|Home | About | Journals | Submit | Contact Us | Français|
The A2A adenosine receptor plays a critical and non-redundant role in suppressing inflammation at sites of hypoxia and tissue damage. The tumor microenvironment has high levels of adenosine as a result of hypoxia and ectopic expression of enzymes responsible for the generation of extracellular adenosine. Thus, we sought to determine the ability of A2A receptor null mice to immunologically reject tumors. We observed that mice lacking the A2A adenosine receptor showed significantly delayed growth of lymphoma cells when compared to WT mice. Furthermore, when immunized with a low dose of tumor or with an irradiated GM-CSF–secreting tumor vaccine, A2A receptor null mice showed significantly enhanced protection from a subsequent high-dose challenge from both immunogenic and poorly immunogenic tumor lines. This increase in protection was accompanied by an increase in the number of tumor-antigen-specific CD8 T cells at the vaccine-site draining lymph node. Finally, we found that A2A receptor null mice displayed more robust anti-tumor responses than WT mice when they were treated with a soluble B7-DC/Fc fusion protein designed to antagonize B7-H1-mediated co-inhibition. This combinatorial immunotherapy strategy could also be recapitulated with pharmacological A2A receptor blockade paired with B7-DC/Fc administration. In light of these data, we believe that blockade of the A2A adenosine receptor is an attractive target for tumor immunotherapy that synergizes with other immunomodulatory approaches currently in clinical trials.
Extracellular adenosine is a primordial signaling molecule that serves as a potent regulator of inflammation [1, 2]. This ubiquitous purine nucleoside is normally only found at nanomolar levels in the extracellular environment  due to the activity of the enzyme Adenosine Deaminase which rapidly catalyses the generation of inosine. However, the concentration of extracellular adenosine can increase dramatically at sites of trauma, hypoxia and inflammation . Damaged cells with compromised membrane integrity in these environments release into the surrounding interstitial space not only adenosine, but also AMP, ADP, ATP and other purine catabolites [5, 6]. These nucleotides are quickly broken down to form adenosine by the extracellular ectonucleotidase CD39 and CD73 [7, 8]. There are four known adenosine receptors, A1, A2A, A2B and A3, all of which modulate the activity of adenylyl cyclase to influence the generation of intracellular cAMP [9–11]. The A1 and A3 receptors are Gi protein linked and inhibit the generation of cAMP when stimulated, while the A2A and A2B receptors are Gs linked and activate the production of cAMP . These receptors are expressed widely throughout the body, including the heart, brain and lung , as well as epithelial cells  and lymphocytes .
It is becoming clear that this extracellular adenosine is not only the product of inflammation but can profoundly regulate inflammation. Seminal studies by the Sitkovsky group demonstrated the critical role of the adenosine A2A receptor in regulating inflammation . In spite of the fact that there are 4 adenosine receptors, knocking out the A2A receptor led to exaggerated and lethal inflammation in response to normally moderate stimuli. The A2A receptor is found on cells of hematopoietic origin and has been shown to inhibit inflammatory gene expression from activated neutrophils, macrophages, dendritic cells, NK cells and lymphocytes [17–20]. Indeed, acutely, A2A receptor activation is a potent inhibitor of TCR-induced cytokine production . Further, our group has shown that signaling via the A2A adenosine receptor can induce anergy in CD4 T cells despite strong TCR engagement and co-stimulation . In addition, adenosine has been shown to play a pivotal role in regulating the generation and functionality of regulatory CD4 T cells [22, 23] and the expression of inhibitory molecules such as LAG3 on activated T cells . Treatment of mice with agonists to the A2A adenosine receptor has been shown to decrease the severity of several models of autoimmunity, such as Inflammatory Bowel Disease (IBD) .
Inasmuch as adenosine is released into the inflammatory milieu, it is thought that signaling via the A2A receptor serves as a negative feedback loop preventing collateral tissue destruction by overzealous immune responses . However, many tumors have subverted the normally beneficial function of this signaling pathway to evade immune destruction [17, 25]. As a result of the highly hypoxic and/or inflammatory conditions, the tumor microenvironment has been shown to have high levels of adenosine which is an ideal situation for A2A receptor activation . In addition to being intrinsically hypoxic due to a high rate of metabolism and insufficient vascular development, many tumors ectopically express the ectonucleotidase responsible for the generation of extracellular adenosine [26, 27].
