Myeloid-derived suppressor cells (MDSCs), which are known to be accumulated in the blood, spleen, and bone marrow of tumor-bearing mice and cancer patients, were tested as APCs for a cellular vaccine because they have phenotypical similarity with inflammatory monocytes and may be differentiated from the same precursors as monocytes. Although MDSCs have immunosuppressive properties, in vivo transferred MDSCs, which present tumor Ag and NKT cell ligand (α-galactosylceramide), significantly prolonged survival time in metastatic tumor-bearing mice in a CD8+ cell-, NK cell-, and NKT cell-dependent manner vs a CD4+ T cell- and host dendritic cell-independent manner. Major concerns about using MDSCs as APCs in a vaccine are their suppression of CTLs and their induction of Foxp3+ regulatory T cells. However, α-galactosylceramide-loaded MDSCs did not suppress CD4+ and CD8+ T cells and allowed for the generation of Ag-specific CTL immunity without increasing the generation of regulatory T cells. Furthermore, stimulation with activated NKT cells induced changes on MDSCs in phenotypical or maturation markers, including CD11b, CD11c, and CD86. Taken together, these findings suggest that NKT cells facilitate the conversion of immunosuppressive MDSCs into immunogenic APCs, eliciting successful antitumor immunity and providing the basis for alternative cell-based vaccines.
Cell-based vaccines using APCs have been shown to activate CD8+ T cells and CD4+ T cells effectively, resulting in excellent antitumor effect (1). The most frequently used APCs for cell-based vaccine are dendritic cells (DCs),3 which take up Ag and present it to T cells along with costimulatory signals, meaning that DCs can induce strong immune responses against the presented Ag (2, 3, 4, 5). Despite the advantages of DCs as APCs, the development of cellular immunotherapeutic agents using DCs remain limited because DCs are present in blood and lymphoid tissues in only small numbers and are difficult to isolate. In addition, several days are required for ex vivo generation of DCs from precursors with the addition of cytokines (6, 7). Therefore, an alternative cellular immunotherapeutic agent is needed.
Immature myeloid cells, also called myeloid-derived suppressor cells (MDSCs), are accumulated in cancer patients (8). Among the immature myeloid cells are immature macrophages, granulocytes, immature dendritic cells, monocytes, and other myeloid cells in early differential stages (9) that cannot further differentiate and that accumulate because of the influence of tumor-derived factors such as IL-6, IL-10, vascular endothelial growth factor, and GM-CSF (10). The MDSCs are known to foster a tumor-suppressive environment by inhibiting the functions of Ag-specific or nonspecific T lymphocytes; they do so by taking advantage of arginase I (11), nitrogen oxide (12), reactive oxygen species (13, 14), and TGF-β (15). MDSCs, which express Gr-1 and CD11b simultaneously on their surface and are constituted by mononuclear cells (including monocytes and macrophages), neutrophils, and eosinophils (16, 17, 18), are known to proliferate and accumulate in mice transplanted with tumor. Given that MDSCs accumulate in the blood of cancer patients, a large quantity of them can easily be obtained. In addition, when monocytes are separated to produce a dendritic cell vaccine in a cancer-bearing host, it is difficult to selectively isolate monocytes from the mixture of MDSCs, a difficulty that could bar development of DC-based vaccines. Thus, conversion of immunosuppressive MDSCs into immunogenic APCs could provide profound benefit for successful immunotherapy.
Recently, it has been shown that, after the loading of α-galactosylceramide (αGalCer) on their CD1d, DCs obtain enhanced immunity with the help of activated invariant NKT (iNKT) cells (19, 20). Furthermore, we have shown that B cells can also be used as effective APCs with the help of activated NKT cells, which elicited equivalent levels of anticancer immunity compared with a DC-based vaccine (21, 22). However, no study has investigated whether MDSCs or even monocytes can enhance immunity with the help of activated NKT cells. In the present study, we investigate whether monocytes and MDSCs can be effectively used for the production of a cell-based vaccine by loading αGalCer, an invariant NKT ligand, and presenting antigenic peptide on an MHC class I molecule. We show that MDSCs loaded with an Ag and NKT ligand can be converted into immunogenic APCs and can lead to significant anticancer effects, including Ag-specific cytotoxic immune responses and innate immune responses (such as NK cytotoxicity) against cancer cells. MDSC-based vaccines show promise as a novel immunotherapeutic and prophylactic anticancer agent.
Materials and Methods
BALB/c and C57BL/6 mice (both from Charles River Laboratories) were purchased at 6 wk of age. The CD11c-DTR, CD45.1, DO11.10, and OT-1 breeding pairs were purchased from The Jackson Laboratory. The CD1d−/− mice were provided by Dr. Se-Ho Park, Korea University, Seoul, South Korea. All mice were kept under specific pathogen-free conditions in the Animal Center for Pharmaceutical Research at Seoul National University, Seoul, South Korea. The experiments were approved by the Institutional Animal Care and Use Committee of Seoul National University.
