IL-21 is a key factor in the transition between innate and adaptive immune responses. We have used the cytokine gene therapy approach to study the antitumor responses mediated by IL-21 in the B16F1 melanoma and MethA fibrosarcoma tumor models in mice. Retrovirally transduced tumor cells secreting biologically functional IL-21 have growth patterns in vitro similar to that of control green fluorescent protein-transduced cells, but are completely rejected in vivo. We show that IL-21 activates NK and CD8+ T cells in vivo, thus mediating complete rejection of poorly immunogenic tumors. Rejection of IL-21-secreting tumors requires the presence of cognate IL-21R and does not depend on CD4+ T cell help. Interestingly, perforin, but not IFN-γ or other major Th1 and Th2 cytokines (IL-12, IL-4, or IL-10), is required for the IL-21-mediated antitumor response. Moreover, IL-21 results in 50% protection and 70% cure of nonimmunogenic tumors when given before and after tumor challenge, respectively, in C57BL/6 mice. We conclude that IL-21 immunotherapy warrants clinical evaluation as a potential treatment for cancer.
Interleukin-21 and its receptor (IL-21R) have been recently characterized and described (1). Secreted by activated T cells, IL-21 is a 131-aa residue, four-helix bundle cytokine with sequence homologies to IL-2 and IL-15 (1). IL-21R is expressed in lymphoid tissues, in particular by NK, B, T, and dendritic cells, macrophages, and endothelial cells (1) (data not shown). IL-21R has the highest sequence homology to IL-2Rβ chain and IL-4Rα chain (2), which associates with the common γ cytokine receptor chain (γc)3 (2, 3) upon ligand binding. The widespread lymphoid distribution of IL-21R suggests that IL-21 may potentially play a substantial role in immune regulation. Indeed, in vitro studies have shown that IL-21 significantly modulates the function of B, T, and NK cells (1, 4, 5, 6). IL-21 potentiates maturation of NK cell from bone marrow progenitors and activation of peripheral NK cells in human assay systems (1). In murine systems, Kasaian et al. (4) have demonstrated that IL-21 limits ongoing NK cell expansion while promoting NK effector functions and Ag-specific CD8+ T cell responses. More recently, Wurster et al. (7) have shown that IL-21 specifically inhibits IFN-γ production from developing Th1 cells and is preferentially expressed by Th2 cells. These data suggest that IL-21 may play a unique role in fine tuning the response of B, T, and NK cells, depending on the type of stimulus and the phenotype of immune cells.
The cytokine gene therapy approach as a form of molecular pharmacology applied to tumor models has contributed significantly to identifying immune responses mediated by cytokines that were previously either unknown or not fully appreciated (8). In this study we have used the cytokine gene-transfer technology in two tumor models and studied the in vivo biology of IL-21. Our results show that IL-21/IL21R interactions have a unique role in sequentially activating both innate and adaptive immune responses against poorly immunogenic tumors, leading to tumor rejection that is perforin (pfp) dependent but IFN-γ independent. More important, we demonstrate that IL-21 stimulates potent prophylactic and therapeutic immunity that leads to the cure of tumors.
Materials and Methods
Female C57BL/6, BALB/c, IL-4−/−, IL-10−/−, IFN-γ−/−, IL-12−/− (i.e., p35−/−), and pfp−/− deficient mice in C57BL/6 background were purchased from The Jackson Laboratory (Bar Harbor, ME). C57BL/6 SCID and nude mice were purchased from Taconic Farms (Germantown, NY). IL-21R−/− mice (4) were maintained at Charles River Breeding Laboratories (Andover, MA). All mice were maintained and treated in accordance with National Institutes of Health and American Association of Laboratory Animal Care regulations.
Tumor cell lines and reagents
B16F1 melanoma cells were maintained in culture in DMEM medium supplemented with 10% heat inactivated FBS, 2% glutamine, and 1% penicillin-streptomycin. MethA fibrosarcoma cells were maintained by i.p. passage in BALB/c mice. B16F1 melanoma-specific tyrosinase related protein-2 (TRP-2) (SVYDFFVWL) (9, 10) and control peptide OVA257–264 (SIINFEKL) were synthesized at Wyeth Research (Cambridge, MA). mAbs anti-CD3, anti-CD28, and rat IgG2a and IL-2 cytokine used in this paper were all purchased from BD PharMingen (San Diego, CA).
