Invariant natural killer T cells are preserved in patients with glioma and exhibit antitumor lytic activity following dendritic cell-mediated expansion
Article first published online: 3 FEB 2004
Copyright © 2004 Wiley-Liss, Inc.
International Journal of Cancer
Volume 109, Issue 6, pages 893–899, 10 May 2004
How to Cite
Dhodapkar, K. M., Cirignano, B., Chamian, F., Zagzag, D., Miller, D. C., Finlay, J. L. and Steinman, R. M. (2004), Invariant natural killer T cells are preserved in patients with glioma and exhibit antitumor lytic activity following dendritic cell-mediated expansion. Int. J. Cancer, 109: 893–899. doi: 10.1002/ijc.20050
- Issue published online: 11 MAR 2004
- Article first published online: 3 FEB 2004
- Manuscript Accepted: 23 SEP 2003
- Manuscript Revised: 12 AUG 2003
- Manuscript Received: 19 MAR 2003
- National Institutes of Health. Grant Numbers: CA 84512, MO-1 RR-00102
- innate immunity;
Brain tumors carry a poor prognosis, and newer approaches to their therapy are urgently needed. Natural killer T (NKT) cells are distinct innate lymphocytes with antitumor potentials. Defects in NKT cell function have been observed in patients with other forms of cancer. Here we show that both the frequency and interferon-γ-producing function of NKT cells are well preserved in adult patients with glioma (n = 9) and comparable to findings in healthy controls (n = 9). These cells can be readily expanded in culture using autologous mature dendritic cells loaded with the NKT ligand, α-galactosyl ceramide. The expanded NKT cells from glioma patients are functional and, importantly, kill glioma cells in a ligand- and CD1d-dependent manner. Expression of CD1d is detected both on primary glioma cells as well as endothelial cells in infiltrating new blood vessels by immunohistochemistry of glioma tissue sections. These data suggest that targeting NKT cells may provide a novel strategy for immunotherapy of glioma. © 2004 Wiley-Liss, Inc.
Brain tumors remain one of the most challenging of human cancers with limited treatment options and high mortality.1, 2 The prognosis of patients with brain tumors, particularly those with high-grade tumors, remains poor, with a median survival of only 12 months.1 Newer approaches for therapy of these tumors are therefore urgently needed. Several studies have tried to harness the immune system to resist human glioma, the most common primary brain tumor. Most of these studies have focused on the antitumor T-cell immunity. However, it remains challenging to identify methods that yield effector cells for killing human glioma.3
Invariant natural killer T (NKT) cells are distinct innate lymphocytes characterized by a restricted TCR usage (Vα24 Vβ11 in humans) and coexpression of T- and natural killer cell markers.4, 5, 6 These cells recognize glycolipid ligands in the context of antigen-presenting CD1d molecules and respond to a synthetic glycolipid, α-galactosyl ceramide (α-GalCer).7, 8 NKT cells have been postulated to play protective as well as tolerogenic roles in antitumor immunity.9, 10, 11 In mice, NKT cells may play a role in surveillance from carcinogen-induced sarcomas12 and are important effectors in IL-12- and α-GalCer-mediated antitumor immunity.13 In cell cultures from human blood samples, it is feasible to expand functional NKT cells using mature dendritic cells (DCs) pulsed with α-GalCer.14
Recent studies have pointed to a quantitative or qualitative deficiency of NKT function in cultured NKT cells from patients with advanced solid tumors.15, 16 Here, we study patients with primary brain tumors, including high-grade glioma, and show in distinct contrast to these reports that freshly isolated NKT cells from patients with glial tumors are functionally comparable to those in healthy controls and have a Th1 phenotype. NKT cells from these patients can be readily expanded in vitro using mature autologous DCs loaded with NKT ligand, α-GalCer. Importantly, the expanded NKT cells reliably lyse glioma targets. Furthermore, we demonstrate the expression of CD1d by primary tumor cells as well as tumor-infiltrating vessels in vivo by immunohistochemistry. These data suggest that stimulation of NKT cells, particularly via dendritic cells, may be of therapeutic benefit in patients with brain tumors, a tumor with poor prognosis and limited treatment options.
