IL-21 Enhances Tumor-Specific CTL Induction by Anti-DR5 Antibody Therapy1

Tumor cell apoptosis is the basis of many cancer therapies, and tumor-specific T cells are the principal effectors of successful anti-tumor immunotherapies. In this study, we show that induction of tumor cell apoptosis by agonistic mAb against DR5, combined with delayed IL-21 treatment, suppressed tumor growth and pre-established tumor metastases. Synergistic effects of the combination were observed in several tumor models where the target tumor was sensitive to DR5-mediated apoptosis. IL-21 promoted tumor-specific CTL activity and enhanced memory responses to tumor rechallenge. These results indicate that a rational combination of Ab-based therapy that causes tumor cell apoptosis and a cytokine that promotes T cell memory is a useful new strategy for cancer immunotherapy.

C ancers are generally treated by surgical resection, radiotherapy, chemotherapy, and more recently, Ab-based immunotherapy (e.g., trastuzumab (Herceptin) or rituximab (Rituxan); Ref. 1). The therapeutic effect of nonsurgical treatments is based on induction of tumor cell death. Elucidating the molecular basis of apoptosis induced by some members of the TNF family has provided opportunities to develop novel therapeutic strategies (2)(3)(4). TRAIL is one of the most attractive candidates because it preferentially induces apoptosis in a wide variety of transformed cells, but typically not in normal cells (2,5,6). The administration of recombinant soluble forms of TRAIL or agonistic mAb specific for human death-inducing TRAIL receptors (DR4 or DR5) exhibited potent tumoricidal activities against xenografted human tumors (7)(8)(9)(10)(11).
Moreover, we have reported recently that anti-mouse DR5 mAb treatment inhibited tumor growth and metastasis of TRAIL-sensitive tumor cells in an FcR-dependent fashion (12). These data strongly supported the potential utility of anti-DR5 and/or DR4 mAb as a cancer therapeutic capable of inducing tumor cell apoptosis, and such agents have entered early clinical trials. It is appreciated that tumor-specific CTL induction plays a critical role in the successful anti-tumor effects of some mAb-based therapies (13,14). Recently, we demonstrated that anti-DR5 mAb treatment eradicated TRAIL-resistant tumor variants by priming tumor-specific CTL following initial DR5-mediated tumor cell apoptosis (12). However, this treatment alone did not induce regression of pre-established tumors.
Because combining tumor cell apoptosis with T cell activation may result in a more dramatic induction of anti-tumor responses (15,16), we have now examined the combined therapeutic effect of inducing tumor cell death (via anti-DR5 mAb) with the addition of cytokines that promote effector cell function. IL-21, the newest member of the IL-2 family of cytokines, is made by CD4 ϩ T cells and acts on T cells, B cells, NK cells, and dendritic cells (17)(18)(19)(20). More specifically, IL-21 is required for normal Ab responses in mice (21), and it has potent anti-tumor activity of its own in a variety of mouse tumor models (22)(23)(24)(25). In part, it appears that the anti-tumor activity of IL-21 can be attributed to its ability to induce terminal differentiation of NK cells (24) and regulate T cell proliferation and differentiation (25,26). IL-21 enhances NK cellmediated tumor suppression via an NKG2D-dependent mechanism (27). IL-21 also has been used in combination with IL-2, IL-12, IL-15, and the CD1d-reactive glycolipid, ␣-galactosylceramide (␣-GalCer), 3 and great synergy in anti-tumor activity has been observed for IL-15/IL-21 and ␣-GalCer/IL-21 combinations (28,29). Furthermore, IL-21 also has shown to promote generation of CD8 ϩ T cell memory (25). However, it remains unclear how effectively IL-21 may promote tumor-specific T cell activity generated in the context of an Ab-or vaccine-based immunotherapy.
In this study, we have examined the ability of IL-21 to suppress tumor metastases and growth when used sequentially in combination with an anti-DR5 mAb directed at TRAIL-sensitive tumors. We have demonstrated in a series of in vivo models that IL-21 promoted tumor-specific CTL activity and enhanced memory responses to tumor rechallenge. This study illustrates the principle that IL-21 may be used to support the anti-tumor activity of many clinically useful mAbs in cancer treatment.

