Sustained intratumoral delivery of IL-12 and GM-CSF can overcome tumor immune suppression and promote T cell-dependent eradication of established disease in murine tumor models. However, the antitumor effector response is transient and rapidly followed by a T suppressor cell rebound. The mechanisms that control the switch from an effector to a regulatory response in this model have not been defined. Because dendritic cells (DC) can mediate both effector and suppressor T cell priming, DC activity was monitored in the tumors and the tumor-draining lymph nodes (TDLN) of IL-12/GM-CSF–treated mice. The studies demonstrated that therapy promoted the recruitment of immunogenic DC (iDC) to tumors with subsequent migration to the TDLN within 24–48 h of treatment. Longer-term monitoring revealed that iDC converted to an IDO-positive tolerogenic phenotype in the TDLN between days 2 and 7. Specifically, day 7 DC lost the ability to prime CD8+ T cells but preferentially induced CD4+Foxp3+ T cells. The functional switch was reversible, as inhibition of IDO with 1-methyl tryptophan restored immunogenic function to tolerogenic DC. All posttherapy immunological activity was strictly associated with conventional myeloid DC, and no functional changes were observed in the plasmacytoid DC subset throughout treatment. Importantly, the initial recruitment and activation of iDC as well as the subsequent switch to tolerogenic activity were both driven by IFN-γ, revealing the dichotomous role of this cytokine in regulating IL-12–mediated antitumor T cell immunity.

Active tumor immune therapy strategies have traditionally focused on inducing tumor-specific cytotoxic T cells (1). However, the inability to establish a direct correlation between the intensity of a systemic antitumor T cell response and efficacy of tumor regression (2, 3) eventually led to a shift in focus from quantitative to qualitative analysis of posttherapy T effector activity (4). These studies revealed the critical roles of tumor-associated immune suppressive mechanisms and T cell-intrinsic negative checkpoint molecules in premature loss of T cell cytotoxicity in tumors (5, 6). Consequently, recent work has focused on combinatorial approaches that are designed to simultaneously activate antitumor T cells and block suppressive and regulatory pathways (7).

Paracrine delivery of IL-12 and GM-CSF to the tumor microenvironment overcomes tumor immune suppression and promotes effective eradication of established disease in mice (810). Specifically, studies in our laboratory revealed that treatment restored cytolytic function to tumor-resident CD8+ T effector/memory cells, induced the elimination of CD4+CD25+Foxp3+ T suppressor cells, and promoted de novo priming of a secondary CD8+ T cell response in the tumor-draining lymph node (TDLN) (1113). All posttherapy changes in T cell immunity were dependent on IFN-γ (1113). The antitumor effector window, however, was transient and short-circuited by a potent T suppressor cell rebound that developed within 4–7 d of treatment (10, 14). Recent analysis of posttherapy counterregulation established a link between the IL-12–IFN-γ–IDO axis and T suppressor cell expansion (10). These findings suggested that IFN-γ was not only responsible for the mobilization of antitumor effectors, but also played a role in driving feedback inhibition. The dual role of IFN-γ in immune stimulation and suppression has been reported (15, 16). However, the cellular and molecular pathways that govern these opposing functions are not yet completely understood.

IFN-γ is a pluripotent cytokine with direct functional effects on innate and adaptive effector cell populations including the dendritic cell (DC) (17). This is important, as DC ultimately control the T effector versus T suppressor decision during an immune response (18). IFN-γ–driven molecular pathways that influence DC phenotype and function are complex and poorly defined (17, 19). For example IFN-γ, in conjunction with TLR ligands, can enhance immunogenic activation of DC, yet conversely promote tolerogenic function via induction of IDO (17, 19, 20). To this end, we examined the role of the IFN-γ–DC axis in the development and progression of post–IL-12/GM-CSF antitumor T cell immunity. Our data demonstrate that conventional myeloid DC (cDC) that are recruited to the tumor following IL-12/GM-CSF therapy mediate both the antitumor effector priming and the subsequent suppressor rebound in the TDLN and that both pathways are driven by IFN-γ.

All mice used were in the BALB/c background and maintained in our breeding colony. Male wild-type and IFN-γ knockout (KO) mice were used for experiments at 6–10 wk of age. Breeder pairs of TCR-hemagglutinin (HA) (21) and Clone-4 mice (22) were kindly provided by Dr. Sandra Gollnick (Roswell Park Cancer Institute, Buffalo, NY). Clone-4 mice were tested for the transgenic TCR via flow cytometric analysis of the TCR from retro-orbital blood samples (FITC-conjugated anti-Vβ 8.1, 8.2 TCR Ab, clone MR5-2; BD Pharmingen).

Line-1, a weakly immunogenic MHC I low metastatic lung alveolar carcinoma of the BALB/c mouse, was used to induce tumors in all experiments as previously described (11). In brief, 1 × 106 cells were injected in 0.1 ml PBS s.c. in the subscapular area of mice. Tumors were allowed to grow to ∼150–200 mm3 before treatment.

Polylactic acid microspheres with a cytokine loading of 0.025% (w/w) were prepared using the phase inversion nanoencapsulation method as described previously (12). Mice were treated with 4 mg each microsphere preparation (equivalent to 1 μg each GM-CSF and IL-12) suspended in 0.1 ml sterile PBS via direct injection into the tumor.

Single-cell suspensions from tumors, TDLN, and spleens were prepared as described previously (11).

Twenty microliters 0.5-μm FITC-conjugated latex beads (Polysciences; diluted 1:25 in PBS) were administered simultaneously with microsphere treatment. Single-cell suspensions prepared from tumor and TDLN were analyzed by flow cytometry for CD11c+MHC II+FITC+ DCs. For control tumors, FITC-latex beads were administered into the tumor with blank microspheres.

All cell populations used for RT-PCR, ELISA, and in vitro coculture T cell priming assays were isolated by magnetic cell sorting according to the manufacturer’s recommendations (Miltenyi Biotec, Auburn, CA). CD11c (N418) microbeads were used to isolate total DCs. For separation of plasmacytoid DCs (pDC) and cDCs, TDLN cell suspensions were first sorted with murine pDC Ag (mPDCA)+ microbeads to select for pDCs. The PDCA-negative (pDC-depleted) fraction was then used to purify CD11c+ cDCs. CD4+ (L3T4) and CD8 (Ly-2) microbeads were used to isolate naive CD4+ T cells and CD8+ T cells from spleens of TCR-HA and Clone-4 mice, respectively, for the in vitro T cell priming assay.

