We have previously reported that a distinct subset of splenic CD4− rat dendritic cells (DC) induces a rapid and caspase-independent apoptosis-like cell death in a large number of tumor cells in vitro. The killing activity of these killer DC (KDC) was restricted to their immature state and was immediately followed by their engulfment of the apoptotic target cells, suggesting that these KDC could directly link innate and adaptive immunity to tumors. Here, we addressed this question using a transplantable model of rat osteosarcoma. First, we showed that rat KDC have an MHC II+CD103+CD11b+NKp46− phenotype and are therefore distinct from natural killer cells, which are MHC II−CD103−CD11b−NKp46+. KDC numbers could be specifically and strongly (up to 10-fold) enhanced by Flt3L in vivo. The OSRGa cell line derived from the osteosarcoma tumor was killed and phagocytosed in vitro by both normal and Flt3L-induced splenic KDC. Such tumor antigen–loaded KDC were used to s.c. vaccinate progressive tumor-bearing rats. Vaccination with OSRGa-loaded KDC but not KDC loaded with irrelevant tumor cells (Jurkat) delayed tumor progression or even induced tumor regression. This vaccine effect was not observed in CD8 T cell–depleted animals and protective against tumor rechallenge. These results suggest that KDC possess the intrinsic capability not only to kill and then engulf tumor cells but also to efficiently cross-present tumor cell–derived antigen in vivo and subsequently induce an adaptive antitumor immune response. [Cancer Res 2008;68(22):9433–40]
- dendritic cells
Dendritic cells (DC) are a rare population of hematopoietic cells present in all lymphoid organs as well as in numerous tissues and non lymphoid organs. The main function of DC is to present antigen to T lymphocytes. Unlike other antigen-presenting cells, DC are capable of inducing primary T-cell responses. As such, DC play an important role in antitumor immunotherapy ( 1), raising the possibility of novel therapeutic strategies for the treatment of tumor pathologies such as melanoma or prostate adenocarcinoma ( 2). Many studies have focused on the administration of autologous or allogeneic DC generated in vitro and pulsed with tumor antigens in various forms, including tumor peptide, tumor lysates ( 3, 4), apoptotic or necrotic tumor cells ( 5), exosomes ( 6), native RNA or RNA amplified from the tumor, stress proteins ( 7), or viral vectors ( 8). The fusion of osteosarcoma cells to allogeneic bone marrow–derived DC has also been shown to induce the rejection of tumors in rats ( 9).
Recent observations indicate that in addition to their pivotal role in the induction of adaptive immune responses, DC or more precisely, defined subsets of DC, exhibit direct antitumor cytotoxic activity in vitro (reviewed in ref. 10). We have previously reported that a subset of rat DC (hereafter called killer DC, KDC) exhibiting a MHC II+ CD4− CD103high CD11b+ CD172α− phenotype, induces an apoptosis-like cell death in various tumor cells ( 11, 12). KDC-induced tumor cell death was independent of caspases as well as the classic effectors of apoptosis such as perforin-granzyme, tumor necrosis factor (TNF)-related apoptosis inducing ligand (TRAIL), FasL, and TNF-α. The killing mechanism of rat KDC therefore differs from those described for human DC, which in many cases involved TRAIL ( 10). Importantly, the killing of tumor cells by KDC was followed by a specific and rapid phagocytosis of the target cells ( 12). Nevertheless, the in vivo role of cytotoxic DC and this particular subset of KDC has not yet been elucidated. We hypothesized that KDC could use their cytolytic activity to capture tumor antigens for presentation to T cells, and this unusual function of DC could be used in antitumor immunotherapy to elicit a tumor-specific cytotoxic T-cell response. Rare populations of so-called KDC (natural KDC, NKDC, and IFN-producing KDC, IKDC) were recently described in mice ( 13– 15). These populations produce large quantities of IFN-γ upon encounter with tumor cells, probably through NKG2-D–mediated recognition of tumor antigens ( 15). However, the capacity of NKDC or IKDC to phagocyte material from dead tumor cells and to present tumor antigens in vitro or in vivo has not been shown, and recent data indicate that IKDC are in fact a subset of natural killer (NK) cells and are unrelated to DC ( 16– 18). Of note, IKDC expressed NKp46, which is indeed known as a NK lineage–specific marker ( 19).
