Anti-Inflammatory Preconditioning by Agonists of Adenosine A1 Receptor

Background Adenosine levels rise during inflammation and modulate inflammatory responses by engaging with four different G protein-coupled receptors. It is suggested that adenosine exhibits pro-inflammatory effects through its A1 receptor (A1R), and anti-inflammatory effects through A2A receptor (A2AR). Therefore, understanding of the mechanisms that govern adenosine receptor regulation may advance treatment of various inflammatory disorders. We previously reported that peak A1R expression during leukocyte recruitment, is followed by a peak in A2AR during inflammation resolution. Principal Findings Here, we examined whether A1R activation sequentially induces A2AR expression and by this reverses inflammation. The effect of adenosine on A1R mediated A2AR expression was examined in peritoneal macrophages (PMΦ) and primary peritoneal mesothelial cells (PMC) in vitro. Induction of A2AR was inhibited by pertussis toxin (PTX) and partly dependent on A2AR stimulation. Administration of A1R agonists to healthy mice reduced A1R expression and induced A2AR production in PMC. Mice that were preconditioned with A1R agonists 24 hours before E. coli inoculation exhibited decreased TNFα and IL-6 sera levels and reduced leukocytes recruitment. Preconditioning was blocked by pretreatment with A1R antagonist, as well as, or by late treatment with A2AR antagonist, and was absent in A2AR−/− mice. Conclusions Our data suggest that preconditioning by an A1R-agonist promotes the resolution of inflammation by inducing the production of A2AR. Future implications may include early treatment during inflammatory disorders or pretreatment before anticipated high risk inflammatory events, such as invasive surgery and organ transplantation.


Introduction
Over the past few years, a vast number of investigations have reported the involvement of adenosine in the anti-inflammatory process [1,2]. Adenosine is an endogenous purine nucleoside that is constitutively present in the extracellular spaces at low concentrations. However, in metabolically-stressful conditions such as tissue damage, ischemia and inflammation, adenosine dramatically increases its extracellular levels. Extracellular adenosine levels have been observed to increase by dephosphorylation of ATP in non-immune and immune cells [1] and then to be released through the action of specialized nucleoside transporters [3]. Extracellular adenosine interacts with at least four different receptor subtypes [4][5][6]. The A 2A receptor (A 2A R) interacts with the G protein G s and the A 2B receptor (A 2B R) interacts with the G proteins G s and G q to induce adenylyl cyclase activity and elevate cAMP levels. In contrast, ligation of adenosine to the A 1 receptor (A 1 R) or to the A 3 receptor (A 3 R), through interaction with members of the G i /G o family, inhibits adenylyl cyclase activity and decreases cAMP levels [7]. A 1 R exerts a pro-inflammatory response by enhancing phagocytosis [8], promoting chemotaxis [9,10] and enhancing neutrophils adherence to endothelium during inflammatory process [11]. In contrast, engagement of A 2A R inhibits neutrophils adherence to endothelium during inflammation [12] and inhibits the activation of neutrophils, monocytes platelets and T-cells [13][14][15]. In animal models, A 2A R-agonists can prevent lethal response to bacterial LPS and sepsis [16,17].
Since each of these receptor subtypes has a unique physiological profile and a particular affinity to its ligand, the inflammatory state is determined by both extracellular adenosine concentrations and by the distribution and expression levels of its receptor subtypes. It has been shown that the expression of adenosine receptors is regulated by factors that are involved in the inflammatory response, such as LPS [18], pro-inflammatory cytokines [19][20][21], growth factors [22,23] and glucocorticoids [24]. Recently, we have shown in a model of peritonitis that shortly following inoculation, A 1 R mRNA and protein levels are upregulated on peritoneal mesothelial cells (PMC), reaching a peak in the initial phase of the inflammatory process [19]. Interestingly, concomitant with the resolution phase of peritonitis, we observed a decrease in A 1 R expression levels and an elevation of adenosine and A 2A R levels. The coordinated kinetics of adenosine and its receptors led to the hypothesis that adenosine differentially regulates its own receptors. Since the two receptors, A 1 R and A 2A R, have opposing biological effects, and A 1 R domination precedes the elevation of A 2A R, we sought to examine whether A 1 R activation would be one of the factors that trigger the anti-inflammatory phase, and whether this action is mediated by upregulation of the A 2A R.
To test our hypothesis, we examined the effect of adenosine receptor agonists and antagonists in vivo in a model of peritonitis induced by E. coli inoculation. This model has particular clinical significance because peritonitis is commonly caused by pathological processes of the gastrointestinal tract or as a complication of abdominal surgery. In vitro, we examined the regulation of the receptors on the cell surface of PMW, which are the first line of cellular defense against bacterial invasion in the peritoneum [25], and on PMC, the cells that line the peritoneal membrane and therefore play an important role in transferring inflammatory signals from the peritoneal cavity to the blood vessels [26][27][28][29][30]. We demonstrate that A 1 R activation triggers the switching of adenosine receptor subtype from A 1 R to A 2A R. By the antiinflammatory effects of the ligation of adenosine to the A 2A R, the described receptor subtype switch alters the progression of inflammation toward resolution.

