Abstract
It has been suggested that dendritic cells (DCs) are capable of ingesting apoptotic tumor cells (ATCs) and presenting tumor-associated antigens to immune cells. We evaluated the potential of human DCs, which have ingested ATCs, to serve as a source of antigenic epitopes for presentation to T cells specific for PCI-13, a squamous cell carcinoma of the head and neck cell line. Immature DCs (DCimm) generated in the presence of interleukin 4 and granulocyte machrophage colony-stimulating factor from peripheral blood monocytes of HLA-A2+ healthy donors were incubated in the presence of ATCs. Uptake of ATCs by DCs was monitored by flow cytometry and confocal microscopy after 2–18 h of coincubation. When DCs were matured (DCmat) in the presence of proinflammatory cytokines, their capacity to uptake ATCs was reduced. Responses of PCI-13-specific CD8+ T cells to unmodified PCI-13 cells and to DCimm or DCmat +/− ATCs or +/− tumor lysates were tested in γ-IFN enzyme-linked immunospot and cytotoxicity assays. Unmodified tumor cells were found to be the best stimulators of antitumor activity of the established T-cell line, and ATCs alone were minimally stimulatory. However, DCs that ingested ATCs were able to present tumor antigens to CTLs, and DCimm and DCmat were almost equally stimulatory. When DCs plus various tumor-derived preparations were used as antigen-presenting cells with autologous HLA-A2+ T cells obtained from normal donors, DCs that had ingested ATCs were more effective in generating CD8+ CTLs than tumor cells alone or DCs pulsed with tumor lysates. The results indicate that human DCs fed with ATCs and then matured effectively generated T cell-mediated antitumor responses in vitro.
INTRODUCTION
CTLs 4 are a critical component of the immune response to human tumors, and induction of strong CTL responses is the goal of most current cancer vaccine strategies. CTL target tumors through the recognition of a self-MHC class I molecule and an antigenic peptide generally derived from endogenous tumor cell proteins. However, for CTL induction and expansion, the antigenic peptide has to be presented to precursor T cells in the context of costimulatory molecules usually provided by professional APCs. Delivery of exogenous antigens to the endogenous MHC class I-restricted processing pathway in professional APCs is a critical challenge in cancer vaccine designs. DCs are potent APCs, which can uptake exogenous proteins. Peptides generated from these proteins are cross-presented by DCs on class I MHC molecules to T cells and, on successful T-cell receptor-mediated recognition, can induce antigen-specific CTL responses (1, 2, 3, 4, 5) . However, priming with DCs for generation of tumor-specific responses usually requires prior definition of tumor-derived antigens and characterization of the epitopes involved. With the exception of melanoma and renal cell carcinoma, however, few TAAs have been defined and cloned thus far (6) . In the majority of human cancers, where the TAAs are unknown, DCs coincubated or pulsed with tumor cells or various tumor-derived preparations could be used as vaccines. This type of strategy, broadly referred to as cross-priming, could potentially result in polyvalent immunization of the host to multiple (unknown) TAAs.
Recently, it has been demonstrated that human DCs can acquire viral antigens from apoptotic cells and stimulate antigen-specific MHC class I-restricted CD8+ T cells to mediate antiviral CTL responses (7 , 8) . In this and other in vitro models (9, 10, 11) , apoptotic death was a critical trigger for the antigen processing pathway, and apoptotic cells were a preferred source of antigen, because antigens derived from necrotic cells were not presented on the MHC class I molecules. In addition, the use of ATCs fed to DCs led to effective priming of tumor-specific CTLs in several recent in vivo animal studies (11 , 12) .
In the present study, the ability of human DCs coincubated with ATCs or tumor cell lysates to stimulate tumor-specific T cells was evaluated in an in vitro TAA presentation model. We compared monocyte-derived DCimm and DCmat for their capacity to phagocytose apoptotic SCCHN and cross-present TAAs to tumor-specific CTLs. In a separate in vitro model of TAA cross-priming, we observed that DCs fed with apoptotic SCCHN were capable of cross-priming naïve T cells obtained from HLA-A2-matched healthy donors for tumor-specific responses.
MATERIALS AND METHODS
Cells and Cell Culture.
The HLA-A2+ PCI-13 cell line was established in our laboratory from freshly harvested squamous cell carcinoma of the retromolar trigone and characterized by Heo et al. (13) . Tumor cells were cultured in plastic culture flasks (Costar, Cambridge, CA) under standard conditions (37°C, 5% CO2 in a fully humidified atmosphere) using serum-free AIM-V medium (Life Technologies, Inc., Grand Island, NY). For subculturing cells were detached from plastic using 0.05% trypsin/0.02% EDTA solution (Life Technologies, Inc.). The cultures were routinely tested and found to be free of Mycoplasma contamination (GEN-PROBE, San Diego, CA).
Human DCs were generated according to a modified method by Sallusto and Lanzavecchia (14) . Briefly, peripheral blood or a leukapheresis product was obtained from HLA-A2+ normal donors, and PBMCs were isolated by sedimentation over Ficoll-Hypaque (Amersham Pharmacia Biotech, Piscataway, NJ). The PBMCs were incubated for 1 h at 37°C in AIM-V medium, and nonadherent cells were removed by gentle washing with warm medium. The remaining (adherent) cells were incubated in AIM-V medium + 1000 units/ml granulocyte macrophage colony-stimulating factor (Immunex, Seattle, WA) and IL-4 (Schering Plough, Kennilworth, NJ). The cultures were supplemented with additional cytokines on day 4 of culture. DCs were harvested at day 6 using cold Hanks’ solution (Life Technologies, Inc.).
