-cell assays, statistical differences between groups were determined by analyzing means of replicates by two-tailed Student's t test.

Anti-GITR mAb enhances effector CD8⁺ T-cell responses to self-tumor antigens. We have shown previously that xenogeneic DNA immunization of mice can induce immune responses to self-differentiation antigens leading to antibody and T-cell–mediated immunity against syngeneic, poorly immunogenic tumors (24, 25, 29). For the melanoma differentiation antigens gp100 and TRP2, activated CD8⁺ T cells (but not CD4⁺ T cells or antibodies) are required at the time of tumor challenge for tumor rejection (24, 25).⁶

Therefore, we sought to determine if signaling through GITR during immunization enhances CD8⁺ T-cell responses against these antigens. Mice were immunized thrice at weekly intervals with hgp100 DNA and treated with 0.25 or 1 mg DTA-1 anti-GITR mAb or a control antibody (rat IgG, 1 mg) 1 day before the second immunization (day 6). Five days after the final vaccination, CD8⁺ T-cell responses were measured against the immunodominant gp100_25-33 D^b-restricted epitope (human, KVPRNQDWL; mouse, EGSRNQDWL; ref. 23).

By direct staining with hgp100_25-33-D^b-tetramer (Fig. 1A), the frequency of hgp100-specific CD8⁺ T cells was significantly increased in mice receiving DTA-1 (Fig. 1B). In repeated experiments, a 2- to 4-fold enhancement in CD8⁺ T-cell responses was seen both locally in draining lymph nodes and systemically in the spleen. Enhancement was greatest in the spleen with the 1-mg DTA-1 dose, with comparable enhancement of responses in lymph nodes at both doses (Fig. 1B; data not shown). Similar results were seen for effector function measured by intracellular IFN-γ production (Fig. 1C,, left) and CD107a mobilization (a surrogate for lytic degranulation; Fig. 1D; refs. 32, 33). These results show that DTA-1 enhances responses against the relevant mouse self-epitope. In addition, combining DTA-1 with hTRP2 DNA vaccination led to an increase in CD8⁺ T cells producing IFN-γ following restimulation with the mouse TRP2_181-188 K^b-restricted epitope (34) as well as restimulation with B16 melanoma cells (Fig. 1C , right). These observations show that these effects are not limited to a single antigen and that in vivo anti-GITR treatment can enhance vaccine-induced CD8⁺ T-cell responses against B16 melanoma.

Anti-GITR mAb enhances rejection of B16 melanoma. Although xenogeneic DNA immunization can induce immune responses to self, complete rejection of a subsequent i.d. challenge with B16 melanoma typically occurs in only 30% to 60% of hgp100- or hTRP2-immunized mice (23, 25). Injecting 1 mg DTA-1 improved tumor protection for mice immunized with hgp100 (P = 0.05, by log-rank analysis) and hTRP2 (P = 0.01) compared with mice receiving a control antibody (Fig. 2A and B). The 0.25-mg DTA-1 dose did not significantly enhance long-term tumor-free survival. In four independent experiments using 1 mg DTA-1, tumor protection was consistently improved by 20% to 40%, with 30 of 40 (75%) total mice remaining tumor-free 60 days after challenge compared with 15 of 40 (38%) control antibody-treated mice (P = 0.02) and 0 of 40 untreated mice. In three independent experiments combining 1 mg DTA-1 with hTRP2 immunization, 24 of 30 (80%) DTA-1-treated mice remained tumor-free at 60 days compared with 15 of 30 (50%) control antibody-treated mice (P = 0.01) and 1 of 30 untreated mice. In contrast, combining 1 mg DTA-1 with syngeneic mTRP2 immunization failed to induce any TRP2-specific CD8⁺ T-cell response or protection from tumor challenge (data not shown). This indicates that GITR ligation alone is not sufficient to break tolerance in this setting but that immunizing with the xenogeneic form of the antigen remains critical.

