Murine Vα14+ NKT cells, characterized by expression of a single invariant receptor encoded by the Vα14 and Jα281 gene segments, have been identified as a novel lymphocyte lineage.1, 2, 3, 4, 5 Upon stimulation with a glycolipid, αGalCer, Vα14+ NKT cells assume various functions, including the ability to regulate cytokine-mediated Th1/Th2 differentiation and perforin/granzyme B-mediated antitumor activity.2, 4, 6, 7 In addition, a significant role of Vα14+ NKT cells in tumor surveillance has been suggested.6 Human Vα24+ NKT cells bearing the invariant receptors Vα24JaQ and Vβ11, the counterpart of murine Vα14+ NKT cells, also recognize αGalCer in a CD1d-dependent fashion8 and display potent antitumor activity in vitro.9, 10, 11, 12, 13 A series of promising results in studies of the antitumor effects of activated Vα14+ NKT cells in murine tumor metastasis models14, 15, 16, 17 and the recognition of the same ligand, αGalCer, by human Vα24+ NKT cells9, 10, 11, 12, 13 encouraged us to establish an effective new approach to cancer immunotherapy using the αGalCer/CD1d-Vα24+ NKT cell system as a target.
For clinical trials aimed at Vα24+ NKT cell activation in cancer patients, a simple procedure would be to inject soluble αGalCer i.v. into cancer patients to activate endogenous Vα24+ NKT cells. However, results in murine models indicated that αGalCer treatment was no longer effective when αGalCer treatment started 3 days after melanoma cell injection.14, 17 Moreover, multiple injections of soluble αGalCer induced an anergy state in NKT cells in murine tumor-bearing models18, 19 or a shift from Th1- to Th2-type cytokine production.20, 21 Indeed, a phase I clinical study involving i.v. injection of αGalCer into advanced cancer patients resulted in neither Vα24+ NKT cell expansion nor clinical responses, although feasibility was proved.22 Several investigators, including us, have noted that DCs expressing CD1d present αGalCer efficiently to murine Vα14 and human Vα24 NKT cells. We reported that in a murine tumor-metastasis model, αGalCer-loaded DCs showed potent antitumor activity that eradicated multiple small metastatic nodules generated in the liver.14 It was shown that injection of αGalCer-loaded DCs induced prolonged IFN-γ production in NKT cells.18 The results of a clinical study using IL-4 and GM-CSF–cultured DCs loaded with αGalCer suggested the feasibility of the procedure.23, 24 Thus, inducing the activation and expansion of endogenous Vα24+ NKT cells by αGalCer-loaded DCs would be a promising strategy for cancer immunotherapy.
In tumor immunotherapy using DCs loaded with tumor peptides, plastic-adherent or CD14+ PBMCs were cultured with GM-CSF and IL-4 to cause them to differentiate into DCs,25, 26 which were recently refined for clinical application.27 Since the frequency of Vα24+ NKT cells is very low (<0.3% of PBMCs), very large numbers of CD1d-expressing APCs would be required to induce efficient activation and expansion of endogenous Vα24+ NKT cells. However, it is very difficult to obtain large numbers of moDCs by standard procedures, making an alternative APC preparation method a necessity. In fact, CD1d expression is inducible on human T cells upon activation, although it has not been clarified whether activated CD1d-expressing T cells present αGalCer to human Vα24+ NKT cells.28, 29, 30
The results shown here indicate that αGalCer-loaded, IL-2/GM-CSF-cultured PBMCs induce very efficient expansion of autologous Vα24+ NKT cells, which maintain the ability to produce IFN-γ compared to moDCs. Interestingly, the maturation of DCs induced by CD11c− T cells through TNF-α production appeared to be critical for the efficient activation of Vα24+ NKT cells.