The PCI-13-specific CTL bulk cell line was established from peripheral blood lymphocytes of a patient with SCCHN, as described previously (15) . The CTLs were thawed and maintained in the presence of IL-2 and IL-4 with repeated sensitization on tumor cell monolayers. The CD8+ T cell line was derived from the original bulk T cell line by negative selection, using anti-CD4 Ab-coated magnetic beads. The CD8+ cells recognized a shared antigen on SCCHN: they lysed autologous SCCHN targets as well as HLA-A2+ allogeneic (but not HLA-A2−) SCCHN targets, including PCI-13 (16) . This lysis was blocked with anti-CD3, anti-CD8, anti-TCR α/β, and anticlass I MHC Abs (w6/32) as well as anti HLA-A2 Abs. The CTL line or clones derived from it did not lyse K562 or Daudi targets, normal tissue cells, or HLA-A2+ phytohemagglutinin-stimulated T cells. For the experiments described here, the CD8+ T cell line was cultured in AIM-V containing 10% FCS and 300 IU IL-2/IL-4. It was stimulated twice with γ-irradiated PCI-13 cells (10,000 rad) and incubated for 7 days before being used in 51Cr-release assays or ELISPOT assays.
Cytokines and Antibodies.
The following cytokines were used for cell cultures: IL-1β (National Cancer Institute, Biological Resources Branch, Frederick, MD), IL-2 (Chiron-Cetus, Emeryville, CA), IL-4 (Schering Plough), IL-6 (Sandoz, Basle, Switzerland), PGE2 (Sigma Chemical Co., St. Louis, MO), IFN-γ (Genentech, San Francisco, CA), granulocyte macrophage colony-stimulating factor (Immunex), and TNF-α (Knoll Pharmaceuticals, Whippany, NJ).
The antibodies used for staining of cells or blocking of responses were either unlabeled or labeled with PE or FITC and included: anti-MHC class I mAbs (HB95; w6/32), as well as anti-HLA-A2 mAbs (BB7.2) obtained from Dr. Albert DeLeo (University of Pittsburg Cancer Institute); anti-MHC class II, anti-CD14, anti-CD25, and anti-CD80 (Becton Dickinson, San Jose, CA); anti-CD40 and anti-CD86 (Ancell, Bayport, MN), anti-CD83 mAbs (Immunotech, Marseille, France) and respective IgG isotype controls (either from Becton Dickinson or PharMingen, San Diego, CA).
DCs, lymphocytes, or tumor cells (2 × 105/200 μl) were incubated with mAbs on ice for 30 min and washed twice in PBS containing 0.1% (w/v) BSA and 0.1% (w/v) NaN3. After staining, the cells were fixed with 1% (w/v) paraformaldehyde in PBS for 30 min at room temperature prior to flow cytometry. Flow cytometry analysis was performed as described previously (17) , using a FACScan (Becton Dickinson) equipped with a single 488-nm argon ion laser. At least 10,000 events were acquired for each sample.
Apoptosis Induction and Detection.
PCI-13 cells cultured in AIM-V were irradiated with 1500μ W/cm2 UVB (UVB bulb BLE-GT 302; Spectronics Corp., Westbury, NY) for 2 min or 15 min. To minimize the UVB absorbing effect of phenol red in AIM-V, the medium level was reduced to a minimum during irradiation. Apoptosis was detected by DiOC6 staining and in the TUNEL assay. Additionally, apoptotic bodies were stained by propidium iodide (10μ g/ml; Sigma Chemical Co.) for 15 min at room temperature after cell membrane permeabilization and examined by confocal microscopy, as described below.
For DiOC6 staining, aliquots (50 nm) of the lipophilic cationic fluorochrome DiOC6 (Molecular Probes, Eugene, OR) were added to 5 × 105 cells/ml of culture medium and incubated for 15 min at room temperature. In apoptotic cells, mitochondria show a decrease in green fluorescence intensity, which is quantitated by flow cytometry (18) . For the TUNEL assay, tumor cells were fixed with 2% (w/v) paraformaldehyde in PBS and permeabilized with 0.1% (w/v) sodium citrate in PBS containing 0.1% (w/v) Triton X-100 for 7 min on ice. After washing, cells were incubated with FITC-conjugated dUTP in the presence of terminal deoxynucleotidyl transferase enzyme solution for 1 h at 37°C, using reagents purchased from Boehringer Mannheim (Indianapolis, IN). After incubation, the cells were washed, and 10,000 events were acquired and analyzed by flow cytometry. Negative controls included cells incubated without the enzyme in the labeling buffer, and positive controls included the same cells treated with DNase (Sigma Chemical Co.).
Tumor Lysates.
Lysates were produced by exposing tumor cells to four rapid freeze-and-thaw cycles until the cell membrane integrity was lost. Cell debris was removed by centrifugation (30 min at 15,000 × g), and the protein content was measured by a Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA). Aliquots of the lysate (1 mg/ml) were used to pulse DCs.
Tumor Uptake by DCs.
