INTRODUCTION

Adenosine, acting outside the cell, exerts potent actions on awide variety of physiological systems, including the nervous,cardiovascular, gastrointestinal, urogenital, respiratory, andlymphatic systems. Many of these actions are mediated viaspecific receptors named P1 purinoceptors (Daly et al., 1983).From studies on the variation of adenylate cyclase activityproduced by adenosine analogues, and on the rank order ofpotency of agonists, P1 purinoceptors were subdivided intotwo classes: A1, which mediate decreases in cAMP levels,and A2, which mediate increases in cAMP levels (VanCalcker et al., 1979; Londos et al., 1980). Thepharmacological characterization of the A2 subtype indifferent tissues led to the finding of two different populationswhich were named A2A and A2B, and had very differentaffinities for the agonist CGS21680. It is now becoming clearthat the subtypes of adenosine receptors can be linked to avariety of signal transduction systems apart from adenylatecyclase (Stiles, 1992; Linden, 1994; Schubert et al., 1994;Fredholm, 1995). These receptor subtypes (A1, A2A and A2B)as well as another member of the family, which wasdiscovered more recently (A3), have been cloned (Zhou et al.,

1992; Linden et al., 1993; Salvatore et al., 1993; Tucker andLinden, 1993). All these P1 purinoceptors belong to thesuperfamily of G protein-coupled receptors having seventransmembrane domains.

The relevance of adenosine for the development andfunction of the immune system is deduced from the severecombined immunodeficiency syndrome (SCID) associatedwith the congenital defect of adenosine deaminase, theenzyme which degrades the nucleoside. In adenosinedeaminase-deficient children, elevated levels of adenosine(and 2′-deoxyadenosine) are found in body fluids. Severalmechanisms whereby adenosine deaminase deficiency mayaffect the development and function of lymphoid cells havebeen suggested but none can satisfactorily explain all therelationships found in SCID. Most of the proposedmechanisms (see Hershfield and Mitchell, 1989, for review)have rested upon the assumption that accumulation of the twophysiological substrates of adenosine deaminase, adenosineand 2′-deoxyadenosine, are toxic for lymphocytes (Franco etal., 1998, and references therein). There is recent datasuggesting that signaling through purinergic receptors inlymphocyte and lymphocyte precursors may contribute to thepathogenesis of adenosine deaminase-related SCID (Apasov

491Journal of Cell Science 112, 491-502 (1999)Printed in Great Britain © The Company of Biologists Limited 1999JCS0069

Extracellular adenosine has a key role in the developmentand function of the cells of the immune system. Many ofthe adenosine actions seem to be mediated by specificsurface receptors positively coupled to adenylate cyclase:A2A and A2B. Despite the fact that A2A receptors (A2ARs)can be easily studied due to the availability of the specificagonist CGS21680, a pharmacological and physiologicalcharacterization of adenosine A2B receptors (A2BRs) inlymphocytes has not been possible due to the lack ofsuitable reagents. Here we report the generation andcharacterization of a polyclonal antipeptide antibodyraised against the third extracellular loop of the A2BRhuman clone which is useful for immunocytochemicalstudies. This antibody has permitted the detection of A2BR+

cells in lymphocyte samples isolated from humanperipheral blood. The pharmacology of cAMP-producingcompounds is consistent with the presence of functional

A2BRs but not of A2A receptors in these human cells. Thepercentage of A2BR-expressing cells was similar in theCD4+ or CD8+ T cell subpopulations. Interestinglyactivation signals delivered by either phytohemagglutininor anti-T cell receptor/CD3 complex antibodies led to asignificant increase in both the percentage of cellsexpressing the receptor and the intensity of the labeling.These receptors are functional since interleukin-2production in these cells is reduced by NECA but not by R-PIA or CGS21680. These results show that A2BRexpression is regulated in T cell activation and suggest thatthe role of adenosine in lymphocyte deactivation ismediated by A2BRs.