Generation of IL-21 and green fluorescent protein (GFP)-expressing tumor cells
B16F1 and MethA tumor cells were engineered to express IL-21 and GFP, or only GFP. Retroviral vectors encoding mIL21-IRES-GFP or IRES-GFP were constructed using GFP-RV vector (kindly provided by Dr. K. Murphy, Washington University, St. Louis, MO) (11). High titer retrovirus was obtained by transfecting the 293-vesicular stomatitus virus G ecotropic packaging cell line (12). Spin infections were performed at 1800 rpm for 40 min at room temperature. Cells were infected three times. Tumor cells expressing GFP were enriched by flow sorting, and the purity of GFP-expressing cells was >90%.
In vivo tumor studies
C57BL/6 mice were shaved on the back and injected i.d. with 105 B16F1-IL-21 or control B16F1-GFP. BALB/c mice were injected with either 106 or 2 × 106 MethA-IL-21 or control MethA-GFP cells. Tumor growth was monitored by measuring perpendicular diameters with a caliper. Mice were sacrificed when the tumors displayed severe ulceration or reached a size of 200 mm2. In general, 10 mice per group were used in each experiment, and tumor size averages from each group are shown. Results of this study (except in vivo depletion studies) represent experiments repeated at least two times with similar results. The difference in tumor size between the control and experimental groups was statistically analyzed using Student’s t test.
In vivo depletion studies
CD4+ or CD8+ T cell depletion was accomplished by i.p. injecting 400 μg of either anti-CD4 (GK1.5), anti-CD8 (53-6.7) mAbs, or rat IgG isotype control per mouse for three consecutive days before tumor cell injection. Ab injections were continued every other day after tumor cell injection for 12 days. For NK cell depletion, 50 μl of rabbit anti-mouse/rat asialo GM1 polyclonal Ab (Cedarlane Laboratories, Ontario, CA) was injected i.p. 1 day before tumor cell injection. Mice similarly injected with normal rabbit serum were used as controls. After tumor cell inoculation, Ab injections were continued twice per week for 2 wk. T and NK cell depletion was confirmed in lymph nodes and spleens 1 day before tumor challenge (for T cells) or on the same day of tumor cell challenge (for NK cells) by flow cytometry using relevant primary Abs followed by biotinylated secondary Abs. FACS analysis showed that >99% of the relevant population of T cells or NK cells were depleted in mice treated with anti-CD4, anti-CD8, and anti-asialo GM1. In contrast, mice treated with isotype controls displayed T lymphocyte profiles similar to the profiles of untreated mice (data not shown).
Overnight supernatants from 106 B16F1-IL-21, B16F1-GFP, MethA-IL-21, and MethA-GFP tumor cells were assayed for IL-21 levels by ELISA. In brief, murine IL-21RmIgG2a Fc fusion protein (Wyeth Research) was used as coating Ab, and anti-mouse IL-21 (R&D Systems, Minneapolis, MN) was used as capture Ab. Purified mIL-21 (Wyeth Research) (4) was used as control. The detection limit for IL-21 is 12.5 ng/ml.
IL-21R mRNA expression detected by TaqMan
RNA was isolated from different tumor cell lines according to the manufacturer’s instructions (Promega, Madison, WI). mRNA extracted from splenocytes that had been activated with plate-bound anti-CD3 (1 μg/ml) and anti-CD28 (1 μg/ml) mAbs (BD PharMingen) for 24 h was used for positive control. Purified RNA was treated with DNase (Ambion, Austin, TX) and adjusted to a concentration of 50 ng/μl before mRNA analysis by quantitative TaqMan PCR analysis. IL-21R and cyclophylin-specific primer pairs and probes were designed using PrimerExpress software and were prepared by Wyeth Research (primers: 5′-GCCTTCTCAGGACGCTATGAT-3′ and 5′-CCCTACAGCACGTAGTTGGA-3′ and probe TCCTGGGACTCAGCTTATGACGAACC). Standard curves for each gene were generated with RNA from known IL-21R-expressing cells. mRNA expression in control and transduced cell lines was normalized based on cyclophilin expression in each cell line, and the results are presented as relative units of mRNA.