Cell preparation and culture of APCs including IL-2/GM-CSF-cultured PBMCs and moDCs
Written informed consent was obtained from HV1 and HV2 before sampling PBMCs. PBMCs were separated by density gradient centrifugation, washed once with PBS and twice with PBS (or RPMI-1640) supplemented with 3% heat-inactivated FBS and then used for experiments. The frequency of Vα24+ NKT cells in PBMCs was 0.18% in HV1 and 0.008% in HV2. Vα24+ NKT cells were defined as Vα24+Vβ11+ cells.13, 31
For preparing IL-2/GM-CSF-cultured PBMCs, whole PBMCs were cultured for 7 days in the presence of rhIL-2 (100 JRU/ml; Imunace, Shionogi, Japan) and rhGM-CSF (800 U/ml; NCPC Gene Tech Biotechnology Development, Shi jiazhuang, People's Republic of China) in RPMI-1640 supplemented with 10% FBS, 0.01 mM 2-ME and 50 U/ml penicillin-streptomycin. For the preparation of moDCs, whole PBMCs were allowed to adhere to culture flasks for 1.5–2 hr at 37°C, and then adherent cells were cultured for 5–7 days in the presence of rhIL-4 (500 U/ml, R&D Systems, Minneapolis, MN) and rhGM-CSF (800 U/ml). CD11c+ cells were purified by using a MACS separation column (Miltenyi Biotech, Bergisch Gladbach, Germany) according to the manufacturer's protocol. Briefly, moDCs or IL2/GM-CSF-cultured PBMCs were incubated with PE-conjugated anti-CD11c MAb (Pharmingen, La Jolla, CA) for 15 min, washed twice and then incubated with microbeads conjugated with anti-PE MAb (Miltenyi Biotech) for 30 min on ice in PBS containing 3% FBS. Magnetically labeled cells were applied to the auto-MACS apparatus, and trapped cells (the positive-selected fraction) and untrapped cells (the negative-selected fraction) were used as APCs. In the positive-selected fraction, >88% of cells were CD11c+. CD11c− cells were purified from the nonadherent fraction of IL-2/GM-CSF-cultured PBMCs by depleting macrophages and adherent CD11c+ cells using the MACS separation column. CD11c+ cells accounted for <5% of the cells in the negative-selected fraction. The CD11c− cells contained marginal numbers (<1%) of macrophages (CD11b+CD11c−/low cells), NKT cells (Vα24+Vβ11+ cells), myelomonocytic cells (CD14+CD3−) and B cells (CD19+CD3− cells), as well as CD11c+ DCs. Mature CD11c+ moDCs were obtained by culturing immature CD11c+ moDCs with LPS (1,000 ng/ml), cytokine cocktail (IL-1β, 10 ng/ml; TNF-α, 10 ng/ml; and PGE2, 1 μg/ml), PGE2 (5 μg/ml) plus LPS (100 ng/ml) or OK432 (0.1 KE/ml) for 1–3 days.
Expansion and detection of Vα24+ NKT cells
αGalCer (KRN7000) was provided by Kirin Brewery (Gunma, Japan) and prepared as described previously.16 To detect Vα24+ NKT cell expansion, APCs were loaded with αGalCer (100 ng/ml) or vehicle for 12–18 hr at 37°C in a 5% CO2 incubator. Autologous PBMCs (0.2 to 1 × 106) prepared from the same volunteer were cultured for 6 days in the presence of various types of APC. Flow cytometry to detect Vα24+Vβ11+ cells was performed on an EPICS-XL flow cytometer (Beckman Coulter, Fullerton, CA), and the results were analyzed by Flow JO software (Tree Star, Inc., Ashland, OR). Vα24+ NKT cell expansion was quantified by the following formula: NKT fold expansion = (whole live cell count after culture × % of NKT cells after culture)/(whole live cell count before culture × % of NKT cells before culture). For the NKT cell proliferation assay using [3H]-thymidine, in vitro activated Vα24+ NKT cells, prepared by 2 cycles of cultivation with IL-2 and αGalCer for 7 days, were stimulated with irradiated APCs for 72 hr at 37°C, with 0.5 μCi/well of [3H]-thymidine added to the stimulation culture for the final 16 hr. Incorporated radioactivity was measured by a β-plate scintillation counter.
In general, 0.2 to 1 × 106 cells were stained with antibodies according to the standard method described previously.9, 13 The antibodies used were as follows: anti-Vα24-FITC (C15) and anti-Vβ11-PE (C21) (Coulter-Immunotech, Miami, FL) and anti-HLA-DR-FITC, anti-CD4-FITC, anti-CD56-FITC, anti-CD83-FITC, anti-CD11c-PE, anti-CD80-PE, anti-CD8-PE, anti-CD16-PE, anti-CD1d-PE, anti-CD86-Cy and anti-CD3-Cy (Pharmingen). For intracellular staining, anti-Vα24-biotin (C15, Coulter-Immunotech), anti-streptavidin-Cy-Chrome, an FITC-conjugated anti-IFN-γ antibody and a PE-conjugated anti-IL-4 antibody (Pharmingen) were used. For neutralization, anti-TNF-α (MABTNF-A5), anti-IFN-γ (NIB42), anti-IL-4 (MP4-25D2), anti-IL-12 (C8.6), anti-CD1d (CD1d42) or anti-MHC-I (G46-2.6) antibodies were purchased (Pharmingen). Different lymphocyte subsets were defined as follows: CD11c+ DCs, SShighFShighCD11c+ cells; CD4+ T cells, SSlowFSlowCD4+CD8−CD3+ cells; CD8+ T cells, SSlowFSlowCD4−CD8+CD3+ cells; NK cells, SSlowFSlowCD56+CD16+CD3− cells; and NKT cells, SSlowFSlowVα24+Vβ11+CD3+ cells.