Our previous experiments using anti-CTLA-4 mAb in combination with xenogeneic gp100 DNA vaccination showed the importance of timing of antibody administration, with enhancement of CD8⁺ T-cell responses and tumor immunity when the mAb was given with the second, but not the first, immunization (35). Timing of DTA-1 administration was similarly critical. Injecting the antibody after completion of the three immunizations (1 day before the tumor challenge) or just before the third immunization failed to enhance T-cell responses or protection (data not shown). Interestingly, treatment with 1 mg DTA-1 1 day before the first (priming) immunization not only persistently failed to enhance tumor rejection but also, in two of five experiments, led to a significant loss of the protection compared with immunization alone (Fig. 2C). Finally, no improvement in tumor protection was observed in control groups treated with DTA-1 mAb alone (in the absence of vaccination) at equivalent time points before tumor challenge (i.e., 1, 13, and 20 days before the tumor challenge; data not shown).

One explanation for the loss of protection with anti-GITR mAb before the first vaccination is that strong GITR signaling at initial activation of naïve effector T cells can promote activation-induced cell death (AICD). This phenomenon has been shown in CD4⁺ effector cells in vivo (19, 21) and in vitro (13, 15). To address this possibility, we adoptively transferred splenocytes from Thy1.1⁺ pmel-1 transgenic mice, in which 60% to 80% of CD8⁺ T cells have a TCR specific for the immunodominant gp100_25-33 epitope (28), into naïve Thy1.2⁺ C57BL/6 mice and vaccinated recipients with hgp100 DNA and 1 mg DTA-1 (Fig. 2D). The proportion of donor CD8⁺ T cells undergoing apoptosis (Annexin V positive and DAPI negative) following restimulation with mgp100_25-33 peptide was significantly increased when GITR ligation was initiated before the first vaccination compared with mice receiving no DTA-1 or DTA-1 before the second vaccination. This finding supports differential susceptibility of antigen-specific CD8⁺ T cells to AICD as an explanation for the negative effect on tumor immunity of GITR ligation at the first immunization compared with the second immunization.

Anti-GITR mAb enhances autoimmune hypopigmentation following hTRP2 DNA immunization. Immunization with hTRP2 DNA typically induces a CD8⁺ T-cell–dependent autoimmune response against normal melanocytes, manifested as coat hypopigmentation that appears 3 to 4 weeks after vaccination (24). Administration of DTA-1 with hTRP2 immunization led to an increase in the intensity and distribution of hypopigmentation with greater spreading of hypopigmentation to nondepilated areas (Fig. 2E). No hypopigmentation was seen following hgp100 immunization plus IgG, but 10 of 40 mice (from four independent experiments) developed abdominal hypopigmentation following hgp100 DNA plus DTA-1 given just before the second immunization (P = 0.001, by two-sided Fisher's exact test). This observation provides further evidence that signaling through GITR during active immunization enhances breaking of tolerance or ignorance against self-antigens.

Anti-GITR mAb treatment during immunization enhances memory CD8⁺ T-cell responses. The “programming” of memory CD8⁺ T cells is influenced by CD4⁺ T-cell help, strength and persistence of antigenic stimulation, magnitude of the initial effector response, and Tregs (36–38). We hypothesized that by influencing the latter two factors using an anti-GITR mAb, we could enhance the generation of a memory CD8⁺ T-cell population against self-tumor antigens. When mice were immunized with hgp100 DNA and DTA-1, significantly higher numbers of functional gp100-specific CD8⁺ T cells were seen 5, 12, 19, and 33 days following the final immunization (Fig. 3A and B). We then assessed the effect of GITR signaling during primary immunization on CD8⁺ T-cell recall responses using the immunization schedule shown in Fig. 4A. The CD8⁺ T-cell recall (day 33) response following a booster immunization was significantly greater in mice that received DTA-1 during the primary immunization series 4 weeks earlier (P = 0.004; Fig. 4B). The same increased response was also seen by intracellular cytokine assay for IFN-γ production (data not shown).

