A_2A adenosine receptors and C/EBPβ are crucially required for IL-10 production by macrophages exposed to Escherichia coli

The innate immune system is responsible for identifying infectious pathogens and eliciting specific immune responses to eliminate the pathogens. Pathogen recognition can be mediated by a set of pattern recognition receptors. One class of pattern recognition receptors involves the Toll-like receptors (TLRs), which consist of a family of transmembrane molecules found in a broad range of organisms from Drosophila to higher mammals.^1,,,–5  TLR4 was first recognized as the receptor mediating the macrophage stimulatory effect of endotoxins (lipopolysaccharide, LPS) of Gram-negative bacteria, such as Escherichia coli. Other Gram-negative bacterial components can also induce macrophage activation via various other pattern recognition receptors, which include TLR2 (lipopeptide receptor), TLR5 (flagellin receptor), and TLR9 (CpG DNA receptor), as well as nucleotide-binding oligomerization domain (nucleotide-binding oligomerization domain 1, peptidoglycan receptor) and nucleotide-binding oligomerization domain 2 (muramyl dipeptide receptor).^4,5  Lipoteichoid acid, a component of Gram-positive bacteria, activates cellular responses via TLR2.^4,5 

The engagement of pattern recognition receptors by microbial components triggers the activation of signaling cascades leading to the induction of genes involved in antimicrobial host defense such as cytokines and chemokines.^6,–8  Myeloid differentiation factor 88 (MyD88) has been shown to be a critical adaptor protein linking TLRs to several downstream intracellular pathways.^4,5  Many of the activating signals originating from TLRs and MyD88 converge on tumor necrosis factor receptor–associated factor 6 (TRAF6), which transduces the signals toward the nuclear factor-κB system and mitogen-activated protein kinases (MAPKs), resulting in induction of genes involved in pro-inflammatory responses.^4,5 

Much less is known about the regulation of anti-inflammatory cytokines and their genes, the most prominent of which is IL-10.^9,10  IL-10 is an inducible gene and several transcriptional factors have been implicated in its regulation, which include Sp1,^11,–13  Sp3,¹²  C/EBPδ and C/EBPβ,¹⁴  STAT3,¹⁵  and c-Maf.¹⁶  IL-10 gene expression also requires transient remodeling of the IL-10 promoter that occurs as a result of MAPK activation.¹⁷  In addition, we and others^18,19  have recently shown that IL-10 production is also regulated at various posttranscriptional levels, including alterations in mRNA stability and translation efficacy.

Adenosine, an endogenous purine nucleoside, is a biologically active extracellular signaling molecule that is formed at sites of metabolic stress associated with hypoxia, ischemia, trauma, or inflammation. Adenosine mediates its effects via engaging A₁, A_2A, A_2B, and A₃ adenosine receptors,²⁰  which are all expressed abundantly on monocytes and macrophages.²¹  Stimulation of adenosine receptors has been shown to convert the phenotype of TLR4 (LPS)-stimulated macrophages from a pro-inflammatory to an anti-inflammatory one.^21,,,,,–27  Most notably, adenosine receptor activation has been documented to upregulate the TLR4-induced production of IL-10.^28,–30  In addition to TLR4-stimulated macrophages, adenosine enhanced IL-10 production by tumor necrosis factor-α–stimulated or hydrogen peroxide–stimulated human monocytes.³¹ 

We recently demonstrated that genetic inactivation of the A_2A receptor almost abolished IL-10 production in the cecal ligation and puncture model of murine polymicrobial sepsis, which suggested that A_2A receptor activation is required for IL-10 production after bacterial stimuli.³²  Because during sepsis the host is challenged with whole bacteria that possess multiple components capable of triggering IL-10 production via various pattern recognition receptors, the relevance of findings with limited stimuli such as LPS, tumor necrosis factor-α, or hydrogen peroxide to bacterial infections is not clear. In addition, the cellular mechanisms by which A_2A receptor activation regulates IL-10 production in response to clinically relevant stimuli, such as whole bacteria, are unknown. Here we show that A_2A receptor activation and bacteria synergistically induce IL-10 production by macrophages, a process that is critically dependent on the transcription factor C/EBPβ.