ELISAs for detecting cytokines in culture supernatants and ELISpot assays for detecting IFN-γ-producing Vα24+ NKT cells
Concentrations (pg/ml) of IL-4 and IFN-γ in the culture supernatants derived from 2 × 105 cell cultures were determined by ELISA (OPT-EIA Set, Pharmingen). Mean concentrations of cytokines in 3 wells are shown. For ELISpot assays, fresh whole PBMCs (adherent cell-depleted) were incubated with various APC preparations for 14–18 hr in 96-well plates. Cultured cells were washed and transferred into an ELISpot assay kit (Pharmingen) with 96-well filtration plates coated with antihuman IFN-γ capture antibody for 3 hr. After wells were washed extensively with PBS, biotinylated antihuman IFN-γ antibody was added. Two hours later, spots were detected by avidin-biotin-peroxidase complex and aminoethyl carbazole solution. Mean values of the spots in 3 wells are shown.
The Cr-release assay was performed as described.9 NKT-sensitive U937 lymphoma target cells were labeled with 100 μCi sodium chromate (Amersham, Little Chalfont, UK) for 1 hr. Cultured cells containing effector cytotoxic cells were seeded into 96-well round-bottomed plates at the indicated E: T ratios on the 51Cr-labeled target cells (1 × 104). Percent specific 51Cr release was calculated by the following formula: %specific lysis = (sample cpm − spontaneous cpm) × 100/(maximum cpm − spontaneous cpm). Spontaneous cpm was calculated from the supernatant of target cells alone, and maximum release was obtained by adding 1N HCl to target cell suspension. Data are expressed as mean values of triplicate cultures with SD.
Expansion of Vα24+ NKT cells by αGalCer-loaded IL-2/GM-CSF-cultured PBMCs
Whole PBMCs from HV1 or HV2 were cultured for 7 days in the presence of rhIL-2 (0–100 JRU/ml) and/or rhIL-4 (0–500 U/ml) with rhGM-CSF (800 U/ml), loaded with αGalCer (100ng/ml) or vehicle for the last 18 hr, irradiated and then used as stimulators. Freshly prepared autologous PBMCs (2 × 105, adherent cell-depleted) from the same volunteers were cocultured for 6 days in the presence of stimulator cells (1 × 105 or 4 × 104). Flow-cytometric analysis was performed to detect Vα24+Vβ11+ cells. Vα24+ NKT cell expansions of more than 20-fold (in HV1) and 50-fold (in HV2) were detected when αGalCer-loaded PBMCs cultured with IL-2 (100 JRU/ml) and GM-CSF (800 U/ml) were used as stimulators (Fig. 1a). The magnitude of the expansion was decreased when PBMCs were cultured in the presence of IL-4 (125 U/ml and 500 U/ml for HV1, 500 U/ml for HV2). Next, the ability of αGalCer-loaded IL-2/GM-CSF-cultured PBMCs (PBMCs cultured in the presence of 100 JRU/ml of IL-2 and 800 U/ml of GM-CSF) to induce Vα24+ NKT cell expansion was compared to that of CD11c+ moDCs, prepared by cultivating adherent PBMC cells with IL-4 (500 U/ml) and GM-CSF (800 U/ml) for 7 days (Fig. 1b). For both HV1 and HV2, the expansion of Vα24+ NKT cells induced by IL-2/GM-CSF-cultured PBMCs was comparable to that induced by purified CD11c+ moDCs. Concurrently, we assessed the number of IFN-γ-producing cells following stimulation with IL-2/GM-CSF-cultured PBMCs (Fig. 1c). Freshly prepared PBMCs (1 × 105) depleted of CD8+ cells, CD56+ cells and adherent cells were cocultured for 15 hr with αGalCer-loaded, IL-2/GM-CSF-cultured PBMCs or purified CD11c+ moDCs. ELISpot assays were performed to detect IFN-γ-producing cells in the cultures. As shown in Figure 1c, the number of IFN-γ-producing cells in the culture with IL-2/GM-CSF-cultured PBMCs was significantly higher than that in the culture with CD11c+ moDCs.
