Differential Expression of Adenosine A3 Receptors Controls Adenosine A2A Receptor-Mediated Inhibition of TLR Responses in Microglia

The Journal of Immunology

Ce´line van der Putten,* Ella A. Zuiderwijk-Sick,* Linda van Straalen,* Eveline D. de Geus,‡ Leonie A. Boven,‡ Ivanela Kondova,† Ad P. IJzerman,§ and Jeffrey J. Bajramovic1*

Microglia activation is a prominent feature in many neuroinﬂammatory disorders. Unrestrained activation can generate a chronic inﬂammatory environment that might lead to neurodegeneration and autoimmunity. Extracellular adenosine modulates cellular activation through adenosine receptor (ADORA)-mediated signaling. There are four ADORA subtypes that can either increase

(A2A and A2B receptors) or decrease (A1 and A3 receptors) intracellular cyclic AMP levels. The expression pattern of the subtypes thus orchestrates the cellular response to extracellular adenosine. We have investigated the expression of ADORA subtypes in

unstimulated and TLR-activated primary rhesus monkey microglia. Activation induced an up-regulation of A2A and a down-

regulation of A3 receptor (A3R) levels. The altered ADORA-expression pattern sensitized microglia to A2A receptor (A2AR)- mediated inhibition of subsequent TLR-induced cytokine responses. By using combinations of subtype-speciﬁc agonists and

antagonists, we revealed that in unstimulated microglia, A2AR-mediated inhibitory signaling was effectively counteracted by

A3R-mediated signaling. In activated microglia, the decrease in A3R-mediated signaling sensitized them to A2AR-mediated in-

hibitory signaling. We report a differential, activation state-speciﬁc expression of ADORA in microglia and uncover a role for A3R

as dynamically regulated suppressors of A2AR-mediated inhibition of TLR-induced responses. This would suggest exploration of

combinations of A2AR agonists and A3R antagonists to dampen microglial activation during chronic neuroinﬂammatory conditions. The Journal of Immunology, 2009, 182: 7603–7612.

icroglia are macrophage-like resident cells of the sine receptor (ADORA)2-mediated signaling. Documented effects brain. Under inflammatory conditions, microglia be- of adenosine include the inhibition of LPS-induced production of M come activated, as indicated by an altered morphol- proinflammatory cytokines such as TNF-␣ and IL-12 by macro- ogy, the production of inflammatory mediators, and by the in- phages and monocytes and the facilitated production of the anti- creased expression of molecules involved in Ag presentation (1, inflammatory cytokine IL-10 by LPS-treated macrophages (13– 2). Activated microglia are a prominent feature in a wide variety of 15). During periods of high metabolic activity such as ischemia neuroinflammatory disorders where they are attributed roles as and inflammation, extracellular levels of adenosine are strongly APCs as well as effector cells. increased, possibly forming an endogenous brake on inflammation TLR are part of the innate immune system and recognize patho- (16, 17). gen-associated molecular patterns (3, 4). To date, ten members of The cellular response to extracellular adenosine is orchestrated

the TLR family have been described for human and nonhuman by the expression pattern of four different ADORA subtypes: A1,

primates (5). Ligand binding to TLR leads to cellular activation by A2A,A2B, and A3, which are characterized by their capacity to ␬ activation of the transcription factor NF- B, which in turn induces either increase or decrease intracellular cAMP levels (16). A1 and ␣ the production of cytokines such as TNF- and IL-12 (4, 6–8). A3 receptors (A1R and A3R) are coupled to Gi protein signaling

Human microglia express mRNA for a broad variety of TLR, and and mediate biological effects opposite to A2A and A2B receptors

activation of rhesus monkey microglia via TLR1/2, -2/6, -3, -4, -5 (A2AR and A2BR), which are coupled to Gs protein signaling. Ro-

and -8 was recently described to trigger a rapid inﬂammatory re- dent microglia have been reported to express functional A1,A2A,

sponse (9–12). Such a response must be tightly regulated because and A3, but not A2B receptors (18), whereas studies detailing unrestrained TLR signaling can generate a chronic inflammatory ADORA expression profiles on human or nonhuman primate mi- environment that might lead to neurodegeneration. croglia are currently lacking. In addition, little is known about the Several studies have shown that extracellular adenosine modu- dynamics of ADORA expression after TLR-mediated activation lates TLR-mediated activation through G-protein coupled, adeno- and the consequences thereof during chronic inflammatory conditions.

*Alternatives Unit and †Animal Science Department, Biomedical Primate Research Centre, Rijswijk, The Netherlands; ‡Department of Immunology, Erasmus MC, Rot- terdam, The Netherlands; and §Division of Medicinal Chemistry, Leiden/Amsterdam Center for Drug Research, Leiden, The Netherlands 2 Abbreviations used in this paper: ADORA, adenosine receptor; A1R, A1 receptor; Received for publication October 9, 2008. Accepted for publication April 14, 2009. A3R, A3 receptor; A2AR, A2A receptor; A2BR, A2B receptor; CGS21680, 2-p-(2- carboxyethyl)phenethylamino-5Ј-N-ethylcarboxamidoadenosine hydrochloride; The costs of publication of this article were defrayed in part by the payment of page DPCPX, 8-cyclopentyl-1,3-dipropylxanthine; I␬B, inhibitor of NF-␬B proteins; charges. This article must therefore be hereby marked advertisement in accordance NECA, 5Ј-(N-ethylcarboxamido)adenosine; SCH58261, 7-(2-phenylethyl)-5-amino- with 18 U.S.C. Section 1734 solely to indicate this fact. 2-(2-furyl)-pyrazolo-[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine; VUF5574, N-(2-me- 1 Address correspondence and reprint requests to Dr. Jeffrey J. Bajramovic, Alterna- thoxyphenyl)-NЈ-[2-(3-pyridinyl)-4-quinazolinyl]-urea. tives Unit, Biomedical Primate Research Centre, Lange Kleiweg 139, 2280 GH Ri- jswijk, The Netherlands. E-mail address: [email protected] Copyright © 2009 by The American Association of Immunologists, Inc. 0022-1767/09/$2.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.0803383 7604 DYNAMIC A3R EXPRESSION CONTROLS A2AR-MEDIATED TLR INHIBITION

Table I. Primer and probe sequences

Target Forward Primer (5Ј–3Ј) Reverse Primer (5Ј–3Ј) Probe Amplicon Size (bp) Used in

A1R CTATGTTTGGCTGGAACAAT GCTGCTTGCGGATTAGGTAG 207 RT-PCR A2AR CTCATGCTGGGTGTCTATTT TGAAGCAGTTGATGATGTGT 193 RT-PCR A2BR ATTTCTTTGGGTGTGTTCTG ACTTGGGCTTATTTTTACCC 252 RT-PCR A3R CTGGAACATGAAACTGACCT GAGTTTGTTCCGAATGATGT 181 RT-PCR ␤-actin GGTCATCACCATTGGCAATGA ACGTCACACTTCATGATGGAGTTG 122 RT-PCR