Adenosine, the selective A₁ receptor agonist CCPA (2-chloro-N⁶-cyclopentyladenosine), A_2A receptor agonist 2-p-(2-carboxyethyl)phenethyl-amino-5′-N-ethyl-carboxamidoadenosine (CGS21680), A₃ receptor agonist IB-MECA [N⁶-(3-iodobenzyl)adenosine-5′-N-methyluronamide], nonselective agonist 5′-N-ethylcarboxamidoadenosine (NECA), and actinomycin D were purchased from Sigma-Aldrich (St. Louis, MO). The selective A_2A receptor antagonist 4-(2-[7-amino-2-(2-furyl)[1.2.4]triazolo[2.3-a][1.3.5]triazin-5-ylamino] ethyl) phenol (ZM241385) was purchased from Tocris Cookson (Ellisville, MO). The p38 pathway inhibitor SB203580 and p42/44 pathway inhibitor PD98059 were purchased from Calbiochem (San Diego, CA). Stock solutions of the various agonists and protein kinase inhibitors were prepared using dimethylsulphoxide. Lipoteichoic acid prepared from Staphylococcus aureus was from Invitrogen (Carlsbad, CA).

Male CD-1 mice were purchased from Charles River Laboratories (Wilmington, MA). A_2A receptor knockout (KO) mice and their wild-type (WT) littermates^32,33  and MyD88 KO and WT mice on the C57Bl/6J background,³⁴  kindly provided by Dr Shizuo Akira (Osaka University, Osaka, Japan) were bred in a specific pathogen-free facility using founder heterozygous male and female mice. Male TLR4 KO (C57BL/10ScNJ) and WT (C57BL/10ScSnJ) littermates mice were obtained from the Jackson Laboratory (Bar Harbor, ME). A_2B receptor KO mice were acquired from Deltagen Inc. (San Mateo, CA). Mice were backcrossed 10 generations onto a C57Blk/6J background. A_2B-null mice were generated by the insertion of a lacZNeo cassette into exon 1 of an A_2B receptor WT allele (Figure S1A, on the Blood website via the Supplemental Data link at the top of the online article, shows genotyping details). Measurement of A_2B receptor transcripts in large intestine and bladder revealed an absence of A_2B receptor transcripts in mice homozygous for the targeted A_2B receptor allele (Figure S1B). A normal Mendelian ratio of inheritance of the mutant allele was seen after heterozygous matings. A_2B receptor KO mice survive a normal lifespan with no obvious external defects and reproduce normally (data not shown). All mice were maintained in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals, and the experiments were approved by the New Jersey Medical School Animal Care Committee. Thioglycolate-elicited mouse peritoneal macrophages,¹⁸  C/EBPβ-deficient macrophages, control macrophages (kind gifts from Valeria Poli, Department of Genetics, Biology, and Biochemistry, University of Turin, Turin, Italy),^35,36  and RAW 264.7 macrophages (ATCC) were grown in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum, 50 U/mL penicillin, 50 μg/mL streptomycin, and 1.5 mg/mL sodium bicarbonate in humidified atmosphere of 95% air and 5% CO₂.

E coli strain K12 was grown in Luria Bertani (LB) medium overnight at 37°C with shaking at 220 rpm. Cells were centrifuged at 3000g for 15 minutes at 4°C and resuspended in phosphate-buffered saline at a density of 10⁸/mL. Thereafter, bacteria were heat-inactivated by incubation at 65°C for 90 minutes. The killed cells were collected by centrifugation at 3000g for 15 minutes at 4°C and resuspended in phosphate-buffered saline at a density of 6 × 10⁹/mL.

Total RNA was prepared from macrophages and reverse-transcribed, as previously described.³²  For detection of IL-10 and TRAF6 mRNA, real-time polymerase chain reaction commercial kits (Applied Biosystems, Foster City, CA) were used, and all data were normalized to constitutive rRNA values. The Applied Biosystems 7700 sequence detector was used for amplification of target mRNA, and quantitation of differences between treatment groups was calculated according to the manufacturer's instructions.