To study their uptake by DCs, tumor cells were stained green with 2 μg/ml DiOC16 (Molecular Probes) for 30 min at 37°C in PBS and washed three times in medium before induction of apoptosis. After a 12–24-h incubation in medium to allow for the tumor cells to undergo apoptosis, they were cocultured with DCs at various DC:tumor cell ratios. The cells were harvested 2–18 h later, and DCs were stained with PE-labeled anti-CD80 Ab. Two-color flow cytometry was performed to determine the percentage of cells that phagocytosed apoptotic SCCHN, based on the number of double-positive cells (green/red). The same experiments were also performed at 4°C to show that the uptake of tumor cells by DCs was inhibited at low temperatures.
To prepare cells for confocal microscopy, sterilized glass coverslips were placed on the bottom of a 6-well plate. DiOC16-stained PCI-13 cells were added to these wells and exposed to UVB light as described above. DCs were added 12–24 h after induction of apoptosis. After overnight coculture, the glass coverslips were removed and washed with PBS. The DCs attached to glass were stained with anti-CD80 Ab in combination with a secondary Cy3-conjugated rabbit antimouse Ab (Jackson ImmunoResearch Laboratories, West Grove, PA). After fixation with 1% (w/v) paraformaldehyde, coverslips were mounted on a slide and analyzed by confocal laser scanning microscopy at ×600 original magnification (Leica TCS NT confocal LSM; Leica Lasertechnik, Heidelberg, Germany). Images were edited using the Adobe Photoshop software program (Adobe Systems, Mountain View, CA).
Maturation of DCs.
Maturation of DCs was induced by the addition of proinflammatory cytokines (10 ng/ml IL-1β, 1000 units/ml IL-6, 10 ng/ml TNF-α, and 1 μg/ml PGE2), as described previously (19) . Changes in expression of MHC class I and II molecules as well as CD14, CD25, CD40, CD80, and CD86 on DCs were monitored by flow cytometry, and the level of expression is shown as mean fluorescence intensity for DCimm and DCmat.
Processing and Cross-Presentation of Tumor-derived Epitopes.
DCimm and DCmat that had phagocytosed ATCs or DCs pulsed with tumor lysates, as described above, were harvested, washed, and counted. To determine the ability of these DCs to process and cross-present tumor-derived epitopes to the PCI-13-specific CD8+ CTL line and, thus, be recognized by the CTLs, the DCs were used as stimulators in 24-h ELISPOT assays for IFN-γ production or as targets in 4 h 51Cr-release cytotoxicity assays.
IFN-γ ELISPOT Assay.
The ELISPOT assay was performed as described elsewhere (20) . Briefly, wells of 96-well plates with nitrocellulose membrane inserts (Millipore, Bedford, MA) were coated with 50 μl of primary Ab solution [10 μg/ml in 1× PBS (pH 7.4), clone MAB1-D1K; Mabtech, Nacka, Sweden] and incubated for 24 h at 4°C. Then, the plates were washed four times with PBS, and a 100-μl aliquot of AIM-V supplemented with 10% (w/v) human serum was added for 1–3 h to block nonspecific binding. Next, 1 × 104 to 2 × 104 responder T cells with an equal number of stimulator cells (PCI-13 or DC) were added in a final volume of 200μ l of AIM-V medium. The assay was performed in quadruplicate wells for each experimental condition. The plates were then incubated in a humidified atmosphere of 5% CO2 in air at 37°C for 24 h. After the incubation period, cells were removed by washing the plates six times with 0.05% (w/v) Tween 20 in PBS (Fisher Scientific, Pittsburgh, PA). A 50-μl aliquot of biotinylated secondary anti-IFN-γ Ab (2 μg/ml, clone Mab7-B6–1; Mabtech) was added to each well. The plates were again incubated in a humidified atmosphere of 5% CO2 in air at 37°C for 2 h. The washing steps were repeated, and after a 1-h incubation at room temperature with the avidin-peroxidase complex reagent (Vectastain Elite Standard ABC-Kit; Vector Laboratories, Burlingame, CA), the plates were washed again three times with PBS/0.05% Tween and then three times with PBS alone. Aliquots (100 μl) of the aminoethylcarbazole staining solution (Sigma Chemical Co.) were added to each well to develop the spots. The reaction was stopped after 4–6 min under running tap water. The spots were counted by computer-assisted image analysis (Zeiss ELISPOT 4.14.3.; Zeiss, Jena, Germany). If the mean number of spots against DCs plus tumor preparation (experimental values) was significantly different from the mean number of spots against nonpulsed DCs (background values), as determined by a two-tailed Wilcoxon rank sum test, the background values were subtracted from the experimental values.
For Ab-blocking experiments, PCI-13 cells were preincubated with w6/32 Ab, anti-HLA-A2 (BB7.2) Ab or anti-HLA-DR Ab (clone L 243; kindly provided by Dr. Albert DeLeo), or purified mouse IgG1 (clone S1-68.1; PharMingen) for 30 min. As a control for the assay reproducibility, PBMCs obtained from the same normal donor and cryopreserved in a series of vials were used each time the assay was performed. These control cells were thawed, washed, and used at the concentration of 2 × 104/ml in AIM-V in the ELISPOT assay. The control PBMCs were stimulated with phorbol 12-myristate 13-acetate (1 ng/ml) and ionomycin (1μ m), both from Sigma Chemical Co. The coefficient of variation for this assay was determined as 15% based on 30 independent determinations.