MATERIAL AND METHODS
Peripheral blood was drawn from patients after obtaining institutional review board-approved informed consent. All patients had a histologic diagnosis of primary brain tumor. The tumors were graded according to the WHO criteria for glial tumors. Blood was drawn on all patients prior to initiation of therapy or at least 1 year after completion of therapy in the case of recurrent disease.
Isolation of blood mononuclear cells and generation of dendritic cells
Peripheral blood monocytes (PBMCs) were obtained by density gradient centrifugation using Ficoll-Hypaque (Amersham Pharmacia Biotech, Uppsala, Sweden) and used in immune assays (Elispot, flow cytometry) and for the generation of DCs. DCs were generated from the PBMCs as previously described17 by culturing plastic adherent PBMCs in RPMI-1640 medium with L-glutamine (Mediatech, Herndon, VA) supplemented with 1% single-donor plasma in the presence of gentamicin (20 μg/ml; Bio Whittaker, Walkersville, MD). GMCSF (20 ng/ml; Immunex, Seattle, WA) and IL-4 (25 ng/ml; R&D Systems, Minneapolis, MN) were added to the culture on days 0, 2 and 4. On days 5–6 of culture, the DCs were matured overnight by the addition of the maturation cytokine cocktail consisting of TNF-α (10 ng/ml), IL-6 (1,000 ng/ml), IL-1β (10 ng/ml; all from R&D Systems) and PGE2 (1 μg/ml; Sigma, St. Louis, MO) as previously described.18, 19 In some wells, α-GalCer (100 ng/ml) was added overnight along with the maturation cocktail to load the CD1d molecule on the DCs with this glycolipid ligand for the Vα24 Vβ11 TCR on the NKT cell.
Expansion of NKT cells
Monocyte-derived mature DCs pulsed with α-GalCer were utilized to stimulate expansion of NKT cells in culture as previously described.14 Bulk nonadherent cells (predominantly lymphocytes; 1–2 × 105 cells/well) were cultured with either mature DCs alone or mature DCs loaded with 100 ng/ml α-GalCer at a lymphocyte:DC ratio of 10:1 in 96-well U-bottom plates in 200 μl of RPMI-1640 supplemented with 5% pooled human serum (Labequip, Ontario, Canada). Recombinant IL-2 (50 U/ml; Chiron, Emeryville, CA) was added on days 4 and 7 of culture.
Flow cytometric analyses
Flow cytometry was performed on fresh PBMCs and on day 10–14 after expansion for enumeration of the NKT cells. Fluorescein isothiocynate (FITC)-labeled murine IgG1, antihuman CD4, CD3, phycoerythrin (PE)-labeled IgG2a (isotype for Vβ11), antihuman CD8, anti-CD56, anti-CD25, antihuman interferon-γ (IFN-γ) and Cychrome-labeled antihuman CD3 were purchased from Becton Dickinson (San Jose, CA). FITC-labeled mouse antihuman Vα24 and PE-labeled Vβ11 were purchased from Beckman Coulter (Miami, FL). Flow cytometry was performed using FACS Calibur and data were analyzed using the Cell Quest software.
Single-cell enzyme-linked immunospot (Elispot) assay for determination of T- and NKT cell function in fresh blood
Antigen-specific cytokine-producing NKT or T cells were quantified using an Elispot assay as previously described.14, 17 Briefly, PBMCs (5 × 105) were cultured overnight either alone or in the presence of α-GalCer (100 ng/ml) in Elispot plates precoated with anti-IFN-γ or IL-4 antibody (10 μg/ml; Mabtech, Stockholm, Sweden). As positive controls, we used staphyloccal enterotoxin A (SEA) (50 ng/ml) and influenza virus-specific T-cell responses. For the latter, the PBMCs were infected with influenza virus (multiplicity of infection = 2).17 After overnight culture for 18–20 hr, cytokine-producing spot-forming cells were quantified.17
Evaluation of IFN-γ production by NKT cells
On days 10–14, intracellular cytokine staining assay was performed by stimulating IFN-γ production in 5% PHS alone or in the presence of PMA/ionomycin (50 ng/mL and 1 μM, respectively). GolgiStop (Pharmingen, San Diego, CA) was added at the time of culture to accumulate IFN-γ in the golgi apparatus of the T cells to be examined. After 5 hr of stimulation, the cells were stained with anti-CD3 and Vα24, fixed and permeabilized using Permfix (Pharmingen) and stained with antihuman IFN-γ (Pharmingen) for the detection of intracellular cytokine using the FACS Calibur. The NKT cells were defined as CD3+ and Vα24+ cells.
