Adenosine A_2A Receptors Intrinsically Regulate CD8⁺ T Cells in the Tumor Microenvironment

Solid tumors produce high concentrations of adenosine in response to hypoxia, cell necrosis, and the rapid metabolism of extracellular adenine nucleotides by ecto-nucleotidases expressed on tumor cells, tumor cell exosomes, and T regulatory cells (1–4). Adenosine engages four adenosine receptor subtypes: A₁, A_2A, A_2B, and A₃. The adenosine A2A receptor (A_2AR) is the predominant subtype found on T cells, and is induced upon cell activation (4, 5).

A_2AR signaling inhibits innate and adaptive immune responses (5, 6). Global deletion of A_2ARs facilitates activation of CD8⁺ T cells and enhances rejection of certain tumors that were genetically engineered to be highly sensitive to cytotoxic T-cell killing due to overexpression on tumor cells of MHC-I molecules (7). A_2AR deletion also enhances lymphoma killing and the effectiveness of an antilymphoma tumor vaccine (8). Hence, adenosine has been viewed as an inhibitor of T-cell–mediated tumor surveillance (9, 10), and blockade of lymphocyte A_2ARs has been advocated to facilitate tumor immunotherapy. Curiously, global deletion of A_2ARs did not affect the growth of B16F10 melanomas or MB49 bladder carcinomas that were not modified by genetic engineering (7, 11), despite the fact that these tumors produce immune cell activation (12). This might occur because adenosine levels are high in solid tumors, and A_2AR signaling can inhibit activation-induced death of T cells and thus facilitate their survival (13). It is also possible that some of the effects of global A_2AR deletion on tumor growth are due to disinhibition of tumor macrophages, dendritic cells (DC), or natural killer (NK) cells.

In contrast with the failure of global A_2AR deletion to inhibit B16F10 growth, reduced adenosine production due to deletion of CD73, an ecto-enzyme that converts AMP to adenosine, was found to consistently enhance antitumor adaptive immune responses (6, 14–17). This could be in part due to the involvement of A_2BRs in tumor suppression (11) or to differential effects on immune cell function caused by moderately reducing A_2AR stimulation by deleting CD73 as opposed to eliminating A_2AR signaling by deleting receptors.

In the current study, we evaluated the effects on B16F10 melanoma growth and tumor-associated T-cell survival of: (i) global A_2AR deletion, (ii) LckCre-mediated T-cell selective deletion of floxed A_2ARs, and (iii) adoptive cotransfer of T cells to tumor-bearing mice with and without A_2ARs. The results indicate that T-cell–specific A_2AR deletion does activate T cells, but can also lead to reduced numbers of tumor-associated T cells and an increase over time in the growth rate of large solid tumors. Hence, some degree of A_2AR signaling is needed for maintenance and effector differentiation of tumor-associated T cells. Opposing effects of A_2AR deletion to enhance T-cell activation but to reduce effector cell numbers in solid tumors provide an explanation for why global deletion of the A_2AR causes inconsistent effects on tumor growth.