The mutated IL-10 promoter was prepared by gene synthesis by GenScript (Piscataway, NJ). The following nucleotides were changed in the C/EBP consensus sequences of the 5′-flanking region of the IL-10 gene between −410/−390 relative to the transcription start site: −410-TGGAGGAAACAATTATTTCTC-390, The mutated construct had the following sequence: −410 TGGActgtgCAATTAgtaCTC-390. The mutated DNA fragment was inserted into pGL2-Basic vector. The mutation sites were confirmed by DNA sequencing.

RAW 264.7 cells were transiently transfected using FUGENE 6.0 transfection reagent (Roche, Indianapolis, IN). For transfection, 0.5 × 10⁶/mL cells were plated in a 24-well plate. The next day, the cells were transfected with 0.4 μg of IL-10 reporter plasmids (kind gifts from Stephen T. Smale, University of California, Los Angeles, School of Medicine, Los Angeles, CA), and an A_2A adenosine receptor overexpressing pA_2A–cytomegalovirus construct (Origene) or C/EBP reporter plasmid (Stratagene). All transfections were performed at 37°C overnight, after which procedure the cells were washed with Dulbecco modified Eagle medium and treated with heat-killed E coli and/or adenosine for 8 hours. For reporter assays, whole-cell extracts were prepared using 80 μL 1 × passive lysis buffer (Promega, Madison, WI). Luciferase activity was determined from 20 μL cell extract.

Specific hairpin siRNA (shRNA) oligonucleotide constructs for murine TRAF6 silencing were designed using the Ambion siRNA design algorithm, synthesized in the New Jersey Medical School Biotechnology Central facility, and cloned into the pSilencer3.1-H1neo expression vector (Ambion, Austin, TX), following the manufacturer's protocol. The shRNA constructs were designed targeting both the 5′ end and the 3′ of the gene. As controls, nonspecific shRNA with the same nucleotide composition as the specific inserts, but lacking significant homology with any sequence in the murine genome database, were also prepared. RAW 264.7 cells were transfected with shRNA-containing plasmids using Superfect transfection reagent (Qiagen, Valencia, CA), following the manufacturer's protocol. G418-resistant colonies were selected after approximately 2 weeks of growth. Six single clones of each shRNA transfection were selected and grown. G418 was maintained in the medium for the growth of the clones.

RAW 264.7 cells were stimulated with adenosine and E coli for various periods and nuclear protein extracts were prepared as described previously.³⁷  The C/EBP consensus oligonucleotide probe used for electrophoretic mobility shift assay was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Electrophoretic mobility shift assays were performed as described previously for nuclear factor-κB³⁷  or CREB.³⁸  For supershift studies, samples were preincubated with a C/EBPβ or C/EBPδ antibody (Santa Cruz Biotechnology).

C/EBPβ and C/EBPδ protein levels were analyzed using 10 μg of nuclear extracts using antibodies raised against C/EBPβ and C/EBPδ, respectively. Western blotting to detect p38 and p42/44 phosphorylation was performed as we have described previously³⁸  using rabbit anti-phospho-p42/44 and monoclonal mouse anti-phospho-p38 (both from Cell Signaling, Danvers, MA).

Peritoneal macrophages in 96-well plates (2 × 10⁵/mL) were treated with adenosine or various adenosine receptor agonists, followed immediately by addition of E coli. IL-10 levels in cell supernatants were determined by enzyme-linked immunosorbent assay.³² 

Values in the figures are expressed as mean plus or minus the SEM of the indicated number of observations. Statistical analyses of the data were performed by Student t test or one-way analysis of variance followed by the Dunnett test, as appropriate.

We have shown recently that genetic inactivation of A_2A receptors almost completely abolishes IL-10 production induced by a polymicrobial challenge in mice, suggesting that A_2A receptor activation is required for IL-10 production after bacterial stimuli.³²  To begin to examine the nature of the interaction of bacterial stimuli and A_2A receptor signaling in vitro, peritoneal macrophages were obtained from A_2A adenosine receptor KO and WT mice and treated with E coli and adenosine. Macrophages from A_2A receptor WT mice produced low levels of IL-10 after exposure to E coli but not adenosine (Figure 1A). Challenging E coli–treated WT cells with adenosine dramatically increased IL-10 levels (Figure 1A). However, neither adenosine (Figure 1A) nor E coli (Figure 1A) was capable of eliciting IL-10 release by A_2A KO mouse macrophages. Moreover, the combination of these 2 agents was also ineffective in triggering IL-10 release by A_2A KO macrophages (Figure 1A). We next treated macrophages with E coli in the presence or absence of various adenosine receptor agonists. Both the selective A₁ receptor agonist CCPA and A₃ receptor agonist IB-MECA failed to mimic the stimulatory effect of adenosine on IL-10 release. However, both the selective A_2A receptor agonist CGS21680 and nonselective agonist NECA increased IL-10 release by E coli–challenged macrophages, with CGS21680 being the most potent (Figure 1B).