A1R GTCAAGATCCCTCTCCGGTA CCACCACGAAGGAGAGGA GGCTGCTG 91 Real-time RT-PCR A2AR CCAACTACTTCGTGGTGTCG GTAATGGCAAAGGGGATGG GGCGGCGG 73 Real-time RT-PCR A2BR TCTGTGTCCCGCTCAGGT GATGCCAAAGGCAAGGAC TGCTGTCC 89 Real-time RT-PCR A3R GGGTCAAGCTTACCGTCAGA ATGACACCAGCCAGCAAAG GGCCCTGG 84 Real-time RT-PCR GAPDH TCCACTGGCGTCTTCAC GGCAGATGATGACCCTTTT AGCCCCAG 78 Real-time RT-PCR

Using primary rhesus monkey microglia, we characterized the Primary cell isolation and culture expression of ADORA subtypes in unstimulated and TLR-acti- Primary microglia were obtained from adult rhesus monkeys at necropsy as vated microglia. Microglia simultaneously up-regulated A2AR and described previously (12). Briefly, tissue samples from prefrontal subcor- ϳ down-regulated A3R expression levels upon TLR-mediated acti- tical white matter were collected, divided into cubes of 3 g, and meninges vation. As a consequence, ADORA-mediated inhibition of subse- and visible blood vessels were removed before mincing the tissue into 3 quent TLR-induced TNF-␣ and IL-12p40/p70 production was cubes of less than 2 mm . Tissue fragments were incubated at 37°C for 20 min in HBSS (Ϯ8 ml/g tissue) containing 0.25% w/v porcine trypsin (Sig- much more potent in activated than in unstimulated microglia. This ma-Aldrich), 0.2 mg/ml EDTA, 1 mg/ml glucose, and 0.1 mg/ml bovine could be attributed to an increase in A2AR-mediated signaling and pancreatic DNase I (Sigma-Aldrich). The supernatant (no centrifugation) involved suppression of NF-␬B activation. By using combinations was discarded and the pellet was resuspended in microglia medium, i.e., of subtype-specific agonists and antagonists, we revealed that in 1:1 v/v DMEM (high glucose)/HAMF10 (with L-glutamine) with 10% v/v FCS and antibiotic supplement (penicillin 100 U/ml and streptomycin 0.1 unstimulated microglia, A2AR-mediated inhibitory signaling was mg/ml) (all purchased from Life Technologies), washed once, and passed effectively counteracted by A3R-mediated signaling. In activated through a 100-␮m nylon cell strainer (BD Biosciences). Following cen- ϫ microglia, the decrease in A3R-mediated signaling shifted the bal- trifugation at 300 g for 7 min, the resuspended pellet was subjected to ance toward A R-mediated signaling, sensitizing the cells to in- Percoll gradient centrifugation and hypotonic shock to remove erythro- 2A ϫ 5 hibitory signaling. We report a differential, activation state-specific cytes. Cells were plated at a density of 2.2–2.5 10 /ml in tissue culture- treated 6- or 24-well plates (Corning Costar). After 24 h incubation at 37°C expression of ADORA in microglia and uncover a role for A3Ras in a humidified atmosphere containing 5% CO2, unattached cells and my- dynamically regulated suppressors of A2AR-mediated inhibition of elin debris were removed by washing and replaced by microglia medium TLR-induced responses. These findings contribute significantly to supplemented with Ն4 U recombinant human M-CSF/ml (PeproTech). our understanding of how inhibitors of cellular activation are reg- Half of the medium was replaced by fresh medium containing new growth factors every 3–4 days. ulated themselves, and might have relevance for therapeutic inter- vention in neuroinflammatory disorders where chronic microglia RT-PCR and quantitative RT-PCR activation often is an unwanted feature. Total cellular RNA was isolated using TriReagent (Sigma-Aldrich) according to manufacturer’s protocol. Subsequently, mRNA was reversely tran- Materials and Methods scribed into cDNA using the Omniscript Reverse Transcription System Animals according to manufacturer’s protocol (Qiagen) using 0.5 ␮g RNA as tem- plate and 0.25 ␮g oligo(dT) primers (Promega). RT-PCR were performed Adult brain donor rhesus monkeys (Macaca mulatta) without neurological 15 on the iCycler Thermal cycler (Bio-Rad). mRNA levels of ADORA sub- disease became available from the outbred breeding colony; no monkeys types and GAPDH as reference genes were determined by real-time quan- were killed for the exclusive purpose of primary cell culture initiation. titative RT-PCR. Reactions were performed using the ABI PRISM 7700 Individual details are listed in Supplementary Table I.3 Better use of ex- sequence detection system (Applied Biosystems) and the iQ5 multicolor perimental animals contributes to the active refinement program within the PCR detection system (Bio-Rad). Sequences of primer and probe (Probe- Biomedical Primate Research Centre (Rijswijk, The Netherlands). Library, Roche) combinations are listed in Table I. ADORA mRNA expression levels were standardized to GAPDH or ␤-actin mRNA expression Reagents levels using the Pfaffl method (19). The nonspecific adenosine receptor agonist 5Ј-(N-ethylcarboxamido)adenosine (NECA), adenosine A2AR agonist 2-p-(2-carboxyethyl)phenethyl- Cytokine analysis amino-5Ј-N-ethylcarboxamidoadenosine hydrochloride (CGS21680), ␣ adenosine A1R antagonist 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), TNF- and IL-12p40/p70 levels were determined by ELISA according to adenosine A2AR antagonist 7-(2-phenylethyl)-5-amino-2-(2-furyl)-pyra- manufacturer’s protocol (U-CyTech Biosciences). zolo-[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine (SCH58261), adenosine A2BR antagonist benzo[g]pteridine-2,4(1H,3H)-dione (alloxazine) and the aden- Ј Western blot analysis osine A3R antagonist N-(2-methoxyphenyl)-N -[2-(3-pyridinyl)-4-quinazolinyl]-urea (VUF5574; all purchased from Sigma-Aldrich), were diluted Protein extracts were prepared from untreated and R848-activated micro- Ϫ in DMSO, aliquotted and stored at 20°C. DMSO controls were included glia as well as from A2AR-transfected human embryonic kidney 293 cells. in the functional assays. The cells were lysed in lysis buffer (1% Triton X-100, 10 mM Tris-HCl, 50 ␮ TLR-ligands used were E. coli O26:B6 LPS (Sigma-Aldrich), mM NaCl, 30 mM Na4P2O7, 50 mM NaF, 5 M ZnCl2, 1 nM DTT, 100 ␮ Pam3CSK4, poly(I:C), Ultrapure LPS, Flagellin, FSL-1, R848, and CL075 MNa3VO4, and protease inhibitors; Roche). Lysates were heated at 85°C (Invivogen). During the course of this study, the distribution of R848 was for 10 min before centrifugation at 14000 rpm for 10 min. Supernatants discontinued and replaced by CL075, a TLR8 agonist with similar were separated on 12% Bis-Tris gels in combination with MOPS buffer properties. using the Novex Xcell Surelock miniCell system (Invitrogen) and subsequently transferred onto Hybond-ECL nitrocellulose membranes (Amer- sham Biosciences). The membranes were blocked for 30 min at room tem- 3 The online version of this article contains supplemental material. perature in TBST containing 5% nonfat dry milk, and probed using goat The Journal of Immunology 7605

