Abstract
Adenosine is elaborated in injured tissues where it suppresses inflammatory responses of essentially all immune cells, including production of proinflammatory cytokines such as tumor necrosis factor-α (TNF-α). Most of the anti-inflammatory actions of adenosine have been attributed to signaling through the A2A adenosine receptor (A2AAR). Previously, however, it has been shown that the A3AR agonist N6-(3-iodobenzyl)adenosine-5′-N-methylcarboxamide (IB-MECA) potently inhibited TNF-α release from macrophages obtained from A2AAR “knockout” (A2AKO) mice, suggesting that the A3AR may also regulate cytokine expression. Here, we confirmed that the A2AAR is the primary AR subtype that suppresses TNF-α release from thioglycollate-elicited mouse peritoneal macrophages induced by both Toll-like receptor-dependent (TLR) and TLR-independent stimuli, but we determined that the A2BAR rather than the A3AR mediates the non-A2AAR actions of adenosine since 1) the ability of IB-MECA to inhibit TNF-α release was not altered in macrophages isolated from A3KO mice, and 2) the A2BAR antagonist 1,3-dipropyl-8-[4-[((4-cyanophenyl)carbamoylmethyl)oxy]phenyl]xanthine (MRS 1754) blocked the ability of the nonselective AR agonist adenosine-5′-N-ethylcarboxamide (NECA) to inhibit TNF-α release from macrophages isolated from A2AKO mice. Although A2BARs seem capable of inhibiting TNF-α release, the A2AAR plays a dominant suppressive role since MRS 1754 did not block the ability of NECA to inhibit TNF-α release from macrophages isolated from wild-type (WT) mice. Furthermore, the potency and efficacy of adenosine to inhibit TNF-α release from WT macrophages were not influenced by blocking A2BARs with MRS 1754. The data indicate that adenosine suppresses TNF-α release from macrophages primarily via A2AARs, although the A2BAR seems to play an underlying inhibitory role that may contribute to the anti-inflammatory actions of adenosine under select circumstances.
Proinflammatory cytokines, including tumor necrosis-α (TNF-α), interleukin (IL)-1, and IL-6, are generated in tissues infected by microbial pathogens as well as in tissues subjected to generalized trauma such as ischemia/reperfusion injury. Generation of inflammatory mediators serves a necessary function to facilitate wound healing, in part by recruiting various immune cell populations. Nevertheless, excessive or chronic inflammation can also lead to additional tissue injury and is the target for therapy in a variety of diseases (Rankin, 2004). In the context of microbial invasion and sepsis, proinflammatory cytokine expression is induced by engagement of Toll-like receptors (TLR) by bacterial components such as lipopolysaccharide (LPS) (Beutler et al., 2003). During generalized tissue injury, cytokine expression is induced by lipid mediators (e.g., leukotrienes, platelet-activating factor), oxygen-derived free radicals, complement components, and possibly by activation of TLR or N-formyl-methionyl-leucyl-phenylalanine receptors activated by intracellular debris released from necrotic cells (Frangogiannis et al., 2002; Le et al., 2002; Beutler et al., 2003).
Adenosine is a purine nucleoside generated by metabolically stressed or inflamed tissues that is recognized as an endogenous anti-inflammatory agent because of its potent suppressive action on virtually all cells of the immune system (Hasko and Cronstein, 2004). Of the four adenosine receptor (AR) subtypes (A1, A2A, A2B, and A3), the Gs protein-coupled A2AAR is most widely recognized to attenuate inflammation via a cAMP-mediated pathway (Sitkovsky, 2003; Hasko and Cronstein, 2004). A critical role has been established for the A2AAR in inhibiting neutrophil superoxide production, degranulation, and adhesion (Hasko and Cronstein, 2004); T-cell expansion and differentiation (Huang et al., 1997); platelet aggregation (Paul et al., 1990); and endothelial cell adhesion molecule expression (Bouma et al., 1996). The use of A2AAR-deficient mice (A2AKO) has confirmed the importance of the A2AAR in limiting proinflammatory responses (Ohta and Sitkovsky, 2001).