Animal experiments were approved by the ACUC of the La Jolla Institute (La Jolla, CA). B16F10 cells stably expressing luciferase were obtained from Caliper Life Sciences. MB49 bladder carcinoma cells were from Dr. Timothy Ratliff of Purdue University (West Lafayette, IN). MB49 Bladder carcinomas were characterized as indicated by Luo and colleagues (18) and further tested at the time of experimentation for adherence, freeze thaw viability, growth properties, and mouse MHCI expression, without further authentication. Ovalbumin-expressing B16F10 cells produced as described previously (19) were a gift of Dr. Stephen Schoenberger of the La Jolla Institute. Ovalbumin and luciferase-expressing B16F10 cells were obtained from Dr. Andreas Limmer of the University of Bonn (Bonn, Germany). Both ovalbumin-expressing melanoma cell lines were received within 6 months of experimentation and evaluated at the time of experimentation by morphology, adherence, freeze thaw viability, growth properties, mouse MHCI expression before and after IFNγ treatment, cell surface expression of MHCI/Ova peptide complexes, and antigen-specific recognition of TRP2 or OVA peptides by respective transgenic T cells. B16F10 cells were cultured in R5F (RPMI-1640 medium containing 10% heat-inactivated FBS, 2 mmol/L l-glutamine, 1 mmol/L sodium pyruvate, 50 U/mL penicillin, and 50 μg/mL streptomycin). Tumor cells were injected into mice after reaching 60% to 80% confluence. A_2AR KO mice produced by Chen and colleagues (20) on a mixed genetic background were backcrossed onto C57BL/6. Six-week-old C57BL/6J, EGFP+, and OT-I Rag^−/− mice were purchased from Jackson Laboratories, crossed with Adora2a^−/− mice, and used for experiments after being acclimated for 2 to 6 weeks. LckCre⁺ mice (21) were obtained from Taconic (B6.Cg-Tg(Lck-cre)1Cwi N9) and used to create Adora2a^f/f-LckCre^+/− mice. Tail DNA from all mice was genotyped (Transnetyx, Inc.) to detect the presence of Cre recombinase and to quantify by qPCR lckCre-mediated excision of floxed Adora2a DNA. Global versus lck-mediated Cre expression was found to increase the amount of excision by >20-fold in tail DNA. Hence, qPCR was used to exclude from experiments occasional mice with non-lymphoid deletion. As further evidence of lymphoid-selective deletion, we have shown previously by qPCR that thymocyte expression of A2AR mRNA in lckCre/Adora2a^f/f mice is only deleted after thymocytes activate lck (22). Yellow or aqua fluorescent reactive dyes were from Invitrogen. SIINFEKL-loaded H2K^b tetramers with human β-2 microglobilin were provided by the NIH tetramer core facility. Fluorescent antibodies used in this study, their sources, and dilutions are listed in Supplementary Table S1.

Single-cell suspensions from indicated tissues were prepared by sequential pressing through 100 and 40 μm cell strainers. Dead cells were removed from tumor samples by Ficoll gradient centrifugation at 2,000 rpm (900 g) for 20 minutes at room temperature. After RBC lysis (Biolegend) of spleen samples, remaining cells were washed and resuspended in R10F, and counted in a Z2-Coulter particle counter (BeckmanCoulter). Cells (3–5 × 10⁶) were preincubated for 10 minutes in 100 μL FACS buffer with antibody to block Fc receptors. Each sample tube received 100 μL fluorescently labeled antibody cocktail and was incubated for 30 minutes at 4°C in the dark. Cells were analyzed using an LSRII equipped with 4 lasers or a LSR Fortessa equipped with 5 lasers and FACS Diva software (BD Biosciences). Live/dead fixable yellow, aqua, or blue (Invitrogen) was used to exclude dead cells before analysis. Flow-cytometric data were analyzed using FlowJo software (9.5.3 version, TreeStar Software Inc.).

B16F10 or MB49 cells (10⁵) were injected into the right flanks of mice. B16F10 melanoma cells expressing luciferase were injected into Adora2a^f/f –LckCre^−/+ and used for in vivo imaging. Tumor volumes were measured using digital calipers and calculated as height × width²/2. Luciferase activity was determined using an IVIS 200 Bioluminescence imager (Caliper Life Sciences) after intravenous injection of 1 mg D-Luciferin (Caliper Life Sciences) in 100 μL PBS to validate that tumor size differences were not due to infiltration of host cells. To measure tumor metastasis, 3 × 10⁵ B16F10 melanoma cells expressing luciferase were injected intravenously into mouse tail veins and luciferase activity was measured in the lungs 1 and 2 weeks later. After measuring luciferase activity, lungs were removed, photographed, and weighted to validate that luciferase activity correlated with tumor mass.

B16F10 cells (10⁵) expressing ovalbumin (B16F10-OVA) were injected into mouse flanks and allowed to expand for 16 days. Mixtures of 3 × 10⁶ OT-1 Rag^−/− and 7 × 10⁶ OT-1 Rag^−/−Adora2a^−/− cells were injected intraperitoneally. Greater numbers of OT-1 Adora2a^−/− cells were included in the mixture because A_2AR deficiency substantially reduced their numbers. On days 3 or 5, tumors and spleens were harvested and stained for analysis by flow cytometry. For adaptive transfer experiments, 10⁷ OT-1 Rag^−/− or OT-1 Rag^−/−Adora2a^−/− cells were injected intraperitoneally into the mice bearing B16F10-OVA tumors established for 2 weeks. Tumor growth was measured after T-cell transfer and on day 21. Mice were sacrificed and single-cell suspensions from tumors and spleen were analyzed for Annexin V staining, cell surface CD44 and CD127 expression, and cell number and density.