Here, we showed that rat KDC do not express NKp46. A rat osteosarcoma model was used to address the capacity of KDC to induce a specific immune response to tumor in vivo. First, we show that Flt3L specifically expanded KDC in vivo. In vitro, normal or Flt3L-induced KDC killed and rapidly phagocytosed the OSRGa cell line derived from the osteosarcoma tumor. This direct antigenic capture pathway enabled the KDC to induce long-lasting CD8 T cell–dependent tumor regression upon vaccination of osteosarcoma-bearing rats.
Materials and Methods
Animals. Sprague-Dawley (SPD) rats obtained from the Centre d'Elevage Janvier and SPD OFA rats obtained from the Centre d'Elevage Charles River (IFFA-CREDO) were housed under pathogen-free conditions in the Experimental Therapeutic Unit (Nantes University Medical School) in accordance with the institutional guidelines of the French Ethical Committee and under the supervision of authorized investigators. Animals were used at ages 4 to 10 wk.
Reagents. Carboxyfluorescein diacetate succinimidyl ester (CFDA SE: CFSE) was purchased from Molecular Probes (Invitrogen). A Dead Cell removal kit was purchased from Miltenyi Biotec.
Adenovirus. An adenovirus encoding human Flt3L (AdFlt3L) and the corresponding control adenovirus (Adl324) were produced as previously described ( 20). The titer was 5.8 × 1011 infectious particles per milliliter. AdFlt3L and Adl324 were injected i.v. to SPD OFA rats.
Cell lines. The YAC-1 cell line was obtained from the European Collection of Cell Culture. The Jurkat cell line was kindly provided by Dr. I. Anegon (Institut National de la Santé et de la Recherche Médicale, Unité U643, Nantes, France). The OSRGa cell line was derived from the corresponding osteosarcoma model ( 21). Nonadherent cell lines were cultured in complete RPMI and adherent cell lines in complete DMEM.
Flow cytometry. The MHC II-allophycocyanine (APC)-Cy7, CD11b-PerCP-Cy5.5, CD4-phycoerythrine (PE)-Cy7, and CD45R-PE monoclonal antibody (mAb) were purchased from BD Biosciences. R7/3 (TCRαβ), OX12 (Igκ), W3/25 (CD4), and OX62 (CD103) hybridomas were obtained from the European Collection of Cell Culture, and mAb were purified from supernatants followed by coupling to AlexaFluor 488, FITC, and AlexaFluor 647, respectively, in our laboratory using kits from Molecular Probes. Anti-rat NKp46 (clone WEN23) mAb ( 22) was kindly provided by Erik Dissen (University of Oslo, Oslo, Norway). Cells were stained with mAbs in 96-well plates and were analyzed using a LSR II flow cytometer (BD Biosciences). Results were analyzed using the FlowJo software (Ashland).
Isolation of DCs. Spleens were chopped into small pieces and digested in 2 mg/mL collagenase D (Roche Diagnostique). Low-density cells were separated on a 14.5% Nycodenz gradient (Abcys) as previously described ( 12). CD4− CD103+ DC (KDC) purification was performed using magnetic beads after CD4+ cell depletion. Low-density cells were incubated with W3/25 mAb at 4°C for 10 min, washed, and then incubated with antimouse IgG-conjugated Dynabeads (Dynal; Invitrogen) for 20 min followed by magnetic depletion of positive cells (BD Biosciences). Cells were then washed, incubated with CD103-conjugated microbeads at 4°C for 15 min (Miltenyi Biotec), and positive selection was performed using an Automacs (Miltenyi Biotec). Purity was routinely >90%.
Cytotoxic assays. The cytotoxic activity of DC was assessed in a standard 5-h 51Cr-release assay as previously described ( 12). Specific release was calculated as (experimental release − spontaneous release)/(maximal release − spontaneous release) × 100. Results are expressed as mean ± SD of triplicate wells.