Materials and Methods
Mice, bacterial strains and drugs CD1 female mice aged 10 to 12 weeks (Harlan, Jerusalem, Israel) were maintained in the animal laboratory of the Soroka Medical Center. Experiments were conducted with the permission of the Israel Committee for Animal Experiments. A 2A R 2/2 mice whose phenotype is well established in the literature were graciously kindly donated by Catherine Ledent (Université Libre de Bruxelles) [31].

Induction of peritonitis and treatment protocol
Peritonitis was induced in mice by intraperitoneal (i.p.) inoculation of a sub-lethal dose of E. coli (3.6610 9 CFU). Adenosine agonists and antagonists were injected i.p. before E. coli inoculation.

Sera and peritoneal lavage fluids collection, leukocyte counting and cytokine detection
At different time points after E. coli inoculation, animals were anesthetized. 1 ml syringe flushed with heparin was used to draw intracardial blood sample. The samples were stored on ice before centrifugation at 1,000 g at 4uC for 10 minutes. The cell-free supernatants were collected and frozen at 220uC until assayed by ELISA. Peritoneal lavage was performed with 5 ml phosphate buffer saline (PBS) containing 2% BSA and 5 mM EDTA. After centrifugation at 400 g for 10 minutes, the cell-free supernatants were removed and frozen at 220uC until analysis. TNFa and IL-6 levels were determined by commercial ELISA kits (Biolegend, San Diego, CA and R&D Systems, Minneapolis, MN, respectively). Cells were washed once, and total leukocytes were counted after trypan blue staining using an improved Neubaur hemocytometer. Cell counts and ELISA were performed blindly on coded samples.

Scraping of mice PMC
Following treatment, animals were anesthetized and PMC were scraped from the peritoneal membrane. The cells were stored on ice before centrifugation at 400g and 4uC for 10 minutes. Cells were harvested with lysis buffer for analyzing mRNA levels or with RIPA (150 mM NaCl, 50 mM Tris HCl pH-7.4, 1% NP-40, 0.25% Na deoxycholate, 1 mM EGTA) including protease inhibitor cocktail (Sigma) for analyzing protein levels.

Preparation of cultured PMC and PMW
To prepare PMC, the peritoneum was removed from eight newborn (two-week old) mice and isolated, as previously described [32]. To assess the purity of mesothelial cells, samples of each PMC preparation were morphologically inspected, as previously described [33]. Cells were grown in M199 and supplemented with 10% heat-inactivated FCS, 2 mmol/l L-glutamine and 100 U/ml penicillin and 100 mg/ml streptomycin (Biological Industries, Bet Haemek, Israel). Experiments were performed on cells from the second to fourth passages. To prepare PMW, mice were injected intraperitoneally with 3 ml of 3% thioglycollate (Difco, Sparks, MD). After 3 days, peritoneal cells were collected by lavage and seeded onto 12-well plates in RPMI supplemented with 10% heatinactivated FCS, 2 mmol/l L-glutamine and 100 U/ml penicillin and 100 mg/ml streptomycin. Non-adherent cells were subsequently removed by washing with PBS. In experiments, to simulate the graduate increase in adenosine levels found in vivo, cells were treated with increasing doses of adenosine or CHA with or without DPCPX (9 hours with 0.1 mM or 3 hours with 0.1 mM and then 6 hours with 1 mM or 3 hours with 0.1 mM, then 3 hours with1 mM and then 3 hours with 10 mM).

mRNA analysis
Total RNA was extracted from PMC or PMW using the Versagene RNA cell kit (Gentra systems, Minneapolis, MN). cDNA was prepared as previously described [29]. Reaction was carried out in Rotor-Gene real time PCR machine (Corbett-Research, Northlake, Australia).