Analysis of leukocyte subsets was performed on a four-color BD FACSCalibur (BD Biosciences) using chromofluor-labeled Abs to surface and intracellular markers as previously described (11). For analysis of intracellular expression of IDO, single-cell suspensions were first labeled for extracellular markers. Cells were then fixed and permeabilized using the BD Fixation/Permeabilization kit and labeled with 1.0 μl anti-IDO per 2 × 106 cells (mouse IgG3, clone 10.1; Millipore, stock resuspended at 200 μg/ml) and incubated for 40 min at 4°C. Cells were then washed and stained with the secondary Ab (FITC rat anti-mouse IgG3, clone R40-82; BD Pharmingen) for 30 min at 4°C followed by a final wash/fix cycle prior to analysis. For analysis of Foxp3 expression, the eBioscience Foxp3 staining kit (eBioscience) was used as per the manufacturer’s recommendations.

DC were isolated using magnetic cell sorting as described above. DC were resuspended in 1 ml MLR media (DMEM plus 5% FBS with 10 mM HEPES [pH 7.4], 1% sodium pyruvate, 1% penicillin/streptomycin, 1% l-glutamine, 0.4% L-arginine HCl, 1% folic acid/l-asparagine, and 0.2% 2-ME) including HA peptides (MHC I sequence, IYSTVASSL; MHC II sequence, SFERFEIFPK; 5 μg/ml) and incubated for 1 h at 37°C. CD8+ and CD4+ T cells from Clone-4 and TCR-HA mice, respectively, were isolated as described above and labeled with CFSE (SA Biosciences) as recommended by the manufacturer. Peptide-loaded DCs were cultured with CFSE-labeled T cells at a DC to T cell ratio of 1:5 (for CD8+ T cell priming) or 1:10 (for CD4+ T cell priming) for 2 to 3 d. CFSE dilution was quantified via flow cytometry. To determine CD4+ T cell polarization after in vitro priming, cocultured cells were stained for IFN-γ and Foxp3 as described above.

The IDO enzyme activity assay was performed as described (23). In brief, 1 × 105 DCs isolated via magnetic sorting as described above were resuspended in 300 μl sterile DMEM-5% FBS containing 500 μM tryptophan and aliquoted in a 96-well plate. After 24 h incubation, 50 μl 30% TCA was mixed with 100 μl culture supernatant and centrifuged at 11,000 rpm (9,000 × g) in a benchtop microcentrifuge for 5 min. Sixty microliters supernatant was then mixed with 60 μl Ehrlich reagent (100 mg P-dimethylbenzaldehyde [Sigma-Aldrich] in 5 ml glacial acetic acid). The OD was measured at 492 nm. Purified l-kynurenine (0–500 μM; Sigma-Aldrich) was used to construct a standard curve.

Student t test was used to determine the significance of the differences between control and experimental groups in pairwise comparisons. In experiments with multiple groups, homogeneity of intergroup variance was analyzed by ANOVA. In all analyses, p ≤ 0.05 was considered significant.

We hypothesized that treatment-generated DC were central to both the initial CD8+ T effector cell priming and the subsequent CD4+CD25+Foxp3+ T suppressor cell rebound. To this end, we first undertook quantitative and qualitative characterization of tumor and TDLN DC in posttherapy mice. Quantitative analysis of DC between days 0 and 7 revealed a rapid 4-fold increase in intratumoral DC numbers on day 2 followed by a gradual decline between days 2 and 7 (Fig. 1A). A similar pattern was observed in the TDLN, except the numbers peaked on day 4 and remained stable thereafter (Fig. 1A). The sequential nature of the tumor and TDLN infiltration kinetics was consistent with initial recruitment of DC to tumors followed by migration to the TDLN. To determine whether the recruited DC were actively taking up Ag, tumors were injected with FITC-labeled latex beads on the day of treatment, and the numbers of FITC-positive DC were monitored in control and experimental mice. The data shown in Fig. 2B demonstrate effective uptake by tumor-infiltrating DC in experimental but not control mice during the first 48 h. Importantly, following uptake, FITC+ DC rapidly migrated to and persisted in the TDLN for up to 7 d. To confirm that recruitment and Ag uptake were followed by maturation and activation, DC were analyzed for the expression of lymph node-homing chemokine receptor CCR7 and costimulatory molecules CD40 and CD86. Characterization of tumor-infiltrating DC for membrane CCR7 revealed a 2-fold increase in expression on day 2, suggesting that the cells matured rapidly upon entry into tumor. Moreover, analysis of the day 2 TDLN DC showed that these cells had upregulated CD40 and CD86, suggesting that migration to the TDLN was accompanied with activation. DC activation results in cytokine production, which can provide signal 3 during T cell priming and determine the type of the T cell response (24). To this end, posttreatment TDLN DC were analyzed for the production of proinflammatory cytokines TNF-α and IL-12 and the immune suppressive cytokine IL-10. The results shown in Fig. 1D demonstrate that therapy resulted in a 3–5-fold increase in TNF-α and IL-12 with a concurrent decline in IL-10 consistent with immunogenic activation on day 2. Both costimulatory molecule and inflammatory cytokine expression returned to basal levels on day 7. IL-10 levels, however, continued to decline, displaying a 3-fold reduction between days 0 and 7, suggesting that day 7 DC were phenotypically distinct from day 0 cells. Collectively, the data shown in Fig. 1A–D are consistent with recruitment and activation of DC between days 0 and 2 followed by migration to the TDLN and a return to quiescence on day 7.

FIGURE 1.