In prior studies, global deletion of A_2ARs failed to slow the growth rate of B16F10 melanomas transplanted into syngeneic mice (7, 11). In the current study, we performed similar experiments in mice inoculated with B16F10 melanoma or MB49 bladder carcinomas and confirmed that A_2AR deletion failed to decrease the rate of growth of either tumor; in fact, the growth rates of both tumors were significantly increased at days 14 to 18 after inoculation as the tumors became large (Fig. 1A). By preparing single-cell suspensions of tumors grown for 18 days after tumor inoculation, we next determined whether increased B16F10 growth was associated with reduced accumulation and/or impaired function of particular immune cell types within the tumor. Adora2a deletion significantly reduced the frequencies of CD8⁺ T cells (Fig. 1B) but not the frequencies of CD4⁺ T cells (Fig. 1B), myeloid cells, or CD11b^dimCD11c+ cells (Fig. 1C). Adora2a deletion also caused a significant increase in frequencies of NK1.1+TCRβ- cells (henceforth referred to as NK cells; Fig. 1C). Therefore, we calculated the density in tumors of NK and T cells by dividing the absolute numbers of these cells by tumor volume. Figure 1D shows that a large reduction in CD8⁺ T-cell density is associated with an increase in NK cell density in tumors. Local intratumoral injection of an irreversible A_2AR blocker, 5-amino-7-[2-(4-fluorosulfonyl)phenylethyl]-2-(2-furyl)-pryazolo[4,3-∈]-1,2,4-triazolo[1,5-c]pyrimidine (FSPTP) also reduced CD8 T-cell density but not CD4 T-cell or NK density within tumors, suggesting that local effects rather than global effects of Adora2a deletion are responsible for reduced T-cell numbers (Fig. 1E) and these effects are not dependent on elevated NK cell density. Additional experiments will be required to determine whether the increase in NK cell density in tumors caused by global A_2AR deletion is due to a cell intrinsic effect of A_2AR deletion on NK cells. Cell surface expression of CD44, KLRG1, and PD-1 was significantly lower in tumor-associated CD8⁺ T cells isolated from A_2AR-deficient mice as compared with tumor-associated CD8⁺ T cells from control animals (Fig. 2A). However, expression of CD25 tended to increase in A_2AR-deficient CD8⁺ T cells, suggesting that CD8⁺ T cells in tumors are activated but fail to become effector–memory cells in the absence of A_2ARs (Fig. 2A, top). CD4⁺ T-cell effector differentiation (as measured by CD44 and KLRG1) was also significantly inhibited in the absence of Adora2a (Fig. 2A, bottom). After global A_2AR deletion, CD80 expression on tumor-associated APCs increased, whereas CD86 expression and NK markers were unchanged (Fig. 2B and C). We also evaluated phenotypic markers in tumor-associated myeloid APCs such as MHCII, PD-1, and PD-L1, which regulate T-cell activation and CD39, which mediates tolerogenic activity of DCs by converting immunostimulatory ATP to ADP and AMP. A_2AR deficiency did not cause significant changes among any of these markers except for a reduction in cell surface PD-L1 expression (Supplementary Fig. S1). PD-L1, although inhibitory for T-cell activation, can be upregulated by inflammatory signals. It is possible that a reduction in the production of inflammatory cytokines due to reduced T-cell accumulation and activation contributes to reduced PD-L1 expression in A_2AR-deficient mice. Overall, these results suggest that decreased CD8⁺ T-cell infiltration and effector–memory differentiation in A_2AR^−/− mice are not due to APC inactivation. In fact, myeloid-selective deletion of A_2ARs decreases melanoma growth and increases the number of tumor-associated T cells and NK cells (23).

As in mice with B16F10 tumors, in mice with solid MB49 carcinomas A_2AR deletion reduced CD8⁺ T-cell frequency and expression of the effector–memory marker CD44 (Fig. 3A and B). One possible explanation for the reduction in CD8⁺ T cells in the tumors of A_2AR^−/− mice is reduced expression of CXCR3, which is required for activated T cells to home to inflamed sites. We did not observe any reduction in CXCR3 expression after A_2AR deletion (Fig. 3B). Furthermore, local inhibition by intratumoral injection of the A_2AR antagonist FSPTP, but not by selective A_2BR blocker ATL-801, also significantly reduced the frequency of tumor-associated CD8⁺ T cells (Fig. 3C), suggesting that as in melanomas, A_2AR signaling facilitates the accumulation CD8⁺ T cells within bladder tumors as well.

