Antitumor Monoclonal Antibodies Enhance Cross-Presentation of Cellular Antigens and the Generation of Myeloma-specific Killer T Cells by Dendritic Cells

MHC class I molecules are generally complexed with peptides derived exclusively from newly synthesized cytosolic proteins (1). However, CD8⁺ T cell responses can also be directed to exogenous cell–associated antigens derived from tumors, transplants, and virus-infected nonhematopoietic cells (2). This requires presentation of these antigens by bone marrow–derived cells (a process termed cross-presentation), with dendritic cells (DCs)^*being one of the major candidate cell types. DCs can acquire antigen from dying tumor cells and elicit tumor-specific CD8⁺ T cell responses in vitro (3–8). Strategies to optimize cross-presentation of antigen from tumor cells are of great interest for immunotherapy of tumors.

Studies in animal tumor models have implicated both cellular and humoral responses in protective antitumor immunity (9). Interest in tumor-specific humoral immunity has been intensified by certain successes of mAbs in cancer therapy. The mechanism of antitumor effects of these antibodies is not fully understood. Direct effects on tumor cells, as well as enhancement of innate effectors (complement and antibody-dependent cytotoxicity) have been proposed (10–12). Fc receptors are required for both active and passive immunity to melanoma, (11) and the protective effect of therapeutic mAbs in some murine models (12). Together, these studies point to the importance of Fc-dependent innate effector mechanisms in the protective effects of tumor-specific antibodies. However, whether coating of tumor cells by antibodies also affects antitumor cellular immunity is not known.

Here we show that coating of tumor cells with antitumor mAbs leads to enhanced cross-presentation of tumor-derived cellular antigens and generation of tumor specific–killer T cells by DCs. This effect is dependent on Fcγ receptors (FcγRs) on the DCs, but is exerted at a step after the uptake of tumor cells by DCs.

Myeloma cell lines were obtained from American Type Culture Collection (U266, RPMI 8226), or provided by J. Epstein, Arkansas Cancer Center, Little Rock, AR; cag, arp, ark cells). HLA A2.1 status on cell lines was assessed by serotyping. All cell lines were grown in RPMI 1640/10–20% FCS/glutamine/gentamicin.

The expression of a panel of cancer-testis (C-T) antigens (MAGE1, MAGE3, MAGE4, MAGE10, CT-7, LAGE-1, and NY-Eso-1) by myeloma cells was examined using RT-PCR, as described previously (13).

Tumor cells were killed by repeated freeze thaw cycles (necrosis) or by γ irradiation (30 Gy) (apoptosis). The induction of apoptosis was monitored using staining with Annexin V-FITC. For antibody coating, tumor cells (10⁷ cells per milliliter) were incubated with anti–syndecan-1 antibody (14) (1 μg/ml, B-B4; Serotec) or isotype (IgG1) control, for 30 min at 4°C. Syndecan-1 is a heparan sulfate proteoglycan highly expressed on myeloma cells. After antibody coating, cells were washed, irradiated with 3 Gy, and immediately added to immature DCs as live (Annexin V-negative) cells. To determine if antibody coating enhanced presentation of antigens from dying cells, tumor cells were killed either by γ-irradiation (apoptosis) or freeze thaw (necrosis) as above, and then treated with anti–syndecan-1 or isotype control antibody as described above, or left untreated, before feeding to DCs at DC/tumor ratio of 1:1.