Cytotoxicity Assay.
The 4-h 51Cr-release assay was performed at four
E:T ratios, as described previously
(21)
. Briefly, targets
(PCI-13, K562 or DC +/− different tumor preparations) were labeled
with 51Cr for 45 min at 37°C, washed, and added
to wells of 96-well plates (1 × 104 cells/well). Effector T cells were then added
to give various E:T ratios. The assays were performed in triplicate.
The percentage of specific lysis was calculated according to the
formula:
Enrichment of CD8+ Cells.
Cultured PBMCs or the bulk CTL line were enriched for CD8+ cells by positive immunoselection, using magnetic beads (MiniMacs; Miltenyi Biotec, Auburn, CA) according to the manufacturer’s recommendations. The purity of selected CD8+ cell fractions was checked by flow cytometry.
Cross-Priming of T Cells.
PBMCs were obtained as leukapheresis products from normal HLA-A2+ donors, and monocytes were separated by adherence to plastic. The adherent cells were used for DC generation, whereas the recovered lymphocytes were stimulated with autologous DCs, which have ingested ATCs at the ratio of 10:1. The lymphocytes were cultured in AIM-V medium + 10% human serum supplemented with 25 ng/ml IL-7 for the first 72 h and then in AIM-V supplemented additionally with 20 IU/ml IL-2 for the remaining time in culture. The lymphocytes were restimulated after the first week and weekly thereafter for up to four total stimulations. Responses of T cells to PCI-13, PCI-13 + w6/32 Ab, PCI-13 + HLA-A2.1, PCI-13 + anti-HLA-DR, and PCI-13 + IgG or the controls HR (HLA-A2+ gastric carcinoma), Fem-X (HLA-A2+ melanoma), and HLA-A2+ normal human fibroblasts were tested in 24-h ELISPOT and cytotoxicity assays.
Statistical Analysis.
A two-tailed Wilcoxon rank sum test was performed to analyze ELISPOT data. Unpaired two-tailed Student’s t test was used for statistical analysis of flow cytometry data.
Differences were considered significant when P was <0.05.
RESULTS
Uptake of ATCs by Human DCs.
For induction of apoptosis, PCI-13 cells were treated with UVB light for various periods of time. Apoptosis was already induced after 2 min of UVB exposure but was more pronounced using UVB for 15 min, as indicated by reduced DiOC6 staining, reflecting a decrease in mitochondrial transmembrane potential (Fig. 1A) ⇓ or by increased TUNEL reactivity, which measures DNA fragmentation (Fig. 1B) ⇓ . To allow for apoptosis to progress, UVB-treated tumor cells were incubated for 4–48 h before flow cytometry analysis. Optimal effects were observed following a 24–48-h incubation of UVB-treated PCI-13 cells (Fig. 1) ⇓ .
UVB treatment results in apoptotic death of PCI-13 cells. Apoptosis was induced by a 15-min treatment with UVB (1500μ W/cm2) and measured by flow cytometry after a 4-h, 24-h, or 48-h incubation of tumor cells in medium. A, DiOC6-stained tumor cells show a decrease in the mitochondrial transmembrane potential during the apoptotic process, as evidenced by reduced fluorescence (FL1, X axis). In a TUNEL assay (B), FITC-labeled nucleotides are incorporated into DNA of apoptotic cells, leading to an enhancement in fluorescence (FL1, X axis).
To obtain additional evidence for apoptosis at 24 h after exposure to UVB light for 15 min, PCI-13 cells were permeabilized, stained with propidium iodide, and examined in a confocal microscope. They showed typical morphological features of apoptosis: chromatin condensation and fragmentation of cell nuclei into apoptotic bodies (data not shown). UVB irradiation induced apoptosis in nearly all tumor cells.
We next coincubated apoptotic PCI-13 cells with DCimm or DCmat for various periods of time to determine the optimal conditions for internalization of ATCs. We observed that DCimm ingested ATCs already after 2 h of coincubation (data not shown). Overnight coculture of viable PCI-13 cells with DCs resulted in a significant increase in the double-stained cell population (27%), as shown in Fig. 2 ⇓ . This proportion of double-stained DCs was further increased to 69% when DCs were cocultured with ATCs. After a 24-h coincubation, no“ free” ATCs could be detected (Fig. 2F) ⇓ .
Monocyte-derived DCs efficiently engulf ATCs. DCs, labeled with PE-conjugated anti-CD80 Ab (A), and PCI-13 tumor cells, stained with DiOC16 (B), were analyzed by flow cytometry. A brief coincubation of DCs with ATCs at the 2:1 ratio resulted in two distinct populations (C). Overnight coculture of DCs with alive PCI-13 cells led to internalization of tumor cells in 27% of DCs (D). The number of double-positive DCs increased up to 69%, when ATCs [PCI-13 cells treated either for 2 min with UVB (E) or for 15 min with UVB (F)] were used. The higher proportion of double-positive DCs was accompanied by a decreased proportion of tumor cells, additionally suggesting an active uptake of tumor cells by DCs.
Several findings indicated an active uptake rather than a tight association of DCs with PCI-13 cells: (a) the remaining lymphocytes in the DC culture did not show a shift to green fluorescence (FL1, X axis); (b) the green fluorescence of DCs was always lower than that of tumor cells alone, presumably because the green color of tumor cells inside DCs was absorbed by the DC membrane; (c) the process of uptake was substantially inhibited at 4°C, a temperature that blocks phagocytosis (data not shown); and (d) by confocal microscopy, we were able to confirm that ATCs were internalized by DCs (Fig. 3) ⇓ .