FIGURE 1. TLR-mediated activation alters the ADORA expression proﬁle in primary rhesus microglia. A, Analysis of ADORA subtype mRNA expression by RT-PCR in unstimulated and TLR-activated microglia. B, ADORA mRNA expression levels were quantitated by real-time RT-PCR in unstimulated and in TLR-activated microglia. Cells were stimulated for 16 h with 1 ␮g R848/ml (open symbols) or with 100 ng LPS/ml (ﬁlled symbols). p Ͻ 0.05; Wilcoxon signed-rank test. ADORA mRNA expression levels are expressed relative to reference ,ء .Different symbols represent different donors

gene (GAPDH) mRNA expression levels. A1R mRNA expression levels were below detection limits and low levels of A2BR prohibited reliable quanti- ␮ ﬁcation of all samples. C, Western blot analysis of cell lysates of unstimulated microglia and of microglia stimulated for 16 h with 1 g R848/ml. A2AR expression levels were quantitated and normalized to GAPDH protein expression levels. Data are shown for three donors. D, FACS analysis of A3R expression levels on unstimulated microglia and microglia stimulated for 16 h with 1 ␮g R848/ml. The ﬁlled curves in the histogram (left) represent isotype

controls, the open curves A3R-staining. The solid lines represent unstimulated microglia and the dashed lines represent microglia that were stimulated for ␮ Ϫ 16 h with 1 g R848/ml. The MFI (mean ﬂuorescence intensity) indices given in the right panel were calculated as follows: (MFI A3R-staining MFI isotype control)/MFI isotype control.

ϩ anti-GAPDH (Imgenex) and goat anti-A2AR (Everest Biotech) Abs fol- performed on ice. Microglia were washed in FACS buffer (PBS 2% lowed by anti-goat IgG-HRP-conjugated secondary Abs (Jackson Immu- BSA) and incubated for 10 min with 1/10 diluted Fc receptor-block (Milte- noResearch Laboratories) for 1 h at room temperature. Blots were devel- nyi Biotec) in FACS buffer, to prevent aspeciﬁc binding of the Abs. Cells

oped using Femto ECL substrate (Pierce) and quantitated with the were washed and incubated for 30 min with anti-adenosine A3 receptor Abs ChemiDoc XRS system (Bio-Rad). (SP055P, Acris Antibodies) or total rabbit IgG (AbD Serotec) as an isotype control. Cells were washed and incubated for 30 min with PE-labeled Immunofluorescence CD11b Abs (clone D12; BD Biosciences) and allophycocyanin-labeled secondary anti-rabbit Abs (Jackson ImmunoResearch Laboratories). Cells Cells grown on glass coverslips were fixed for 20 min at 4°C with 2% were analyzed on the LSRII and data were analyzed using FACS Diva paraformaldehyde, washed with PBS, and permeabilized for 15 min in software (both purchased from BD Biosciences). PBS ϩ 0.1% Triton X-100. After two washes with PBS, cells were incubated for1hatroom temperature with normal rabbit serum (1 ␮g IgG/ml) and with rabbit polyclonal anti NF-␬B p65 Abs (0.4 ␮g IgG/ml; Santa Statistics Cruz) in PBS, washed with PBS ϩ 0.01% Tween20, and followed bya1h incubation at room temperature with FITC-labeled goat anti-rabbit second- GraphPad Prism 5 (GraphPad Software) for Macintosh was used for sta- ary Abs (Jackson ImmunoResearch Laboratories). After extensive washes, tistical analysis. Statistical analysis of the differential effect of activation on coverslips were mounted using Vectashield ϩ DAPI (Vector Laboratories, NECA-induced suppression was analyzed using linear mixed effects mod- Burlingame, CA) and images were captured on a fluorescence microscope. els (20) in the R environment (21). The following model was used: Log cytokine level ϭ b0 ϩ b1 ⅐ stimulus ϩ b2 ⅐ activation ϩ b3 ⅐ Flow cytometry stimulus ⅐ activation, where stimulus was coded 0 for LPS only and 1 for LPS ϩ NECA Microglia were harvested by incubation with 4 mg lidocaine/ml (Sigma- and activation was coded 0 for unstimulated and 1 for activated micro- Aldrich) for maximally 20 min at 37°C. All following incubations were glia. b0 represents the log value of the intercept, which is the mean of 7606 DYNAMIC A3R EXPRESSION CONTROLS A2AR-MEDIATED TLR INHIBITION

FIGURE 2. ADORA-mediated inhibition of LPS-induced TNF-␣ and IL-12p40/p70 is enhanced in activated microglia. Unstimulated and activated microglia were exposed to 100 ng LPS/ml in the presence or absence of 10 ␮M NECA. After 16 h, TNF-␣ (A) and IL-12p40/p70 (B) levels were measured in the culture su- ;p Ͻ 0.05 ,ء .pernatants using ELISA Student’s t test. TNF-␣ (C) and IL- 12p40/p70 (D) production after exposure to 100 ng LPS/ml ϩ 10 ␮M NECA is expressed relative to production after exposure to 100 ng LPS/ml alone for unstimulated and activated microglia of six different donors. TNF-␣ (E) and IL-12p40/p70 (F) production after exposure to 100 ng LPS/ml ϩ different concentrations of NECA is expressed relative to production after exposure to 100 ng LPS/ml alone for unstimulated and ;p Ͻ 0.05 ,ء .activated microglia 2-way ANOVA. The data are representative for three independent experiments using different donors.

the log cytokine production levels when stimulated with LPS only (the were below detection limits in both unstimulated and TLR-ac- antilog is the geometric mean). b1 represents the difference in log cy- tivated microglia from all donors, prohibiting quantitation. Fur- tokine levels when NECA is added. Thus, the antilog of b1 represents thermore, comparison of the threshold cycle values suggests the ratio of cytokine levels induced by LPS ϩ NECA/LPS; b2 represents the difference in intercept for unstimulated and activated micro- that A2AR- and A3R-encoding mRNA levels were generally glia, again the antilog represents a ratio; and b3 represents the interac- higher than those that encoded for A2BR. TLR-mediated acti- tion term for activation on NECA. The antilog of b3 is the ratio of vation of microglia resulted in an 11-fold (95% CI 7.34–15.28, ϩ NECA LPS-induced cytokine production in unstimulated and acti- Ͻ vated microglia. p 0.05) increase in A2AR-encoding mRNA expression levels. By contrast, A3R-encoding mRNA expression levels were Ͻ Results 3-fold decreased (95% CI 0.14–0.48, p 0.05; Fig. 1B). A2BR- TLR-mediated activation alters the ADORA expression profile in encoding mRNA levels were not significantly altered after rhesus monkey microglia TLR-mediated activation. The predicted relative contribution of stimulatory G protein signaling-coupled A R over G protein The expression profile of ADORA subtypes in human and non- 2A i signaling-coupled A R to ADORA-mediated signaling would human primate microglia had not been characterized before and 3 thereby be enhanced Ͼ30-fold in TLR-activated microglia. The it was unknown whether TLR-mediated activation would affect presented effects of TLR-mediated activation on A R and A R their expression levels. As published recently, signaling via 2A 3 TLR1/2, -2/4, -3, -5, or -8 potently activates rhesus monkey mRNA levels are in line with earlier observations in LPS-acti- microglia (12). We therefore exposed adult primary rhesus vated human and mouse macrophages (22), and preliminary re- monkey microglia from two different donors to TLR8 ligands sults indicate that they hold true for primary rhesus bone mar- and analyzed the expression of all ADORA subtypes. mRNA row-derived macrophages as well (data not shown). To examine whether the changes in mRNA expression levels transcripts encoding A2AR, A2BR, and A3R were readily de- were reflected by changes in protein levels, A receptor protein tected by RT-PCR, whereas transcripts for A1R were low or 2A below detection levels (Fig. 1A). After TLR8-mediated activa- expression levels were quantified by Western blotting. Results demonstrate that A R protein levels in TLR8-activated microglia tion, A2AR-encoding mRNA levels appeared up-regulated, 2A whereas A3R-encoding mRNA levels appeared down-regulated. were increased 2- to 4-fold (Fig. 1C), consistent with our data on To quantitate ADORA subtype mRNA expression levels, we mRNA expression levels. A3R protein expression levels were exposed microglia from different donors to TLR2/4 and TLR8 quantified by flow cytometry analysis. These data reveal that ligands and determined mRNA levels by quantitative RT-PCR. TLR8-mediated activation indeed resulted in decreased A3R pro-