Characterization of IL-2/GM-CSF-cultured PBMCs
Next, IL-2/GM-CSF-cultured PBMCs and moDCs prepared as in Figure 1b were subjected to flow-cytometric analysis. Figure 2a shows SSC/FSC profiles of the cultured cells (upper) and representative HLD-DR/CD11c profiles of cells present in each gate (a–f): cells of both small and large size (gates a and d), cells of small size containing mostly lymphocytes (gates b and e) and cells of large size containing mostly CD11c+ cells (gates c and f). IL-2/GM-CSF-cultured PBMCs included about 15% large and forward scatter cells (gate c) and >23% (23.8%) phenotypically typical DCs (HLA-DR+ and CD11c+, see HLA-DR/CD11c profile of gate a). Among moDCs, 76% were large and forward scatter cells (gate f). The small cells (gate b) in the IL-2/GM-CSF-cultured PBMCs included only a few (4.9%) HLA-DR+ and CD11c+ cells. Substantial numbers of CD11c− cells among IL-2/GM-CSF-cultured PBMCs expressed low but significant levels of HLA-DR (arrows in gates a and b); however, most large cells (gate c) in IL-2/GM-CSF-cultured PBMCs (>90%) expressed high levels of HLA-DR and CD11c. In contrast, almost 100% of the large cells (gate f) in moDCs expressed high levels of CD11c, but expression of HLA-DR was more heterogenous, with only 36% of high HLA-DR-expressing cells. Concurrently, we determined the expression of CD1d antigen-presenting molecules of αGalCer for Vα24+ NKT cells. Representative CD1d profiles of the cells present in each gate are shown in Figure 2b. In both IL-2/GM-CSF-cultured PBMCs and moDCs, the majority of the cells, regardless of their FSC or SSC, expressed similar levels of CD1d.
Cell surface expression of CD86, CD83 and CD80 on IL-2/GM-CSF-cultured PBMCs
In addition, we assessed the expression of other cell surface marker antigens (CD86, CD83 and CD80) on SSChighFSChigh cells in IL-2/GM-CSF-cultured PBMCs and moDCs from HV1 and HV2. In both HV1 and HV2, SSChighFSChigh cells in IL-2/GM-CSF-cultured PBMCs expressed substantial levels of CD86, CD83 and CD80 but those in moDCs expressed only CD86 and CD80 (Fig. 3a).
Next, whole PBMCs or adherent cells were cultured for 7 days in the presence of IL-2 and GM-CSF or IL-4 and GM-CSF, and the maturation of DCs was induced by stimulation with OK432 for 24 hr. As shown in Figure 3b, expression levels of CD86 on SSChighFSChighCD11c+ cells in IL-2/GM-CSF-cultured PBMCs (upper left) were significantly higher than those in IL-4/GM-CSF-cultured PBMCs (lower left) or immature moDCs (lower right) (compare filled areas in Fig. 3b). Expression of CD86 was significantly higher in whole PBMC cultures than in cultures of purified adherent cells stimulated with IL-2 and GM-CSF (compare profiles upper left and right). In all 4 CD11c+ cell preparations, levels of CD86 were increased and were similar when the cells were stimulated with OK432 (compare dashed lines in Fig. 3b). These results suggest that CD11c+ cells in IL-2/GM-CSF-cultured PBMCs express high levels of CD86 and are in a relatively more mature state, while CD11c+ cells in moDCs need to be stimulated to become mature.
IL-2/GM-CSF-cultured PBMCs induce efficient expansion of Vα24+ NKT cells that produce IFN-γ preferentially
We then tested the ability of whole IL-2/GM-CSF-cultured PBMCs to expand Vα24+ NKT cells and examined their production of cytokines (IL-4 and IFN-γ) in the absence of IL-2 in cultured medium compared to the CD11c+ fraction of IL-2/GM-CSF-cultured PBMCs and moDCs with or without further stimulation by LPS (1,000 ng/ml) and a cytokine cocktail (IL-1β, 10 ng/ml; TNF-α, 10 ng/ml; and PGE2, 1 μg/ml) for 2 days30 or PGE2 (5 μg/ml) plus LPS (100 ng/ml). As shown in Figure 4a, the frequency of Vα24+Vβ11+ Vα24 NKT cells was highest in cultures with whole IL-2/GM-CSF-cultured PBMCs. Slightly lower but substantial levels (approx. 4%) of Vα24NKT cells were induced in the culture with CD11c+ IL-2/GM-CSF-cultured PBMCs and moDCs after 2-day stimulation with LPS (second and third groups in Fig. 4a). Essentially no expansion was induced by nonstimulated immature moDCs (bottom group in Fig. 4a).
To assess the production of cytokines by expanded Vα24+ cells, we performed cytoplasmic cytokine staining with anti-IL-4 and anti-IFN-γ in conjunction with cell surface staining with anti-Vα24 (Fig. 4b). Most of the APCs prepared as indicated induced substantial levels of IFN-γ-producing Vα24+ cells. Immature moDCs (with medium) showed the lowest ability to induce IFN-γ-producing Vα24+ cells. Interestingly, IL-2/GM-CSF-cultured PBMCs induced very low levels (<10%) of IL-4-producing Vα24+ cells in culture compared to the other preparations (Fig. 4b, top group). It is of interest that immature moDCs induced the highest frequency of IL-4-producing Vα24+ cells (>30%). These results suggest that IL-2/GM-CSF-cultured PBMCs possess a potent ability to expand Vα24+ cells that produce substantial levels of IFN-γ and small amounts of IL-4.
CD11c− cells in IL-2/GM-CSF-cultured PBMCs induce expansion of Vα24+Vβ11+ NKT cells
In addition to DCs and macrophages, human activated T cells express CD1d molecules.28, 29, 30 Therefore, we studied whether CD11c− cells themselves induced αGalCer-dependent expansion of Vα24+ NKT cells. We prepared CD11c+ and CD11c− populations from whole IL-2/GM-CSF-cultured PBMCs and compared their abilities to expand Vα24+ NKT cells (Fig. 5a). Whole IL-2/GM-CSF-cultured PBMCs, the CD11c+ and CD11c− fractions of whole IL-2/GM-CSF-cultured PBMCs and mature CD11c+ moDCs stimulated with a cytokine cocktail were pulsed with αGalCer and used as stimulators. Nonadherent fresh PBMCs (2 × 105, adherent cell-depleted) were cultured with stimulators for 6 days in the presence of IL-2. Although levels of Vα24+ NKT cell expansion by CD11c− populations (approx. 2%) were slightly less than those of whole (approx. 3%) or the CD11c+ fraction (approx. 4%) of IL-2/GM-CSF-cultured PBMCs, certain levels of expansion were consistently observed. Next, adherent cell-depleted PBMC responders were cocultured with titrated doses of CD11c− cell stimulator cells. As shown in Figure 5b, experiment 1 (Exp. 1), levels of Vα24+ NKT cell expansion produced by 20 × 103 CD11c− cells were slightly lower than those produced by 20 × 103 CD11c+ cells, while levels became higher when 50 × 103, 100 × 103 and 200 × 103 CD11c− cells were used. We included αGalCer-nonpulsed cells in Exp. 2 to demonstrate αGalCer dependence in the expansion of Vα24+ NKT cells. Next, levels of CD1d expression on CD4+CD3+, CD8+CD3+ and CD3− cells in the CD11c− lymphocyte fraction (SSClowFSClowCD11c−) of IL-2/GM-CSF-cultured PBMCs were compared to those on SSChighFSChighCD11c+ cells in HV1 and HV2 (Fig. 5c). Although the levels were slightly lower, CD4+CD3+, CD8+CD3+ and CD3− lymphocytes expressed certain levels of CD1d on their cell surface. These results suggest that CD11c− cells in IL-2/GM-CSF-cultured PBMCs possess some level of αGalCer antigen-presenting ability to expand Vα24+ NKT cells.
Ability to expand Vα24+Vβ11+ cells in moDCs is enhanced by cocultivation with CD11c− cells in whole IL-2/GM-CSF-cultured PBMCs
Since CD11c+ cells in IL-2/GM-CSF-cultured PBMCs showed potent activity in Vα24 NKT cell expansion at levels comparable to mature moDCs induced by LPS stimulation (Fig. 4a), we next assessed whether CD11c− cells in IL-2/GN-CSF-cultured PBMCs induced maturation of DCs. CD11c+ moDCs were stimulated with LPS, cytokine cocktail (IL-1β, TNF-α and PGE2) or irradiated CD11c− cells derived from IL-2/GM-CSF-cultured PBMCs for 1–3 days. Stimulated CD11c+ moDCs were irradiated and cocultured with nonadherent fresh PBMCs (2 × 105, adherent cell-depleted) containing Vα24+ NKT cells in the presence of IL-2 for 6 days. Numbers of Vα24+Vβ11+ cells in the cultures were assessed by flow cytometry. Data are shown as Vα24+Vβ11+ frequency and fold increase in Figure 6a and b, respectively. By either indicator, when irradiated CD11c− cells in IL-2/GM-CSF-cultured PBMCs were used as stimulators for moDC maturation for 3 days, levels of expansion of Vα24+Vβ11+ cells were equivalent to those in LPS- or cytokine cocktail–stimulated moDCs. One day was enough for LPS or the cytokine cocktail to induce maturation of moDCs, but 3 days were needed for the CD11c− cells to give moDCs the ability to expand NKT cells.