Adenosine also potently inhibits the expression of proinflammatory cytokines, including TNF-α, from several different cell types in response to TLR activation (Hasko and Cronstein, 2004). Previous work has shown that the A2AAR is the primary receptor subtype responsible for this effect acting by a variety of mechanisms such as inhibition of nuclear factor-κB (NF-κB) (Bshesh et al., 2002; Majumdar and Aggarwal, 2003; Lukashev et al., 2004; Sands et al., 2004) and inhibition of p38-induced promotion of RNA stability (Fotheringham et al., 2004). However, it has been suggested that the A3AR is also capable of suppressing proinflammatory cytokine expression (Hasko et al., 2000; Hasko and Cronstein, 2004). Furthermore, several studies have shown that A3AR agonists are beneficial using in vivo models of inflammation (Hasko et al., 1998; Mabley et al., 2003; Baharav et al., 2005). Our laboratory has shown that A3AR agonists are effective at protecting against myocardial ischemia/reperfusion injury when administered at the time of reperfusion (Auchampach et al., 2003), a therapeutic window associated with a powerful inflammatory reaction involving cytokine production initiated by free radical generation, complement activation, calcium overload, and tissue necrosis (Frangogiannis et al., 2002).
The goal of this investigation was to further characterize the AR subtypes involved in regulating proinflammatory cytokine expression, focusing on the potential involvement of the A3AR. Specifically, we sought to determine the AR subtypes involved in inhibiting TNF-α expression from mouse peritoneal macrophages in response to TLR agonists as well as non-TLR stimuli generated in the context of ischemia/reperfusion injury. Our experimental approach involved the use of thioglycollate-elicited peritoneal macrophages obtained from A2A and A3KO mice and newly developed antagonists for rodent ARs. Our results indicate that the A3AR plays no role in regulating TNF-α expression. Rather, we demonstrate that the A2AAR is the predominant receptor subtype responsible for inhibiting TNF-α production and that the A2BAR plays an underlying inhibitory role. Previous work implicating the A3AR is explained by the use of high concentrations of A3AR agonists capable of activating A2A and A2BARs.
Materials and Methods
Materials. Cell culture reagents, TRIzol reagent, pcDNA3.1, Geneticin (G418), Lipofectamine, and ThermoScript RT-PCR kits were purchased from Invitrogen (Carlsbad, CA). TNF-α enzyme-linked immunosorbent assay kits were purchased from BD Biosciences (San Jose, CA). SYBR Green Supermix and Bradford reagent were purchased from Bio-Rad (Hercules, CA). ZM241385 was from Tocris Cookson Inc. (Ellisville, MO), adenosine deaminase was from Roche Applied Science (Indianapolis, IN), BG 9928 was a gift from Dr. Barry Ticho (Biogen Idec., Cambridge, MA), and all remaining drugs and reagents were purchased from Sigma-Aldrich (St. Louis, MO). N6-(4-Amino-3-[125I]iodobenzyl)adenosine-5′-N-methylcarboxamide ([125I]I-AB-MECA) and 125I-ZM241385 were synthesized and purified by high-performance liquid chromatography, as described previously (Olah et al., 1994; Palmer et al., 1995; Auchampach et al., 1997). [3H]MRS 1754 was custom-synthesized according to the procedure of Ji et al. (2001).
Mice. C57BL/6 wild-type (WT) mice were purchased from Taconic Farms (Germantown, NY). Homozygous A3KO mice were a kind gift from Dr. Marlene Jacobson (Merck Research Labs, West Point, PA), and A2AKO mice were created by Dr. Jiang-Fan Chen (Boston University, Boston, MA). All of the A2A and A3KO mice used in these studies were back-crossed to the C57BL/6 genetic background.
Isolation of Mouse Peritoneal Macrophages. Mice were injected intraperitoneally with 2 ml of 2% thioglycollate. After 4 days, peritoneal cells collected by lavage were seeded onto 24-well plates in RPMI 1640 medium with 10% calf serum and gentamicin (50 μg/ml) for 4 h to allow the macrophages to adhere to the plates. Nonadherent cells were subsequently removed by washing with RPMI 1640 medium, and the adherent macrophages were refed with RPMI 1640 medium with 10% calf serum and gentamicin. Macrophages were used for experiments immediately following isolation.