Splenocytes (2 × 105 cells/well) from either C57BL/6 or BALB/c mice were stimulated with various concentrations of syngeneic irradiated tumor cells that expressed either GFP or IL-21 in 96-well plates. [3H]Thymidine at 1 μCi/well (PerkinElmer, Boston, MA) was added during the last 6 h of culture. After harvesting the supernatant onto glass fiber filter mats, [3H]Thymidine incorporation was determined by liquid scintillation counting.
TRP-2-specific T cell responses were determined by the IFN-γ ELISPOT kit (R&D Systems) following manufacturer’s instructions. Splenocytes (2 × 105–4 × 105) in 200 μl of medium containing 20 U/ml murine IL-2 (BD PharMingen) were placed in each well in the presence of 5 μg/ml specific TRP-2 peptides (9) or nonspecific OVA peptides. The plate was incubated for 24 h at 37°C in a CO2 incubator. Plates were then incubated overnight at 4°C with detection Ab, followed by a 2-hr incubation with streptavidin-alkaline phosphatase conjugate. Spots were visualized with 5-bromo-4 chloro-3′ indolylphosphate p-toluidine salt/nitro bluetetrazolium chloride alkaline phosphatase substrate (R&D Systems). Plates were washed with tap water and air dried, and spots were counted with a stereomicroscope and recalculated to 106 cells with background spots subtracted. Generally, <10 spots/well were detected when OVA peptide was used as Ag.
In vitro restimulation of splenocytes from tumor cell-inoculated mice
Tumor peptide-specific T cell lines were generated as described elsewhere (13). In brief, mice were inoculated with either B16F1-GFP or B16F1-IL-21 cells. After 8–11 days, splenocytes were harvested and cultured with 5 μg/ml TRP-2 peptide (9). On the third day of culture, 20 U/ml IL-2 (BD PharMingen) was added to each culture. After 5 days, cells were used for 51Cr release assay.
Cytotoxicity against targets was quantified using a 4-h 51Cr release assay. RMA-S cells (generously provided by Dr. K. Karre, Karolinska Institute, Stockholm, Sweden) were pulsed with TRP-2 peptide at 10 μg/ml and labeled with Na251CrO4 (PerkinElmer) for 1 h at 37°C. After washing, 51Cr-labeled target cells were incubated with T cell lines generated from C57BL/6 mice injected with tumor cells described earlier at different E:T ratios in 96 round-bottom plates. After a 4-h incubation at 37°C, supernatants were collected and radioactivity was detected in a scintillation counter (Wallac, Turku, Finland). Percent-specific lysis was calculated as 100 × [(release by CTL − spontaneous release)/(maximal release − spontaneous release)]. Maximal release was determined by the addition of 1% Triton X-100. Spontaneous release in the absence of CTL was generally <15% of maximal release.
Detection of tumor Ag-specific CD8+ cell using tetramers
APC-conjugated MHC Kb tetramer complexes, as described elsewhere, (10) were purchased from Beckman Coulter (San Diego, CA). Draining lymph node cells from naive mice and mice injected with B16F1-IL-21 or B16F1-GFP tumor cells were stained with APC-labeled tetramers, FITC-labeled anti-CD3 (BD PharMingen), and PE-labeled anti-CD8 (BD PharMingen). The percentage of tetramer-positive cells was gated on CD8+ and CD3+ double-positive populations.
Prophylactic and therapeutic treatment of tumor
For prophylactic treatment, C57BL/6 mice were first vaccinated with 2 × 105 of irradiated (4000 rad) B16F1-GFP or B16F1-IL-21 tumor cells once or twice (1 wk apart) into the left flank. One week after vaccination, mice were challenged with 105 B16F1 tumor cells into the right flank. For therapeutic treatment, mice were first challenged with 105 of B16F1 cells on the back. One day or later, as indicated, treatment was initiated by injecting 106 of irradiated B16F1-IL-21 cells s.c into the left flank.
