Efficient cross-priming of tumor antigen-specific T cells by dendritic cells sensitized with diverse anti-MICA opsonized tumor cells

Abstract

Dendritic cells (DCs) have the capacity to prime tumor-specific T cell responses and are considered as potentially effective vaccines for immunotherapy of cancer. Critical parameters in the development of DC vaccines are the source of tumor antigen (TA) and the mode of DC-loading. Whole tumor cells contain complex assortments of TA, which has been exploited to enhance cross-presentation to CD8 T cells by DCs loaded with anti-syndecan mAb-opsonized myeloma cells. This approach may be broadly improved by targeting the MHC class I chain-related protein A (MICA), which is frequently and abundantly expressed on most if not all types of epithelial cancers but not in normal tissues except intestinal mucosa. Loading of DC with anti-MICA mAb-coated breast, melanoma, or ovarian tumor lines or uncultured ovarian cancer cells efficiently promoted TA cross-presentation and priming of multivalent anti-tumor CD8 and CD4 T cell responses. These were of substantially greater breadth and magnitude than those of T cells primed by peptide-pulsed or apoptotic tumor cell-loaded DCs. These results may advance DC vaccine development and provide a platform for adoptive T cell therapy and TA discovery. These results further suggest that antibody targeting of MICA might be applicable to elicit T cell immunity against tumors of diverse tissue origins in cancer patients.

Dendritic cells (DCs) have a central role in the initiation and control of T cell-mediated immunity. Immature DCs residing in tissues endocytose soluble antigens, microbes, or apoptotic cells, and receive microbial or inflammatory maturation cues depending on the type of pathogen and the nature and extent of tissue damage (1-3). Maturing DCs migrate to lymph nodes via afferent lymphatics where they complete their maturation and present peptides derived from protein degradation on MHC class I and class II molecules to CD8 and CD4 T cells, respectively (2). The type and composition of maturation signals received by DCs determine whether they induce productive T cell responses or tolerance (4, 5).

Because of their potent immunostimulatory capacity, there is much interest in employing DCs as tumor vaccines to induce effector and long-term memory CD4 and CD8 T cells with broad tumor antigen (TA) specificities (6-8). Early-stage trials using ex vivo TA-loaded and matured monocyte- or CD34⁺ progenitor-derived DCs have provided some evidence for clinically beneficial immunostimulatory effects (9, 10). However, many variables remain to be explored, including antigen sources and modes of DC antigen-loading. Most commonly, DCs are pulsed with synthetic TA-derived MHC-binding peptides (11, 12). This approach is constrained by limited knowledge of TA, their natural and immunoselected variation within and among tumors, and by the MHC allele-specific restrictions of peptide-binding, thus producing narrow repertoires of antigen-specific T cells. Other methods include transfection of DCs with tumor-derived RNA, the use of viral vectors for expression of TA, and facilitating DC uptake of tumor-derived exosomes, apoptotic tumor cells, or recombinant proteins for antigen processing and presentation (13-19). Moreover, DC expression of Ig Fc receptors (FcγR) can be exploited for targeting immune complexes and antibody-coated tumor cells to DCs (20, 21). Exposure of DCs to anti-syndecan mAb-opsonized myeloma cells promotes efficient in vitro cross-priming of cytotoxic T lymphocytes, yielding results that are superior to cross-presentation of TA from apoptotic tumor cells (22).

This immunization mode is appealing because it may operate in vivo and contribute to the beneficial effects of some antibody-based cancer therapies (23, 24). Hence, antibody targeting of surface molecules that are absent from most cell types and tissues but are frequently tumor-associated could be a viable approach to inducing tumor immunity. Suitable candidate molecules are MHC class I chain-related proteins A (MICA) and B (MICB), which are distantly related to MHC class I and function as ligands of the NKG2D receptor on natural killer cells and T cell subsets (25, 26). In healthy individuals, the tissue distribution of MIC is restricted to variable areas of gastrointestinal epithelium (25, 27). However, MICs are abundantly expressed in many lung, breast, kidney, ovarian, prostate, gastric, and colon carcinomas, and melanomas, as well as in certain leukemias and on epithelial tumor cell lines (28-33). Hence, because MIC proteins represent uncommonly ubiquitous tumor markers, we investigated the ability of DCs to take up anti-MICA mAb-opsonized breast, melanoma, and ovarian tumor lines and uncultured ovarian tumor cells, and to prime multivalent antitumor CD8 and CD4 T cell responses in vitro. The results suggest that this approach may be effective for immunization against a large variety of histologically distinct tumors.