Phagocytosis. Jurkat cells were stained with CFSE for 2 min at room temperature, centrifuged through a Ficoll gradient to remove dead cells, and washed twice in PBS. OSRGa cells stably transfected with eGFP were harvested using Trypsin-EDTA. Jurkat cells and OSRGa cells were then cultured with CD4− DC for 5 h at an effector/target ratio of 25:1. Cells were then harvested and DC were stained with MHC II-APC-Cy7 mAb and analyzed by fluorescence-activated cell sorting (FACS). Alternatively, cells were sorted using a FACS ARIA (BD Biosciences), cytospun onto glass slides, and the cells were observed by fluorescence microscopy.
Cell sorting. For cell sorting, CD4− CD103+ DC were cultured for 5 h in the presence of OSRGa-GFP+ cells (ratio, 25:1), which were first depleted of dead cells and cell fragments using Annexin V–coated magnetic beads (Miltenyi Biotec). Nonadherent cells were then harvested and stained with a MHC-APC-Cy7 mAb. MHC II+ GFP+ cells were then sorted after exclusion of dead cells and doublets using a FACS Aria (BD Biosciences).
Osteosarcoma model. The osteosarcoma was initially induced in rats by a local injection of colloidal radioactive 144Cerium ( 21). The tumor can be regrafted as described below and maintained in vivo for many months, or fragments can be frozen until reuse. Using a right tibial approach, the periostum of the diaphysis was opened and resected along a length of 5 mm, leaving underlying bone intact. A 10-mm3 fragment of osteosarcoma was placed contiguous to the exposed bone surface without periostum, as previously described ( 23). The evolution of the tumor is comparable at the temporal (ratio, 1:100 between rats and humans) and physiologic levels to the development of human osteosarcoma. Lung metastases are observed in 75% to 90% of rats bearing advanced malignant bone tumors.
Vaccination of osteosarcoma-bearing rats with KDC. OFA rats were injected i.v. with 2.5 × 1010 infectious particles of AdFlt3L. Spleens of injected animals were harvested 8 d later and CD4− CD103+ DC were purified as described previously. DC were cultured overnight with tumor cells at a 25:1 ratio in the presence of 1.5 ng/mL of murine granulocyte macrophage colony-stimulating factor (GM-CSF) to enhance DC survival. Nonadherent cells were harvested the next day, and DC were separated from residual tumor cells by centrifugation on a Nycodenz gradient. Four to five millions DC were then injected into tumor-bearing rats 14 d after tumor implantation, a time corresponding to the development of a progressive tumor with a volume exceeding 1,200 mm3. Vaccinations were performed s.c. at different injection points and repeated weekly for 3 wk.
For the injection of KDC that had phagocyted OSRGa GFP+ cells, KDC were previously cultured for 5 h in the presence of OSRGa GFP+ cells (ratio, 25:1) first depleted of dead cells, with 1.5 ng/mL of murine GM-CSF. Cells were then stained with an MHC II-APC Cy7 mAb, and MHC II+ GFP+ cells were sorted with a 97.5% purity. Tumor-bearing rats were then vaccinated with 1.7 to 5 × 105 DC. Vaccinations were performed weekly for 3 wk as previously described.
In vivo CD8 T-cell depletion. Rat received i.p. injection of purified OX8 mAb (depleting anti-rat CD8α, mouse IgG1) twice a week at 3 mg/kg for 3 wk, starting 2 d before the first DC vaccination.
Statistical analyses. For in vivo experiments, the ANOVA (Tukey/Bonferroni) test was used to compare the tumor volume (quantitative data) between controls and vaccinated animals. The cumulative rate of overall survival was calculated according to the actuarial method, the end point considered being either the animal death or tumor volume superior to 40,000 mm3. The differences in actuarial survival were determined by the χ2 test.
KDC are not related to NK cells. Rat CD4− CD103+ DC (KDC) were isolated by magnetic selection of CD103+ cells after selection of spleen low density cells and depletion of CD4+ cells. CD103 is a classic and thus far the most specific marker for rat conventional DC ( 24) but is not expressed on plasmacytoid DC ( 25). CD103 has also been identified on some murine DC ( 26). The recent description of IKDC and then the demonstration that these cells were in fact NK cells ( 16– 18) prompted us to show that rat KDC were not related to NK cells. As shown in Fig. 1 , KDC were MHC II+ CD11b+ and NKp46−, whereas the rat NK cell line RNK16 ( Fig. 1A) as well as splenic NK cells (data not shown) were MHC II+ CD11b− NKp46+. As we previously showed, KDC were also homogenously CD103high ( 27), whereas NK cells did not express CD103. Although both KDC and NK cells could kill various target cells in vitro, the mechanism of killing of both subsets were different as NK cells use a Ca2+-dependent mechanism, whereas KDC used a Ca2+-independent mechanism ( Fig. 1B; ref. 12).