Key words: Adenosine receptor, T cell activation, Antipeptideantibody

SUMMARY

Expression of A 2B adenosine receptors in human lymphocytes: their role in T

cell activation

Maribel Mirabet 1, Carolina Herrera 1, Oscar J. Cordero 2, Josefa Mallol 1, Carmen Lluis 1 and Rafael Franco 1,*1Department of Biochemistry and Molecular Biology, Faculty of Chemistry, University of Barcelona, Barcelona, Catalonia, Spain2Department of Biochemistry and Molecular Biology, Faculty of Biology, University of Santiago de Compostela, Spain*Author for correspondence (e-mail: r.franco@sun.bq.ub.es; homepage: www.bq.ub.es/recep/franco.html)

Accepted 9 December 1998; published on WWW 25 January 1999

492

et al., 1997; Resta and Thompson, 1997). On the other hand,adenosine receptors seem to be involved in the regulation byadenosine of T-cell receptor-triggered activation-relatedevents, such as antibody production, cell proliferation, IL-2production, upregulation of IL-2 receptor α-chain (CD25)and lymphocyte-mediated cytolysis (Wolberg et al., 1975;Dos Reis et al., 1986; Antonysamy et al., 1995; Huang et al.,1997). Adenosine receptors regulating these different aspectsof lymphoid function are those leading to increases in cAMPlevels, i.e. A2AR and/or A2BR (Dos Reis et al., 1986). In fact,adenosine and 2-chloroadenosine lead to increases in humanlymphocyte cAMP levels (Marone et al., 1978; Dinjens et al.,1986). Recently, Huang et al. (1997) have shown that A2ARsare involved in adenosine-mediated inhibition of murine T-cell activation and expansion. In these events the role ofA2BRs has not been established due to the lack of suitabletools for their study. In this paper we describe the generationand characterization of two polyclonal antipeptide antibodiesraised against different regions of the A2BR human clone.One of these antibodies has proven to be inestimable for theidentification of A2BR-expressing human blood lymphocytes.Further analysis of these cells have shown that the expressionof A2BR is regulated in T cell activation events thus indicatingthat A2BRs are involved in the regulation by adenosine ofsignaling processes occurring in human lymphocytes.Furthermore, upregulated A2BRs are functional and elicitsignificant reductions in interleukin-2 production. Theseresults are similar to those found in murine macrophages inwhich A2BRs are upregulated in response to interferon-γ. Aswe described here for T lymphocytes, adenosine, actingthrough A2BRs, contributes to the deactivation ofmacrophages since it reduces the upregulation of MHCclass II molecules, the activity of nitric oxid synthase orthe production of pro-inflammatory cytokines (Xaus et al.,1999).

MATERIALS AND METHODS

Materials5′-N-ethylcarboxamidoadenosine (NECA), N6-(R)-phenylisopropyl-adenosine (R-PIA), fluorescein isothiocyanate (FITC),tetramethylrhodamine isothiocyanate (TRITC), saponin,paraformaldehyde, Indo-1/AM, Triton X-100, EGTA and bovineserum albumin (BSA) were purchased from Sigma Chemical Co. (StLouis, MO, USA). Ficoll was from Biochrom KG (Berlin, Germany)and 2-(p-[2-carboxyethyl]phenylethylamino)-5′-N-ethylcarboxamido-adenosine (CGS21680) from Research Biochemicals International(Natick, MA, USA). Keyhole limpet hemocyanin was purchased fromCalbiochem (La Jolla, CA, USA), glycine and IL-2 were fromBoehringer Mannheim (Barcelona, Spain) and Immuno-FluoreMounting Medium was from ICN Biomedical Inc. (Costa Mesa, CA,USA). Thiol-Sepharose 4B and Sephadex G-25 fine grade wereobtained from Pharmacia (Uppsala, Sweden) and OKT3 antibody waskindly provided by Dr Terhorst (Beth Israel-Deaconess MedicalCenter, Boston, MA, USA). The anti-M2 flag antibody was purchasedfrom Eastman Kodak company (New Haven, USA). Phycoerythrin(PE)-anti CD4 (T4), PE-anti CD8 (T8) and PE-anti CD3 (CD3, cloneHIT3A) mAbs were from Coulter (Hialeah, FL, USA). All otherproducts were of the best grade available and were purchased fromMerck (Darmstad, Germany) and Sigma. Deionized water furtherpurified with a Millipore Milli-Q system (Bedford, MA, USA) wasused throughout.