To further examine whether CD11c− cells induced maturation of CD11c+ cells, CD11c+ moDCs were stimulated with various doses of irradiated CD11c− cells derived from IL-2/GM-CSF-cultured PBMCs for 3 days, and then expression of CD86 on CD11c+ cells was assessed. Figure 6c shows a representative result of MFI of CD86 staining on moDCs. As expected, a CD11c− cell dosage-dependent increase in the MFI of CD86 staining on moDCs was detected. Moreover, to identify functional effector molecules that are produced by CD11c− cells and induce DC maturation, various MAbs (5 μg/ml) were added to the culture with CD11c+ moDCs and irradiated CD11c−, and then the cell surface expression of CD86 on CD11c+ moDCs was assessed (Fig. 6d). Among MAbs tested, anti-TNF-α MAb substantially inhibited the induction of CD86 on moDCs. No apparent effects were seen in the groups with MAb specific for other cytokines, including INF-γ, IL-4 and IL-12. These results suggest that CD11c− cells in IL-2/GM-CSF-cultured PBMCs are able to support the maturation of CD11c+ moDCs to induce the expansion of Vα24+ NKT cells and that TNF-α plays important roles in the CD11c− cell–mediated maturation of CD11c+ moDCs.
CD11c− CD3+ T cells produce substantial amounts of TNF-α and are involved in the maturation of DCs
To investigate further which subpopulations of CD11c− cells in IL-2/GM-CSF-cultured PBMCs are involved in moDC maturation, we depleted the PBMC cultures of CD11c−CD56+ cells, CD11c−CD3+ cells or both CD11c−CD56+ and CD11c−CD3+ cells prior to IL-2/GM-CSF cultivation and tested the ability of the cultures to induce CD86 molecules on moDCs. As shown in Figure 7a, upregulation of CD86 was not affected by the depletion of CD56+ cells but was substantially inhibited by the depletion of CD3+ cells. The depletion of both CD56+ cells and CD3+ cells resulted in almost no induction of CD86 on moDCs. These results suggest that CD3+ T cells in PBMCs are important for the maturation of moDCs and that CD56+ NK cells may play some role in the maturation process. Next, we isolated CD11c−CD3+ cells, CD11c−CD3−CD56+ cells and CD3−CD56− cells from IL-2/GM-CSF-cultured PBMCs and assessed their production of TNF-α. A substantial amount of TNF-α was secreted from CD3+ cells. Levels of TNF-α were approximately 1/5 and 1/3 in the CD3−CD56+ and CD3−CD56− cell cultures, respectively (Fig. 7b). These results suggest that CD3+ T cells in IL-2/GM-CSF-cultured PBMCs are the major producer of TNF-α. Next, to examine the requirement for IL-2 in PBMC culture for the induction of CD86 on DCs, PBMCs were cultured with GM-CSF alone, IL-2 alone or IL-2 and GM-CSF and CD86 expression on SSChighFSChigh cells (DC) and CD69 expression on CD3+ cells (T cells) assessed (Fig. 7c). Cultivation with GM-CSF alone generated 56.4% of CD11c+ cells, but CD86 molecules were not induced significantly on DCs (MFI = 16). Cultivation with IL-2 alone induced CD86 molecules on DCs (MFI = 58) but generated only 5.0% of CD11c+ cells. Cultivation with both IL-2 and GM-CSF generated 15.3% of CD11c+ cells with sufficient CD86 induction (MFI = 61). More than 30% of T cells were activated to express CD69 in cultures containing IL-2. These results indicate that IL-2 is required for the maturation of DCs and that GM-CSF is important for the generation of CD11c+ cells. Thus, it is most likely that TNF-α is produced by CD11c− T cells upon activation with IL-2 and then TNF-α induces the maturation of DCs generated by GM-CSF in PBMC culture.
Cytotoxic activity of Vα24+ NKT cells stimulated with αGalCer-loaded IL-2/GM-CSF-cultured PBMCs
We assessed whether Vα24+ NKT cells activated with αGalCer-loaded IL-2/GM-CSF-cultured PBMCs show cytotoxic activity against U937 cells (Fig. 8). Freshly prepared PBMCs from HV1 were cultured with αGalCer-loaded IL-2/GM-CSF-cultured PBMCs for 7 days, activated Vα24+ cells were purified with MACS and their cytotoxic activity on U937 cells was examined with a standard Cr-release assay. Significantly increased cytotoxic activity against NKT cell-sensitive U937 target cells was induced by Vα24+ NKT cells activated with αGalCer-loaded IL-2/GM-CSF-cultured PBMCs compared to those with αGalCer-unloaded IL-2/GM-CSF-cultured PBMCs or Vα24− cells containing activated T cells.
