Central Role of Tumor-Associated CD8⁺ T Effector/Memory Cells in Restoring Systemic Antitumor Immunity¹

The majority of tumor immunotherapy protocols are designed to induce antitumor T cell responses due to the exquisite specificity and potent cytotoxicity of T cells (1). However, such vaccines have generally been ineffective in achieving regression of established tumors in murine tumor models and in cancer patients (2). The observed lack of therapeutic efficacy is not simply due to the qualitative and/or quantitative inferiority of vaccine-induced antitumor T cells but rather appears to be associated with the immune-suppressive characteristics of the tumor microenvironment (3, 4); that is, even when potent and long-lasting tumor-specific T cell activity is generated, vaccine-induced cells are either unable to accumulate in tumors or are inactivated rapidly upon infiltration into tumor (3, 4, 5, 6). Functional analyses of either vaccine-induced or natural tumor-infiltrating T cell populations have demonstrated that their quiescent phenotype is primarily associated with signaling defects (7, 8). At the same time, others have shown that this phenotype is reversible and that purification and ex vivo culture of tumor-infiltrating T lymphocytes (TIL)³ can result in the recovery of their cytotoxic function (9, 10). To this end, T cells expanded from tumors have been utilized in adoptive T cell transfer therapy to achieve durable clinical regressions in melanoma patients (11).

Whether activation of TIL in situ can overcome tumor-mediated immune suppression and achieve tumor regression has been addressed in several studies. Targeting of inhibitory/regulatory mechanisms in tumors, including selective elimination of T suppressor cells or blocking of coinhibitory molecules can rescue concomitant immunity and lead to tumor regression even in the absence of additional immune stimulation (12, 13, 14). Separately, targeting of intratumoral T cells with proinflammatory cytokines or TLR ligands has also been shown to overcome tumor immune suppression and activate TIL (15, 16, 17). These findings demonstrate that cytotoxic function of preexisting tumor-infiltrating T cells can be restored effectively in vivo. To this end, recent studies in our laboratory revealed that sustained delivery of IL-12 and GM-CSF directly to the tumor microenvironment induced a rapid yet transient activation of tumor-associated CD8⁺ T effector memory cells (Tem), the concurrent elimination of tumor-infiltrating CD4⁺CD25⁺Foxp3⁺ T suppressor cells, and the subsequent infiltration of tumors with CD8⁺ T cells displaying full effector phenotype (18). Importantly, restoration of TIL function not only resulted in the eradication of advanced s.c. primary tumors (15) but also in the elimination of established lung metastases, consistent with the notion that TIL-mediated tumor kill results in the priming of systemic effector responses (19). Collectively, the above studies identify TIL as a potent resource for the induction of effective tumor regression.

Whereas the above studies established that intratumoral IL-12/GM-CSF promoted both immediate and latent T cell activity in tumors, the mechanistic link between these responses was not defined. More specifically, whether secondary T effectors developed from a subset of preexisting TIL or were primed de novo in tumor-draining lymph nodes (TDLN) following TIL-mediated tumor destruction was not clear. Previous studies demonstrated that IL-12/GM-CSF-mediated activation of preexisting CD8⁺ Tem was transient due to rapid apoptosis (18). This finding, combined with the lack of an increase in intratumoral CD8⁺ T cell numbers until day 5, supported the notion that the latent CD8⁺ T effector cell expansion was likely due to a second wave of T cells arriving from the draining lymph nodes. However, these data could not rule out the possibility that a small subset of preexisting CD8⁺ Tem avoided apoptotic cell death and proliferated, becoming detectable only later. Here we undertook a detailed analysis of the quantitative and functional changes in tumor and TDLN CD8⁺ T cells in posttherapy mice to test the hypothesis that activation of preexisting tumor-resident CD8⁺ Tem is important not only to local tumor destruction but also to subsequent re-priming of systemic antitumor T cell immunity. The results demonstrate that CD8⁺ T effector cells infiltrating the tumors on day 7 posttherapy were in fact primed in the TDLN and, importantly, activation of preexisting tumor-associated CD8⁺ Tem was essential to the development of the secondary response.