Cells and culture conditionsChinese hamster ovary (CHO) cells transfected with cDNA coding forthe human A2B adenosine receptor (Pierce et al., 1992) were kindlyprovided by Dr Peter R. Schofield from the Garvan Institute ofMedical Research at St Vincent’s Hospital in Sidney, Australia.Chinese hamster ovary (CHO) cells transfected with a cDNA codingfor a recombinant protein composed of the human A2A adenosinereceptor containing the M2 flag in its C-terminal end (Rivkees et al.,1995) were kindly provided by Dr Scott A. Rivkees from YaleUniversity, New Haven, CT, USA. Transfected CHO cells were grownin Dulbecco’s modified Eagle’s medium/Hams F12 nutrient mixture(1:1) containing 10% (v:v) fetal calf serum, 2 mM L-glutamine,antibiotics (100 i.u./ml penicillin, 100 µg/ml streptomycin and 0.25µg/ml fungizone and 1.6 mg/ml of the neomycin analog G-418), at37°C in an humid atmosphere of 5% CO2. Wild-type CHO cells werecultured in the same conditions described for transfected cells but inthe absence of G-418. Jurkat cells were maintained in RPMI 1640medium supplemented with 10% (v:v) fetal bovine serum, 2 mM L-glutamine and antibiotics at 37°C under a 5% CO2 atmosphere. HEP-G2 cells were cultured in DMEM supplemented with 4.5 g/l glucoseand 10% fetal serum. All the culture reagents were from Gibco (GrandIsland, NY, USA).

Peripheral blood mononuclear cells (PBMC) from healthy donorswere isolated from buffy coats (kindly provided by Dr Hernándezfrom the Banco de Sangre of the Hospital General Vall d’Hebrón deBarcelona) using the Ficoll gradient method described by Boyum(1968). Further purification of lymphocytes (PBL) from PBMC wasperformed by depletion of contaminating cells by adherence to plasticplates. An homogeneous population corresponding to totallymphocytes was observed according to forward and side light scatterparameters by flow cytometry analysis with less than 7% ofcontaminating monocytes. PBMC or PBL were split at 106 cells/mland cultured in the same conditions as described above for Jurkatcells. In vitro activation of cells was carried out with 1 µg/mlphytohemagglutinin (PHA; Difco Laboratories, Detroit, MI, USA) orwas conducted in well plates coated with OKT3 antibody. For OKT3antibody immobilization, 300 µl of 10 mM PBS, pH 7.4, containing2.5 µg/ml purified OKT3 antibody were placed for 3 hours at 37°Cin a 24-well flat bottom plate. After 2 washes with 2 ml of cold PBS,cells were added to wells at a density of 106 cells/ml to begin theactivation experiments.

Anti-A 2B adenosine receptor antibody generation andpurificationThe procedure for obtaining the antibodies is patent pending.

Antipeptide antisera were generated in New Zealand white rabbitsby Biokit, S.A. (Barcelona). The peptides used for immunizationcorrespond to the deduced amino acid sequence from the putativethird extracellular loop (FQPAQGKNKPKWA) and from a portion ofthe C-terminal domain (AYRNRDFRYTFHKI) of the cloned humanA2B adenosine receptor (Pierce et al., 1992) (Fig. 1). They wereselected on the basis of hydropathy analysis (Kyte and Doolitte, 1982)and, whenever possible, low homology with the A1, A2A and A3adenosine receptors. The peptides were synthesized by the PeptideSynthesis Service at the University of Barcelona with an extracarboxy-terminal cysteine residue, to facilitate conjugation withkeyhole limpet hemocyanin via m-maleimidobenzoyl-N-hydroxysuccinimide coupling (Kitagawa and Aikawa, 1976). Thecomposition of each peptide was verified by high performance liquidchromatography. Each keyhole limpet hemocyanin-coupled peptidewas injected into two rabbits using the immunization protocol methodof Tanaka et al. (1985). The reactivity of the resulting antisera, MEin the case of the extracellular peptide and MI in the case of theintracellular peptide (Fig. 1), against the appropriate peptide wastested by enzyme-linked immunosorbent assay (ELISA). For thispurpose, 96-well Maxisorp microtiter plates (Nunc) were coated withthe peptides (2.5 µM), blocked, and incubated with various dilutions