Treatment with Adenosine Agonists and Antagonists. Macrophages were pretreated with AR agonists for 30 min at 37°C, followed by stimulation with various activating agents including LPS, zymosan-activated serum (ZAS), A23187 (calcimycin), or phorbol 12-myristate 13-acetate (PMA) at the concentrations indicated. ZAS was prepared by incubating mouse serum with zymosan A (15 mg/ml) followed by removal of zymosan by centrifugation. Antagonists were given 30 min before treatment with the AR agonists. After 24 h, culture media were collected and assayed for TNF-α by enzyme-linked immunosorbent assay. Subsequently, the cells were lysed with 0.4 N NaOH and assayed for total protein by the Bradford assay (Bradford, 1976). TNF-α released was expressed as picograms per milligram of protein, or as a percentage of maximal TNF-α released from vehicle-treated cells.
Quantitative Real-Time RT-PCR. Total RNA was isolated from macrophages using TRIzol reagent. Subsequently, 1 μg of total macrophage RNA was reverse transcribed using a mixture of random and poly-T primers according to the manufacturer's protocol (Invitrogen). Primers were designed for the mouse A1 (FWD, 5′-TGGCTCTGCTTGCTATTG-3′; REV, 5′-GGCTATCCAGGCTTGTTC-3′), A2A (FWD-5′ TCAGCCTCCGCCTCAATG-3′; REV, 5′-CCTTCCTGGTGCTCCTGG-3′), A2B (FWD, 5′-TTGGCATTGGATTGACTC-3′; REV, 5′-TATGAGCAGTGGAGGAAG-3′), and A3AR (FWD, 5′-CGACAACACCACGGAGAC-3′; REV, 5′-GCTTGACCACCCAGATGAC-3′) using Beacon Design software (Bio-Rad). PCR amplification (in SYBR Green Supermix) was performed using an iCycler iQ thermocycler (Bio-Rad) for 40 cycles of 25 s at 95°C followed by 45 s at an optimized annealing temperature for each AR. The cycle threshold, determined as the initial increase in fluorescence above background, was ascertained for each sample. Melt curves were performed upon completion of the cycles to ensure that nonspecific products were absent. For quantification of AR transcripts, a standard curve plotting cycle threshold versus copy number was constructed for each receptor subtype by analyzing 10-fold serial dilutions of plasmids containing the full-length mouse AR clones. AR transcript levels were expressed as copies/50 ng of total RNA.
Radioligand Binding Assays. Binding assays were conducted with membranes prepared from isolated macrophages. In brief, macrophages cultured on 150-mm plates were washed with phosphate-buffered saline, scraped into homogenization buffer (10 mM Na-HEPES, pH 7.4, 10 mM EDTA, and 0.1 mM benzamidine), homogenized in a glass Dounce homogenizer, and then centrifuged at 20,000g for 30 min. Cell pellets were washed in binding buffer (10 mM Na-HEPES, pH 7.4, 1 mM EDTA, and 0.1 mM benzamidine) and then resuspended in binding buffer containing 10% (w/v) sucrose. Membranes were stored in aliquots at –20°C until used for binding assays.
For radioligand binding studies, 50 μg of membrane protein was incubated in a final volume of 100 μl of binding buffer containing 5 mM MgCl2, 1 unit/ml adenosine deaminase, and either ∼0.5 nM 125I-ZM241385 to label A2AARs (Palmer et al., 1995), ∼0.5 nM [125I]I-AB-MECA to label A1 and A3ARs (Olah et al., 1994; Auchampach et al., 1997), or 10 nM [3H]MRS 1754 to label A2BARs (Ji et al., 2001). In competition experiments, inhibitors were included in the reactions at the concentrations indicated. After incubating at 21°C for 3 h, the incubations were terminated by rapid filtration over glass-fiber filters using a 48-well Brandel cell harvester. Filter discs containing trapped membranes bound with radioligand were quantified using a gamma or liquid scintillation counter. Nonspecific binding was determined in the presence of 1 μM ZM241385, 1 μM[125I]I-AB-MECA, or 100 μM adenosine-5′-N-ethylcarboxamide (NECA), respectively.
Competition radioligand binding assays were also conducted with membranes prepared from HEK 293 cells expressing recombinant mouse ARs to determine the affinity of the antagonists used in the study for mouse ARs. The full-length cDNA sequences of the four mouse ARs were obtained by RT-PCR using total RNA isolated from mouse brain tissue. The cDNA clones were verified by sequencing and then subcloned into the mammalian expression vector pcDNA3.1, transfected into HEK 293 cells using Lipofectamine, and selected with 2 mg/ml G418. After antibiotic selection, the cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum with 0.6 mg/ml G418. Cell membranes were prepared and then incubated with radioligands ([125I]I-AB-MECA for A1 and A3ARs, 125I-ZM241385 for A2AARs, and [3H]MRS 1754 for A2BARs) and antagonist competitors. The radioligand binding data were analyzed as described previously (Auchampach et al., 1997).