Materials and Methods

Peripheral Blood Samples and Tissue Materials. Peripheral blood was obtained from HLA-A2⁺ healthy volunteers and from three and six patients with metastatic melanoma and stage III-IV ovarian cancer, respectively. Matched primary ovarian tumor surgical material and malignant ascites fluid were previously obtained from one, and matched primary ovarian cancer from two of these patients (28, 29). Malignant pleural effusion was obtained from one stage-IV breast cancer patient. These activities were approved by local institutional review boards, and all subjects gave written informed consent.

Tumor Cell Suspensions and Cell Lines. Tumor cell suspensions were prepared as described in ref. 28, tested for MICA by staining with mAb 2C10 (IgG1) (25) and flow cytometry, and cryopreserved until use for DC-loading. Ovarian and breast cancer lines OT140 and JB were generated from ascites fluid- and pleural effusion-derived cell pellets, respectively. Cells were resuspended in RPMI medium 1640/20% FCS supplemented with epidermal growth factor (12.5 ng/ml; Invitrogen), insulin (1 μg/ml; Sigma), and hydrocortisone (1 μg/ml; Sigma). After several passages, cells were adjusted to RPMI medium 1640/10% FCS without supplements. OT140 and JB cells were positive for surface MICA, and for NY-ESO-1 and NY-BR-1 as shown by using RT-PCR with primers 5′-AGCCGCCTGCTTGAGTTCT-3′/3′-AGCGCTGTGGGTACCTT-5′ and 5′-TCGAAGAGCAGCATAGGAAA-3′/3′-CAGAACTTTAAGCTGCCCACT-5′, respectively. Other tumor lines were used as antigen sources and targets or stimulators in cytotoxicity and IFN-γ secretion assays. Melanoma cell lines used were A2058 (MICA⁺, HLA-A0201^-, MART-1⁺, gp100⁺, and tyrosinase⁺), MZ2 (MICA⁺, HLA-A0201^-, MART-1^-, gp100^-, and tyrosinase^-), mel526 (MICA^-, HLA-A0201⁺, MART-1⁺, gp100⁺, and tyrosinase⁺), Malme-3M (MICA⁺, HLA-A0201⁺, MART-1⁺, and gp100⁺), and A375 (MICA⁺, HLA-A0201⁺, MART-1^-, gp100^-, and tyrosinase^-). Ovarian cancer lines used were OVCAR-3 (MICA⁺, HLA-A0201^-, and NY-ESO-1⁺), HTB-77 (MICA⁺, HLA-A0201⁺, and NY-ESO-1⁺), OT140 (MICA⁺, HLA-A2⁺, HLA-DP4⁺, and NY-ESO-1⁺), and CRL-2183 (MICA⁺, HLA-A0201⁺, and NY-ESO-1^-). Breast cancer lines used were MDA-453 (MICA⁺, HLA-A0201^-, and NY-BR-1⁺), JB (MICA⁺, HLA-A0201^-, and NY-BR-1⁺), HTB-21 (MICA⁺, HLA-A0201⁺, and NY-BR-1⁺), and MCF-7 (MICA⁺, HLA-A0201⁺, and NY-BR-1^-). MZ2 and mel526 were gifts from T. Boon (Ludwig Institute, Brussels) and Y. Kawakami (University of Keio, Japan), respectively. All other cell lines were from American Type Culture Collection. Cell lines were tested for MICA and HLA-A2 by flow cytometry using mAbs 2C10 and MA2.1, respectively. TA expression was determined by using RT-PCR (NY-ESO-1 and NY-BR-1, see above) or by immunohistochemistry using sections of paraffin-embedded cell pellets, anti-MART-1/Melan A, anti-gp100/HMB45 (NovoCastra, Newcastle, U.K.) and anti-tyrosinase mAbs (NeoMarkers, Lab Vision, Fremont, CA), and peroxidase-streptavidin-biotin conjugate staining.