Materials and Methods
Mice accordance with guidelines set out by the Peter Mac animal experimental ethics committee.

4T1 mammary tumor growth
Groups of five BALB/c WT mice were inoculated in the mammary fat pad with various doses of 4T1 tumor cells on day 0. Groups of mice were then treated i.p. with the following: control Ig (cIg) and PBS, anti-DR5 (MD5-1) and PBS, cIg and IL-21, or a combination of MD5-1 and IL-21 at the doses and times indicated. Tumor size was measured every second day with a caliper as the product of two perpendicular diameters (cm 2 ). Some groups of mice were injected in the mammary fat pad with 4T1 tumor cells, and 28 days later, the primary developing tumor was resected and groups of mice were then treated i.p. with cIg and PBS, anti-DR5 (MD5-1) and PBS, cIg and IL-21, or a combination of MD5-1 and IL-21 at the doses and times indicated (34,36). Survival of the mice was monitored for 150 days. Spleen cells from MD5-1-and MD5-1/IL-21-treated mice that were tumor free after 6 wk were then adoptively transferred (i.v. at the doses indicated) into groups of five BALB/c SCID mice that had received 5 ϫ 10 3 4T1 tumor cells in the mammary gland 24 h earlier. SCID mice injected with 4T1 cells but not receiving a spleen T cell transfer were used as controls. Mammary tumor growth was measured every second day, and tumor sizes represent the mean Ϯ SE of five mice in each group.

Therapy of R331 tumor growth
Groups of five BALB/c WT mice were inoculated s.c. with 1 ϫ 10 4 R331 tumor cells on day 0. Groups of mice were then treated i.p. with cIg on days 0, 3, and 6 and PBS on days 9 -11; anti-DR5 (MD5-) on days 0, 3, and 6 and PBS on days 9 -11; cIg (as above) and IL-21 (IL-21) on days 9 -11; or combinations of MD5-1 and IL-21 at similar times as indicated. In some experiments, commencement of the treatment with MD5-1/IL-21 was delayed until days 12 or 18 after tumor inoculation. Mice treated with either MD5-1 and PBS or MD5-1 and IL-21 that remained free of their primary R331 tumor were then rechallenged s.c. in the opposite flank 12 wk after the primary tumor inoculation with increasing doses of R331 tumor cells as shown. Growth after rechallenge was compared with growth at the same doses in naive BALB/c WT mice. In primary and secondary challenge experiments, R331 tumor growth was measured every second day, and tumor sizes represent the mean Ϯ SE of five mice in each group. Tumor Enhanced suppression of s.c. 4T1 tumors by early anti-DR5 mAb and IL-21 therapy. Groups of five BALB/c WT mice were inoculated in the mammary fat pad with 5 ϫ 10 4 4T1 tumor cells on day 0. A, Groups of mice were then treated i.p. with the following: 100 g of cIg on days 0, 3, and 6 and PBS on days 9 -11; 100 g of anti-DR5 (MD5-1) on days 0, 3, and 6 and PBS on days 9 -11; cIg (as above) and 20 g of mouse IL-21 (IL-21) on days 3-5; cIg (as above) and 20 g mouse IL-21 (IL-21) on days 6 -8; cIg (as above) and 20 g of mouse IL-21 (IL-21) on days 9 -11; or combinations of MD5-1 and IL-21 at similar times as indicated. In another experiment, groups of mice received cIg on days 0, 3, and 6 and PBS or IL-21 on days 9 -11 (B); MD5-1 on days 0, 3, and 6 and PBS on days 9 -11 (C); or MD5-1 on days 0, 3, and 6 and IL-21 on days 9 -11 (D), alone or in combination with 100 g mAbs to deplete CD4 ϩ T, CD8 ϩ T, NK cells (anti-asGM1), or neutralize CD11b (5C6) as indicated on days Ϫ1, 0, 7, and 14. Mammary tumor growth was measured every second day, and tumor sizes represent the mean Ϯ SE of five mice in each group. Similar results for A were obtained in two independent experiments. size was measured periodically with a caliper as the product of two perpendicular diameters (cm 2 ).