DCs were generated as described previously (15), by culture of plastic adherent blood mononuclear cells, obtained from leucocyte concentrates, or whole blood of A2.1+ve healthy donors, in GM-CSF (Immunex) and IL-4 (R&D Systems). The nonadherent blood cells were used as a source of T cells. On day 5 or 6 of culture, the immature DCs were fed in 96-well plates with apoptotic, necrotic, or live antibody coated HLA A2.1-negative tumor cells at a ratio of 1:1 and then 4–12 h later, a cytokine cocktail consisting of IL-1β (10 ng/ml), IL-6 (1,000 U/ml), TNF-α (10 ng/ml), and PGE₂ (1 μg/ml) was added to induce maturation (16, 17). As controls, some mature DCs were pulsed for 2 h with 1 μM HLA A2.1-restricted peptides from MAGE-3 (271–279; FLWGPRALV) and NY-Eso-1 (157–167; SLLMWITQCFL). For some experiments, DCs were pretreated with a cocktail of anti–human Fc receptor blocking anti-CD16 (clone 3G8) and CD32 (clone IV.3) antibodies (both at 10 μg/ml for 60 min; Medarex), before tumor cell feed. For some experiments, DCs were fed with a mixture of antibody-treated and untreated tumor cells, differing in the expression of tumor antigens.

To evaluate phagocytosis of tumor cells by DCs, tumor cells were dyed red with PKH26 (Sigma-Aldrich) before cell death or antibody coating, and immature DCs were stained green with PKH67 (Sigma-Aldrich) before coculture with tumor for 0–16 h at 4°C or 37°C (5). Phagocytosis was monitored by flow cytometry as double-positive cells. Tumor cell uptake was also evaluated by applying tumor cell–loaded DCs to alcian blue coated slides for two color fluorescence microscopy using an Olympus epifluorescence microscope. DCs were stained with anti-CD11c and tumor cells were stained with an antibody for C-T antigen, CT-7 (a gift of A. Jungbluth, Ludwig Institute, New York, NY). The motorized stage allowed 0.5-mm optical sections with a cooled CCD camera (Hammatsu) and Metamorph software (Universal Imaging Corp.).

Antigen loaded or unpulsed DCs were cocultured in 96-well round-bottomed microtest plates with autologous nonadherent cells at a ratio of 1:30 in 0.2 ml of 5% pooled human serum. As controls, irradiated (3 Gy) antibody/isotype treated tumor cells were also used to present antigen. The culture media was supplemented with recombinant IL-2 (50 U/ml) on days 4 and 7 of culture. At 1 wk, cultures were restimulated with the identical APCs as the initial stimulation. For some experiments, DCs loaded with tumor cells were purified from the DC-tumor mixtures after initial coculture for 4 h, using anti-CD11c-coupled magnetic beads (Miltenyi Biotech). The DCs were matured before their use as APCs. For some experiments, DCs loaded with tumor cells were cultured in transwell inserts, physically separated from autologous T cells and unpulsed DCs.

Antigen-specific T cell response against defined HLA A*0201-restricted peptide epitopes from MAGE-3 and NY-Eso-1 was determined using IFN-γ enzyme linked immunospot (ELISPOT) and cytotoxicity assays, after 2 wk of culture with tumor/peptide loaded DCs. For the ELISPOT assay, autologous mature DCs pulsed with 1 μM of specific (MAGE-3, NY-Eso-1) or irrelevant (HIV-gag) peptide were used as APCs at ratio of 1:30, and spot-forming cells were quantified after overnight culture as described previously (15). For evaluation of antigen-specific killers, T2 cells pulsed with 1 μM of specific (MAGE-3, NY-Eso-1) peptide were used as targets in a standard 5 h ⁵¹Cr release assay. Other targets were HLA A*0201 +ve (U266 cells) or –ve (ark cells) myeloma cells known to express both MAGE-3 and NY-Eso-1, and K562 cells to monitor NK cell–mediated lysis.

The expression of several members of C-T antigen family by myeloma cell lines was evaluated by RT-PCR. All cell lines tested expressed at least one member of this family (Table I).