In this study, we show that mice lacking the A2A adenosine receptor have significantly decreased rates of tumor growth relative to their WT counterparts. Furthermore, A2A receptor null mice show increased responses to both lymphoma and melanoma tumor vaccines, resulting in a greater expansion of tumor-specific CD8 T cells and increased survival of tumor-bearing hosts. Finally, the combination of mitigating A2A receptor signaling and blocking checkpoint-mediated inhibition of anti-tumor immunity with a B7-DC/Fc fusion protein resulted in significantly increased survival in tumor-bearing mice. These results suggest that signaling via the A2A adenosine receptor plays a critical role in regulating tumor growth in vivo and that the inhibition of A2A receptor signaling, when combined with blockade of other inhibitory signals, offers an exciting possibility for enhancing the potency of tumor immunotherapy.
A2A receptor KO and C57Bl/6 WT mice were housed and bred in pathogen-free conditions prior to treatment on the Johns Hopkins School of Medicine East Baltimore campus. All mouse work was done in accordance with the Animal Care and Use Committee guidelines for Johns Hopkins University School of Medicine.
B16 and EL4 cell lines were maintained in RPMI supplemented with 10% fetal bovine serum and 50 μg/mL penicillin, 50 μg/mL streptomycin. Luciferase expressing EL4 cells were generated by retrovirally transfecting EL4 cells with luciferase construct (a generous gift of the Levitsky lab), followed by the enrichment by FACS sorting and purification by limiting dilution.
EL4-luciferase bearing mice were anesthetized by an i.p injection of room temperature ketamine (80 μg/g of mouse) and xylene (17.5 μg/g of mouse). Once unresponsive, the tumor inoculation site was shaved and the mice injected with 1.5 mg of luciferin-potassium salt dissolved in PBS (Caliper LifeSciences). After 15 min to allow complete diffusion of the substrate, mice were imaged using an IVIS system (Caliper LifeSciences).
Superficial inguinal lymph nodes from immunized mice were harvested and passed through a 60um strainer to create a single-cell suspension. Cells were then stained at 21°C with MHC class I OVA tetramer (Beckman Coulter), diluted in FACS buffer (PBS + 10% FBS) followed by surface staining for CD8 (BD). Cells were then analyzed on a FACSCalibur (BD), and post-collection analysis performed using FlowJo (TreeStar).
During initial tumor dose-finding experiments, it was discovered that both WT and A2A receptor KO mice completely rejected a subcutaneously (s.c.) dose of 103 live EL4 cells. In light of these findings, WT and A2A receptor null mice were primed with 103 live EL4 cells and rested for 60 days, whereupon they were challenged with 106 live EL4 cells along with age-matched naïve controls. This high tumor dose was previously determined to be lethal in both WT and A2A KO mice, albeit with a slight delay in the A2A receptor null mice. Tumor growth was monitored by palpation and caliper measurement. Mice were killed once tumor volume exceeded 100 mm3.
WT and A2A receptor KO mice were immunized with 106 irradiated (10,000 rad) GM-CSF–secreting B16 melanoma cells, producing ~300 ng GM-CSF in a 24 h period. After 27 days, immunized mice, along with naive age-matched controls, were injected i.v. via the lateral tail vein with 106 B16 cells. After 15 days, all mice were killed by CO2 inhalation and their lungs harvested, weighed and placed in Fekete’s Solution (62% EtOH, 3.3% Formaldehyde, 1.5% Glacial Acetic Acid). After a week of fixation, lungs were removed from solution and visible B16 nodules were counted under a dissecting microscope.