Six- to 8-wk-old male or female BALB/c mice were used in the experiments. Thy1.2 BALB/c and IFN-γ-deficient mice in BALB/c background were purchased from The Jackson Laboratory. Thy1.1 BALB/c mice (a gift from Dr. Richard Dutton, Trudeau Institute, Saranac Lake, NY) and perforin-deficient BALB/c mice (a gift from Dr. John Harty, University of Iowa) were bred at our institute. Mice were injected with 1 × 10⁶ line 1 cells (a syngeneic BALB/c lung carcinoma) or 4T1 (a syngeneic BALB/c mammary carcinoma) in 0.1 ml DMEM behind the neck just above the scapula (or in the flank in some cases) as described previously (18). Tumors were monitored until they reached 300–400 mm³ in size and were treated.

Microspheres with a cytokine loading of 0.025% (w/w) were prepared using the phase inversion nanoencapsulation method as described previously (18). Mice were treated with 4 mg of each microsphere preparation (1 μg equivalent of each cytokine) suspended in 0.1 ml of sterile PBS via direct injection into the tumor.

Single-cell suspensions from tumor were prepared by enzymatic digestion as described previously (18). For TDLN, brachial and axillary (or inguinal in the case of flank tumors) lymph nodes draining 14-day s.c. line 1 tumors (300–400 mm³) were removed, mechanically teased apart in DMEM, and filtered through 70-μm mesh.

Mice were sacrificed at indicated times and single-cell suspensions were prepared. Samples were labeled with Abs to various T cell markers and were analyzed on a four-color FACSCalibur flow cytometer (BD Biosciences). The following panel of commercially available and directly fluorochrome-conjugated anti-mouse mAbs was included in this study: CD3 (clone 17A2), CD8 (clone 53-6.7), CD69 (clone H1.2F3), CD62L (clone MEL-14), CCR5 (clone C34-3448), and Thy1.1 (clone OX-7). All Abs were purchased from BD Pharmingen. Flow cytometry data were analyzed using CellQuest software (BD Biosciences). Intracellular cytokine staining for granzyme B was performed as described previously (18).

Animals were injected with BrdU in 0.1 ml of sterile PBS (100 mg/kg body weight) i.p. 4 h before sacrifice. Immunostaining for BrdU was performed using the BD Pharmingen BrdU flow kit as per the supplier’s instructions.

Single-cell suspensions were pepared from posttherapy day 3 axillary and brachial TDLN. CD8⁺ T effector cells were magnetically separated using the CD8a (Ly-2) MicroBeads with autoMACS Pro separator as per the manufacturer’s instructions (Miltenyi Biotec). The purity of the CD8⁺ cell preparation was >95% as determined by flow cytometry. Target line 1 cells were incubated in DMEM/F12 plus 10% FBS containing 5 U/ml IFN-γ (PeproTech) for 48 h to induce MHC class I expression (20), and they were stained with PKH67 using the PKH67-GL dye kit (Sigma- Aldrich) as recommended by the manufacturer. PKH67-labeled target cells were then suspended at 4 × 10⁵ cells/ml in DMEM/F12 plus 10% FBS, and 0.1 ml of aliquots was added to 96-well plates. CD8⁺ T effector cells at various concentrations were mixed with target cells to measure cell-mediated cytotoxicity by a dye exclusion assay as previously described by us (21). Briefly, cells were incubated at 37°C for 24 h and stained for 7-aminoactinomycin D before analysis by flow cytometry. Cytotoxicity was reported as percentage dead cells within the PKH67-positive targets (dead-labeled targets/dead-labeled targets + live-labeled targets × 100). Percentage target cell death was corrected for background (spontaneous death) by subtracting the percentage of dead cells in control samples (PKH26-labeled targets alone) from the percentage of dead cells within the test samples.

Tumors were surgically resected after reaching a size of 300–400 mm³, as described previously (22). Mice were allowed to recover from surgery for 48 h and treated via direct injection of microspheres into the site of resection.