We used 2 HLA A2.1-negative cell lines (cag, arp) expressing both MAGE-3 and NY-ESO-1 (cag cells) or MAGE-3 alone (arp cells), as sources of defined C-T antigens to elicit HLA A*0201-restricted peptide-specific T cells in vitro. Both cell lines expressed high levels of syndecan-1, and binding of anti-syndecan antibody did not induce apoptosis (data not shown). The myeloma cells were induced to undergo apoptosis by high dose (30 Gy) γ irradiation, necrosis by repeated freeze thaw cycles, or treated with anti–syndecan-1 antibody/isotype, as live, low dose (3 Gy) γ-irradiated cells. Myeloma cells were then fed to immature DCs from healthy donors. Phagocytosis studies performed using tumor cells and DCs labeled with different color fluorochromes indicated that label from anti–syndecan-1, isotype control antibody treated, or untreated tumor cells was taken up by DCs upon 2–20 h of coculture at 37°C (but not at 4°C) with similar efficiency (Fig. 1). Uptake of tumor cells was also verified using confocal microscopy after staining for tumor cells with an antibody to the CT-7 tumor antigen (data not shown). Therefore the coating with anti–syndecan-1 antibody was not required for tumor cell uptake in this system.

Ligation of FcγR is known to induce the maturation of mouse DCs. In our system however, the uptake of antibody treated (both anti–syndecan-1 or isotype control), or untreated tumor cells was associated with partial, but equivalent phenotypic maturation of DCs, as measured by a slight increase in the surface expression of CD83, and considerable increases in CD80 and CD86. However, upregulation of HLA-DR, CD40, and CD25 was not noted, as with DCs matured using inflammatory cytokines (Fig. 2). However these DCs could be fully matured by the addition of inflammatory cytokine cocktail after tumor cell uptake (data not shown). Therefore the anti–syndecan-1 antibody does not detectably influence DC uptake of myeloma cells or DC maturation.

After phagocytosis and ex vivo maturation using cytokine cocktail, the tumor-loaded, peptide-pulsed, or unpulsed DCs, as well as tumor cells alone, were used as APCs to stimulate T cells in culture. No T cell reactivity to either MAGE3 or NY-Eso-1 peptide was noted at baseline before in vitro expansion in any subject (data not shown). However, DCs loaded with apoptotic-, necrotic-, or antibody-coated myeloma cells expanded C-T antigen-specific T cells using two cycles of stimulation in vitro (Fig. 3 A). If the tumor expressed both MAGE-3 and NY-Eso-1 (cag cells), then both MAGE-3 and NY-Eso-1–specific T cells developed. In contrast, expansion of only MAGE-3 but not NY-Eso-1–specific T cells was seen using DCs loaded with arp cells, consistent with the pattern of antigen expression on these cells. Peptide-pulsed DCs also led to the expansion of peptide-specific T cells, while no expansion was seen using tumor cells alone. The antigen-specific T cell responses were lost after depletion with anti-CD8 antibody before the ELISPOT, indicating the CD8⁺ nature of the IFN-γ–secreting cells (data not shown).

In experiments with five consecutive donors, the generation of tumor-specific T cells was the greatest using DCs loaded with antibody-coated tumor cells (5/5 donors), higher than the responses observed using DCs loaded with only dying cells (3/5 experiments; Fig. 3 B). Thus antibody coating of tumor cells enhances the cross-presentation of cellular antigens after tumor cell uptake by DCs.

To examine the possibility that contaminating cells in the DC-tumor cocultures led to T cell expansion, we purified CD11c⁺ DCs from the DC-myeloma mixtures by positive selection, and then used CD11c⁺ or CD11c−ve fractions as APCs. The CD11c purified DCs expanded antigen specific T cells (Fig. 3 C). No T cell expansion was seen with the CD11c −ve fraction (data not shown). To examine if soluble factors/peptides generated in these cultures could lead to T cell expansion, we cultured these tumor cell–loaded DCs in transwell inserts, physically separated from mixtures of T cells and unpulsed DCs. No expansion of antigen-specific T cells was seen in these cultures. In summary, activation of tumor-specific T cells requires direct contact of T cells with the tumor-loaded DCs (Fig. 3 C).