Data Analysis. Data are reported as means ± S.E.M. Differences between groups were analyzed by one-way analysis of variance followed by post hoc analyses with unpaired Student's t test with the Bonferroni correction. A p value <0.05 was considered statistically significant.
Results
Antagonist Pharmacology of Mouse ARs. Preliminary radioligand binding studies were conducted to assess the affinity and selectivity of the A1AR antagonists 1,3-dipropyl-8-cyclopentylxanthine (CPX) and BG 9928, the A2AAR antagonist ZM241385, the A2BAR antagonist MRS 1754, and the A3AR antagonist MRS 1523 for recombinant mouse ARs expressed in HEK 293 cells. The data from these studies are presented in Table 1 and were used to choose appropriate concentrations of the chemicals in subsequent studies with macrophages.
AR Expression in Mouse Peritoneal Macrophages. Our initial goal was to determine which AR subtypes are expressed in mouse peritoneal macrophages. To begin, we quantified AR transcript levels by quantitative real-time RT-PCR. The absolute copy numbers of the AR transcripts were calculated based on standard curves generated with mouse AR cDNA clones. As illustrated in Fig. 1A, we detected mRNA expression of A2A, A2B, and A3ARs in mouse peritoneal macrophages. mRNA expression of the A2AAR was highest (5395 ± 657 copies/50 ng of RNA), followed by the A2BAR (649 ± 92) and the A3AR (455 ± 59). We did not detect expression of A1AR mRNA above background levels.
We subsequently conducted radioligand binding assays with crude membrane preparations to assess expression of ARs at the protein level in mouse peritoneal macrophages. Membranes were incubated with ∼0.5 nM A2AAR antagonist 125I-ZM241385, ∼0.5 nM A1/A3AR agonist [125I]I-AB-MECA, or 10 nM A2BAR antagonist [3H]MRS 1754. We detected specific binding of 125I-ZM241385 to membranes, defined by inclusion of 1 μM ZM241385 (Fig. 1B). Specific binding of 125I-ZM241385 was not displaced by the A1AR antagonist BG 9928 (100 nM), the A2BAR antagonist MRS 1754 (100 nM), or the A3AR antagonist MRS 1523 (5 μM), indicating that 125I-ZM241385 was specifically labeling A2AARs. Given that we included 125I-ZM241385 in our assays at a concentration equal to its Kd value for the A2AAR (Table 1), we estimated the Bmax of 125I-ZM241385 to be ∼40 fmol/mg protein.
We also detected specific binding of [125I]I-AB-MECA to membranes prepared from mouse peritoneal macrophages (Fig. 1C), defined by inclusion of 1 μM nonradiolabeled I-AB-MECA. Since [125I]I-AB-MECA binds with relatively high affinity to both A1 and A3ARs (Olah et al., 1994; Auchampach et al., 1997), it could have labeled either of these AR subtypes in our assays. However, specific binding of [125I]I-AB-MECA was displaced solely by the A3AR antagonist MRS 1523 (5 μM) and not by BG 9928 (A1AR antagonist; 100 nM) or ZM241385 (A2AAR antagonist; 100 nM), indicating that [125I]I-AB-MECA was binding to the A3AR. We estimated the Bmax of [125I]I-AB-MECA binding to mouse peritoneal macrophages to be ∼18 fmol/mg protein.
[3H]MRS 1754 has recently been characterized as a useful high-affinity antagonist radioligand for recombinant A2BARs (Ji et al., 2001). Therefore, we attempted to use [3H]MRS 1754 to detect protein expression of A2BARs in mouse peritoneal macrophages. However, we were unable to detect specific binding of [3H]MRS 1754 with macrophage membranes (data not shown) most likely because of a combination of low specific activity of [3H]MRS 1754, high nonspecific binding, and the potentially low expression of endogenous A2BARs in mouse peritoneal macrophages.
In summary, our quantitative real-time RT-PCR and radioligand binding data indicate that A2A and A3ARs are expressed in mouse peritoneal macrophages, correlating with previous work implicating these two receptor subtypes in the regulation of TNF-α release (Hasko et al., 2000; Hasko and Cronstein, 2004). Our PCR data also suggest that A2BARs are expressed in mouse macrophages.
AR Activation Inhibits TNF-α Release from Macrophages in Response to TLR and Non-TLR Stimuli. Proinflammatory cytokine production by macrophages is classically known to be stimulated by TLR agonists such as LPS (Beutler et al., 2003). In preliminary studies, we tested a panel of agents associated with ischemia/reperfusion injury for their ability to stimulate TNF-α release from peritoneal macrophages. Among those chosen were hydrogen peroxide as a source of oxygen-derived free radicals, platelet-activating factor, PMA as an activator of protein kinase C, ZAS as a source of complement factor C5a, A23187 to mimic calcium overload, and TNF-α. Of these, LPS, PMA, A23187, and ZAS induced TNF-α release from isolated peritoneal macrophages (Fig. 2). LPS at a concentration of 10 μg/ml was the most potent stimulant increasing TNF-α from a basal level of 130 ± 35 pg/mg in vehicle-treated cells to 557,936 ± 172,636 pg/mg protein (∼5000-fold increase), whereas ZAS (6 μl) was the least potent producing increases of ∼5-fold above basal levels.
We subsequently assessed the ability of AR stimulation to inhibit TNF-α release from mouse peritoneal macrophages. Isolated macrophages were preincubated with either 1 or 10 μM of the nonselective AR agonist NECA for 30 min before stimulation with LPS, PMA, A23187, or ZAS. As shown in Fig. 2, pretreatment with NECA inhibited LPS-, A23187-, and ZAS-induced TNF-α release with varying efficacies (∼95, 50, and 70%, respectively). Although there seemed to be a trend, NECA did not significantly inhibit PMA-induced TNF-α release (Fig. 2C). Thus, the results demonstrate that AR activation inhibits TNF-α release from peritoneal macrophages induced by both TLR-dependent and TLR-independent stimuli.
The A3AR Plays No Role in Regulating TNF-α Production in Mouse Peritoneal Macrophages. To determine the potential involvement of the A3AR in inhibiting TNF-α release, we next examined the effect of the AR agonist IB-MECA. IB-MECA is an N6-substitued 5′-methyluronamide derivative of adenosine developed as a potent and selective agonist for the A3AR (Gallo-Rodriguez et al., 1994). As shown in Fig. 3, treatment with IB-MECA at a concentration of 1 μM potently inhibited LPS-stimulated TNF-α release (64 ± 7%). However, when the cells were pretreated with the A3AR antagonist MRS 1523 (10 μM) or the A1AR antagonist CPX (30 nM), the inhibitory action of IB-MECA on TNF-α release was not affected, whereas it was successfully blocked by ZM241385 at concentrations (30 and 100 nM) capable of blocking A2A and A2BARs (Fig. 3). These results suggest that IB-MECA inhibited TNF-α release via the A2A and/or A2BAR rather than the A3AR.
To conclusively exclude a role of the A3AR in regulating TNF-α release, we compared concentration-response curves generated with IB-MECA using macrophages isolated from either WT or A3KO mice. For purpose of comparison, concentration-response curves were also conducted with the A1AR agonist 2-chloro-N6-cyclopentyladenosine (CCPA) and the A2AAR agonist CGS 21680. As revealed in Fig. 4, the concentration-response curves generated with IB-MECA and the other subtype-selective AR agonists were similar when macrophages from WT and A3KO mice were compared. EC50 values calculated for each AR agonist are presented in Table 2. The potency order of the agonists to inhibit TNF-α release from both WT and A3KO macrophages was CGS 21680 > IB-MECA > CCPA, which is indicative of an effect mediated by the A2AAR.