Consistent with abovementioned results, A1R mRNA levels tein expression levels on primary microglia (Fig. 1D). The Journal of Immunology 7607

FIGURE 3. Kinetics of the enhanced inhibitory phenotype of activated microglia. Microglia were either unstimulated or activated by exposure to 1 ␮g R848/ml for 2, 4, 8, or 16 h. After extensive washing and a 4-h recovery period, microglia were exposed to 100 ng LPS/ml in the presence or absence of 10 ␮M NECA. After 16 h, TNF-␣ (A) and IL-12p40/p70 (B) levels were measured and expressed as relative values of exposure to 100 ng LPS/ml alone. Microglia were either unstimulated or activated by exposure to 1 ␮g R848/ml for 16 h. After extensive washing, microglia were allowed to recover for 4, 8, 24, 48, or 72 h before exposure to 100 ng LPS/ml in the presence or absence of 10 ␮M NECA. After 16 h, TNF-␣ (C) and IL-12p40/p70 (D) levels p Ͻ 0.05; Student’s t test. The data are representative for three ,ء .were measured and expressed as relative values of exposure to 100 ng LPS/ml alone independent experiments using different donors.

ADORA-mediated inhibition of LPS-induced TNF-␣ and stimulated microglia in all donors tested (0.448-fold of unstimu- IL-12p40/p70 production is enhanced in activated lated, 95% CI 0.157–1.279, p ϭ 0.123; Fig. 2D), but due to a microglia smaller effect of activation and higher within-donor variability in Adenosine-induced signaling through ADORA can modulate IL-12p40/p70 production levels this difference was not statistically LPS-induced production of TNF-␣ and IL-12 (13, 23, 24). To significant. assess the functional relevance of the altered ADORA expres- To better characterize the ADORA-mediated inhibitory ef- sion profile on TLR-activated microglia, we used the following fects, we measured LPS-induced cytokine production in the experimental set-up. Microglia were either unstimulated or ex- presence of different concentrations of NECA. Under nonsat- Յ ␮ posed for 16 h to TLR8-mediated activation, yielding activated urating conditions (NECA 1 M), inhibition of LPS-induced ␣ microglia. Both populations were washed, rested for 4 h, TNF- and IL-12p40/p70 was stronger in activated microglia washed again, and subsequently stimulated with LPS in the than in unstimulated microglia at all concentrations used (Fig. presence or absence of NECA, a potent nonselective ADORA 2, E and F). LPS-induced cytokine production in activated mi- ␮ agonist that is more stable than adenosine. After 16 h, TNF-␣ croglia was already inhibited by 0.02 M NECA, whereas this and IL-12p40/p70 levels in the culture supernatants were mea- concentration increased LPS-induced cytokine production in sured by ELISA. unstimulated microglia. Under saturating conditions (NECA Ն ␮ ␣ Both unstimulated and activated microglia produced compara- 5 M), inhibition of LPS-induced TNF- and IL-12p40/p70 ble amounts of TNF-␣ and IL-12p40/p70 upon stimulation with in unstimulated microglia did not attain the same level as in LPS (Fig. 2, A and B). The addition of NECA alone did not induce activated microglia (Fig. 2, E and F). Activated microglia were TNF-␣ or IL-12p40/p70 production in either unstimulated or ac- thus not only sensitized to ADORA-mediated inhibition of LPS- tivated microglia (data not shown). In unstimulated microglia, induced cytokine production, the potency of ADORA-mediated NECA inhibited the LPS-induced production of TNF-␣ and IL- inhibition was enhanced as well. 12p40/p70 to 86 and 78%, respectively of control levels. Remark- ␣ ably, in activated microglia, NECA inhibited the LPS-induced pro- Enhanced inhibition of LPS-induced TNF- production is duction of TNF-␣ and IL-12p40/p70 to 9 and 39%, respectively of rapidly gained and lost in activated microglia, whereas control levels (Fig. 2, A and B). Analysis of six additional donors inhibition of IL-12p40/p70 production remains enhanced for up confirmed that there were no significant differences in LPS-in- to 72 h duced production of TNF-␣ or IL-12p40/p70 between unstimu- Whereas it is conceivable that microglia receive multiple TLR lated or activated microglia ( p ϭ 0.39 and p ϭ 0.47, respectively). stimuli during inflammatory conditions in vivo, the timing of such As expected, the effects of NECA on LPS-induced TNF-␣ pro- stimuli will be variable. To investigate how fast after activation duction differed significantly between unstimulated and activated microglia became sensitized to ADORA-mediated inhibition, we microglia (0.125-fold of unstimulated, 95% CI 0.061–0.255, p Ͻ varied the duration of the activation period. Interestingly, there 0.001; Fig. 2C). In addition, inhibition of LPS-induced IL-12p40/ was a discordance between inhibition of LPS-induced TNF-␣ and p70 production by NECA was stronger in activated than in un- IL-12p40/p70 production. Where LPS-induced TNF-␣ production 7608 DYNAMIC A3R EXPRESSION CONTROLS A2AR-MEDIATED TLR INHIBITION

FIGURE 4. The enhanced inhibitory effect of NECA is mediated by the enhanced contribution of A2AR-mediated signaling and the decreased ␮ contribution of A3R-mediated signaling. Unstimulated and activated microglia were exposed to 100 ng LPS/ml in the presence or absence of 10 M ␣ NECA in combination with A1,A2A,A2B, and A3 receptor antagonists. After 16 h, TNF- (A) and IL-12p40/p70 (B) levels were measured and

expressed as relative values of exposure to 100 ng LPS/ml alone. DPCPX was used as an A1R, SCH58261 as an A2AR, alloxazine as an A2BR and ␮ VUF5574 as an A3R antagonist. All antagonists were used at 10 M concentrations. Solid lines in the graph indicate relative cytokine production levels of exposure to 100 ng LPS/ml ϩ 10 ␮M NECA without antagonists of unstimulated microglia, the dashed lines indicate activated microglia. p Ͻ 0.05; Student’s t test. TNF-␣ (C) and IL-12p40/p70 (D) production after exposure to 100 ng LPS/ml in the presence or absence of 10 ␮M ,ء ␮ CGS21680, an A2AR-selective agonist, and/or of 10 M VUF5574, an A3R antagonist. Cytokine production is expressed relative to production after p Ͻ 0.05; Student’s t test. TNF-␣ (E) and IL-12p40/p70 (F) production ,ء .exposure to 100 ng LPS/ml alone for unstimulated and activated microglia ϩ after exposure to 100 ng LPS/ml different concentrations of CGS21680, in the presence or absence of A3R antagonist, is expressed relative to production after exposure to 100 ng LPS/ml alone for unstimulated and activated microglia. All data are representative for three independent experiments using different donors.

