MHC Class II–Transduced Tumor Cells Originating in the Immune-Privileged Eye Prime and Boost CD4⁺ T Lymphocytes that Cross-react with Primary and Metastatic Uveal Melanoma Cells

Uveal melanoma, the most common malignancy of the eye, has a 50% rate of liver metastases among patients with large primary tumors. Several therapies prolong survival of metastatic patients; however, none are curative and no patients survive. Therefore, we are exploring immunotherapy as an alternative or adjunctive treatment. Uveal melanoma may be particularly appropriate for immunotherapy because primary tumors arise in an immune-privileged site and may express antigens to which the host is not tolerized. We are developing MHC class II (MHC II)–matched allogeneic, cell-based uveal melanoma vaccines that activate CD4⁺ T lymphocytes, which are key cells for optimizing CD8⁺ T-cell immunity, facilitating immune memory, and preventing tolerance. Our previous studies showed that tumor cells genetically modified to express costimulatory and MHC II molecules syngeneic to the recipient are potent inducers of antitumor immunity. Because the MHC II–matched allogeneic vaccines do not express the accessory molecule, Invariant chain, they present MHC II–restricted peptides derived from endogenously encoded tumor antigens. We now report that MHC II–matched allogeneic vaccines, prepared from primary uveal melanomas that arise in the immune-privileged eye, prime and boost IFNγ-secreting CD4⁺ T cells from the peripheral blood of either healthy donors or uveal melanoma patients that cross-react with primary uveal melanomas from other patients and metastatic tumors. In contrast, vaccines prepared from metastatic cells in the liver are less effective at activating CD4⁺ T cells, suggesting that tumor cells originating in immune-privileged sites may have enhanced capacity for inducing antitumor immunity and for serving as immunotherapeutic agents. [Cancer Res 2007;67(9):4499–506]

Primary ocular or uveal melanoma is the most common malignancy of the eye and can be effectively treated with a variety of therapies, such as plaque radiotherapy, laser photocoagulation, transpupillary thermotherapy, trans-scleral resection, or enucleation of the tumor-bearing eye. Although these treatments limit the growth of the primary tumor and may partially preserve vision, they do not prevent the development of metastases, which occurs in ∼50% of patients with large tumors (1–3) and is universally fatal within ∼4 to 9 months of diagnosis (4). Although several treatments are available that increase median survival time to ∼15 months, metastatic uveal melanoma remains universally fatal (5, 6).

We are exploring immunotherapy as an alternative or adjunctive treatment for metastatic uveal melanoma. Whereas most tumors arise from somatic cells and express an array of antigens to which the host is tolerant (7), uveal melanomas arise in the eye, an immune-privileged site, and may express molecules to which the host is not tolerized (8, 9). Consequently, cell-based tumor vaccines composed of primary uveal melanomas may be effective at inducing antitumor immunity in patients with metastatic disease, provided that the immunity is cross-reactive with metastatic uveal melanoma cells. Our efforts are focused on the activation of CD4⁺ T lymphocytes, which have long been recognized as critical for optimal CD8⁺ T-cell–mediated immunity (10–13), either through their classic role as “helper” T cells that provide cytokine support for CD8⁺ T cells (14, 15) or through their induction of CD40 expression on dendritic cells (“licensing”), which in turn activate CD8⁺ T cells (16–18). CD4⁺ T cells are also essential for generating CD8⁺ T memory cells and for preventing CD8⁺ T cells from being tolerized (19–22). In addition, IFNγ production by CD4⁺ T cells facilitates tumor rejection by up-regulating tumor-expressed MHC molecules that improves CTL recognition, blocking neovascularization, and directly inhibiting tumor cell proliferation (23, 24).