We then investigated the role of A_2B receptors using A_2B receptor KO and WT mice, because previous studies have implicated A_2B receptors in regulating IL-10 release.^18,22 E coli induced the release of IL-10 by macrophages from both A_2B WT and KO mice to the same extent. Adenosine enhanced substantially this E coli–induced IL-10 release in A_2B WT mice, which was slightly (approximately 10%) but consistently decreased in A_2B KO mice (Figure 1C).

Recent studies have illustrated that bacteria contain a number of components that are able to elicit IL-10 secretion independently of LPS and TLR4.³⁹  To determine whether bacterial components that are not TLR4 ligands could be involved in the synergistic upregulation of IL-10 release after combined administration ofE coli and adenosine, we treated peritoneal macrophages obtained from TLR4 KO and WT mice with adenosine and heat-killed E coli. We found that E coli was capable of inducing low levels of IL-10 in TLR4 WT macrophages, which was decreased in TLR4 KO cells (Figure 2A). Exogenous adenosine synergistically upregulated (> 10-fold) E coli–induced IL-10 release in TLR4 WT cells. Although the combined exposure of adenosine and E coli induced lower levels of IL-10 in TLR4 KO macrophages than the same treatment in TLR4 WT cells, adenosine upregulated E coli induced IL-10 release to the same degree (> 10-fold) in TLR4 WT and TLR4 KO cells (Figure 2A). We next examined whether adenosine affected the release of IL-6, a pro-inflammatory cytokine, independently of TLR4. E coli triggered IL-6 release in TLR4 WT and to a lesser extent in TLR4 KO macrophages, and adenosine decreased this IL-6 release in both WT and KO cells approximately to the same degree (by approximately 30%; Figure 2B).

To confirm that adenosine can upregulate bacteria-induced IL-10 release independently of TLR4, we challenged peritoneal macrophages with heat-killed S aureus, a Gram-positive bacterium, which, unlike Gram-negative E coli, is thought to act in a TLR4-independent fashion to elicit macrophage activation. We observed that S aureus increased IL-10 levels, which were further enhanced by adenosine (Figure 2C). We then investigated the effect of adenosine on IL-10 release that was induced by the specific TLR2 agonist lipoteichoic acid prepared from S aureus. This TLR2 agonist elicited the release of IL-10 by macrophages, and adenosine boosted the TLR2 agonist-induced IL-10 level (Figure 2D).

Collectively, our results demonstrate that E coli increases IL-10 release in macrophages, and this effect is partially dependent on TLR4. Moreover, adenosine upregulates bacteria-induced IL-10 secretion and down-regulates IL-6 release by a mechanism that does not require TLR4.

MyD88 has been documented to contribute to intracellular signaling from all TLRs except TLR3. In addition, recent evidence indicates that TLR signaling consists of both MyD88-dependent and a MyD88-independent pathways.⁴⁰  Because of the central role of MyD88 in many bacteria-induced macrophage responses,^41,42  we first studied whether MyD88 regulated IL-10 release in macrophages challenged with E coli, adenosine, or the combination of E coli and adenosine. There was no difference in IL-10 release by peritoneal macrophages obtained from MyD88 WT and KO mice after E coli treatment and the IL-10 level was upregulated to the same degree in WT and KO macrophages after adenosine treatment (Figure 3A).

Previous work has shown that in the absence of MyD88 the production of IL-6 is decreased or completely abolished depending on which TLR activates IL-6 gene expression.^34,42  In addition, contrary to its stimulatory effect on IL-10 production, adenosine has been shown to suppress IL-6 production by TLR4-stimulated macrophages.⁴³  IL-6 release in E coli–stimulated MyD88 KO macrophages was decreased compared with WT macrophages, and adenosine decreased the IL-6 level in both MyD88 KO and WT peritoneal macrophages (Figure 3B). These results indicate that the release of the anti-inflammatory cytokine IL-10 and pro-inflammatory IL-6 were regulated differentially by adenosine, and that while IL-10 release is completely MyD88-independent, IL-6 release is partially MyD88-dependent.