IL-21-transduced B16F1 and MethA cells secrete biologically functional IL-21
B16F1 melanoma and MethA fibrosarcoma tumor cells were transduced to express GFP plus IL-21 (B16F1/MethA-IL-21) or GFP (B16F1/MethA-GFP), respectively. GFP-positive cells were sorted and expanded, and IL-21 levels secreted by these cell lines were determined by IL-21 ELISA. B16F1-IL-21 and MethA-IL-21 cells, but not B16F1-GFP or MethA-GFP tumor cells, secreted a substantial amount of IL-21 in overnight cultures (Fig. 1,A). To determine whether the IL-21 cytokine secreted by the transduced cells was biologically functional, irradiated IL-21 or GFP-expressing tumor cells were used to stimulate naive syngeneic splenocytes from C56BL/6 or BALB/c mice in the presence of suboptimal amounts of anti-CD3 and anti-CD28. B16F1-IL-21 enhanced naive splenocyte proliferation when compared with control GFP-expressing cells at all concentrations tested (Fig. 1 B). Similar results were obtained with MethA-IL-21 cells (data not shown). These results suggest that IL-21 secreted by transduced tumor cells is biologically functional.
IL-21 does not affect tumor cell growth in vitro, but inhibits tumor formation in vivo
The in vitro growth kinetics of IL-21-producing tumor cells were very similar to that of the GFP-expressing control tumor cells (data not shown). This indicates that IL-21 does not have any apparent effect on the in vitro growth characteristics of transduced tumor cells. The unresponsiveness of tumor cells to IL-21 was further confirmed with the lack of IL-21R expression by tumor cells in real time PCR analysis. IL-21R expression was detected in anti-CD3- and anti-CD28-activated splenocytes, but not in any of the transduced tumor cell lines (Fig. 2 A).
To assess the effect of paracrine-secreted IL-21 on the immune system during tumor growth in vivo, B16F1-IL-21 or MethA-IL-21 tumor cells were inoculated i.d. into the flank of syngeneic mice. B16F1 tumor cells are poorly immunogenic in that previous vaccination with irradiated wild-type tumor cells only protects 20% of vaccinated mice against subsequent live B16F1 challenge (14). In contrast, MethA, a methycholantheren-induced fibrosarcoma, is an immunogenic tumor model in which vaccination with irradiated MethA cells leads to almost 100% protection against subsequent live MethA challenge (data not shown). Of interest, the only cytokines that have been reported to prevent B16F1 tumor formation in mice are IL-2 and IL-10 (14, 15).
In our experiments there was no tumor formation in any of the C57BL/6 mice inoculated with B16F1-IL-21 tumor cells for >27 wk after tumor injection (Fig. 2,B). By contrast, all C57BL/6 mice bearing B16F1-GFP cells grew tumors starting as early as day 9. Control tumors increased rapidly in size, and these mice had to be sacrificed 2–3 wk after tumor cell inoculation because of a heavy tumor burden. In the MethA model, small but palpable tumor masses were detected 1 wk after tumor inoculation with either MethA-IL-21 or MethA-GFP cells in BALB/c mice (Fig. 2 C). However, MethA-IL-21 tumors gradually reduced in size starting from wk 2 (day 11) and eventually regressed completely in 100% of mice, whereas 80% of control MethA-GFP tumors continued to grow in size until the mice were sacrificed. These results reveal the potency of IL-21 to trigger immune responses that lead to eradication of both immunogenic and nonimmunogenic tumors.
IL-21-induced anti-tumor response requires the presence of its cognate receptor (IL-21R)
The homology of IL-21 and IL-21R with other cytokines and cytokine receptors raises the question whether IL-21-induced anti-tumor responses require the interaction of IL-21 with its cognate receptor IL-21R. To address this question, B16F1-IL-21 cells were injected into the flank of IL-21R−/− mice (4) or control C57BL/6 mice. IL-21-expressing B16F1 tumors grew out in 100% of IL-21R−/− mice, but not in the control mice (Fig. 3). Thus, cognate interaction between IL-21 and IL-21R is crucial for the development of immune responses that lead to the rejection of IL-21-secreting tumor cells.
IL-21 requires both innate and adaptive immunity in tumor rejection
Studies with genetically modified tumor cells have highlighted the participation of several types of cells, including neutrophils, eosinophils, mast cells, lymphocytes, and NK cells in tumor rejection (15, 16, 17, 18, 19, 20). To determine the relative roles of lymphocytes in the rejection of IL-21-transduced tumor cells, equal numbers (105) of B16F1-IL-21 or control B16F1-GFP cells were injected into T and B cell-deficient (SCID) mice. Both B16F1-IL-21 and B16F1-GFP cells showed similar growth kinetics in SCID mice (Fig. 4,A), indicating that lymphocytes (B and/or T cells) are required for IL-21-mediated tumor rejection. The indispensable role of T cells was confirmed in experiments with C57BL/6 nude/nude mice. All nude mice inoculated with B16F1-IL-21 cells developed tumors, albeit with slower growth kinetics (Fig. 4 B).