Therapy of R331 tumor metastases
Groups of BALB/c WT mice were inoculated i.v. with the indicated dose of R331, Renca, or R331-FLIP tumor cells on day 0. Groups of mice then received i.p. cIg on days 0, 3, and 6 and PBS on days 9 -11; anti-DR5 (MD5-1) on days 0, 3, and 6 and PBS on days 9 -11; cIg and IL-21 (IL-21) on days 9 -11; or combination MD5-1 and IL-21 as indicated. All groups of mice were sacrificed at day 14, and lung tumor metastases counted with the aid of a dissecting microscope. Data are recorded as the mean Ϯ SE of each group.

Cytotoxicity assay
Cytotoxic activity mediated by MD5-1 was tested by a 4-h 51 Cr release assay as described previously (12,37). Susceptibility of tumor cells to TRAIL-mediated cytotoxicity was examined using mouse TRAIL-transfected 2PK-3 (2PK-3-mTRAIL) or mock-transfected 2PK-3 (2PK-3-m) as the effector cells. Susceptibility to MD5-1 cytotoxicity was tested in the presence of FcR-expressing P815 cells or spleen NK cells prepared from PBS or IL-21-treated B6 RAG-1 Ϫ/Ϫ mice. Mice received PBS or IL-21 (20 g i.p.) on days Ϫ3, Ϫ2, and Ϫ1 before spleen harvest. In some experiments, the cytotoxicity assay was performed in the presence of 5 g/ml MD5-1 or cIg. CTLs reactive with tumor were induced as described previously (12). Briefly, splenocytes were prepared from either naive or tumor-bearing mice treated with a combination of anti-DR5 or cIg and IL-21 or PBS. Following treatment and on day 28 of tumor growth, splenocytes were harvested from three mice of each group, pooled, and 20 million splenocytes were cocultured with 2 ϫ 10 6 of mitomycin C (MMC; Kyowa Hakko)-treated (200 g/ml, 2 h) 4T1 cells for 7 days. Seven days later, the cytotoxic activities of these cultures were then tested against either 4T1 or R331 tumor targets at the E:T ratio shown in a 4-h 51 Cr release assay as described (34). Results were expressed as the mean Ϯ SE of triplicate samples.

Statistical analysis
Statistical analysis was performed by a Mann-Whitney U test for the tumor metastases and growth data. Values of p Ͻ0.05 were considered as significant.