To facilitate the activation of tumor-specific CD4⁺ T cells, we have made cell-based vaccines consisting of tumor cells that constitutively express MHC class I (MHC I) molecules, do not constitutively express MHC class II (MHC II) molecules, and are genetically modified to express CD80 costimulatory molecules and MHC II alleles that are syngeneic to the recipient. Because the MHC II–matched allogeneic “MHC II vaccine” cells do not constitutively express the MHC II accessory molecule, Invariant chain (Ii), they preferentially present endogenously synthesized tumor peptides rather than exogenously derived peptides (25). Expression of both CD80 and MHC II allows the vaccine to directly present antigens that prime MHC II–matched naive T cells (26–28). Recent studies indicate that tumor cell vaccines also activate CD4⁺ T cells through the process of cross-dressing, in which the MHC II-peptide complexes are transferred from the vaccine cells onto the surface of host dendritic cells (29). Therefore, tumor cell vaccines possess both a direct and indirect route of activating tumor-specific CD4⁺ T cells.

The MHC II vaccines have several advantages that favor the activation of tumor-specific CD4⁺ T cells. (a) MHC II⁺Ii⁻ cells present additional MHC II–restricted peptides that are not presented by MHC II⁺Ii⁺ cells (26–28, 30); therefore, the recipient is exposed to a larger repertoire of peptides than the repertoire presented by professional antigen-presenting cells (APC). (b) If the additional tumor peptides are novel, then the recipients will not previously have been exposed to them and hence will not be tolerized to them. (c) The vaccine cells synthesize many proteins that are potential tumor antigens; hence, multiple MHC II–restricted tumor peptides will be presented. (d) It is not necessary to identify or characterize tumor antigens because the MHC II–restricted peptides are constitutively processed and presented by the vaccine cells. (e) If the MHC II vaccines coexpress MHC I molecules that are shared with the host, then tumor-specific CD8⁺ T cells may also be activated.

MHC II vaccines prepared from mouse sarcoma and mammary carcinoma activate tumor-specific CD4⁺ T cells and mediate rejection of established primary (31) and spontaneously metastatic (32) tumors, respectively, validating the MHC II vaccine concept in animal models. We now report that MHC II vaccines prepared from human primary uveal melanoma cells activate naive CD4⁺ T cells from either healthy donors or uveal melanoma patients. Activated T cells produce high levels of IFNγ and cross-react with primary tumors from other patients and metastatic uveal melanoma cells. In contrast, vaccines prepared from metastatic uveal melanoma cells are much less efficient at activating CD4⁺ T cells, suggesting that tumor cells originating in immune-privileged sites may have enhanced capacity for inducing antitumor immunity and for serving as immunotherapeutic agents.

Cell lines. Primary uveal melanoma cell lines MEL202 and MEL270 and metastatic uveal melanoma cell line OMM2.3 were established from uveal melanoma patients and cultured as described for MEL202 (33, 34). Sweig, Jurkat, and SUM159PT cells were maintained as described (34). H358 lung adenocarcinoma cells were obtained from the American Type Culture Collection (ATCC) and cultured as recommended by the ATCC.

Blood samples and peripheral blood mononuclear cells. Blood samples were obtained from healthy donors and uveal melanoma patients by venipuncture. Patient 308 was a 68-year-old male and blood was collected 6 months after enucleation of the right eye. Patient M-185 was a 45-year-old male whose left eye was enucleated in 1998. In January 2002, liver metastases were diagnosed, and in March to April 2002, he underwent segmental hepatectomy and insertion of an intrahepatic artery catheter and received four weekly courses of 100 mg/msq of intra-arterial fotemustine. In June 2002, he was inoculated with irradiated (18 Gy) autologous liver metastatic cells. Subsequent delayed type hypersensitivity responses to M-185 cells were negative. In October 2002, he presented with brain, liver, lung, and pelvic metastases. In January 2003, he was given dendritic cells loaded with M-185 cell lysates. He died in August 2003. Peripheral blood mononuclear cells (PBMC) used in the current study were obtained in January 2003 before dendritic cell inoculation. PBMCs were isolated from whole blood by Ficoll gradient, stored frozen in liquid nitrogen until used, and cultured as described previously (35). Thawed PBMCs that were >75% viable as measured by trypan blue exclusion were used. All cell lines and procedures with human materials were approved by the Institutional Review Boards of the participating institutions.