TRAF6 has been shown to be a crucial intracellular protein for the induction of pro-inflammatory genes in macrophages in response to bacteria or TLR agonists.^44,–46  Therefore, we assessed using an shRNA approach whether TRAF6 would be required for the effect of E coli and/or adenosine also on IL-10 release. First, we confirmed that TRAF6 shRNA–expressing cells expressed 83% less TRAF6 mRNA than cells stably transfected with a control vector expressing scrambled TRAF6 sequences (Figure 4A). Then, we treated cells with heat-killed E coli and/or adenosine and measured cytokine levels. In agreement with previous data,⁴⁴  IL-6 release by E coli–induced macrophages was decreased in TRAF6 shRNA–expressing cells compared with controls (scrambled shRNA-expressing cells; Figure 4B), indicating that IL-6 release after inflammatory stimuli requires TRAF6. Moreover, adenosine decreased IL-6 release in control cells, but did not reduce it in TRAF6 shRNA–expressing cells (Figure 4B), indicating that the suppressive effect of adenosine was TRAF6-dependent. In contrast, IL-10 levels were markedly higher in TRAF6 shRNA–expressing cells than in control cells after both E coli and E coli/adenosine treatment (Figure 4C). The fact that adenosine was less efficacious in potentiating the stimulatory effect of E coli on IL-10 release in RAW 264.7 cells (Figure 4) than in peritoneal macrophages (Figures 1,Figure 2–3) is probably a reflection of the low expression of A_2A receptors on RAW cells.³⁷  In sum, these findings indicate that TRAF6 is not required for the effect of E coli and adenosine in inducing IL-10, but TRAF6 negatively modulates this synergistic interaction.

To investigate the mechanisms that are responsible for the increased release of IL-10 by bacteria-activated macrophages both in the absence and presence of adenosine, we determined IL-10 mRNA levels from CD-1 peritoneal macrophages using real-time polymerase chain reaction. We found that E coli increased IL-10 mRNA levels by approximately 4-fold, and adenosine further augmented by approximately 8-fold the E coli–induced accumulation of IL-10 mRNA (Figure 5A). In addition, adenosine had no effect on IL-10 mRNA levels in the absence of E coli (Figure 5A). The transcriptional inhibitor actinomycin D prevented the synergistic effect of E coli and adenosine in inducing IL-10 mRNA and protein accumulation arguing for the transcriptional nature of this synergism (Figure 5A,B).

We also studied the effect of adenosine on E coli–induced IL-10 promoter activity by transfecting RAW 264.7 cells with a construct in which luciferase expression was driven by the full-length IL-10 promoter.¹¹  We found that E coli increased IL-10 promoter activity, and adenosine enhanced IL-10 promoter activity approximately 2-fold in E coli-induced but not control macrophages that were not exposed to E coli (Figure 5C). Because RAW 264.7 macrophages express low levels of the A_2A receptor endogenously,³⁷  we transfected these cells with an A_2A receptor-expressing construct (Figure 5C). This enforced expression of the A_2A receptor resulted in a more pronounced increase in IL-10 promoter activity both after E coli and the combination of E coli plus adenosine, underscoring the importance of A_2A receptors in augmenting IL-10 production.

To identify the DNA sequences that are necessary for adenosine to increase IL-10 promoter activity in E coli–challenged cells, a series of promoter mutants that contain successive deletions from the 5′ end were inserted upstream of the luciferase reporter gene¹¹  (Figure 5D). After transfection of RAW 264.7 cells with these constructs, luciferase activity was detected after adenosine/E coli treatment. Analysis of luciferase activity from the 5′ deletion mutants revealed that deletion of sequences between −1538 and −438 from the transcription start site did not affect the effect of adenosine in stimulating promoter activity. In contrast, the effect of adenosine was completely abolished by deletion of sequences between −438 and −376 (Figure 5D). These results show that DNA sequences in the IL-10 promoter between −438 and −376 are sufficient for the enhancing effect of adenosine on IL-10 promoter activity in E coli–treated macrophages. Using Searching Transcription Factor Binding Sites program we found that there were 2 potential binding sites for C/EBP (−410/−399 and −398/−385) in this promoter region.