Synergistic anti-tumor efficacy of combined anti-DR5 mAb/IL-21 therapy
We have demonstrated previously that an agonistic anti-DR5 mAb exhibits potent anti-tumor effects against TRAIL-sensitive tumors through induction of tumor cell apoptosis, recruiting Fc receptorbearing APCs, and generating tumor-specific CTL (12). In addition, IL-21 has been reported to induce CTL-mediated anti-tumor effects by costimulating T cells (25). Given the points of action of these agents, we expected that the combination of anti-DR5 and IL-21 might have enhanced efficacy against primary tumors and metastases. Initially, various early treatment schedules of anti-DR5 mAb (MD5-1) and IL-21 alone or in combination were compared for their anti-tumor efficacy against orthotopic growth of the experimental 4T1 mammary tumor (Fig. 1A). If treatment was commenced on day 0, anti-DR5 mAb alone substantially inhibited the growth of tumors, but never caused tumor rejection (Fig. 1A). IL-21 alone was ineffective unless treatment was administered 3-5 days after tumor inoculation, and even at that early time, tumor growth was inhibited only to a minor extent. However, strikingly, combined anti-DR5 and IL-21 treatment completely suppressed tumor growth in a significant proportion of mice, and tumors did not recur Ͼ120 days after the treatment ceased (Fig. 1A). Importantly, combined treatment was most effective when IL-21 treatment was commenced 3 days (days 9 -11) after the third and final MD5-1 treatment. A dose response was performed with IL-21 from 1 to 50 g per injection, and 20 g per injection was found to be optimal in this experimental setting (data not shown). Earlier administration of IL-21 in combination with MD5-1 was less effective, suggesting that IL-21 was simply not significantly enhancing Ab-dependent cellular cytotoxicity (ADCC) triggered by MD5-1. Indeed, MD5-1 has been reported to mediate apoptosis of tumor cells strictly via DR5 and a caspase-8-dependent pathway rather than via classical ADCC (12). When we examined the cytotoxic activity of MD5-1 against tumor targets in vitro, in the presence of effector splenocytes or P815-expressing Fc receptor, IL-21 treatment did not substantially enhance ADCC by the MD5-1 mAb (data not shown). Thus, the improved in vivo efficacy with delayed IL-21 treatment was consistent with a similar improved effect when using ␣-GalCer and delayed IL-21 in combination (29) and suggested that IL-21 was promoting adaptive tumor immunity.
We next examined which leukocyte subsets were responsible for the primary anti-tumor effect of combined MD5-1 and IL-21 therapy by depleting NK cells, CD4 ϩ T cells, or CD8 ϩ T cells, or neutralizing CD11b function. As shown previously, cIg and IL-21 were ineffective in suppressing growth of 4T1 in the mammary gland, and depletion of NK cells, CD4 ϩ T cells, or CD8 ϩ T cells or neutralization of CD11b did not suppress or enhance 4T1 tumor growth (Fig. 1B). We have reported previously that suppression of tumor growth by MD5-1 was mediated in part by FcR-expressing NK cells and CD11b ϩ cells (12), and our finding in the orthotopic 4T1 tumor model was in concert with this report (Fig. 1C). In vitro studies have shown previously that both NK cells and macrophages enable DR5-triggered apoptosis of tumor cells after FcRmediated cross-linking of MD5-1 (12). Anti-CD11b mAb (5C6) inhibits both macrophage and neutrophil recruitment to inflammatory sites. Strikingly, the combination MD5-1 and IL-21 therapy critically required CD11b function and, to a lesser extent, NK cell and CD8 ϩ T cell function (Fig. 1D). The dependence on CD8 ϩ T cell function was supported by the relative ineffectiveness of MD5-1 and IL-21 combination against larger tumors in SCID, compared with WT mice (data not shown). CD4 ϩ T cell function was not required for combined MD5-1 and IL-21 anti-tumor efficacy.
To further evaluate the efficacy of the anti-DR5 and IL-21 combination, we assessed suppression of tumor metastases in the 4T1 tumor model when the primary mammary tumor was resected when it reached ϳ8 mm in diameter on day 28 (Fig. 2). Typically, at day 28, the mice have a high level of 4T1 metastases in their lung and metastases at additional sites such as liver, lymph node, brain, and bone. All mice eventually succumbed to tumor; however, groups of mice receiving a combination of MD5-1 and delayed IL-21 survived significantly longer than those receiving MD5-1 or IL-21 alone or MD5-1 and IL-21 concurrently. Future experiments will be designed to include additional cycles of treatment with the aim of completely suppressing metastatic disease.

