A2AR eGFP reporter mouse enables elucidation of A2AR expression dynamics during anti-tumor immune responses

There is significant clinical interest in targeting adenosine-mediated immunosuppression, with several small molecule inhibitors having been developed for targeting the A2AR receptor. Understanding of the mechanism by which A2AR is regulated has been hindered by difficulty in identifying the cell types that express A2AR due to a lack of robust antibodies for these receptors. To overcome this limitation, here an A2AR eGFP reporter mouse is developed, enabling the expression of A2AR during ongoing anti-tumor immune responses to be assessed. This reveals that A2AR is highly expressed on all tumor-infiltrating lymphocyte subsets including Natural Killer (NK) cells, NKT cells, γδ T cells, conventional CD4+ and CD8+ T lymphocytes and on a MHCIIhiCD86hi subset of type 2 conventional dendritic cells. In response to PD-L1 blockade, the emergence of PD-1+A2AR- cells correlates with successful therapeutic responses, whilst IL-18 is identified as a cytokine that potently upregulates A2AR and synergizes with A2AR deficiency to improve anti-tumor immunity. These studies provide insight into the biology of A2AR in the context of anti-tumor immunity and reveals potential combination immunotherapy approaches.

There is significant clinical interest in targeting adenosine-mediated immunosuppression, with several small molecule inhibitors having been developed for targeting the A 2A R receptor.Understanding of the mechanism by which A 2A R is regulated has been hindered by difficulty in identifying the cell types that express A 2A R due to a lack of robust antibodies for these receptors.To overcome this limitation, here an A 2A R eGFP reporter mouse is developed, enabling the expression of A 2A R during ongoing anti-tumor immune responses to be assessed.This reveals that A 2A R is highly expressed on all tumorinfiltrating lymphocyte subsets including Natural Killer (NK) cells, NKT cells, γδ T cells, conventional CD4 + and CD8 + T lymphocytes and on a MHCII hi CD86 hi subset of type 2 conventional dendritic cells.In response to PD-L1 blockade, the emergence of PD-1 + A 2A R -cells correlates with successful therapeutic responses, whilst IL-18 is identified as a cytokine that potently upregulates A 2A R and synergizes with A 2A R deficiency to improve anti-tumor immunity.These studies provide insight into the biology of A 2A R in the context of antitumor immunity and reveals potential combination immunotherapy approaches.
Adenosine is an immunosuppressive metabolite that modulates antitumor immunity and is produced from the degradation of adenine nucleotides by the ecto-enzymes CD73 and CD39 on tumor cells, stroma and fibroblasts or through direct export from the intracellular compartment of cells undergoing hypoxia and/or stress [1][2][3][4][5][6][7] .Adenosine binds to four known G-coupled receptors, A 1 R, A 2A R, A 2B R, and A 3 R on immune cells with the majority of immunosuppressive effects thought to be mediated through A 2A R and A 2B R. A 2A R and A 2B R elicit immunosuppressive effects on tumor-infiltrating immune cells due to activation of adenylate cyclase, which results in increased intracellular cAMP.Pioneering studies by Sitkovsky and colleagues have demonstrated that extracellular adenosine potently modulates immune responses and antitumor immunity 2,3 .
Accordingly, enhanced anti-tumor responses are observed in A 2A R deficient mice 3,5,8,9 , an observation that led to the development of small molecule antagonists, which elicit anti-tumor immunity against solid tumors in mice 3,5,8,[10][11][12] .A 2A R antagonists have been evaluated in clinical trials against a range of solid tumors including renal cell carcinoma, non-small cell lung carcinoma, prostate, breast, and head and neck cancers 13 .Moreover, preclinical investigations have highlighted that targeting the A 2A R either genetically or pharmacologically enhances the efficacy of immune checkpoint blockade 5,8,14 or adoptive cellular therapy 3,12,[15][16][17][18][19][20][21] .Despite the intense interest in targeting this pathway, the mechanism of action is not completely understood.While A 2A R is known to be expressed on a range of immune cell types including CD8 + T cells, CD4 + T cells 22 , Tregs 23 , NK cells 24,25 , NKT cells 26 , and myeloid cells 27 , this has predominantly been determined from mRNA analyses due to a lack of an effective antibody recognizing this receptor that is suitable for flow cytometry.Therefore, a detailed analysis of the immune cell subpopulations expressing A 2A R have not been determined and, in addition, the factors driving A 2A R expression both within tumors at baseline and in response to therapy remain largely unknown.To overcome this, we describe a transgenic mouse model enabling the quantification of A 2A R through the expression of a GFP reporter gene.Using this model enables the expression of A 2A R on immune cells within solid tumors, draining lymph nodes (DLN) and the spleen to be assessed, allowing for analyses of A 2A R expression at baseline and in response to immunotherapies (anti-PD-L1, anti-CTLA-4 and IL-18) or chemotherapy (carboplatin).
This reveals that A 2A R is highly expressed in NK, CD8 + , CD4 + , NKT, and γδ T cells within tumors, with NK cells and conventional T cells constituting more than 50% of all A 2A R positive cells.Relatively limited A 2A R expression is observed on myeloid cells within the tumor, apart from cDC2s, and expression is absent from tumor-infiltrating B cells.When comparing the expression of A 2A R in the tumor and DLNs, T cells in the tumor uniquely express significantly more A 2A R than their counterparts in the DLN.We observe that whilst A 2A R expression is elevated in activated tumor antigen-specific CD8 + T cells in the DLN, A 2A R expression is fairly ubiquitous amongst CD8 + T cell subpopulations within the tumor, regardless of their antigen specificity or differentiation status.Treatment of tumor-bearing mice with anti-PD-L1 alone or combined anti-PD-L1 and anti-CTLA-4 leads to the emergence of a population of PD-1 + A 2A R -CD8 + T cells that correlate with therapeutic outcome.Transcriptomic analysis of A 2A R + cells reveals that the most apparent difference between A 2A R + and A 2A R - CD8 + T cells is reduced expression of STAT5 target genes in A 2A R + cells, a phenotype that is restored following anti-PD-L1 treatment.Lastly, these studies reveal that IL-18 is a potent inducer of A 2A R expression and that the therapeutic activity of IL-18 is accentuated in the context of A 2A R-deficient mice.In summary, we report the development of a mouse model that provides insight into the biology of A 2A R expression in the context of anti-tumor immunity and reveals potential mechanisms of action and combinatorial approaches for A 2A R blocking therapies.