Qualitative effects of DTA-1 treatment on the memory CD8⁺ T-cell population were assessed by phenotypic analyses of hgp100-tetramer⁺ CD8⁺ T cells (Fig. 4C , representative plots). At day 33 after immunization, a greater proportion of gp100-specific CD8⁺ T cells from DTA-1-treated mice expressed an activated effector/memory phenotype compared with IgG-treated mice [mean percentage CD44^hiCD122^hi, 10.2 ± 1.2% for DTA-1 versus 7.3 ± 0.4% for IgG (P = 0.08); mean percentage CD44^hiCD62L^lo, 38.1 ± 2.3% for DTA-1 versus 25.3 ± 1.4% for IgG (P = 0.009)]. These differences in activation and memory markers were maintained in mice that have received a booster vaccination at day 28 [mean percentage CD44^hiCD122^hi, 45.6 ± 1.2% for DTA-1 versus 29.8 ± 0.7% for IgG (P = 0.0003); mean percentage CD44^hiCD62L^lo, 48.5 ± 1.3% for DTA-1 versus 41.5 ± 0.4% for IgG (P = 0.007)]. These differences in phenotype suggest that providing an agonist signal through GITR during the primary phase of immunization against self-antigens not only produces quantitative increases in antigen-specific memory CD8⁺ T cells but also enhances their ability to achieve an activated, effector state on reexposure to antigen.

However, these effects did not directly translate into improved survival from tumor challenge (Fig. 4D). Mice treated with DTA-1 during primary immunization, boosted with hgp100 DNA at day 28, and challenged with B16 at day 33 had a delay in tumor growth compared with mice treated with control antibody but ultimately had no significant improvement in long-term tumor-free survival (P = 0.34). Both groups of mice receiving booster immunizations, with or without previous DTA-1, had poorer long-term tumor-free survival (17% at day 60) than a group challenged during the peak effector response 5 days after a primary hgp100 immunization series (40%; P = 0.10; data not shown).

In an attempt to further strengthen memory responses, we investigated GITR ligation at the time of the CD8⁺ T-cell recall response. Mice were immunized as in Fig. 4A, except that groups also received DTA-1 or IgG 1 day following the day 28 booster vaccination. GITR ligation during the primary immunization series (“DTA-1 with primary”) again led to modest enhancement of gp100-specific CD8⁺ T-cell recall responses in both spleen (Fig. 5A) and draining lymph nodes (data not shown). A similar magnitude of enhancement was seen when mice received DTA-1 only with the booster vaccination (“DTA-1 at boost”; Fig. 5A). When anti-GITR mAb was given during both primary and booster immunization, CD8⁺ T-cell recall responses were markedly enhanced (mean percentage tetramer⁺ CD8⁺ T cells in the spleen, 2.92 ± 0.08%; P < 0.001, compared with the other three boosted groups), suggesting a synergistic effect of ligating GITR during both primary and recall response. A similar pattern of CD8⁺ T-cell responses was seen in draining lymph nodes (data not shown) as well as following overnight restimulation with mouse gp100_25-33 peptide (Fig. 5B) or irradiated B16 melanoma cells (Fig. 5C).

This strategy was then tested for rejection of B16 tumor challenge (Fig. 5D). In two experiments, ligating GITR during both primary and recall response led to an increase in tumor-free survival compared with mice receiving control antibody (P = 0.05). Injecting DTA-1 only with the primary or recall immunization did not significantly enhance tumor-free survival, although a trend toward improvement (P = 0.11) was again seen when DTA-1 was given during the primary immunization series. These observations show the potential of this approach for strengthening recall responses against self-tumor antigens.