Cell lines and PBMCs were human leukocyte antigen (HLA) typed and analyzed using MicroSSP HLA class I and II ABDR DNA typing trays and analysis software (One Lambda, Inc.) according to the manufacturer's instructions. Genomic DNA was isolated from 4 × 10⁶ cells using a DNeasy Tissue kit (Qiagen, Inc.) and amplified by PCR using recombinant Taq DNA polymerase (Fermentas, Inc.). HLA genotypes are referred to by their short-hand form (i.e., HLA-DRB1*0101 is DR1). Figure 1A shows the HLA alleles of the PBMCs and cell lines used in these studies.

The pLNCX2/DR1 and pLHCX/CD80 retroviral constructs, retrovirus production, transductions, and drug selections for DR1 (600 μg/mL G418; Sigma) and CD80 (75 μg/mL hygromycin; Calbiochem) were done as described (34). Transduced cells were grown in the same medium as their parental cells.

DR1-restricted HER2/neu peptide 776 to 790 (GVGSPYVSRLLGICL; refs. 36, 37) was synthesized at the University of Maryland Biopolymer Laboratory. HLA-DR-phycoerythrin (PE), CD80-PE, and FITC and PE isotype (mouse IgG2a) control monoclonal antibodies (mAb) were purchased from BD PharMingen; goat anti-mouse IgG-FITC from ICN; c-neu (Ab-2) from Oncogene; CD4-FITC and CD8-FITC from Miltenyi Biotech; and CD3-PE from eBioscience. Ii mAb PIN1.1 was prepared, and tumor cells and PBMCs were stained and analyzed by flow cytometry as described (34).

Western blot analyses for Ii were done as described (28, 35) using culture supernatant from hybridoma PIN1.1 at a 1:100 dilution followed by sheep anti-mouse-horseradish peroxidase (Amersham) at a 1:10,000 dilution.

Tumor cells were incubated for 48 h in culture medium supplemented with 100 units/mL recombinant human IFNγ (Pierce Biotechnology) and washed with culture medium to remove IFNγ.

PBMCs were thawed and resuspended in T-cell medium, and viability was determined by trypan blue exclusion. Viable PBMCs were plated at 1 × 10⁷/4 mL T-cell medium/well in six-well plates with 2 μg/mL HER2/neu peptide 776 and incubated at 37°C and 5% CO₂. After 5 days, nonadherent cells were harvested, washed twice with T-cell medium, counted, and plated at 1 × 10⁶/2 mL T-cell medium/well in 24-well plates with 20 units/mL recombinant human interleukin (IL)-2 (R&D Systems). Seven days later, nonadherent cells were harvested, washed, counted, and plated at 1 × 10⁶/2 mL T-cell medium/well without exogenous IL-2 and used the following day.

HER2/neu antigen presentation assays were done as described (35) with the following modifications: for endogenous HER2/neu presentation, stimulator cells (2.5 × 10⁴ per well) and HER2/neu peptide 776–primed PBMCs (5 × 10⁴ per well) in 200 μL/well T-cell medium were cultured at 37°C and 5% CO₂. After 48 h, the 96-well plates were spun at 1,200 rpm for 3 min, and the supernatants were assayed for IFNγ by ELISA. Values are the averages of triplicate data points with their SD. For exogenous HER2/neu peptide presentation, HER2/neu peptide 776 at 2 μg/mL was added to each well at the beginning of the 2-day culture period.

PBMCs (2.5 × 10⁶) were mixed with irradiated (10,000 Rads) transductants or parental tumor cells (2.5 × 10⁵) and cultured in 2 mL T-cell medium/well in 24-well plates at 37°C and 5% CO₂. After 3 days of culture, nonadherent cells were harvested, washed twice, and replated with 20 units/mL IL-2 in 24-well plates at 1 × 10⁶/2 mL T-cell medium. Five days later, nonadherent cells were harvested, washed, and plated at 1 × 10⁶/2 mL T-cell medium/well without IL-2 for an additional day. The resulting “primed” cells were then boosted with live transductants or parental cells as described (35) at a ratio of 1:2 boosting cells/PBMCs. PBMCs were depleted for CD8⁺ or CD4⁺ T cells as described (34, 35), except depletions were done on day 0 before priming. For experiments with patients' PBMCs, the same protocol was used except recombinant human IL-15 (20 ng/mL; PeproTech) was used instead of IL-2.