To further dissect which T cell subset(s) is (are) important for the IL-21-induced effect, we performed in vivo depletion of lymphocyte subpopulations by administering anti-CD4 or anti-CD8 mAbs. B16F1-IL-21 tumors did not grow in CD4+ T cell-depleted mice and in control rat IgG-treated mice (Fig. 4,C). However, palpable tumors grew out in 7 of 10 CD8+ T cell-depleted mice, suggesting that CD8+, but not CD4+, T cells are necessary for the IL-21-induced anti-tumor response (Fig. 4 C). Of interest, the growth of B16F1-IL-21 in CD8+ T cell-depleted mice was significantly delayed, suggesting that cells other than CD8+ T cells may be responsible for the suppression of early phase tumor growth.
Accumulated experimental evidence supports the role of NK cells as a first line of defense in promoting anti-tumor immunity (21, 22, 23). We examined this possibility by injecting equal numbers of B16F1-IL-21 tumor cells into anti-asialo GM1-treated mice or control C57BL/6 mice. As expected, no tumor formation was observed in control C57BL/6 mice, whereas all of the anti-asialo GM1-treated mice grew tumors as early as 2 wk after B16F1-IL-21 tumor cell inoculation (Fig. 4 D). Of note, we also used anti-NK1.1 Ab for NK cell depletion and observed the same results as in anti-asialo GM1-treated mice. This further confirms that NK cells are required for the rejection of B16F1-IL-21 tumors. Taken together, these results indicate that IL-21 activates NK and CD8+ T cells and that tumor eradication requires the participation of both innate and adaptive immunity.
IL-21 supports the generation of IFN-γ-secreting tumor Ag-specific T cells and enhances tumor Ag-specific CTL activity
We have used the B16F1 tumor-specific TRP-2 peptide to evaluate the effect of IL-21 on T cell responses in vivo (9, 10). The presence of TRP-2-specific CD8+ T cells in tumor cell-injected mice was determined with tetramers containing TRP-2 peptides. Single cell suspensions from draining lymph nodes of naive (Fig. 5,A) B16F1-GFP- (Fig. 5,B) or B16F1-IL-21- (Fig. 5C) injected mice were stained with TRP-2 tetramer. Compared with B16F1-GFP-inoculated mice, B16F1-IL-21 cell-injected mice possessed more than a 2-fold increase in the number of tumor-specific CD8+ T cells. As expected, cells from naive mice stained at a background level of tetramer. To examine whether these T cells are functional, splenocytes from mice injected with either IL-21 or GFP-expressing B16F1 cells were stimulated with TPR-2 peptide or OVA control peptide in an IFN-γ ELISPOT assay. After the background with OVA peptide was subtracted, the number of IFN-γ-producing cells in B16F1-IL-21-injected mice was 3-fold higher than that of B16F1-GFP-injected mice (Fig. 5,D). Although TNF-α and IL-10 ELISPOT were performed similarly, no significant differences in spot-forming units were detected between B16F1-GFP and B16F1-IL21 tumor cell-injected mice (data not shown). To further characterize IL-21-mediated anti-tumor T cell responses, splenocytes from either IL-21 or GFP-expressing tumor-injected mice were first stimulated with TRP-2 peptide in vitro before CTL assays. Splenocytes from B16F1-IL-21-injected mice had enhanced cytolytic activity toward TRP-2 peptide-pulsed RMA-S cells compared with splenocytes from GFP-expressing tumor-bearing mice at all E:T ratios (Fig. 5,E). These results indicate that IL-21 enhances the development of tumor Ag-specific cytolytic T cell responses. Of note, TRP-2 is one of the Ags shared between B16F1 tumor cells and normal melanocytes. Thus, CTLs detected in this study are actually autoreactive T cells. Indeed, 10–20% of C57BL/6 mice developed local hair and skin depigmentation at the injection site of B16F1-IL-21 tumor cells, but not at the injection site of control cells (Fig. 5 F).