CD8⁺ T cell isolation from both treated and control TDLN was performed using the CD8a (Ly-2) MicroBeads with an autoMACS Pro separator as above. Tumor-bearing Thy1.2 mice received i.p. transfer of Thy1.1 CD8⁺ T cells (3 or 4 × 10⁶ cells in 0.1 ml of PBS) on days indicated in the text. Some recipients were depleted of CD8⁺ T cells via injection of anti-CD8 Ab (22) before adoptive transfer. At the designated time points after adoptive transfer, mice were sacrificed and checked for in vivo trafficking by flow cytometry analysis.

Stimulation of day 4 posttherapy TDLN CD8⁺ T cells by tumor lysate-pulsed monocytes was performed as previously described by us (19) except that CD8⁺ T cells were purified from the TDLN of treated mice and were cocultured with APCs for 16 h. Following stimulation, CD8⁺ T cell activation was analyzed by intracellular IFN-γ staining (18) without PMA/ionomycin treatment. The ability of TDLN dendritic cells obtained from CD8-depleted and control mice to prime naive T cells was determined using CD8⁺ T cells expressing a TCR specific for the influenza virus hemagglutinin (HA) peptide IYSTVASSL (23) and peptide-pulsed DC (24). Briefly, responder CD8⁺ T cells were isolated from spleens of clone 4 mice (23) with magnetic beads and labeled with CFSE (24). Stimulator CD11c⁺ dendritic cells were purified from the lymph nodes of CD8-depleted or control mice and cultured with HA peptide (10 μg/ml) for 1 h at 37°C. Stimulator and responder cells were then mixed at a 1:5 ratio and incubated at 37°C for 2 days. CD8⁺ T cell proliferation (as determined by CFSE dilution) was then analyzed by flow cytometry.

Student’s t test was utilized to determine the significance of the differences between groups, and p ≤ 0.05 was considered significant.

Intratumoral delivery of IL-12 plus GM-CSF promoted rapid activation of tumor-associated CD8⁺ Tem with subsequent CD8⁺ T effector cell expansion. To determine whether the secondary CD8⁺ T effector cell wave was associated with the priming of a de novo T cell response in the TDLN, functional kinetics of draining lymph node CD8⁺ T cells were analyzed. The absolute numbers of and the phenotypic changes in CD8⁺ T cells were quantified by flow cytometry in both the tumor and the draining lymph nodes between days 0 (pretreatment) and 10. The results are shown in Fig. 1. Treatment resulted in a 4-fold increase in the absolute numbers of CD8⁺ T cells in the TDLN on day 3 posttreatment and declined gradually thereafter (Fig. 1,A). A similar increase in the numbers of TDLN CD4⁺ T cells was also observed, demonstrating that posttherapy T cell expansion was not limited to the CD8⁺ subset (Fig. 1,A, inset). In contrast, the number of tumor-associated CD8⁺ T cells did not change between days 0 and 4 but increased 2.5-fold between days 4 and 7 (Fig. 1,A). The sequential nature of CD8⁺ T cell expansion, first in the TDLN and then in the tumor, supported the hypothesis that CD8⁺ T effector priming in the TDLN was followed by migration of these cells to the tumor. To determine whether expansion was concurrent with activation, TDLN CD8⁺ T cells were analyzed for changes in the expression of early activation markers CD69 and CD62L. The data shown in Fig. 1,B demonstrate that expansion was accompanied with an 8-fold increase in the proportion of CD8⁺ T cells displaying a CD69⁺CD62L⁻ phenotype, consistent with activation. Priming of T cells in the lymph node is followed by migration to the periphery (25). To this end, TDLN CD8⁺ T cells were also analyzed for the manifestation of a CD62L⁻CCR5⁺ phenotype (26). The data shown in Fig. 1, C and D, demonstrate that treatment resulted in a rapid decline in membrane CD62L expression with concomitant CCR5 up-regulation on day 4, consistent with the development of a periphery-homing phenotype. Collectively, these data support the notion that treatment resulted in rapid priming of a CD8⁺ T cell response in the lymph node with egress to the periphery on day 4.