Cytolytic T-cell assay
Effector cells were added to 51Cr-labeled targets at the respective effector to target ratios in 5% PHS in 96-well round-bottom plates. After 6 hr of incubation, lysis of target cells was measured by a standard 51Cr release assay. For some experiments, CTL targets were pretreated with a CD1d-blocking antibody (30 μg/ml; Pharmingen) or isotype control (IgG1, BD) for 1 hr.
Immunohistochemical detection of CD1d
Fresh-frozen sections of surgically resected tumors were air-dried, fixed with cold acetone and stained using the avidin-biotin peroxidase system. After quenching the endogenous peroxide by incubation with 0.3% hydrogen peroxide, the sections were blocked with 10% horse serum. The sections were then incubated with either mouse antihuman CD1d antibody (Biosource, Camarillo, CA) or mouse IgG1 isotype control (Becton Dickinson) at a concentration of 10 μg/ml for 1 hr at room temperature. Biotinylated horse antimouse antibody 1:200 (Vector labs, Burlingame, CA) was used as secondary antibody followed by development using the avidin-biotin peroxidase staining kit (Vectastain, Vectorlabs, Burlingame, CA). 3-amino-4-ethylcarbazole was used for peroxidase detection and hematoxylin was used as counterstain.
Groups were compared using Student's t-test. Significance was set at p < 0.05.
Blood samples were obtained from 9 patients with histologically proven primary brain tumor, and healthy adult controls (n = 9) following informed consent approved by the institutional review board. Patient characteristics are shown in Table I. Blood was drawn prior to the initiation of chemotherapy, either at the time of initial diagnosis (8 of 9 patients) or at the time of relapse (1 patient). Four patients had a low-grade tumor (WHO grades I and II) and 5 were diagnosed to have a high-grade tumor (WHO grades III and IV). Seven patients had undergone prior subtotal resection of their tumor, while in 2 patients only a biopsy was performed. Only 1 of the 9 patients was receiving steroids at the time of the blood draw.
Responses in fresh PBMCs
First, we quantified T-cell subsets (CD4+ T helper, CD8+ killer T and CD4+CD25+ suppressor T cells) and natural killer cells (CD3−, CD56+) by flow cytometry in fresh blood mononuclear cells from patients with glioma (n = 9) and healthy donors (n = 9). The proportions of CD8+ T cells and natural killer (NK) cells in glioma patients were comparable to those in healthy donors (mean, CD8+: 27% vs. 30%, p = 0.6; NK: 14% vs. 13%, p = 0.8; Fig. 1a). Glioma patients had a lower proportion of CD4+ T cells (mean, 31% vs. 46%; p = 0.01) and higher proportion of CD4+CD25+ cells with a regulatory phenotype (mean, 7.3 vs. 4.5; p = 0.08; Fig. 1a). A reduction in CD4+ T cells in glioma patients was mostly seen in those with high-grade tumors (mean, CD4+ T cells, 29.6% vs. 46%; p = 0.01), and the proportions in patients with low-grade glioma were comparable to those in healthy donors (mean, 39% vs. 46%; p = 0.2). We also measured IFN-γ-producing T cells in response to influenza-infected antigen presenting cells (APCs) and stimulation with SEA using the Elispot assay as previously described.17 Both influenza- and SEA-reactive IFN-γ-producing T cells were again comparable between glioma patients and healthy donors, indicating that T-cell immunity to these 2 positive controls was not impaired in glioma patients (Fig. 1b).
Next, we focused on NKT cells that bear the invariant Vα24 Vβ11 T-cell receptor. Vα24+ Vβ11+ NKT cells were detectable in the peripheral blood of all patients examined (Fig. 2a). The NKT population ranged from 0.01% to 0.2% of the total lymphocyte population in glioma patients (mean ± SD, 0.07% ± 0.06%), which was comparable to the findings in healthy controls (0.08% ± 0.14%; p = 0.78).