M. Mirabet and others

493A2B adenosine receptors and T cell activation

of the ME and the MI antisera. The antibodies were detected withperoxidase-conjugated sheep anti-rabbit IgG antibodies (BoehringerMannheim). Reactivity could be detected in the ELISA down to a 105-fold dilution for both antisera (data not shown).

Purification of antibodies from whole rabbit serum was achievedby affinity chromatography using the appropriate immunizing peptidecoupled to Thiol-Sepharose 4B. Columns were prepared using 3 mgof peptide/ml of swollen activated Thiol-Sepharose 4B, in accordancewith the manufacturer’s instructions. For antibody purification, 8 mlof antiserum was diluted (3-fold) in 10 mM PBS, pH 7.4, and thismaterial was recirculated for 2 hours at room temperature or overnightat 4°C through 3 ml of the peptide-Sepharose column. The columnwas then washed with PBS and antibodies were eluted by addition of50 mM glycine/HCl, pH 2.3. Fractions (1 ml) were collected in tubescontaining 20 µl of 1 M Tris, to give a final pH of 7.4. Protein-containing fractions were pooled, dialyzed against 2000 volumes ofPBS and aliquoted for storage at −80°C. The affinity purifiedantibodies obtained from ME and MI antisera were called MPE1 andMPI1, respectively.

Antibody conjugation to fluorocromesPurified antibodies were dialyzed against labeling buffer (50 mMH3BO3, 200 mM NaCl, pH 9.2) at 4°C. Subsequently, 20 µl of 5 mg/mlFITC (for MPE1) or TRITC (for MPI1) in DMSO were added for eachmilligram of antibody. After 2 hours at room temperature, unboundfluorocrome was separated by gel filtration using a Sephadex G-25column. Fluorochrome-conjugated antibodies were stored at −80°C.

Immunostaining and immunofluorescence assaysTransfected and wild-type CHO cells grown onto glass coverslips andperipheral blood lymphocytes and Jurkat cells (4×106) were washedwith PBS and fixed in 2% paraformaldehyde, 60 mM sucrose in PBS,pH 7.4, for 15 minutes at room temperature. In some experiments cellswere permeabilized by adding 0.05% saponin to the fixation solution.Then cells were washed twice with PBS containing 20 mM glycine

(buffer A) and were incubated for 15 minutes at room temperaturewith buffer A containing 1% BSA and 0.05% NaN3 (buffer B) beforeaddition of the antibodies. For indirect immunostaining, samples wereincubated for 45 minutes at 37°C with 50 µg/ml MPE1 in buffer B(50 ml), washed 3 times with the same buffer and stained with FITC-conjugated anti-rabbit IgG antibody (dilution 1:50; BoehringerMannheim). For direct immunostaining, incubation (45 minutes at37°C) was performed with 100 µg/ml FITC-MPE1 and/or 100 µg/mlTRITC-MPI1. For staining of lymphocytic populations, samples wereincubated (30 minutes in ice) with PE-anti CD3, PE-anti CD4 or PE-anti CD8 mAb and washed before adding FITC-MPE1 antibody (30minutes in ice). After 3 washes with buffer B, stained cells weremounted in Immuno-Fluore Mounting Medium for confocalmicroscopy analysis. Alternatively, cell suspensions wereresuspended in buffer B for flow cytometry analysis.

