Activation of Th1 and Tc1 cell adenosine A_2A receptors directly inhibits IL-2 secretion in vitro and IL-2-driven expansion in vivo

Adenosine, which is released from metabolically active cells and generated extracellularly by degradation of released adenosine triphosphate (ATP), is a potent biologic mediator that modulates numerous cell functions. Adenosine protects cells and tissues during inflammation and ischemia^1,2  and mediates its effects via 4 receptor subtypes: the A₁, A_2A, A_2B, and A₃ adenosine receptors.²  G protein-coupled adenosine receptors are emerging as potential therapeutic targets, particularly for cardiac ischemia and neurodegenerative diseases.^1-3  Importantly, recent work indicates that regulation of inflammation through A_2A receptor signaling cannot be compensated by other pathways. Inflammatory cell damage causes adenosine to accumulate in tissues, which subsequently initiates a negative feedback signal in monocytes, neutrophils, and lymphocytes via the A_2A receptor.³  Therefore, adenosine is believed to “put the brake on inflammation.”⁴ 

Immunosuppression through adenosine A_2A receptor activation of T lymphocytes has been previously documented.^5,6  However, these studies were performed with naive T cells under initial stimulation conditions; therefore, little is known regarding A_2A receptor activation of polarized effector T-cell populations. This question is of significant interest, as tissue damage and subsequent adenosine accumulation likely occurs relatively late in the inflammatory process, after T-cell maturation and differentiation. Therefore, it is possible that mature T cells may be a more relevant target of A_2A stimulation than naive T cells.

Furthermore, depending on the cytokine environment during initial antigen encounter, T cells can differentiate either into type 1 T cells that secrete primarily interleukin 2 (IL-2) and interferon γ (IFN-γ) or type 2 T cells that secrete primarily IL-4, IL-5, and IL-10. This type 1/type 2 cytokine polarization exists for both CD4⁺ T cells (Th1/Th2 subsets) and for CD8⁺ T cells (Tc1/Tc2 subsets). Importantly, Th1/Th2 and Tc1/Tc2 states of T-cell differentiation appear to influence disease pathogenesis, especially inflammatory processes.⁷  Because Tc1 and Tc2 subsets of CD8⁺ T cells each can mediate potent target cell lysis and secrete cytokines that exert differential effects on inflammation, we hypothesized that adenosine receptors, and in particular, those of the A_2A subtype, may operate differentially in these immune effector subsets.

Female, 6- to 8-week-old (C57Bl/6 x Balb/c)F₁ (CB6F₁, H-2K^b/d), C57Bl/6 (H-2K^b), and congenic C57Bl/6 Ly5.1 mice were obtained from the Frederick Cancer Research Facility (FCRF, Frederick, MD) and maintained in a specific pathogen-free facility. All experiments were performed according to an animal protocol approved by the National Cancer Institute (NCI). Transgenic mice [B6;SJL-Tg(TcrAND)53Hed/J] expressing the T-cell receptor (TCR) specific for pigeon cytochrome c (PCC) in the context of major histocompatibility complex (MHC) class II I-E^k were obtained from Jackson Labs (Bar Harbor, ME). For adoptive transfer experiments involving transgenic T cells, hosts were obtained from Jackson Labs that were either capable of presenting PCC antigen (B10.BR; I-E^k) or not capable of presenting PCC (C57BL/10J; I-E^b).

Complete media (CM) consisted of RPMI 1640 (Life Technologies, Rockville, MD) supplemented with penicillin, streptomycin, nonessential amino acids, 2-mercaptoethanol (Life Technologies), and 10% fetal calf serum (Gemini, Woodland, CA). ACK lysing buffer (Quality Biological [QBI], Gaithersburg, MD) and Hanks balanced salt solution (HBSS; Mediatech Cellgro, Herndon, VA) were used. Monoclonal anti-CD3 (clone 145-2C11) and anti-CD28 (clone 37.51) used for generating anti-CD3/anti-CD28-coated beads (3/28 beads) were purchased from BD PharMingen (San Diego, CA). Magnetic tosyl-activated M450 Dynabeads were purchased from Dynal Biotech (Lake Success, NY). The method for conjugating antibody to beads is detailed elsewhere.⁸  CPA (N⁶-cyclopentyl-adenosine), NECA (5′-N-ethyl-carboxamidoadenosine), CGS21680 (CGS), Cl-IB-MECA (2-chloro-N⁶-(3-iodobenzyl)adenosine-5′-N-methyluron-amide; Cl-IB), and dimethyle sulfoxide (DMSO) were purchased from Sigma (St Louis, MO), ZM241385 from Tocris (Ellisville, MO), and SCH58261 was provided by Dr E. Ongini (Schering Plough, Milan, Italy). For T-cell depletion, anti-CD4 (clone GK1.5) and anti-CD8 (clone 2.43) were obtained from NCI Biological Resource Branch (NCI-BRB, Frederick, MD). Goat anti-rat Ig (GAR) and goat anti-mouse Ig (GAM) coated bioparticles were from Polysciences (Warrington, PA).

Cytokines and antibodies for in vitro culture were recombinant (r) human IL-2 and anti-IL-4 clone 11B11 (NCI-BRB); human rIL-7, murine rIL-12, and murine rIL-4 (Peprotech, Rocky Hill, NJ); and 20% N-acetylcysteine solution (NAC; Bedford Laboratories, Bedford, OH). For CTL assays, ⁵¹Cr was purchased from Amersham (Piscataway, NJ), the anti-CD3 (clone 145-2C11) from PharMingen; EGTA (ethylene glycol-bis(beta-aminoethyl ether)-N,N,N′,N′-tetraacetic acid), PMA (phorbol 12-myristate 13-acetate), and calcium ionophore (CI) were from Sigma; magnesium chloride was from QBI; HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid) was from Mediatech; and sodium dodecyl sulfate (SDS) was from Oncor (Gaithersburg, MD). For cyclic AMP (cAMP) accumulation assay, [³H] cAMP (40 Ci[2.22TBq]/mmol) was from Amersham Pharmacia Biotech (Buckinghamshire, United Kingdom); rolipram, hydrochloric acid (HCl), K₂HPO₄, EDTA (ethylenediaminetetraacetic acid), and protein kinase A (PKA) were from Sigma. For flow cytometry, rat IgG2a (phycoerythrin [PE] and fluorescein isothiocyanate [FITC]), anti-CD4 (clone RM4-5), anti-CD8 (clone 53-6.7), and anti-CD45.1 (clone A20) were all from BD PharMingen. For flow cytometric identification of PCC-specific T cells, anti-Vα11 FITC (clone RR8-1; PharMingen) and anti-Vβ3 (clone KJ25; PharMingen) were used. Bovine serum albumin (BSA) was from ICN Biochemicals (Aurora, OH); sodium azide and propidium iodide (PI) were from Sigma.

Splenic single-cell suspensions were obtained from CB6F₁ mice, red blood cells were lysed with ACK, and B cells were removed with GAM beads. T-cell-enriched cells (0.3 × 10⁶ cells/mL) were stimulated with 3/28 beads (bead-to-cell ratio of 3:1) at 37°C in a 5% humidified CO₂ chamber. The Tc1 culture condition contained human rIL-2 (20 IU/mL), human rIL-7 (20 ng/mL), murine rIL-12 (10 ng/mL), anti-IL-4 (10 μg/mL), and 3.3 mM NAC. The T2 culture condition consisted of human rIL-2 (1000 IU/mL), human rIL-7 (20 ng/mL), murine rIL-4 (1000 U/mL), and 3.3 mM NAC. Cells were maintained at 0.2 × 10⁶ cells/mL to 0.5 × 10⁶ cells/mL by daily addition of T1- or T2-specific media. However, IL-12 was added only at day 0 of Tc1 culture. In some experiments, the culture was enriched for CD8⁺ T cells on day 5 by negative selection using anti-CD4 and GAR beads. CD4 cell contamination of negatively selected CD8 cells was less than 1%.

On day 5 of culture, after T1 or T2 culture with or without CD8⁺ enrichment, T cells were washed, and 1 × 10⁶ cells were restimulated with 3/28 beads (bead-to-cell ratio of 3:1) in 2 mL fresh CM in the presence of either vehicle (DMSO) or A_2A agonist CGS for 24 hours. Supernatants were collected and cytokine content was measured by 2-site enzyme-linked immunosorbent assay (ELISA). TNF-α kit (tumor necrosis factor alpha) was from CytImmune (College Park, MD), IL-2 reagents were from R&D Systems (Minneapolis, MN), and IL-4, IL-5, IL-10, and IFN-γ reagents were from Biosource (Camarillo, CA).