To test the functional aspects of NKT cells from glioma patients, we took advantage of the fact that NKT cells respond to stimulation with a synthetic ligand, α-GalCer, with production of cytokines. IFN-γ producers in response to 18–20 hr of α-GalCer stimulation were detectable by an Elispot assay in fresh PBMCs from 8 of 9 patients tested (Fig. 2b). The numbers of α-GalCer-reactive IFN-γ-secreting cells in glioma patients were comparable to those in healthy controls (mean, 41 vs. 55/105 PBMCs; p = 0.7). No α-GalCer-reactive IL-4 producers could be detected by this assay, similar to prior results in healthy donors and confirming prior reports of a Th1 phenotype of freshly isolated human NKT cells.14, 20 Also, the findings in the 2 patients who had undergone biopsy only were not significantly different from those with prior subtotal resection (mean, 31 vs. 44/105 PBMCs; p = 0.7). The number of α-GalCer-reactive cytokine producers in glioma patients correlated with the number of NKT cells (r2 = 0.8) but not NK cells (r2 = 0.17) enumerated by flow cytometry (data not shown), as was also reported previously in healthy donors.14 The α-GalCer-reactive interferon-γ producers were depleted by specific depletion of Vα24 or Vβ11+ cells (data not shown). Thus, functional NKT cells can be quantified in the blood of glioma patients, including those with high-grade brain tumors, and the NKT cells carry a Th1 phenotype similar to healthy controls.
Dendritic cell-mediated expansion of NKT cells
We have recently shown that DCs are superior antigen-presenting cells for the expansion of NKT cells in culture.14 To test if NKT cells from glioma patients could also be expanded in culture, we stimulated nonadherent cells from these patients with autologous mature DCs loaded with α-GalCer, using unpulsed DCs as controls. α-GalCer-loaded mature DCs could successfully expand NKT cells from the blood of all 9 patients tested. In just 2 weeks of culture, we obtained a 100- to 250-fold expansion of NKT cells in 8/9 patients (Fig. 3a and b). In one patient who was receiving steroids at the time of the blood draw, we obtained only a 10-fold expansion at 10 days. After expansion with α-GalCer-loaded DCs, the mean ± SD frequency of NKT cells was 12.5% ± 15% of the total lymphocyte population. The degree of NKT expansion in response to α-GalCer-loaded DCs in PBMCs from glioma patients was thus again comparable to prior findings in healthy donors.14
Recent studies have suggested that CD4+, double-negative, or CD8+ subsets of NKT cells may represent distinct lineages with important functional differences.20, 21 Since we had not seen a quantitative change in the NKT cells, we wanted to study the phenotype of these cells to see whether there was a qualitative difference from healthy donors. In 4/4 patients and healthy controls studied, the predominant NKT cell population was either CD4+ or double-negative (CD4−, CD8−; Fig. 3c). To test the function of these NKT cells expanded by α-GalCer-loaded DCs, the expanded NKT cells were tested for the secretion of IFN-γ in response to stimulation by PMA and ionomycin using intracellular cytokine staining and flow cytometry. The expanded NKT cells were functional and responded with IFN-γ production in response to PMA/ionomycin (Fig. 3d). Thus, NKT cells from glioma patients do not have a proliferative defect in response to stimulation with ligand-pulsed DCs and can be readily expanded to yield functional NKT cells capable of IFN-γ production.
Expression of CD1d by glioma cell lines
To test the cytolytic function of the expanded NKT cells, we first verified that the glioma cell lines expressed antigen-presenting CD1d molecules. Cell surface expression of CD1d by glioma cell lines was monitored by flow cytometry. All cell lines expressed detectable levels of CD1d except A172, where the expression of CD1d was weak/negative (Fig. 4).
Lysis of glioma cells by NKT cells from glioma patients
A major challenge in harnessing antiglioma immunity is to elicit glioma-specific lytic effectors reliably from patients. We therefore looked for cytolysis of CD1d high- or low-expressing glioma cell lines by expanded NKT cells with a standard 51Cr release assay. We used 2 different glioma cell lines, U251, which expresses CD1d, and A172, which is CD1d-negative, as detected by flow cytometry (Fig. 4). We also used the U937 cell line as a positive control, since this target is lysed by NKT cells, and the K562 cell line as a positive control for NK-mediated lysis (data not shown). Following expansion of NKT cells by DC-α-GalCer for 10–14 days, the cultures were restimulated with mature DCs loaded with α-GalCer overnight prior to addition in the 4-hr 51Cr release assay. In 4/4 patients tested, the bulk NKT cells expanded with DC-α-GalCer, but not the control populations expanded using unpulsed DCs, were able to lyse U251 (CD1d+) cells but not A172 (CD1d−) cells (Fig. 5a). The restimulation of the expanded NKT cells with ligand-bearing DCs prior to use in the CTL assays was essential for the detection of glioma-specific lysis (data not shown). Lysis of glioma cells could be inhibited by preincubation of targets with an anti-CD1d-blocking antibody (Fig. 5b), confirming that the lytic mechanism required recognition of CD1d.