GFP expression accurately maps A 2A R expression in A 2A R eGFP reporter mice
To characterize the A 2A R eGFP reporter mice and ensure that GFP was a bona fide readout of A 2A R expression we first examined the expression of GFP within the spleens of mice.Immunofluorescent analysis indicated detectable expression of GFP in the spleens of A 2A R eGFP reporter mice, particularly in the T cell-rich region of the arterioles (Fig. 1A).Flow cytometry analyses of splenocytes revealed that GFP expression was highest in NK cells that are known to express high levels of A 2A R (Fig. 1B) 24,28 and CD8 + CD44 + T cells, which is also consistent with a previous report 29 (Fig. 1C).FACS sorting of splenocytes from A 2A R eGFP reporter mice indicated that GFP + splenocytes expressed significantly higher levels of A 2A R mRNA (Fig. 1D), confirming that GFP expression correlated with increased expression of the A 2A R. Consistent with previous reports indicating that activation of T cells through the TCR led to increased A 2A R mRNA expression 30 , stimulation of splenocytes with anti-CD3/anti-CD28 led to a significant increase in GFP expression by both CD8 + and CD4 + T cells (Fig. 1E).Moreover, elevated GFP expression was associated with increased expression of PD-1 and CD44, highlighting that A 2A R expression was associated with an activated T cell phenotype in vitro (Fig. 1F).To confirm that A 2A R expression in reporter mice was functional we evaluated their response to adenosine receptor stimulation.Suppression of TNF by NECA, a pan adenosine receptor agonist, and reversal of this phenotype by SCH58261, an A 2A R antagonist, was observed following activation of splenocytes from A 2A R eGFP reporter mice or wildtype controls (Fig. 1G).To assess the impact of tumor antigen recognition on A 2A R expression, we generated anti-Her2 CAR T cells from A 2A R eGFP reporter splenocytes in line with our previous work using anti-Her2 CAR T cells 16 .Generation of CAR T cells led to increased expression of GFP relative to naïve T cells (Fig. 1H) and coculture of anti-Her2 CAR T cells with AT-3-, E0771-, or MC38-Her2 expressing tumor cells led to a further and significant induction of GFP in both CD8 + and CD4 + CAR T cells (Fig. 1I).
A 2A R is expressed in a broad range of immune cells within the tumor microenvironment A 2A R expression has been reported on a broad range of immune cells including conventional αβ T lymphocytes and NK cells but the expression of this receptor in the context of the tumor microenvironment has not been extensively investigated, and much less in a comparative analysis between subsets.Therefore, we examined the expression of GFP (A 2A R) within the tumor microenvironment.Having confirmed an important role for A 2A R in the control of AT3ova tumors (Fig. 2A), consistent with our previous observations 8 , GFP expression was determined in the tumors and draining lymph nodes of mice bearing AT3ova tumors (representative flow cytometry Fig. 2B, gating strategy; Supplementary Fig. 1A).This analysis revealed that, within the draining lymph nodes, A 2A R expression was most abundantly expressed on NK cells on a per cell basis (mean 74% positive), followed by γδ T cells (49.3% positive) and type I NK T cells (33.7% positive) (Fig. 2C).Expression of A 2A R was lower on CD8 + and CD4 + lymphocytes (4.7% and 5.4% positive respectively) and virtually absent on B lymphocytes (1.9% positive).However, over 50% of the A 2A R positive cells within draining lymph nodes were either CD4 + or CD8 + lymphocytes because of their higher overall frequency relative to innate lymphocyte subsets (Fig. 2D).We next investigated the expression of A 2A R within tumor-infiltrating lymphocytes.Notably, A 2A R expression was highly expressed in all subsets except for B lymphocytes (Fig. 2E), with NK cells, CD4 + lymphocytes and CD8 + lymphocytes together constituting over 50% of A 2A R + cells within the tumor (Fig. 2F).Comparison of the expression of A 2A R in these lymphocyte subsets indicated that A 2A R was significantly upregulated in tumor-infiltrating lymphocytes relative to their counterparts in the tumor-draining lymph node, but this was more apparent with conventional CD8 + and CD4 + lymphocytes (5 fold increased expression in tumor relative to draining lymph node) than with NK, NKT and γδ T cells (all ~1.5 fold increased expression in the tumor) (Fig. 2B, G).Within the tumors, NK cells expressed a significantly higher level of A 2A R than all other immune subsets analyzed on a per cell basis (Fig. 2H).A similar pattern of A 2A R expression was observed in MC38 and E0771 tumors, that is to say A 2A R was highly expressed on innatelike cells (NK cells, NKT cells, γδ T cells) isolated from either DLNs or tumors and was significantly increased on conventional T lymphocytes isolated from tumors relative to counterparts isolated from draining lymph nodes (Supplementary Fig. 1B-F, Supplementary Fig. 2A, B).To investigate the relationship between A 2A R and tumor progression we assessed the expression of GFP on each of these subsets within AT3ova tumors at days 10, 17, 21 and 28 post tumor inoculation.Interestingly, A 2A R expression was at its highest point within total CD45 + cells at day 10 post-tumor inoculation and gradually decreased over time (Fig. 2I).This effect was largely caused by reduced A 2A R expression within CD8 + T cells, whilst the expression of A 2A R in other subsets remained relatively stable.A similar reduction of A 2A R expression in CD8 + T cells over time was observed in E0771 tumors, albeit that the time course was shorter than in the AT3ova model due to the more aggressive growth of this tumor line (Supplementary Fig. 2C).These data suggest that early intervention with A 2A R antagonists may be critical for achieving maximum impact on CD8 + T cell mediated anti-tumor immunity.Given previous reports indicating a role for A 2A R expression on myeloid cells, we next assessed the expression of A 2A R on myeloid subsets isolated from either AT3ova or E0771 tumors using a panel including antibodies for CD11b, CD11c, F4/80, Ly6C, CD103, XCR1, CD64, and MHCII.These analyses revealed that, amongst all myeloid subsets, A 2A R expression was significantly higher on CD11b + cDC2 cells (defined as ), relative to all other myeloid cell subsets (Fig. 2J, Supplementary Fig. 2D).Interestingly the GFP + subset of cDC2s exhibited significantly higher expression of MHCII, CD80 and CD86 (Fig. 2K, L, Supplementary Fig. 2D), indicating that this population may represent a more mature subset of cDC2 cells with enhanced capacity to prime T cell responses.