Induction of cytokine production and proliferation of in vitro activated Vα24+ NKT cells by CD11c+ and CD11c− cells in IL-2/GM-CSF-cultured PBMCs
Finally, we examined whether CD11c+ and CD11c− cells exert significant activity to expand in vitro activated Vα24+ NKT cells. PBMCs from HV1 were cultured with IL-2 and αGalCer for 7 days, and the Vα24−positive cells were enriched using a MACS separation column. Enriched Vα24+ NKT cells were subjected to another 7-day stimulation with IL-2 and αGalCer-loaded irradiated fresh PBMCs from HV1. More than 96% of the cells were Vα24+Vβ11+ cells (Fig. 9a). Activated Vα24+ NKT cells were cocultured with irradiated αGalCer-loaded immature moDCs, CD11c+ cells or CD11c− cells prepared from IL-2/GM-CSF-cultured PBMCs. The proliferative responses (Fig. 9b) and production of IFN-γ and IL-4 (Fig. 9c) from in vitro activated Vα24+ NKT cells were examined. Proliferative responses of Vα24+ NKT cells were induced by immature moDCs, CD11c+ cells and CD11c− cells in IL-2/GM-CSF-cultured PBMCs; and levels were highest in immature moDCs. All these responses were significantly inhibited by the presence of an anti-CD1d MAb but not a control anti-MHC class I MAb, suggesting that the proliferation is dependent on CD1d. As can be seen in Figure 9c, CD11c+ cells induced substantial levels of both IFN-γ and IL-4 production in activated Vα24+ NKT cells. Although levels were low, CD11c− cells also induced IFN-γ and IL-4 production in a cell dose-dependent manner. These results suggest that, although levels are lower, both CD11c+ and CD11c− cells in IL-2/GM-CSF-cultured PBMCs are able to stimulate in vitro activated Vα24+ NKT cells in a CD1d-dependent manner.
Here, we demonstrate that IL-2/GM-CSF-cultured PBMCs are able to expand freshly prepared autologous Vα24+ NKT cells very effectively and induce their ability to produce large amounts of IFN-γ and exert significant cytotoxicity to tumor target cells. Especially, CD11c+ cells in IL-2/GM-CSF-cultured PBMCs expressed high levels of CD86 (Fig. 3b), suggesting that these cells were in a relatively more mature state and induced efficient expansion of Vα24+ NKT cells and preferential production of IFN-γ at levels equivalent to those of mature moDCs (Fig. 4a,b). Maturation of CD11c+ cells in IL-2/GM-CSF-cultured PBMCs was induced by CD11c− cells present in culture (Fig. 6a–c). TNF-α appeared to play important roles in the maturation of CD11c+ cells. In addition, although levels were slightly lower, even CD11c− cells expressed CD1d and showed substantial stimulatory activity on freshly prepared Vα24+ NKT cells, resulting in expansion and IFN-γ production (Fig. 5). Thus, IL-2/GM-CSF-cultured PBMCs can be used as potent APCs for αGalCer presentation to naive Vα24+ NKT cells.
The maturation of CD11c+ DCs is well established by cultivation of moDCs with TNF-α alone or a cytokine cocktail including TNF-α, IL1β, PGE2 with or without IL-632 as well as other material such as LPS and CD40-L.27 However, maturation also appears to occur through cell–cell contact, such as with lymphokine-activated lymphocytes or activated NK cells.33 In particular, it has been reported that NK cells can induce maturation of DCs via cell–cell interactions of MHC and NK receptors, as well as secreted cytokines such as IFN-γ and TNF-α.34 In our study, phenotypic maturation of moDCs indeed occurred after cocultivation with CD11c− cells in IL-2/GM-CSF-cultured PBMCs (Fig. 6c). Moreover, we demonstrate that TNF-α is a critical cytokine for the maturation of moDCs by CD11c− cells (Fig. 6d). In the CD11c− fraction of IL-2/GM-CSF-cultured PBMCs, CD11c−CD3+ T cells appeared to play important roles in DC maturation (Fig. 7a). In addition, TNF-α was preferentially produced by CD11c−CD3+ T cells (Fig. 7b). Thus, the maturation of DCs induced by CD11c− T cells through TNF-α production appears to be critical for the efficient expansion and activation of Vα24+ NKT cells to produce IFN-γ preferentially.