A B7-DC/Fc fusion protein was provided by Amplimmune (Rockville, MD). For experiments using WT and A2A receptor null mice, EL4 tumor-bearing mice were treated with 100 μg B7-DC/Fc or with 100 μg control mouse IgG, by i.p. injection on day 1 post-tumor inoculation and every 3 days following for the duration of the experiment. Pharmacological A2A receptor blockade was achieved by daily i.p. injections of 100 ul of the A2A antagonist ZM-241385 (Tocris) diluted in a 1:1:4 emulsion of DMSO, Cremophor and ddH20 at a concentration of 3 mg/ml. This treatment was supplemented with either 25 μg of control mouse IgG or 25 μg B7-DC/Fc.
Unless otherwise stated error bars were generated using SEM, and P values by a Mann–Whitney t test. Curves were compared using a 2-way ANOVA. All analysis was performed by Prism 5.0c (Graphpad).
Mice lacking the A2A adenosine receptor (A2A KO) have previously been described  and show no gross baseline immunologic phenotype compared to WT C57Bl/6 mice. However, in light of the increased severity of inflammation observed in the A2A KO mice in several models of autoimmunity, we decided to determine whether any difference could be detected in their ability to reject a syngeneic tumor challenge. To this end, A2A KO and WT control mice were injected s.c. with 106 luciferase expressing cells from the EL4 thymoma line. Tumor growth was monitored twice weekly by i.p injection of luciferase substrate, followed by whole-body imaging. WT mice showed steady tumor growth, as measured by photon flux, during the 26 day course of the experiment. All mice in the WT group were killed on day 26 due to unacceptably large tumor size. In contrast, mice lacking the A2A adenosine receptor showed a much-reduced rate of tumor growth compared to the WT controls, with several mice showing near complete rejection (Fig. 1a). The difference between the two groups was most evident on day 26 of the experiment, where there was almost a 100-fold difference in the luminescent output (Fig. 1b). However, there was a significant difference in the rate of tumor growth even at early timepoints of the experiment (Fig. 1c) *P < 0.05. These data suggest that absence of the A2A adenosine receptor can increase anti-tumor responses.
During the course of the initial tumor challenge experiments utilizing the non-luciferase expressing EL4 line, several doses of the tumor were used to determine the necessary number of cells required for successful tumor implantation. A dose of 106 live EL4 cells was found to be universally lethal in both WT and A2A receptor null mice, albeit with a slight increase in survival time in the A2A KO mice, while a dose of 105 cells proved lethal to the majority of WT mice but not to A2A receptor KO mice. However, mice from both genotypes easily rejected tumor doses of 104 and 103 cells (data not shown). We wondered whether these low-level tumor doses could act as a tumor vaccine, providing protection against a later high-dose challenge.
WT and A2A KO mice were initially primed with a non-lethal dose of 103 live EL4 cells, which both genotypes easily rejected. Sixty days post-priming, both groups of mice, along with naive controls, were challenged with 106 EL4 cells. Both naive WT and A2A KO mice showed a relatively fast rate of tumor growth, but there was a slightly higher rate of survival in the naive A2A KO group compared to WT controls despite the high tumor dose (Fig. 2a). Both primed WT and A2A KO mice showed greater overall survival compared to their naive counterparts. However, primed A2A KO mice showed the highest rate of overall survival of all experimental groups (P < 0.05) and had a significantly slower rate of tumor volume growth (Fig. 2b). These observations suggested that elimination of the anti-inflammatory A2A adenosine receptor signaling not only enhanced the ability of mice to initially reject tumor but also led to enhanced recall responses.