Western blotting analysis
Cell lysates was centrifuged at 13,000 g for 30 minutes and then supernatants were collected for total protein determination by the BCA protein assay kit (Pierce, Rockford, IL). 30 mg of total protein from each sample was subjected to 10% SDS-PAGE under reducing conditions and after heating. The gels were blotted onto a PVDF membrane (Bio-Rad, Hercules, CA) and probed with the following specific antibodies: rabbit anti-adenosine A 2A R (Santa Cruz Biotechnology, Santa Cruz, CA) or rabbit anti-A 1 R (Alpha Diagnostic International, San Antonio, TX) or goat anti-b-actin (Santa Cruz Biotechnology). The membrane was then probed with goat anti-rabbit immunoglobulins Ig-conjugated to peroxidase agent (Santa Cruz Biotechnology) or with donkey anti-goat IgG conjugated to peroxidase agent (Jackson Immuno Research laboratories, West Grove, PA). Antigen-antibody complexes were subsequently visualized by the EZ-ECL Chemiluminescence Detection kit for HRP (Biological Industries).

Statistical Analysis
Data are presented as mean6SEM. Statistical analysis was performed by t-test or ANOVA followed by Tukey post test. P values below 0.05 were considered significant.

Adenosine receptors exhibit unique expression kinetics in peritoneal leukocytes following bacterial inoculation
It has been shown that adenosine is upregulated during peritonitis [19]. We therefore examined the regulation of adenosine receptors in peritoneal leukocytes and found that the A 1 R and A 2A R are upregulated during the first 48 hours of peritonitis. However each of the subtypes exerted unique kinetics. As shown in figure 1, A 1 R mRNA levels were maximal at 6 hours after inoculation and returned to basal levels at 24 hours, while A 2A R mRNA levels gradually increased and reached maximum at 24 hours.

Adenosine induces the expression of A 2 R in a dosedependent manner
Since both adenosine and adenosine receptors are upregulated upon bacterial inoculation [19], we wanted to elucidate whether the regulation of adenosine receptors is adenosine-dependent. In order to simulate the gradual and accumulative increase of adenosine that is observed in vivo, we treated cultured PMCs with multiple and increasing concentrations of adenosine (0.1, 1 and 10 mM at 3 hours intervals). As shown in Figure 2, adenosine induced the expression of A 2A R mRNA levels in a dose dependent manner. However, there was no change in A 1 R mRNA levels upon treatment with the different concentrations of adenosine.

Adenosine regulates A 2A R expression through A 1 R
Since A 1 R is elevated shortly after bacterial inoculation ( Figure 1) and is followed by elevation of A 2A R expression, we wanted to examine whether the induction of A 2A R by adenosine may be mediated by the A 1 R. Therefore, we treated PMC and PMW with 0.1, 1 and 10 mM at 3 hour intervals with A 1 R agonist (CHA) or adenosine in the presence or absence of the A 1 R antagonist (DPCPX, 50 nM). As shown in Figure 3A and B, CHA upregulated mRNA levels of the A 2A R while treatment with adenosine in the presence of the DPCPX blocked A 2A R upregulation both in PMW and PMC respectively. In contrast, stimulation with CGS, an A 2A R agonist failed to induce A 2A R ( Figure 3D).
Ligation of adenosine to the A 1 R is mediated through the interaction with members of the G i /G o family and inhibits adenylyl cyclase activity. To elucidate the mechanism by which A 1 R induces A 2A R elevation, we pretreated PMC with PTX, a G i inhibitor ( Figure 3C). Pretreatment with PTX blocked the effect of CHA on A 2A R mRNA levels.
For effective induction of A 2A R a sequential induction with increasing doses of adenosine or CHA (0.1, 1, 10 mM) were  necessary suggesting the involvement of an additional adenosine receptor. CCPA, a specific A 1 R agonist, was less effective than CHA, an A 1 R agonist with lower specificity ( Figure 3D). ZM241385, an A 2A R antagonist, partially blocked the induction of A 2A R mRNA that was induce by adenosine ( Figure 3D) or CHA (data not shown), which suggests that in addition to the requirement of A 1 R stimulation, A 2A R ligation supports its own induction. Treatment with adenosine in the presence of A 3 R (MRS1220, 100nM) or A 2B R antagonist (MRS1754, 50nM) did not alter on A 2A R mRNA levels (data not shown).

Effect of A 1 R agonist on the expression of A 2A R and A 1 R in vivo
We examine whether the A 1 R agonist also regulates the levels of the A 2A R in vivo. We determined the mRNA and protein levels of the A 2A R and the A 1 R in mice that were administered an A 1 R agonist (CHA, 0.1 mg/kg). We found that A 2A R mRNA levels increase ,3 fold and that A 2A R protein levels increase ,2.5 fold, compared to vehicle. In contrast, as shown in Figure 4, both A 1 R mRNA and protein levels decreased in the presence of A 1 R agonist by ,6 and ,2 fold, respectively.