Abs and reagents
Hybridomas producing depleting anti-CD4 (clone GK1.5) and anti-CD8 (clone 2.43) mAb were purchased from the American Type Culture Collection. Anti-CD25 (clone PC61) mAb were obtained from ascites as described previously (23). Mice received i.p. injections of 150 μg of depleting Abs. For depletion of NK cells, mice were treated with 20 μl of polyclonal rabbit anti-asialo-GM-1 Abs purchased from Wako Chemicals, and the same amount of rat IgG (Sigma-Aldrich) was used as a control. For depletion of DCs, CD11c-DTR mice were injected with 150 ng of diphtheria toxin (DTX; Sigma-Aldrich) at days, −1, 2, and 6 after immunization. DTX treatment depleted >85% of CD11c+ DC at both day 0 and day 10 (data not shown).
CT26 (ATCC), Her-2/neu-expressing transfectoma Her-2/CT26 cells (24), or mHer-2/CT26, murine Her-2/neu-expressing CT26 cells (25), were used to develop solid tumor or lung metastasis models in BALB/c mice. A murine thymoma cell line, EL-4 (ATCC), was used for the induction of MDSCs from C57BL/6 mice.
+ T cells or CD8+ +
Construction of recombinant adenovirus
Recombinant adenovirus encoding the extracellular and transmembrane domains of Her-2/neu (AdHM) was constructed by Viromed (Seoul, Korea) as previously described (22). The transduction efficacy of AdHM in primary cells, especially in CD11b+ cells obtained from splenocytes, was ∼40% after 24 h of incubation with 100 multiplicity of infection (MOI) of AdHM when the expression of Her-2/neu on the cell surface was analyzed by FACS.
Preparation of monocyte or MDSC vaccine
PBMCs and splenocytes were isolated from BALB/c mice. Granulocytes and RBCs were removed by Ficoll (Sigma-Aldrich) density gradient centrifugation. After depletion of B220+ and CD11c+ cells from the cells by anti-B220 and anti-CD11c microbeads, CD11b+ cells were purified using anti-CD11b microbeads (all from Miltenyi Biotec). To produce a peptide-loaded monocyte vaccine, CD11b+ cells incubated with αGalCer (1.5 μg/ml) or vehicle (0.5% Tween in PBS) for 14 h were additionally pulsed with 2.5 μg/ml Her-2/neu63–71 epitope peptide (Anygen) for 1 h. To produce Ag-expressing monocyte vaccine, CD11b+ monocytes were transduced with 100 MOI of AdHM in a serum-free medium for 60–90 min in a CO2 incubator. Serum was added thereto and αGalCer (1.5 μg/ml) or vehicle (0.5% Tween in PBS) was also added and incubated for an additional 14 h.
To prepare the MDSC vaccine, Her-2/CT26, mHer-2/CT26, or EL-4 were injected s.c. into syngenic mice. When the tumor volume was grown to at least 1500 mm3, spleens were isolated from the mice followed by homogenization. B220+ cells or CD11c+ cells were eliminated by anti-B220 and anti-CD11c microbeads. Obtained cells were incubated with αGalCer (1.5 μg/ml), vitamin A (20 μM all-trans+ cells isolated by using anti-CD11b microbeads were pulsed with 2.5 μg/mL peptide for 1 h or transduced with 100 MOI of AdHM for 5–6 h. Both monocyte- and MDSC-based vaccines (2 × 106 cells/mouse) were i.v. injected.
To assess how NKT functioned in the generation of antitumor immunity by MDSC vaccine, MDSCs were obtained from splenocytes of wild-type C57BL/6 or CD1d−/− mice that had an EL-4 tumor (∼1500 mm3) transplanted in the left flank. C57BL/6 mice were immunized with MDSC/peptide (loaded with OVA257–264 peptide) or MDSC/peptide/αGalCer.
Preparation of bone marrow-derived DC vaccine
Dendritic cells were generated from bone marrow cells of naive or tumor-bearing mice as described previously (26). To prepare DC vaccine, DCs were incubated with 1.5 μg/ml αGalCer. On the following day, these cells were pulsed with Her-2/neu63–71 peptide (2.5 μg/ml) for 1 h.