SD and Student's t test were calculated using Excel version 2002.

Primary and metastatic uveal melanoma MHC II vaccines. MHC II vaccines must express MHC II but not express Ii to allow endogenous tumor peptides within the endoplasmic reticulum access to the peptide binding groove of MHC II molecules. Because the MHC II and Ii genes are coordinately regulated by the MHC II transactivator (CIITA; ref. 38), tumor cells that constitutively express MHC II, or are induced by IFNγ to express MHC II, also express Ii and are not suitable for vaccine development. Conveniently, some uveal melanomas methylate the CIITA gene, preventing expression of either MHC II or Ii (39). To identify uveal melanoma cell lines that cannot be stimulated to express either MHC II or Ii, uveal melanoma cell lines derived from primary tumors (MEL202 and MEL270) or from liver metastasis (OMM2.3) were cultured with or without 100 units/mL recombinant human IFNγ. Cell surface expression of MHC I and MHC II was analyzed by flow cytometry (Fig. 1B) and expression of Ii was analyzed by Western blotting (Fig. 1C). Both primary and metastatic uveal melanoma cell lines expressed MHC I but did not express MHC II and were not induced by IFNγ to express either MHC II or Ii (Fig. 1B and C). Therefore, these cell lines are suitable candidates for MHC II vaccine development.

MHC II vaccine cells were generated by transducing the primary and metastatic uveal melanoma cell lines with retroviruses encoding HLA-DRB1*0101 (DR1) and/or the costimulatory molecule CD80 (Fig. 2). The transductants have maintained stable expression of their transgenes in culture for >6 months. Therefore, vaccine cells prepared from the primary (MEL202/DR1/CD80 and MEL270/DR1/CD80) and metastatic (OMM2.3/DR1/CD80) uveal melanoma cell lines stably express the transduced DR1 and CD80 genes but do not express Ii.

Vaccine cells express functional MHC II molecules that present endogenously synthesized tumor peptides. To ascertain that the transduced HLA-DR1 molecules of the vaccine cells are functional, DR1⁺ PBMCs from healthy donor 1 were primed to HER2/neu peptide 776 and boosted with MEL202/DR1/CD80 or OMM2.3/DR1/CD80 vaccine cells, which constitutively express HER2/neu (Fig. 3A). MEL202/DR1/CD80 and OMM2.3/DR1/CD80 vaccine cells boost HER2/neu–primed PBMC, which produce IFNγ (Fig. 3B). DR1 is the functional restriction element for the response because DR1⁻ parental cells (MEL202 or OMM2.3) are not effective. Coexpression of CD80 by the vaccine cells enhances antigen presentation because transductants without CD80 (MEL202/DR1 or OMM2.3/DR1) are not as effective APC as vaccine cells expressing both DR1 and CD80. Although the vaccine cells are MHC I allogeneic with respect to the PBMC, there is no allogeneic response because MEL202/CD80 cells do not activate T cells. Therefore, the transduced DR1 molecules of the uveal melanoma vaccines are functional antigen presentation elements for endogenously synthesized tumor peptides.

MHC II⁺CD80⁺ vaccine cells made from primary uveal melanomas prime and boost naive CD4⁺ T cells. The MHC II vaccines are designed to prime naive CD4⁺ T cells to novel, endogenously synthesized tumor antigens. To determine if the vaccines have this capability, DR1⁺ PBMCs from healthy donor 1 were cocultured (primed) with irradiated vaccine cells prepared from primary uveal melanomas (MEL202/DR1/CD80 or MEL270/DR1/CD80) and boosted with either parental or transduced uveal melanoma cells. Vaccines prepared from mammary carcinoma (SUM159PT/DR1/CD80; ref. 34) and lung adenocarcinoma (H358/DR1/CD80)⁵