Cytokine secretion analysis by capture flow cytometry was performed using Miltenyi reagents and protocol (Miltenyi Biotec, Bergisch Gladbach, Germany). Generated Th1/Tc1 cells were restimulated with anti-CD3, anti-CD28-coated beads with CGS or DMSO vehicle for 4 hours. T cells were harvested and coated with anti-IFN-γ or anti-IL-2 antibody. Coated cells were incubated (37°C, 45 minutes) with anti-IFN-γ detection or anti-IL-2 detection antibody. Cells were stained with directly conjugated anti-CD4 or anti-CD8 and evaluated by 3-color flow cytometry. Percentage of cells secreting cytokine and mean fluorescence intensity (MFI) of cytokine secretion were calculated.

T1 cell cultures were generated from congenic Ly5.1 mice. On day 6, Ly5.1⁺ Th1/Tc1 cells (10 × 10⁶) were transplanted by intravenous injection into irradiated syngeneic (C57Bl/6) or semiallogeneic (CB6F₁) recipients (¹³⁷Cs radiation source, 100 cGy/min; Gamma Cell 40; Atomic Energy of Canada, Ottawa, ON, Canada). The irradiation dose of 1050 cGy was split into 2 fractions separated by 5 hours (550 cGy/500 cGy). After T-cell transplantation, recipients were injected daily (intraperitoneally) with DMSO vehicle, CGS (2 mg/kg per day), or CGS plus IL-2 (10 000 IU/day). After 4 days (experiment no. 1) or 5 days (experiment no. 2), spleens were removed, spleen cell number was determined, splenic single cells were stained for CD4, CD8, and the congenic marker CD45.1, and fluorescence activated cell sorting (FACS) was performed. Absolute number of CD45.1⁺CD4⁺ Th1 and CD45.1⁺CD8⁺ Tc1 cells per spleen was then calculated. Mean values from n = 5 mice were determined and P values were obtained using t test; values less than .05 were considered statistically significant.

Th1 cell cultures were generated from transgenic mice expressing CD4⁺ T cells with TCR specific for PCC. On day 6, transgenic Th1 cells (5 × 10⁶) were transplanted by intravenous injection into irradiated hosts that were either capable or not capable of presenting PCC to transgenic Th1 cells (B10.BR or C57/BL10 hosts, respectively). Irradiation dose of 1050 cGy was split into 2 fractions separated by 5 hours (550 cGy/500 cGy). At the time of adoptive T-cell transfer, PCC (Sigma; 50 μg/recipient) was mixed with complete Freund adjuvant (CFA; Gibco BRL, Gaithersburg, MD; 1:1 mix of PCC to CFA) and administered by intraperitoneal injection. Experimental cohorts were injected daily (intraperitoneally) with DMSO vehicle, CGS (2 mg/kg per day), CGS plus IL-2 (50 000 IU twice per day), or IL-2 alone (50 000 IU twice per day). On day 8 after adoptive T-cell transfer, spleens were removed, spleen cell number was determined, splenic single cells were stained for CD4, CD8, and the markers of the transgenic T cells (Vα11 and Vβ3), and FACS was performed. The absolute number of transgenic T cells per spleen was then calculated. Mean values from n = 7 to 8 mice per cohort were determined and P values were obtained using nonparametric Wilcoxin rank sum analysis; values less than .05 were considered statistically significant.

The cAMP assay was performed as described previously.^9,10  On day 5 of culture, polarized Tc1 and Tc2 cells were washed, incubated with phospho-diesterase inhibitor rolipram (10 μM) for 30 minutes, and restimulated with 3/28 beads for 3 hours at 37°C on a rotator with or without the A_2A agonist CGS. Cells were pelleted and lysed by addition of 200 μL ice-cold 0.1 M HCl. The cell lysate was resuspended and stored at -20°C. For determination of cyclic AMP production, protein kinase A (PKA) was incubated with [³H] cAMP (2 nM) in K₂HPO₄/EDTA buffer (K₂HPO₄, 150 mM; EDTA, 10 mM), 20 μL of the cell lysate, and 30 μL 0.1 M HCl or 50 μL of cyclic AMP solution (0-16 pmol/200 μL for standard curve). Bound radioactivity was separated by rapid filtration through Whatman GF/C filters, washed once with cold buffer, and measured by liquid scintillation counter.

Cytolytic capacity of Tc1 and Tc2 cells was assessed on day 6 of culture. For measurement of total cytolytic function, a modified redirected CTL assay was performed.¹¹  The murine Fc receptor expressing cell line P815 (H-2K^d; provided by Ron Gress, NIH) was labeled with ⁵¹Cr Tc1 and Tc2 populations were incubated with CGS for 2 hours and loaded with anti-CD3 (clone 2C11), which binds to CD3 on T cells and to Fc receptor on P815 cells. P815 target cells were incubated with Tc1 and Tc2 effector cells for 4 hours in CM containing either CGS or DMSO.

To measure nongranule exocytosis-based killing, ⁵¹Cr release assays were performed in calcium-neutralized conditions by addition of EGTA (5 mM), magnesium chloride (0.5 mM), and HEPES (25 mM) to CM.^12,13  Effector Tc1 and Tc2 populations were incubated with CGS for 2 hours; PMA (5 ng/mL) and CI (375 ng/mL) were added for an additional 2 hours to induce FasL expression. Fas-transfected L1210 cells (L1210-fas; H-2K^d), or wild-type L1210 cells were used as targets in 8-hour CTL assays with or without CGS (lines provided by Pierre Henkart, NIH).

Supernatants were harvested, transferred to LumaPlates (Packard, Downers Grove, IL) and read using a Microplate Scintillation and Luminescence Counter (Packard). Minimal lysis was determined by incubation of each target in CM, whereas maximum lysis was determined by target incubation in SDS 20%. Percent specific target lysis was calculated as % lysis = (cpm sample - cpm minimum lysis)/(cpm maximum lysis - cpm minimum lysis) × 100.

For surface staining of T1 and T2 cells, 1 × 10⁶ cells were washed in HBSS containing 1% BSA and 0.5% sodium azide, incubated with relevant antibodies or isotype controls, and washed. Two-color flow cytometry was performed on a FACSCalibur (Becton Dickinson Immunocytometry Systems, Mountain View, CA) using CELLQuest software (Becton Dickinson). Five thousand to 10 000 events were acquired per sample; dead cells were gated out by PI.

By day 4 of culture, T cells grown under type 1 polarizing conditions primarily produced IFN-γ and IL-2 upon 3/28 restimulation, whereas T cells grown under type 2 conditions produced preferentially IL-4 and IL-10 (Figure 1). After in vitro culture, CD8⁺ T cells predominated (65%-80%), whereas CD4⁺ T cells represented the remaining 20% to 35% of cells. Cytokine secretion was polarized in both mixed CD4/CD8 cells and purified CD8⁺ T cells. Expanded T cells had an effector phenotype by flow cytometry (reduced CD62L and increased CD25 and CD44; data not shown).

To evaluate the effects of adenosine activation on Th1/Tc1 and Th2/Tc2 subsets, cytokine production was measured in the presence of subtype-selective adenosine receptor agonists² : CPA was used for A₁ receptor activation, CGS for A_2A activation, NECA for A_2A and A_2B activation, and Cl-IB-MECA for A₃ activation. Adenosine receptor agonist signaling in Th1/Tc1 cells markedly inhibited IL-2 secretion and occurred only through A₂ receptor agonists (Figure 2, CGS and NECA conditions). Since NECA binds to A_2A and A_2B receptors, but CGS21680 binds to A_2A receptors exclusively,¹⁴  adenosine direct regulation of T-cell cytokine secretion likely occurs primarily via the A_2A receptor. In the face of this potent inhibition of IL-2 secretion, A₂ receptor agonists did not significantly reduce Th1/Tc1 cell IFN-γ secretion. Adenosine signaling in Th2/Tc2 cells moderately inhibited IL-10 secretion through A₂ receptor activation, but did not reduce Th2/Tc2 cell secretion of the hallmark type 2 cytokine, IL-4.¹⁵ 