T cells elicited using mAb tumor–loaded DCs efficiently killed not only peptide pulsed targets, but also killed tumor cells HLA matched to DCs (Fig. 4 A and B). In contrast, T cells elicited using peptide-pulsed DCs only killed peptide pulsed targets, but not HLA A2.1⁺ myeloma cells expressing these antigens (Fig. 4 A). The killing of tumor cells was inhibited by preincubation of targets with anti-MHC I, but not anti-MHC II (data not shown), consistent with the response being due to CD8⁺ T cells and specific for MHC class I binding peptides. Again, the generation of tumor-specific killer T cells was higher and more consistent using DCs loaded with antibody- coated cells (3/3 donors) than using dying cell loaded DCs (2/3 donors; Fig. 4 B). Thus DCs loaded with antibody-coated tumor cells are superior to peptide pulsed DCs or DCs loaded with dying cells to stimulate tumor-specific killer T cells.

In the experiments described above, we cocultured DCs with live antibody-coated tumor cells after low dose (3 Gy) irradiation. However, several studies have shown that DCs can cross-present antigen from apoptotic or necrotic tumor cells. Therefore, we next examined if antibody coating also enhanced cross-presentation of antigen from dying cells. We first confirmed the expression of syndecan-1 on tumor cells rendered apoptotic after γ-irradiation, by double staining for Annexin V and CD138 (data not shown). Next, we fed DCs with apoptotic or necrotic tumor cells that had been stained with anti–syndecan-1 antibody, isotype control, or left unstained. These DCs were then matured using an inflammatory cytokine cocktail and used as APCs to stimulate autologous T cells. The DCs that were fed with antibody-coated apoptotic tumor cells expanded antigen-specific T cells more efficiently relative to DCs fed with apoptotic cells alone (Fig. 5 A). No enhancement of presentation of necrotic cells was observed by antibody-mediated targeting, which is likely due to the lack of effective binding of the antibody to necrotic fragments. T cells expanded using DCs loaded with antibody-coated apoptotic cells were also superior at killing peptide-pulsed or tumor cell targets (Fig. 5 B). Thus antibody coating of tumor cells also leads to enhanced cross-presentation of dying cells by DCs.

To examine the role of antibody and FcγR in the cross-presentation of antigens by DCs, DCs were first fed with a mixture of antigen (NY-Eso-1)-positive (cag cells) and negative (arp cells) tumors. The tumor cells were internalized similarly with or without antibody coating, implying substantial direct recognition of tumor cells by DCs. However, NY-Eso-1–specific T cells were efficiently expanded in these cultures only if the NY-Eso-1-positive tumor cells (cag cells) were coated with anti–syndecan-1 antibody (Fig. 6 A). Therefore, antibody coating directly promotes the cross-presentation of tumor cells.

Preincubation of DCs with a cocktail of Fc receptor blocking antibodies (anti-CD16 plus CD32) also significantly inhibited the generation of antigen-specific IFN-γ–producing T cells after uptake of tumor cells (Fig. 6 B), although the uptake of tumor cells was not blocked (data not shown). Likewise, the generation of antigen-specific killer T cells was also inhibited in these cultures (data not shown). Culture of DCs and T cells in heat-inactivated serum to inactivate complement, did not inhibit the generation of tumor-specific T cells (data not shown). Thus enhanced cross-presentation of mAb-coated tumor cells is dependent on Fc receptor mediated interactions, that act at a step subsequent to standard opsonization or uptake.

Cross-presentation of cellular antigens is of special interest in tumor immunology, as tumors themselves are poor APCs (2). Recent studies have shown that DCs can acquire cellular antigen from dying tumor cells and expand tumor reactive CD8⁺ killer T cells in vitro (3–8). Our studies now demonstrate that cross-presentation of cellular antigens by human DCs is enhanced by mAb-mediated targeting. DCs, positively selected after just 4-h coculture with live mAb treated tumor cells, are able to present tumor antigen with superior efficiency than DCs loaded with dying cells or defined peptides.