Uptake of apoptotic PCI-13 cells by DCs. Tumor cells were stained with DiOC16 (green) before induction of apoptosis and cocultured with DCs on glass coverslips. After an overnight incubation, attached cells were stained with an anti-CD 80 Ab in combination with a rabbit antimouse Cy3-conjugated secondary Ab (red). Apoptotic PCI-13 cells (green) were detected inside DCs (red) by confocal microscopy (midplane image). Bar, 10 μm.
In a separate series of experiments, DCmat were generated by incubating DCimm in the presence of proinflammatory cytokines for 36 h. This led to up-regulation of various molecules on DCs, including HLA-DR and CD83 as well as CD80 or CD86, as shown in Table 1 ⇓ .
Effects of proinflammatory cytokines on the maturation of monocyte-derived human DCsa
Effects of DC maturation on the ability to uptake ATCs were next examined. When DCs were treated with the cytokine mixture at the beginning of coculture (0 h) with ATCs, the uptake of tumor material was slightly increased, judging by the percentage of double-positive DCs in these cocultures (78%). However, at 8–12 h after the start of DC maturation, the proportion of DCs ingesting ATCs was significantly and consistently reduced by about 50%. Because of this observation, the following procedure was selected to optimize the ability of DCs to present ATC-derived material to T cells: DCimm were cocultured with ATCs overnight to guarantee sufficient uptake, and afterward they were matured for 36 h by the addition of the cytokine mixture prior to ELISPOT and cytotoxicity assays.
Cross-Presentation by DCs of TAAs to PCI-13-specific CD8+ T Cells.
To determine whether human DCs ingesting ATCs are able to process and present tumor-derived epitopes to T cells, we used an in vitro antigen-presentation model, consisting of a tumor-specific CTL line and monocyte-derived DCs. The semi-allogeneic, HLA-A2-restriced and PCI-13-specific CTL line was generated as described previously by us (15 , 16) , and cryopreserved T cells were thawed, maintained in culture, and restimulated with PCI-13 tumor cells at weekly intervals to expand the cells, as needed. The characteristics of this CTL line were extensively evaluated and described before (16 , 22) . Before their use as responders in cross-presentation experiments with DCs, the expanded CTLs were tested in 4-h 51Cr-release assays against PCI-13 as well as K562 targets. As shown in Fig. 4 ⇓ , the CTLs were HLA-A2 restricted, and they efficiently killed HLA-A2+ PCI-13 but not K562 targets. Not shown are results demonstrating that the CTLs did not lyse HLA-A2+ T cell blasts, HLA-A2+ tumor targets that were not SCCHN or HLA-A2− SCCHN targets.
Cytotoxicity of the CTL line against PCI-13 and K562 cells as control for nonspecific lysis. Anti-MHC-class I Ab and anti-HLA-A2 Ab were used in blocking experiments to confirm that the CTLs are HLA-A2 restricted. A representative experiment of four performed is shown.
In cross-presentation experiments, an ELISPOT assay was used as the read-out. PCI-13-specific CD8+ T cells were used as responders, and viable PCI-13 cells, ATCs alone, DCimm+ ATCs, or DCmat+ATCs were used as stimulators. As shown in Table 2 ⇓ , the highest number of IFN-γ spots was always observed with viable PCI-13 cells (P < 0.05), indicating that tumor cells were the best stimulators. This finding was confirmed in the cytotoxicity assays performed in parallel with ELISPOT assays (data not shown). This was not surprising because the CTL line was generated by coincubation with PCI-13 and had been repeatedly stimulated by irradiated PCI-13 cells. ATCs alone were poor stimulators of CTLs, possibly because these ATCs expressed significantly lower levels of MHC class I molecules than PCI-13 tumor cells (mean fluorescence intensity of 77 for viable PCI-13 cells versus 15 for ATCs and 10 for IgG control). DCs ingesting ATCs or pulsed with tumor lysates were recognized better than ATCs alone, and DCmat + ATCs tended to be somewhat more stimulatory than DCimm, although the differences between experiments using different DC preparations were not statistically significant (Table 2) ⇓ .
Responses of the PCI-13-specific CTLs to stimulation with the tumor, ATCs or ATCs as well as tumor lysates, presented by mature or immature human DCsa
Cross-Priming with DCs to Generate Tumor-specific CTLs.
To demonstrate that cross-priming of naïve T cells with DCs, which have ingested ATCs, results in the generation of highly responsive antitumor CTLs, we obtained PBMCs from HLA-A2+ normal donors (n = 4) and used them as a source of both DCs and T cells. Priming of T cells (two to four stimulations) was performed using γ-irradiated PCI-13 cells, ATCs alone, DCs + ATCs, DCs + nonapoptotic PCI-13, DCs pulsed with tumor lysates, or DCs alone as a control. Various priming conditions +/− DCs were compared for the best generation of tumor-reactive T cells, as detectable in ELISPOT assays. The data in Table 3 ⇓ indicate that in three of four cultures the percentage of CD3+/CD8+ T cells was increased after priming with DCs + ATCs compared with the other priming conditions. Also, in two of four cultures primed with irradiated tumor cells alone, increased proportions of natural killer cells were observed. Overall, however, priming with DCs + ATCs favored an enrichment in T cells.