Given our findings concerning the enhanced rejection of tumor in the A2A receptor knockout mice, we wanted to demonstrate that this was associated with increased anti-tumor T-cell responses. Indeed, our findings concerning the enhanced protection upon rechallenge suggested that A2A receptor null mice would demonstrate robust responses to tumor vaccines. Irradiated GM-CSF–secreting tumor vaccines initiate potent immune responses against a wide range of tumor antigens without the need to immunize with individual, isolated peptides . While this immunization technique is most often used with a single irradiated transformed cell line secreting GM-CSF, it is also possible to induce a bystander response by mixing the irradiated GM-CSF–secreting line with a second, non-cytokine secreting, irradiated target line . To this end, we immunized both WT and A2A KO mice s.c. with a 1:1 mixture of irradiated ovalbumin (OVA) expressing EL4 cells and a GM-CSF-secreting B16 melanoma line. Generation of an OVA-specific CD8 T-cell response was monitored using MHC class I tetramer staining. Flow cytometric analysis of the injection site draining lymph nodes of immunized mice (Fig. 3a) 7 days post-immunization revealed significantly more OVA-specific CD8+ T cells in the A2A KO mice compared to WT mice (Fig. 3b) *P < 0.05. There was no significant increase in the percentage of tetramer-positive CD8 T cell in the draining LN of the immunized A2A receptor null mice compared to their WT counterparts, but the significantly larger size of the A2A KO draining lymph node resulted in the significant increase in the overall number of antigen-specific CD8 T cells. Thus, while both the WT and the A2A receptor null mice demonstrated a robust CD8 +T-cell response, the A2A KO mice were able to generate a markedly increased response to the tumor vaccine.
To further elucidate the potential for tumor vaccine enhancement by the loss the A2A receptor, we decided to again utilize an irradiated GM-CSF–secreting tumor vaccine, this time in the context of a systemic tumor model. To this end, we immunized WT and A2A KO mice with 106 irradiated GM-CSF–secreting B16 cells and rested the mice for 60 days. Immunized mice, along with naive controls, were injected i.v. with 106 B16 melanoma cells. After 17 days, the mice were harvested and the number of lung nodules counted and lungs weighed.
Naive WT mice showed extremely heavy tumor burden at the time of harvest, both in terms of the number of countable metastases and the wet-weight of the lungs. As would be predicted by our previous data, the naive A2A KO mice showed a significantly lower level of tumor growth than the naive WT mice (Fig. 4a, b). Thus, in addition to lymphoma, the A2A receptor null mice demonstrate an increased ability to reject melanoma, even in the absence of vaccine. Lungs from WT mice that were previously immunized with the irradiated GM-CSF–secreting B16 line showed a lower tumor burden than either naive group demonstrating the efficacy of the GM-CSF-secreting tumor vaccine in this model. As we would have predicted, the lungs from previously immunized A2A KO mice showed near complete protection from the growth of the transferred melanoma. Overall these findings demonstrate the ability of A2A receptor null mice to respond more vigorously to GM-CSF-secreting whole cell tumor vaccines than their WT counterparts. The significance of these data is compounded by the fact that the B16 melanoma line is considered poorly immunogenic , suggesting that even non-ideal vaccine strategies can be enhanced by elimination of A2A receptor signaling.
Extracellular adenosine is just one of many inhibitory signals that tumors use to evade the immune system. It is clear that tumors have developed mechanisms to activate co-inhibitory receptors on T cells to effectively shut down anti-tumor responses. For example, expression of the inhibitory receptor Programed Death (PD)-1 on activated T cells is a negative feedback signal that under normal conditions serves to limit proliferation and effector function following the resolution of an immune response, but is frequently subverted by tumors to inhibit infiltrating lymphocytes . The canonical ligand for PD1, B7-H1, is normally found on activated antigen presenting cells, including B cells and macrophages . B7-DC is a recently discovered, dendritic cell restricted PD-1 ligand that provides potent co-stimulation to T cells . The exact mechanism by which B7-DC enhances T-cell stimulation is still unknown, but it has been proposed to antagonize the inhibitory PD1/B7-H1 interaction, or interacts with an as-of-yet undiscovered receptor . Many tumors take advantage of the ability of B7-H1 expression to limit the proliferation and cytokine production of PD1 expressing T cells by ectopically expressing the ligand . We wondered whether we could further enhance an anti-tumor immune response by simultaneously mitigating A2A signaling and providing additional co-stimulation signal via B7-DC. To this end, we utilized a B7-DC/Fc fusion protein in the context of an EL4 thymoma challenge in both WT and A2A receptor null mice.