DC and T Cell Preparations. Peripheral blood mononuclear cells (PBMC) were isolated by density-gradient centrifugation (Ficoll/Hypaque, Pharmacia). Immature DCs were derived from monocytes by culture of adherent PBMC in the presence of granulocyte/macrophage colony-stimulating factor (800 units per ml; LEUKINE sargramostim, Amgen Biologicals) and IL-4 (500 units per ml; R & D Systems) in AIM-V media (Invitrogen) (34). DC maturation was induced by a 36- to 48-h culture in the presence of TNF-α (10 ng/ml), IL-1β (2 ng/ml), IL-6 (1,000 units per ml) (all from R&D Systems), and prostaglandin E₂ (1 μg/ml) from Sigma. CD8 and CD4 T cells were purified from PBMC by using Dynabeads and DETACHaBEAD (DynalBiotech, Brown Deer, WI).

Tumor Cell Uptake and Maturation of DCs. MICA⁺ tumor lines labeled with the PKH26 red fluorescent cell linker kit (Sigma) were treated with anti-MICA mAb 2C10 (1 μg/ml, 30 min at 4°C) or control IgG1, or left untreated, and low-dose (10-50 Gy) or high-dose (30-90 Gy) irradiated to arrest proliferation or induce apoptosis, respectively. Tumor cells were cocultured at 1:1 ratios with the PKH67 green fluorescent cell linker kit (Sigma)-labeled immature (day 6) DCs in AIM-V media for ≤20 h (22). Tumor cell uptake was evaluated by using two-color flow cytometry, and DC maturation was assessed by stainings with conjugated anti-CD11c, -CD40, -CD80, -CD83, and -CD86 (BD Pharmingen).

Generation of Cross-Presenting DCs, T Cell Stimulation, and Cloning. HLA-A2⁺ immature (day 6) DCs were loaded with equal numbers of mAb 2C10-opsonized live (low-dose irradiated) or apoptotic (high-dose irradiated) tumor lines A2058, MZ2, mel526, OVCAR-3, OT140, MDA-543, and JB, or with cryopreserved and thawed ovarian tumor cell suspensions, and were cytokine-matured and cocultured with DC autologous CD8 or CD4 T cells at a DC-to-T cell ratio of 1:10 in RPMI medium 1640/10% human serum/55 μM mercaptoethanol/antibiotics. For comparison, DCs were pulsed with 10 μM HLA-A2-restricted peptide epitopes of MART-1 (M27; AAGIGILTV), gp100 (G154; KTWGQYWQV), tyrosinase (368D; YMDGTMSQV), NY-ESO-1 (157-165; SLLMWITQC), NY-BR-1 (904; SLSKILDTV), or the HLA-DP4-restricted NY-ESO-1 peptide (157-170; SLLMWITQCFLPVF), or were loaded with untreated apoptotic tumor cells. T cell stimulations were in the absence of exogenous cytokines. On day 7 after the second weekly stimulation, cultures were analyzed for HLA-A2-peptide tetramer positive T cells, for cytotoxicity against tumor cell targets, and for IFN-γ secretion upon tumor cell stimulation. To establish TA-specific T cell clones, IL-6 (10 ng/ml) and IL-12 (5 ng/ml) were added during the first weekly stimulation, and IL-7 (10 ng/ml; all from R&D Systems) and IL-2 (10 units/ml; aldesleukin, Chiron) were added during the second weekly stimulation. Tetramer-positive CD8 T cells or INF-γ-secreting CD4 T cells were sorted on day 7 after the second stimulation, by using a FACS-Vantage cell sorter (Becton Dickinson), into 96-well round-bottom plates containing RPMI medium 1640/10% human serum, 55 μM mercaptoethanol, antibiotics, anti-CD3 (OKT3, 30 ng/ml; Orthoclone, Ortho Biotech, Bridgewater, NJ], irradiated allogeneic feeder cells (10⁶ PBMCs per ml and 2 × 10⁵ lymphoblastoid cells per ml), and IL-2 (25 units/ml added every 2-3 days). Ten to 14 days after plating, T cells were screened in microcytoxicity assays against ⁵¹Cr-labeled, peptide-pulsed T2 cells (2 × 10³) and unpulsed controls. Restimulated T cell clones were tested against T2 cells pulsed with titered peptide concentrations. CD4 T cell clones were screened for IFN-γ secretion in response to peptide-pulsed and tumor cell-loaded DCs. The HLA-DP4-restricted NY-ESO-1-specific CD4 T cell clones MS#8 and MS#55 were established from PBMCs from an NY-ESO-1⁺ melanoma patient (N.N.H. and C.Y., unpublished data).