A2BARs Mediate Inhibition of TNF-α Release in Macrophages from A2AKO Mice. The data suggest that the A2AAR is the primary AR subtype that inhibits TNF-α release from mouse peritoneal macrophages. However, to examine the potential contribution of additional AR subtypes, we evaluated the effect of the nonselective AR agonist NECA on TNF-α released by macrophages isolated from A2AKO mice. As presented in Fig. 5, NECA continued to suppress TNF-α released in response to LPS, ZAS, and A23187 in macrophages isolated from A2AKO mice. These results illustrating an A2AAR-independent suppression of TNF-α release from mouse peritoneal macrophages concur with those reported earlier by Hasko et al. (2000). Since the A1AR does not seem to be expressed in mouse peritoneal macrophages and since our data do not support the involvement of the A3AR, we predicted that the A2BAR might be suppressing TNF-α production in macrophages from A2AKO mice. To address this theory, we examined the effect of blocking A2BARs with MRS 1754. As shown in Fig. 6, MRS 1754 (but not the A3AR antagonist MRS 1523) completely blocked the inhibitory effect of NECA on LPS-, ZAS-, and A23187-induced TNF-α release from A2AKO macrophages.
The A2BAR Does Not Influence Suppression of LPS-Induced TNF-α Release from WT Macrophages. We next questioned the relative contribution of the A2BAR in suppressing TNF-α release when all ARs are present at physiological densities, i.e., in macrophages isolated from WT mice. To address this issue, two experiments were performed. First, we examined whether blockade of the A2BAR with MRS 1754 (300 nM) reduces the ability of the nonselective AR agonist NECA to inhibit LPS-induced TNF-α release. In the second experiment, we examined whether the concentration-response relationship with adenosine, the endogenous AR ligand, is shifted in the presence of MRS 1754 (100 nM). As shown in Fig. 7A, the ability of NECA to inhibit LPS-induced TNF-α production was not antagonized by MRS 1754. MRS 1754 also failed to appreciably shift the concentration-response curve generated with adenosine to inhibit LPS-induced TNF-α release in macrophages isolated from WT mice (Fig. 7B). EC50 values in the absence and presence of MRS 1754 were 20.6 ± 5.6 and 48.8 ± 13.7 μM (p > 0.05), respectively.
Discussion
The results of the present study indicate that the A2AAR is the predominant AR subtype that suppresses TNF-α production from murine peritoneal macrophages in response to TLR-dependent and TLR-independent stimuli and that the A2BAR plays an underlying inhibitory role. Although we detected the expression of the A3AR at both the mRNA and protein level in mouse macrophages, our experiments using AR gene KO mice and specific AR antagonists argue against a role for the A3AR in regulating TNF-α expression.
An earlier study by Hasko et al. (2000) suggested that the A3AR may regulate proinflammatory cytokine expression. These investigators demonstrated that adenosine continued to suppress LPS-induced TNF-α and IL-12 production from murine peritoneal macrophages isolated from A2AKO mice.
Since it was observed that the A3AR agonist IB-MECA (but not the A2AAR agonist CGS 21680) was also capable of suppressing cytokine expression from A2AKO macrophages, it was cautiously postulated that the A3AR may be responsible for the A2AAR-independent actions of adenosine (Hasko et al., 2000; Hasko and Cronstein, 2004). In the present investigation, however, we have determined that IB-MECA inhibits TNF-α release from macrophages devoid of A2AARs via the A2BAR rather than the A3AR. This conclusion is based on the following pieces of evidence. 1) The A2BAR antagonist MRS 1754 blocked the ability of the nonselective AR agonist NECA to inhibit TNF-α production from A2AKO macrophages, whereas the A3AR antagonist MRS 1523 was ineffective, and 2) the potency and efficacy of three different AR agonists to inhibit LPS-induced TNF-α release were essentially identical in macrophages isolated from WT and A3KO mice. The low potency of IB-MECA to inhibit TNF-α production (EC50 = 467 ± 13 nM) is also evidence to suggest that it is not acting via the A3AR in our studies. Notably, Sajjadi et al. (1996) reported several years ago that activation of the A3AR inhibited LPS-induced TNF-α production from PMA-differentiated U937 human monocytic cells based on the agonist potency order of IB-MECA = N6-(3-iodo-4-aminobenzyladenosine > CGS 21680 and on the inability of CPX or the A2AAR antagonist 3,7-dimethyl-1-propargylxanthine to block the inhibitory effect of N6-(3-iodo-4-aminobenzyladenosine on LPS-induced TNF-α production. In retrospect, however, it seems likely that the inhibitory AR involved was also the A2BAR, since all of the agonists used in the study displayed very low potency (EC50 > 1 μM; Sajjadi et al., 1996).