Whereas the above data established that intratumoral IL-12/GM-CSF promoted T cell priming in the TDLN, these experiments did not determine whether primed cells directly homed to tumors. To this end, BrdU labeling studies were performed to reveal if cells labeled with BrdU in the TDLN during early stages of priming eventually infiltrated the tumor. Initially, pulse labeling of CD8⁺ T cells was performed on days 0–7 to delineate the proliferation kinetics of tumor-resident vs TDLN CD8⁺ T cells. Mice were treated with a single injection of BrdU daily and the CD8⁺ T cells were harvested from tumors and the TDLN 4 h after BrdU injection on each day for analysis. The results are shown in Fig. 2,A. These data demonstrate that tumor-associated CD8⁺ T cells did not proliferate in response to treatment, as the percentage of BrdU-positive cells did not change significantly between day 0 (pretreatment) and day 7. In contrast, TDLN cells divided rapidly with maximum proliferation occurring on day 2 (Fig. 2,A). Proliferation continued between days 3 and 5, albeit at lower levels, and returned to background levels on day 7 posttherapy. The finding that CD8⁺ T cells proliferated in the TDLN but not in the tumor allowed us to determine whether the TDLN cells eventually migrated to tumors. To this end, CD8⁺ T cells were pulse labeled on day 3 (1 day before the development of full peripheral migratory phenotype) and the percentages of BrdU-positive CD8⁺ T cells in the tumor and the TDLN were determined on days 3 (4 h after labeling) and 7 (4 days after labeling). The results are shown in Fig. 2,B. Whereas the percentage of CD8⁺ T cells that were BrdU⁺ decreased from an average of 25% to 6.6% in the TDLN between days 3 and 7, their proportion increased from 8.5% to 20.8% in the tumor during the same period (Fig. 2,B). The >9-fold change in the tumor-to-TDLN ratio of BrdU⁺ CD8⁺ T cells between days 3 and 7 established that the TDLN cells that had incorporated BrdU on day 3 left the TDLN and migrated to the tumor. These findings, that is, that primed CD8⁺ T cells selectively homed to tumors, were further confirmed in a separate study involving adoptive transfer of primed TDLN CD8⁺ T cells to tumor-bearing recipients. In this study, tumors were induced in Thy1.1 BALB/c mice and treated with IL-12/GM-CSF microspheres. On day 4 posttreatment, CD8⁺ T cells were isolated from the TDLN and were adoptively transferred to tumor-bearing Thy1.2 recipients. In the control group, CD8⁺ T cells were isolated from day 4 TDLN following treatment with blank microspheres. Tumors and spleens of recipients were then analyzed for infiltration of adoptively transferred Thy1.1 CD8⁺ T cells. The results are shown in Fig. 2 C. Whereas the CD8⁺ T cells obtained from control and IL-12/GM-CSF-treated mice migrated to the spleen with similar efficacy, primed CD8⁺ T cells from treated mice targeted tumors much more effectively (7-fold better) than did CD8 T cells of control mice, as indicated by the relative numbers of Thy1.1 CD8⁺ T cells found in tumors of recipient mice. These data further confirmed that treatment-induced TDLN CD8⁺ T cells were able to preferentially home to tumors.