MHC Tetramers and Stainings. Tetramers complexed with antigenic peptides were made as described in ref. 35. To quantify antigen-specific T cells in mixed populations, 0.5-1 × 10⁶ T cells were stained with FITC-conjugated anti-CD8 (Pharmingen) and phycoerythrin-conjugated tetramer (25-50 μg/ml) for 30 min at 4°C and examined by using a BD LSR benchtop analyzer (Becton Dickinson). Dead cells were excluded by staining with 7-AAD (5 μg/ml; BD Pharmingen).

Cytotoxicity and IFN-γ Capture and Release Assays. T cell cytotoxicity was tested in standard 4-h ⁵¹Cr- release assays. The presence of IFN-γ-producing T cells after stimulation with tumor cells at a ratio of 5:1 was tested by using an IFN-γ secretion detection kit (Miltenyi Biotec, Auburn, CA) as recommended by the manufacturer. For INF-γ release assays, resting (14-18 days after stimulation) T cells (10⁵ per well) were stimulated with equal numbers of tumor-loaded or peptide-pulsed autologous DCs. After 24 h, INF-γ was quantitated in pooled supernatants from triplicate wells by using commercial ELISA with matched antibody pairs in relation to standards (R & D Systems).

Results and Discussion

Enhanced TA Cross-Presentation to CD8 T Cells by DCs Loaded with Anti-MICA-Opsonized Melanoma, Ovarian, or Breast Cancer Cells. HLA-A2^- and MICA⁺ tumor lines with previously known or newly determined antigenicities were selected to evaluate the effects of anti-MICA opsonization on DC tumor cell phagocytosis and maturation. These tumor lines included the melanoma lines A2058 (MART-1⁺, gp100⁺, and tyrosinase⁺) and MZ2 (MART-1^-, gp100^-, and tyrosinase^-), the ovarian cancer line OVCAR-3 (NY-ESO-1⁺), and the breast cancer lines MDA-453 and JB (NY-BR-1⁺). The melanoma mel526 line (HLA-A2⁺, MART-1⁺, gp100⁺, and tyrosinase⁺) served as a MICA^- control. Green fluorescent dye-labeled immature monocyte-derived DCs from healthy HLA-A2⁺ donors were cocultured with low-dose irradiated red fluorescent dye-labeled tumor cells that were coated with anti-MICA mAb 2C10 or isotype control Ig, or were left untreated. Two-color flow cytometry showed that after 4 and 20 h, ≈50% and 90% of DCs were loaded with tumor cells, respectively. Among nine such experiments, no reproducibly increased uptake of anti-MICA-opsonized tumor cells was observed. Similarly, there were no notable differences in DC maturation, which was marginal with only small increases of CD83 and CD40 and required exogenous cytokines for full induction (data not shown and ref. 22).

To test whether anti-MICA opsonization promotes TA cross-presentation, DCs derived from HLA-A2⁺ healthy donors were loaded with low-dose irradiated melanoma, ovarian, or breast cancer lines coated with mAb 2C10, cytokine-matured, and cocultured with autologous peripheral-blood CD8 T cells. In parallel, DCs were pulsed with TA peptides or loaded with Ig isotype-treated or apoptotic tumor cells. Frequencies of TA-specific CD8 T cells were determined by staining with phycoerythrin-conjugated HLA-A2-peptide tetramers and flow cytometry. Responses to MART-1, gp100, tyrosinase, NY-ESO-1, and NY-BR-1 were tested in seven and five independent experiments by using different donors, respectively. Freshly isolated control CD8 T cells were negative for binding of the tetramers before stimulation, except for the occurrence of small numbers of MART-1-specific T cells in four of seven individuals (Fig. 1) (36). All CD8 T cell populations stimulated by peptide-pulsed DCs included small numbers of tetramer-positive cells. Slightly stronger responses were recorded with apoptotic tumor cells as TA sources (14, 15). However, severalfold-larger, tetramer-positive T cell populations were generated in stimulations with DCs loaded with anti-MICA-opsonized tumor cells (Fig. 1), with expected variabilities in the magnitudes of T cell expansions among TA and donors (Fig. 2 A-E). Tetramer-binding corresponded to the TA expression profiles of the various melanoma, ovarian, and breast cancer lines used as immunogens and was absent when melanoma MZ2 cells, which were negative for the typed TA, were used as antigen source. No or minimal responses were seen with MICA^- mel526 cells or with isotype control Ig-treated tumor cell lines (data not shown) (Figs. 1 and 2 A-E). Using apoptotic tumor cells together with anti-MICA produced no synergistic effect (data not shown). Independent of the mode of stimulation, tetramer-binding CD8 T cells were uniformly CD45RO⁺ and CD27^- and showed heterogeneous expression of CD28, thus exhibiting an effector-memory phenotype (data not shown).