KDC kill and phagocytose osteosarcoma cells. As previously shown ( 12), rat CD4− CD103+ DC (KDC) freshly extracted from the spleen induced rapid (5 hours) cell death in vitro in tumor target cells such as murine YAC-1 cells ( Fig. 1), human Jurkat cells, or the rat OSRGa osteosarcoma cell line ( Fig. 2A ). We then assessed whether KDC could acquire cellular material from killed OSRGa cells in vitro, as previously shown for YAC-1 cells ( 12). To this end, KDC were cocultured for 5 hours with a suspension of live eGFP-expressing OSRGa cells (OSRGa-GFP+) and then recovered for analysis. For FACS analyses, DC were stained with a MHC II mAb, and the phagocytosis of target cells by KDC was assessed by the number of double-positive eGFP+ MHC II+ cells ( Fig. 2B). As shown in Fig. 2B, 20.1% of DC were also eGFPlow, suggesting they had phagocyted eGFP-OSRGa cell fragments. CFSE-labeled Jurkat cells used as a positive control were also efficiently killed by KDC ( Fig. 2B). To confirm the phagocytosis of OSRGa cells, the MHC II+ eGFP− and MHC II+ eGFP+ cell populations were sorted by FACS and analyzed by microscopy and compared with eGFP OSRGa cells. ( Fig. 2C). Various numbers of small to large GFP+ fragments were detected inside MHC II+ eGFP+, but not MHC II+ eGFP− cells, suggesting that DC phagocytosed part or whole killed tumor cells. This observation excludes simple exchange of membranes between the target cells and the DC. These results indicate that in vitro, KDC can kill OSRGa osteosarcoma cells and then acquire cellular fragments from dead cells.
Flt3L expands functional KDC numbers in vivo. A major hurdle in using KDC for tumor immunotherapy is their low numbers in vivo. Flt3L is a key factor for the generation of DC in vivo ( 28, 29), and administration of Flt3L has been shown to expand DC subsets in both mice ( 30, 31) and rats ( 32). We assessed whether Flt3L increased the number of KDC as well as other splenic DC subsets in the rat. For this purpose, rats were injected i.v. with 1011 infectious particles of an AdFlt3L. The presence of high levels of hFlt3L was confirmed in AdFlt3-L but not AddL324-treated rat serum by ELISA with a peak at day 5 ( Fig. 3C ). The frequency and absolute numbers of splenic DC subsets were determined at various intervals by flow cytometry ( Fig. 3A). CD103+ CD4−, CD103+ CD4+, and pDC ( 33) were analyzed after gating on MHC II+ TCR− Igκ− cells as shown in Fig. 3. In control adenovirus-injected animals, the frequency and numbers of each DC subset did not change notably compared with unmanipulated animals. In contrast, AdFlt3L-injected rats exhibited a dramatic expansion of CD103+ CD4− DC (KDC) at day 7 (up to 10-fold compared with a control adenovirus; Fig. 3B). On the other hand, CD103+ CD4+ DC were only expanded by 2-fold, whereas the number of pDC did not change appreciably. The expansion of CD4− DC peaked at day 7 and diminished radically at day 10, correlating with the decrease in serum Flt3L concentration ( Fig. 3C). We next assessed whether Flt3L-expanded CD103+ CD4− DC still exhibited KDC properties. Splenic CD103+ CD4− DC were prepared 7 days after injection of 2 different AdFlt3L doses. Flt3L-expanded CD103+ CD4− DC killed the YAC-1 target cells ( Fig. 3D, left) as well as Jurkat cells (data not shown) as efficiently as CD103+ CD4− DC from control animals, independently of the dose of adenovirus administered. The OSRGa cell line was also killed by AdFlt3L-expanded DC in the same proportion as with control CD4− DC ( Fig. 3D, left), indicating that the expansion did not influence their tumoricidal properties. In addition, Flt3L-expanded KDC exhibit similar to higher phagocytic activity of cellular material from killed OSRGa cells than control KDC ( Fig. 3D, right). Flt3L-expanded KDC displayed the same immature phenotype as those from naive animals (data not shown). This is important as we previously showed that both the cytotoxic and phagocytic activities of KDC are strongly down-regulated upon maturation ( 12). Finally, it is intriguing that in the rat, Flt3L did not seem to expand pDC in vivo, although it is possible that pDC expansion would require longer exposure to Flt3L ( 34).