We next addressed the question of whether postprime CD8⁺ T effectors could mediate cell killing in a tumor-specific manner. Since T cell activation and proliferation is accompanied with enhanced cytotoxic function (25), day 4 TDLN CD8⁺ T cells were first analyzed for cytotoxic effector molecule expression. Comparison of granzyme B levels in pre- and posttreatment cells demonstrated a significant increase in both the percentage of cells that expressed granzyme B and the amount of granzyme B per cell, demonstrating that priming and peripheral migration was accompanied with development of cytotoxic effector function (Fig. 3,A). Importantly, CD8⁺ T cells isolated from the TDLN 4 days after treatment were capable of lysing line 1 targets but not a different tumor in vitro, establishing tumor specificity (Fig. 3,B). In contrast, CD8⁺ T cells that were obtained from control-treated mice did not lyse either target. In further studies, Ag specificity of primed day 4 CD8⁺ T cells was tested in a coculture assay where purified day 4 CD8⁺ T effector cells were stimulated either with line 1 or 4T1 tumor lysate-pulsed monocytes. Data shown in Fig. 3,C establish that while CD8⁺ T effector cells isolated from treated mice produced IFN-γ at levels 12-fold higher than those purified from untreated controls in response to line 1, there was no difference between treated and control cells following coincubation with 4T1-pulsed monocytes. Finally, tumor specificity of T effectors was evaluated in vivo where the ability of posttherapy CD8⁺ T cells to infiltrate experimental and control tumors in the same animal was monitored. Mice bearing line 1 and 4T1 tumors in close proximity were treated with a single injection of IL-12/GM-CSF microspheres into the line 1 tumor, and CD8⁺ T cell infiltration was quantified in both tumors. The data in Fig. 3 D show that while the CD8⁺ T cell quantity in 4T1 tumors remained unchanged between days 0 and 7, a 2-fold increase was observed in line 1 tumors. Collectively, these results demonstrate that therapy-induced CD8⁺ T effectors were tumor-specific.

The above studies established that intratumoral delivery of IL-12/GM-CSF resulted in the development of a tumor-specific CD8⁺ T effector cell response in the draining lymph nodes. On the other hand, the mechanism underlying the cytokine-induced CD8⁺ T effector priming remained undefined. For example, whether cytokines directly activated Ag-experienced CD8⁺ T cells in the TDLN or mediated priming indirectly via the induction of an inflammatory response within the primary tumor was not determined. Subcutaneous injection of IL-12/GM-CSF microspheres into nontumor-bearing naive mice did not result in CD8⁺ T cell priming in the draining lymph nodes (Fig. 4,A), demonstrating that CD8⁺ T effector cell activation was not indiscriminate, and that factors associated with the tumor microenvironment and/or the tumor-conditioned draining lymph nodes were required. To determine whether this requirement involved factors that were uniquely associated with the tumor microenvironment, the efficacy of CD8⁺ T effector priming was tested in a surgical resection model. Tumors were induced and were allowed to reach a size of 300–400 mm³ to establish tumor conditioning of the draining lymph nodes. The primary tumors were then surgically removed and the mice were treated by direct injection of IL-12/GM-CSF microspheres into the site of resection. CD8⁺ T effector cells were isolated from the TDLN 3 days after treatment and analyzed for activation. The results are shown in Fig. 4 B. These data demonstrate that treatment failed to induce significant CD8⁺ T effector activation in postsurgical mice. In contrast, in control mice, which did not experience surgery, treatment resulted in a substantial decrease in the proportion of CD62L⁺ cells (2.5-fold) and a parallel increase in CD62L⁻CCR5⁺ cells (3-fold). The minor changes observed in percentage CD62L⁺ and CD62L⁻ CCR5⁺CD8⁺ T cells in the surgical groups were not statistically significant and likely represented a nonspecific effect of postsurgical inflammation and/or cytokine delivery. The inability to induce effective activation of CD8⁺ T effector cells in tumor-conditioned draining lymph nodes in the absence of the primary tumor suggested that factors present in the tumor were critical to priming in the draining lymph nodes following intratumoral IL-12/GM-CSF therapy.

Since previous studies demonstrated that tumor-resident CD8⁺ Tem were important to posttherapy activation of innate effectors (22), we hypothesized that this subset could be the putative tumor-associated factor that was essential to T effector priming. This notion was tested in an adoptive cell transfer model in which tumor-bearing mice were first depleted of CD8⁺ T cells and then reconstituted selectively with lymph node-homing CD8⁺ T cells before treatment. The goal of this protocol, which is outlined in Fig. 5,A, was to create a mouse in which the TDLN were repopulated with CD8⁺ T cells but the tumor largely remained CD8⁺ T cell-free. To determine whether selective lymph node reconstitution could be achieved, CD8⁺ T cells purified from the TDLN of Thy1.1 BALB/c donors bearing established tumors were adoptively transferred to tumor-bearing Thy1.2 recipients. The results from one such experiment are shown in Fig. 5,B. Adoptive transfer of CD62L⁺CD8⁺ T cells isolated from the TDLN of donor Thy1.1 BALB/c mice resulted in selective migration of these cells to the TDLN of Thy1.2 recipients with an average of 601 ± 96 cells homing to a single TDLN vs 21 ± 8 cells migrating to the tumor, validating the above model (Fig. 5 B).