A 2A R expression is indicative of CD8 + T cell activation status within draining lymph nodes but less so within tumors Given that A 2A R expression was significantly enhanced on tumorinfiltrating CD8 + and CD4 + lymphocytes within the tumor microenvironment, we further analyzed the phenotype of A 2A R + T cells relative to their A 2A R -counterparts.This analysis included the expression of markers associated with differentiation status (CD62L, SLAMF6, CD69, CD44), the immune checkpoint PD-1 and CD39, the ectoenzyme responsible for the breakdown of ATP to AMP.Notably, CD39 is also a recognized marker of tumor-reactive terminally differentiated CD8 + T cells and its expression limits anti-tumor immunity [31][32][33] .
A SIINFEKL-loaded tetramer reagent was also used to identify tumorantigen (Ova) specific CD8 + T cells.We first investigated the expression of A 2A R on tetramer + and tetramer -subsets.Within tumor-draining lymph nodes A 2A R expression was significantly upregulated on CD44 + tetramer + CD8 + T cells relative to other CD8 + T cell subsets (Fig. 3A), consistent with the notion that A 2A R was upregulated following activation of these cells by tumor antigens presented by APCs in the draining lymph nodes.However, within the tumors there was only a modest increase in A 2A R expression on CD62L -tetramer positive cells relative to CD62L -tetramer negative counterparts (Fig. 3B, Supplementary Fig. 1G).Within tumors, CD8 + T cells follow a pathway of differentiation from TCF7 + precursor exhausted cells to TCF7 − terminally differentiated cells.Given the technical complexities of TCF7 staining in GFP reporter mice due to quenching of GFP following nuclear permeabilization, we employed the gating strategy described by Beltra et al. using the markers CD69 and SLAMF6 whereby T cells transition from CD69 + SLAMF6 + > CD69 − SLAMF6 + > CD69 -SLAMF6->CD69 + , SLAMF6 −34 .Analysis of these distinct subsets indicated that A 2A R was modestly but significantly enriched within the CD69 + SLAMF6 + subset relative to SLAMF6-counterparts (Fig. 3C).Similarly, within E0771 tumor cells it was observed that A 2A R expression was significantly higher on SLAMF6 + CD39 − or SLAMF6 − CD39 + cells relative to SLAMF6 − CD39 − counterparts (Supplementary Fig. 2E).We next investigated whether these differences contributed to the decreased expression of A 2A R observed in CD8 + tumor infiltrating lymphocytes over time (Fig. 2I).We observed that in the course of tumor progression that the proportion of both the antigen specific tetramer positive population and the SLAMF6 + CD69 + progenitor population were progressively diminished (Supplementary Fig. 1H).
Given that these cell types were the highest expressers of A 2A R this partly accounts for the reduced expression of A 2A R in total CD8 + T cells over time.Moreover, on a per cell basis these cell populations elicited significantly higher expression of A 2A R at day 10 than other timepoints but there was no significant difference from day 17 onwards (Supplementary Fig. 1H).Therefore, the reduction in A 2A R expression in CD8 + T cells over time is partly explained by a reduced frequency of the tetramer positive and SLAMF6 + CD69 + cells and partly due to reduced expression of A 2A R by these subsets on a per cell basis at later timepoints.To confirm this using a complementary approach we determined the expression of A 2A R following treatment of mice with FTY720.FTY720 acts to block S1PR1 and thus prevent the egress of immune cells from lymph nodes into the tumor site.We reasoned that if A 2A R was expressed more highly on terminally differentiated cells that GFP expression would be increased following FTY720 treatment since this treatment would prevent cells egressing from the draining lymph node and replenishing the SLAMF6 + CD69 + population.FTY720 treatment was effective, as demonstrated by significantly reduced numbers of CD8 + and CD4 + T cells in the peripheral blood (Supplementary Fig. 1I).Within tumors, FTY720 treatment resulted in a reduction in the proportion of CD69 + SLAMF6 + (less-differentiated population) CD8 + T cells and A 2A R expression was significantly reduced (Fig. 3D), supporting our earlier observations that A 2A R expression was not enhanced on more terminally differentiated CD8 + T cells.
To further investigate this relationship, we determined the expression of A 2A R on PD-1 + and CD39 + subsets.Somewhat surprisingly and counter to our observations in vitro, A 2A R was expressed equally between PD-1 + / PD-1 − and CD39 + /CD39 − subsets, whereas PD-1 and CD39 were largely co-expressed.Thus, A 2A R was not co-expressed with other known inhibitory checkpoint receptors on CD8 + T cells suggesting that A 2A R blockade may modulate the function of distinct subsets of CD8 + T cells (Fig. 3E).
In contrast to CD8 + T cells, both CD39 and PD-1 were expressed to a significantly greater extent on CD4 + A 2A R + cells relative to their A 2A R − counterparts (Fig. 3F).This may reflect an increased expression of A 2A R on Treg cells, which are known to express high levels of CD39 and PD-1 in the tumor microenvironment [35][36][37][38] , particularly given CD4 + GFP + CD39 + cells were enriched for CD25 + cells relative to CD4 + GFP + CD39 − counterparts (Fig. 3G).However, this could not be confirmed with Foxp3 counterstaining due to the issues of GFP degradation upon nuclear permeabilization as aforementioned.Taken together, these results suggest that TCR-mediated activation/ differentiation may not be the major factor driving A 2A R expression within the tumor microenvironment on CD8 + T cells.