WT and A2A KO mice were injected with 106 EL4 thymoma cells s.c. and treated i.p. with 100 μg of either control IgG or B7-DC/FC on day 1 of the experiment and every 3 days following. IgG-treated WT mice demonstrated steady tumor growth, first showing tumor development by day 8 post-inoculation (Fig. 5a) and reaching the cutoff size for killing by day 37 (Fig. 5b). Also, at this dose of tumor, A2A KO mice showed steady tumor growth during the course of the experiment. A slight increase in overall survival was observed in IgG-treated A2A KO mice compared to similarly treated WT controls, but no difference was observed in the duration of tumor-free survival. WT mice receiving B7-DC/Fc treatment show a marked increase in the duration of tumor-free survival compared to IgG-treated groups (P < 0.001). Interestingly, the B7-DC/Fc-treated A2A KO mice demonstrated a significant increase in the duration of both tumor-free survival and overall survival compared to all other treatment groups, even the B7-DC/Fc-treated WT mice (P = 0.05). In addition, combining pharmacological A2A receptor blockade with B7-DC/Fc treatment resulted in a significant reduction in tumor growth (Fig. 5c). However, this combinatorial effect of pharmacological A2A receptor blockade and B7-DC/Fc administration was only observed during the first week of therapy, after which all groups receiving B7-DC/Fc showed similar reduced tumor growth. These data suggest that combination therapy attacking A2A signaling and co-inhibitory molecules might represent a novel avenue of enhancing tumor immunotherapy.
In this report, we demonstrate the ability of A2A receptor null mice to readily reject syngeneic tumors. In doing so, our data support the findings of Ohta et al.  This group, by demonstrating the ability of A2A KO mice to immunologically reject tumors, unequivocally established a role for adenosine in the tumor microenvironment as contributing to the ability of tumors to evade immune responses. Herein we reproduce and additionally extend these findings to demonstrate that mitigating A2A receptor signaling can enhance the efficacy of irradiated GM-CSF–secreting vaccines and additionally can synergize with agents which target co-inhibitory pathways. As such, our findings further support the development of A2A receptor antagonists as a means of enhancing anti-tumor immunity. Interestingly, A2A receptor antagonists have been safely tested in Phase III clinical trials for Parkinson’s disease paving the way for the initiation of clinical trials in cancer .
Extracellular adenosine is emerging as a potent regulator of inflammation at sites of extensive tissue damage and hypoxia. In cancer, the hypoxic environment of tumors leads to an increase in the local concentrations of extracellular adenosine . In addition, it is becoming clear that tumors upregulate enzymes that lead to the generation of adenosine [26, 27]. For example, it has recently been shown that the ecto-5′-nucleotidase CD73 that catalyzes AMP breakdown to adenosine is upregulated on the cell surface of many tumor cell lines. Indeed, knocking down the expression of CD73 on tumors led to enhanced anti-tumor responses in vivo . This increase in adenosine not only serves to inhibit anti-tumor immunity in terms of T-cell responses but also down modulates the activity of macrophages, neutrophils, dendritic cells and NK cells [17–20]. It is important to keep in mind that the extracellular adenosine found in hypoxic tumor environments can not only influence the host immune response to the tumor, but also the ability of a tumor to proliferate and survive. Several studies have suggested that extracellular adenosine signaling via the A2A adenosine receptor can negatively impact the viability of several tumor lines . These observations highlight the balance a tumor must strike between generating a suppressive microenvironment that can inhibit the ability of the immune system initiate tumor rejection, while still permitting the tumor itself to proliferate.
A2A adenosine receptor exhibits a similar expression pattern and shares some functional redundancy with the A2B adenosine receptor. Recent studies have shown that mice lacking the A2B adenosine receptor show an increased resistance to tumor growth, which is attributable to a decrease in VEGF production by tumor infiltrating lymphocytes . Our current study adds to the numerous previous reports demonstrating a central non-redundant role for the adenosine A2A receptor in mediating the immune-inhibitory effects of adenosine.