was already stronger when inhibited by NECA in microglia that activation, whereas NECA-mediated inhibition of IL-12p40/ were activated for 2 h, enhanced inhibition of IL-12p40/p70 pro- p70 production remained enhanced for up to 72 h after activa- duction was observed only after activation for Ͼ8 h (Fig. 3A tion (Fig. 3, C and D). and B). Next we investigated how long activated microglia remained sensitized by varying the duration between activation and LPS Important roles for A2AR as well as for A3R exposure. Again there was a discordance between the inhibition Although the altered ADORA expression pattern on activated mi- of LPS-induced TNF-␣ and IL-12p40/p70 production. NECA- croglia was likely correlated to the enhanced inhibitory phenotype, mediated inhibition of TNF-␣ production was comparable in the contribution of individual ADORA subtypes remained unclear. unstimulated and activated microglia already within 8 h after Because selective agonists for all four ADORA subtypes are not The Journal of Immunology 7609

FIGURE 5. The enhanced inhibitory phenotype of activated microglia inhibits proinflammatory cytokine production induced via suppression of NF-␬B activation. A, Unstimulated and activated microglia were exposed to 100 ng LPS/ml in the presence or absence of 10 ␮M NECA for 1 h. Binding of normal rabbit serum and of Abs recognizing NF-␬B p65 was visualized by FITC-conjugated anti-rabbit Ab (left panels). Cell nuclei were visualized using 4Ј,6-diamidino-2-phenylindole (DAPI; right panels). Original magnifications, ϫ800 (upper panels) and ϫ400 (lower panels). B, Quantification of the intracellular localization of NF-␬B. Intracellular localization was scored as exclusively cytosolic, cytosolic and nuclear or exclusively nuclear. Total cell numbers were determined by counting DAPI-stained nuclei and data are represented as a percentage of total cell numbers. A minimum of 500 cells per variable were counted. available, we used the nonselective agonist NECA in combination unstimulated and activated microglia, in line with the idea that with subtype-selective antagonists to address this issue. A2AR-mediated signaling generates opposite effects from A3R-me- Addition of an A1R antagonist slightly enhanced the inhibitory diated signaling. potential of NECA in unstimulated microglia as measured by the To further investigate the contribution of A2AR and A3R, we per- effect on LPS-induced TNF-␣ and IL-12p40/p70 production (Fig. formed experiments using A2AR-specific agonists (CGS21680), A3R 4, A and B). Although analysis of A1R mRNA expression levels antagonists (VUF5574), and a combination of both. Addition of indicate that they were low at best (Fig. 1A), A1R expression levels CGS21680 alone mimicked our earlier observations using NECA, appear sufficient to affect LPS-induced cytokine production in un- confirming an important role for A2AR-mediated signaling in ac- stimulated microglia. In activated microglia however, NECA-me- tivated microglia (Fig. 4, C and D). Addition of A R antagonist diated inhibition of LPS-induced cytokine production was similar 3 alone inhibited LPS-induced cytokine responses in both nonacti- in the presence and absence of the antagonist, indicating that A R 1 vated and activated microglia. Because no exogenous adenosine are not responsible for the enhanced inhibition. Addition of an was added, the inhibitory effects are most likely attributable to A R antagonist did not affect the inhibitory effect of NECA on 2A endogenously produced extracellular adenosine in response to LPS LPS-induced TNF-␣ production in unstimulated microglia, but did (25–27) and demonstrate that A R-mediated signaling indeed annul NECA-mediated inhibition of IL-12p40/p70. In activated 3 microglia, NECA-mediated inhibition of both TNF-␣ and IL- counteracts adenosine-mediated inhibitory signaling. When A2AR 12p40/p70 production was completely abrogated, revealing an im- agonists were added in combination with A3R antagonists, inhibition was maximal in both nonactivated and activated microglia, portant role for A2AR. Addition of an A2BR antagonist did not affect the inhibitory effect of NECA on LPS-induced cytokine pro- confirming this idea. duction in either unstimulated or activated microglia, indicating Additional experiments using different concentrations of that the role of A2BR in the inhibition of LPS-induced cytokine CGS21680 either alone or in combination with A3R antagonists production is probably limited. Most interestingly, addition of an indicate that the contribution of A3R-mediated signaling is more

A3R antagonist enhanced the inhibitory effect of NECA in both important in unstimulated microglia than in activated microglia 7610 DYNAMIC A3R EXPRESSION CONTROLS A2AR-MEDIATED TLR INHIBITION

(Fig. 4, E and F). In addition, they indicate that, even in the ab- pression levels of A2AR are sufﬁcient to inhibit LPS-induced cy- sence of A3R-mediated signaling, activated microglia are more tokine levels to a considerable extent. However, it is only upon sensitive to A2AR-mediated inhibition and are more potently in- TLR-mediated activation that the full inhibitory capacity of A2AR hibited at saturating concentrations than unstimulated microglia. is unleashed and this is primarily due to the decreased contribution

Together, these results demonstrate that the potential to inhibit of A3R-mediated signaling in the response to adenosine. Because LPS-induced cytokine production via A2AR is already present in these effects were also found when saturating amounts of NECA unstimulated microglia to a considerable extent, but counteracted were used, they cannot be explained by mere competition of

by A3R-mediated signaling. Combined with our data on ADORA ADORAs for available NECA. Based on our data, we propose a expression levels, these data are most consistent with the idea that model in which, under homeostatic conditions, A2AR-mediated in-

the simultaneous up-regulation of A2AR and down-regulation of hibitory responses are kept under tight control by A3R-mediated, A3R sensitize activated microglia to ADORA-mediated inhibitory counter-regulatory mechanisms. When microglia have become ac- signaling. tivated and inhibition has become necessary, A2AR-mediated inhibitory responses are rapidly unleashed by removal of A3R. It is Cross-inhibition of proinﬂammatory cytokine production induced unclear yet whether A R-mediated signaling suppresses A R-me- ␬ 3 2A by different TLR involves suppression of NF- B activation diated inhibition of TLR-induced cytokine responses by directly

Next we investigated whether the altered ADORA proﬁle on ac- inhibiting A2AR-mediated signaling or rather by stimulating TLR- tivated microglia also resulted in a stronger inhibition of cytokine induced signaling (Fig. 6). A3R-mediated signaling has been re- production induced by other TLR ligands. We measured cytokine ported to induce p38 MAPK and ERK1/2 phosphorylation in pri- production induced by TLR1/2, -2/6, -2/4, -3, and -4-mediated mary murine microglia and in the N13 microglial cell line (33, 34),

activation and all these responses were stronger when inhibited by while other studies have reported on A3R-mediated suppression of NECA in activated microglia than in unstimulated microglia (Sup- LPS-induced TNF-␣ in murine BV2 microglial cells (24) and LPS- plementary Fig. 1). induced IL-12 in human monocytes (35). Biochemical studies us- Because activation of microglia enhanced the ADORA-medi- ing reporter cell lines transfected with different ADORAs might

ated inhibition of cytokine production as induced by a variety of reveal possible interactions between A2AR- and A3R-mediated sig- TLR ligands, the mechanism is likely to involve a shared element naling but are currently lacking. Although our data strongly sug- of TLR-mediated signaling. All TLR-induced intracellular signal- gest that the changes in ADORA expression patterns are respon- ing cascades ﬁnally lead to the activation of NF-␬B, a transcription sible for the enhanced inhibition of TLR-induced cytokine factor that controls among others. The mRNA transcription levels production in activated microglia, we cannot exclude the possibil- of TNF-␣ and IL-12p40/p70, followed by its translocation from ity that other effects, such as altered coupling of ADORAs to in- the cytosol to the nucleus. We therefore evaluated the effect of tracellular signaling cascades (36, 37), TNF-␣-induced prevention ␬ ADORA-mediated signaling on TLR-induced NF- B transloca- of A2AR desensitization (38), or the altered expression of down- tion in rhesus microglia by using immunoﬂuorescence (Fig. 5A). stream signaling elements (39), play a role as well.