Analysis of IL-12/GM-CSF treatment-mediated changes in DC phenotype and function. A, Quantitative changes in DC. CD45+MHC II+CD11c+ DC numbers were quantified in tumors and TDLN of control and IL-12/GM-CSF–treated mice. Error bars indicate SE. n = 14–24 per time point in IL-12/GM-CSF groups; n = 3 for control. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.0001). *Statistical significance compared with day 0 (p ≤ 0.017). B, Ag uptake and migration. Numbers of FITC+ DC found in the tumor and TDLN of IL-12/GM-CSF–treated mice were determined. The flow cytometry panels are representative of the analytical approach used to identify FITC+ tumor-infiltrating DC. Control mice received blank microspheres and displayed negligible uptake (data not shown). Error bars indicate SE. n = 3 to 4 per time point. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.05). #,*Significant compared with day 0 (p ≤ 0.021 and ≤ 0.037, respectively). C, Activation markers. Membrane expression of CD40, CD86, and CCR7 were analyzed. The panels represent typical profiles for CCR7 and CD86 (CD40 profile was similar to CD86). Error bars indicate SE. n = 6–16 per time point. *Statistical significance compared with day 0 (p ≤ 0.014). D, Cytokine profile. The mRNA (IL-10, TNF-α) or protein (IL-12) levels were determined by quantitative RT-PCR or IL-12 p70-specific ELISA, respectively. Error bars indicate SE. n = 7–14 per time point. *Statistical significance compared with day 0 (p ≤ 0.038). E, Priming function. The ability of DC to promote Ag-specific CD8+ T cell proliferation was determined in an in vitro CFSE dilution assay (1Materials and Methods). Error bars indicate SE. n = 7–10 per time point.

FIGURE 1.

Analysis of IL-12/GM-CSF treatment-mediated changes in DC phenotype and function. A, Quantitative changes in DC. CD45+MHC II+CD11c+ DC numbers were quantified in tumors and TDLN of control and IL-12/GM-CSF–treated mice. Error bars indicate SE. n = 14–24 per time point in IL-12/GM-CSF groups; n = 3 for control. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.0001). *Statistical significance compared with day 0 (p ≤ 0.017). B, Ag uptake and migration. Numbers of FITC+ DC found in the tumor and TDLN of IL-12/GM-CSF–treated mice were determined. The flow cytometry panels are representative of the analytical approach used to identify FITC+ tumor-infiltrating DC. Control mice received blank microspheres and displayed negligible uptake (data not shown). Error bars indicate SE. n = 3 to 4 per time point. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.05). #,*Significant compared with day 0 (p ≤ 0.021 and ≤ 0.037, respectively). C, Activation markers. Membrane expression of CD40, CD86, and CCR7 were analyzed. The panels represent typical profiles for CCR7 and CD86 (CD40 profile was similar to CD86). Error bars indicate SE. n = 6–16 per time point. *Statistical significance compared with day 0 (p ≤ 0.014). D, Cytokine profile. The mRNA (IL-10, TNF-α) or protein (IL-12) levels were determined by quantitative RT-PCR or IL-12 p70-specific ELISA, respectively. Error bars indicate SE. n = 7–14 per time point. *Statistical significance compared with day 0 (p ≤ 0.038). E, Priming function. The ability of DC to promote Ag-specific CD8+ T cell proliferation was determined in an in vitro CFSE dilution assay (1Materials and Methods). Error bars indicate SE. n = 7–10 per time point.

Close modal
FIGURE 2.

Individual effects of IL-12 and GM-CSF on posttherapy DC activity. A, Migration. The number of DC in the tumor and the TDLN were determined on days 0 and 2 for each treatment group. Error bars indicate SE. n = 6–9 per group. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.009). *Significance compared with day 0 (p ≤ 0.012). B, Activation. Membrane expression of CD86 and CCR7 (percent positive cells and mean fluorescence intensity, respectively) were quantified. Error bars indicate SE. n = 3 to 4 per group. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.03). *Significance compared with day 0 (p ≤ 0.028). C, Function. Ability to induce CD8+ T cell proliferation was determined as in Fig. 1E. Error bars indicate SE. n = 5 per group. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.0001). *Significance compared with day 0 (p ≤ 0.003).

FIGURE 2.

Individual effects of IL-12 and GM-CSF on posttherapy DC activity. A, Migration. The number of DC in the tumor and the TDLN were determined on days 0 and 2 for each treatment group. Error bars indicate SE. n = 6–9 per group. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.009). *Significance compared with day 0 (p ≤ 0.012). B, Activation. Membrane expression of CD86 and CCR7 (percent positive cells and mean fluorescence intensity, respectively) were quantified. Error bars indicate SE. n = 3 to 4 per group. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.03). *Significance compared with day 0 (p ≤ 0.028). C, Function. Ability to induce CD8+ T cell proliferation was determined as in Fig. 1E. Error bars indicate SE. n = 5 per group. Homogeneity of intergroup variance was analyzed by ANOVA and found to be significant (p < 0.0001). *Significance compared with day 0 (p ≤ 0.003).

Close modal

The ultimate measure of DC function is the ability to prime naive T cells. To this end, DC isolated from the TDLN on days 0, 2, and 7, were evaluated for their capacity to induce the proliferation of TCR-transgenic naive CD8+ T cells in an in vitro CFSE dilution assay. Specifically, DC pulsed with an MHC I peptide of influenza HA were coincubated with peptide-specific CD8+ T cells isolated from TCR-transgenic mice, and the proportion of CD8+ T cells that divided was quantified. The results are shown in Fig. 1E. These data demonstrate that pre-existing DC (day 0) were functionally impaired as their ability to expand T cells was reduced by 2-fold compared with DC from naive mice. In contrast, DC isolated from the TDLN of day 2 mice were just as effective as their naive counterparts, establishing that treatment-induced DC were fully functional. By day 7, however, these cells had again become functionally impaired (i.e., their ability to expand T cells was severely reduced) (∼4-fold). Moreover, day 7 DC were significantly less effective in promoting T cell proliferation than day 0 population, demonstrating that posttherapy tolerogenic DC (tDC) were functionally distinct from pretherapy tDC.