Combination MD5-1/IL-21 suppresses TRAIL-sensitive renal carcinoma metastases
We next examined the efficacy of combined MD5-1/IL-21 treatment against experimental renal carcinoma metastases. R331 is an extremely TRAIL-sensitive variant of the Renca renal carcinoma (35). Despite the fact that IL-21 treatment commenced as late as day 9, once again the combination was more effective than either MD5-1 or IL-21 alone against TRAIL-sensitive Renca and R331 renal carcinoma cells (Fig. 3, A and B). By contrast, the combination was ineffective against TRAIL-resistant R331-FLIP tumor cells (Fig. 3C). The mechanism of metastases suppression by MD5-1/IL-21 was once again assessed by depleting various subsets of lymphocytes and neutralizing CD11b (Fig. 3D). Consistent with the 4T1 tumor model, the combination MD5-1 and IL-21 therapy critically required CD11b function and, to a lesser extent, NK cell and CD8 ϩ T cell function to suppress R331 lung metastases (Fig. 3D).

Combination MD5-1/IL-21 enhances memory T cell response to secondary tumor challenge
We next examined the ability of combined MD5-1/IL-21 treatment to suppress low doses of s.c. R331 renal carcinoma cells. At this low dose of R331, both groups of WT mice receiving MD5-1 or combination MD5-1/IL-21 remained free of their primary R331 tumor (Fig. 4A). By contrast, all mice receiving control Ig and/or IL-21 developed tumors. Tumor-free mice were then rechallenged s.c. in the opposite flank 12 wk after the primary tumor inoculation with increasing doses of R331 tumor cells (Fig. 4, B-D). Mice that had originally rejected R331 following MD5-1/IL-21 treatment were able to resist even the highest dose of secondary challenge with R331 tumor cells (Fig. 4D). By contrast, mice originally receiving MD5-1 alone were unable to reject secondary R331 challenge at doses of 10 5 cells (Fig. 4C) and above (Fig. 4D). Groups of MD5-1/IL-21-treated mice that rejected primary R331 tumor challenge were unable to resist a secondary 4T1 tumor challenge in . Groups of mice then received i.p. the following: 100 g of cIg on days 0, 3, and 6 and PBS on days 9 -11; 100 g of anti-DR5 (MD5-1) on days 0, 3, and 6 and PBS on days 9 -11; cIg (as above) and 20 g of mouse IL-21 (IL-21) on days 9 -11; or combination MD5-1 (as above) and 20 g of mouse IL-21 (IL-21) on days 9 -11 as indicated. D, Groups of mice receiving 5 ϫ 10 5 R331 tumor cells were treated with: cIg on days 0, 3, and 6 and PBS on days 9 -11; MD5-1 on days 0, 3, and 6 and PBS on days 9 -11; or MD5-1 on days 0, 3, and 6 and IL-21 on days 9 -11, alone or in combination with 100 g of mAbs to deplete CD4 ϩ T, CD8 ϩ T, NK cells (anti-asGM1), or neutralized CD11b (5C6) as indicated on days Ϫ1, 0, and 7. All groups of mice were sacrificed at day 14, and lung tumor metastases were counted with the aid of a dissecting microscope. Data were recorded as the mean Ϯ SE of each group. A significant reduction in lung metastases in groups receiving the MD5-1/IL-21 combination, compared with mice receiving MD5-1 alone are shown ‫,ء(‬ p Ͻ 0.05). Similar results for A-C were obtained in two independent experiments. the mammary gland, demonstrating the specificity of the memory response (data not shown). Additional data (Fig. 4E) illustrated that primary rejection of R331 conferred protective immunity against both R331 (TRAIL-sensitive) and R331-FLIP (TRAIL-resistant) tumor cells.
In a similar fashion, tumor-specific immunity could be transferred by splenic T cells from MD5-1 or MD5-1/IL-21-treated mice into SCID mice that otherwise could not resist a lethal chal-lenge with 4T1 tumor cells. Both MD5-1 and MD5-1/IL-21 suppressed the outgrowth of primary 4T1 tumor inoculation in BALB/c WT mice, compared with cIg or IL-21 alone (Fig. 5A). Spleen T cells purified from MD5-1-and MD5-1/IL-21-treated mice that were tumor free after 6 wk were then adoptively transferred at two different doses (5 ϫ 10 6 , 5 ϫ 10 7 ) into groups of five BALB/c SCID mice that had received 5 ϫ 10 3 4T1 tumor cells in the mammary gland 24 h earlier. SCID mice injected with 4T1 cells, but not receiving a T cell transfer, were used as controls. Clearly SCID mice were well protected from lethal 4T1 tumor challenge when receiving even lower doses of T cells from MD5-1/IL-21-treated WT mice that had rejected primary 4T1 tumor challenge (Fig. 5B). T cells from MD5-1-treated mice also offered some protection at the higher T cell dose transferred. These data illustrated that memory for 4T1 tumor can be more effectively generated and transferred from MD5-1/IL-21-treated mice.