If the hypothesis that tumor-associated CD8⁺ T cells are critical to priming is correct, the minimal numbers of Thy1.1 CD8⁺ T cells found in tumors of Thy1.2 recipients would not be expected to induce effective priming. To test this notion, experimental recipient mice were injected with an anti-CD8 Ab, resulting in the elimination of >98% of preexisting CD8⁺ T cells both in the secondary lymphoid organs and the tumor (supplemental Fig. 1)⁴ for up to 7 days. Three days after Ab administration, both groups received Thy1.1 CD8⁺ T cells isolated from the TDLN of tumor-bearing donors. Two days after adoptive transfer, when adoptively transferred cells had achieved peak accumulation in the TDLN (data not shown), both groups were treated with a single intratumoral injection of IL-12/GM-CSF microspheres, and Thy1.1 CD8⁺ T cell activation was monitored. The results from such an experiment are shown in Fig. 5,C. Transfer of Thy1.1 CD8⁺ T cells resulted in similar TDLN migration efficiency in CD8⁺ T cell-depleted mice and in nondepleted controls, demonstrating that prior CD8⁺ T cell depletion did not affect the migration efficiency or survival of donor cells. Treatment induced a >5-fold increase in Thy1.1 CD8⁺ T cell quantity and a switch from a CD62L⁺ to a CD62L⁻ phenotype in the TDLN of control mice. In contrast, the absence of CD8⁺ Tem from tumors of CD8⁺ T cell depleted recipients resulted in the complete abrogation of Thy1.1⁺CD8⁺ T cell activation in the TDLN. It was unlikely that the loss of CD8⁺ T effector cell priming in the TDLN of CD8⁺ T cell-depleted mice was due to the elimination of CD8⁺ dendritic cell (DC) subset as the CD8⁺ DC constituted a very small proportion of the TDLN DC population (<1%, data not shown) and that DC isolated from the lymph nodes of control and Ab-treated mice were equally effective in stimulating the proliferation of CD8⁺ T cells in vitro (supplemental Fig. 2). However, these findings do not unequivocally rule out the possibility that depletion of the CD8⁺ DC, which are particularly effective in priming CD8⁺ T cells via cross-presentation, contributed to the observed loss of T effector cell expansion in vivo. Regardless, the data presented in Figs. 4 and 5 strongly suggest that activation of tumor-resident CD8⁺ Tem was critical to the priming of the secondary CD8⁺ effector T cell response in the TDLN.

The above data established the direct connection between CD8⁺ Tem activation and T effector priming but did not address the underlying mechanism. Previous studies had shown that treatment with IL-12/GM-CSF microspheres induced IFN-γ and granzyme B production by tumor-resident CD8⁺ Tem within 24 h of treatment (18). We therefore postulated that CD8⁺ Tem mobilized T effector cell priming via rapid tumor destruction and Ag release. To this end, T effector cell activation was evaluated in IFN-γ knockout (GKO) and perforin knockout (PfKO) mice. Analysis of posttreatment CD8⁺ T cell populations in the TDLN of mice defective in these cytotoxic effector molecules demonstrated that CD8⁺ T effector cell priming was completely abolished in both the GKO and the PfKO mice (Table I). The inability to achieve priming in the GKO or PfKO mice was not associated with reduced numbers of preexisting tumor-resident CD8⁺ T cells in these strains (28.8 ± 10.9 × 10⁻⁴, 50.8 ± 24 × 10⁻⁴, and 28.7 ± 6 × 10⁻⁴ cells/g tumor for GKO, PfKO, and wild-type mice, respectively). These data are therefore consistent with the notion that restoration of CD8⁺ Tem cytotoxicity and the resulting tumor kill was the primary mechanism through which preexisting CD8⁺ Tem mediated secondary effector priming in the TDLN.