Enhanced cross-presentation of antibody-coated tumor cells is dependent on FcγR-mediated interactions and inhibited by preincubation of DCs with FcγR blocking antibodies. Increased presentation is not simply explained by increased phagocytosis of mAb-treated tumor cells. Thus, the uptake of tumor cells treated with isotype control antibody is comparable to the cells treated with anti-syndecan, yet the former are presented poorly under identical conditions. Indeed, when DCs are fed with a mixture of antigen-positive and -negative cells, efficient generation of antigen-specific T cells occurs only when the cells that express the antigen are coated with the mAb.

Although uptake of antigen as immune complexes or FcγR targeted liposomes delivers a maturation signal to murine DCs, we were unable to show a difference in phenotypic maturation between DCs fed with antibody-treated or untreated tumor cells (18, 19). Taken together, our data suggest that the enhanced cross-presentation of antibody-coated tumor cells by DCs is primarily due to enhanced efficiency of cross-presentation for antigen via FcγR. Thus while DCs may be efficient at phagocytosing myeloma cells, efficient cross-presentation of the phagocytosed antigen by DCs may depend on the specific receptors involved in tumor antigen uptake. This result is reminiscent of the findings of Albert et al., who found that monocytes could phagocytose apoptotic cells as well as immature DCs, and expressed high levels of MHC I, but only DCs could cross-present the cell-associated antigen (20). Enhancement of cross-presentation observed here may be due to novel pathways of antigen processing for FcR-mediated uptake (21) or FcR-mediated signaling (22). Indeed, several studies have shown efficient FcγR-mediated uptake and presentation of immune complexes in murine and human APCs including DCs (18, 23–28).

Whole (either autologous or allogeneic) tumor cells are an attractive source of antigen to load DCs simultaneously with multiple tumor-specific epitopes for immunotherapy. Our findings that DCs loaded with antibody-coated cells are superior to both peptide-pulsed, as well as dying cell-loaded DCs, provides a novel and potentially more effective antigen loading strategy for DC immunotherapy and the generation of tumor-specific killer T cells for adoptive transfer. T cells elicited using peptide-pulsed DCs killed peptide pulsed targets, but not tumor cells. Lack of tumor cell killing by some (specially MAGE-3) peptide-specific T cells may relate to the lack of processing and presentation of this epitope by tumor cells (29). Importantly, T cells elicited using mAb-tumor loaded DCs were superior to peptide-loaded DCs not only for killing of peptide-pulsed targets but also lysed HLA-matched tumor cells. The mechanism for eliciting superior antigen-specific lytic effectors after mAb-tumor loaded DCs is not known, and may be due to the elicitation of tumor-specific CD4⁺ T cells, the polyepitope nature of the response or the unique efficiency of FcγR-mediated intracellular antigen processing in DCs, as compared with loading with preprocessed peptide (21, 25, 30).

Antitumor effects of mAbs in cancer are thought to be mediated by direct effects on cancer cells, as well as enhancement of innate effectors. Binding of antibodies to tumors cells can sensitize them for complement or antibody-dependent cytotoxicity (10). However the impact of this therapy on T cell immunity is unclear. In some tumor models, both CD8⁺ T cells and functional FcγRs were shown to be important for mAb-mediated eradication of established tumor (31, 32). Increase in T cell infiltration at the tumor site has also been observed in some human mAb trials, however specificity of the T cell response was not determined (33). Our findings suggest that mAb therapy may also promote the induction of antigen-specific CD8⁺ T cell responses. These data may also allow the application of mAbs to settings wherein such therapy may not be clinically feasible. For example, selective targeting of anti–syndecan-1 to myeloma cells in vivo is limited by the fact that this molecule is expressed on nonhematopoietic cells. Our data now suggest a strategy for ex vivo targeting of this molecule on myeloma tumor cells to DCs.

The authors thank Dr. Ralph Steinman for many helpful suggestions and critical reading of the manuscript, and Drs. Lloyd Old and Jeffrey Ravetch for helpful discussions.

This work was supported in part by an investigator award from the Cancer Research Institute and grants from National Institutes of Health (CA81138) to M.V. Dhodapkar; MO-RR00102 to the Rockefeller GCRC and the Rockefeller University Clinical Scholars program (to K.M. Dhodapkar).