Antigen-presenting KDC can induce tumor regression. Next, using a tumor vaccination model of transplanted osteosarcoma in the rat, we explored whether this unusual tumor Ag capture pathway enables KDC to present tumor Ag in vivo. For this purpose, KDC were purified from the spleens of AdFlt3L-treated SPD OFA rats and cultured in the presence of OSRGa cells or Jurkat cells as a control (DC/target ratio, 25:1). Human Jurkat cells were chosen as negative controls as they are efficiently killed and phagocytosed by KDC ( Fig. 1B) but are likely to be irrelevant for the rat osteosarcoma model. Importantly, target cells were systematically depleted of dead cells and cell fragments. After overnight coculture, DC were harvested and separated from the remaining and adherent tumor cells that had escaped KDC-induced cell death. These tumor antigen-loaded KDC were used to vaccinate tumor-bearing rats 14 days after implantation of the tumor fragment, which could be considered as bearing progressive tumors. One group of animals received DC that had been cocultured with OSRGa cells and another received DC that had been cultured in the presence of Jurkat cells (n = 9 per group). To evaluate spontaneous tumor growth, a control group of animals was included that received injections of PBS alone. Osteosarcoma-bearing rats received weekly s.c. injections of 4 to 5 × 106 DC at days 14, 21, and 29 posttumor implantation. The mean tumor outgrowth was calculated over a period of 27 days until the first animals died as a result of lung metastasis or were sacrificed because of their tumor volume ( Fig. 4A ). Tumor growth measurement revealed a rapid increase in tumor size, as shown in the control group, as soon as 15 days posttumor implantation (mean tumor volume, 5,523 mm3; Fig. 4A). The mean tumor growth was strongly and significantly delayed in the group of rats immunized with OSRGa-cultured KDC compared with the control group (6,523 versus 25,396 mm3 at day 27; P = 0.01). In contrast, vaccination with KDC cultured in the presence of Jurkat cells did not significantly prevent the increase in tumor size (18,143 versus 25,396 mm3 in the control group; not significant). Animal survival was also monitored for >50 days ( Fig. 4B). Animals vaccinated with KDC-OSRGa exhibited a higher survival rate compared with controls (66.7% versus 22.2% at day 36 posttumor implantation; P < 0.01), whereas Jurkat-DC treated animals showed no significant variation compared with control rats (33.3 versus 22.2 at day 36–49 posttumor implantation; not significant).
These results suggest that KDC can efficiently present tumor antigens captured from tumor cells.
To exclude the possibility that the vaccination effect was due to a passive transfer of tumor cellular antigens, the DC that phagocytosed the OSRGa cells were sorted. KDC were cultured for 4 hours in the presence of OSRGa GFP+ cells (ratio, 25:1) that were first depleted of dead cells and cell fragments using Annexin V–coated magnetic beads. Next, cells were collected, stained with a MHC-APC Cy7 mAb, and MHC II+ GFP+ cells were sorted by FACS after exclusion of doublets ( Fig. 5A ). Two to 5 × 105 sorted cells were injected s.c. into osteosarcoma-bearing rats every week for 3 weeks, beginning at day 14 posttumor transplantation. As a control, a group of animals received only PBS. As shown in Fig. 5B, a significant delay in the tumor outgrowth was observed in the group of rats vaccinated with KDC that had previously phagocytosed OSRGa cells compared with the control group (P = 0.01 at day 29; Fig. 5B). The fact that the antitumor effect was less potent than in the previous experiments could be explained by the lower number of injected cells (10-fold lower) and a potential difference in the maturation state, as the cells were injected directly after the 4-hour culture with OSRGa cells.