Combination MD5-1/IL-21 enhances tumor-specific CTL generation
We also examined the ability of mice that initially resisted a low dose of 4T1 mammary tumor after MD5-1/IL-21 treatment to secondarily reject 4T1 inoculated in the opposite mammary gland (Fig. 6). Early MD5-1/IL-21 treatment effectively suppressed the outgrowth of primary 4T1 tumor inoculation in all mice (Fig. 6A). These tumor-free mice were then re-challenged 7 wk after the primary tumor inoculation with 4T1 tumor cells. Groups of these mice were treated with cIg or 5C6 mAb or were depleted of CD4 ϩ T, CD8 ϩ T, CD4 ϩ and CD8 ϩ T, or NK cells. Growth after rechallenge was compared with growth at the same doses in naive cIg-treated BALB/c WT mice (Fig. 6B). Clearly, mice that had received MD5-1/IL-21 treatment and had resisted primary 4T1 tumor challenge were all resistant to a lethal secondary challenge with 4T1 tumor cells, compared with naive WT mice. By contrast, those mice additionally depleted of CD8 ϩ T cells at the time of secondary 4T1 tumor challenge were as susceptible as naive WT mice to 4T1 tumor growth (Fig. 6B). Groups of mice neutralized for CD11b or depleted of CD4 ϩ T cells or NK cells remained resistant to 4T1 tumor challenge. These data support the results observed in the R331 tumor model and illustrated that CD8 ϩ T cells were responsible for secondary responses to tumor. These responses also were specific for 4T1 as demonstrated by the inability of WT mice to resist secondary R331 tumor challenge after FIGURE 6. Memory response to secondary tumor challenge is mediated by CD8 ϩ T cells. BALB/c WT mice were inoculated in the mammary fat pad with 5 ϫ 10 3 4T1 tumor cells (day 0). A, Groups of mice were then treated i.p. with the following: 100 g of cIg on days 0, 3, and 6 and PBS on days 9, 10 and 11 (n ϭ 5); or 100 g of anti-DR5 (MD5-1) on days 0, 3, and 6 and 20 g of IL-21 on days 9 -11 as indicated (n ϭ 25). Groups of 25 WT mice treated with MD5-1 and IL-21 that remained tumor free of their primary 4T1 tumor were then rechallenged in the opposite fat pad 7 wk after the priming tumor inoculation with 5 ϫ 10 3 4T1 tumor cells. Some groups additionally received 100 g of mAbs to neutralize CD11b or deplete CD4 ϩ T, CD8 ϩ T, CD4 ϩ and CD8 ϩ T cells, and NK cells (anti-asGM1) as indicated on days Ϫ1, 0, 7, and 14. Growth after rechallenge was compared with growth at the same doses in naive cIg-treated BALB/c WT mice. In primary and secondary challenge experiments, 4T1 tumor growth was measured every second day, and tumor sizes represent the mean Ϯ SE of five mice in each group.

