It has been shown previously that the suppression of tumor immunosurveillance may be a mechanism by which tumors resist immune detection and elimination. In this study, we evaluated the role of the immunoregulatory natural killer T (NKT) cells in the biology of immunosurveillance of osteosarcoma. The K7M2 mouse osteosarcoma cell line was implanted orthotopically into wild-type and NKT cell–deficient CD1d knockout (KO) BALB/c mice, and mice were monitored for growth of primary tumors. Further, we examined the role of CD4+ and/or CD8+ cells by depleting the cells in vivo and measuring CTL activity in vitro. We also asked the role of interleukin (IL)-4 receptor α (IL-4Rα)-signal transducer and activator of transcription 6 (STAT6) signaling, including IL-13, and transforming growth factor β (TGF-β) by using gene-disrupted mice or treating mice with cytokine antagonists. We were surprised to find a high rate of rejection of osteosarcoma primary tumors in 88% (14 of 16) of CD1d KO mice compared with syngeneic wild-type BALB/c mice that showed rejection of tumor in <24% of mice. Further studies suggested that the rejection of tumor in CD1d KO mice was dependent on CD8+ lymphocytes. Distinct from other murine tumor models, the negative regulation induced by CD1d-restricted NKT cells was not dependent on IL-4Rα-STAT6 signaling, including IL-13, or on TGF-β. These data suggest that a novel CD1d-restricted NKT cell–mediated mechanism for tumor immunosuppression is active in the K7M2 osteosarcoma model and that NKT cells can regulate immunosurveillance through more than one pathway. (Cancer Res 2006; 66(7): 3869-75)
- NKT cell
- tumor resistance to immune response
Osteosarcoma is a highly metastatic primary bone cancer most commonly seen in pediatric and adolescent patients. In spite of effective control of the primary tumor and multimodality adjuvant therapy, death from metastasis, most commonly to the lung, continues to be a problem for >30% of patients diagnosed with localized disease ( 1). To improve patient outcomes, an improved understanding of the biology of metastasis in osteosarcoma is needed. Anecdotal experience and recent preclinical and clinical studies support the role of the immune system in the surveillance and elimination of metastatic cancer cells; however, the mechanisms guiding this immune response are poorly understood ( 2– 5).
There is accumulating evidence that tumors use mechanisms that suppress the immune system to evade host immunity ( 6– 8). These mechanisms include host immune components CD4+CD25+ regulatory T cells, myeloid suppressor cells, and natural killer T (NKT) cells ( 6– 10). We have recently described a novel immunoregulatory pathway operative in both regressor, in which natural immunosurveillance is apparent, and nonregressor, in which no natural immunosurveillance is apparent, mouse tumor models ( 11– 14). In this pathway, interleukin (IL)-13 produced by CD4+ CD1d-restricted NKT cells induces transforming growth factor β (TGF-β) production by CD11b+Gr-1+ myeloid cells. We showed that blocking the immunoregulatory pathway by inhibiting these cytokines or depleting CD4+ CD1d-restricted NKT cells or myeloid cells unmasked natural immunosurveillance that can lead to tumor rejection. It is also reported that, in the syngeneic breast cancer model 4T1, although NKT cell–deficient CD1d knockout (KO) mice are highly resistant to lung metastasis, IL-13 or its receptor component IL-4 receptor α chain (IL-4Rα) seems not to be necessary for the suppression of antitumor immunity against lung metastasis ( 15). Therefore, it is important to understand whether CD1d-restricted NKT cells use different mechanisms to suppress antitumor immunity in different tumor models.
In this study, we took advantage of a recently described syngeneic murine model of osteosarcoma characterized by appendicular tumor growth at orthotopic sites, a period of minimal residual/micrometastatic disease accomplished by primary tumor resection, spontaneous metastases to distant sites, and, most importantly, host animals that have competent immune systems ( 16– 19) to examine the role of NKT cells that are CD1d restricted in the regulation of antitumor immunity. Application of this transplantable model system to NKT cell–deficient CD1d gene KO mice resulted in complete rejection of primary osteosarcoma tumors. This rejection was CD8 dependent and associated with enhanced CTL lytic activity. Distinct from other murine tumor models, the negative regulation induced by CD1d-restricted NKT cells in this case was not dependent on IL-4Rα-signal transducer and activator of transcription 6 (STAT6) signaling, including IL-13, or on TGF-β. We therefore propose that a novel CD1d-restricted NKT cell–mediated mechanism for tumor immunosuppression is active in the K7M2 osteosarcoma model and that NKT cells can regulate immunosurveillance through more than one distinct pathway.