The above studies established that codelivery of IL-12 and GM-CSF to the tumor microenvironment resulted in a rapid and highly effective rescue of DC activity in the tumor and the TDLN. At the same time, the individual roles of IL-12 and GM-CSF in mediating these changes were not determined. The ability of IL-12 to directly induce T and NK cell activation (25) and of GM-CSF to promote DC generation (26) is well known. Consistent with these functions, previous work in our laboratory demonstrated that rescue of tumor-resident T cell cytotoxicity required IL-12, whereas systemic long-term effects were enhanced by GM-CSF (8, 11). To define the individual roles of each cytokine in DC activation in our model, we characterized selected DC functional parameters in mice treated with IL-12 alone, GM-CSF alone, or IL-12 plus GM-CSF. The results are shown in Fig. 2. Initially, DC chemotaxis to the tumor and the TDLN was monitored via quantitative analysis of DC numbers on days 0 and 2. Either cytokine alone induced effective recruitment of DC to the tumor (Fig. 2A). In contrast, both cytokines were needed to promote effective homing of the DC to the TDLN. Analysis of the homing and activation markers revealed that CD86 was induced primarily by IL-12, but both cytokines were required for upregulation of CCR7 (Fig. 2B) and efficient migration (Fig. 2A). Finally, consistent with the data shown in Fig. 2A and 2B, whereas either cytokine promoted a modest (50%) enhancement of TDLN DC priming function, both were required to achieve full T cell expansion (a 3-fold increase compared with day 0). These data demonstrate that the synergistic activity of IL-12 and GM-CSF was largely due to superior DC maturation and homing to the TDLN.

We next examined the physiological basis of the post–IL-12/GM-CSF functional switch in DC. Previous studies had demonstrated that the progression from an effector to a suppressor T cell response was accompanied by the induction of IDO in the tumors and TDLN (10). This tolerogenic enzyme can be expressed by both pDC and cDC subsets (2730). To determine whether DC were a significant source of IDO in treated mice, pDC and cDC populations found in the TDLN were analyzed. Intracellular staining revealed that pretherapy cDC expressed negligible levels of IDO with no change on day 2. In contrast, a subset of day 7 cDC displayed a robust 3-fold increase in intracellular IDO levels (Fig. 3A, Supplemental Fig. 1). In pDC, the basal level of IDO was higher than that seen in cDC on day 0. Treatment resulted in further increases on days 2 and 7; however, these changes were not statistically significant. Analysis of IDO function in an in vitro assay confirmed the intracellular staining data, demonstrating a significant increase in the ability of day 7 cDC to metabolize tryptophan to kynurenine in comparison with day 0 cDC. No change was observed in pDC IDO activity between days 0 and 7 in the same assay. These findings suggested that therapy-induced changes in IDO expression were linked to the cDC subset and that these cells were the primary modulators of the tolerogenic rebound.

FIGURE 3.

Expression of IDO and its effects on T cell priming by cDC versus pDC. A, Changes in IDO levels and activity. The histogram demonstrates intracellular staining of TDLN cDC for IDO (MHC II+CD11c+B220 cells were gated on and analyzed for IDO). Conventional DC found in the TDLN of untreated tumor-bearing mice or the LN of non–tumor-bearing naive mice expressed similar levels of IDO (data not shown). Approximately 12–15% of day 7 cDC were routinely found to be positive for IDO in treated mice. Quantitative analysis of IDO expression and bioactivity in pDC and cDC subsets (purified as described in 1Materials and Methods) are shown in adjacent plots. Error bars indicate SE. n = 6–9 per time point. *Statistical significance compared with days 0 and 2 (p ≤ 0.002). B, Priming. Expansion of CD8+ T cells by pDC versus cDC subsets purified from the TDLN on days 0, 2, and 7 posttherapy is shown. Significance was obtained only between day 2 cDC and day 0 or day 7 cDC. Error bars indicate SE. n = 9 per time point. C, Rescue of priming function by D1-MT. The effect of inhibition of IDO by D1-MT on the priming ability of cDC and pDC was determined using the CD8+ T cell proliferation assay. Significance was obtained only for day 7 cDC. Error bars indicate SE. n = 7–10 per time point.

FIGURE 3.

Expression of IDO and its effects on T cell priming by cDC versus pDC. A, Changes in IDO levels and activity. The histogram demonstrates intracellular staining of TDLN cDC for IDO (MHC II+CD11c+B220 cells were gated on and analyzed for IDO). Conventional DC found in the TDLN of untreated tumor-bearing mice or the LN of non–tumor-bearing naive mice expressed similar levels of IDO (data not shown). Approximately 12–15% of day 7 cDC were routinely found to be positive for IDO in treated mice. Quantitative analysis of IDO expression and bioactivity in pDC and cDC subsets (purified as described in 1Materials and Methods) are shown in adjacent plots. Error bars indicate SE. n = 6–9 per time point. *Statistical significance compared with days 0 and 2 (p ≤ 0.002). B, Priming. Expansion of CD8+ T cells by pDC versus cDC subsets purified from the TDLN on days 0, 2, and 7 posttherapy is shown. Significance was obtained only between day 2 cDC and day 0 or day 7 cDC. Error bars indicate SE. n = 9 per time point. C, Rescue of priming function by D1-MT. The effect of inhibition of IDO by D1-MT on the priming ability of cDC and pDC was determined using the CD8+ T cell proliferation assay. Significance was obtained only for day 7 cDC. Error bars indicate SE. n = 7–10 per time point.

Close modal

To determine whether the above hypothesis was correct, the ability of cDC and pDC populations isolated from the TDLN of day 0, 2 and 7 mice to prime naive CD8+ T cells was evaluated using the in vitro system described above. The results presented in Fig. 3B reveal that pDC were not able to expand CD8+ T cells following treatment. In contrast, cDC were highly active on day 2, mirroring the functional kinetics of total TDLN DC populations shown in Fig. 1E. These data confirmed that the immunogenic effects of treatment on APC populations were uniquely associated with the cDC subset in the TDLN. We next determined whether day 7 DC dysfunction was associated with IDO. The abilities of purified DC subsets to expand CD8+ T cells were evaluated in the absence or presence of D-1-methyl tryptophan (D1-MT), an inhibitor of IDO function. The data are shown in Fig. 3C. The results demonstrate that addition of D1-MT to day 7 cDC, but not to day 0 cDC, reversed the dysfunctional phenotype, resulting in the complete rescue of priming function. In contrast, D1-MT had no effect on pDC. These findings confirm that day 7 cDC quiescence was directly associated with IDO and that pDC did not play a major role in therapy-mediated changes in T cell immunity. It is also noteworthy that whereas impaired function of day 7 DC was associated with IDO, that of day 0 DC was not, suggesting that day 0 and 7 cDC were physiologically distinct with regard to the mechanisms underlying their inability to prime CD8+ T cells. This notion was further supported by the finding that whereas initial treatment reversed the suppressive phenotype, a second treatment on day 7 not only failed to restore immunogenic DC (iDC) activity but also exacerbated both the IDO expression and the loss of priming function in TDLN cDC (data not shown).