Our results demonstrate that exposure of microglia to LPS is fol- Different groups have reported that A2AR-mediated signaling lowed by a rapid translocation of NF-␬B from the cytosol to the can suppress TLR-mediated activation of the NF-␬B pathway and

nucleus. Simultaneous triggering of ADORA-mediated signaling we demonstrate that the A2AR-mediated inhibitory effects on LPS- by the addition of NECA partially blocked translocation of NF-␬B, induced cytokine production in primate microglia involves as exempliﬁed by the enhanced proportion of cells where NF-␬B suppression of NF-␬B activation as well. Although we describe was localized in the cytosol as well as in the nucleus. Quantitative different effects of adenosine receptor-mediated signaling on LPS- analysis of the images (Fig. 5B) demonstrates that NECA inhibited induced TNF-␣ and IL-12p40/p70 production, both in kinetics and the LPS-induced nuclear translocation of NF-␬B in both unstimu- in strength of the inhibitory effect, this dissociation is not incon- lated as well as in activated microglia. Exposure of unstimulated sistent with our description of NF-␬B signaling as a likely target

and activated microglia to LPS resulted in 81 and 80%, respec- for A2AR-mediated signaling. It would rather indicate that, in ad- tively of total cell numbers with an exclusive nuclear localization dition to NF-␬B, there are other factors that differentially modulate of NF-␬B. Addition of NECA decreased these percentages to 54% the synthesis and secretion of TNF-␣ and IL-12 (e.g., AP-1). The in unstimulated microglia and to 23% in activated microglia, in molecular basis for the suppression of NF-␬B activation appears to

line with the enhanced inhibition in activated microglia. be cell type-specific. In rat glioma cells, A2AR-mediated signaling blocked the phosphorylation and subsequent degradation of inhib- Discussion itor of NF-␬B proteins (I␬B) in contrast to the I␬B-independent In this study we explored how ADORA-mediated signaling mod- suppression of the NF-␬B pathway in vascular endothelial cells ulates microglial innate immune responses during chronic inflam- (40). In addition, recent reports have shown that ADORA-medi- matory conditions. To do so, we characterized the ADORA ex- ated signaling can inactivate the ubiquitination of I␬B by promot- pression profile of primary adult rhesus monkey microglia before ing the deneddylation of cullin-1 in lung tissue from mice that and after TLR-mediated activation. We show that activation alters were exposed to hypoxia (41). Preliminary experiments using p38 the ADORA expression pattern and that ADORA-mediated inhi- MAPK inhibitors, MAPK/ERK kinase inhibitors, and protein ki- bition of LPS-induced TNF-␣ and IL-12p40/p70 is much stronger nase A inhibitors to assess the relative importance of these path- ␬ in activated than in unstimulated microglia. We demonstrate that ways for A2AR-mediated suppression of NF- B activation in rhe- both enhanced A2AR-mediated signaling as well as decreased sus microglia have thus far yielded inconclusive results (our

A3R-mediated signaling are responsible for this effect. Whereas unpublished observations), suggesting the involvement of redun- the inhibitory potential of A2AR-mediated signaling on LPS-in- dant, multiple, or other signaling cascades in microglia. duced cytokine production is widely acknowledged (13, 14, 16, 18, Whereas both A2AR- and A3R-mediated signaling contribute to

23, 28–32), the dynamically regulated counteracting role of A3R the effects of adenosine on LPS-induced cytokine responses, we is new. found little evidence for contributions of A1R- or A2BR-mediated Our data were obtained by using combinations of agonists and signaling. Our results on A1R expression levels are most consistent

antagonists and show that even on unstimulated microglia the ex- with the idea that the limited contribution of A1R-mediated effects The Journal of Immunology 7611

FIGURE 6. Schematic representation of our working hypothesis. A, In resting microglia, A3R-mediated signaling counteracts A2AR-mediated signaling either by stimulating TLR-mediated signaling or directly by inhibiting A2AR-mediated inhibition of TLR-mediated signaling. B, In activated microglia the contribution of A3R-mediated signaling is decreased and the contribution of A2AR-mediated signaling is increased, unleashing the inhibitory potential of extracellular adenosine on TLR-induced proinﬂammatory cytokine production. IKK, inhibitor of NF-␬B kinase; IRAK, IL1-receptor associated kinase; MyD88, myeloid differentiation primary response gene 88; RIP, receptor-interacting protein; TRAF, TNF-receptor associated factor; TRAM, TRIF-related adaptor molecule; TRIF, TIR-domain containing adaptor protein inducing IFN-␤.

is due to absence of expression or low expression levels on rhesus A2AR drives macrophages to switch from the production of proin- microglia, contrasting reports on A1R expression levels on rodent flammatory cytokines to the production of cytokines that are as- microglia (42–45). In addition to sequence differences (see Table sociated with repair and angiogenesis, such as vascular endothelial II and supplementary Fig. 2, A–D), such species-specific expres- growth factor (47, 48). This, in combination with the sustained sion patterns should be considered when rodents are used to model inhibition of the typical Th1-associated cytokine IL-12, might fur- human ADORA-mediated signaling. Although A2BR mRNA was ther contribute to the prevention of neurodegeneration and auto- expressed by rhesus microglia, we did not measure A2BR-medi- immunity during chronic inflammatory conditions. ated effects on the production of LPS-induced TNF-␣ and IL- In conclusion, our data define a novel role for A3R as dynam- 12p40/p70. It is, however, conceivable that A2BR-mediated sig- ically regulated suppressors of A2AR-mediated inhibition of TLR- naling may lead to other (anti-)inflammatory effects in microglia. induced proinflammatory cytokine responses. It is only during ␥ A2BR-mediated signaling has been described to inhibit IFN- -in- chronic neuroinflammatory conditions, such as encountered in duced cytokine production and MHC class II expression in bone multiple sclerosis, that the reciprocal regulation of A R and A R marrow-derived murine macrophages (46). 2A 3 unleashes the full inhibitory potential of adenosine. In situ studies In vivo, it is likely that microglia receive multiple TLR stimuli on ADORA subtype expression in humans and rhesus macaques during inflammatory conditions, with variable timing. Our data on suffering from acute and chronic CNS inflammatory disorders the kinetics of ADORA-mediated inhibition of LPS-induced cy- should reveal whether such a mechanism is likely to play a role in tokine production revealed that there is discordance between the vivo as well. One could speculate that the use of A R agonists sustained enhanced inhibition of LPS-induced IL-12p40/p70 re- 2A sponses and the rapid loss of the inhibition of TNF-␣ responses, during chronic inflammatory conditions might selectively inhibit suggesting different regulatory mechanisms. Interestingly, recent proinflammatory cytokine production of activated microglia, reports have demonstrated that synergistic signaling of TLR4 and whereas the addition of A3R antagonists might potentiate inhibitory effects of adenosine on unstimulated microglia as well. The recent description of a synthetic compound that demonstrates both