were also used as boosting agents to determine specificity of the activated CD4⁺ T cells for uveal melanoma cells and to control for potential alloreactivity. Priming and boosting with either uveal melanoma vaccine induced significant IFNγ release (Fig. 4A). Activated T cells were highly specific for uveal melanoma cells and minimally reactive with breast and lung cancer cells. The minimal reactivity to breast and lung cells could be due to cross-reactivity to shared DR1-restricted antigens, such as HER2/neu, which are expressed by SUM159PT and H358 cells (Fig. 3A).⁵ Vaccine cell coexpression of CD80 enhanced the response, which was DR1 restricted, because transductants without CD80 or DR1 induced significantly less IFNγ (Fig. 4B). Despite the potential for alloreactivity against HLA-A3, which is expressed by the priming MEL202/DR1/CD80 cells and boosting H358/DR1/CD80 cells, the minimal reactivity with the lung cancer cells indicates that the vaccines do not stimulate a significant alloresponse. Similar results were obtained with PBMC from donors 2 and 3 (data not shown).

To identify the activated cells, PBMCs from healthy donor 1 were depleted for either CD4⁺ or CD8⁺ T cells before priming with MEL202/DR1/CD80 vaccine cells. Depletion of CD4⁺ T cells virtually eliminated IFNγ release, whereas depletion of CD8⁺ T cells had no effect (Fig. 4C). Therefore, MHC II–matched allogeneic uveal melanoma cells, expressing CD80 and HLA-DR alleles matched to the responding T cells, efficiently prime and boost healthy donor CD4⁺ T cells that are specific for uveal melanoma tumor cells.

Vaccines made from metastatic uveal melanomas are less efficient activators of CD4⁺ T cells. If the efficacy of the uveal melanoma vaccines prepared from primary tumors is due to their origin in the immune-privileged eye, then MHC II⁺CD80⁺ vaccines prepared from metastatic uveal melanoma cells may be less capable of activating CD4⁺ T cells. This hypothesis was tested using transductants prepared from metastatic OMM2.3 cells, which are derived from a liver metastasis of the same patient from which the primary MEL270 line was derived. Priming with OMM2.3/DR1/CD80 and boosting with OMM2.3/DR1/CD80 or MEL202/DR1/CD80 cells repeatedly gave <10% the amount of IFNγ as priming and boosting with MEL202/DR1/CD80 (Fig. 5A). The inability of the metastatic transductants to prime T cells could be due to individual variation between uveal melanoma patients 202 and 270. To eliminate this possibility, PBMC from DR1⁺ healthy donor 1 were primed in parallel with MEL270/DR1/CD80 or OMM2.3/DR1/CD80 transductants and boosted with MEL270/DR1/CD80 or OMM2.3/DR1/CD80 (Fig. 5B). Although priming with MEL270/DR1/CD80 vaccines induced IFNγ release, no IFNγ was detectable following priming with metastatic OMM2.3/DR1/CD80 cells, and only very low levels of IFNγ were produced following priming with MEL270/DR1/CD80 and boosting with the metastatic cells. Therefore, vaccines prepared from metastatic uveal melanoma cells are much less effective for activating T cells than vaccines prepared from primary uveal melanomas.

Uveal melanoma vaccines prime and boost CD4⁺ T cells that cross-react with metastatic and other primary uveal melanomas. If vaccines prepared from primary tumor cells are to be useful clinically, then they must prime CD4⁺ T cells that cross-react with metastatic tumor cells. To determine if the MHC II uveal melanoma vaccines have the capability to induce cross-reactivity, DR1⁺ PBMCs from healthy donor 1 were primed with MEL202/DR1/CD80 (Fig. 5C) or MEL270/DR1/CD80 (Fig. 5D) vaccine cells and boosted with MEL202, MEL270, or OMM2.3 transductants. Both vaccines prepared from primary tumor cells primed T cells that cross-react with metastatic tumor and with the other primary tumor cells. Similar results were obtained with PBMC from healthy donors 2 and 3 (data not shown). Therefore, vaccines made of primary uveal melanoma cells prime and boost T cells that are cross-reactive with other primary cells and with metastatic uveal melanoma cells.