It is notable that A_2AR blockade consistently increased the frequencies of tumor-associated NK cells (Figs. 1B and 3A and C). This observation agrees with findings by Beavis and colleagues (2) who found that blockade or global deletion of A_2ARs reduced lung metastasis of CD73-expressing tumors by increasing NK cell activity and numbers, presumably by blocking A_2AR-mediated effects of high adenosine in the tumor.

Because global A_2AR deletion activates APCs (Fig. 2B), we hypothesized that A_2AR signaling helps to maintain T-cell numbers in the solid tumor microenvironment in a T-cell intrinsic manner. To evaluate the effects of cell-intrinsic A_2AR signaling on tumor growth and on T-cell responses, we crossed mice with a floxed Adora2a gene with mice expressing Cre recombinase under control of the Lck promoter. Adora2a^f/f –LckCre^+/− mice have normal numbers of thymic T-cell precursors in the absence (22) or presence (Supplementary Fig. S2) of solid tumors, suggesting that Adora2a^f/f –LckCre^+/− mice have normal T-cell development compared with littermate controls. However, T-cell selective deletion of A_2ARs markedly accelerated the growth rate of tumors after they reached a volume >500 mm³ (Fig. 4A and B; see Supplementary Fig. S3 for results of individual experiments). These findings suggest that paradoxically, some degree of A_2AR signaling in T cells is required to mount an optimal antitumor immune response in large solid tumors. A_2AR signaling increases cAMP production (24). It has been suggested that a temporary increase in cAMP levels may be required for T-cell activation (25). However, A_2AR deletion failed to affect the expression of CD69 in tumor-associated T cells (Supplementary Fig. S4). We reasoned that the absence of A_2AR signaling in the tumor microenvironment might cause T cells to polarize toward a regulatory phenotype. Figure 4C shows that T-cell A_2AR deletion does not enhance regulatory T-cell differentiation in the tumor or tumor draining lymph nodes. Therefore, we measured CD25 expression in antigen-experienced T-cell populations. The deletion of the A_2AR from lymphocytes increased CD25 expression in tumor-associated/antigen-experienced CD44^hiFoxp3⁻ CD4⁺ T cells and did not affect CD8⁺ T cells or lymph node CD44hiFoxp3- CD4⁺ T cells (Fig. 4D). These data suggest that deletion of A_2AR signaling does not hamper T-cell activation in tumors.

We next considered the possibility that A_2AR signaling sustains normal numbers of tumor-associated T cells. We found previously that A_2AR signaling, by activating PKA, reduces the activity of the PI3K/Akt pathway (22). This suppresses TCR-mediated downregulation of antiapoptotic CD127, which is upregulated in long-lived effector–memory cells and required for their maintenance. LckCre-mediated deletion of A_2ARs significantly reduced CD127 expression in antigen-experienced T cells in the tumor (Fig. 5), and significantly reduced the frequencies of tumor-associated T cells (Fig. 6A and B) but not NK cells (Fig. 6C). T-cell–selective A_2AR-deletion also significantly reduced the frequency of CD44⁺ effector–memory T cells in tumors, but not lymph nodes (Fig. 6D and E). Figure 6F shows a reduction in A_2AR^−/− T-cell density in tumors. Hence, although A_2AR activation during TCR stimulation inhibits T-cell activation, the data suggest a role for adenosine in maintaining effector T cells within the tumor microenvironment. These opposing effects of A_2AR deletion to enhance T-cell activation but to reduce effector cell numbers provide an explanation for why global deletion of the A_2AR causes small or inconsistent effects on tumor growth.