In DC therapy with HLA-ABC-restricted antigens, several preclinical investigations and clinical trials using ex vivo generated moDCs succeeded in producing tumor regression in patients with limited malignant tumors.27, 35, 36 Mature DCs are much more effective at activating naive T cells and inducing antigen-specific T-cell responses.37, 38 As for Vα24+ NKT cells, mature moDCs pulsed with αGalCer appear to have a better ability to induce Vα24+ NKT cell expansion than either immature moDCs or monocytes.32 Here, we show that the expansion of Vα24+ NKT cells induced by naturally matured CD11c+ DCs in IL-2/GM-CSF-cultured PBMCs was comparable to that induced by mature CD11c+ moDCs stimulated additionally with LPS or cytokine cocktail.
Another interesting observation is that CD11c− cells in IL-2/GM-CSF-cultured PBMCs imparted a certain stimulatory activity to Vα24+ NKT cells. CD11c− cells were shown to express CD1d (Fig. 2), and the majority (>75%) were T cells (data not shown). CD1d expression has been reported to be inducible on human T cells upon activation.28, 29, 30 However, it has not been clarified whether activated CD1d-expressing T cells can successfully present αGalCer to human Vα24+ NKT cells. We characterized CD11c− cells in IL-2/GM-CSF-cultured PBMCs very carefully to determine their ability to present αGalCer to human Vα24+ NKT cells, causing them to proliferate and produce cytokines (Fig. 5), and concluded that CD11c− cells induced activation and expansion of Vα24+ NKT cells, particularly freshly prepared Vα24+ NKT cells.
There are many reports in the literature about the function of DCs in the regulation of T cell–mediated immune responses. The nature of the regulation has become more complicated due to the discovery of activation status and ontogenetically diverse subtypes of DC in murine models,27, 37 as well as the results of in vitro studies using human PBMCs. For instance, DCs cultured under some conditions, including the presence of IL-10, TGF-β, or steroids and at low DC/T cell ratios, induce naive CD4 T cells to differentiate into Th2 cells.39, 40 However, it is not known how the balance of Th1/Th2 cytokine production in Vα24+ NKT cells is regulated by human αGalCer-loaded APCs. Here, we examined the Th1/Th2 cytokine profiles of Vα24+ NKT cells activated by various αGalCer-loaded APCs and found that whole IL-2/GM-CSF-cultured PBMCs and CD11c+ cells in IL-2/GM-CSF-cultured PBMCs induced preferential IFN-γ production in Vα24+ NKT cells. The preferential IFN-γ production was superior or equal to that of mature moDCs. Moreover, the separation process of nonadherent cells is not required for the preparation of IL-2/GM-CSF-cultured PBMCs. In addition, there is no loss in cell number during preparation. Thus, IL-2/GM-CSF-cultured PBMCs represent an alternative potent material for tumor immunotherapy aimed at the in vivo activation and expansion of Vα24+ NKT cells.
It has been reported that the number and/or function of Vα24+ NKT cells in PBMCs from patients with malignant tumors is decreased compared to PBMCs from healthy volunteers and the amount of the decrease appears to be dependent on tumor type and clinical stage.13, 41, 42, 43, 44 We reported that the number of Vα24+ NKT cells among PBMCs is reduced in patients with lung cancer but their ability to produce IFN-γ remains intact.13 In patients with advanced gastrointestinal cancers, the in vitro expansion of Vα24+ NKT cells was impaired, although it recovered partially by coculture with G-CSF.44 In patients with progressive malignant multiple myelomas, freshly prepared Vα24+ NKT cells did not produce IFN-γ even after stimulation with αGalCer-loaded moDCs, although the dysfunction was considered to be reversible because functional Vα24+ NKT cells could be expanded after in vitro culture.41 Thus, it is speculated that αGalCer-loaded APC therapy will be useful for treating patients with early-stage malignancies who show no severe dysfunction of their Vα24+ NKT cells; patients with advanced cancers may need to receive additional supportive therapy, such as adoptive administration of functional Vα24+ NKT cells, prior to αGalCer-loaded APC therapy.
In summary, our results indicate that maturation of DCs induced by CD11c− T cells through TNF-α production appears to be critical for the efficient activation of Vα24+ NKT cells by αGalCer to produce IFN-γ preferentially. In addition, activated Vα24+ NKT cells with IL-2/GM-CSF-cultured PBMCs showed significant cytotoxic activity against tumor cells in vitro. Thus, use of IL-2/GM-CSF-cultured PBMCs is appropriate for tumor immunotherapy aimed at Vα24+ NKT cell activation and expansion.
We thank Kirin Brewery for providing clinical grade αGalCer (KRN7000) for these studies and Ms. K. Sugaya and Ms. A. Nitta for excellent technical assistance.