Fig. 1.

Enhanced cross-presentation of MART-1, gp100, tyrosinase, NY-ESO-1, and NY-BR-1 TAs to CD8 T cells by autologous DCs loaded with anti-MICA-opsonized tumor lines. Dot plots show the proportions of HLA-A2-TA peptide tetramer-positive CD8 T cells among bulk populations that were primed by DCs pulsed with relevant peptide (DC+peptide) or loaded with apoptotic (DC+apoTU), control IgG-treated (DC+TU+IgG) or anti-MICA-opsonized (DC+TU+mAb 2C10) melanoma, ovarian, or breast cancer cells. (Top) Shown is the lack of tetramer-binding, with the exception of the HLA-A2-MART-1 peptide tetramer, by unstimulated control CD8 T cells. T cells were stained with phycoerythrin-conjugated tetramers (y axis) and anti-CD8 (x axis), respectively, and analyzed by using two-color flow cytometry. Percentages of tetramer-positive T cells were derived from quadrant statistics. The MART-1, gp100, and tyrosinase and NY-ESO-1 and NY-BR-1 responses are representative of seven and five separate experiments, respectively, which include those shown in Fig. 2.

Fig. 2.

Consistently increased frequencies of TA-specific CD8 T cells after stimulation of T cell cultures established from healthy donors (A-E) and cancer patients (F-H) by anti-MICA (mAb 2C10)-opsonized melanoma (MEL) (A-C, F, and G), ovarian (OT) (D and H), or breast tumor (BT) (E) cell line-loaded DCs. HLA-A2-TA peptide tetramer (MART-1, gp100, tyrosinase, NY-ESO-1, and NY-BR-1) staining data (plotted as percent of tetramer-positive CD8 T cells) are shown for T cells derived from three or four donors. Also shown are tetramer stainings of unstimulated control CD8 T cells, and of T cells stimulated by DCs pulsed with relevant antigenic peptide or loaded with apoptotic (apo) or IgG1 isotype control-treated tumor cells. In F-H, DCs and CD8 T cells were derived from two patients with metastatic melanoma and a patient with ovarian cancer, respectively.

The potential significance of these results was scrutinized by using DCs and CD8 T cells derived from HLA-A2⁺ tumor patients. These patients included each three individuals with metastatic melanoma or stage III-IV ovarian cancer. T cell responses were recorded against gp100, tyrosinase, and NY-ESO-1, respectively. As with the healthy donors, patient DCs sensitized with anti-MICA-coated A2058 melanoma or OVCAR-3 ovarian tumor cells primed CD8 T cell responses that were quantitatively superior to those elicited by the control antigen-presenting cells (Fig. 2 F-H). Thus, there was no quantitative deficiency associated with DCs and CD8 T cells from tumor patients, confirming that the in vitro approach was both effective and practical in enhancing DC cross-presentation of antigens from tumors of diverse histological origins.