IL-21-induced anti-tumor response is independent of IFN-γ, IL-12, IL-4, and IL-10, but requires pfp
It is well established now that IFN-γ and IL-12 are the cytokines primarily involved in activating immune cells against tumors, both in murine models and in humans (24, 25, 26). To definitively determine whether these Th1/Th2 cytokines actively participate in IL-21-mediated anti-tumor response, equal numbers of B16F1-IL-21 or B16F1-GFP tumor cells were injected into C57BL/6 mice deficient of IFN-γ, IL-12, IL-10, or IL-4. Interestingly, none of the aforementioned cytokine-deficient mice formed tumors after B16F1-IL-21 cell injection (Fig. 6 A–D). However, B16F1-GFP control cells grew tumors in those mice. The observation that tumor rejection is IFN-γ independent may have clinical importance, because production of IFN-γ during an overwhelming anti-tumor response has been accompanied by severe side effects (27, 28).
Besides cytokines, NK and CD8+ T cells can also reject tumors via a pfp-mediated cytotoxic mechanism (29, 30). To investigate whether IL-21-induced anti-tumor response requires the pfp-mediated pathway, IL-21- or GFP-expressing tumor cells were injected into pfp−/− mice. In these experiments, 100 and 80% of the pfp−/− mice injected with B16F1-GFP or B16F1-IL-21 cells, respectively, developed tumors (Fig. 6,F). Interestingly, B16F1-IL-21 cell-injected mice showed delayed tumor growth compared with control-injected mice (Fig. 6 E). The growth of B16F1-IL-21 cells in pfp−/− mice indicates that IL-21-induced anti-tumor response involves pfp-mediated cytotoxicity.
IL-21 mediates a potent prophylactic and therapeutic effect
In preliminary vaccination experiments, we observed that 30–40% of mice that rejected primary B16F1-IL-21 tumors were also protected from subsequent B16F1 tumor challenge when it was administered 3 wk later in the opposite flank. Nevertheless, none of these mice was protected when challenged with bladder carcinoma MB49 tumor cells (data not shown). Therefore, we sought to evaluate the role of IL-21 in prophylactic and therapeutic treatment of tumors. For prophylactic treatment, mice were vaccinated once or twice with irradiated B16F1-IL-21 or B16F1-GFP 1 wk before B16F1 challenge. Mice vaccinated with irradiated B16F1-IL-21 showed substantially delayed tumor growth when compared with the mice that received irradiated B16F1-GFP cells (Fig. 7,A). Interestingly, irradiated B16F1-IL-21 vaccination followed by boosting conferred better protection than a single vaccination (Fig. 7,A). This suggests that IL-21 can induce effective anti-tumor memory response against subsequent challenge of tumor cells. To evaluate the therapeutic potential of IL-21, irradiated B16F1-IL-21 cells were administrated s.c. 1, 3, and 5 days after mice were challenged s.c. with 105 live B16F1 cells. Injection of irradiated B16F1-GFP cells did not show any therapeutic effect upon B16F1 cell challenge (data not shown). However, significant delay in tumor growth was observed in mice treated with irradiated B16F1-IL-21 cells, and 70% of tumor-bearing mice treated with three doses of irradiated B16F1-IL-21 cells rejected tumors completely (Fig. 7 B). Of note, all of the mice that were cleared of tumors also developed autoimmune depigmentation of the skin in the surrounding as well as more distant areas (data not shown).
In this study we have examined the in vivo biology of IL-21 using tumor models. We validate previous in vitro observations by showing that IL-21 has a very potent in vivo effect on both NK and CD8+ T cells that results in eradication of both immunogenic and nonimmunogenic tumors. Of interest, IL-21-mediated tumor rejection is independent of major Th1 and Th2 cytokines, i.e., IFN-γ, IL-12, IL-4, and IL-10, but requires the pfp-mediated cytotoxic pathway. Finally, we demonstrate that, when used as prophylactic or therapeutic vaccine, IL-21 leads to 50 and 70% tumor rejection, respectively.
The rapid and definitive elimination of IL-21-transduced B16 melanoma tumor cells observed in our studies is in clear contrast with data obtained in other cytokine gene vaccination models, underscoring a unique role of IL-21 in modulating the immune system. Mice immunized with IL-4-, IL-5-, IL-6-, IL-12-, IFN-γ-, TNF-α-, or GM-CSF-transduced B16 vaccines, as reported previously, displayed moderate delays in tumor formation. However, eventually all mice succumbed to lethal tumors (14, 31). Of note, GM-CSF-, IL-5-, IL-6-, and TNF-α-expressing cells caused significant side effects (14, 31). To date, only IL-2 and IL-10 have been reported to induce complete regression of transduced B16 tumors in vivo (14, 15). In our study, syngeneic mice injected with B16F1-IL-21 or MethA-IL-21 tumor cells did not develop any clinically overt tumor for a period of >41 wk after tumor inoculation.