FIGURE 5.
Enhanced tumor-specific T cell function transferred from MD5-1/IL-21-treated mice. Groups of five BALB/c WT mice were inoculated in the mammary fat pad with 5 ϫ 10 3 4T1 tumor cells on day 0. A, Groups of mice were then treated i.p. with the following: 100 g of cIg on days 0, 3, and 6 and PBS on days 9 -11; 100 g of anti-DR5 (MD5-1) on days 0, 3, and 6 and PBS on days 9 -11; cIg (as above) and 20 g of mouse IL-21 (IL-21) on days 9 -11; or a combination of MD5-1 and IL-21 at similar times as indicated. B, Splenic T cells from MD5-1-and MD5-1/ IL-21-treated mice that were tumor free after 6 wk were then adoptively transferred (at the doses indicated) into groups of five BALB/c SCID mice that had received 5 ϫ 10 3 4T1 tumor cells in the mammary gland 24 h earlier. SCID mice injected with 4T1 cells, but not receiving a T cell transfer, were used as controls. Mammary tumor growth was measured every second day, and tumor sizes represent the mean Ϯ SE of five mice in each group. Similar results were obtained in two independent experiments for A.

MD5-1/IL-21 treatment and elimination of primary 4T1 tumor inoculation (data not shown).
Given the ability of MD5-1/IL-21-treated mice to resist secondary tumor challenge in a CD8 ϩ T cell-dependent manner, we next assessed whether this treatment protocol enhanced tumor-specific CTL generation in tumor-bearing mice. Splenocytes from treated naive and 4T1 tumor-bearing mice were harvested and cocultured with 4T1 tumor cells in vitro. Seven days later, the cytotoxic activities of these cultures were then tested against either 4T1 or R331 tumor targets at various E:T ratios (Fig. 7). Strikingly, only splenocytes from 4T1 tumor-bearing mice treated with either MD5-1 or MD5-1/IL-21 displayed significant levels of cytotoxicity against the 4T1 tumor targets (Fig. 7, A and B). Those splenocytes from MD5-1/IL-21-treated mice displayed Ͼ10 times the cytotoxic capacity of splenocytes from MD5-1-treated mice (Fig.  7B). By contrast, all these same effector cells did not kill R331 tumor targets, illustrating the specificity of the CTL generated in response to the priming tumor (Fig. 7D).

Combination MD5-1/IL-21 suppresses established experimental tumor burden
To examine the efficacy of combined MD5-1/IL-21 treatment against more established tumors, the commencement of treatment was delayed until 12 (2-3 mm in diameter) or 18 (4 -5 mm in diameter) days after tumor inoculation. Neither IL-21 nor MD5-1 therapy alone had any discernable effect on R331 tumor growth (Fig. 8). By contrast, the combination of MD5-1 and IL-21 reduced s.c. R331 tumor growth by Ͼ50% when commenced from day 12. These data in this instance demonstrate the synergistic activity of anti-DR5 and IL-21 in an advanced tumor model. In this setting, the mechanism of tumor suppression likely involves CTL generation, but it may additionally involve effects on tumor angiogenesis, suppressor macrophages, or suppressor T cell populations, and these possibilities remain to be investigated. Groups of these mice were then treated i.p. with the following: 100 g of cIg on days 0, 3, and 6 and PBS on days 9 -11; 100 g of anti-DR5 (MD5-1) on days 0, 3, and 6 and PBS on days 9, 10, and 11; cIg (as above) and 20 g of mouse IL-21 (IL-21) on days 9 -11; or combinations of MD5-1 and IL-21 at similar times as indicated. On day 28, splenocytes harvested from three mice from each group were pooled and cocultured with the MMC-treated 4T1 tumor cells in vitro. Seven days later, the cytotoxic activities of these cultures were then tested against either 4T1 (A and B) or R331 (C and D) tumor targets at the E:T ratio shown. Results represent the mean Ϯ SE of triplicate samples.