Intracellular cytokine staining
To analyze cytokine production of NKT cells, splenocytes were incubated for 4 h in RPMI 1640 supplemented with 10% FBS and GolgiPlug (BD Pharmingen) with or without PMA (1 μg/ml) and ionomycin (100 ng/ml) (all from Sigma-Aldrich). Cells were fixed after surface Ag staining and permeabilized with Cytofix/Cytoperm reagents (BD Pharmingen) before being stained with PE-conjugated control, anti-IFN-γ, or anti-IL-4 Abs. For detection of intracellular Foxp3 expression in CD4+
To determine the degree of CD8+ T cell activation induced by the MDSC vaccine, CD45.1 congenic mice were given 5 × 106 OT-I cells 1 day before MDSC vaccination. Twelve days after vaccination, splenocytes were isolated and restimulated with or without 2.5 μg/ml OVA peptide257–264 for 2 days. After an additional 6 h of incubation in GolgiPlug-containing medium, cells were stained with biotin-labeled CD45.2, streptavidin-allophycocyanin- and PE-labeled CD3 Abs, and intracellular IFN-γ production was analyzed.
In vivo cytotoxicity assay
An in vivo CTL assay was performed as previously described (21r = percentage of CFSElow/percentage of CFSEhigh and percentage of lysis = [1 − (runprimed/rprimed)] × 100, where r is the ratio.
Metastasis tumor models
To assess the antimetastasis effect of the vaccine, mice were challenged with 2 × 105 cells of Her-2/CT26 on day 0. On the following day, 1 × 106 cells of manipulated monocytes or MDSC vaccine were administered i.v. to tumor-challenged mice. To assess the specific immune cells for the antitumor effect, mice were treated with anti-CD4, anti-CD8, anti-CD25, and anti-asialo-GM-1–depleting Ab i.p. at intervals of 3–4 days beginning 2 days before Her-2/CT26 cancer cells were injected i.v. into the BALB/c mice (2 × 105 cells/mouse). On the day after the cancer cells were injected, mice received an i.v. injection of the indicated MDSC-based vaccines. Mice showing abnormal tumor-related symptoms were sacrificed for humanitarian reasons and pulmonary tumor nodules were identified (data not shown).
CD4+ T cells or CD8+ T cells were isolated from naive BALB/c mice using anti-CD4 or anti-CD8 microbeads (all from Miltenyi Biotec), respectively. T cells (2 × 105 cells/well) were stimulated with mitomycin C-treated allogenic lymphocytes (5 × 105 cells/well) in 96-well plates for 2 days and pulsed for 18 h with 4 μCi/ml [3H]thymidine. Serially diluted MDSCs were added during incubation to assess their suppression of T cell proliferation.
The Kaplan-Meier method was used to determine the statistical significance of differences in survival time. We performed the log-rank test (Mantel-Cox) using SPSS 12.0K for Windows. To compare the differences between two groups, Student’s t test was used. To compare multiple groups we conducted one-way ANOVA, followed by the Tukey’s HSD (honestly significant differences) post hoc test.
αGalCer-loaded monocytes induced tumor Ag-specific CTL responses and significant antitumor effects against metastatic tumor
We tested whether monocytes can be used in an APC-based vaccine without an in vitro maturation process. As previously reported, monocytes are poor at eliciting Ag-specific CTLs compared with DCs when pulsed with antigenic CTL epitope peptide (Fig. 1⇓A). Based on our previous data showing that activated NKT cells help poorly immunogenic B cells to successfully elicit Ag-specific CTLs against an H-2Kd-restricted CTL epitope peptide (Her-2/neu63–71 peptide, designated hP63) of human Her-2/neu (21), we tested whether αGalCer-loaded monocytes presenting Her-2/neu63–71 peptide (designated Mo/hP63/αGC) could also generate Her-2/neu63–71 peptide-specific CTL responses. Immunization of mice with Mo/hP63/αGC, but not with αGalCer-unloaded monocytes presenting Her-2/neu63–71 peptide (designated Mo/hP63), induced significant levels of CTL responses against the Her-2/neu63–71 peptide, as did bone marrow-derived DC (designated BmDC in Fig. 1⇓A).
To test the antitumor effect elicited by the manipulated monocyte vaccine, we injected BALB/c mice i.v. with 2 × 105 cells of Her-2/CT26 tumor on day 0, followed by vaccination on day 1. Although both Mo/αGC and Mo/hP63/αGC induced significant antitumor activity, Mo/hP63/αGC led to a significant extension in mean survival time of tumor-challenged mice as compared with Mo/αGC-treated mice (Fig. 1⇑B). The antitumor effect of αGalCer-loaded monocytes without peptide can be ascribed to the direct effect of αGalCer on antitumor immunity or to the activation of monocytes by αGalCer. We next assessed whether a monocyte-based vaccine expressing whole tumor Ag instead of peptide could also induce antitumor immunity. Using AdHM, an adenovirus encoding the transmembrane and extracellular domains but not the cytoplasmic kinase domain of Her-2/neu, to remove its oncogenic potential, we expressed Her-2/neu on the cell surface of monocytes (data not shown). The Mo/AdHM/αGC-immunized group showed a significant increase in survival over the group immunized with Mo/AdHM or Mo/αGC (Fig. 1⇑C). Collectively, these findings show that vaccination of mice with αGalCer-loaded monocytes presenting a tumor Ag induced strong Ag-specific CTL responses and successful antitumor immunity against circulating metastatic tumor cells.
MDSCs could be converted into immunogenic APCs with the help of activated NKT cells
Because MDSCs are greatly increased in tumor-bearing mice and in patients with cancer, and because they have phenotypical similarity and share a common lineage with monocytes, we asked whether MDSCs obtained from tumor-bearing mice would also be able to induce antitumor immunity when loaded with αGalCer. We isolated CD11b+Gr-1+ MDSCs from splenocytes (Fig. 2⇓A) or bone marrow (data not shown) of tumor-bearing mice after depleting CD11c+ or B220+ cells and confirmed their suppressive activity (Fig. 2⇓, B and C). In addition, the lowered surface expression of CD80, CD86, CD40, MHC class I, MHC class II, and CD1d in CD11b+Gr-1+ MDSCs compared with CD11c+MHC class II+ DCs indicated immature phenotypes among MDSCs (Fig. 1⇑D).
To our surprise, despite their well-known suppressive characteristics, MDSCs loaded with Ag peptide and αGalCer (designated MDSC/hP63/αGC) induced significantly higher hP63-specific cytolysis than that observed in MDSC/hP63-immunized mice (Fig. 2⇑D). Furthermore, immunization with MDSC/hP63/αGC increased protection against the development of metastases (Fig. 2⇑E). However, MDSC treated with all-trans retinoic acid (designated ATRA) or GM-CSF to induce differentiation-elicited marginal increases in both the hP63-specific CTL responses and antitumor effects (Fig. 2⇑, D and E). Next, we compared the antitumor effect of these MDSC vaccines with the DC-based vaccine in the metastatic tumor model. Vaccination with MDSC/hP63/αGC led to a significant extension in survival time, one that was comparable to the antitumor effect of bone marrow-derived DCs loaded with hP63 and αGalCer (designated BmDC/hP63/αGC in Fig. 2⇑F). Although BmDC/hP63 vaccination induced strong CTL activity in itself, MDSC-based vaccine required αGalCer loading to generate successful CTL activity in vivo (Fig. 2⇑G). The survival data (Fig. 2⇑F) show MDSC/hP63/αGC induced CTL responses comparable to those BmDC/hP63/αGC vaccine induced (Fig. 2⇑G). When groups of mice were immunized with MDSC-based vaccines expressing whole tumor Ag such as Her-2/neu, mice vaccinated with MDSC/AdHM/αGC showed significantly higher resistance against tumor challenge than those immunized with MDSC/AdHM (Fig. 2⇑H). In contrast to its weak antitumor effect, MDSC/AdHM elicited levels of cytotoxicity against hP63-pulsed target cells that were comparable with those elicited by MDSC/AdHM/αGC (Fig. 2⇑I), suggesting that factors other than CTL could also account for the antitumor effects of the MDSC/AdHM/αGC vaccine (Fig. 2⇑H). Furthermore, depletion of CD25+ cells, encompassing Foxp3+ regulatory T (Treg) cells, dramatically increased the survival of tumor-challenged mice by eliminating the subsisting tumor cells (Fig. 2⇑J). In addition, we repeated the challenge 115 days after the first tumor challenge, confirming the establishment of memory responses against the same tumor (Fig. 2⇑J). As a negative control, depletion of Treg cells alone could not generate an antitumor effect (Fig. 2⇑K). To our surprise, however, depletion of Treg cells during MDSC/AdHM could not extend the mean survival time of immunized mice (Fig. 2⇑K). Collectively, these data show that MDSCs are attractive candidates for APC-based antitumor vaccines and that depletion of Treg cells significantly increased the antitumor effects of the MDSC vaccine.
Cellular prerequisites for the generation of successful antitumor immunity by MDSC/AdHM/αGC vaccine
We next assessed whether NKT cells are necessary for the induction of Ag-specific CTL activity. When we used MDSC from OVA-expressing tumor-bearing CD1d−/− mice, αGalCer-loading did not increase the CTL activity of the MDSC/OVA peptide257–264 (designated MDSC/pep) vaccine as it had done in C57BL/6 wild-type mice (Fig. 3⇓A). These data suggest that NKT cell activation by CD1d-loaded αGalCer is indispensable for the successful generation of CTL responses by MDSC vaccines. We also detected the increase of IFN-γ in activated NKT cells after MDSC/αGC treatment, but not after MDSC treatment, either in naive (data not shown) or in tumor-bearing mice (Fig. 3⇓B). Taken together, these findings suggest that αGC-loaded MDSCs induced NKT cell activation even in tumor-bearing mice and that activated NKT cells augmented Ag-specific CTL responses induced by MDSC immunization.
To investigate which types of immune cells are involved in generation of antitumor immunity using the MDSC-based vaccine, we depleted CD4+, CD8+, or NK cells in the immunized mice using anti-CD4, anti-CD8, or anti-asialo-GM-1 Abs, respectively. As expected, depletion of CD8+ cells in MDSC/AdHM/αGC-treated mice significantly abrogated the antitumor effect, indicating that the therapeutic antitumor effect was dependent on CD8+ CTLs (Fig. 3⇑C). In contrast, depletion of CD4+ cells did not affect the antitumor effect of the MDSC/AdHM/αGC vaccine (Fig. 3⇑C). Likewise, CD4+ cells were not required for the generation of hP63-specific CTL responses (Fig. 3⇑E). Surprisingly, however, anti-asialo-GM-1 Ab treatment in MDSC/AdHM/αGC-immunized mice abrogated the antitumor effect, showing that the therapeutic antitumor effect was highly dependent on asialo-GM-1-expressing NK cells (Fig. 3⇑C). On the contrary, the depletion of CD4+, CD8+, or asialo-GM-1+ cells in MDSC/AdHM-immunized mice significantly abrogated the antitumor effect of MDSC/AdHM vaccine (Fig. 3⇑D). These data suggest that both CD8+ cells and NK cells are necessary for the induction of antitumor effects by the αGalCer-loaded MDSC vaccine and that NKT cell activation by αGalCer in MDSC-based vaccine can compensate the helper function of CD4+ T cells for the generation of successful antitumor CTL activity in the absence of CD4+ T cells.
To clarify the roles of asialo-GM-1-expressing NK cells after MDSC vaccination, we first assessed whether it was possible to generate CTLs when NK cells were depleted by anti-asialo-GM-1 Abs. Although treatment with anti-asialo-GM-1 Ab significantly reduced the specific lysis of target cells by MDSC/pep/αGC, there were higher levels of CTL activity than with MDSC/pep (Fig. 3⇑F). Thus, the abrogation of antimetastatic activity by anti-asialo-GM-1 Ab treatment cannot be solely ascribed to deficient CTL activity in the NK-depleted group. The lowered expression in the large tumor mass of MHC class I (H-2Kd) molecules (Fig. 3⇑G), not in the small tumor mass, might make them susceptible to NK-mediated killing. Furthermore, the spleen of MDSC/hP63/αGC-vaccinated mice showed a profound decrease in the population of NKT cells after anti-asialo-GM-1 Ab treatment, but the spleen of naive mice did not (Fig. 3⇑H), demonstrating that activated NKT cells might also be depleted by anti-asialo-GM-1-depleting Ab treatment. However, we failed to detect significant asialo-GM-1 expression in B220−TCRβ+αGC/dimer+ NKT cell from MDSC/hP63/αGC-vaccinated mice 6 and 24 h after vaccination (data not shown), suggesting more complex interaction among innate immunity after NKT cell activation. Collectively, these data show that NK cells, CD8+ T cells, and NKT cells, but not CD4+ T cells, are necessary to produce the antimetastatic effects of the MDSC vaccine.
MDSCs function as APCs after interaction with NKT cells
Next, we sought to verify the characteristics of transformed MDSC after activation by NKT cells by assessing the phenotypes and maturation of MDSC in vivo. CD11c−B220−CD11b+ MDSCs, which were obtained from the splenocytes of tumor-bearing mice, were adoptively transferred into naive or tumor-bearing mice after CFSE labeling; CFSE-labeled MDSCs were analyzed in the splenocytes at day 3 after immunization. The percentage of CFSE+ cells among splenocytes was very low but similar in number between MDSC- or MDSC/αGC-transferred groups (data not shown), suggesting that there was no significant NKT-dependent lysis of αGalCer-loaded MDSC in vivo. However, the level of CD11b was decreased in both naive (Fig. 4⇓, A and C) and tumor-bearing mice (Fig. 4⇓, B and D) when loaded with αGalCer. Although we could detect a significant increase in CD11c expression in MDSC transferred to naive mice (Fig. 4⇓, A and C) but not to tumor-bearing mice (Fig. 4⇓, B and D), there was a significant decrease in the expression level of CD11b on MDSC in both naive and tumor-bearing mice, suggesting the differentiation of MDSC. Further, αGalCer-loaded MDSCs increased the expression of CD86, demonstrating the shift of MDSCs into activated APCs after stimulation by activated NKT cells.
We next checked whether the combination of antigenic peptide on class I molecule and αGalCer on CD1d in the same MDSC was necessary for the generation of effective CTL responses. Interestingly, the cytolytic activity could be fully induced only when αGalCer and peptide were copulsed to the same MDSC (Fig. 5⇓A). Furthermore, αGalCer-loaded MDSC-mediated NKT cell activation did not afford the same protection against tumor as administration of MDSC/AdHM/αGC (Fig. 5⇓B).
It could be argued that MDSCs acted not as APCs but instead as reservoirs of Ag that was cross-presented by host DCs to induce CTL responses. To explore this possibility, we used a CD11c-DTR mouse model for the conditional depletion of CD11c+ DCs by DTX treatment. The Ag-specific killing of target cells was not influenced by DTX treatment of CD11c-DTR mice immunized with MDSC vaccine (Fig. 5⇑C), suggesting that MDSCs were able to present Ag to T cells as APCs.
MDSC vaccines are immunogenic rather than suppressive when activated with NKT cells
Major concerns raised by using MDSC as an APC vaccine are their suppression of CTL responses (12) and their promotion of Foxp3+ CD4+ Treg cell generation (15). To investigate whether NKT stimulation could change the effect of MDSCs on T cells from suppressive to immunogenic, we adopted OVA-specific TCR transgenic CD8+ (OT-1) or CD4+ (DO11.10) T cells. To analyze the changes in CD8+ T cells after MDSC vaccination, OVA-specific CD8+ T cells were adoptively transferred to mice 1 day before immunization. At day 2 after MDSC/pep immunization, peptide presentation on H-2Kb of MDSCs significantly increased proliferation of CD8+ T cells in vivo, which was further increased by αGalCer treatment (Fig. 6⇓A). To assess Ag-specific IFN-γ secretion of CD8+ T cells, CFSE-labeled OT-I cells of the spleen were obtained 24 h after vaccination and analyzed after 24 h of in vitro restimulation with cognate peptide for CD8+ T cells. When compared with proliferation results, there were only a few IFN-γ-secreting CD8+ T cells in the MDSC/pep-immunized group, whereas OVA-specific IFN-γ secretion was significantly increased by NKT stimulation in both naive and tumor-bearing mice (Fig. 6⇓B).
We then investigated the influence of MDSC vaccine on CTL responses in an established tumor-suppressive environment. Groups of mice with an established tumor on their left flank received MDSC vaccines 1 day after CD45.2+ OT-I cell transfer to congenic CD45.1+ mice. Twelve days after vaccination, IFN-γ secretion could be detected in mice immunized with MDSC/pep or MDSC/αGC, perhaps because of the priming by MDSCs presenting Ag originating from tumor cells. However, MDSC/pep/αGC treatment significantly enhanced the Ag-specific IFN-γ secretion even in tumor-bearing mice (Fig. 6⇑C), suggesting that MDSC vaccines induced Ag-specific CTLs when loaded with αGalCer rather than suppressing CTL function.
We next sought to determine whether NKT stimulation of MDSCs would influence the proliferation of OVA-specific CD4+ T cells or the generation of Foxp3+ Tregs in vitro. To obtain αGalCer- or GM-CSF-conditioned MDSCs, CD11c- and B220-depleted splenocytes of tumor-bearing mice were cultured with αGalCer or GM-CSF in the presence of NKT cells. Two days later, they were cocultured with CFSE-labeled DO11.10 CD4+ T cells for 3 days in vitro. When stimulated with OVA peptide (OVA323–339), both αGC- and GM-CSF-treated splenocytes containing MDSCs led to efficient OVA-specific proliferation of DO11.10 CD4+ T cells, whereas control cells did not (Fig. 7⇓A). Interestingly, GM-CSF-conditioned MDSCs significantly increased Foxp3+ cells among proliferating cells, whereas αGalCer-conditioned MDSCs limited the generation or expansion of Treg cells after in vitro stimulation with peptide (Fig. 7⇓B). When assessed in vivo, both MDSC/pep and MDSC/pep/αGC vaccination induced comparable levels of proliferation of CD4+ T cells despite differences in antitumor immunity between groups, suggesting that CD4+ T cell activation might not be essential for establishment of successful antitumor immunity by MDSC/pep/αGC vaccine. However, the lowered level of Treg cells in MDSC/pep/αGC-vaccinated mice compared with that of MDSC/pep-immunized mice could at least partially explain the superiority of MDSC/pep/αGC vaccine to MDSC/pep vaccine (Fig. 7⇓C). These data suggest that the αGC-loaded MDSC vaccine could be immunogenic for CD4+ T cells rather than increase Treg cells.
Despite breakthroughs in anticancer therapies such as microsurgery, radiotherapy, and chemotherapy, the effects of treatment on malignant tumors are still limited and side effects resulting from nonspecific anticancer effects and relapse of cancer are still problems that need to be addressed. In particular, disseminated metastases have been reported to be the primary cause of mortality in cancer patients (27, 28). To protect cancer patients against tumor recurrence, the induction of tumor-specific long-term memory responses seems to be necessary after the elimination of the major tumor burden by classic treatments. In these aspects, cell-based immunotherapy using APCs has the advantages of reducing side effects caused by systemic toxicity and improving the results of conventional cancer therapy by establishing positive memory response against the tumor Ag, resulting in excellent antitumor effects. In addition to DCs, we have used αGalCer-loaded, Ag-presenting B cells in a novel APC-based vaccine that increased the immunogenicity of B cells and led to cytotoxic T cell responses against Ag that were comparable to those achieved with the DC-based vaccine (21). However, there has been no report of NKT cell activation-enhanced immunity in monocytes, in the precursors of DCs, or even in MDSCs. We provided here a novel MDSC-based vaccine strategy, comparable to DC vaccine, that eliminated tumors by inducing Ag-specific adaptive immunity, especially CTL activity generated through the help of initially activated innate immunity, including NKT cells and NK cells.
MDSCs are known to foster immunosuppressive environments that dampen tumor rejection by mediating the development of Treg cells through secretion of IL-10 and TGF-β (15) and by suppressing NK or T cell functions. Monocytes, which are known to develop from common myeloid precursors, give rise to macrophages or DCs according to their surrounding cytokines (29, 30). In contrast, MDSCs, which also emerge from common myeloid precursors, accumulate as immature cells in cancer patients and tumor-bearing mice. Because high levels of MDSCs accumulate in the blood of cancer patients, it is easy to obtain a large quantity of the cells. Furthermore, it is difficult to separate a large number of MDSCs during the preparation of monocytes for a DC vaccine. For these practical reasons, we tried to find a way to overcome MDSC-mediated immune suppression and, by extension, to convert MDSCs into immunogenic APCs.
Recently, it was reported that human-activated NKT cells direct monocytes to differentiate into immature DCs (31) and that cytokines produced by the NKT cells, including IL-13 and GM-CSF, could induce differentiation in monocytes. In the current study, we found that the levels of CD11c and CD86 on MDSCs were increased by activated NKT cell-mediated stimulation in naive mice. After transfer of MDSC vaccine into tumor-bearing mice, the increase of CD11c expression on MDSCs was minimal, but the lowered levels of CD11b on MDSCs supported the conversion and maturation of MDSCs even in tumor-bearing mice (32, 33). We further confirmed that αGalCer-loaded MDSC vaccines successfully elicited Ag-specific proliferation of IFN-γ-secreting CD8+ T cells even in tumor-bearing mice and also stimulated CD4+ T cell proliferation without expanding the Treg population, suggesting immunostimulation by αGalCer-loaded MDSCs rather than immunosuppression. Taken together, these findings show that activated NKT cells are able to provide a signal for the differentiation of MDSCs that is sufficient to elicit Her-2/neu-specific antitumor immunity.
αGalCer, the ligand of type I NKT cells, has been widely used as an adjuvant to enhance antitumor immunity, not only as a free form (34) but also as a loaded form onto various CD1d-expressing cells, including DCs (19, 20), B cells (21, 22), or tumor cells (35, 36, 37). Paradoxically, however, NKT cells have also been reported to suppress tumor immunosurveillance (38, 39) rather than induce protective immunity (40, 41) in several tumor models. Their suppressive activity is known to be mediated by either of the following: 1) abrogation of CTL function through increased TGFβ secretion in CD11b+Gr-1+ myeloid cells affected by IL-13-producing CD1d-restricted NKT cells (42); or 2) through an IL-13-independent mechanism (43). In the current study, however, αGalCer-loaded MDSCs specifically activated type I NKT cells and induced the secretion of IFN-γ rather than stimulating immunosuppressive type II NKT cells (44, 45). Type II NKT cells were characterized by TCRs, other than Vα14Jα18 chain composition in the mouse, by reactivity to sulfatide (46) and by a lack of response to αGalCer. Recent studies have shown that type I NKT cells conferred protection against B cell lymphoma, whereas type II NKT cells were suppressive (44, 47). Furthermore, there is a possibility that type II NKT cells are cross-regulated by type I NKT cells in antitumor immunity (44).
In addition to activated NKT cells, activated NK cells might be required to establish effective antitumor immunity by providing the required signal for the successful generation of CTL and innate immunity via interaction with differentiated MDSCs. Activated NK cells have been shown to activate DCs, inducing their maturation and promoting the induction of tumor Ag-specific CD4+ and CD8+ T cell responses (48, 49); they also provide IFN-γ for Th1 priming (50) and kill malignant cells that express low levels of MHC Ags (51), especially in circulating rather than solid tumors (52). Recently, activated DCs have been shown to be necessary and sufficient for the priming of NK cells that required the trans-presentation of IL-15 by DCs to resting NK cells (53). Based on the previous findings, we presume that the induction of NK cell activation generated by activated DC-like MDSCs or NKT cells would be another major contributor to effective antitumor immunity.