MHC II uveal melanoma vaccines prime and boost T cells from patients with primary and metastatic uveal melanoma. To determine if the vaccines prime T cells from the blood of uveal melanoma patients, MEL202/DR1/CD80 vaccine cells were cocultured with DR1⁺ PBMCs from (a) a patient with primary uveal melanoma (patient 308); (b) a patient with metastatic uveal melanoma (patient M-185); or (c) healthy donor 1. PBMCs were collected from patient 308 6 months after enucleation of the tumor-bearing eye when he had no clinically detectable metastatic disease. Patient M-185 had extensive metastatic disease of the liver at the time the PBMCs were collected. Priming and boosting with the primary MEL202/DR1/CD80 vaccine maximally activated PBMC from all donors, with the healthy donor giving the highest level of IFNγ (Fig. 6A). Priming with MEL202/DR1/CD80 and boosting with the metastatic OMM2.3/DR1/CD80 vaccine similarly activated PBMC from the healthy donor and the two patients; however, the level of activation for all three donors was reduced relative to boosting with the primary vaccine. Therefore, MHC II⁺CD80⁺ uveal melanoma cell vaccines activate T cells from either healthy donors or uveal melanoma patients, and vaccines prepared from primary tumors are the most effective.

Uveal and skin melanomas are clinically and genetically distinct, although both tumors are derived from melanocytes. The differences may be due to the unique structure and function of melanocytes within the choroid of the eye. Although a few antigens are shared between skin and uveal melanomas (40, 41), the differences between the tumors suggest that immunotherapies developed for skin melanomas are unlikely to be effective in uveal melanoma patients. This hypothesis is supported by anecdotal reports on small numbers of metastatic uveal melanoma patients who have been included in skin melanoma clinical trials. Therefore, we are developing vaccines specifically for patients with uveal melanoma. Our MHC II uveal melanoma vaccines were designed to activate HLA-DR–restricted CD4⁺ T cells and thereby generate protective tumor immunity and immune memory in uveal melanoma patients who are at high risk of developing metastatic disease. The studies reported here show that MHC II uveal melanoma vaccines efficiently activate tumor-reactive, IFNγ-secreting, MHC II–restricted CD4⁺ T cells from healthy donors and uveal melanoma patients; therefore, these data are an important step toward showing the therapeutic efficacy of this approach.

Vaccines prepared from individual patients' primary uveal melanoma cells activated CD4⁺ T cells that cross-reacted with aggressive primary and metastatic tumor cells derived from other uveal melanoma patients, suggesting that the genetically modified, nonautologous vaccines may be useful reagents for stimulating tumor immunity in uveal melanoma patients. This cross-reactivity also suggests that a “cocktail vaccine” (Fig. 6B) may be the most effective and feasible approach for adapting the MHC II vaccines for clinical use. Vaccine cocktails would consist of a pool of four to six individual primary uveal melanoma cell lines each transduced with CD80 (or an equivalent costimulatory molecule) and a HLA-DR allele shared with the recipient. It is likely that some of the cell lines within the cocktail will share tumor antigens with the patient; hence, immunization with the cocktail will induce cross-reactivity with the patient's tumor. By maintaining a frozen bank of individual uveal melanoma cell lines, each transduced with a common HLA-DR allele and CD80, a vaccine cocktail could readily be customized for an individual patient. For example, for a HLA-DR4⁺DR7⁺ patient, four to six uveal melanoma cell lines from the bank of DR4⁺CD80⁺ and DR7⁺CD80⁺ transductants would be pooled.