Discussion
In this study, we have demonstrated that induction of tumor cell apoptosis by agonistic mAb against DR5, combined with delayed IL-21 treatment, suppressed tumor growth and pre-established tumor metastases by enhancing tumor-specific CTL induction. Synergistic effects of the combination were observed in several tumor models where the target tumor was sensitive to DR5-mediated apoptosis. Although combined therapy could not eradicate larger preestablished tumors, IL-21 both promoted tumor-specific CTL activity and enhanced memory responses to tumor rechallenge when smaller tumor burdens were resisted on primary challenge. This combined therapy did not cause any apparent toxicity or autoimmunity (data not shown). IL-21 was most effective when given 3 days after a consecutive course of MD5-1 treatment. These results indicated that a rational combination Ab-based therapy that directly causes tumor cell apoptosis and a cytokine that promotes CD8 ϩ T cell function and memory may be a useful new strategy for cancer immunotherapy.
Tumor Ags are readily supplied by apoptotic tumor cells, and induction of tumor cell apoptosis has been hypothesized as an attractive initial step in cancer immunotherapy, when followed by appropriate T cell activation. Previously, we have provided ample in vitro evidence that anti-DR5 mAb induces tumor cell apoptosis, whereas in vivo, this apoptosis of TRAIL-sensitive tumor cells is inferred by the resistance of FLIP-expressing variants. Perhaps more importantly, the anti-DR5 mAb triggers the secondary recruitment of APCs, such as macrophages and DC, via activating Fc receptors. Subsequent uptake of tumor Ag and cross-presentation evokes tumor-specific CTLs that can then eliminate both TRAILsensitive and TRAIL-resistant variants (12). However, induction of this pathway is relatively weak, and only small numbers of TRAIL-sensitive tumor cells can be resisted by MD5-1 alone (data herein and Ref. 12). IL-21 has been shown to regulate T cell proliferation and differentiation and thereby facilitate distinct steps in the generation of an anti-tumor immune response (25,26,28). It appears to be one of the most promising cytokines for promoting T and B cell responses to Ag. The anti-tumor effects of IL-21 are mediated by activation of NK cells as well as eliciting tumorspecific memory (25) (P. Sivakumar, unpublished observations). In this study, we have reiterated that IL-21 can promote CTL generation and enhance tumor-specific memory responses in mice in a setting where apoptotic cells provide the initiating signal for the CD8 ϩ T cell response. CD8 ϩ T cells appeared critical for this memory response to tumor challenge. Additional experiments have been performed in various gene-targeted mice, and we have shown that both perforin and IFN-␥ are essential for secondary responses (data not shown).
Agonistic anti-DR5 mAb can trigger tumor cell death directly via caspase activation (12), but despite different pathways of cell death induction, primary tumor cell apoptosis and secondary induction of tumor-specific T cell immunity have been implicated in the successful anti-tumor effects of other anti-tumor mAbs (12)(13)(14). Indeed, tumor cells may acquire TRAIL resistance, and thus other modes of triggering tumor cell apoptosis may be preferred. It remains to be determined whether IL-21 may enhance the capacity of other tumor-specific Abs to generate tumor-specific CTL. In some cases, IL-21 additionally improves primary tumor suppression by ADCC mediated via NK cells when mAbs such as Herceptin or Rituxan are used (38 -40). Cytokines, like IL-21, also may augment the therapeutic effects of other modalities that induce tumor cell apoptosis, including chemotherapy and radiotherapy (15,16,41).
In summary, it is becoming clear that IL-21 is capable of enhancing CD8 ϩ T cell expansion and function and is necessary for an optimal CD8 ϩ T cell response to Ag. Many immunotherapies that induce tumor-specific CTL may be enhanced further by combination with IL-21. Pending safety trials of IL-21 in patients with advanced cancer, it should be possible to combine this cytokine with many new CTL-based vaccines.