Stained cell suspensions were analyzed using an Epics Elite flowcytometer (Coulter Corporation; Hialeah, FL, USA). FITC and PEexcitation was obtained by a 488-nm Argon laser lamp and theirfluorescences were collected using 525 nm and 575 nm band-passfilters, respectively. The parameters used to select cell populations foranalysis were forward and side light scatter.

Confocal microscope observations were made with a Leica TCS 4Dconfocal scanning laser microscope adapted to an inverted LeitzDMIRBE microscope (Leica Lasertechnik GmbH, Heidelberg,Germany). The light source was an Argon Krypton laser and FITCand TRITC were respectively excited at 488 and 568 nm. In thissystem a 580 nm short-pass filter directs the emitted green (FITC) andred (TRITC) fluorescences towards two photomultipliers, one with a513-527 nm band-pass filter and the other with a 590 nm long-passfilter. The colocalization analysis was made by means of the MultiColor software (version 2.0; Leica Lasertechnik GmbH).

Analysis of intracellular Ca 2+ by double wavelengthfluorimetryCells (5×106/ml) were loaded with 5 µM Indo-1/AM in HBSS buffer

EXTRACELLULAR

INTRACELLULAR

NH2

HOOC F

FRYT

DR

YE

I

V

AL VA L

S VLA

G N V LA

V C A A

P T

D G

IFL

SA IL G

V AIA

L A A AV

F L V S

C FVL

SQTI

V

FISLLSAVA V

T PGI

LGFA

VWLVI

L

GR S

S

MV

N

S

FY FY M

C VGF

P P L LL

I M V

TNP

F

VC VQ P

L CAH

V G I FW

AL M I

NI

L

M

SLA

AH

V V N PN

YVI

V

LAF

F

AL

F

GF

LTC

A

FD

YCL

MLLE

DAL

TQ

G

TQN

AT

LT

TP

YVN

A

PN

M

GQ A

KK

KW

V

I

L

LP

RY

A

CV

VT

L

GT

R

SKR Y

DA

LV

G T

WG

N

C

SK

D

T

SA C

NNCT E P D

TNES

ENFLCKV

L

VP

I

LV

RQ

IH

E

A

RQL

R

TT

SHDMLET

LQR

C

YI

KIF

A AK

L

I K HY R SV A L L IKSGNGQAGVQPA D Q C

AYRN

LGVGL

W

Antiserum: MEPurified antibody: MPE1

Antiserum: MIPurified antibody: MPI1

Fig. 1.Sequence and predicted membrane-spanning topography of the human A2Badenosine receptor. The amino acid sequencesused for antibody generation are indicatedwith dark circles. The figure is adapted fromthe amino acid sequence reported by Pierce etal. (1992) and from the receptor structureproposed by Stehle et al. (1992). Consensussequences for glycosylation (Y) and acylation(` /̀) are indicated.

494 M. Mirabet and others

495A2B adenosine receptors and T cell activation

(140 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1 mM MgSO4, 1.2 mMCaCl2, 10 mM Hepes, 5 mM glucose, 0.3 mM KH2PO4 and 2 mMNa2HPO4), pH 7.0, for 30 minutes at 37°C. Then, an equal volumeof HBSS, pH 7.4, containing 10% heat inactivated fetal calf serum,was added and the cell suspension was incubated for 30 minutes. Afterwashing with HBSS, pH 7.2, containing 5% heat inactivated fetal calfserum and 10 µg/ml bovine pancreatic deoxyribonuclease I (Sigma),cells were resuspended at 5×106 cells/ml and maintained at roomtemperature in the dark until use. Just before the analysis, cells werediluted to 1×106 cells/ml and warmed to 37°C. Fluorescence was

monitored with a RF-5000 Shimadzu spectrofluorimeter in cuvettesthermostatically controlled at 37°C and continuously stirred. The cellsuspension was excited at 355 nm and fluorescence emissions weredetected at 405 (Ca2+-bound dye) and 485 nm (Ca2+-free dye). Final[Ca2+]i values were calculated from the ratio of emissionfluorescences (405/485 nm) using the equation described byGrynkiewickz et al. (38), with a Kd value of 250 nM for Indo-1. TheRmax value was obtained by lysing the cells with 0.1% Triton X-100,followed by an addition of excess EGTA for Rmin.