CD4⁺ T cells facilitate tumor immunity by contributing to the activation of CD8⁺ cytotoxic T cells and enhancing the generation of long-term immune memory. They mediate their effects by interacting with CD8⁺ T cells and/or dendritic cells (14, 15, 18) and, therefore, once activated, do not need to directly react with tumor cells. Studies by others using uveal melanoma cells as priming agents for CD8⁺ T cells have given equivocal results. One study reported that uveal melanoma cells inhibited the activation of CD8⁺ T cells and attributed the inhibition to the lack of costimulation and MHC II expression (33). Two other studies reported the activation of CD8⁺ T cells; however, most of these responses were not MHC I restricted, and it was unclear if the activated CD8⁺ T cells were responding to uveal melanoma tumor antigens or to alloantigens (42, 43). Because the MHC II uveal melanoma vaccines described here coexpress MHC I, MHC II, and costimulatory molecules and activate tumor-reactive CD4⁺ T cells, it is likely that they will activate MHC I–restricted, tumor-reactive CD8⁺ T cells, provided that the vaccine cells share at least one MHC I allele with the recipient. MHC I matching will be feasible for at least 50% of patients because the HLA-A2 allele is expressed by ∼50% of uveal melanoma patients (44). Due to HLA polymorphism, a cocktail vaccine is likely to be partially MHC I allogeneic to the recipient. However, allo-MHC I differences neither adversely affect nor dominate the generation of tumor-specific CD4⁺ T cells (see Fig. 4A; ref. 35). In addition, the studies of others show that alloreactivity may be beneficial and have an adjuvant effect (45).

Uveal melanoma may be a particularly appropriate cancer to treat with active immunotherapy because of the progression of the disease. Typically, patients with primary uveal melanoma are diagnosed before they develop overt metastasis, and they have a lengthy disease-free interval before metastases become clinically detectable (3, 4). Therefore, once the primary tumor is eliminated, patients have minimal residual disease, reducing the likelihood of immune suppression, which is associated with large, bulky tumor burdens (46). In the absence of tumor-induced immune suppression, patients are more likely to actively respond to vaccination and to produce tumor-reactive CD4⁺ and CD8⁺ T cells.

The location of primary tumor in the eye may also be advantageous for immunotherapy against metastatic uveal melanoma. Because the eye is an immune-privileged site, tumor cells residing there may express molecules to which the host is not tolerized (8, 9) and, therefore, be inherently more immunogenic than tumor cells from nonprivileged sites. This hypothesis is supported by data showing that tumors formed in an immunodeficient environment are more immunogenic than tumors that develop in immunocompetent hosts (47, 48). Our finding that MHC II uveal melanoma vaccines made from primary tumor cells are significantly better activators of CD4⁺ T cells than vaccines prepared from metastatic cells also supports this concept.

Microarray and cytogenetic studies of primary uveal melanomas have identified chromosomal aberrations and genes that may be predictive of progression to metastatic disease. For example, primary tumors exhibiting monosomy of chromosome 3 are believed to be significantly more metastatic than primary tumors with normal chromosome numbers (49). Likewise, primary tumors that express high levels of E-cadherin and β-cadherin in combination with certain epithelial characteristics are thought to metastasize at much higher frequency than primary tumors with low levels of these gene products (50). If classification of primary tumors is sufficiently prognostic of tumor progression, then MHC II uveal melanoma vaccines may not only be useful for the treatment of established metastatic disease but could also be used as prophylactic reagents for the treatment of the 50% of patients with large primary tumors who are identified as being at high risk for developing metastatic disease.

Grant support: NIH R01 CA52527 and R01 CA84232 (S. Ostrand-Rosenberg) and R01 EY016486 (B.R. Ksander). Fight for Sight, Inc. postdoctoral fellowship and the following Dutch foundations: Rotterdamse Vereniging Blindenbelangen, Stichting Blindenhulp, Stichting Blinden-Penning, Stichting Dondersfonds, Stichting Nelly Reef Fund, Gratama Stichting, Stichting Admiraal van Kinsbergen Fonds, and Foundation ‘De Drie Lichten’ (J.J. Bosch). Department of Defense Breast Cancer Program predoctoral fellowship grant DAMD17-03-1-0337 (J.A. Thompson). MARC-U-STAR training grant NIH/NIGMS GM08663 (U.K. Iheagwara).

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.

We thank Virginia Clements for her excellent technical assistance, Wan-Ju Liu for transducing MEL270, Dr. Dean Mann for providing PBMC from healthy donors, and Drs. Peter Chen, Martine Jager, Kees Melief, and Jacob Pe'er for their helpful discussions.