NKT cells may kill tumor targets directly or indirectly via stimulation of NK cells.22 To characterize the mechanism of killing of glioma cells by NKT cells, we separated the expanded bulk NKT cell cultures into Vα24+ and Vα24− fractions using magnetic beads. The Vα24+ NKT cells, but not the negative fraction, mediated killing of glioma cells (Fig. 5c). The killing was again blocked by anti-CD1d and was directed to CD1d-expressing U251 cells, and not CD1d-negative A172 cells. Thus, activated NKT cells can directly lyse human glioma cell lines.
Expression of CD1d by glioma cells in vivo
Immunohistochemistry of fresh-frozen tumor tissue demonstrated cytoplasmic expression of the CD1d molecule by the tumor cells. This expression was heterogeneous. Some tumor cells in high-grade glioma particularly express high levels of CD1d (Fig. 6a). Similar cells were not seen in sections of low-grade tumors. Strikingly, there is a very strong expression of the CD1d molecule on the angiogenic vessels that are typically numerous in high-grade glioma (Fig. 6b).
We find that adult patients with both high- or low-grade glioma have functional IFN-γ-producing NKT cells and that these can be readily expanded using α-GalCer-loaded DCs. The expanded NKT cells are likewise functional, secreting IFN-γ and exhibiting strong lytic activity against glioma cells. Our finding that NKT cells are preserved in glioma patients at levels comparable to healthy controls is in contrast to recent reports suggesting a deficiency of IFN-γ production and proliferative potential of NKT cells from patients with advanced cancer.15, 16 However, these studies did not utilize DCs to activate NKT cells in culture.
The reason that NKT cell function is well preserved in glioma while being reduced in the other cancers is worth further exploration. In mice, presentation of a synthetic NKT ligand, α-GalCer, by DCs leads to prolonged activation of NKT cells in vivo, but presentation of this ligand by non-DCs can lead to induction of NKT cell energy.23 Therefore, one possibility is that systemic tumors may promote loss of NKT function by presentation of endogenous tumor-associated glycolipids. The preserved NKT cell function in glioma patients that we observed may reflect the lack of systemic metastases in this tumor and therefore an inability of the tumor cells to anergize NKT cells. The nature of physiologic and tumor-associated ligands that are presented on the CD1d molecule to NKT cells is not known. However, a growing body of data suggests a role for endogenous CD1d ligands and NKT cells in immunopathology in the CNS, particularly in the models of experimental autoimmune encephalitis.24, 25 Further work is needed to characterize more fully the phenotype and heterogeneity of NKT cells in these patients.
The finding that NKT cells from glioma patients can be reliably recruited to kill glioma cells suggests that these effectors may be important for immunotherapy of glioma. NKT cells have also been implicated in resistance to experimental glioma during some immunotherapeutic approaches in mice, such as recombinant vaccine therapy with IL-2 or IL-12 constructs.26 We now show that both tumor cells as well as the angiogenic vessels in the tumor express CD1d. Indeed, recent studies point to antiangiogenesis as one of the mechanisms of action of α-GalCer.27 Although we find direct killing effect of tumor cells by the NKT cells, the latter may also inhibit tumor growth via recruitment of other immune effectors such as the NK cells and CD4+, CD8+ T cells and exert antiangiogenesis effects through the release of IFN-γ or direct lysis.22, 28 Enhancement of antitumor NKT function in patients needs to be considered as a new form of immunotherapy of glioma, either via active immunization using α-GalCer loaded DCs or via adoptive transfer of NKT cells expanded ex vivo using these specialized antigen-presenting cells.14, 29
Supported in part by grants from the National Institutes of Health (CA 84512 to R.M.S. and MO-1 RR-00102 to the Rockefeller General Clinical Research Center).