Pretreatment with the A 1 R agonist reduces serum cytokine levels and peritoneal leukocyte recruitment during inflammation
Since we showed that A 2A R levels are upregulated through the activation of A 1 R both in vitro and in vivo, we wanted to elucidate whether pretreatment of A 1 R agonist before inoculation would upregulate the expression of A 2A R and lead to advancement of the anti-inflammatory response via A 2A R. For this, mice were treated with an A 1 R agonist (CHA, 0.1 mg/kg) 24 hours before inoculation of E. coli, after which sera were analyzed for IL-6 and TNFa levels. As shown in Figure 5A, we found a significant reduction both in serum IL-6 and TNFa levels 12 hours after inoculation (to 25% and 38% from vehicle, respectively).
Since PMC express an array of chemokines which cause accumulation and activation of leukocytes in tissues, we wanted to examine changes in the levels of CXC chemokines, MCP-1 and MIP-2, following pretreatment with A 1 R agonist. As a result of pretreatment with the A 1 R agonist (CHA 0.1mg/kg), MCP-1 and MIP-2 mRNA level decreased in comparison to vehicle, as determined 12 hours after inoculation ( Figure 5B). In accordance with reduced chemokine levels, leukocyte recruitment significantly decreased 24 hours after inoculation to 66% from vehicle, as determined in lavage fluid ( Figure 5C).

A 1 R-agonist preconditioning is blocked by a selective A 1 R antagonist
To ensure that the anti-inflammatory state was mediated by selective activation of the A 1 R, we examined the anti-inflammatory effect of low-dose CHA and an additional specific A 1 Ragonist CCPA, in the presence of a specific A 1 R antagonist (DPCPX). As shown in Figure 6, treatment with either CCPA (A) or CHA (B) significantly reduced serum and lavage IL-6 and TNFa levels. However, pretreatment with an A 1 R antagonist (DPCPX, 1 mg/kg) 2 hours before administration of A 1 R agonist blocked the effect of 0.02 mg/kg CHA, 0.1 mg/kg CHA (data not shown) and 0.1 mg/kg CCPA.  Modulation of the inflammatory response due to pretreatment with the A 1 R agonist is A 2A R-dependent To prove that the modulation in the inflammatory response ( Figure 5) is mediated by A 2A R, we treated animals with an A 2A R antagonist (30 min before inoculation, ZM241385, 1 mg/kg). As shown in figure 7, blockade of the A 2A R caused an increase in serum and lavage IL-6 and TNFa levels to similar levels found in infected mice administrated with vehicle alone. As expected, administration of A 2A R agonist (30 minutes before inoculation, CGS21680, 1 mg/kg) reduced IL-6 and TNFa levels in serum and lavage fluids to levels comparable to those found in CHA-treated animals. In concordance, pretreatment of A 2A R 2/2 mice with A 1 R agonist resulted in unchanged serum IL-6 and TNFa levels ( Figure 7C), as well as chemokine mRNA levels in PMC (data not shown). However, in WT mice there was a significant reduction both in cytokine levels and mRNA chemokine levels (data not shown). These data suggest that the modulation of the inflammatory response caused by pretreatment with A 1 R agonist is, indeed, mediated by A 2A R.