We only observed the A2BAR to functionally inhibit LPS-induced TNF-α production in studies with macrophages isolated from A2AKO mice but not from WT mice. Specifically, we found that blockade of A2BARs with MRS 1754 did not reduce the ability of NECA or adenosine to inhibit LPS-induced TNF-α release from WT macrophages (Fig. 7). These data suggest that the A2BAR plays relatively minor role in regulating cytokine production in macrophages that express the A2AAR at normal levels, likely because abundantly expressed A2AARs mask the inhibitory actions of the A2BAR. However, it is important to consider that our studies were performed with macrophages that had been stimulated previously in the isolation process using thioglycollate. Since it has been shown that expression of the A2AAR, and to a lesser extent the A2BAR, is induced in response to inflammatory stimuli (Murphree et al., 2005), it is possible that the A2BAR plays a more important role in the initial stages of inflammation before the induction of the A2AAR. Future studies with unstimulated macrophages are necessary to test this theory and to exclude the possibility that genetic deletion of the A2AAR produces adaptive changes that may increase the influence of A2BAR signaling in macrophages.
It is interesting that the A2BAR has generally been considered to be a proinflammatory receptor. Activation of the A2BAR has been shown to stimulate IL-8 production by the human mast cell line HMC-1 (Feoktistov and Biaggioni, 1995) and to stimulate IL-6 and monocyte chemotactic protein-1 production from human primary bronchial smooth muscle cells (Zhong et al., 2004). In the gut, the A2BAR has been shown to be expressed at high levels in epithelial cells, which promotes chloride/water secretion as well as production of IL-6 in response to adenosine released from infiltrating inflammatory cells (Strohmeier et al., 1995). Finally, we have previously shown that the A2BAR mediates degranulation of dog BR mastocytoma cells (Auchampach et al., 1997). Based on these observations, it has been theorized that specific antagonists of the A2BAR may be effective anti-inflammatory agents. It has also been speculated that the mechanism of action of theophylline and enprofylline for the treatment of asthma may be, in part, because of its relatively high potency as an A2BAR antagonist (Feoktistov and Biaggioni, 1995; Auchampach et al., 1997). Nevertheless, the results of the present study suggest that the A2BAR is also capable of suppressing TNF-α expression in macrophages, supporting the alternative view that selective agonists of the A2BAR may be useful anti-inflammatory agents in certain disease states.
We did not address the specific mechanism by which A2AR signaling suppressed TNF-α release. However, A2AR activation is likely to interfere with central components controlling cytokine expression since we observed that AR stimulation inhibited TNF-α release induced by a variety of stimuli, including LPS, A23187, and ZAS. A2A and A2BARs are both Gs protein-coupled receptors that increase intracellular cAMP levels upon activation. Previous studies have suggested that, in some cell types (endothelial cells, C6 glioma cells, myeloid cells, and lymphoid cells), the A2AAR/cAMP pathway inhibits NF-κB activation, an important transcription factor that drives the expression of many inflammatory genes, including TNF-α (Bshesh et al., 2002; Majumdar and Aggarwal, 2003; Lukashev et al., 2004; Sands et al., 2004). In other cell types, it has been suggested that A2AAR signaling interferes with stimulus-induced p38 kinase activation, which along with NF-κB promotes transcription of proinflammatory genes (including TNF-α) and promotes mRNA stability by interfering with destablizing AU-rich elements in the 3′ untranslated region (Fotheringham et al., 2004). It seems that the signaling pathways by which ARs inhibit cytokine expression vary depending on the specific cell type and the proinflammatory stimulus. Although we observed that AR stimulation suppressed TNF-α release in response to LPS, A23187, and ZAS, it was ineffective when PMA was used as the stimulus. PMA may increase TNF-α release by post-translational mechanisms involving activation of tumor necrosis factor converting enzyme that cleaves soluble TNF-α from preformed parent molecules within the cell membrane (Doedens et al., 2003). Thus, the lack of effect of NECA on PMA-induced TNF-α seems to support the theory that AR signaling suppresses TNF-α release at the level of gene expression or RNA stability.