While there is a plethora of studies demonstrating the ability of tumor vaccines to both prevent and treat cancer in a wide variety of preclinical models, such findings have yet to be robustly translated in Phase III clinical trials. Of note, however, recently the FDA granted approval to the first therapeutic cancer vaccine. Sipuleucel-T is an autologous dendritic cell vaccine loaded with prostate-specific peptides which has been shown to be efficacious in extending survival in patients with metastatic castration-resistant prostate cancer . Our data suggest that inhibiting A2A receptor signaling has the ability to enhance GM-CSF-secreting whole cell vaccines. Recent Phase III trials using this tumor vaccine strategy failed to demonstrate efficacy in prostate cancer . However, there are a number of other clinical trials demonstrating safety and in some cases hints of efficacy using this vaccine platform . Of note, several studies suggest that the efficacy of irradiated GM-CSF–secreting tumor vaccines can be markedly enhanced when it is combined with agents which block co-inhibition. Specifically, in preclinical models, there is evidence to suggest that blocking PD-1 and CTLA-4 can increase anti-tumor responses in irradiated GM-CSF–secreting tumor vaccines treated mice . Our current studies suggest that blocking A2A receptor signaling represents an additional avenue to enhance vaccine efficacy. Our current data specifically employs GM-CSF-secreting “whole cell” vaccines. Based on these studies, we predict that A2A receptor blockade would also enhance other “whole cell” vaccines such as Sipuleucel-T. Of note, preliminary studies in our laboratory suggest that in addition to “whole cell” vaccines, A2A receptor blockade can also enhance the generation of effector cells in response to viral vaccines (data not shown).
Many tumor-associated antigens have been identified, and indeed various cancer vaccine strategies have been able to induce immunity to these antigens [42, 43]. What is clear, however, is that the tumor microenvironment can inhibit the ability of tumor-specific recognition to lead to immune-mediated tumor destruction. The tumor microenvironment is replete with inhibitory cytokines such as TGF-β, IL-10 and IL-13 [44–46]. Such cytokines not only inhibit anti-tumor Th1-mediated immunity, but also promote the generation of regulatory T cells and monocytic-derived cells [47, 48]. Accumulating data, including our current study, suggest that adenosine in the tumor micro-environment also represents a potent means by which tumors convert anti-tumor recognition by T cells into tumor-antigen-specific tolerance .
A2A receptor activation inhibits both CD4+ and CD8+ T-cell proinflammatory cytokine expression [15, 22, 49–51]. Alternatively, A2A receptor signaling does not appear to inhibit anti-inflammatory cytokine expression . In addition, adenosine acting via the A2A receptor has previously been shown to promote T-cell anergy even in the presence of costimulation . Further, A2A signaling has been shown to play a role in not only the generation of Foxp3 and Lag-3+ regulatory cells, but also in mediating their inhibitory effects [22, 23]. Thus, in the setting of infection, adenosine acts as an inhibitory metabokine  here preventing self-destruction by effectively down modulating immune responses. In the tumor microenvironment, cancer has usurped the adenosine-induced negative feedback loop to inhibit the anti-tumor immune response.
Co-inhibitory molecules are emerging as potentially potent adjuvants to cancer immunotherapy [53, 54]. Specifically, targeting CTLA-4- and PD-1-mediated negative regulatory pathways has shown promise in a number of clinical trials . Our current data demonstrate that administering B7-DC/FC fusion protein to A2A receptor null mice led to marked enhancement of anti-tumor immunity. In these experiments, mice were challenged with doses of tumor that led to death in the A2A receptor null mice and in the WT mice treated with B7-DC/FC. However, when we treated A2A receptor null mice with B7-DC/FC, there was a significant decrease in mortality. Such data suggest that simultaneously blocking the A2A receptor- and B7-H1-mediated co-inhibition can act additively/synergistically to enhance anti-tumor immune responses. Experiments are underway to determine whether A2A receptor blockade in combination with anti-CTLA4 or anti-PD1 also leads to enhanced responses.
We thank Amplimmune for their generous gift of B7-DC/FC for use in this study, and to Charles Drake and Ivan Borrello for their critical reviews. Funding for this study was provided by the NIH grant R01CA114227.
Conflict of interest Jonathan Powell is a Scientific Founder of Amplimmune, Inc.
Adam T. Waickman, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Angela Alme, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Liana Senaldi, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Paul E. Zarek, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Maureen Horton, Department of Pulmonary and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Jonathan D. Powell, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA.