Antimetastatic Effects of Blocking PD-1 and the Adenosine A2A Receptor

Tumor-induced immunosuppression is a major hurdle to the efficacy of current cancer therapies. Perhaps because of their remarkable clinical efficacy against a broader range of cancers, recent successes with immune checkpoint blockade inhibitors such as anti-CTLA-4 (ipilumumab; refs. 1 and 2) and anti-PD-1/PDL1 (nivolumab, MK-3475/MPDL3280A, MDX-1105; refs. 3–5) are revolutionizing cancer treatment. Immune checkpoints refer to a plethora of inhibitory pathways hardwired into the immune system that are crucial for maintaining self-tolerance and modulating the duration and amplitude of physiologic immune responses in peripheral tissues to minimize collateral damage. Ipilumumab and nivolumab generated very effective responses in advanced melanoma when used in combination (6), but the former triggered a significant number of immune-related adverse events, even when used as single therapy (7), which may limit its potentially broader use in combinations. Because of the number of immunosuppressive mechanisms utilized by tumors, it is becoming increasingly apparent that targeting multiple immunosuppressive pathways has the potential to enhance therapeutic efficacy without an increase in adverse events associated with excessive inflammation/autoimmunity (8).

One promising molecule that is a new target in preclinical studies is immunosuppressive adenosine. This metabolite is produced by the CD73 ectoenzyme expressed on host suppressor cells and tumor cells (9). Inhibition of CD73 using a monoclonal antibody (mAb) has been shown to reduce tumor growth and metastasis by enhancing antitumor immunity (10). Both host and tumor-expressed CD73 are important suppressive mechanisms, because knockdown or overexpression of CD73 on tumor cells can modulate tumor growth and metastasis (9–11). In addition, CD73^−/− mice are protected from transplanted and spontaneous tumors (10, 12, 13). In humans, high CD73 expression in triple negative breast cancer has been shown to be a negative prognostic marker and correlates with a high risk of metastasis (14, 15). The protumor effects of CD73 are believed to be largely because of adenosine-mediated immunosuppression. Adenosine binds to four known receptors A₁, A_2A, A_2B, and A₃, with the activation of A_2A and A_2B receptors known to suppress the effector functions of many immune cells (10). In the microenvironment of the tumor, both A_2A and A_2B receptor activation has been demonstrated to suppress antitumor immunity (11, 16, 17) and increase the spread of CD73⁺ tumors (9). In addition, either A_2A or A_2B blockade with small molecule antagonists can reduce tumor metastasis. Because A_2A receptor expression is increased in lymphocytes following activation (18), we hypothesized therapies that liberate lymphocyte effector responses, such as anti-CTLA-4, anti-PD-1, and anti-Tim3 (19–21), may also increase the effects of A_2A-mediated immunosuppression. In this study, we investigated whether dual immune checkpoint blockade and A_2AR inhibitor could increase the magnitude of immune responses to metastasis.

C57BL/6 and BALB/c wild-type (WT) mice were purchased from the Walter and Eliza Hall Institute for Medical Research or ARC Animal Resource Centre. C57BL/6 perforin-deficient (pfp^−/−) and C57BL/6 DEREG (FoxP3-DTR-GFP) mice were bred in-house at the QIMR Berghofer Medical Research Institute. All mice were maintained at the QIMR Berghofer Medical Research Institute and used between the ages of 6 to 14 weeks. Groups of 5 to 10 mice per experiment were used for experimental and spontaneous tumor metastasis assays. These group sizes were used to ensure adequate power to detect biologic differences. All experiments were approved by the QIMR Berghofer Medical Research Institute Animal Ethics Committee.

The C57BL/6 B16F10 (ATCC) and B16F10-CD73^hi melanomas and 4T1.2 mammary carcinomas were derived and maintained as described (9, 10). 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 in a 200 μL volume.

Purified anti-mouse PD-1 mAb (RMP1-14), anti-mouse CTLA-4 mAb (UC10-4F10), anti-mouse Tim3 (RMT3-23) and control Ig (2A3) were purchased from BioXCell (West Lebanon) and used in the schedule and dose as indicated. SCH58261 was purchased from Sigma and used at 1 mg/kg i.p. per dose. Anti-IFNγ (H22), Rabbit anti-asialoGM1 antibody (Wako Chemicals), anti-CD4 (GK1.5), and anti-CD8β (53-6.7) were resuspended in PBS and injected at the indicated time points.

For spontaneous metastasis and postsurgery survival experiments, 5 × 10⁴ 4T1.2 tumor cells were inoculated into the fourth mammary fat-pad of BALB/c mice. On day 20 to 22, mice were anesthetized, the primary tumor was surgically removed, the wound was closed with surgical clips, and treatments commenced immediately as indicated. For experimental metastasis, 1 or 2 × 10⁵ B16F10 or B16F10-CD73^hi cells were injected intravenously in 200 μL PBS and the treatment commenced immediately after (day 0) and day 3. B16F10 and B16F10-CD73^hi macrometastases on the surface of the lungs were counted using a dissecting microscope (Nikon SMZ 745T).

Eight- to 12-week-old C57BL/6 mice were injected with 2.5 × 10⁵ B16F10-CD73^hi cells intravenously into the tail vein (10 mice per group). Mice were treated with SCH58261 (1 mg/kg) or anti-PD-1 (250 μg/mouse) or the combination of both or with isotype control mAb (clone 2A3, 250 μg/mouse) on day 0 and day 3. Mice were sacrificed on day 5 and lungs were collected after perfusion with PBS and mechanically dissociated using the Medimachine (Dako) to obtain cell suspensions. Hematopoietic cells were then purified using a discontinuous Percoll gradient and stained for flow cytometry.

The following antibodies were used: CD3 PerCPefluor710 (500A2), CD8 APCefluor780 (53-6.7), Foxp3 efluor605NC (FJK-16s), NK1.1 CD106a PE (1D4B), CD45.2 PECy7 (104), CD4 Alexa Fluor 700 (RM4-5) from BD Biosciences and NK1.1 Alexa Fluor 488 (PK13) from Biolegend. Stained cells were acquired and analyzed by multicolor flow cytometry using an LSR Fortessa (BD Biosciences).

Total RNA was isolated from cancer cell lines and flow cytometry-sorted splenic lymphocytes by using RNAzol (Sigma-Aldrich) as per the manufacturer's instructions. For cDNA synthesis, 500 ng of total RNA was reverse transcribed in 20 μL reaction containing 500 μmol/L dNTP mix, 500 nmol/L Oligo-dT, 200U of Tetro Reverse Transcriptase and 20 U of Ribosafe RNase Inhibitor (Bioline) at 42°C for 50 min. The reactions were heated at 85°C for 5 minutes, cDNA product was diluted to 10 ng/μL and 2 μL of cDNA was subjected to 10 μL real-time PCR reaction using SensiFAST SYBR Lo-ROX Kit (Bioline) and primers (Sigma-Aldrich) on Applied Biosystems ViiA 7 Real-Time PCR system. Ct values for A_2A receptor were compared with the housekeeping gene RPL32 using the following mouse primers: L32 forward-TTCCTGGTCCACAATGTCAAG, L32 Reverse- TGTGAGCGATCTCAGCAC, A_2A receptor Forward- CAGAGTTCCATCTTCAGCCTC. A_2A receptor Reverse- CACCCAGCAAATCGCAATG.

Statistical differences were analyzed by Mann–Whitney test with P < 0.05 considered significant. A log-rank test was performed to assess the statistical significance of differences between survival curves. Prism (Graph pad Software) was used for graphs and statistical analysis.

We have recently demonstrated the therapeutic activity of the A_2ARi, SCH58261, in protecting mice from B16F10-CD73^hi experimental lung metastasis (9). Here we have confirmed that activity and examined A_2ARi in the context of checkpoint blockade by additionally treating mice with anti-CTLA-4, anti-PD-1, or anti-Tim3 (Fig. 1A). Early treatment of B16F10-CD73^hi lung metastases by anti-CTLA-4, anti-PD-1, or anti-Tim3 mAb alone was relatively ineffective, compared with A_2ARi treatment alone over the same period. However, each of 3 checkpoint blockade mAbs, further enhanced the activity of A_2ARi when given coincidently in combination (Fig. 1A). This combination and effect of A_2ARi alone was specific for CD73^hi expressing B16F10, because neither A_2ARi alone, nor in combination, was effective in suppressing lung metastases of the parental B16F10 cells (Fig. 1B). This suggests that combination therapy involving A_2ARi may be successfully used for patients with high tumor cell expression of CD73, such as those with poor prognosis triple negative breast cancer (15). Given the promise of anti-PD-1 mAbs in the clinic, we decided to pursue further general utility and mechanism of this combination. Initially, we confirmed the effect of the combination against B16F10-CD73^hi melanoma metastases by varying the schedule of A_2ARi and checkpoint blockade (Fig. 1C). Coincident treatment of A_2ARi and anti-PD-1 mAb on days 0 and 3 after tumor inoculation was superior to other schedules, although A_2ARi given twice on day 0 and anti-PD-1 mAb twice on day 3; or A_2ARi given on day 0 and 1 and anti-PD-1 given on days 2 and 3 were also effective. Late treatment with A_2ARi seemed to be less effective (Fig. 1C) and delayed combination therapy was not significant when treatment commenced at day 2 onwards post tumor inoculation (Fig. 1D). These data indicated the combinatorial promise of checkpoint blockade and A_2ARi and the potential use of tumor CD73 expression as a predictive marker of response to A_2ARi or A_2ARi and checkpoint blockade.

The 4T1.2 mammary carcinoma expresses high levels of CD73 constitutively (10) and may be used as a model of primary tumor resection and metastasis to test therapies in a clinically relevant setting. We have previously shown that the A_2ARi, SCH58261, can inhibit spontaneous lung metastases when given early to mice (from day 3) with primary mammary tumor intact (9). When mice were administered A_2ARi four times every three days from the day of surgery, a very significant prolongation in lifespan was achieved (Fig. 2A, P <0.01). By contrast, mice that were additionally depleted of NK cells or T cells (either CD8⁺ alone or CD4⁺ and CD8⁺) did not enjoy that survival benefit, indicating the nonredundant function of these lymphocyte subsets in the mechanism of antimetastatic activity of A_2ARi (Fig. 2A). We next assessed the combinatorial effect of A_2ARi and anti-PD-1 mAb in the same experimental model and found that anti-PD-1 and A_2ARi were superior to either monotherapy alone (Fig. 2B). Thus, in both experimental and spontaneous metastases models, the combination of anti-PD-1 mAb and A_2ARi was more effective than monotherapy and significantly reduced disease burden.

We next assessed the cell-mediated immune effector function responsible for the effectiveness of the A_2ARi and anti-PD-1 combination in both the experimental and spontaneous metastases tumor models. In the B16F10-CD73^hi lung metastases model, the therapeutic effect of combined anti-PD-1 mAb and A_2ARi was more dependent on NK cells, than CD8⁺ T cells, and CD4⁺ T cells did not seem to have a critical additional role (Fig. 3A). In the surgical resection and spontaneous metastasis of 4T1.2 tumor model, both CD8⁺ T cells and NK cells were key in the optimal antitumor activity generated by the combination anti-PD-1 mAb and A_2ARi (Fig. 3B). Thus, in both experimental and spontaneous tumor metastasis models, the combination of A_2AR and PD-1 blockade was able to engage CD8⁺ T and NK-cell effector function. This effector function was displayed by the critical need for host perforin and IFNγ pathways in the optimal antimetastatic activity of the anti-PD-1 mAb and A_2ARi combination (Fig. 4A).

To our knowledge, there is no functional antibody available to detect mouse adenosine receptors. To assess the target of A_2ARi, SCH58261, we analyzed the mRNA expression of A_2AR on B16F10-GFP and B16F10-CD73^hi mouse melanoma and 4T1.2 mammary carcinoma cell lines. The A_2AR expression on tumor cells was at minimum 1,000-fold lower than in lymphocytes and it was independent of the tumor cell expression of CD73 (Fig. 4B). Interestingly, the expression of A_2AR on splenic NK cells was 5-fold higher than other lymphocytes (Fig. 4B), suggesting that lymphocytes (CD4⁺ T cells, CD8⁺ T cells, and regulatory T cells) and in particular NK cells are the possible target of A_2ARi in the tumor microenvironment. Our data are consistent with a previous finding where SCH58261 and NECA antagonized one another when added to NK cell:B16F10 cocultures, NECA significantly suppressing NK-cell–mediated killing activity in vitro (9). In addition, we also found significantly increased NK-cell proportions in the lungs of mice with 5-day B16F10-CD73^hi metastases and treated with monotherapy or combination therapy compared with control treated mice (Fig. 4C). NK-cell numbers were enhanced in the combination group (Fig. 4D), whereas the numbers of other immune cell populations were not significantly different between the treatment groups (data not shown).

We conclude that treatment of mice with a combination of A_2ARi and anti-PD-1 mAb substantially reduces experimental and spontaneous metastases and increases the survival of mice, compared with either monotherapy. The antimetastatic activity of the A_2ARi and combination immunotherapy is dependent upon CD73 expression on tumor cells whereas the mechanism of the combination is largely dependent on NK cells and IFNγ, and to a lesser extent on CD8⁺ T cells and perforin-dependent effector functions. This effector mechanism is possibly also a combination of the reported perforin-dependent activity of A_2ARi (9) and IFNγ activity of anti-PD-1 (22). Notably, adenosine or CD73 blockade has not yet been attempted in patients with cancer, despite the safety profile of A_2ARi in human neurologic disease. The great promise of anti-PD-1 in the treatment of several human malignancies, including melanoma, non–small cell lung cancer and renal cancer, is being realized, and new combinations with anti-PD-1 are being explored and having further impact (6). Treatment of metastatic disease should be a priority and these results strongly advocate a combination of A_2ARi with anti-PD-1 in the treatment of metastatic disease.

No potential conflicts of interest were disclosed.

Conception and design: D. Mittal, J. Stagg, M.J. Smyth

Development of methodology: D. Mittal, B. Allard, M.J. Smyth

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Young, K. Stannard, M.W.L. Teng, B. Allard, M.J. Smyth

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Young, B. Allard, J. Stagg, M.J. Smyth

Writing, review, and/or revision of the manuscript: D. Mittal, A. Young, M.W.L. Teng, J. Stagg, M.J. Smyth

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

Study supervision: M.W.L. Teng, J. Stagg, M.J. Smyth

The authors thank the assistance of K. Elder and L. Town for mouse genotyping and the QIMR Berghofer Medical Research Institute animal facility technicians. The authors also thank M. Hughes and K. Powell for mouse care and maintenance.

M.J. Smyth and K. Stannard are supported by a National Health and Medical Research Council of Australia (NHMRC) Australia Fellowship (628623) and Program Grant (1013667). D. Mittal is supported by a Susan G. Komen Breast Cancer Foundation Program Grant. A. Young is supported by a Cancer Council Queensland Ph.D. fellowship. M.W.L. Teng and M. Yong are supported by a NH&MRC Career Development Award (1025552) and Project Grant (1021139). J. Stagg is supported by the Canadian Institutes of Health Research (CIHR) and Famille Jean-Guy Sabourin Research Chair. B. Allard is supported by a Fonds de Recherche du Québec - Santé (FRQS) postdoctoral fellowship. This research was supported by a grant from Susan G. Komen for the Cure (#IIR12221504).

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.