Superior Functional Proficiency and Antitumor Reactivity of Cross-Primed CD8 T Cells. For functional analysis of CD8 T cells primed by two consecutive stimulations with melanoma A2058-loaded DC, T cell clones derived by tetramer staining-enabled sorting and unsorted bulk T cell lines were tested in peptide titration and cytotoxicity assays by using antigen processing-deficient T2 cells and tumor cell targets with defined TA profiles, respectively. At effector-to-target cell ratios of 10:1, all of six MART-1-specific T cell clones efficiently lysed peptide-pulsed T2 cells (effecting 50% specific lysis at peptide concentrations of 0.07-0.45 nM) and the mel526 melanoma line (specific lysis ranging from 46.8% to 67.7%) but not the MART-1^- A375 melanoma line (data not shown and ref. 37). Similarly, the bulk T cell lines lysed the HLA-A2 and TA-matched targets mel526 and Malme-3M in the absence, but not in the presence, of the anti-MHC class I mAb W6/32. No cytoxicity was scored against HLA-A2^- A2058 cells (Fig. 3A Left and Right). Corresponding results were obtained with T cells cross-primed by DCs loaded with ovarian OVCAR-3 or breast tumor MDA-453 cells, which lysed the HLA-A2⁺ and NY-ESO-1- or NY-BR-1-matched target cells HTB-77 and HTB-21 but not the HLA-A2^- lines OVCAR-3 and MDA-453, respectively (Fig. 3 B and C Left and Right). Notably, all T cell lines also responded strongly in a T cell antigen receptor-dependent manner against HLA-A2⁺ targets that were negative for the expression of the profiled TA (Fig. 3 Center). These results implied recognition of other TA and, possibly, of minor antigens. Further evidence was obtained by comparisons of the proportions of CD8 T cells primed by OVCAR-3-loaded DCs that were positive for tetramer staining or for IFN-γ secretion after short-term stimulation with HLA-A2-matched OT HTB-77 cells. In three independent experiments, 21-29% (mean 25%) of the T cells were positive for surface IFN-γ capture, whereas only 1.1-3.4% (mean 2.2%) bound the HLA-A2-NY-ESO-1 peptide tetramer (Fig. 4 A and B) (data not shown). Stimulation with HLA-A2^- OVCAR-3 cells gave negative results (Fig. 4C). Hence, the presence of T cells reactive against tumor cell antigens other than NY-ESO-1 that were at least partially shared among different tumor cells was highly probable, although it could not be precluded that alloreactivity due to HLA mismatching contributed to the overall antitumor cytotolytic and INF-γ responses.

Fig. 3.

Broad antitumor reactivities of CD8 T cell lines cross-primed by DCs loaded with anti-MICA-opsonized tumor cells. Shown are cytolytic responses of T cell lines (plotted as percent of specific lysis) generated by stimulation with melanoma (A2058)-loaded (A), ovarian (OVCAR-3)-loaded (B), or breast cancer line (MDA-453)-loaded (C) DCs against HLA-A2⁺ tumor lines with (Left) or without (Center) expression of the defined TA. Lysis was inhibited by anti-HLA class I mAb W6/32. No cytotoxicity was scored against HLA-A2^- tumor lines used for DC-loading (Right). All assays were done in triplicate, with deviations ≤3%. Data are representative of three independent experiments. E:T, effector-to-target cell ratios.

Fig. 4.

Substantial proportions of cross-primed CD8 T cells produce IFN-γ upon tumor cell stimulation. T cells were primed by DCs loaded with anti-MICA-opsonized ovarian tumor OVCAR-3 cells (HLA-A2^- NY-ESO-1⁺). (A) Staining with HLA-A2-NY-ESO-1 peptide tetramer. (B) IFN-γ production after T cell stimulation with ovarian tumor HTB-77 cells (HLA-A2⁺ NY-ESO-1⁺). (C) Lack of response against HLA-A2^- OVCAR-3 cells. Data shown are representative of three independent experiments.

To attain a more accurate qualitative assessment of the antitumor responses, cross-priming was tested in an autologous HLA-A2⁺ NY-ESO-1⁺ ovarian cancer setting. Moreover, using surgical material allowed for a direct comparison of cryopreserved uncultured tumor cell suspensions (OT cells) with a cultured tumor line (OT140) as antigen sources. Similar to allogeneic tumor lines, stimulations with OT- or OT140-loaded DCs yielded proportions of NY-ESO-1 tetramer-positive CD8 T cells that were ≈3-fold larger than those recorded with peptide-pulsed DCs (Fig. 5A). However, the former T cells disproportionally exceeded the latter in their capacity to respond against autologous OT and OT140 cells, which was evident in strong and negligible cytolytic and INF-γ responses, respectively (Fig. 5 B and C). Moreover, large differences (≈20-fold) were apparent by comparing the proportions of tetramer-positive and IFN-γ-secreting CD8 T cells among populations cross-primed by opsonized OT or OT140 cell-loaded DCs (Fig. 5 A and B). Comparable results were obtained with uncultured tumor cells and autologous DCs derived from two additional ovarian cancer patients. Thus, altogether, these results indicated that DCs loaded with anti-MICA-opsonized tumor cells were effective in cross-priming efficient and multivalent antitumor CD8 T cell responses that were quantitatively and qualitatively superior to those elicited by peptide-pulsed DCs. As an antigen source, cryopreserved ovarian tumor cells were as effective as the ascites-derived OT140 line, suggesting that the use of primary tumor cells is a viable alternative to cell lines that may alter or lose expression of TA over extended periods of time.