Interestingly, mice injected with B16F1-IL-21 tumor cells developed hair and skin depigmentation around the injection sites. Presumably, autoimmunity against normal mouse melanocytes was induced through (a) common epitope(s) shared with B16 tumor cells in the close proximity of tumor injection sites (9, 10, 31). To our knowledge, IL-21 is the only cytokine that has caused local depigmentation upon injection of cytokine gene-modified B16F1 tumor cells, thus validating its unequivocal ability in activating potent immune responses. However, the fact that flow cytometric analysis of spleen and lymph node cells removed from immunized mice did not show any major changes in cell population (data not shown), suggests that paracrine secretion of IL-21 modulates the immune system without causing overt systemic side effects.
Despite the apparent redundancy in the IL-21R signaling pathway with other γc-dependent cytokine receptors, B16F1-IL-21 tumor grew out in IL-21R−/− mice, but not in C57BL/6 mice. This suggests that IL-21 can act only through its cognate receptor IL-21R. Additionally, this also implies that IL-21 expressed by tumor cells is functioning through cells of host origin. Of note, the outgrowth of B16F1-IL-21 tumors in IL-21R−/− mice should not be attributed to an intrinsic defect in T and NK cells, as IL-21R−/− mice have normal NK and T cell development, and those cells respond to cytokines other than IL-21 (4).
IL-21 promotes innate immune responses by enhancing cytotoxicity and IFN-γ production of NK cells (1, 4). B16F1-IL21 tumors grew out in mice depleted of NK cells, demonstrating that IL-21 is acting on NK cells to potentiate tumor rejection. In these mice tumors grow out quickly after inoculation, consistent with an early role for NK cells in tumor rejection, which is likely to account for the significantly delayed B16F1-IL-21 tumor growth in CD8+ T cell-depleted mice. IL-21 may directly enhance anti-tumor lytic activity of NK cells, and/or NK cell activation may enhance macrophage cytotoxicity and facilitate tumor Ag processing and presentation to T cells (32, 33). Thus, NK cells keep tumor growth under control, during which time the adaptive immune response is initiated. Although NK cells are present in both SCID and nude mice (34), there is no significant delay in tumor formation in those mice. However, these mice also lack CD4+ T cells and NKT cells, in contrast to the CD8-depleted mice. Although CD4+ T cells are not required for rejection of B16F1-IL21 tumors, they may contribute to tumor rejection in the absence of CD8+ cells. Alternatively, the explanation may lie in NKT cells; NKT cells are a component of NK cells that express both NK markers and TCR. They can be promptly activated to release cytokines that may contribute to NK cell activation. Thus, the possible absence of NKT cells in SCID and nude mice (35, 36) may result in rapid initial tumor growth caused by lack of activated NK cells.
IL-21 promotes Ag-specific (adaptive) anti-tumor responses as evidenced by the restoration of IL-21-expressing tumor growth in SCID and nude mice. Both CD4+ and CD8+ T cells have been described to be important for the induction of tumor regression and the development of protective immunity (17, 18, 37, 38). At first glance, the finding that CD8+, but not CD4+, T cells are required for the anti-tumor response of IL-21 is not in agreement with the doctrine that CD4+ T helper cells are required for activation of naive CD8+ T cells (39). However, recent studies have demonstrated that in vivo elimination of CD4+ T cells may actually enhance the anti-tumor effect in cytokine gene therapies. This is accomplished by either skewing the cytokine milieu to Th1 phenotype or removing CD4+CD25+ T regulatory cells (10, 31). Because activated CD4+ T cells are a major source of IL-21 (1), we further speculate that IL-21 released by B16F1-IL-21 may replace the help that is normally provided by CD4+ T cells. Thus, IL-21 may act on CD8+ T cells directly by lowering the threshold of costimulation necessary for their activation and/or, indirectly, by augmenting Ag presentation of the IL-21R-bearing APCs (40, 41, 42). Although B16F1-GFP control tumor cells elicited some level of tumor-specific CTL responses, these responses were not sufficient for tumor clearance. In this regard, as our experiments with tetramers demonstrate, IL-21 potentiates the expansion of Ag-specific CD8+ T cells, which ultimately leads to tumor elimination.