In the above studies, we employed the CD8+ T cell priming assay to evaluate DC function. Priming and differentiation of CD4+ T cells is more complex, as these cells can develop into functionally distinct effector and suppressor subsets depending on the signals provided by the DC and the microenvironment (31). In fact, DC that are considered tolerogenic based on their inability to prime effector T cells can efficiently expand suppressor T cells (32). To this end, the ability of TDLN cDC isolated from day 0, 2, and 7 mice to expand naive CD4+ T cells was analyzed in an in vitro assay similar to that used for the CD8+ T cells, except the DC were pulsed with a class II peptide epitope of HA, and the naive CD4+ T cells were purified from the transgenic TCR-HA mice (21). The results shown in Fig. 4A demonstrate that similar to that seen with the CD8+ T cells, day 2 cDC were able to expand CD4+ T cells more effectively than day 0 cDC. Day 7 cDC, in contrast, behaved somewhat differently with respect to their ability to prime CD8+ and CD4+ T cells. Whereas day 7 cDC were virtually incapable of expanding CD8+ T cells, they could prime CD4+ T cells, although somewhat less efficiently than day 2 cDC. This suggested that the CD4+ T cells that are primed on days 2 and 7 might differ in phenotype. Intracellular staining of CD4+ T cells primed by day 2 and 7 cDC for IFN-γ and Foxp3 revealed that the Th1 effector phenotype dominated in the former, whereas the latter were primarily of the suppressor phenotype (Fig. 4B). Addition of D1-MT to the media reduced but did not completely ablate induction of Foxp3, suggesting that factors other than IDO may contribute to the ability of day 7 cDC to induce T suppressor cells (data not shown). These findings suggest that day 7 cDC were not necessarily dysfunctional but rather were qualitatively different from day 2 cDC (i.e., tolerogenic in character).

FIGURE 4.

CD4+ T cell priming and phenotype. A, Priming. HA-peptide–pulsed day 0, 2 and 7 cDC were used to prime naive CD4+ T cells isolated from TCR-HA mice (1Materials and Methods). Error bars indicate SE. n = 9–11 per time point. B, Phenotype. DC-expanded CD4+ T cells (A) were stained for intracellular IFN-γ and Foxp3. Representative dot plots, relative prevalence of each population (pie charts), and the ratio of Foxp3+ to IFN-γ+ T cells on days 0, 2, and 7 are shown. Error bars indicate SE. n = 9 to 10 per time point.

FIGURE 4.

CD4+ T cell priming and phenotype. A, Priming. HA-peptide–pulsed day 0, 2 and 7 cDC were used to prime naive CD4+ T cells isolated from TCR-HA mice (1Materials and Methods). Error bars indicate SE. n = 9–11 per time point. B, Phenotype. DC-expanded CD4+ T cells (A) were stained for intracellular IFN-γ and Foxp3. Representative dot plots, relative prevalence of each population (pie charts), and the ratio of Foxp3+ to IFN-γ+ T cells on days 0, 2, and 7 are shown. Error bars indicate SE. n = 9 to 10 per time point.

Close modal

The above results identified cDC as the primary APC subset controlling therapy-induced changes in T cell immunity and revealed the critical role of IDO+ cDC in the transition from effector to regulatory phase. Next, we investigated the molecular basis of how treatment itself led to the physiological changes in the cDC subset. The essential role of IFN-γ in the proinflammatory function of IL-12 is well known (25). In contrast, we recently showed that the IL-12–IFN-γ–IDO axis represented a feedback inhibitory pathway in IL-12/GM-CSF therapy (10). We therefore hypothesized that the observed therapy-induced changes in cDC function could be due to the dichotomous activity of IFN-γ. To test this notion, posttreatment effector and suppressor responses were analyzed in IFN-γ KO mice. Analysis of cDC migration, activation, and priming function in wild-type versus IFN-γ KO mice demonstrated that IFN-γ was essential to these processes (Fig. 5A–C). Similarly, analysis of IDO levels in day 0 versus day 7 cDC revealed that in the absence of IFN-γ, treatment with IL-12/GM-CSF failed to induce IDO. These findings confirmed the dual role of IFN-γ in both immune activation and regulation and identified cDC as the primary conduit mediating the immune-modulatory activity of this cytokine in our model.

FIGURE 5.

The requirement for IFN-γ in the development of immunogenic and tDC. A, Migration to TDLN. CD45+MHC II+CD11c+PDCATDLN cDC were quantified on days 0 and 2 in IL-12/GM-CSF–treated wild-type and IFN-γ KO mice. B, Activation. Percent CD86+ cDC were quantified as in A. C, Priming. The ability of day 0, 2, and 7 TDLN cDC to expand HA-specific CD8+ T cells was determined. D, IDO induction. TDLN cDC were stained for intracellular IDO on days 0, 2, and 7 posttherapy. Error bars indicate SE. n = 4–10 per group for all panels. No significant differences were observed in IFN-γ KO mice on days 0, 2, and 7 in any of the studies.

FIGURE 5.