Table II. Rhesus monkey, mouse, and rat adenosine receptor protein adenosine A2AR agonist and A3R antagonist activity therefore homology with human adenosine receptors holds pharmacological promise (49). However, the widespread expression of adenosine receptors on different cells and tissues and A1R (%) A2AR (%) A2BR (%) A3R (%) the described neuroprotective effects of A2AR antagonists in ani- Macaca mulatta 99.4 98.3 98.5 97.5 mal models for stroke and Parkinson’s disease (50, 51) warrant Mus musculus 96.9 86.5 92.8 84.7 caution when extrapolating these results to therapeutic Rattus norvegicus 96.6 86.0 92.2 86.9 applications. 7612 DYNAMIC A3R EXPRESSION CONTROLS A2AR-MEDIATED TLR INHIBITION

Acknowledgments 26. Sperlagh, B., M. Baranyi, G. Hasko, and E. S. Vizi. 2004. Potent effect of interleukin-1␤ to evoke ATP and adenosine release from rat hippocampal slices. We thank T. Haaksma and T. Mulder for expert technical assistance. We J. Neuroimmunol. 151: 33–39. thank Dr. E. Remarque for expert assistance with the statistical analyses 27. Fredholm, B. B. 2007. Adenosine, an endogenous distress signal, modulates tis- and Drs. R.E. Bontrop, S.B. Geutskens, L.A. ‘t Hart, and J.M. van Noort sue damage and repair. Cell Death Differ. 14: 1315–1323. for critically reading the manuscript. 28. Boucsein, C., R. Zacharias, K. Farber, S. Pavlovic, U. K. Hanisch, and H. Kettenmann. 2003. Purinergic receptors on microglial cells: functional expression in acute brain slices and modulation of microglial activation in vitro. Disclosures Eur. J. Neurosci. 17: 2267–2276. The authors have no financial conflict of interest. 29. Cunha, R. A. 2005. Neuroprotection by adenosine in the brain: from A(1) receptor activation to A (2A) receptor blockade. Purinergic Signal. 1: 111–134. References 30. Dare, E., G. Schulte, O. Karovic, C. Hammarberg, and B. B. Fredholm. 2007. Modulation of glial cell functions by adenosine receptors. Physiol. Behav. 92: 1. Aloisi, F., F. Ria, G. Penna, and L. Adorini. 1998. Microglia are more efficient 15–20. than astrocytes in antigen processing and in Th1 but not Th2 cell activation. 31. Jijon, H. B., J. Walker, F. Hoentjen, H. Diaz, J. Ewaschuk, C. Jobin, and J. Immunol. 160: 4671–4680. K. L. Madsen. 2005. Adenosine is a negative regulator of NF-␬B and MAPK 2. Santambrogio, L., S. L. Belyanskaya, F. R. Fischer, B. Cipriani, C. F. Brosnan, signaling in human intestinal epithelial cells. Cell. Immunol. 237: 86–95. P. Ricciardi-Castagnoli, L. J. Stern, J. L. Strominger, and R. Riese. 2001. De- 32. Palmer, T. M., and M. A. Trevethick. 2008. Suppression of inflammatory and velopmental plasticity of CNS microglia. Proc. Natl. Acad. Sci. USA 98: immune responses by the A(2A) adenosine receptor: an introduction. 6295–6300. Br. J. Pharmacol. 153(Suppl 1): S27–S34. 3. Janeway, C. A., Jr., and R. Medzhitov. 2002. Innate immune recognition. Annu. 33. Hammarberg, C., B. B. Fredholm, and G. Schulte. 2004. Adenosine A3 receptor- Rev. Immunol. 20: 197–216. mediated regulation of p38 and extracellular-regulated kinase ERK1/2 via phos- 4. Kawai, T., and S. Akira. 2005. Pathogen recognition with Toll-like receptors. phatidylinositol-3Ј-kinase. Biochem. Pharmacol. 67: 129–134. Curr. Opin. Immunol. 17: 338–344. 34. Hammarberg, C., G. Schulte, and B. B. Fredholm. 2003. Evidence for functional 5. Iwasaki, A., and R. Medzhitov. 2004. Toll-like receptor control of the adaptive adenosine A3 receptors in microglia cells. J. Neurochem. 86: 1051–1054. immune responses. Nat. Immunol. 5: 987–995. 35. la Sala, A., M. Gadina, and B. L. Kelsall. 2005. G(i)-protein-dependent inhibition 6. Akira, S. 2006. TLR signaling. Curr. Top. Microbiol. Immunol. 311: 1–16. of IL-12 production is mediated by activation of the phosphatidylinositol 3-ki- 7. O’Neill, L. A. 2006. How Toll-like receptors signal: what we know and what we nase-protein 3 kinase B/Akt pathway and JNK. J. Immunol. 175: 2994–2999. don’t know. Curr. Opin. Immunol. 18: 3–9. 36. Gsandtner, I., C. Charalambous, E. Stefan, E. Ogris, M. Freissmuth, and 8. Takeda, K., and S. Akira. 2005. Toll-like receptors in innate immunity. Int. Im- J. Zezula. 2005. Heterotrimeric G protein-independent signaling of a G protein- munol. 17: 1–14. coupled receptor: direct binding of ARNO/cytohesin-2 to the carboxyl terminus 9. Bsibsi, M., R. Ravid, D. Gveric, and J. M. van Noort. 2002. Broad expression of of the A2A adenosine receptor is necessary for sustained activation of the ERK/ Toll-like receptors in the human central nervous system. J. Neuropathol. Exp. MAP kinase pathway. J. Biol. Chem. 280: 31898–31905. Neurol. 61: 1013–1021. 37. Gsandtner, I., and M. Freissmuth. 2006. A tail of two signals: the C terminus of 10. Jack, C. S., N. Arbour, J. Manusow, V. Montgrain, M. Blain, E. McCrea, the A(2A)-adenosine receptor recruits alternative signaling pathways. Mol. Phar- A. Shapiro, and J. P. Antel. 2005. TLR signaling tailors innate immune responses macol. 70: 447–449. in human microglia and astrocytes. J. Immunol. 175: 4320–4330. 38. Khoa, N. D., M. Postow, J. Danielsson, and B. N. Cronstein. 2006. Tumor ne- 11. Olson, J. K., and S. D. Miller. 2004. Microglia initiate central nervous system ␣ ␣ innate and adaptive immune responses through multiple TLRs. J. Immunol. 173: crosis factor- prevents desensitization of G s-coupled receptors by regulating 3916–3924. GRK2 association with the plasma membrane. Mol. Pharmacol. 69: 1311–1319. 12. Zuiderwijk-Sick, E. A., C. van der Putten, M. Bsibsi, I. P. Deuzing, W. de Boer, 39. Nguyen, D. K., M. C. Montesinos, A. J. Williams, M. Kelly, and B. N. Cronstein. C. Persoon-Deen, I. Kondova, L. A. Boven, J. M. van Noort, B. A. t Hart, 2003. Th1 cytokines regulate adenosine receptors and their downstream signaling S. Amor, and J. J. Bajramovic. 