Phenotypic characteristics of T cells generated from PBMCs of healthy donors by priming with different stimulatorsa
The CTL lines generated by cross-priming under different conditions were further enriched in CD3+/CD8+ T cells by positive selection with Ab-coated magnetic beads (purity >95% as detected by flow cytometry). These T cells were then used as responders in ELISPOT assays, whereas PCI-13 tumor cells served as stimulators. The representative data for donor 1 (Table 3) ⇓ are shown in Fig. 5 ⇓ . The ELISPOT results indicate that T cells derived by priming with DCs + ATCs were significantly more responsive (P < 0.05) to the tumor (PCI-13) than were T cells derived by priming of PBMCs with γ-irradiated (viable) PCI-13 cells (Fig. 5A) ⇓ , ATCs alone (data not shown), or DCs alone (Fig. 5, A and B) ⇓ . More importantly, the response was blocked by anti-MHC class I and HLA-A2.1 mAbs but not isotype control immunoglobulin. These T cells did not respond to irrelevant HLA-A2+ HR (gastric carcinoma) or Fem-X (melanoma) tumor cells, or HLA-A2+ normal fibroblasts used as control (Fig. 5, A and C) ⇓ . In only one of four cross-priming experiments γ-irradiated PCI-13 cells were as stimulatory as DCs + ATCs.
Recognition of PCI-13 targets by T cells generated by cross-priming of PBMCs with DCs alone or DCs + various tumor preparations or tumor cells alone. IFN-γ ELISPOT assays were performed after two to four IVS cycles. T cells generated in these cocultures were enriched in CD8+ T cells by positive selection with magnetic beads prior to ELISPOT assays. A, T cells (as responders) were tested against PCI-13 or HR cells (HLA-A2+ gastric carcinoma used as a negative control) as stimulators at the ratio of 1:1. To confirm HLA-class-I restriction, PCI-13 cells were incubated with w6/32 or IgG Ab. Spots were counted by computer-assisted image analysis. A representative experiment (donor 1) of four performed with autologous DCs and T cells obtained from different HLA-A2+ healthy donors is shown. B, PBMCs of a normal HLA-A2+ donor (donor 3) were cross-primed with DCs, DCs + tumor lysate, or DCs + ATCs. The generated T cells were tested in ELISPOT assays against PCI-13 or HR cells at the ratio of 1:1. To confirm HLA-class-I restriction, PCI-13 cells were incubated with w6/32 or IgG Ab. Spots were counted by computer-assisted image analysis. A representative experiment of four performed with autologous DCs and T cells obtained from different HLA-A2+ donors is shown. C, PBMCs of two normal HLA-A2+ donors were cross-primed with DCs + ATCs. To confirm specificity of the generated T cells were tested in ELISPOT assays against PCI-13 +/− anti-MHC class I mAb (w6/32), PIC-13 +/− anti-HLA-A2 mAb (BB7.2), or against irrelevant targets such as HLA-A2+ melanoma (Fem-X) or HLA-A2+ normal human fibroblasts. Asterisks indicate that the number of spots obtained when T cells were cross-primed with DCs + ATCs was significantly higher (P < 0.05) than the numbers of spots obtained with all other priming protocols and the use of PCI-13 or PCI-13 + IgG as stimulator resulted in a significantly higher number of spots (P < 0.05) compared with controls (PCI-13 + w6/32 Ab; PCI-13 + BB7.2 Ab; HR cells; Fem-X; fibroblasts).
Experiments performed to compare the priming capacity of DCs + ATCs with that of DCs pulsed with PCI-13 lysates show that DCs + ATCs, but not DCs pulsed with tumor lysate, were effective in cross-priming autologous T cells (Fig. 5B) ⇓ . Most important, blocking experiments with anti-HLA class I and anti-HLA-A2 mAbs confirmed that recognition of PCI-13 by these T cells was MHC class I and HLA-A2 restricted, respectively (Fig. 5C) ⇓ . The results obtained with T2 cells, expressing unoccupied HLA-A2 molecules, which were pulsed with fresh tumor lysate and used as stimulators in ELISPOT assays, confirmed the results obtained with PCI-13 or DC stimulators (data not shown). In a limited number of experiments, we had sufficient numbers of T cells available to perform cytotoxicity assays in addition to ELISPOT assays. Our data consistently showed that DCs + ATCs generated MHC class Irestricted CTLs, whereas the other priming regimens did not (data not shown). Overall, these results clearly indicated that the cross-priming protocol, using DCs + ATCs was optimal for the generation of tumor-specific CTLs ex vivo.
DISCUSSION
SCCHN are generally considered to be poorly immunogenic and/or immunosuppressive tumors (23) . This perception is largely based on recent observations indicating that SCCHN show reduced expression of costimulatory molecules (24) , have alterations in the MHC class I-associated epitope processing pathway (25) , and are able to induce functional defects and apoptosis in immune cells (17) . Therefore, there is a need for immune restoration or potentiation of the immune system in patients with SCCHN, perhaps via vaccination strategies involving DCs as APCs. This novel therapeutic modality could be considered today because it is now feasible to generate human DCs ex vivo and pulse them with antigenic peptides for delivery to patients with cancer (26 , 27) . However, only a few TAAs have been characterized in SCCHN, including CASP-8, SART-1, and p53 wild-type peptides (28, 29, 30, 31, 32) . Vaccination trials with specific peptides ± DCs similar to those ongoing in patients with melanoma (26 , 27) have to await the definition of immunogenic peptides in SCCHN. Meanwhile, it might be feasible to use DCs plus tumor cells or ATCs as vaccine components, provided preclinical studies confirm that such DCs can be successfully used for cross-priming of T cells.