Enhancement of CD8⁺ T-cell responses by anti-GITR antibody is only partially dependent on CD4⁺ cells. In previous studies, DTA-1 has been shown to have effects on multiple T-cell populations, including inhibition of suppressive activity of CD4⁺CD25⁺ Tregs and costimulation of CD4⁺CD25⁻ and CD8⁺ effector T cells (12–19, 21). We wished to determine if the observed effects on CD8⁺ T-cell responses were due to a direct effect of DTA-1 on the responding CD8⁺ T cells, an indirect effect through CD4⁺ cells, or both. Depletion of CD4⁺ cells before immunization or immunization of MHC class II-deficient (Abb−/−) mice, which lack functional CD4⁺ T cells, led to a reduction in CD8⁺ T-cell responses to near-baseline levels (presumably due to a lack of CD4⁺ help during initial priming), making it difficult to discern specific effects of DTA-1 (data not shown). We therefore developed a modified CD4 depletion schedule, which allows CD4⁺ T-cell help to be initiated against melanoma antigens during priming following the first immunization (22) but depletes CD4⁺ cells by the time DTA-1 is given (day 6) and maintains CD4⁺ depletion for the duration of the immunizations (data not shown). DTA-1-induced enhancement of gp100-specific CD8⁺ T-cell responses was observed even in the absence of CD4⁺ cells, consistent with a direct effect on DTA-1 on CD8⁺ T cells in vivo (Fig. 6A and B). However, a contribution from CD4⁺ cells cannot be ruled out because the magnitude of the increase in gp100-specific CD8⁺ T cells was lower in CD4-depleted mice (mean percentage tetramer⁺ CD8⁺ cells, 7.35 ± 0.29% in DTA-1-treated mice versus 5.5 ± 0.07% in DTA-1 plus anti-CD4-treated mice; P = 0.004). No conclusions could be drawn about the cellular targets of DTA-1 that mediate the enhanced tumor rejection because the modified CD4 depletion schedule abrogated any protection from B16 challenge provided by the hgp100 immunization despite expansion of gp100-specific effector CD8⁺ T cells (data not shown).

We next investigated the effects of depleting CD25⁺ cells before immunization. DTA-1 injection still enhanced both CD8⁺ T-cell responses (Fig. 6C) and B16 melanoma rejection (Fig. 6D) in the absence of CD25⁺ cells, showing that GITR ligation on CD4⁺CD25⁺ Tregs is not necessary for its effect. Interestingly, CD25 depletion without DTA-1 also led to enhanced tumor protection despite no apparent quantitative effect on peak CD8⁺ T-cell responses. This suggests that depleting Tregs before the vaccination against self-antigens can lead to greater tumor immunity, perhaps due to qualitative effects on activated CD8⁺ cells (e.g., greater avidity or effector function and improved memory) and/or effects on other effector cells, such as CD4⁺ T cells, natural killer cells, or B cells. Thus, DTA-1-induced enhancement of CD8⁺ T-cell responses can occur independently of any effect on CD4⁺ cells (including CD4⁺CD25⁺ Tregs), although this enhancement is weaker in the absence of CD4⁺ cells consistent with one or more GITR⁺CD4⁺ populations also playing a role.

We report that giving an agonist anti-GITR antibody during immunization enhances both effector and memory CD8⁺ T-cell responses against defined self-antigens expressed by tumors and that these responses translate into improved rejection of a poorly immunogenic tumor. Although GITR ligation can abrogate Treg-mediated suppression and increase the proliferation of both CD4⁺ and CD8⁺ T cells in vitro (12–19, 21), less is known about the effects of GITR ligation in vivo on T-cell function, particularly CD8⁺ T cells. Muriglan et al. (19) showed that the proliferation of alloreactive CD8+ T cells following adoptive transfer into MHC-mismatched recipients was increased with in vivo administration of anti-GITR antibody along with the severity of CD8⁺ T-cell–mediated GVHD. Giving DTA-1 in combination with adoptive transfer of viral antigen-specific CD8⁺ T cells into virus-infected mice led to greater production of IFN-γ and TNF and greater reduction of viral load than adoptive transfer alone (39). Together with our results, these studies show that GITR ligation can lead to enhanced CD8⁺ T-cell effector function in vivo whether the T-cell specificity is for an alloantigen, viral antigen, or self-antigen expressed by tumors.