2007. Differentiation of primary adult microglia elements in human microvascular endothelial cells. J. Immunol. 171: 3991–3998. alters their response to TLR8-mediated activation but not their capacity as APC. 40. Sands, W. A., A. F. Martin, E. W. Strong, and T. M. Palmer. 2004. Specific ␬ Glia 55: 1589–1600. inhibition of nuclear factor- B-dependent inflammatory responses by cell type- 13. Hasko, G., P. Pacher, E. A. Deitch, and E. S. Vizi. 2007. Shaping of monocyte specific mechanisms upon A2A adenosine receptor gene transfer. Mol. Pharma- and macrophage function by adenosine receptors. Pharmacol. Ther. 113: col. 66: 1147–1159. 264–275. 41. Khoury, J., J. C. Ibla, A. S. Neish, and S. P. Colgan. 2007. Antiinflammatory 14. Link, A. A., T. Kino, J. A. Worth, J. L. McGuire, M. L. Crane, G. P. Chrousos, adaptation to hypoxia through adenosine-mediated cullin-1 deneddylation. R. L. Wilder, and I. J. Elenkov. 2000. Ligand-activation of the adenosine A2a J. Clin. Invest. 117: 703–711. receptors inhibits IL-12 production by human monocytes. J. Immunol. 164: 42. Wittendorp, M. C., H. W. Boddeke, and K. Biber. 2004. Adenosine A3 receptor- 436–442. induced CCL2 synthesis in cultured mouse astrocytes. Glia 46: 410–418. 15. Nemeth, Z. H., C. S. Lutz, B. Csoka, E. A. Deitch, S. J. Leibovich, W. C. Gause, 43. Wittendorp, M. C., J. von Frijtag Drabbe Kunzel, A. P. Ijzerman, H. W. Boddeke, M. Tone, P. Pacher, E. S. Vizi, and G. Hasko. 2005. Adenosine augments IL-10 and K. Biber. 2004. The mouse brain adenosine A1 receptor: functional expres- production by macrophages through an A2B receptor-mediated posttranscrip- sion and pharmacology. Eur. J. Pharmacol. 487: 73–79. tional mechanism. J. Immunol. 175: 8260–8270. 44. Fiebich, B. L., K. Biber, K. Gyufko, M. Berger, J. Bauer, and D. van Calker. 16. Fredholm, B. B., A. P. IJzerman, K. A. Jacobson, K. N. Klotz, and J. Linden. 1996. Adenosine A2b receptors mediate an increase in interleukin (IL)-6 mRNA 2001. International Union of Pharmacology. XXV. Nomenclature and classifi- and IL-6 protein synthesis in human astroglioma cells. J. Neurochem. 66: cation of adenosine receptors. Pharmacol. Rev. 53: 527–552. 1426–1431. 17. Jacobson, K. A., and Z. G. Gao. 2006. Adenosine receptors as therapeutic targets. 45. Fiebich, B. L., K. Biber, K. Lieb, D. van Calker, M. Berger, J. Bauer, and Nat. Rev. Drug Discov. 5: 247–264. P. J. Gebicke-Haerter. 1996. Cyclooxygenase-2 expression in rat microglia is 18. Hasko, G., P. Pacher, E. S. Vizi, and P. Illes. 2005. Adenosine receptor signaling induced by adenosine A2a-receptors. Glia 18: 152–160. in the brain immune system. Trends Pharmacol. Sci. 26: 511–516. 46. Xaus, J., M. Mirabet, J. Lloberas, C. Soler, C. Lluis, R. Franco, and A. Celada. 19. Pfaffl, M. W. 2001. A new mathematical model for relative quantification in 1999. IFN-␥ up-regulates the A2B adenosine receptor expression in macro- real-time RT-PCR. Nucleic Acids Res. 29: e45. phages: a mechanism of macrophage deactivation. J. Immunol. 162: 3607–3614. 20. Pinheiro, J. C., and D. M. Bates. 2000. Mixed Effects Models in S and S-plus. 47. Macedo, L., G. Pinhal-Enfield, V. Alshits, G. Elson, B. N. Cronstein, and Springer-Verlag, New York. S. J. Leibovich. 2007. Wound healing is impaired in MyD88-deficient mice: a 21. R Development Core Team. 2007. R: a language and environment for statistical role for MyD88 in the regulation of wound healing by adenosine A2A receptors. computing. R Foundation for Statistical Computing, Vienna, Austria. Am. J. Pathol. 171: 1774–1788. 22. Murphree, L. J., G. W. Sullivan, M. A. Marshall, and J. Linden. 2005. Lipopoly- 48. Pinhal-Enfield, G., M. Ramanathan, G. Hasko, S. N. Vogel, A. L. Salzman, saccharide rapidly modifies adenosine receptor transcripts in murine and human G. J. Boons, and S. J. Leibovich. 2003. An angiogenic switch in macrophages macrophages: role of NF-␬B in A(2A) adenosine receptor induction. Biochem. J. involving synergy between Toll-like receptors 2, 4, 7, and 9 and adenosine A(2A) 391: 575–580. receptors. Am. J. Pathol. 163: 711–721. 23. Hasko, G., D. G. Kuhel, J. F. Chen, M. A. Schwarzschild, E. A. Deitch, 49. Bevan, N., P. R. Butchers, R. Cousins, J. Coates, E. V. Edgar, V. Morrison, J. G. Mabley, A. Marton, and C. Szabo. 2000. Adenosine inhibits IL-12 and M. J. Sheehan, J. Reeves, and D. J. Wilson. 2007. Pharmacological characteri- TNF-␣ production via adenosine A2a receptor-dependent and independent mech- sation and inhibitory effects of (2R,3R,4S,5R)-2-(6-amino-2-{[(1S)-2-hydroxy- anisms. FASEB J. 14: 2065–2074. 1-(phenylmethyl)ethyl]amino}-9 H-purin-9-yl)-5-(2-ethyl-2H-tetrazol-5-yl)tetra- 24. Lee, J. Y., B. S. Jhun, Y. T. Oh, J. H. Lee, W. Choe, H. H. Baik, J. Ha, hydro-3,4-furandiol, a novel ligand that demonstrates both adenosine A(2A) K. S. Yoon, S. S. Kim, and I. Kang. 2006. Activation of adenosine A3 receptor receptor agonist and adenosine A(3) receptor antagonist activity. Eur. J. Phar- suppresses lipopolysaccharide-induced TNF-␣ production through inhibition of macol. 564: 219–225. PI 3-kinase/Akt and NF-␬B activation in murine BV2 microglial cells. Neurosci. 50. Chen, J. F., and F. Pedata. 2008. Modulation of ischemic brain injury and neu- Lett. 396: 1–6. roinflammation by adenosine A2A receptors. Curr. Pharm. Des. 14: 1490–1499. 25. Eigler, A., T. F. Greten, B. Sinha, C. Haslberger, G. W. Sullivan, and S. Endres. 51. Kalda, A., L. Yu, E. Oztas, and J. F. Chen. 2006. Novel neuroprotection by 1997. Endogenous adenosine curtails lipopolysaccharide-stimulated tumour ne- caffeine and adenosine A(2A) receptor antagonists in animal models of Parkin- crosis factor synthesis. Scand. J. Immunol. 45: 132–139. son’s disease. J. Neurol. Sci. 248: 9–15.