KDC vaccination–induced tumor regression is CD8 T-cell dependent. To determine whether the vaccination effect induced by tumor antigen-presenting KDC was indeed related to the induction of adaptive T-cell response and more specifically a CD8 T-cell response, we compared the vaccination effect in CD8+ T cell–depleted animals versus unmodified recipients. Prolonged CD8 T-cell depletion was induced by i.p. injection of a mouse anti-rat CD8α mAb ( Fig. 6A ). We previously showed that such a treatment did not affect NK cells whose 50% express CD8α in the rat ( 35). As shown in Fig. 6B, 5 of 7 KDC-vaccinated animals in the CD8 T cell–depleted group exhibited a progressive tumor and 1 animal died at day 33. In contrast, in unmodified and vaccinated animals, we observed 1 progressive tumor of 7 rats.
Finally, we rechallenged in two independent experiments five animals from the DC-OSRGa vaccinated group that completely rejected the tumor with an osteosarcoma tumor on one side and an irrelevant chondrosarcoma tumor on the other side. No osteosarcoma tumor growth was observed in these animals, whereas chondrosarcoma tumor developed in all rats, suggesting that vaccination induced a long lasting and specific antitumor response.
The present study provides evidence that rat KDC may efficiently capture antigen from their victims, at least in vitro, and then present tumor antigens to T cells in vivo. The fact that vaccination was inefficient in CD8 T cell–depleted animals indicates that tumor regression was dependent on the generation of CTL directed against OSRGa tumor antigens. This would imply that KDC can efficiently cross-present tumor antigens. Although not tested in this study, it is possible that cross-presentation in this model is also CD4-dependent. The possibility that injected KDC could have a direct cytotoxic activity on tumors can be excluded on the following grounds: (a) KDC were injected s.c. and were therefore unlikely to gain access to the tumor; (b) KDC were injected after overnight culture, which induces a dramatic down-regulation of their killing properties ( 12); and (c) the effect was antigen-specific as KDC that had been cultured with Jurkat cells did not induce a significant delay in tumor progression.
The physiologic role of KDC is still unknown and it remains unclear why some professional APC exhibit cytotoxic activity (reviewed in ref. 10). Because of their low numbers in vivo, it seems unlikely that DC with killing properties play an important direct role in tumor regression by killing tumor cells. However, tumors can be heavily infiltrated by various DC subsets and a recent report in humans showed that after local treatment with a TLR7/8 agonist, cutaneous carcinomas are infiltrated and surrounded by perforin-granzyme–expressing CD11c+ DC and TRAIL-expressing pDC ( 36). Whether tumor regression is directly related to KDC is unknown. In the mouse, so-called IKDC were recently shown to play a role in tumor regression in vivo. Indeed, intratumor injection of IKDC purified from imatinib mesylate+IL2-treated mice into melanoma-bearing Rag2−/−IL2Rγ−/− mice led to an impaired tumor outgrowth ( 15). However, the APC function of IKDC has not been fully shown, and in fact, IKDC were recently shown to be a subset of NK cells that are probably activated and devoid of APC function ( 16– 18). We believe that the KDC subset we described in rats are not related to NK cells because: (a) KDC exhibit a homogeneous MHCII+ CD11b+ CD103high NKp46− phenotype, whereas NK cells are MHCII− CD11b− CD103− NKp46+ ( Fig. 1); (b) KDC produce very little, if any, IFN-γ ( 33); (c) target cells are killed and are immediately phagocyted by KDC ( 12); and (d) upon TLR or CD40L stimulation, KDC produce large amounts of IL-12, up-regulate MHCII, CD80, and CD86 molecules, and become potent APC in vitro ( 33), whereas strongly down-regulating their tumoricidal and phagocytic properties ( 12).