We observed enhanced immunity when DTA-1 was given just before the second vaccination (day 6) but no effect or worsened tumor protection (Fig. 2C) and CD8⁺ T-cell responses (data not shown) when the antibody was given before the initial priming vaccination (day 1). GITR is expressed constitutively at high levels by CD4⁺CD25⁺ Tregs and at low levels on resting CD4⁺CD25⁻ and CD8⁺ T cells. Up-regulation of GITR begins 6 hours after TCR engagement in vitro and peaks at 72 hours (13, 16). Considering the time required for antigen production, processing, and cross-presentation by APCs following DNA vaccination, it is likely that GITR expression on activated effector T cells is still high at the time of anti-GITR antibody administration on day 6. In contrast, ligation of GITR before the initial priming immunization, in the absence of any TCR stimulation, may not only fail to costimulate T cells during subsequent activation but also impair activation and expansion, perhaps by interfering with normal GITR-GITR ligand interactions between T cells and APCs important for priming T-cell responses (16). Support for this hypothesis comes from Shimizu et al. (17) who noted that the ability of DTA-1 to abrogate Treg suppression in coculture assays was only seen when DTA-1 was added to the culture together with anti-CD3 antibody and irradiated APCs. When effector cells were preincubated with DTA-1 followed by anti-CD3-mediated activation, Tregs still suppressed proliferation.

An alternative explanation is that early GITR ligation followed by repetitive restimulation with antigen may increase susceptibility of T cells to AICD. The effects of GITR ligation on apoptotic signals remain unclear, with reports of both antiapoptotic (via nuclear factor-κB activation; refs. 10, 40–42) and proapoptotic effects (via interactions with Siva; ref. 43). What is clear is that the in vitro effect of GITR ligation on CD4⁺CD25⁻ and CD8⁺ T-cell proliferation is most pronounced under conditions of suboptimal TCR stimulation regardless of the presence of Tregs. When strong TCR stimulation is combined with higher concentrations of anti-GITR antibody, T-cell proliferation can actually decrease (13, 15). This effect has also been seen in vivo. Muriglan et al. (19) showed that GITR ligation paradoxically ameliorated GVHD mediated by CD4⁺CD25⁻ T cells. DTA-1 treatment was inducing apoptosis in alloactivated CD4⁺CD25⁻ cells. Similarly, Valzasina et al. (21) found that GVHD mediated by alloreactive CD4+CD25 T cells was exacerbated by a low DTA-1 dose (300 μg) but mitigated by a high dose (1,200 μg), again suggesting AICD of the alloreactive cell population with stronger GITR signaling. Our data show that DTA-1 given before the first, but not the second, immunization, increases the susceptibility of antigen-specific CD8⁺ T cells to AICD (Fig. 2D). We are currently investigating whether this phenomenon is responsible for the differences observed with the administration of DTA-1 before the first immunization versus the second immunization.

The circumstances surrounding the initial priming of effector T cells profoundly influence the subsequent memory T-cell population. Factors important during priming include CD4⁺ T-cell help, the duration and strength of antigenic stimulation, costimulatory molecules, the magnitude of the initial response, and the expression of cytokines and cytokine receptors (e.g., interleukin-7 receptor α; refs. 36, 37). In addition, depleting Tregs either during the primary response or at the time of a recall response can enhance the magnitude and function of memory CD8⁺ T cells recognizing pathogens (38, 44), implying a role for Tregs in memory T-cell generation.

Our work shows that a single dose of agonist anti-GITR mAb during a primary immunization series increases not only the magnitude of the initial effector CD8⁺ T-cell response but also the persistence of functional antigen-specific T cells (Fig. 3A and B). Moreover, GITR ligation during primary hgp100 DNA immunization led to a greater recall response following booster immunization 4 weeks later, with a higher proportion of gp100-specific CD8⁺ T cells expressing an activated effector/memory (CD44^hiCD62L^lo and CD44^hiCD122^hi) phenotype (Fig. 4B and C). Ultimately, however, both recall CD8⁺ T-cell responses and tumor rejection after secondary boosting were weaker than at the peak of the effector response (4-5 days after the primary immunization series). This pattern of weak memory response is different from the typically robust recall responses against viral and bacterial pathogens (37) and likely reflects inherent difficulties in maintaining a functional, nontolerized memory population specific for constitutively expressed self-antigens. Recall CD8⁺ T-cell responses were improved, however, by ligating GITR during both primary and booster vaccination, approaching levels induced during the peak effector response (Fig. 5). This combined approach also led to improved memory CD8⁺ T-cell responses against B16 melanoma ex vivo (Fig. 5C) as well as significantly improved protection following late (day 33) tumor challenge after a boost (Fig. 5D), showing that these difficulties can be partially overcome through this approach.