To confirm that these 2 potential C/EBP binding sites are necessary for the stimulatory effect of adenosine on IL-10 promoter activity, we mutated these sites and this modified sequence was inserted upstream of the luciferase reporter gene into pGL2. Analysis of luciferase activity from this construct revealed that mutating the C/EBP consensus sites prevented the stimulatory effect of adenosine on E coli–induced promoter activity (Figure 5E).

Because the IL-10 promoter sequence between −438 and −376 contains binding sites for C/EBP transcription factors, we first determined the effect of E coli as well as E coli plus adenosine on C/EBP transcriptional activity. To this end, RAW 264.7 macrophages were transfected with a construct in which luciferase expression is driven by C/EBP. The transfected cells were treated with adenosine and E. coli, and luciferase activity was measured after 8 hours of treatment. We found that E coli stimulated C/EBP luciferase activity by approximately 2-fold and adenosine upregulated this activity by another approximately 2-and-one-half-fold (Figure 6A). Adenosine alone had no effect on C/EBP luciferase activity (Figure 6A). We then determined the effect of E coli and adenosine on C/EBP DNA binding. Using nuclear extracts from E coli–stimulated macrophages, we observed a significant increase in protein binding to a C/EBP consensus sequence at 30 and 60 minutes compared with untreated cells, and adenosine further enhanced this induction of C/EBP DNA binding (Figure 6B). In competition binding assay, a 50-fold molar excess of cold C/EBP consensus probe completely abrogated the binding of the radiolabeled C/EBP probe (Figure 6B). The C/EBP family contains several members and within this family C/EBPβ and C/EBPδ are both expressed in macrophages.⁴⁷  Therefore, we determined the nature of the different C/EBP isoforms binding to the C/EBP consensus oligo using antibodies specific for C/EBPβ or δ in supershift studies using nuclear extracts from E coli/adenosine–treated RAW 264.7 macrophages. Figure 6C shows that only the C/EBPβ antibody shifted the DNA–protein complex. To obtain additional evidence for the role of C/EBP, we performed immunodetection of C/EBPβ and C/EBPδ from nuclear and cytosolic fractions of macrophages after adenosine and/or E coli treatment using antibodies raised against C/EBPβ and C/EBPδ. This analysis revealed that C/EBPδ did not accumulate in the nuclear fraction after either E coli or adenosine treatment (data not shown). However, 3 C/EBPβ isoforms corresponding to the previously described^31,48,49  liver-enriched inhibitory protein, liver-enriched activating protein and full-length liver-enriched activating protein accumulated in nuclear fractions obtained from E coli-stimulated macrophages (Figure 6D,E). Furthermore, adenosine enhanced the nuclear accumulation of all 3 isoforms in the presence but not absence of E coli (Figure 6D,E). To provide further insight into the role of C/EBPβ in regulating IL-10 production, C/EBPβ WT and KO immortalized macrophage cell lines³¹  were stimulated with heat-killed E coli and adenosine for 5 hours, and IL-10 release was measured by enzyme-linked immunosorbent assay. We found that E coli or the combination of E coli and adenosine failed to induce IL-10 release by C/EBPβ KO cells, whereas these stimuli efficiently triggered IL-10 production by C/EBPβ WT macrophages (Figure 6F). These data identify C/EBPβ as the major transcription factor mediating the stimulatory effect of adenosine on IL-10 release in E coli–challenged cells.