The efficacy of immune checkpoint blockade is correlated with the emergence of PD-1 + A 2A R -cells following therapy Given previous studies indicating that A 2A R blockade enhances the efficacy of conventional immune checkpoint blockade or chemotherapy 5,8,14 and the clinical interest in targeting this pathway, we next assessed the effect of anti-PD-L1, anti-CTLA-4, the combination of anti-PD-L1 and anti-CTLA-4 or carboplatin on the expression of the expression by each cell type isolated from AT3ova tumors.Data represented as mean ± SEM of 5 mice per group.****p < 0.0001, statistics determined by one-way ANOVA I Expression of GFP (A 2A R) in indicated cell types isolated from tumors at indicated timepoints.Numbers represent the GFP MFI.Data represent concatenated samples from n = 6 mice.J Expression of GFP (A 2A R) in indicated myeloid populations.ΔGFP expression calculated as GFP MFI minus the background fluorescence for each cell type as observed in WT mice.Data represented as the mean ± SEM from 24 mice per group, pooled from 4 independent experiments.****p < 0.0001, statistics determined by one way ANOVA.K Expression of MHCII and GFP on cDC2 cells (Gated as CD45 + TCRβ -NK1.1 -Ly6C -CD64 -CD11c + MHCII + F4/ 80 low CD103 -XCR1 -).L Expression of MHCII, CD80, and CD86 on total cDC2 (Red), GFP -cDC2 (Blue) and GFP + cDC2 (Green).K, L. Data concatenated from 6 individual mice.Source data are provided as a Source Data file.
A 2A R within the tumor microenvironment.As expected, immune checkpoint blockade therapy resulted in significantly reduced tumor weights and an increase in tumor-infiltrating CD8 + lymphocytes (Supplementary Fig. 3A, B).Immunofluorescent analyses confirmed that combination (anti-PD-L1 and anti-CTLA-4) therapy resulted in increased numbers of CD45 + cells infiltrating the tumors, and this was associated with an increased abundance of GFP + cells (Fig. 4A).
In terms of T lymphocyte phenotype, whilst the overall frequency of CD8 + A 2A R + (but not CD4 + A 2A R + ) T cells was increased by therapy due to an increase in the proportion of CD8 + T cells as determined by flow cytometry analyses (Fig. 4B), the expression of A 2A R on a per cell basis was not significantly modulated by these treatment regimens at day 7 post-therapy (Fig. 4C).Moreover, carboplatin treatment also failed to modulate A 2A R expression on CD8 + or CD4 + lymphocytes (Supplementary Fig. 3C).To further interrogate the impact of immune checkpoint blockade on A 2A R expression, and to investigate the possibility that A 2A R was transiently upregulated following treatment, further experiments were performed to determine the expression of GFP on CD8 + tumor-infiltrating lymphocytes (TILs) at days 2 and days 4 post-treatment with either anti-PD-L1 or anti-PD-L1 and anti-CTLA-4.Consistent with our analyses at day 7, no significant modulation of A 2A R expression was observed (Supplementary Fig. 3D).Although immune checkpoint blockade did not modulate A 2A R expression on a per cell basis, the expression of PD-1 was increased on CD8 + T cells as expected (Supplementary Fig. 3E, F) and further analysis of the activated PD-1 + cells that emerged following therapy revealed that anti-PD-L1 or anti-PD-L1 and anti-CTLA-4 treatment induced a significant increase of PD-1 + A 2A R -cells (Fig. 4D).Similarly, a significant increase in CD8 + PD-1 + A 2A R -cells (but not PD-1 + A 2A R + cells) were observed in MC38 tumors following treatment with anti-PD-L1 (Supplementary Fig. 3G).
Interestingly, the emergence of these cells was positively correlated with increased therapeutic efficacy in response to anti-PD-L1 single agent therapy (Fig. 4E, F).Lastly, the expression of A 2A R in draining lymph nodes was assessed in the context of therapy.In contrast to the lack of modulation of A 2A R expression within the tumor by anti-PD-L1 therapy, a significant increase in the proportion and absolute number of tetramer + A 2A R + CD8 + T cells within the draining lymph nodes was observed, suggesting that these cells may be more susceptible to adenosine mediated immunosuppression post-treatment (Fig. 4G).Finally, given that we observed the expression of A 2A R on myeloid subsets including cDC2s within the tumor microenvironment (Fig. 2J-L), we also assessed the impact of immunotherapy on the expression of A 2A R on myeloid cells.Interestingly, anti-CTLA-4 significantly enhanced the expression of A 2A R on cDC2 and, to a lesser extent, cDC1s, when combined with anti-PD-L1 (Supplementary Fig. 4A, B).CTLA-4 blockade has previously been shown to induce upregulation of CD80 and CD86 39 and thus we hypothesized that A 2A R expression may be indicative of an activated cDC2 population.To investigate this, we generated dendritic cells from bone marrow of A 2A R eGFP reporter mice.Consistent with our observations in tumors, a subset of cDC2 were identified to express A 2A R that also exhibited increased expression of CD80 and CD86 (Supplementary Fig. 4C, D).Furthermore, activation of bone marrow derived dendritic cells with poly IC resulted in cDC2 maturation, as evidenced by increased expression of MHCII, CD80 and CD86, which was concomitant with enhanced A 2A R expression (Supplementary Fig. 4C, D).To determine the consequence of adenosine signaling on bone marrow-derived dendritic cell (BMDC) function, they were cocultured with NECA, a pan adenosine receptor agonist, prior to coculture with naïve T cells in a T cell priming assay.These experiments revealed that adenosine signaling resulted in significantly attenuated ability of BMDCs to induce T cell proliferation (Supplementary Fig. 4E).

A 2A R + CD8 + TILs exhibit reduced STAT5 transcriptional signatures
To further investigate the phenotype of A 2A R + CD8 + T cells within tumors, we performed 3'RNA-sequencing on CD8 + GFP + and CD8 + GFP − tumor-infiltrating T cells both at baseline and in the context of anti-PD-L1 treatment.As expected GFP -and GFP + cells clustered differently as did anti-PD-L1 and non-treated samples (Fig. 5A).Analysis of differentially expressed genes between GFP + and GFP -cells confirmed that A 2A R was more highly expressed in CD8 + GFP + cells (Fig. 5B).Pathway analysis of differentially expressed genes revealed significant negative enrichments for "Interferon alpha response", "Interferon gamma response" and for the expression of STAT5 target genes based on a signature developed by Grange et al. 40 within A 2A R + cells (Fig. 5C, D, Supplementary Fig. 5).This gene signature included notable effector genes such as Ifng, Tnf, Prf1 and Gzmb.The association between A 2A R + status and reduced STAT5 target genes was of interest because of our previous observations that A 2A R agonists negatively regulate JAK-STAT signaling 16 and a previous report from Cekic and colleagues that IL7R signaling protects CD8 + T cells from adenosine mediatedsuppression 41 .Although Cekic et al. attributed the impact of IL7 signaling on adenosine-mediated suppression to the inactivation of FOXO1, the activation of STAT5 signaling by IL7 represents a possible complementary mechanism of action.Interestingly, treatment with anti-PD-L1 significantly enhanced STAT5 target genes in both A 2A R + and A 2A R -cells, such that there was no longer a significant difference in the expression of these genes between the two subsets (Fig. 5E).This infers that one consequence of anti-PD-L1 treatment is to overcome reduced STAT5 signaling in A 2A R + CD8 + T cells.

A 2A R is potently upregulated by IL-18 and limits the anti-tumor efficacy of IL-18
We next investigated the capacity of cytokines to upregulate A 2A R expression.Beginning with a broad screen of 23 cytokines on the expression of A 2A R on CD8 + T lymphocytes derived from splenocytes, we identified that amongst the most potent inducers of A 2A R expression were members of the IL-1 family of cytokines including IL-1β, IL-18 and IL-36 (Fig. 6A).We proceeded to investigate the impact of the top hits from this in vitro screen on the expression of A 2A R in tumorinfiltrating lymphocytes ex vivo.These results revealed that IL-12 and IL-18 increased the expression of A 2A R on multiple immune lineages including CD8 + T cells and NK cells (Fig. 6B, Supplementary Fig. 6A).Given these data and the clinical interest in IL-18 based therapies 42,43 , we further investigated the relationship between IL-18 and A 2A R. In repeat experiments we confirmed that IL-18 was a potent inducer of A 2A R expression in both murine T cells (Fig. 6C) and human NK cells (Fig. 6D).To investigate this further we engineered AT3ova and MC38 tumor cells to express IL-18.Despite successful transduction of both cell lines, only MC38 secreted detectable levels of IL-18 (Supplementary Fig. 6B).We therefore evaluated the growth of MC38-IL-18 or MC38 mCherry control tumors and observed a significant reduction in growth of IL-18 expressing tumors (Supplementary Fig. 6C).To investigate the relationship between IL-18 and A 2A R expression in this context we inoculated A 2A R eGFP reporter mice with either IL-18 expressing or control MC38 tumors and assessed the phenotype of tumor-infiltrating lymphocytes.This analysis revealed that CD8 + T cells, CD4 + T cells and NK cells derived from IL-18 expressing tumors exhibited significantly increased levels of A 2A R expression, whereas we observed no significant increase in PD-1 expression on CD8 + T cells (Fig. 6E, F).To evaluate whether A 2A R expression limited the efficacy of IL-18 therapy we inoculated control or IL-18 expressing MC38 cells into A 2A R deficient or wild-type mice.Strikingly, A 2A R deficient mice elicited more pronounced anti-tumor responses with significantly reduced tumor growth and increased survival relative to WT counterparts (Fig. 6G, H; Supplementary Fig. 6D).Depletion experiments revealed that the anti-tumor efficacy of IL-18 was dependent on both CD8 + T cells and NK cells (Fig. 6I).Lastly, to evaluate the interplay between IL-18 and A 2A R expression in a more clinically relevant system we determined the impact of recombinant IL-18 treatment on the expression of A 2A R within the context of established MC38 tumors.These experiments revealed that A 2A R expression was significantly increased on CD8 + T cells and NK cells following treatment with IL-18 (Fig. 6J).These data highlight the potential of combining IL-18 therapy with A 2A R blocking therapies such as small molecule antagonists and  A D PC1 (41.5% explained var.)PC2 (25.8% explained var.)

Discussion
Adenosine-mediated immunosuppression is a significant barrier to the formation of effective anti-tumor immune responses.Antibodies targeting the ectoenzymes responsible for the generation of adenosine, CD73 and CD39, and small molecule antagonists of the adenosine receptors A 2A R and A 2B R, have been developed for clinical application 1,44,45 .A 2A R can suppress several immune cell types; however, the pattern of expression of A 2A R within the context of an antitumor immune response has remained largely unknown due to a lack of robust flow cytometry antibodies.We therefore sought to develop a We initially confirmed the veracity of the A 2A R transgenic model through analysis of A 2A R mRNA expression in GFP + and GFP − cells.Moreover, in line with our expectations, analysis of splenocytes from these mice revealed that GFP (A 2A R) was abundantly expressed in cell types known to express high levels of A 2A R such as NK cells 24 and NKT cells 26 but was not expressed in B cells.Furthermore, GFP was strongly upregulated following T cell activation through the TCR or a CAR, in line with previous observations with A 2A R mRNA.We therefore proceeded to investigate the expression of A 2A R in the context of tumor-bearing mice.Consistent with our observations using splenocytes, A 2A R was highly expressed on all innate-like subsets evaluated including NK cells, NKT cells and γδ T cells.Tumor-infiltrating CD8 + and CD4 + T cells also expressed high levels of A 2A R and notably this expression was significantly higher than observed with CD8 + or CD4 + T cells isolated from spleen or draining lymph nodes.We originally hypothesized that this may relate to the activation status of the T cells given that TCR activation is known to drive expression of A 2A R and indeed this is what we observed using the A 2A R eGFP reporter mice in vitro.Whilst antigen (ova)-specific CD8 + T cells expressed significantly higher levels of A 2A R than non-ova-specific CD8 + T cells in draining lymph nodes, this difference was more modest within tumors.Moreover, within tumors A 2A R expression was similar between PD-1 + and PD-1 − subsets and CD39 + and CD39 -subsets, suggesting that within tumors the activation status of the CD8 + T cells is not a major predictor of A 2A R expression.Therefore, whilst TCR activation appears to be a major driver of A 2A R expression in CD8 + T cells within draining lymph nodes other factors appear to govern the expression of A 2A R within tumors.Characterization of CD8 + T cells by differentiation status revealed that A 2A R expression was modestly but significantly higher within the CD69 + SLAMF6 + precursor exhausted subset.This is intriguing given previous studies have indicated that A 2A R signaling is required for the maintenance of naïve cells, partly due to its ability to upregulate IL-7R expression 16,29,46,47 .Whether A 2A R signaling is required for the maintenance and/or expansion of CD69 + SLAMF6 + precursor exhausted CD8 + T cells within tumors remains to be determined but is one potential mechanism by which A 2A R signaling may actually promote a favorable differentiation status for responses to immune checkpoint blockade in some contexts.It will be interesting in followup studies to investigate factors that may influence A 2A R expression such as hypoxia and other local immunosuppressive metabolites found at high concentrations within tumors 48 .For example, systemic oxygenation has been shown to decrease expression of CD73, A 2A R and A 2B R and improve anti-tumor immunity although the cell types that reduced A 2A R expression under these conditions could not be determined in these studies as the determination was based upon mRNA analysis of bulk tumor tissue 6,49 .Amongst CD4 + T cells, A 2A R expression was highly abundant on CD39 + PD-1 + cells, potentially indicative of Treg cells that are known to express these immune checkpoints at high levels within tumors.However, this could not be definitively confirmed due to an inability to counterstain with a Foxp3directed antibody due to loss of GFP signal upon nuclear permeabilization.Future studies may confirm this by FACS sorting GFP + and GFP - cells and then performing analysis of transcription factor expression in each population.
One interesting observation from these analyses is that a subset of cDC2 cells was identified that expressed A 2A R.These cells exhibited higher expression of MHCII, CD80 and CD86 relative to GFP -counterparts, indicating that they may represent an activated subset of cDC2.Indeed, consistent with this, we observed that activation of cDC2s in vitro with poly IC led to a significant increase in A 2A R expression and in vivo, treatment with immune checkpoint blockade, particularly anti-CTLA-4, led to a significant increase in A 2A R expression on cDC2s.Expression of A 2A R and A 2B R on dendritic cells has been reported previously 50 and blockade of A 2B R or CD73 9,51 or myeloidspecific deletion of A 2A R or A 2B R has been shown to enhance antitumor immunity 27,52 in part through modulation of DC function 53 .More recently, it was shown that a component of the mechanism of action for the small molecule A 2A R antagonist AZD4635 was enhancement of cDC1 function 54 .To our knowledge the expression of A 2A R on cDC2 cells has not been previously reported.Interestingly we observed that A 2A R expression was largely restricted to a subset of cDC2 with increased expression of MHCII, CD80 and CD86 that was increased upon treatment with anti-CTLA-4.A 2A R may therefore represent a negative feedback loop that limits the costimulatory capacity of activated cDC2s.cDC2s have been shown to be critical in priming intratumoral CD4 + T cell responses and so it is interesting to postulate that A 2A R expression on these cells may consequently limit the formation of these protective responses 39 .In this study by Binnewies et al. it was shown that Tregs limit the activation and differentiation of cDC2s, which may be linked to the activation and increased expression of A 2A R observed in our study following anti-CTLA-4 blockade given we used the 9H10 clone that is known to induce Treg depletion.Therefore, in future studies, it would be of interest to isolate A 2A R + and A 2A R -cDC2s and compare their functional capacity to prime CD8 + and CD4 + T cell responses ex vivo.
Although we observed increases in A 2A R expression in cDC2s and, to a lesser extent, cDC1s following immune checkpoint blockade, it was somewhat surprising that we did not observe increases in A 2A R expression within tumor-infiltrating T cells following PD-L1 blockade or treatment with anti-CTLA-4 and anti-PD-L1.Previously, an assessment of A 2A R mRNA following PD-1 blockade revealed an approximately 2-3 fold increase in A 2A R expression within tumor-infiltrating CD8 + T cells 8 .However, we did not observe this in our current study Fig. 6 | IL-18 treatment leads to the upregulation of A 2A R which consequently limits anti-tumor immunity.A CD8 + T cells were isolated from C57BL/6 A 2A R eGFP or C57BL/6 WT mice and stimulated with plate-bound anti-CD3 and soluble anti-CD28 (2 μg/ ml) mAbs plus indicated cytokines (100 IU/ ml IL-2, or 50 ng/ ml other cytokines) for 72hrs.Data represent the mean from 2 individual technical replicates.B AT3 ova tumor-infiltrating lymphocytes were isolated and cultured overnight with indicated cytokines (50 ng/ ml).Data represent the mean ± SD of 3 technical replicates from a representative experiment of n = 3. C A 2A R GFP splenocytes were treated for 3 days with anti-CD3/ anti-CD28 and, where indicated, IL-18 (10 ng/ ml).Histogram overlay of concatenated samples (left) or mean ± SD of 2 (Naïve) or 3 (activated) replicate samples from a representative experiment of n = 3 (right).D Human NK cells were stimulated with 20 IU/ml IL-2 (control) or IL-2 and IL-18 (50 ng/ ml) for 16 h.Expression of A 2A R mRNA relative to GAPDH housekeeping gene.Data represent the relative expression (±SD) of triplicate samples.E-J C57BL/6 A 2A R eGFP mice (E, F), C57BL/6 WT mice or A 2A R -/-mice (G, H) were injected with 1 × 10 6 MC38 mCherry or MC38 mCherry-IL-18 expressing tumor cells.E, F Expression of PD-1 and A 2A R (GFP) by indicated subsets.Flow cytometry plots of concatenated samples from a representative experiment.F Data represented as the mean ± SEM of 12 (MC38 Ctrl) or 14 (MC38 IL-18) mice per group pooled from 2 experiments.Data were normalized relative to the average MFI of the relevant cell population in control tumors within each experiment.G Tumor growth represented as the mean ± SEM pooled from 2 representative experiments (n = 15 WT MC38, 17 WT MC38-IL-18, l0 A 2A R KO MC38, 9 A 2A R KO MC38-IL-18) H Survival of mice with 100mm 2 tumor size used to designate survival.(n = 15 WT control, 16 WT IL-18, 7 A 2A R KO control, 8 A 2A R KO IL-18).I C57BL/6 WT mice or A 2A R −/− mice were injected with 1.5 × 10 6 tumor cells and where indicated treated with anti-CD8 or anti-GM-1, n = 5-13 per group.J At day 12 post-tumor inoculation 1 µg IL-18 was administered intra-tumorally at 17 and 2 h prior to analysis of GFP expression.Data are represented as the mean ± SEM of 11 mice per group.*p < 0.05, **p < 0.01, ***p < 0.001 unpaired two-sided t test (F, J), Two-way ANOVA (G) or logrank (H) statistical test.Source data are provided as a Source Data file.
https://doi.org/10.1038/s41467-023-42734-0Nature Communications | (2023) 14:6990 using the A 2A R eGFP reporter mice.The reasons for these differences are not clear but may be related to differences in kinetics in the two studies, the use of anti-PD-L1 as opposed to anti-PD-1 or a true difference between mRNA and protein.Nevertheless, we did observe an increase in A 2A R expression within CD8 + T cells isolated from the draining lymph nodes of mice undergoing immune checkpoint blockade, highlighting the potential for A 2A R blockade to enhance T cell priming in tertiary lymphoid structures beyond the tumor microenvironment in the context of immunotherapy.Moreover, it is likely that both immune checkpoint blockade and chemotherapy result in increased levels of extracellular adenosine due to the destruction of tumor cells, which in itself may make targeting the adenosine pathway therapeutically more relevant in this context.Interestingly, in the context of immune checkpoint blockade we observed that therapeutic efficacy of anti-PD-L1 was associated with the emergence of a PD-1 + A 2A R -subset that may represent a biomarker for response.
Assessment of the transcriptional profile of A 2A R + and A 2A R − CD8 + T cells revealed that A 2A R + cells elicited a reduced expression of a defined set of STAT5 target genes 40 .This is consistent with previous observations we have made with CD8 + T cells treated in vitro with the adenosine mimetic NECA 16 .This may be linked to the observation that IL-2 and IL-7, cytokines that signal through STAT5, can overcome adenosine-mediated suppression 41,55 .Interestingly, we observed that PD-L1 blockade resulted in increased expression of STAT5 target genes within both A 2A R + and A 2A R − subsets, suggesting that this may help CD8 + T cells to overcome adenosine-mediated suppression through upregulation of this pathway.
Lastly, to utilize this model to identify rational combination therapies we interrogated the impact of various cytokines on A 2A R expression as there is remarkably little known about the control of A 2A R expression in this regard.Our analysis identified the IL1 family of cytokines, and particularly IL-18, as a potent inducer of A 2A R expression on CD8 + T cells, NK cells, γδT cells and NKT cells.Given that IL-18 is known to induce anti-tumor immunity, but based upon these data may also concomitantly increase A 2A R expression, we hypothesized that suppression mediated by A 2A R signaling may limit the overall therapeutic effect of IL-18.We therefore evaluated the potential of combining IL-18 therapy with A 2A R blockade through engineering tumor cells to secrete IL-18.These experiments confirmed that IL-18 was a strong inducer of A 2A R expression in vivo, and anti-tumor effects were even greater in A 2A R knockout mice, highlighting the potential for combination therapy targeting the A 2A R in conjunction with IL-18 therapy.There has been significant work in developing a form of IL-18 suitable for clinical application, including the development of a form of IL-18 that does not bind to IL-18DR and therefore elicits greater antitumor potency than wild-type IL-18 42 .It would therefore be of significant interest to interrogate a possible combination strategy utilizing these such reagents and A 2A R antagonists that are currently under clinical evaluation.

Methods
This research and all study protocols have been approved and comply with the Peter MacCallum Animal Experimental Ethics Committee (AEEC) ethical regulations regarding the use of animals.Studies utilizing human PBMCs from healthy donors was approved by the Peter MacCallum Cancer Centre Human Research Ethics committee.Informed consent was obtained from the Australian Red Cross.

Animal models and tumor models
A 2A R eGFP reporter mice (Tg(Adora2a-EGFP)EP141Gsat) were obtained from The Jackson Laboratory repository and backcrossed onto a C57BL/6 background.The transgene for these mice is the cDNA of EGFP, followed by a polyadenylation signal, inserted into the mouse genomic bacterial artificial chromosome (BAC) RP24-238K3.This insertion is at the start codon of the Adora2a gene so that expression of the reporter mRNA is controlled by the regulatory sequences of Adora2a.C57BL/6 wildtype mice were purchased from Australian Bioresources (Moss Vale, New South Wales) or bred in house at the Peter MacCallum Cancer Centre.C57BL/6 A 2A R eGFP reporter mice or A 2A R -/-mice were bred in house at the Peter MacCallum Cancer Centre.Mice used in experiments were between 6 to 16 weeks of age and experiments were approved by the Animal Experimentation Ethics Committee #E672.AT3 tumor cells were obtained from Dr. Trina Stewart and engineered to express chicken ovalbumin as described previously 8 .MC38 tumor cells were obtained from Dr. Nicole Haynes.All studies using the breast cancer cell lines AT3ova or E0771 were performed in female mice.Experiments with MC38 tumors were performed in male mice unless specified in the source data file.AT3-Her2, E0771-Her2 and MC38-Her2 were generated as previously described 16 .Tumor lines were verified to be Mycoplasma negative by PCR analysis and were actively passaged for less than 6 months.Tumor cells were grown in DMEM supplemented with 10% FCS, Glutamax, and penicillin/ streptomycin.For in vivo experiments, the indicated number of cells were resuspended in PBS and injected subcutaneously (100 μL).Tumors were measured using callipers and calculated as mm2 (width*length).Tumor size did not exceed the maximal size approved by the AEEC ethics committee of 150 mm 2 .To engineer AT3ova and MC38 cells to express IL-18, IL-18 cDNA (ATGAACTTTGGCCGACTTC ACTGTACAACCGCAGTAATACGGAATATAAATGACCAAGTTCTCTTCG TTGACAAAAGACAGCCTGTGTTCGAGGATATGACTGATATTGATCAAA GTGCCAGTGAACCCCAGACCAGACTGATAATATACATGTACAAAGAC AGTGAAGTAAGAGGACTGGCTGTGACCCTCTCTGTGAAGGATAGTAA AATGTCTACCCTCTCCTGTAAGAACAAGATCATTTCCTTTGAGGAAAT GGATCCACCTGAAAATATTGATGATATACAAAGTGATCTCATATTCTT TCAGAAACGTGTTCCAGGACACAACAAGATGGAGTTTGAATCTTCAC TGTATGAAGGACACTTTCTTGCTTGCCAAAAGGAAGATGATGCTTTCA AACTCATTCTGAAAAAAAAGGATGAAAATGGGGATAAATCTGTAATGT TCACTCTCACTAACTTACATCAAAGTTAG) was cloned into the MSCV Cherry vector and retrovirus generated from HEK293gp cells prior to transduction of target cells.mCherry + cells were FACS sorted prior to functional experiments.

Fig. 2 |
Fig.2| A 2A R expression is elevated on tumor-infiltrating T cells and cDC2s.A WT, A 2A R -/-or A 2A R eGFP reporter mice were inoculated sub-cutaneously with 5 × 10 5 AT-3 ova tumors.A. Tumor size was measured over time (mm 2 ).Data represent the mean ± SEM of 6 (WT) or 7 (A 2A R -/-) mice per group.B-L Immune cells from draining lymph nodes and tumors were analyzed by flow cytometry at day 21 post-tumor inoculation unless otherwise stated.B Representative flow cytometry plots from concatenated samples.C, E Percentage of indicated cell types expressing GFP (A 2A R) within DLNs (C) or tumors (E).D, F Proportion of all GFP + cells that are of indicated lineage within DLNs (D) or tumors (F).C, D Data represent the mean ± SEM of 10 mice per group from 2 pooled experiments.E, F Data represent the mean ± SEM of 28 mice per group from 5 pooled experiments.G Relative expression of GFP (A 2A R) for indicated cell types in the DLN and tumor.MFI was normalized such that expression in the DLN was equal to 1. Data represent the mean ± SEM of 10 mice per group from 2 pooled experiments.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, unpaired two-sided t test.H MFI of GFP (A 2A R) expression by each cell type isolated from AT3ova tumors.Data represented as mean ± SEM of 5 mice per group.****p < 0.0001, statistics determined by one-way ANOVA I Expression of GFP (A 2A R) in indicated cell types isolated from tumors at indicated timepoints.Numbers represent the GFP MFI.Data represent concatenated samples from n = 6 mice.J Expression of GFP (A 2A R) in indicated myeloid populations.ΔGFP expression calculated as GFP MFI minus the background fluorescence for each cell type as observed in WT mice.Data represented as the mean ± SEM from 24 mice per group, pooled from 4 independent experiments.****p < 0.0001, statistics determined by one way ANOVA.K Expression of MHCII and GFP on cDC2 cells (Gated as CD45 + TCRβ -NK1.1 -Ly6C -CD64 -CD11c + MHCII + F4/ 80 low CD103 -XCR1 -).L Expression of MHCII, CD80, and CD86 on total cDC2 (Red), GFP -cDC2 (Blue) and GFP + cDC2 (Green).K, L. Data concatenated from 6 individual mice.Source data are provided as a Source Data file.

Fig. 3 |
Fig. 3 | A 2A R expression is elevated on tumor antigen-specific CD8 + T cells within draining lymph nodes but not tumors.A-G A 2A R eGFP reporter mice were injected subcutaneously with 5 × 10 5 AT-3 ova tumors.Draining lymph nodes (DLN; A) and tumor infiltrating lymphocytes (B-G) were analyzed by flow cytometry at day 21 post-tumor inoculation.Flow Cytometry plots represent data from concatenated samples in one representative experiment.A Expression of GFP (A 2A R) in CD8 + T cells that are CD44 + Tetramer + or Tetramer -.Data represent the mean ± SEM of 22 individual mice pooled from 4 experiments.B, C Expression of GFP (A 2A R) on indicated subsets of tumor-infiltrating CD8 + T cells.Data represent the mean ± SEM of 17 mice pooled from 3 individual experiments (B) or from 28 mice per group pooled from 5 individual experiments (C).D Mice were treated at days 14, 16, 18, and 20 post-tumor inoculation with 25 µg FTY720.Data represent the mean ± SEM of 5 (Control) or 6 (FTY720) mice per group.E, F Coexpression of A 2A R, PD-1 and CD39 on CD8 + T cells (E) and CD4 + T cells (F).Data represent the mean ± SEM of 28 (E) or 27 (F) mice per group pooled from 5 independent experiments.G Expression of CD25 on indicated CD4 + subsets.Data are concatenated from 6 individual samples.Where normalized MFI is shown, data were normalized relative to the average MFI of total CD8 + T cells in each experiment.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.Statistics determined by one-way ANOVA (B, C) or unpaired two-sided t test (A, D, E, F).Source data are provided as a Source Data file.

Fig. 5 |
Fig. 5 | A 2A R positive CD8 + T cells exhibit decreased expression of STAT5 target genes prior to treatment with anti-PD-L1.A 2A R eGFP reporter mice were injected with 5 × 10 5 AT-3 ova tumors subcutaneously and were indicated treated at days 14 and 18 post-tumor inoculation with anti-PD-L1.At day 21 post tumor inoculation CD8 + tumor-infiltrating lymphocytes were analyzed by 3'RNA-sequencing.Biological samples were pooled from n = 3 mice per replicate.Data represented as mean ± SD.A Principle component analysis based on the top 100 most variable genes.B Normalized CPMs for Adora2a expression from CD8 + GFP − (n = 3) and CD8 + GFP + (n = 2) cells.C-E Gene set enrichment analysis or heatmap of gene expression for indicated pathways comparing CD8 + A 2A R + and CD8 + A 2A R -cells.Raw P values are shown.Source data are provided as a Source Data file.
/doi.org/10.1038/s41467-023-42734-0Nature Communications | (2023) 14:6990 transgenic model to investigate the pattern of expression of A 2A R in the context of anti-tumor immunity and following therapeutic intervention with either immunotherapy, chemotherapy, or cytokines.