Materials and Methods
Reagents. Purified rat anti-mouse CD4 monoclonal antibody (mAb; GK1.5) and anti-CD8 (2.43), as well as monoclonal anti-IL-4 (11B11), were obtained from the Frederick Cancer Research and Development Center, National Cancer Institute (NCI), NIH (Frederick, MD). A fusion protein of murine IL-13Rα2 and human IgG1 (sIL-13Rα2-Fc) was kindly provided by Wyeth-Genetics Institute (Cambridge, MA) ( 20). Monoclonal anti-TGF-β (1D11.16 specific for TGF-β1, TGF-β2, and TGF-β3; ref. 21) and isotype-matched control antibody (13C4; ref 22) were made and kindly provided by Genzyme Corp. (Cambridge, MA). FITC-labeled anti-CD1d mAb and isotype-matched control mAb were purchased from BD Biosciences (San Diego, CA).
K7M2 osteosarcoma cell line. Derivation, characterization, and maintenance of highly metastatic (K7M2) murine BALB/c osteosarcoma cell lines have been described previously ( 17). K7M2 cells were maintained in vitro using complete culture medium [DMEM (Celox Co., Hopkins, MN), 100 μg/mL penicillin-streptomycin, and 2 mmol/L l-glutamine (Sigma, St. Louis, MO)] with 10% FCS (Sigma) at 37°C in 5% CO2. All cell lines used for in vitro and in vivo studies were from the 3rd to the 15th passages. For all in vitro and in vivo assays, cells were harvested using trypsin/Versene from near confluent cultures. Cell viability was assessed using trypan blue, and experiments were not continued if cell viability was <90%. 18Neo is a BALB/c 3T3 cells transfected with a null vector expressing only the Neor gene as a control. Cells were maintained in complete T-cell medium, which consisted of RPMI 1640 with 10% FCS, l-glutamine, sodium pyruvate, nonessential amino acids, penicillin, streptomycin, and 5 × 10−5 mol/L 2-ME, containing 200 μg/mL geneticin (Sigma).
Mice. BALB/c mice were purchased from Animal Production Colonies, Frederick Cancer Research Facility, NCI, NIH. CD1d (N8), IL-4Rα, and STAT6-deficient BALB/c mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Animal care was in accordance with the guidelines of the NIH Animal Research Advisory Committee.
In vivo orthotopic murine osteosarcoma model. The characterization and use of the K7M2 murine osteosarcoma model have been described previously ( 17). Briefly, 4- to 5-week-old female BALB/c mice (Charles River Laboratories, Wilmington, MA) were housed under pathogen-free conditions with a 12-hour light/12-hour dark schedule, fed autoclaved standard chow and water ad libitum, and received 1 million K7M2 tumor cells in 100 μL phenol-free HBSS in the left gastrocnemius muscle group. When primary tumor growth in any single dimension exceeded 1 cm, mice were sacrificed, and tumor tissue was harvested in 3-mm fragments. Tumor fragments were frozen in 10% DMSO or surgically implanted immediately to naive BALB/c mice. Surgical sites were prepared by shaving the skin and then cleansing using Betadine scrub solution (E-Z Prep, Becton Dickinson, Franklin Lakes, NJ) and 70% sterile alcohol. Anesthesia was induced with ketamine (0.45 mg/mouse; Ketaset, Fort Dodge Laboratories, Inc., Fort Dodge, IA) and xylazine (0.45 mg/mouse; Sigma) given by i.p. injection. Anesthesia was then maintained using methoxyflurane (Mallinckrodt Veterinary, Inc., Mundelein, IL). Alternatively, mice were induced and maintained using isoflurane (Forane) inhalation. Postoperative care included i.p. injection of 0.5 mL sterile saline and i.p. buprenorphine hydrochloride (0.04 mg/mouse; Buprenex, Reckitt and Benkiser, Inc., Richmond, VA) as needed. Animal care and use was in accordance with guidelines of the NIH Animal Care and Use Committee. Paraosteal tibial muscle flap implantation was initiated with a transverse incision in the skin overlying the middle patella ligament. The origin of the cranial tibial muscle was exposed. Three millimeters from the muscle origin, a transverse transection of the proximal belly of this muscle was made. Tumor fragments were then surgically placed into the muscle flap. The transected cranial tibial muscle belly was closed using sterile surgical glue (Nexaband, Veterinary Products Laboratory, Phoenix, AZ). The overlying skin was then closed with surgical wound clips (Autoclip 9 mm, Becton Dickinson and Co., Sparks, MD).
Mice were monitored at least thrice weekly for tumor size and evidence of morbidity related to the primary tumor or pulmonary metastases. Tumor size was quantitated in two dimensions using calipers (Vernier Type 6914, Bel-Art Products, Pequannock, NJ). Tumor volume was calculated as follows: Tumor volume (mm3) = π / 6 × D × d2, where D is the largest cross-sectional diameter (mm) of the tumor and d is the cross-sectional diameter (mm) at right angles to D.
Where indicated, mice were treated with 0.2 mg sIL-13Rα2-Fc i.p. every other day for 2 weeks or 0.1 mg of either anti-TGF-β mAb i.p. or 0.1 mg control mAb. It has been reported that this amount of anti-TGF-β mAb maintains a high level of the antibody in the circulation in naive mice ( 23). For in vivo depletion of CD4+ and/or CD8+ cells, the mice were injected i.p. with 0.5 mg anti-CD4 and/or anti-CD8 mAb for 3 consecutive days from the day of tumor challenge and then once weekly. To neutralize IL-4 in vivo, 100 μg anti-IL-4 (11B11) was injected i.p. everyday for 2 weeks and thrice weekly thereafter until the end of the experiment.
Skin graft. Skin grafting from CD1d-intact BALB/c mice to CD1d KO mice was done as described previously ( 24).
Expression of CD1d on K7M2 cell line. The in vitro–cultured K7M2 osteosarcoma cell line was stained with FITC-labeled anti-CD1d mAb (BD Biosciences) and measured by FACSCalibur (BD Biosciences).
Mixed lymphocyte reaction. Spleen cells of CD1d KO mice and CD1d-intact BALB/c mice were suspended in RPMI 1640 containing 10% FCS supplemented with l-glutamine, sodium pyruvate, nonessential amino acids, penicillin, streptomycin, and 5 × 10−5 mol/L 2-ME (complete T-cell medium). Cells (4 × 105) from CD1d KO mice and the same number of irradiated BALB/c spleen cells (3,000 rad) were mixed in a 96-well plate and cultured for 4 days. [3H]thymidine (10 μCi/mL) was added during last 16 hours, and cells were harvested. The thymidine incorporation was measured by β-scintillation counter (Perkin-Elmer Life and Analytical Sciences, Shelton, CT).
Histology/immunohistochemistry. For assessment of early leukocyte infiltration into the implanted tumor fragments, tumor fragments were implanted as described above. Seven days later, the implanted tumor fragment was resected en bloc and frozen in OCT tissue prep medium (HistoPrep, Fisher Scientific, Pittsburgh, PA). Tissues were stained using H&E stain (American HistoLabs, Rockville, MD). Immunohistochemical detection for CD3, CD4, CD8, and MIB101 was accomplished using Vector Elite ABC (Vector Laboratories, Burlingame, CA) with hematoxylin counterstaining (Biogenex, San Ramon, CA).