Both TLR ligands and adenosine have been reported to be able to activate both p38 and p42/44 in macrophages.^38,50,51  Both p38 and p42/44 have been implicated in the regulation of IL-10 production in response to various TLR ligands.^52,53  Therefore, we tested the possibility that either p38 or p42/44 MAPK was involved in mediating the stimulatory effect of E coli or the combination of E coli and adenosine on IL-10 release. We first examined whether E coli or the combination of E coli and adenosine triggered the activation of p38 and p42/44 MAPKs. The results of these experiments showed that E coli increased p42/44 MAPK activation (Figure 7A) but not that of p38 (Figure 7B). When macrophages were treated with E coli and adenosine together, p38 activation was increased (Figure 7B), but the activation of p42/44 was not changed (Figure 7A), compared with E coli treatment alone. To examine whether this activation of p38 caused by adenosine contributed to the stimulatory effect of adenosine on E coli–induced IL-10 release, we investigated whether MAPK inhibition decreased the adenosine-stimulated IL-10 cytokine level. Treatment of peritoneal macrophages with the selective p38 MAPK pathway inhibitor SB203580 but not p42/44 MAPK inhibitor PD98059 completely abolished the IL-10 response to adenosine (Figure 7C).

The most striking finding of this work is that endogenous adenosine acting at A_2A receptors represents a crucial second signal in inducing the production of IL-10. In addition, our results unequivocally demonstrate that C/EBPβ is a central intracellular mediator on which signals from both E coli and adenosine receptors converge to induce IL-10 gene expression.

As we and others demonstrated, the most potent anti-inflammatory effects of adenosine are mediated by A_2A or A_2B receptor stimulation on monocytes/macrophages,^18,54  lymphocytes,^55,56  neutrophils,^57,58  dendritic cells,⁵⁹  and endothelial cells.^22,60  However, the majority of these studies on the immunomodulatory effects of adenosine were confined to investigating TLR4-agonist (LPS)–induced inflammation. Ramakers et al³⁹  have recently shown that the modulatory effects of adenosine receptor stimulation on cytokine production critically depend on the specific TLR ligand used to activate the monocyte inflammatory response. Our group has recently demonstrated that genetic inactivation of A_2A receptors almost abolished (by > 95%) cecal ligation and puncture-induced IL-10 production, which suggested that A_2A receptor activation by endogenous adenosine is required for IL-10 production after bacterial stimuli in vivo. By showing that E coli is unable to trigger IL-10 production by macrophages isolated from A_2A KO mice, we provide in vitro evidence for the first time that A_2A receptors have a crucial role in inducing IL-10 production in conjunction with a physiologic stimulus, such as E coli. In addition, we provide indirect proof that macrophages can serve as a source of endogenous adenosine at concentrations that are sufficient to engage A_2A receptors. Although macrophages have been documented to release adenosine even under resting conditions as well as in response to inflammatory stimuli, such as LPS,^61,62  the current study establishes that this endogenously released adenosine has immunomodulatory effects. In addition, the current results demonstrate that although A_2B receptors have a minor regulatory role in mediating the stimulatory effect of exogenous adenosine on IL-10 production, A_2A receptors are crucially required for the effect of both exogenous and endogenous adenosine.

Recent data have shown that IL-10 production by macrophages was induced via TLR-mediated MyD88-dependent or TRIF-dependent pathways, as well as via non-TLR signals.⁶³  For example, LPS induction of IL-10 via TLR4 was dependent on MyD88 and TRIF. Triggering through TLR9 by CpG showed that the induction of high levels of IL-10 was completely MyD88-dependent in macrophages. Unlike with LPS and CpG, our results clearly indicate that IL-10 production in E coli–stimulated or E coli/adenosine stimulated macrophages is completely MyD88-independent. Because of the complexity of E coli, it is unclear at this juncture what E coli components trigger IL-10 production by macrophages. It is interesting to note that a nucleotide-binding oligomerization domain 2 ligand had synergistic effect on the induction of IL-10 on costimulation with a TLR agonist,⁶⁴  indicating that nucleotide-binding oligomerization domain 2, a TLR-independent protein, may have also contributed to the effect of E coli. It is also worth noting that because S aureus also elicited IL-10 secretion (Figure 2C), ligands common to both Gram-negative and Gram-positive bacteria, such as peptidoglycan and muramyl dipeptide, may have had a role in triggering IL-10 production. However, because adenosine synergistically augmented IL-10 release induced by lipoteichoic acid prepared from S aureus, a true TLR2 agonist, it is also possible that the adenosine–S aureus interaction was mediated by the specific Gram-positive bacterial product lipoteichoic acid prepared from S aureus.