We propose that the tumoricidal properties of KDC would allow these DC to acquire cellular antigens from apoptotic tumor cells. Previously, we showed that the acquisition of cellular materials from target cells by KDC was strictly dependent on their killing by the same DC in a time-dependent fashion ( 12). We did not observed significant fluorescence exchange, either by FACS or confocal microscopy, between KDC and target cells that were resistant to their killing activity. However, we cannot formally exclude that, in the presence of target cells sensitive to KDC-induced cell death, KDC can also acquire antigens from live cells, a process called nibbling, as previously shown with human DC ( 37). Because in vivo apoptotic cells are rapidly cleared by macrophages, a DC population exhibiting both killing and phagocytosis functions should strongly increase their intrinsic potential to both phagocytose tumor cells and present tumor antigens. Our results indicate that rat KDC have the capacity to phagocytose cell fragments from the target cells they previously recognized and killed ( 12), and to present tumor antigens derived from this cellular material for inducing antitumor response. Whether such a phenomenon can occur in vivo remains to be established. However, tolerance could also occur if KDC do not receive a strong maturation signal, a situation that probably occurs in vivo in tumor beds. In the latter case, tumor antigen presentation by immature KDC could favor T-cell tolerance rather than immunity.
Our results suggest that the direct cytotoxic activity of KDC toward tumor cells could be harnessed ex vivo to develop new antitumor therapeutic strategies. Although the generalization of this work is limited by the resistance of certain tumor cell lines to lysis by KDC in the rat ( 12), these results constitute an encouraging approach in the development of new therapeutic tools for the treatment of primitive bone tumor pathologies. Osteosarcomas are the most frequent form of primary bone tumors and develop mainly at a young age. Moreover, despite recent improvements in chemotherapy and surgery, nonresponse to chemotherapy remains a problem and current strategies for the treatment of high-grade osteosarcoma are failing to improve prognosis. In fact, very few immunotherapeutic strategies have shown efficacy in the treatment of osteosarcoma ( 38). Vaccine with CD80-transfected osteosarcoma cells was shown to induce efficient immune response to parental tumors in rats ( 39). The same group reported later that in vivo adenovirus-mediated transduction of CD80 in osteosarcoma cells induced curative immunity to parental tumors ( 40). Other preclinical studies have shown that vaccination of rats with bone marrow–derived DC electrofused with osteosarcoma cells induced protective immunity in tumor-bearing rats ( 41) and osteosarcoma-specific CTL response ( 41). Wongkajornslip and colleagues showed ( 42) a cytotoxic activity exerted by CTL generated in vitro upon stimulation with antigen-presenting DC toward a human Ewing sarcoma cell line. However, the injection of osteosarcoma-specific CTL does not allow the maintenance of the in vivo response upon a tumor rechallenge. The originality of our work resides also in the way of tumor antigen loading of DC. As compared with DC electrofusion with tumor cells or the loading with tumor lysates, KDC can directly acquire antigens from unmanipulated osteosarcoma cells likely allowing the processing and presentation of a large number of tumor antigens.
Another therapeutic approach could consist in the direct intratumoral injection of immature KDC. However, it seems more difficult to use in clinical studies and additional experiments are required to determine whether injected KDC acquire tumor material and migrate to regional lymph nodes. It is moreover possible that the tumor microenvironnement could inhibit DC-mediated killing and DC maturation. Alternatively, intratumoral expression of Flt3L might promote KDC expansion in vivo, but whether these cells could then migrate to the tumor remains to be established. Again, this approach should be associated to the delivery of a delayed maturation signal. Therefore, the development of new immunotherapies such as that described here are of great interest for the improvement of treatments for osteosarcoma-bearing patients.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: Institut National de la Sante et de la Recherche Medicale (INSERM), Association pour la Recherche sur le Cancer (ARC; grant # 3561). C. Chauvin was supported by La ligue Régionale Contre le Cancer (Pays de la Loire), the ARC, and the Progreffe Foundation. F.X. Hubert was supported by INSERM-Région Pays de la Loire. B. Trinité was supported by the ARC.
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 Erik Dissen for providing anti-rat NKp46 mAb.
Note: J.M. Philippeau and C. Hémont contributed equally to this report. F. Rédini and R. Josien share senior authorship.
Current address for C. Chauvin: Nuffield Department of Surgery, John Radcliffe Hospital, Headington, OX3 9DU Oxford, United Kingdom.
Current address for F.X. Hubert: Division of Immunology, The Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, Victoria 3050, Australia.
Current address for B. Trinité: Mount Sinai School of Medicine, Department of Immunobiology, 1425 Madison Avenue, New York, NY 10029.
- Received January 10, 2008.
- Revision received August 20, 2008.
- Accepted September 2, 2008.
- ©2008 American Association for Cancer Research.