A3AR agonists have been shown to be effective in several different experimental animal models of inflammation. For example, IB-MECA or its 2-chloro derivative Cl-IB-MECA have been shown to prevent lethality induced by endotoxemia (Hasko et al., 1998), to reduce the severity of arthritis in adjuvant-induced arthritis (Baharav et al., 2005), to lessen intestinal damage in experimental colitis (Mabley et al., 2003), and to diminish myocardial ischemia/reperfusion injury (Auchampach et al., 2003). Interestingly, in some of these studies the beneficial effects of the agonists correlated with reduced expression of inflammatory cytokines. Based on the results of the present experiments, it is conceivable that IB-MECA and Cl-IB-MECA provided benefit in these in vivo studies by activating the A2AAR instead of the A3AR, especially since recent studies have demonstrated that IB-MECA and Cl-IB-MECA have higher affinity for the A2AAR than originally appreciated (Murphree et al., 2002). However, it is also possible that A3AR agonists effectively reduce inflammation in vivo via an A3AR-mediated mechanism that suppresses cytokine production from cell types other than the macrophage or by reducing the expression of alternative proinflammatory mediators. In this regard, Gi protein-coupled receptors including the A3AR have been suggested to inhibit the expression of IL-12 from human monocytes (la Sala et al., 2005), an important cytokine that links innate and adaptive immunity by activating macrophages and promoting T helper 1 versus T helper 2 development. A final possibility is that A3AR agonists are effective in vivo by inhibiting other proinflammatory responses of immune cells such as chemotaxis (Knight et al., 1997) or superoxide production (Gessi et al., 2002). Additional studies using more selective A3AR agonists in models of inflammation are warranted.
In summary, we have conclusively determined that the A2AAR is the primary AR subtype that mediates inhibition of TNF-α release from mouse peritoneal macrophages induced by both TLR-dependent and TLR-independent stimuli. Additionally, we have identified a previously unrecognized function of the A2BAR to inhibit TNF-α expression, suggesting that this often-considered proinflammatory receptor may be suppressive under select circumstances. Overall, the results of our study highlight the broad anti-inflammatory potential of AR agonists in treatment of inflammatory disorders.
Acknowledgments
We thank Jayasharee Narayanan and the high-performance liquid chromatography core in the Cardiovascular Research Center at the Medical College of Wisconsin for purification of radioligands.
Footnotes
-
This research was supported in part by National Institutes of Health Grants R01 HL 60051 (to J.A.A.), R01 HL 077707 (to J.A.A.), and F32 HL 073643 (to T.C.W.) and by American Heart Association Grants 0315274Z (to L.M.K.) and 0225454Z (to Z.-D.G.).
-
doi:10.1124/jpet.105.096016.
-
ABBREVIATIONS: TNF-α, tumor necrosis factor-α; IL, interleukin; TLR, Toll-like receptor; LPS, lipopolysaccharide; AR, adenosine receptor; RT-PCR, reverse transcription-polymerase chain reaction; KO, knockout; NF-κB, nuclear factor κB; ZM241385, 4-{2-[7-amino-2-(2-furyl)[1,2,4]triazolo-[2,3-a][1,3,5]triazin-5-ylamino]ethyl}phenol; MRS 1754, 1,3-dipropyl-8-[4-[((4-cyanophenyl)carbamoylmethyl)oxy]phenyl]xanthine; BG 9928, 1,3-dipropyl-8-[1-(4-propionate)-bicyclo-[2,2,2]octyl]xanthine; [125I]I-AB-MECA, N6-(4-amino-3-[125I]iodobenzyl)adenosine-5′-N-methylcarboxamide; MRS 1523, 3-propyl-6-ethyl-5[(ethylthio)carbonyl]-2-phenyl-4-propyl-3-pyridine-carboxylate; WT, wild-type; ZAS, zymosan-activated serum; A23187, calcimycin; PMA, phorbol 12-myristate 13-acetate; FWD, forward; REV, reverse; NECA, adenosine-5′-N-ethylcarboxamide; HEK, human embryonic kidney; CPX, 1,3-dipropyl-8-cyclopentylxanthine; IB-MECA, N6-(3-iodobenzyl)adenosine-5′-N-methyluronamide; CCPA, 2-chloro-N6-cyclopentyladenosine; CGS 21680, 2-[p-(2-carboxyethyl)phenethylamino]-5′-N-ethylcarboxamidoadenosine; Cl-IB-MECA, 2-chloro derivative of N6-(3-iodobenzyl)adenosine-5′-N-methyluronamide.
- Received September 21, 2005.
- Accepted December 7, 2005.
- The American Society for Pharmacology and Experimental Therapeutics