Since the currently available adenosine receptor agonists are not absolutely selective for a given subtype, potency of the putative selective agonists were measured to derive a dose-response relationship. At 10^-7 M, both CGS (A_2A agonist) and NECA (A_2A and A_2B agonist) inhibited Th1/Tc1 cell secretion of IL-2; in marked contrast, the A₃ agonist Cl-IB-MECA and the A₁ receptor agonist CPA did not inhibit IL-2 secretion at this concentration (Figure 3A). The potencies for CPA, CGS21680, NECA, and Cl-IB-MECA were 550 nM, 11 nM, 11 nM, and 825 nM, respectively; these values are consistent with each reagent's affinity for the A_2A receptor.²  These results confirm that the A_2A receptor, but not A₁, A_2B, and A₃ receptors, mediates the observed cytokine inhibition. To further confirm specificity of adenosine A_2A receptors for inhibition of Th1/Tc1 IL-2 secretion, additional conditions evaluated CGS in combination with specific A_2A receptor antagonists ZM241 385 and SCH58 261 (Figure 3B). Addition of these antagonists fully abrogated CGS-mediated IL-2 inhibition, thereby further confirming the role of the A_2A receptor in this immune modulation.

We next evaluated the role of adenosine A_2A signaling in purified and cytokine polarized CD8⁺ Tc1 and Tc2 cells. Stimulation of G-protein-coupled A_2A receptors results in accumulation of intracellular cAMP,⁵  which has been associated with inhibition of T-cell effector function. To determine the relative capacity of Tc1 and Tc2 subsets to signal through the adenosine A_2A receptor, Tc1 and Tc2 cell cAMP levels were measured after CD3, CD28 restimulation with or without CGS. Costimulation in the presence of CGS increased intracellular cAMP levels similarly in Tc1 and Tc2 cells, thereby indicating that each subset expressed functional adenosine A_2A receptors (Figure 4).

Using purified and cytokine-polarized effector Tc1 and Tc2 cells, we evaluated the effect of CGS-mediated adenosine A_2A receptor agonism on effector CD8 cell cytokine secretion. Similar to results obtained with mixed Th1/Tc1 cells, A_2A receptor activation of purified Tc1 cells induced a dose-dependent reduction in IL-2 secretion (maximum reduction of ∼ 70% at 10^-7 to 10^-6 M) without reduction in IFN-γ secretion (Figure 5A). The estimated effective concentration (EC₅₀) of CGS for inhibition of Tc1 cell IL-2 secretion was 10 nM to 20 nM, which is similar to that previously described for modulation of rat brain A_2A receptors.¹⁶  With respect to the purified Tc2 effector CD8 subset, results were again similar to that obtained with the mixed Th2/Tc2 population: CGS-mediated A_2A receptor agonism modestly reduced IL-10 secretion, with nominal effect on IL-4 or IL-5 secretion.

In vitro polarization of murine Tc2 cells is typically incomplete, with low-level IL-2 secretion and moderate IFN-γ secretion often detected. It was therefore of interest to determine whether adenosine A_2A receptor activation reduced such type I cytokine secretion emanating from the Tc2 subset. Indeed, similar to the biology demonstrated in Tc1 cells, CGS-mediated adenosine A_2A activation abrogated Tc2 cell secretion of IL-2, and had a nominal effect on Tc2 cell IFN-γ secretion (Figure 5B).

Given the anti-inflammatory role of adenosine, it was of interest to evaluate the direct effect of adenosine signaling on secretion of TNF-α, which is a potent proinflammatory cytokine produced by both Tc1 and Tc2 subsets. Both subsets secreted TNF-α in the absence of adenosine signaling, with Tc1 cells secreting increased levels relative to Tc2 cells. However, similar to the case for IL-2 secretion, adenosine signaling reduced TNF-α secretion by over 90% in both Tc1 and Tc2 subsets (Figure 5C). In sum, these data indicate that IL-2 and TNF-α, and to a lesser extent, IL-10, are preferentially inhibited by adenosine A_2A receptor signaling in polarized effector Tc1 and Tc2 cells.

These results indicate that signaling of T cells through adenosine A_2A receptors differentially inhibits cytokine production; most strikingly, significant IL-2 inhibition occurred concomitant with preserved IFN-γ secretion. This immune modulation may occur either through deletion of cells that secrete high levels of IL-2, or through preferential IL-2 inhibition and preserved IFN-γ secretion at the single cell level. To characterize this biology, we evaluated IL-2 and IFN-γ secretion with or without adenosine A_2A receptor agonism by cytokine capture flow cytometry (Figure 6). In the presence of CGS, MFI of IL-2 secretion in T cells that were IL-2⁺IFN-γ^- or IL-2⁺IFN-γ⁺ was reduced from 3313 to 1372 and from 3226 to 1392, respectively (Table 1). It is important to note that the magnitude of this reduction in IL-2 secretion mirrors that observed with CGS inhibition of IL-2 secretion by ELISA. Concomitantly, and similar to results using ELISA, no reduction in the MFI value for IFN-γ secretion in CGS-treated T cells was observed. Furthermore, CGS did not appear to reduce the frequency of cells capable of secreting IL-2. These results indicate that CGS inhibits IL-2 secretion while preserving IFN-γ secretion at the single-cell level, and does not mediate its effect through deletion of IL-2-secreting cell populations.

Given that A_2A receptor activation directly inhibited polarized effector CD8 cell cytokine secretion, we next evaluated whether adenosine agonism might regulate cytolytic function. Costimulated Tc1 and Tc2 subsets each possessed cytolytic activity (Figure 7). Similar to prior results,¹⁷  costimulated Tc2 cells were more potent than Tc1 cells with respect to exocytosis-based killing, whereas costimulated Tc1 cells were modestly enriched for fas-based killing. Strikingly, adenosine A_2A receptor activation at doses of CGS (10^-8 and 10^-6 M) that potently regulated cytokine secretion did not modulate Tc1 or Tc2 cytolytic function in either exocytosis- or fas-based killing assays.

Given that adenosine agonism significantly reduced IL-2 secretion in Th1/Tc1 effectors in vitro, we hypothesized that activation of adenosine A_2A receptors in vivo through CGS administration might limit the expansion of Th1/Tc1 cells after allogeneic transplantation. Indeed, CGS administration in vivo significantly reduced the absolute number Th1 and Tc1 cells observed after allogeneic transplantation (Figure 8A; Th1 cells reduced from 3.8 ± 0.4 × 10⁶ to 2.0 ± 0.3 × 10⁶ cells/spleen, P = .009; Tc1 cells reduced from 3.2 ± 0.2 × 10⁶ to 2.1 ± 0.3 × 10⁶ cells/spleen, P = .012). In contrast, CGS administration did not significantly reduce Th1 or Tc1 cell number after syngeneic transplantation (Th1 cells, control versus CGS values, 0.4 ± 0.1 × 10⁶ vs 0.5 ± 0.1 × 10⁶ cells/spleen, P = NS; Tc1 cells, control versus CGS values, 3.0 ± 0.6 × 10⁶ vs 2.6 ± 0.3 × 10⁶ cells/spleen, P = NS).

This result indicated that adenosine agonism inhibited antigen-driven but not homeostatic expansion of effector Th1 and Tc1 cells in vivo. To evaluate whether inhibition of the IL-2 pathway contributed to the observed adenosine-mediated reduction in Th1 and Tc1 cell number, a second allogeneic transplantation experiment that evaluated the role of concomitant CGS and exogenous IL-2 was performed (Figure 8B). In this experiment, CGS21680 again significantly reduced the absolute number of Th1 and Tc1 cells observed after allogeneic transplantation. Strikingly, adenosine-mediated reduction in Th1 and Tc1 cell number was completely abrogated by posttransplantation IL-2 therapy.

Our finding that CGS therapy inhibited T cells in vivo after allogeneic but not syngeneic transplantation is consistent with the hypothesis that adenosine A_2A receptor agonism may inhibit T cells undergoing TCR interaction with antigen while sparing T cells not responding to antigen. To further test this hypothesis, we performed adoptive transfer experiments using antigen-specific CD4⁺ T cells obtained from TCR transgenic mice that respond to pigeon cytochrome c (PCC) antigen in the context of MHC I-E^k. Such PCC-specific T cells were expanded and polarized in vitro toward a Th1 phenotype, and administered to hosts capable of presenting PCC antigen (B10.BR hosts; I-E^k) or not capable of PCC antigen presentation (C57/BL10 hosts; I-E^b).

In the setting of transgenic T-cell transfer to hosts capable of PCC antigen presentation, CGS significantly reduced the absolute number of transgenic T cells in vivo (Figure 9A, right panel). In marked contrast, in the setting of transgenic T-cell transfer to hosts not capable of PCC antigen presentation, CGS therapy did not reduce the absolute number of transgenic T cells in vivo (Figure 9A, left panel). These findings indicate that adenosine A_2A receptor agonism through CGS in vivo therapy primarily inhibits T cells as they undergo TCR/MHC-antigen interaction, with sparing of T cells not undergoing TCR activation. Furthermore, we observed that CGS-mediated inhibition of antigen-specific Th1 cells was abrogated by in vivo IL-2 therapy (Figure 9B). In sum, in vivo results from the allogeneic transplantation model and the antigen-specific T-cell transfer model indicate that adenosine A_2A receptor agonism preferentially inhibits antigen-activated T cells by a mechanism that can be abrogated by IL-2 therapy.

Activation of lymphocyte adenosine A_2A receptors by extracellular adenosine released during hypoxia or cell damage results in an anti-inflammatory effect.¹⁸  Effector CD8⁺ T cells contribute to inflammation through cytokine secretion and cytolytic function.¹⁹  In this study, we determined that adenosine modulated both Tc1 and Tc2 effector subsets via A_2A receptor interaction, and found that this regulation involved primarily type 1 cytokine inhibition, with nominal inhibition of cytolytic function.

This is the first study to evaluate direct effects of adenosine signaling on mature CD8⁺ T cells of type 1 and type 2 phenotype. Our studies demonstrate that Tc1 and Tc2 effectors are susceptible to adenosine modulation through adenosine A_2A receptors, but not A₁, A_2B, or A₃ receptors. Our results are consistent with data demonstrating A_2A receptors as the major pathway mediating immunoregulatory effects of extracellular adenosine.⁶  These prior results were obtained in models where naive T cells were challenged with or without adenosine. As such, our results extend this understanding, as the adenosine A_2A receptor appears to be essential for regulation of both afferent T-cell responses and effector Tc1 and Tc2 responses.

Adenosine-mediated inhibition of Tc1 and Tc2 cytokine secretion involved primarily the inflammatory cytokines IL-2 and TNF-α. Our finding that adenosine signaling inhibited IL-2 production in highly expressing Tc1 cells and nominally expressing Tc2 cells is consistent with the concept that the A_2A receptor potently regulates inflammation. That is, by restricting production of a major T-cell growth factor, IL-2, the A_2A receptor likely prevents further autocrine expansion of activated T cells. We further found that IL-2 inhibition through adenosine A_2A receptor activation occurred at the single cell level, and not through deletion of IL-2-secreting T cells. Adenosine signaling also greatly reduced TNF-α secretion in both highly expressing Tc1 and nominally expressing Tc2 cells. Given the long recognized role of TNF-α as a direct inflammatory mediator,²⁰  our results indicate that adenosine may regulate Tc1- and Tc2-mediated inflammation both via limitation of clonal expansion (IL-2 inhibition) and effector molecule production (TNF-α inhibition). Adenosine-mediated inhibition of IL-2 and TNF-α was common to both Tc1 and Tc2 subsets, suggesting that A_2A receptor signaling regulates specific cytokine pathways rather than specific cellular subtypes. Finally, because our experiments used an APC-free method of T-cell costimulation, we conclude that adenosine inhibition of T-cell cytokine secretion can occur through direct T-cell effects independent of adenosine modulation of APC populations.

Surprisingly, we did not observe significant inhibition of IFN-γ production in either Tc1 or Tc2 cells. Because A_2A receptor-deficient mice have elevated serum IFN-γ levels during inflammation, we initially hypothesized that IFN-γ would be a major target for adenosine regulation of Tc1 and Tc2 cell function. Our results suggest that net reduction in IFN-γ by adenosine in vivo may result from limitation of Tc1 and Tc2 cell clonal expansion or adenosine regulation of monocyte IFN-γ production. Given the somewhat paradoxical role of IFN-γ as both immune activator and immune inhibitor,^21,22  it is interesting to speculate that preservation of IFN-γ production in adenosine-exposed Tc1 and Tc2 cells may actually contribute to an anti-inflammatory effect. In contrast, TNF-α, which is strongly inhibited by adenosine in Tc1 and Tc2 cells, is an inflammatory mediator with no known immunosuppressive function.