CTL assay. Spleen cells from tumor-bearing BALB/c or CD1 KO mice were stimulated with mitomycin C (200 μg/mL)–treated K7M2 in complete T-cell medium supplemented with 10% T-stim (BD Biosciences). After 7 days of culture, viable cells were harvested and used as effector cells for the CTL assay. Cytotoxic activity of CD8+ T cells against target cells was measured by a 4-hour 51Cr release assay. The percentage of specific 51Cr release was calculated as follows: 100 × (experimental release − spontaneous release) / (maximum release − spontaneous release). Maximum release was determined from supernatants of cells that were lysed by addition of 5% Triton X-100. Spontaneous release was determined from target cells incubated without added effector cells.
NKT cell–deficient CD1d KO mice rejected primary tumors. We have shown previously that CD1d-restricted NKT cells are a key cell population responsible for down-regulating tumor immunosurveillance. To investigate a possible role of CD1d-restricted NKT cells in tumor immunity against a primary or metastatic nonregressor tumor, we implanted K7M2 murine osteosarcoma tumor fragments adjacent to the proximal tibias of CD1d KO mice that lack NKT cells. In wild-type mice, tumors became palpable on day 17 after the implantation. The tumor incidence was 76% (19 of 25 mice; Table 1 ). In contrast to wild-type BALB/c mice, most of the NKT cell–deficient CD1d KO mice rejected tumors. The tumor incidence was only 12% (2 of 16 mice; Table 1). Both tumors in CD1d KO mice grew more slowly than in wild-type mice ( Fig. 1 ). CD1d KO mice had a mean tumor volume of 0.28 ± 0.02 cm3 35 days following injection of tumor cells compared with 0.81 ± 0.24 cm3 in wild-type mice.
Cellular mechanism for CD1d KO rejection of primary tumors. To begin to define the mechanisms associated with primary tumor rejection in CD1d KO mice, histology and immunohistochemical staining of leukocyte subsets were undertaken 7 days following orthotopic injection of tumor cells to CD1d KO and wild-type BALB/c mice. There was significantly greater infiltration of lymphocytes in the tumor injection bed of CD1d KO mice compared with wild-type mice ( Fig. 2 ). Infiltrating lymphocytes included both CD4+ and CD8+ T cells in the CD1d KO mice. CD4 and CD8 T cells were scantly observed in tumors of wild-type mice. Monocyte infiltration of the primary tumor site was uncommon and similar in both wild-type and CD1 KO mice (data not shown). Primary tumor rejection was not due to the immune response against CD1d as a minor histocompatibility molecule because we could not detect expression of CD1d on the K7M2 cell line by fluorescence-activated cell sorting analysis ( Fig. 3 ). Furthermore, we did not observe any mixed lymphocyte reaction by CD1d KO spleen cells against wild-type BALB/c spleen cells, and BALB/c wild-type skin grafts transplanted onto CD1d KO mice had indefinite graft survival (data not shown). We therefore concluded that absence of a negative regulatory NKT cells relieved antitumor immunosuppression and thereby allowed spontaneous immunosurveillance to mediate rejection of the tumor. We expect that similar mechanisms of tumor rejection would be active against microscopic metastases in this model; however, because most CD1d KO mice did not develop primary tumors, we could not study the effect of NKT cell removal on spontaneous metastases per se.
Tumor rejection in CD1d KO mice is mediated by CD8+ CTL. To understand the mechanism by which CD1d KO mice rejected tumors, we depleted CD4+ and/or CD8+ cells in vivo before tumor implantation. When CD1d KO mice were depleted of CD8+ T cells (CD8 and CD4/CD8 depleted), all mice developed tumors similar to wild-type mice ( Table 2 ). However, the mice depleted of CD4+ T cells continued to reject tumors in the same way as intact CD1d KO mice. These results suggested that tumor rejection seen in CD1d KO mice was mediated by CD8+ CTL. To address whether tumor-specific CD8+ cell activity was increased in tumor-bearing CD1d KO mice, CTL activity against K7M2 was examined in spleen cells from wild-type and CD1d KO mice given K7M2 tumors. The cells from CD1d KO mice lysed the K7M2 tumors but not the unrelated BALB/c-derived fibrosarcoma tumor cell line 18Neo ( Fig. 4 ). However, the cells from wild-type tumor-bearing mice did not show any lytic activity against either target. Because no immunization was given, the CTLs were elicited by the growth of the tumor itself in CD1d KO mice but not in wild-type mice. Interestingly, the cells from those few CD1d KO mice that developed primary tumor did not show lytic activity, similar to the cells from wild-type mice. This correlation strengthened the conclusion that tumor rejection in CD1d KO mice was mediated by CD8+ CTLs. Taken together, these results lead us to conclude that CD1d KO mice that lack NKT cells are better able to raise a CTL response induced just by the growth of the tumor, without a vaccine, and that these CD8+ CTLs are necessary for the tumor rejection observed in CD1d KO mice.
Because we have reported previously in both fibrosarcoma and colon carcinoma transplantable models that CD1d-restricted NKT cells activate a negative immunoregulatory pathway, in which IL-4R-STAT6 signaling activated by IL-13 induces TGF-β production, we investigated the role of IL-13, IL-4R, STAT6, and TGF-β in the K7M2 tumor model. Unexpectedly, in the experiments using IL-4R KO, STAT6 KO, sIL-13Rα2-Fc fusion protein-treated, and anti-TGF-β antibody-treated mice, all the mice developed tumors with high incidence identical to the wild-type BALB/c mice ( Table 3 ). Thus, although NKT cells seem to suppress immunosurveillance against all the tumors studied, the mechanism through which this is achieved seems to be unique for this murine osteosarcoma.
Although we did not see significant difference in the tumor incidence in the mice deficient for IL-4Rα or TGF-β signaling compared with wild-type mice, there was a trend suggesting that those mice with immunologic defects had a higher tumor incidence. These results raise the possibility that these factors may contribute to the enhancement of tumor immunity instead of tumor inhibition. In fact, there are reports suggesting that IL-4 enhances tumor immunity mediated by CTLs ( 25, 26). To address this possibility, we examined the role of IL-4 and TGF-β in the protection observed in CD1d KO mice ( Table 4 ). The protection in CD1d KO mice was not diminished by treatment with either anti-IL-4 or anti-TGF-β. Therefore, the protection observed in CD1d KO mice is not mediated by IL-4 or TGF-β.
We originally conducted this study to investigate the role of CD1d-restricted NKT cells in the development of lung metastases. The syngeneic mouse osteosarcoma K7M2 model used herein has been useful in the study of mechanisms of metastasis and seemed particularly well suited to the study of the immunobiology of metastasis given that primary tumor growth preceded the development of lung metastases and that the rapidly growing and bulky primary tumor could be controlled surgically. Unexpectedly, CD1d KO mice rejected the growth of the aggressive primary tumors. This rejection was mediated by spontaneously induced CD8+ T cells that were not seen in wild-type mice. Because CD1d is a minor histocompatibility antigen, it might be a rejection antigen for mice that are not immunologically tolerant to CD1d. However, we could not detect CD1d on K7M2 tumor cells ( Fig. 3) and did not see any apparent immunologic response of CD1d KO mice against wild-type BALB/c cells (based on mixed lymphocyte response and skin graft rejection). Thus, the rejection was not due to the immunologic response against the CD1d antigen but due to the greater immune responsiveness of CD1d KO mice, resulting in the appearance of tumor-specific CTL.
The finding that tumor-specific CTL activity was observed only in spleen cells of CD1d KO mice but not in wild-type mice after the tumor challenge revealed spontaneous immunosurveillance that was suppressed or masked in wild-type mice. CD1d KO mice lack CD1d-restricted NKT cells. Because no immunization was given, the tumor itself was sufficient to induce specific CTL and able to reject the tumor in CD1d KO but not in wild-type mice. Previously, we have shown that NKT cell–deficient CD1d KO mice could reject or reduce tumor burden in multiple mouse tumor models, in which the effector mechanism of antitumor immunity was CD8+ CTL ( 12– 14). We also have shown that the greater CTL-mediated antitumor immunity in CD1d KO mice was due to lack of activation of the immunoregulatory circuit initiated by CD1d-restricted NKT cells. In this circuit, NKT cells produced IL-13 that activated myeloid cells to produce TGF-β, which directly suppressed CTL activation ( 14). Therefore, we expected to see activation of the same immunoregulatory circuit in the K7M2 osteosarcoma model. However, to our surprise, neither the IL-4R-STAT6 signal pathway, which is induced by IL-13 and/or IL-4, nor TGF-β was necessary for the suppression of tumor immunosurveillance in the osteosarcoma model. These results suggested that CD1d-restricted NKT cells use a mechanism that is different from the immunoregulatory circuit that we have observed in other tumor models to suppress CD8+ CTL in the osteosarcoma model.
It should be noted that lytic activity of CTLs in tumor-bearing mice showed a strong correlation with the outcome of tumor growth. Especially among the CD1d KO mice, only the ones that rejected the tumor showed specific lytic activity against K7M2 tumor targets but not the ones that failed to reject the tumor, although all of them were deficient for CD1d-restricted NKT cells. Thus, the correlation between CTL and protection holds even among CD1d KO mice. Whether other regulatory mechanisms played a role in preventing these two mice from developing CTL cannot be determined.
Immunoregulation of Th1-mediated immune responses and CTL induction by NKT cells has been described in multiple autoimmune diseases ( 27– 29). Although the precise mechanism of the regulation was not well defined yet, some findings have suggested regulation by cytokines, including IL-4, IL-10, and TGF-β ( 30, 31). Although we showed in this study that the immunoregulation by CD1d-restricted NKT cells in the osteosarcoma model is not mediated by IL-4, IL-13, or TGF-β, it is still possible that this is mediated by IL-10. We have not been able to obtain appropriate mice or antibodies to test the role of IL-10. Recently, Ho et al. reported that, in humans, a subset of activated CD1d-restricted NKT cells expressing CD8αα suppressed EBV-specific CD8+ T cells responding to a recall antigen by killing activated CD1d-expressing antigen-presenting cells ( 32). Although it has been believed that there is no CD1d-restricted NKT cell expressing CD8 in the mouse, a subset of CD1d-restricted NKT cell that has similar function might be involved in the immunoregulation of antitumor immunity in the osteosarcoma model.
The biology of TGF-β signaling may be unique in osteosarcoma compared with other tumor models. In addition to an immunosuppressive function, TGF-β is considered to be a major growth factor for osteosarcoma ( 33). Bone has been shown to produce and store a large amount of TGF-β, and recent reports have suggested that TGF-β plays a key role in the biology of osteosarcoma in vitro and in vivo ( 33– 36). We therefore hypothesized that blocking the autocrine loop of TGF-β in osteosarcoma may have facilitated tumor rejection both independent of and dependent on immunosurveillance. However, we could not see any effect of anti-TGF-β treatment on primary tumor growth. Thus, blocking TGF-β is not sufficient to affect the growth of the K7M2 murine primary osteosarcoma in vivo.
NKT cells are strong inducer of antitumor immunity when they are stimulated with the strong agonist α-galactosylceramide ( 37). Several studies indicated that NKT cells are beneficial for induction of antitumor immunity ( 38, 39). On the other hand, there are some studies showing the regulatory role of NKT cells in tumor immunity ( 12– 15, 40). Therefore, it is important to examine the role of NKT cells in different types of tumor models for better understanding of NKT cell function in cancer. Overall, the results shown in this study indicate that NKT cells play a pivotal role in more than one immunoregulatory pathway of tumor immunosurveillance. The NKT cell may be a key cell in several different negative regulatory pathways. Therefore, eliminating NKT cells may be a useful strategy for marshalling natural immunosurveillance to fight cancer.
Grant support: Intramural Research Program of the NIH, NCI, Center for Cancer Research.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Jim Hodge for critically reading this article and giving helpful suggestions; Dr. Terry Fry for technical assistance with skin grafting experiments; Drs. Scott Lonning, Steve Ledbetter, and Jan Pinkas (Genzyme) for providing us anti-TGF-β and control mAb for the experiment; and Dr. Debra Donaldson (Wyeth-Genetics Institute) for providing us sIL-13Rα2-Fc.
Note: M. Terabe and C. Khanna contributed equally to this work.
- Received September 22, 2005.
- Revision received January 20, 2006.
- Accepted February 3, 2006.
- ©2006 American Association for Cancer Research.