We have shown recently that adenosine through A_2B receptors enhanced LPS-induced IL-10 production, and this effect was posttranscriptional, because adenosine had no effect either on LPS-induced IL-10 mRNA accumulation or on IL-10 promoter activity.¹⁸  In contrast, our current results indicate that the mechanism of action of adenosine in upregulating IL-10 production by E coli–treated macrophages is transcriptional. Specifically, adenosine increased E coli–induced IL-10 mRNA accumulation and upregulated IL-10 promoter activity, and the effect of adenosine was blocked by actinomycin D. Our data have clearly demonstrated that C/EBPβ is critical for the stimulatory effect of adenosine on IL-10 production in E coli–treated murine macrophages. First, sequential deletion analysis of the IL-10 promoter as well as site-directed mutagenesis showed that a region containing C/EBPβ binding elements was responsible for the potentiating effect of adenosine in stimulating IL-10 promoter activity. Second, adenosine stimulated E coli–induced nuclear accumulation and DNA binding of C/EBPβ. Third, C/EBPβ-deficient macrophages failed to produce IL-10 in response to adenosine and E coli. Interestingly, C/EBPβ was also important for the stimulatory effect E coli even in the absence of exogenous adenosine, because C/EBPβ-deficient macrophages failed to respond to E coli alone, and E coli alone induced increases in the nuclear levels and DNA binding of C/EBPβ.

From studies performed with different cell types a number of different transcriptional, translational and posttranslational mechanisms have been proposed to regulate C/EBPβ activity. The most frequently reported mechanism is the increased induction of C/EBPβ gene transcription.^65,66  Regulation by nuclear translocation⁶⁷  and alternative translation initiation,⁶⁸  phosphorylation,⁶⁹  and acetylation^49,70  have also been described. Our data suggest that C/EBPβ activity is not regulated by nuclear translocation, because C/EBPβ was undetectable in cytoplasmic fractions both before and after adenosine and/or E coli treatment.

The fact that adenosine alone, unlike E coli, was not able to induce IL-10 production, mRNA accumulation, and C/EBPβ activation suggests that adenosine and E coli act by different mechanisms on the C/EBPβ transcription factor system. Previous studies demonstrated that C/EBPβ promoter activity is induced during monocyte activation, and binding of CREB to C/EBPβ promoter elements is critical for the activation of C/EBPβ transcription.⁷¹  Moreover, we have recently found that adenosine enhances CREB transcriptional activity in macrophages.³⁸  Thus, it is plausible that adenosine up-regulates E coli-induced C/EBPβ accumulation via CREB activation. This explanation is supported by our findings that adenosine stimulates both CREB activation³⁸  and IL-10 production (Figure 7C,D) in a p38-dependent manner. The fact that in contrast to the slow (4 hours) accumulation of C/EBPβ in response to LPS reported previously,⁷²  we observed a rapid (30 minutes) increase in C/EBPβ abundance after E coli exposure of macrophages suggests that the mechanism by which E coli activates C/EBPβ is fundamentally different from not only that of adenosine but also that of LPS. Because adenosine can stimulate the activation of CREB³⁸  but not C/EBPβ in the absence E coli, it is plausible that this CREB activation by exogenous or endogenous adenosine plays a permissive role in activating C/EBPβ in response to E. coli. Further studies are warranted to investigate the mechanisms by which E coli and adenosine activate the C/EBPβ system.

In summary, we propose that A_2A receptor activation provides a crucial permissive and synergistic signal for IL-10 production in response to infectious agents.

This work was supported by the National Institutes of Health (NIH) Grant R01 GM66189 and the Intramural Research Program of NIH, National Institute on Alcohol Abuse and Alcoholism, as well as Hungarian Research Fund OTKA (T 049537) and Hungarian National R&D Program 1A/036/2004.

National Institutes of Health

Contribution: B.C., E.A.D., E.S.V., P.G., L.V., and G.H. designed the research and analyzed data. S.J.L., M.R.B., and C.-X.S. contributed vital new reagents. B.C., Z.H.N., S.J.L., P.P., and G.H. performed research. B.C. and G.H. wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: György Haskó MD, PHD, Department of Surgery, UMDNJ-New Jersey Medical School, 185 South Orange Avenue, University Heights, Newark, NJ 07103; e-mail:haskoge@umdnj.edu.