The studies described above define the source and the functional kinetics of the latent CD8⁺ T effector cell response that is induced by intratumoral IL-12/GM-CSF therapy. More specifically, these data, combined with our previous findings, conclusively establish that treatment promotes de novo priming of antitumor CD8⁺ T effector cells in the TDLN, which subsequently home to tumors. Importantly, the results also establish the critical role of tumor-resident CD8⁺ Tem in the priming of the secondary CD8⁺ T effector response, identifying this population as a potent in situ resource for successful reactivation of systemic antitumor T cell immunity.

The prevailing rationale for restoring TIL function in situ is to achieve rapid tumor elimination without having to resort to logistically challenging vaccination and/or ex vivo cell manipulation protocols. Recent work in our laboratory demonstrated that while intratumoral IL-12/GM-CSF was highly effective in restoring tumor-resident CD8⁺ Tem cytotoxicity, subsequent tumor regression was associated with the induction of a complex, multicomponent inflammatory response. Post-Tem activation immune responses included the elimination of tumor-infiltrating CD4⁺CD25⁺Foxp3⁺ T suppressor cells (18), NK cell recruitment (22), a rapid switch in tumor-associated macrophage phenotype (27), and activation of macrophage-associated cytotoxic effector mechanisms (28). Herein, we demonstrate that rescue of tumor-resident CD8⁺ Tem function also leads to the priming of a de novo CD8⁺ T effector cell response in the TDLN. This finding supports the notion that reactivation of preexisting CD8⁺ Tem not only mediates immediate local tumor kill, but also plays a central role in the re-priming of a long-term systemic antitumor T cell response.

Both IFN-γ and perforin were required for Tem-driven CD8⁺ T effector cell priming in the TDLN. The central role of IFN-γ in the antitumor effects of IL-12 is well known (29). This pleiotropic cytokine induces the expression of more than 200 immune-related genes (30) and augments multiple distinct immune mechanisms including T and NK cell cytotoxicity (31), MHC class I and class II Ag presentation, and macrophage effector function (30). Therefore, the complete loss of T effector cell activation and proliferation observed in the GKO mice was not surprising. At the same time, this finding did not distinguish which of the above effector mechanisms were important to priming. In contrast, perforin release, which occurs immediately downstream of IFN-γ production, is strictly associated with direct cellular cytotoxicity. Accordingly, complete loss of T effector activation in PfKO mice suggests that rescue of intratumoral Tem, and possibly NK cell, cytotoxicity were critical to priming of CD8⁺ T effector cells in the draining lymph node.

Primed CD8⁺ T effector cells up-regulated granzyme B expression before migration to the periphery, were able to home to tumors, and mediated tumor cell lysis in vitro, suggesting that they could achieve tumor kill in vivo. However, postinfiltration cytotoxicity of CD8⁺ T effectors was not directly monitored in this study. Previous work demonstrated that day 7 tumor-infiltrating effectors displayed reduced granzyme B levels in comparison to day 1 CD8⁺ Tem (18) (Fig. 3,A) in treated mice and that the intratumoral CD8⁺ T cell numbers declined rapidly after day 7 (18) (Fig. 1 A). These findings establish that the combined window of cytotoxic activity for preexisting CD8⁺ Tem and the secondary CD8⁺ T effectors was limited to 8–10 days. Whether the transient nature of the T effector response in tumors was due to activation-induced cell death and/or nonresponsiveness (32, 33), to the ability of the tumor to rapidly reinstate immune dysfunction (34), or to a combination of these factors was not investigated here. Defining the mechanisms underlying the rapid loss of the latent CD8⁺ T effector function will be critical to improving the long-term therapeutic efficacy of the above strategy.

We thank Dr. Stan Wolf of Wyeth Pharmaceuticals for providing the recombinant IL-12 and for continued support of our studies.

The authors have no financial conflicts of interest.