The requirement for IFN-γ in the development of immunogenic and tDC. A, Migration to TDLN. CD45+MHC II+CD11c+PDCATDLN cDC were quantified on days 0 and 2 in IL-12/GM-CSF–treated wild-type and IFN-γ KO mice. B, Activation. Percent CD86+ cDC were quantified as in A. C, Priming. The ability of day 0, 2, and 7 TDLN cDC to expand HA-specific CD8+ T cells was determined. D, IDO induction. TDLN cDC were stained for intracellular IDO on days 0, 2, and 7 posttherapy. Error bars indicate SE. n = 4–10 per group for all panels. No significant differences were observed in IFN-γ KO mice on days 0, 2, and 7 in any of the studies.

Close modal

Our results demonstrate that IFN-γ, the primary effector cytokine that is induced by IL-12, promotes a feedback regulatory loop limiting the duration of the antitumor T cell activity in IL-12/GM-CSF–treated mice. The data further establish that the dual function of IFN-γ is mediated via cDC, which orchestrate both the initial T effector cell response and the subsequent IDO-dependent regulatory rebound in the TDLN. Separately, the data also reveal the cellular basis of the synergy observed between IL-12 and GM-CSF (i.e., optimal maturation and homing of treatment-generated DC to the TDLN). These findings further delineate the cellular and molecular mechanisms underlying the transient nature of IL-12–mediated antitumor immune activity in particular and of therapies that are designed to induce Th1 responses in general.

An important hypothesis forwarded in this study is that both posttherapy effector and suppressor T cell responses are primed by the same cDC population in the TDLN. This notion is primarily supported by our observation that immunogenic cDC that internalize FITC+ beads in the tumor rapidly home to and persist in the TDLN for up to 7 d, the interval during which the functional switch occurs. Consistent with our findings, a recent study demonstrated a similar sequential change in DC phenotype following stimulation with LPS and IFN-γ in vitro (20). Although these data support a role for IFN-γ in DC plasticity, our results cannot rule out the possibility that the observed switch is associated with the differentiation of a separate tDC population independent of the iDC. Further characterization of the DC populations in tumors and TDLN will have to be performed to determine whether posttherapy iDC and tDC truly represent the same cell at distinct stages of differentiation or whether they arise from separate populations with different etiology.

Physiological expression of IDO is generally associated with pDC, and it has been suggested that CD11c+B220+CD19+ pDC may play a unique role in IDO-dependent expansion of CD4+CD25+Foxp3+ T cells in TDLN of tumor-bearing mice (28). In the setting of IFN-γ–driven immunity, however, expression of IDO in CD11c+MHC II+B220 cDC, but not in CD11c+MHC II+B220+ pDC, correlated with the switch from an effector to a suppressor response in the TDLN. Importantly, IDO+ pDC, which are dominant over iDC in the steady state (28), could not override IDO immunogenic cDC on day 2. Analysis of pDC and cDC subsets for CD19 expression revealed that both groups contained CD19+ cells, although the proportion of these cells in the cDC population was significantly higher (27 versus 7% in cDC and pDC, respectively; data not shown). Moreover, both CD19+ and CD19 cDC expressed high amounts of IDO on day 7 (data not shown). The respective roles of CD19+ and CD19 cDC subsets in posttherapy T suppressor cell expansion is currently under investigation. Collectively, these data establish that pDC do not play a significant role in IL-12/GM-CSF–induced changes in T cell immunity in our model.

An unexpected finding was the phenotypic and functional differences observed between day 0 and 7 cDC with tolerogenic activity. Day 0 tumor-induced tDC were superior to day 7 treatment-induced tDC in their ability to expand CD8+ T cells. Moreover, day 0 tDC expressed high levels of IL-10 but no appreciable IDO, and their priming ability could not be rescued by D1-MT. In contrast, day 7 tDC expressed low levels of IL-10, high levels of IDO, and their priming function could be fully restored by D1-MT. tDC can be divided into multiple subsets based on their maturation stage, cytokine profile, and the types of suppressive mechanisms they employ (33, 34). The IL-10hiIL-12loTNF-αlo IDO-negative phenotype of day 0 tumor-associated tDC is consistent with tumor-tolerized semimature DC, which are induced via tumor-released immune suppressive cytokines (32, 33, 35) and/or persistent exposure to tumor-infiltrating T suppressor cells (35). To this end, Line-1 cells produce abundant IL-10 and TGF-β (M.O. Kilinc and N.K. Egilmez, unpublished observations), and attract CD4+CD25+Foxp3+ T cells (11). In contrast, treatment-induced IL-10loTNF-αloIL-12lo IDO-positive day 7 tDC are phenotypically more similar to the IFN-γ–induced IDO+CD8α+ tDC (36). Consistent with this notion, a significant proportion (∼30%) of day 7 IDO+ cDC were found to be CD8α positive (data not shown).

The molecular pathways that mediate the dichotomous effects of IFN-γ on DC phenotype and function were not investigated in this study. IFN-γ signaling is complex and involves multiple parallel as well as converging phosphorylation cascades (19, 37). More specifically, in addition to the canonical JAK–STAT pathway, IFN-γ has been shown to signal via MAPK, PI3K, and NF-κB (37), ultimately regulating the expression of >200 immunologically relevant transcripts (17, 19). Among the numerous transcription factors induced, STAT-1–dependent IFN regulatory factor (IRF) family of proteins can influence the effector versus regulatory decision (37). For example, IRF-1 has been shown to be essential to DC immunogenicity (38), whereas IRF-8 has been linked to IDO expression (36). Studies are currently underway to determine whether the STAT-1–IRF axis plays an important role in post–IL-12/GM-CSF DC plasticity. Delineation of these pathways can result in the identification of checkpoints that may be targeted for selective blocking of tolerogenic signals resulting in enhanced immunogenic activity and tumor kill.

This work was supported by National Institutes of Health/National Cancer Institute Grant R01-CA100656 and New York State Office of Science Technology and Academic Research Award C040070.

The online version of this article contains supplemental material.

Abbreviations used in this article:

cDC

conventional myeloid dendritic cells

DC

dendritic cells

D1-MT

D-1-methyl tryptophan

HA

hemagglutinin

iDC

immunogenic dendritic cells

IRF

IFN regulatory factor

KO

knockout

pDC

plasmacytoid dendritic cells

PDCA

plasmacytoid dendritic cell Ag

tDC

tolerogenic dendritic cells

TDLN

tumor-draining lymph nodes.