Global A_2AR deletion significantly reduces the development and peripheral maintenance of naïve T cells (22). Although Lck-mediated Adora2a^f/f deletion did not affect thymic progression of T cells, it did cause a decrease in the number of naive T cells in the periphery. This decrease in the naïve T-cell population may contribute to reducing numbers of T cells in tumors after global or LckCre-mediated deletion of Adora2a^f/f. Also, reduced naïve T-cell numbers in mice lacking T-cell A_2ARs could be a consequence of high tumor burden rather than to an intrinsic effect of A_2AR signaling. To evaluate in vivo competition and phenotypic differentiation of antigen-specific T cells lacking or expressing A_2ARs in the same tumor microenvironment, we performed adoptive cotransfer experiments. When cotransferred into the same host-bearing B16F10-OVA tumors, the proportion of Adora2a^−/−OT-1 T cells was markedly decreased in the tumor relative to Adora2a^+/+OT-1 T cells (Fig. 7A). A_2AR deletion also reduced cell surface expression of PD-1 (Fig. 7B), whereas CD25 expression was increased (Fig. 7C), a phenotype similar to what was observed after global deletion of the A_2AR (Fig. 2A). Figure 7D shows that A_2AR deletion caused a significant decrease in CD127 expression in both spleen and tumor-associated OT-I T cells.

To directly test the effects of Adora2a deletion on T-cell survival and effector–memory differentiation, we adoptively transferred Adora2a^+/+ and Adora2a^−/− OT-I T cells 2 weeks after establishment of B16F10-OVA tumors in congenic hosts. One week after adoptive transfer, we prepared single-cell suspensions from tumor tissue by ficoll gradient and measured cell surface staining of Annexin V as an apoptosis marker, CD44 as marker for effector–memory differentiation, and CD127 as mediator of T-cell survival. Tumor-associated but not splenic Adora2a^−/− OT-I T cells expressed significantly more Annexin V than Adora2a^+/+ OT-I T cells and this was associated with decreased expression of CD44 and CD127 within tumors (Supplementary Fig. S5). Transfer of either Adora2a^+/+ or Adora2a^−/− OT-I cells induced a transient decrease in tumor growth, suggesting that Adora2a^−/− cells are initially functional, but immunostimulatory effects of Adora2a deletion appear to be counteracted by reduced survival/effector–memory differentiation (Supplementary Fig. S5). Overall, these data show that in the tumor environment, A_2AR-deficient T cells have a survival disadvantage as compared with A_2AR-proficeint T cells.

Adenosine accumulates to high levels in solid tumors (1–4). A_2ARs on antigen-presenting cells and T cells, and A_2B receptors on antigen-presenting immune cells are primarily responsible for immunosuppression by adenosine (26, 27). Global deletion or intratumoral blockade of A_2BRs delays the growth of lung and bladder carcinoma and breast cancers, consistent with the immunosuppressive roles of these receptors (11). Curiously, global A_2AR deletion failed to slow the growth of bladder carcinomas and B16BL6 melanomas (7, 11), but was found to enhance the rejection of CL8-1 cells that were genetically engineered to be highly immunogenic by transfection with H-2K^b (7). These findings, and the results of the current study suggest that A_2AR blockade, as a strategy to treat cancer, is more complex than previously thought (9, 10). Although T cells are acutely activated by A_2AR deletion, long-term T-cell–mediated solid tumor rejection is compromised, likely as a result of impaired maintenance of T cells and reduced effector–memory differentiation in the tumor. It is important to point out, however, that the effect of A_2AR deletion on T-cell functions is not necessarily indicative of the effects of A_2AR blocking drugs that lessen, but do not eliminate A_2AR signaling. In this respect, a more clinically relevant reversible A_2AR antagonist rather than an irreversible antagonist may have different properties.

Ohta and colleagues previously showed that silencing of A_2A and A_2B adenosine receptors by siRNA in adoptively transferred, tumor-specific T cells significantly reduced lung metastasis of H2-K^b-expressing RMA T-cell lymphoma cells and improved the survival of tumor-bearing mice (7). The current study shows that Adora2a-deficient OT-I T cells do not slow the growth of solid B16F10 cells expressing ovalbumin, which can be recognized by OT-I T cells. The data suggest that: (i) antitumor effects of Adora2a deletion vary among different types of tumors; (ii) A_2BR signaling in tumor-associated T cells may contribute to adenosine suppression of T-cell activity; (iii) solid versus metastatic tumor growth may be differentially influenced by A_2AR deletion; (iv) highly immunogenic tumors may preferentially elicit antitumor effects of Adora2a deletion; and (v) siRNA silencing of Adora2a expression or pharmacologic inhibition may result in A_2AR residual signaling and contribute to tumor killing by sustaining T cells in the tumor microenvironment.