Fig. 5.

Enhanced cross-priming of autologous antitumor CD8 T cell responses. (A and B) HLA-A2-NY-ESO-1 peptide tetramer stainings and IFN-γ capture assays of CD8 T cells primed by peptide-pulsed DCs (Left) or by DCs loaded with anti-MICA-opsonized autologous cultured OT140 (Center) or uncultured OT (Right) cells. (C) Lysis of the autologous HLA-A2⁺ NY-ESO-1⁺ OT140 and HLA-A2⁺ NY-ESO-1^- CRL-2183 tumor cells by the CD8 T cells primed by peptide-pulsed (Left) or OT140- (Center) or OT cell- (Right) loaded DCs. Data are derived from triplicate assays with deviations not exceeding 3% and are representative of three independent experiments. E:T, effector to target cell ratios.

Cross-Priming of Tumor-Reactive CD4 T Cells. In the development of tumor vaccines, it is desirable that they elicit combined CD4 and CD8 T cell responses. Stimulation of two HLA-DP4-restricted and NY-ESO-1-specific CD4 T cell clones (MS#55 and MS#8) with autologous DCs loaded with anti-MICA-opsonized ovarian OVCAR-3 tumor cells resulted in ≈1.5- and 3-fold-higher IFN-γ responses than exposure to peptide-pulsed DCs (data not shown). Thus, the cross-priming protocol used for the generation of TA-specific CD8 T cell responses also enhanced MHC class II antigen presentation. Corroborating results were obtained in the autologous ovarian tumor setting, which was matched for HLA-DP4⁺ by using OT140 for DC-loading. After short-term stimulation with OT140, the proportions of IFN-γ-secreting CD4 T cells among cross-primed bulk lines were substantially larger than those from the control experiments using peptide-pulsed DCs (Fig. 6A). Of 13 clones derived by sorting of IFN-γ⁺ cells from cross-primed T cell cultures, 4 were specific for NY-ESO-1, whereas 9 exhibited unrelated specificities because they were stimulated by OT140-loaded but not by peptide-pulsed DCs (Fig. 6B). Thus, similar to CD8 T cells, targeting tumor cells to DCs by anti-MICA opsonization was an efficient means for generating broadly TA-reactive effector CD4 T cell responses.

Fig. 6.

Cross-priming of tumor reactive CD4 T cells. (A) Robust vs. minimal IFN-γ capture responses after short-term stimulation with autologous tumor cells of CD4 T cells primed by tumor cell-loaded (Right) or peptide-pulsed (Left) DCs, respectively. (B) IFN-γ responses of two CD4 T cell clones established by sorting the IFN-γ capture positive population. The functional phenotypes of the clones OT#4 and OT#9 were representative of four and nine other clones, respectively. IFN-γ release assays were done in triplicate, and supernatants were pooled for analysis by ELISA. All results are representative of at least three independent experiments.

Conclusions

The experimental approach described takes advantage of the frequent and substantial expression of MICA on most if not all types of epithelial tumors for in vitro antibody-mediated tumor cell targeting to DCs to promote TA cross-presentation and T cell priming. The results demonstrate that both CD8 and CD4 T cells can be efficiently stimulated to respond against autologous tumor cells. The T cell responses elicited are of much greater breadth and magnitude than those primed by peptide-pulsed DCs. This improvement presumably results from favorable effects of antigen processing and from the ability of tumor-loaded DCs to present complex assortments of TA. Thus, this immunization approach, which can be readily standardized, lends itself as a platform for TA discovery, the generation of tumor-reactive T cells for adoptive T cell therapy, and DC vaccine development for a large variety of human tumors with diverse tissue origins. Of particular relevance in these areas are premalignant lesions and tumor stem cells, which have not been extensively altered by immunoediting and may thus express unselected TA repertoires. Preliminary evidence suggests that MICA can be associated with these early stages of tumor development. Moreover, it will be of interest to investigate whether application of an anti-MICA antibody or a derivative thereof can elicit antitumor responses by DC cross-priming of T cells in vivo.