1
Dougan
M.
,
Dranoff
G.
.
2009
.
Immune therapy for cancer.
Annu. Rev. Immunol.
27
:
83
117
.
2
Monsurrò
V.
,
Wang
E.
,
Yamano
Y.
,
Migueles
S. A.
,
Panelli
M. C.
,
Smith
K.
,
Nagorsen
D.
,
Connors
M.
,
Jacobson
S.
,
Marincola
F. M.
.
2004
.
Quiescent phenotype of tumor-specific CD8+ T cells following immunization.
Blood
104
:
1970
1978
.
3
Rosenberg
S. A.
,
Sherry
R. M.
,
Morton
K. E.
,
Scharfman
W. J.
,
Yang
J. C.
,
Topalian
S. L.
,
Royal
R. E.
,
Kammula
U.
,
Restifo
N. P.
,
Hughes
M. S.
, et al
.
2005
.
Tumor progression can occur despite the induction of very high levels of self/tumor antigen-specific CD8+ T cells in patients with melanoma.
J. Immunol.
175
:
6169
6176
.
4
Mantovani
A.
,
Romero
P.
,
Palucka
A. K.
,
Marincola
F. M.
.
2008
.
Tumour immunity: effector response to tumour and role of the microenvironment.
Lancet
371
:
771
783
.
5
Whiteside
T. L.
2008
.
The tumor microenvironment and its role in promoting tumor growth.
Oncogene
27
:
5904
5912
.
6
Peggs
K. S.
,
Quezada
S. A.
,
Allison
J. P.
.
2008
.
Cell intrinsic mechanisms of T-cell inhibition and application to cancer therapy.
Immunol. Rev.
224
:
141
165
.
7
Curran
M. A.
,
Montalvo
W.
,
Yagita
H.
,
Allison
J. P.
.
2010
.
PD-1 and CTLA-4 combination blockade expands infiltrating T cells and reduces regulatory T and myeloid cells within B16 melanoma tumors.
Proc. Natl. Acad. Sci. USA
107
:
4275
4280
.
8
Hill
H. C.
,
Conway
T. F.
 Jr.
,
Sabel
M. S.
,
Jong
Y. S.
,
Mathiowitz
E.
,
Bankert
R. B.
,
Egilmez
N. K.
.
2002
.
Cancer immunotherapy with interleukin 12 and granulocyte-macrophage colony-stimulating factor-encapsulated microspheres: coinduction of innate and adaptive antitumor immunity and cure of disseminated disease.
Cancer Res.
62
:
7254
7263
.
9
Nair
R. E.
,
Jong
Y. S.
,
Jones
S. A.
,
Sharma
A.
,
Mathiowitz
E.
,
Egilmez
N. K.
.
2006
.
IL-12 + GM-CSF microsphere therapy induces eradication of advanced spontaneous tumors in her-2/neu transgenic mice but fails to achieve long-term cure due to the inability to maintain effector T-cell activity.
J. Immunother.
29
:
10
20
.
10
Gu
T.
,
Rowswell-Turner
R. B.
,
Kilinc
M. O.
,
Egilmez
N. K.
.
2010
.
Central role of IFNgamma-indoleamine 2,3-dioxygenase axis in regulation of interleukin-12-mediated antitumor immunity.
Cancer Res.
70
:
129
138
.
11
Kilinc
M. O.
,
Aulakh
K. S.
,
Nair
R. E.
,
Jones
S. A.
,
Alard
P.
,
Kosiewicz
M. M.
,
Egilmez
N. K.
.
2006
.
Reversing tumor immune suppression with intratumoral IL-12: activation of tumor-associated T effector/memory cells, induction of T suppressor apoptosis, and infiltration of CD8+ T effectors.
J. Immunol.
177
:
6962
6973
.
12
Kilinc
M. O.
,
Gu
T.
,
Harden
J. L.
,
Virtuoso
L. P.
,
Egilmez
N. K.
.
2009
.
Central role of tumor-associated CD8+ T effector/memory cells in restoring systemic antitumor immunity.
J. Immunol.
182
:
4217
4225
.
13
Kilinc
M. O.
,
Rowswell-Turner
R. B.
,
Gu
T.
,
Virtuoso
L. P.
,
Egilmez
N. K.
.
2009
.
Activated CD8+ T-effector/memory cells eliminate CD4+ CD25+ Foxp3+ T-suppressor cells from tumors via FasL mediated apoptosis.
J. Immunol.
183
:
7656
7660
.
14
Nair
R. E.
,
Kilinc
M. O.
,
Jones
S. A.
,
Egilmez
N. K.
.
2006
.
Chronic immune therapy induces a progressive increase in intratumoral T suppressor activity and a concurrent loss of tumor-specific CD8+ T effectors in her-2/neu transgenic mice bearing advanced spontaneous tumors.
J. Immunol.
176
:
7325
7334
.
15
Sercan
Ő.
,
Hämmerling
G. J.
,
Arnold
B.
,
Schüler
T.
.
2006
.
Innate immune cells contribute to the IFN-γ-dependent regulation of antigen-specific CD8+ T cell homeostasis.
J. Immunol.
176
:
735
739
.
16
Berner
V.
,
Liu
H.
,
Zhou
Q.
,
Alderson
K. L.
,
Sun
K.
,
Weiss
J. M.
,
Back
T. C.
,
Longo
D. L.
,
Blazar
B. R.
,
Wiltrout
R. H.
, et al
.
2007
.
IFN-γ mediates CD4+ T-cell loss and impairs secondary antitumor responses after successful initial immunotherapy.
Nat. Med.
13
:
354
360
.
17
Boehm
U.
,
Klamp
T.
,
Groot
M.
,
Howard
J. C.
.
1997
.
Cellular responses to interferon-γ.
Annu. Rev. Immunol.
15
:
749
795
.
18
Steinman
R. M.
2007
.
Dendritic cells: understanding immunogenicity.
Eur. J. Immunol.
37
(
Suppl 1
):
S53
S60
.
19
Schroder
K.
,
Hertzog
P. J.
,
Ravasi
T.
,
Hume
D. A.
.
2004