T cells go through an expansion phase after activation but many fail to survive either due to excess activation of inhibitory signals, or due to the absence of costimulation by certain cytokines or homing signals in the tumor (9). Interestingly, the absence of T-cell intrinsic inhibitory A_2AR signaling reduces numbers of tumor-associated T cells after 2 to 3 weeks of solid tumor expansion. Our recent findings indicate that A_2AR signaling can prevent TCR-induced downregulation of antiapoptotic CD127 (22). A_2AR deficiency significantly reduced the development and peripheral maintenance of naïve T cells. This decrease in the naïve T-cell population may contribute to reduced numbers of T cells in tumors after Lck-mediated deletion of Adora2a^f/f. However, the irreversible A_2AR antagonist, FSPTP, injected directly into solid tumors, also reduced tumor-associated T-cell numbers in wild-type recipients. Also, the ratio of antigen-specific A_2AR-deficient/A_2AR-proficient T cells decreased within the same tumor after adoptive cotransfer, suggesting a T-cell survival defect due do either deletion or irreversible blockade of A_2AR signaling in the tumor environment.

We show that CD127 expression by effector–memory cells in tumors is significantly reduced in T cells lacking A_2ARs, whereas CD25 expression is largely intact or even increased. This supports the notion that although A_2AR signaling acutely inhibits initial T-cell activation, A_2AR-dependent control of CD127 expression may be necessary for the maintenance of T cells after they differentiate into long-lived effector–memory cells. Consistent with this idea, we noted a significant impairment in the ex vivo survival of A_2AR^−/− T cells in response to IL7 (22). The pattern of tumor T-cell responses caused by deletion of the A_2AR is very similar to the effect of IL2 in tumor immunotherapy: although IL2 increases the activation and early expansion of T cells, it causes increased activation-induced death (28). Therefore, it is possible that CD127 deficiency and increased IL2 signaling due to increased CD25 expression impair T-cell survival in the tumor microenvironment.

A_2AR signaling may sustain tumor-associated T cells by inhibiting the PI3K/Akt pathway. As with A_2AR agonists, PI3K is also inhibited by rapamycin, which has been shown to increase numbers of long-lived effector–memory T cells in virally infected mice (29). Expansion of tumor-associated effector–memory T cells might account for the observations that despite their immunosuppressive properties, adenosine (this study) and rapamycin (30) have antiapoptotic effects on T cells in some tumors.

The A_2AR is upregulated in multiple immune cell types upon activation (5, 31, 32). Blockade or deletion of A_2ARs in NK cells (2) or myeloid cells (23) significantly inhibits tumor growth and metastasis and is associated with transactivation of cytotoxic lymphocytes. Consistent with these findings, the current study suggests that to avoid apoptosis of tumor-associated T cells, myeloid or NK cell targeted therapies should be considered as preferable targets for A_2AR deletion/blockade as a strategy for tumor immunotherapy. Alternatively, to optimize the beneficial effects of T-cell activation by A_2AR blockade for solid tumor killing, it may be necessary to find complementary strategies to enhance T-cell survival and effector–memory differentiation to counteract activation-induced cell death. These concepts apply not only to the potential use of adenosine receptor blockade or deletion to treat cancer, but also more broadly to the general use of T-cell activators.

No potential conflicts of interest were disclosed.

Conception and design: C. Cekic, J. Linden

Development of methodology: C. Cekic

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Cekic, J. Linden

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Cekic, J. Linden

Writing, review, and/or revision of the manuscript: C. Cekic, J. Linden

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Cekic

Study supervision: J. Linden

The authors thank Drs. Stephen Schoenberger and Michael Croft of the La Jolla Institute for Allergy and Immunology for helpful discussions, Dr. Jiang-Fan Chen of Boston University for A2AR^−/− mice, Dr. Timothy Ratliff of Purdue University for MB49 bladder carcinoma, Dr. Andreas Limmer of the University of Bonn for ovalbumin-expressing B16F10 cells, and Dr. Stephen Schoenberger of the La Jolla Institute for B16F10 cells expressing ovalbumin and luciferase.

This work was supported by NIH grant P01 HL073371 (J. Linden) and by American Heart Association post-doctoral fellowship (C. Cekic).

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.