We have obtained completely contradictory results for the role of CD4+ T cells in NKT cell-dependent tumor rejection after immunization with cell-based vaccines. When mice were immunized with tumor cells loaded with αGalCer, the antitumor effect was dependent on conventional CD4+ T cells (36, 37). Similarly, tumor extract vaccine with αGalCer showed CD4+ T cell-dependent tumor rejection (54). Previously, however, the CD4+ T cell had been shown to be dispensable in the generation of antitumor effects by B cell-based vaccine loaded with αGalCer and antigenic peptide (21). In the current study, depletion of CD4+ T cells did not weaken the antitumor effect of the MDSC-based vaccine loaded with αGalCer even in cases where tumor Ag was expressed on MDSCs. Because depletion of CD4+ cells might have eliminated CD4+ NKT cells and CD4+CD25+ Treg cells at the same time, the removal of CD4+ Treg cells seemed to compensate for the removal of CD4+ type I NKT cells, eliciting a successful antitumor effect after administration of αGalCer-loaded MDSC vaccine. Alternatively, CD4−NK1.1− or CD4−NK1.1+ NKT cell subsets, which were recently suggested as being a major source of NKT-derived IL-17 (55), might participate mainly in the generation of MDSC vaccine-mediated antitumor immunity. In addition, an alternative CD4+ T cell-independent pathway (by DC-NK cross-talk) was suggested for CTL induction (56). We have checked the minor levels of cross-presentation by host DCs and therefore presume that the CD4-independent antitumor effects of the MDSC vaccine are elicited by the manipulated MDSCs themselves that have up-regulated CD86 after being activated by NKT cells. Furthermore, the CD4-independent antitumor effect of the MDSC vaccine could have important implications for the development of a vaccine for HIV-infected cancer patients with immunodeficiency from a lack of CD4+ T cells.
In the current study, depletion of CD25+ Tregs dramatically increased the antitumor effect by αGalCer-loaded, Ag-presenting MDSC vaccine, showing that the involvement of Treg generation by the active cellular vaccine could evoke severe suppressive effect on the effecter cells. In our previous study, likewise, depletion of CD4+ cells containing CD4+CD25+Foxp3+ Treg increased the cytotoxic activity of CTL by an xenogenic adenovirus vaccine expressing human Her-2/neu as an Ag (25). In this regard, recently intratumoral depletion of regulatory T cells has been shown to potentiate antitumor immunity leading to tumor rejection (57, 58). Although TGF-β, which is abundant in the tumor environment, was shown to be related to the expansion of regulatory T cells (59), we presume that MDSCs can induce the expansion of Treg together with effecter cells, because even matured DCs as well as immature DCs could facilitate the expansion of CD25+CD4+ regulatory T cells (60, 61). Thus, the elimination of unnecessarily expanded Treg cells accompanied by active immunotherapy will be a subject that must be taken into consideration for successful antitumor immunotherapy.
In the current study, we have shown that MDSCs, which were known to enhance tumor suppression and therefore to act as a major obstacle together with Treg cells to the rejection of tumor, could be converted into highly immunogenic APCs by activated NKT cells. Thus, NKT cell-licensed MDSCs could be used as immunogenic APCs to induce antitumor immune responses or in strategies to render MDSCs immunogenic rather than immunosuppressive. As such, they have an important role to play in successful antitumor immunotherapy. Combined with currently used chemotherapy and radiation therapy, approaches using easily obtainable cell populations in cancer patients offer a novel strategy to protect cancer patients against metastases.
We thank Dr. Shimon Sakaguchi (Department of Experimental Pathology, Institute for Frontier Medical Sciences, Kyoto University, Japan) for the gift of anti-CD25 (PC61) mAb.
The authors have no financial conflict of interest.
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.
↵1 This study was supported by a grant from the National Research and Development Program for Cancer Control, Ministry of Health and Welfare, Republic of Korea (0720500-1) and a Korea Science and Engineering Foundation National Research Laboratory Program grant funded by the Ministry of Education, Science and Technology, Republic of Korea (R0A-2008-000-20113-0).
↵2 Address correspondence and reprint requests to Dr. Chang-Yuil Kang, Laboratory of Immunology, Institute of Pharmaceutical Sciences, College of Pharmacy, Seoul National University, Kwanak-Gu, Seoul, Korea. E-mail address:
↵3 Abbreviations used in this paper: DC, dendritic cell; AdHM, adenovirus encoding Her-2/neu encoding extracellular and transmembrane domain; αGalCer, α-galactosylceramide; DTX, diphtheria toxin; MDSC, myeloid-derived suppressor cell; MOI, multiplicity of infection; NKT, NKT (cell); Treg, regulatory T (cell).
- Received July 23, 2008.
- Accepted November 26, 2008.
- Copyright © 2009 by The American Association of Immunologists, Inc.