Discussion
The study presented here demonstrates a novel mechanism of adenosine receptor subtype autoregulation. Since adenosine action is mediated through at least four different receptors, each of which exhibits a unique affinity and opposing signaling pathways, the regulation of subtypes expression is critical for determining the outcome of adenosine activity [5]. Others and we have shown that adenosine receptors are regulated by various inflammatory mediators and multiple endogenous factors [24]. For example, we found that A 2A R mRNA and protein levels are upregulated in human PMC following treatment with IL-1b and TNFa, while treatment with IFNc strongly decrease A 2A R expression both alone and in combination with IL-1b and TNFa [19]. In the same study, we show that following inoculation, adenosine receptor levels on PMCs are sequentially upregulated and that adenosine is induced following inoculation and reaches peak levels at 24 hours [19]. The A 1 R is induced during the first phase of leukocyte recruitment and the A 2A R is induced later, at the resolution phase of peritonitis [19]. In the present study, we obtained the same pattern of adenosine receptor expression on peritoneal leukocytes.  These results suggest that both mesothelial cells and the recruited leukocytes are highly synchronized in their response to adenosine. Furthermore, this sequential elevation of the A 1 R and the A 2A R on PMC and leukocytes suggests that adenosine may regulates its receptors. Both our in vitro and in vivo data in the current study support this suggestion; we found that adenosine significantly upregulates A 2A R expression levels in isolated PMC in a dose dependent manner.
Of all adenosine receptor subtypes, A 1 R exhibits the highest affinity for adenosine (K i = 10 nM) [34], implying that A 1 R is activated at the low levels of adenosine produced during the initiation of inflammation. This early activation of A 1 R receptor may enable the induction of A 2A R. The A 1 R agonist, CHA, significantly induced the expression of A 2A R, while treatment with the A 1 R antagonist, DPCPX, or with PTX, a G i inhibitor, blocked A 2A R induction by adenosine, indicating that A 1 R ligation is necessary for the induction of A 2A R. Treatment with CGS21680, an A 2A R agonist, did not induce the expression of the A 2A R. However, treatment with the A 2A R antagonist in the presence of adenosine partially blocked A 2A R induction. Therefore, one can conclude that A 2A R ligation by elevated levels of adenosine is required to support the initial signal of A 1 R.
According to our in vitro data, mice treated with CHA exhibited a significant 2-3 fold increase in A 2A R mRNA and protein levels as determined, in PMCs compared to untreated animals. Interestingly, mRNA and protein A 1 R levels were significantly downregulated by these same treatments in PMCs (6-and 2-fold decrease, respectively), suggesting that A 1 R receptor may be responsible for the ''switching'' between the two receptor subtypes during inflammation. In Support of our findings, Schnurr et al. showed that in immature plasmacytoid dendritic cells (PDCs) adenosine activates A 1 R, which induces chemotaxis; however, in mature PDCs, A 1 R is replaced by the A 2A R, which inhibits cytokine production [9].
In order to understand the physiological role of the exchange between the two receptors, we examined whether ligation of the A 1 R will trigger the induction of the A 2A R and lead to an advancement of the resolution phase of the inflammatory process. We found that preconditioning with an A 1 R agonist significantly reduces the inflammatory response to bacterial challenge. CHA or CCPA administration at 24 hours before inoculation significantly reduced sera and peritoneal levels of the pro-inflammatory cytokines TNFa and IL-6, and reduced mRNA levels of chemokines on PMC as well as leukocyte recruitment to the peritoneum. The anti-inflammatory effect induced by pre-treatment (24 hours) with A 1 R agonist was also achieved by a specific A 2A R agonist (CGS21680) administered to animals 30 minutes before bacterial inoculation. Pre-treatment with CHA or CCPA had no antiinflammatory effect in animals that were administered with the A 1 R antagonist, DPCPX 2 hours before agonists or A 2A R antagonist, ZM241385 30 minutes before inoculation or when A 2A R 2\2 animals were examined. The marked blocking effect of ZM241385 and the lack of effect of CHA in A 2A R knockout animals clearly indicate that the anti-inflammatory effects of the A 1 R agonist are mediated by the A 2A R.
Elevation of cAMP usually down-regulates the inflammatory response [5]. Since A 1 R is a G i coupled receptor that suppresses the induction cAMP, it is not surprising that this receptor had no direct anti-inflammatory effect. High expression of A 1 R implies that immediately after inoculation, decreased cAMP levels give rise to local pro-inflammatory cytokines and leukocyte migration, hence allowing an adequate and effective immune response to the invading microorganisms. In contrast, the increase in A 2A R at late phases of peritonitis is probably associated with elevated cAMP levels, which markedly decrease local pro-inflammatory cytokine levels and leukocyte recruitment, hence restraining inflammatory flames ( Figure 8).
In summary, our study sheds light on the sequential autoregulation of adenosine receptor subtypes. The mechanism we have describes may directly participate in the propagation of the compensatory anti-inflammatory response syndrome (CARS), which follows systemic inflammation in trauma patients. Whether patients with CARS exhibit elevated adenosine levels pursuing traumatic insult should be explored. These findings may also have future implications for clinical treatments by combining pretreatment with an A 1 R agonist and subsequent A 2A R agonist to enhance the anti-inflammatory effect, or to promote antiinflammation by endogenous adenosine at the site of inflammation. As such, preconditioning with an A 1 R-agonist could be used in preparation of tissue for transplantation or to induce an antiinflammatory and immunosuppressive state in patients before invasive surgery and organ transplantation.