Current evidence suggests that significant differences exist in the efficiency of TAA processing and presentation by human DCs, depending on the source or form of tumor-derived materials, the maturation stage of DCs, or responsiveness of T-cell populations available for stimulation with DCs (8, 9, 10, 11, 12 ,, 33 , 34) . These differences are likely to be important for the outcome of immunizations, and, thus, there exists a need to optimize the design of tumor vaccines, using different tumor preparations (tumor cell lysates, ATCs, whole tumor cells, tumor cell fractions) and ex vivo-generated DCs. Clearly, a selection of the optimal method for antigen delivery to be available for future DC-based vaccine clinical trials is important for their success.
In the present study, we have evaluated human monocyte-derived DCs for their ability to: (a) take up ATCs; (b) present tumor-derived epitopes to already sensitized and committed tumor-specific T cells in the cross-presentation ex vivo model; and (c) prime T cells from normal donors to develop into antitumor effector cells in the cross-priming ex vivo model. To this end, two in vitro models had been developed. In the first model (cross-presentation) it was possible to evaluate tumor-specific T-cell responses to human DCs presenting TAAs derived from processed ATCs or tumor lysate. The CTLs used as responders in this model recognized a shared antigen on HLA-A2+ SCCHN, as described previously (16) , and again confirmed in the current experiments. Human DCs were HLA-A2+ semi-allogenic monocyte-derived APCs, which were either immature or were matured by ex vivo culture in the presence of proinflammatory cytokines. ELISPOT assays for IFN-γ production or cytotoxicity assays were used to monitor responses of CTLs to tumor-derived epitopes presented by DCs, tumor cells alone, ATCs alone, or DCs + tumor lysates. This in vitro cross-presentation model was used to quantitate the magnitude of CTL responses, and it allowed for a comparison of variously pretreated DCs for their capability to present a tumor-derived epitope(s) to CTLs known to be able to recognize and efficiently kill tumor cells.
Initially, while developing the model, it was necessary to show that SCCHN cells, subjected to an apoptotic signal (UVB light), were taken up by human monocyte-derived DCs. We showed a high level of uptake of ATCs by these DCs, as determined by flow cytometry as well as confocal microscopy. In fact, nearly all ATCs were taken up by DCs during a 24-h period of coincubation. We also observed that maturation of DCs was not visibly affected by the uptake of ATCs at the ATC:DC ratio of 1:2. On the other hand, previous reports indicated that in mice DC maturation was induced at a much higher (5:1) ATC:DC ratio, whereas lower ratios, similar to those used in our experiments, failed to mature DCs (11) . To achieve maturation of human DCs, we, therefore, resorted to the use of a mixture of proinflammatory cytokines (19) . When the ability of DCmat to uptake ATCs and present TAAs to T cells was compared with those of DCimm, it seemed that the uptake of ATCs was found to be reduced after 8–12 h of coincubation, but the T-cell stimulatory activity of DCmat was improved as compared with that of DCimm. Conceivably, up-regulation of costimulatory molecules and MHC class I molecules or increased stability of peptide-MHC-class I complexes on DCmat could be responsible for this effect. In addition, PGE2 in combination with TNF-α could synergistically induce high levels of IL-12 production in human monocyte-derived DCs, stimulate T-cell proliferation (35) , and increase IFN-γ production by responder CD8+ T cells without inducing type 2 cytokines, as reported previously (19) .
In our cross-presentation model, DCmat ingesting ATCs were recognized by the PCI-13-specific T cells somewhat better than DCimm and ATCs, but untreated PCI-13 cells were always eliciting the best T-cell responses. This observation is as expected because the CTL line used was generated by IVS with PCI-13 cells and was repeatedly stimulated with PCI-13 cells at weekly intervals. Similar results were reported by Bellone et al. (9) in a murine system. It was interesting to note that ATCs alone were not stimulatory, possibly due to a dramatic decrease in the expression of MHC class I molecules in ATCs relative to PCI-13 cells, as observed by flow cytometry. Several recent reports indicated that ATCs have reduced immunogenicity compared with live cells (12 , 36 , 37) . Many different mechanisms have been proposed to account for this low immunogenicity (12 , 36 , 37) , but our findings of low levels of MHC class I expression in ATCs is a novel observation that seems to fit well with the requirement for DCs, which express both MHC and costimulatory molecules, to process and present these ATCs to generate an effective immune response. It is reasonable to assume that ATCs become immunogenic, when they are cross-presented to T cells by professional APCs equipped with an efficient antigen-processing and presenting machinery and expressing costimulatory molecules. Processing of phagocytosed ATCs by DCs yields epitopes that can access the MHC class I pathway via TAP (transporter of antigen-processing)-dependent mechanisms and are ultimately presented to and recognized by antigen-specific CTLs. This type of effective cross-presentation of murine TAAs or viral antigens by macrophages and DCs has been described previously in ex vivo experiments (7, 8, 9, 10) .