The question of which cell subpopulations the anti-GITR antibody is primarily acting upon is debated. Although the ability of GITR ligation to directly costimulate CD4⁺CD25⁻ and CD8⁺ T cells has been well documented, the notion that anti-GITR therapy renders CD4⁺CD25⁺ Tregs unable to suppress (17, 18) has recently been challenged by in vitro data suggesting that GITR ligation allows effector T cells to resist suppression rather than directly affecting Tregs themselves (16). In contradistinction, a recent study using DTA-1 in a GVHD model supports a direct effect of DTA-1 on Tregs (21).

To address this issue, we depleted CD4⁺ cells during hgp100 DNA immunization and found that CD8⁺ T-cell responses were still enhanced by DTA-1 (Fig. 6A and B), supporting a direct effect on CD8⁺ effector T cells. The magnitude of enhancement was lower than in non-CD4-depleted mice, implying that GITR signaling in one or more CD4⁺ populations may also play a role. The enhancement of CD8⁺ T-cell responses (Fig. 6C) and tumor immunity (Fig. 6D) by DTA-1 despite previous depletion of CD25⁺ cells also indicates that GITR ligation on naturally arising Tregs is not required for its effects. In addition, although we did not observe an additive effect of combined PC61 (anti-CD25 mAb) and DTA-1 treatment, this may be due to the concomitant depletion by PC61 of some effector CD4⁺ and CD8⁺ T cells that up-regulated CD25 following activation, diluting the direct costimulatory effect of DTA-1. We speculate that both blockade of Treg suppressive activity and costimulation of effector T cells are contributing to the enhancement of active immunization by DTA-1. To identify the cellular targets of DTA-1 in this system, it will be necessary to test the DTA-1 plus immunization strategy in mice reconstituted with various combinations of GITR-positive and GITR-negative T cells. We are currently backcrossing GITR−/− mice onto the C57BL/6 background for this purpose.

In conclusion, administration of an agonist anti-GITR mAb during DNA immunization against self-antigens expressed by tumors augments vaccine-induced CD8⁺ T-cell responses and immunity against an aggressive tumor and represents a potentially novel immunotherapeutic strategy against cancer. This approach can increase autoimmunity as well (Fig. 2E; refs. 12, 17), requiring thoughtful choice of vaccination targets and careful monitoring for autoimmune toxicity. Finally, DTA-1 antibody given alone (without any vaccination) after tumor challenge can impede tumor growth and modestly increase tumor-induced T-cell responses (45).⁷

Thus, GITR ligation may provide a platform on which to build a successful combination immunotherapy designed to eradicate established tumors.

Grant support: American Society of Clinical Oncology Young Investigator Award, Charles A. Dana Foundation, H-C Fellowship, and National Cancer Institute grant T32 CA009512 (A.D. Cohen); NIH grant K08CA10260 and Byrne Fund (MSKCC; M-A. Perales); Damon Runyon-Lilly Clinical Investigator Award (J.D. Wolchok); NIH/National Cancer Institute grants CA56821, CA47179, CA33049, and CA59350 and Damon Runyon-Lilly Foundation (A.N. Houghton); Swim Across America; Lita Annenberg Hazen Foundation; TJ Martell Foundation; Louis & Anne Abrons Foundation; Mr. and Mrs. Quentin J. Kennedy Fund; Mr. William H. Goodwin and Mrs. Alice Goodman Fund and Commonwealth Cancer Foundation for Research/Experimental Therapeutics Center of MSKCC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Rodica Stan for editorial assistance.