.
Interferon-γ: an overview of signals, mechanisms and functions.
J. Leukoc. Biol.
75
:
163
189
.
20
Jürgens
B.
,
Hainz
U.
,
Fuchs
D.
,
Felzmann
T.
,
Heitger
A.
.
2009
.
Interferon-γ-triggered indoleamine 2,3-dioxygenase competence in human monocyte-derived dendritic cells induces regulatory activity in allogeneic T cells.
Blood
114
:
3235
3243
.
21
Kirberg
J.
,
Baron
A.
,
Jakob
S.
,
Rolink
A.
,
Karjalainen
K.
,
von Boehmer
H.
.
1994
.
Thymic selection of CD8+ single positive cells with a class II major histocompatibility complex-restricted receptor.
J. Exp. Med.
180
:
25
34
.
22
Morgan
D. J.
,
Liblau
R.
,
Scott
B.
,
Fleck
S.
,
McDevitt
H. O.
,
Sarvetnick
N.
,
Lo
D.
,
Sherman
L. A.
.
1996
.
CD8(+) T cell-mediated spontaneous diabetes in neonatal mice.
J. Immunol.
157
:
978
983
.
23
Braun
D.
,
Longman
R. S.
,
Albert
M. L.
.
2005
.
A two-step induction of indoleamine 2,3 dioxygenase (IDO) activity during dendritic-cell maturation.
Blood
106
:
2375
2381
.
24
Curtsinger
J. M.
,
Mescher
M. F.
.
2010
.
Inflammatory cytokines as a third signal for T cell activation.
Curr. Opin. Immunol.
22
:
333
340
.
25
Trinchieri
G.
2003
.
Interleukin-12 and the regulation of innate resistance and adaptive immunity.
Nat. Rev. Immunol.
3
:
133
146
.
26
Jinushi
M.
,
Hodi
F. S.
,
Dranoff
G.
.
2008
.
Enhancing the clinical activity of granulocyte-macrophage colony-stimulating factor-secreting tumor cell vaccines.
Immunol. Rev.
222
:
287
298
.
27
Puccetti
P.
2007
.
On watching the watchers: IDO and type I/II IFN.
Eur. J. Immunol.
37
:
876
879
.
28
Sharma
M. D.
,
Baban
B.
,
Chandler
P.
,
Hou
D.-Y.
,
Singh
N.
,
Yagita
H.
,
Azuma
M.
,
Blazar
B. R.
,
Mellor
A. L.
,
Munn
D. H.
.
2007
.
Plasmacytoid dendritic cells from mouse tumor-draining lymph nodes directly activate mature Tregs via indoleamine 2,3-dioxygenase.
J. Clin. Invest.
117
:
2570
2582
.
29
Belladonna
M. L.
,
Volpi
C.
,
Bianchi
R.
,
Vacca
C.
,
Orabona
C.
,
Pallotta
M. T.
,
Boon
L.
,
Gizzi
S.
,
Fioretti
M. C.
,
Grohmann
U.
,
Puccetti
P.
.
2008
.
Cutting edge: Autocrine TGF-β sustains default tolerogenesis by IDO-competent dendritic cells.
J. Immunol.
181
:
5194
5198
.
30
Matteoli
G.
,
Mazzini
E.
,
Iliev
I. D.
,
Mileti
E.
,
Fallarino
F.
,
Puccetti
P.
,
Chieppa
M.
,
Rescigno
M.
.
2010
.
Gut CD103+ dendritic cells express indoleamine 2,3-dioxygenase which influences T regulatory/T effector cell balance and oral tolerance induction.
Gut
59
:
595
604
.
31
Murphy
K. M.
,
Stockinger
B.
.
2010
.
Effector T cell plasticity: flexibility in the face of changing circumstances.
Nat. Immunol.
11
:
674
680
.
32
Ghiringhelli
F.
,
Puig
P. E.
,
Roux
S.
,
Parcellier
A.
,
Schmitt
E.
,
Solary
E.
,
Kroemer
G.
,
Martin
F.
,
Chauffert
B.
,
Zitvogel
L.
.
2005
.
Tumor cells convert immature myeloid dendritic cells into TGF-β-secreting cells inducing CD4+CD25+ regulatory T cell proliferation.
J. Exp. Med.
202
:
919
929
.
33
Rutella
S.
,
Danese
S.
,
Leone
G.
.
2006
.
Tolerogenic dendritic cells: cytokine modulation comes of age.
Blood
108
:
1435
1440
.
34
Morelli
A. E.
,
Thomson
A. W.
.
2007
.
Tolerogenic dendritic cells and the quest for transplant tolerance.
Nat. Rev. Immunol.
7
:
610
621
.
35
Lin
A.
,
Schildknecht
A.
,
Nguyen
L. T.
,
Ohashi
P. S.
.
2010
.
Dendritic cells integrate signals from the tumor microenvironment to modulate immunity and tumor growth.
Immunol. Lett.
127
:
77
84
.
36
Orabona
C.
,
Puccetti
P.
,
Vacca
C.
,
Bicciato
S.
,
Luchini
A.
,
Fallarino
F.
,
Bianchi
R.
,
Velardi
E.
,
Perruccio
K.
,
Velardi
A.
, et al
.
2006
.
Toward the identification of a tolerogenic signature in IDO-competent dendritic cells.
Blood
107
:
2846
2854
.
37
Gough
D. J.
,
Levy
D. E.
,
Johnstone
R. W.
,
Clarke
C. J.
.
2008
.
IFNgamma signaling-does it mean JAK-STAT?
Cytokine Growth Factor Rev.
19
:
383
394
.
38
Gabriele
L.
,
Fragale
A.
,
Borghi
P.
,
Sestili
P.
,
Stellacci
E.
,
Venditti
M.
,
Schiavoni
G.
,
Sanchez
M.
,
Belardelli
F.
,
Battistini
A.
.
2006
.
IRF-1 deficiency skews the differentiation of dendritic cells toward plasmacytoid and tolerogenic features.
J. Leukoc. Biol.
80
:
1500
1511
.

N.K.E. has ownership interest in TherapyX, Inc., which is developing sustained-release formulations of cytokines for cancer therapy.