IL-21 also enhances IFN-γ production by both NK and CD8+ T cells as demonstrated by our results and others (1, 4). IFN-γ is a pleiotropic cytokine that can act on both tumor cells and host immunity (32, 43). IFN-γ directly inhibits proliferation of some tumor cells in vitro (44) and indirectly inhibits tumor growth in vivo by suppressing tumor angiogenesis (45, 46). Nevertheless, IFN-γ and other cytokines (i.e., IL-12, IL-4, IL-10) are not required for IL-21-induced anti-tumor responses. Thus, it is likely that IL-21 acts either directly or through mediators other than IFN-γ (and IL-12, IL-4, IL-10) to potentiate the expansion of NK and tumor-specific CD8+ T cells and to enhance their cytotoxic activity via the pfp (and/or Fas/Fas ligand) pathway. Indeed, the fact that B16F1-IL-21 tumors grew in pfp−/− mice strongly suggests that the pfp-mediated cytotoxic pathway is indispensable for IL-21-induced anti-tumor responses. Of note, the observed reduced growth rate of B16F1-IL-21 tumors in pfp−/− mice, compared with B16F1-GFP tumors, indicates that other factors induced by IL-21 may also be involved in tumor surveillance.
Finally, we demonstrate in this study that IL-21 alone has the unique potential to prevent or cure poorly immunogenic B16 melanoma tumors. It has been shown previously that GM-CSF-based cancer vaccines protect mice from developing B16 tumors when given prophylactically (47). However, therapeutic GM-CSF vaccines alone have little to no effect on the eradication of pre-existing tumors, and combination treatment (e.g., with anti-cytotoxic T lymphocyte Ag, CTLA-4 Ab) (48) is required. In our study, irradiated B16F1-IL-21 vaccines protected 50% of mice from subsequent tumors and cured 70% of mice with pre-existing tumors. It is noteworthy that, although historically it has been easier to prevent than cure established tumors, in our study prophylactic IL-21 vaccines are less potent than therapeutic vaccines. The better efficacy of IL-21 in therapeutic vaccines may be because of the effects of IL-21 on both NK and CD8+ cells. IL-21 has been shown to enhance NK effector function, but IL-21 does not support expansion of NK cells (4). Thus, IL-21 may be more potent once NK cells have been activated and have begun to expand. This would allow for better NK cytotoxicity, which may also result in enhanced tumor Ag presentation. In addition, IL-21 can potentiate CTL function, and our data suggest that the effects of IL-21 are more potent after priming of the CTL compartment. Nevertheless, in the prophylactic settings, IL-21 is not available to enhance the effector function of NK or CD8+ T cells upon initial tumor challenge, as IL-21 is primarily produced by activated T cells. Of interest, all the mice that were cleared of tumors in the therapeutic setting (but not in the prophylactic setting) also developed autoimmune de-pigmentation (data not shown), a fact that has been correlated both in murine models and human trials, with the establishment of a potent anti-tumor response (9, 49). To our knowledge IL-21 is the first cytokine so far that has been shown to induce autoimmune depigmentation and thus vigorous anti-tumor response. By the time we submitted this manuscript, Ugai et al. (50) published similar findings in murine colon carcinoma models. Their report also highlighted the importance of IL-21 in inducing effective antitumor responses through NK and CD8+ T cells. In conclusion, the unique antitumor effect and the safety profile of IL-21 described in this report opens new possibilities for immunotherapy in cancer.
We thank Dr. Marion Kasaian (Wyeth Research, Cambridge, MA) for critically reviewing this manuscript and helpful discussion about the biology of NK cells, Dr. K. Karre for providing the RMA-S cell line, and Leslie Lowe and Barbara Sibley for providing ELISA protocols.
M.J.G. is supported by National Institutes of Health Grants AI40171 and GM62135 and awards from the Sandler Family Supporting Foundation and the Mathers Foundation.
Abbreviations used in this paper: γc, common γ cytokine receptor chain; GFP, green fluorescent protein; pfp, perforin; TRP-2, tyrosinase related protein-2.