In the second in vitro model (cross-priming), we used DCs, which had internalized ATCs or other tumor-derived epitopes, to generate effector T cells able to recognize and kill PCI-13 targets. In the recent in vivo experiments in rodents, Henry et al. (11) and Ronchetti et al. (12) described generation of tumor-specific CTLs, which were able to mediate tumor rejection and induce long-term memory, when using DCs + ATCs but not DCs pulsed with tumor extracts. These studies in rodents indicated a superior ability of DCs, which had internalized ATCs, to stimulate antitumor responses. Using human DCs and autologous T cells obtained from the circulation of normal HLA-A2+ donors, we demonstrated in the cross-priming model that stimulation with DCs + ATCs yielded T-cell lines strongly responsive to the tumor. To reach this conclusion, we performed several cycles of IVS, under conditions designed to compare γ-irradiated viable tumor cells, non-ATCs, ATCs alone, DCs + ATCs, DCs ± non-ATCs, DCs + lysate, or DCs alone for the ability to prime naïve T cells. This type of cross-priming was successful in generating tumor-specific CTLs in all four attempts using HLA-A2+ PBMCs of normal donors. Lymphocytes cross-primed with DCs + ATCs contained the highest frequency of IFN-γ-producing T cells specific for the tumor. In only one of four experiments (donor 4) γ-irradiated (viable) PCI-13 cells were as stimulatory as DCs + ATCs. ATCs alone did not induce the outgrowth of tumor-reactive T cells in our experiments. Interestingly, DCs pulsed with tumor lysates showed lower, if any, cross-priming capacity compared with DCs + ATCs. This finding was somewhat surprising in view of the widespread practice of using tumor lysates pulsed on DCs as a potentially effective antigen-delivery procedure for CTL generation. On the other hand, it has been shown that the administration of exogenous class I-restricted antigens in the form that requires phagocytosis is essential for their effective presentation to T cells (7 , 8) . Moreover, Inaba et al. (33) have shown that phagocytosed cellular fragments are 3000 times more efficient in forming MHC-peptide complexes than the preprocessed peptide. Because Herr et al. 5 observed the induction of a strong CD4 response, but only a weak CD8 response, when loading DCs with lysates in an EBV/viral model, it is possible that tumor-derived lysates primarily induce CD4 responses. This possibility is currently under investigation in our laboratory.
Overall, this study demonstrates that human monoctye-derived DCs that internalize and process ATCs can cross-prime T cells and generate more effective antitumor-specific T cells in vitro than viable tumor cells or tumor cell lysates pulsed on these DCs. Although these studies were performed in a semi-allogeneic setting, which could enhance T-cell activation and proliferation of both HLA-A2-restricted tumor-specific T cells as well as nonspecific T cells, the antitumor responses we measured in ELISPOT and cytotoxicity assays were mediated by tumor-specific, HLA-A2-restricted T cells. The results available from vaccination experiments in tumor-bearing rodents indicate that a similar approach may be successful in vivo (11 , 12) . Therefore, vaccination of cancer patients with ATCs + autologous DCs should be considered in the future as a reasonable therapeutic strategy, especially applicable when immunogenic tumor epitopes are not available.
Acknowledgments
We thank Dr. Albert DeLeo and Robbie B. Mailliard for helpful discussions as well as Audra Notalio and Sean Alber for assistance with cell imaging performed in the Biological Imaging Facility at the University of Pittsburgh, directed by Dr. Simon Watkins.
Footnotes
-
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 Supported in part by NIH Grant PO1-DE-12321 (to T. L. W.). T. K. H. is funded by a postdoctoral training fellowship from the Dr. Mildred Scheel Stiftung fur Krebsforschung (Grant D/99/08916).
-
↵2 These authors contributed equally to this work.
-
↵3 To whom requests for reprints should be addressed, at University of Pittsburgh Cancer Institute, W 1041 Biomedical Science Tower, 211 Lothrop Street, Pittsburgh, PA 15213-2582. Phone: (412) 624-0096; Fax: (412) 624-0264; E-mail: whitesidetl{at}msx.upmc.edu
-
↵4 The abbreviations used are: CTL, cytolytic T lymphocytes; APC, antigen-presenting cell; Ab, antibody; ATC, apoptotic tumor cell; DC, dendritic cell; DCimm, immature DC; DCmat, mature DC; ELISPOT, enzyme-linked immunospot; IL, interleukin; IVS, in vitro stimulation; mAb, monoclonal Ab; PBMC, peripheral blood mononuclear cell; PE, phycoerythrin; PG, prostaglandin; SCCHN, squamous cell carcinoma of the head and neck; TAA, tumor-associated antigen; TNF, tumor necrosis factor; TUNEL, terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling; DiOC6, 3,3′-dihexadecyloxacarbocyanine perchlorate.
-
↵5 W. Herr, E. Ranieri, W. Olson, H. Zarour, L. Gesualdo, and W. J. Storkus. Mature dendritic cells pulsed with freeze-thaw cell lysates define an effective in vitro vaccine designed to elicit EBV-specific CD4+ and CD8+ T lymphocyte responses, submitted for publication.
- Received October 15, 1999.
- Accepted May 4, 2000.
- ©2000 American Association for Cancer Research.
References
Volume 60, Issue 13