Given the preferential contribution of type 1 cytokines to inflammatory processes, our observation that type 2 cytokines were minimally influenced by adenosine signaling was anticipated. In light of the capacity of type 2 cytokines²³  and type 2 cells^24,25  to counteract inflammation mediated by type 1 processes, our results suggest that adenosine may exert anti-inflammatory effects both by reducing proinflammatory cytokines and maintaining type 2 cytokines. Our data regarding the direct modest inhibitory effect of adenosine on T-cell IL-10 secretion, combined with a previous study that found adenosine A_2A receptor stimulation augmented IL-10 production in vivo,²⁶  suggest that adenosine induction of IL-10 in vivo may emanate from a non-T-cell, APC source. Although inflammation is primarily a type 1-driven process, some inflammatory diseases are driven by type 2 immunity.^27,28  It is possible that, in such cases of type 2-associated inflammation, adenosine A_2A agonism may actually exacerbate disease pathogenesis.

In marked contrast to adenosine A_2A receptor inhibition of Tc1 and Tc2 cell cytokine secretion, we did not observe down-regulation of cytolytic function. This result was somewhat unexpected, as previous investigators concluded that A_2A signaling inhibits CTL lytic capacity.²⁹  This discrepancy may relate to the prior study's usage of relatively nonselective adenosine analogues or usage of a high concentration of the selective agonist CGS, as effective CTL blockade was observed only at the 50 μM concentration; it is unlikely that our result was due to insufficient exposure to the A_2A agonist, as we preincubated CTL for 2 hours prior to chromium release assay, which is an interval longer than the normal agonist A_2A receptor equilibrium time. It should also be noted that prior studies have found that several direct or indirect adenylate cyclase activators, including forskolin and PGE2, can inhibit T-cell cytotoxicity.³⁰  Apparently, the modulatory effect mediated by selective A_2A receptor activation as demonstrated in this study is distinct from that mediated by cAMP itself or by additional receptor activation. Finally, it should be noted that a prior study found that adenosine can act through A₃ receptors to inhibit CTL adhesion to target cells³¹ ; in our studies, we did not evaluate the role of selective A₃ receptor activation on CTL function. Therefore, although adenosine or other modulators of cAMP can inhibit CTL function, we conclude that at a given concentration of agonist, adenosine A_2A receptor signaling of Tc1 and Tc2 cells regulates primarily cytokine secretion, with both perforin- and fas-based killing cytolytic pathways being relatively resistant to adenosine A_2A receptor activation.

In this study, we also found that adenosine A_2A receptor activation reduced Th1 and Tc1 cell numbers in vivo after allogeneic but not syngeneic transplantation. Furthermore, in a separate model involving antigen-specific Th1 cells, we found that adenosine A_2A receptor activation reduced Th1 cell number under host conditions permitting antigen presentation, but not under host conditions lacking antigen presentation. These findings suggest that adenosine modulation of effector T cells through A_2A receptors may selectively reduce antigen-driven T-cell effects while sparing homeostatic T-cell biology. Use of adenosine A_2A receptor reagents to selectively modulate antigen-driven responses may be advantageous relative to established immune suppression approaches. Our novel finding that in vivo T-cell inhibition through adenosine A_2A receptor activation was fully abrogated by IL-2 therapy may be of particular significance. This result predicts that adenosine receptor modulation might be used to treat IL-2-driven diseases involving primarily Th1 and Tc1 cellular pathways, such as graft-versus-host disease (GVHD).^17,32  Conversely, it is also interesting to speculate that the mechanism accounting for antitumor effects after IL-2 administration³³  may operate in part through abrogation of adenosine-mediated T-cell inhibition.

In summary, adenosine signaling of effector Th1/Tc1 and Th2/Tc2 cells can occur directly through the G_s-protein-coupled A_2A type adenosine receptor. Adenosine regulation of Th1/Tc1 and Th2/Tc2 cells is manifested primarily by inhibition of IL-2 and TNF-α production. These effects are relatively specific, as other effector pathways such as cytolysis, IFN-γ secretion, or type 2 cytokine secretion are generally resistant to adenosine A_2A receptor modulation. Modulation of adenosine A_2A receptor biology therefore appears to represent a rational approach to the regulation of effector T-cell-mediated inflammatory disorders, in particular those mediated by Th1 and Tc1 cells.

We would like to thank Dr David Segal for advice and help with setting up the redirected killing assay.