Determination of cAMPPeripheral blood lymphocytes (PBL) were preincubated at 2×106

cells/ml in HBSS, pH 7.2, containing 30 µM RO-20-1724(Calbiochem) for 10 minutes at 37°C. 250 µl of the cell suspensionwere added to tubes containing 2.5 µl of the appropriate stimuli and,after a 10 minute incubation at 37°C, 500 µl of ice-cold ethanol wereused to stop the reactions. Samples were centrifuged at 2000 g for 20minutes at 4°C and supernatants were lyophilized. Quantification ofcAMP levels was determined using an enzyme-immunoassay kit fromAmersham (Amersham Iberica, Madrid, Spain).

Interleukin-2 determinationPeripheral blood lymphocytes were incubated in 96-well plates at 106

cells/ml in the presence or absence of the adenosine receptor agonists.Wells were precoated with 2.5 µg/ml OKT3 and the incubation wasperformed for 72 hours at 37°C. Samples were centrifuged and thesupernatants were tested for interleukin-2 content using a humaninterleukin-2 ELISA kit from Endogen (Woburn, USA).

Fig. 3.Phenotype analysisof A2BR+ lymphocytesfrom PBMC by two colorimmunofluorescence. Theupper dot plot (left)represents side versusforward scatter distributionof total PBMC and B1 inred represents the gatedpopulation. Quadrants wereplaced according to thenonspecific labelingobtained using FITC-goatanti-mouse and PE-goatanti-mouse treated cells. X-axis represents the greenfluorescence (log scale)using FITC-MPE1; Y-axisrepresents the redfluorescence (log scale)using the respective PE-conjugated anti subsetmarker mAb. Thepercentage of cells in eachquadrant is indicated. Datacorrespond to arepresentative experiment.

Fig. 2.Specificity of the anti-A2BR antibodies. (A) Confocal images ofFITC-MPE1 labeling in fixed wild-type CHO cells (left) and TRITC-MPI1 labeling in permeabilized wild-type CHO cells (right).(B) Confocal images of fixed A2BR-transfected CHO cells labeledwith FITC-MPE1 antibody (left) or with TRITC-MPI1 antibody(right). (C) Confocal images of permeabilized A2B-R-transfected CHOcells labeled as in B. The superposition of FITC and TRITC labelingshows a high degree of colocalization (in yellow), as it can be seen alsoin the cytofluorogram. (D) Confocal image of fixed HEP-G2 cellsstained with FITC-MPE1 antibody. (E) Confocal image of fixed andpermeabilized CHO cells overexpressing A2A-M2 flag receptorsstained using 15 µg/ml of anti-M2 flag antibody. (F) Fixed andpermeabilized CHO cells overexpressing A2A-M2 flag receptorsstained using the MPE1 antibody (top) or the MPI1 antibody (bottom).Note the lack of crossreactivity of the MPE1 antibody against the A2Areceptor and some crossreactivity of the MPI1 antibody.

496

Protein determinationProtein was determined by the bicinchoninic acid method (Pierce) asdescribed by Sorensen and Brodbeck (1986) and using BSA asstandard.

RESULTS

Specificity of the anti-A 2BR antibodiesThe specificity of MPE1 and MPI1, which are antipeptideantibodies directed against the third extracellular loop and aportion of the C-terminal domain of the human A2BR,respectively (see Fig. 1 and Materials and Methods), has beenassessed by using CHO cells transfected with the cDNAencoding for human A2BR. Untransfected cells were notlabeled whereas transfected cells showed a very high level oflabeling (Fig. 2), which is consistent with overexpression offunctional A2BRs positively coupled to adenylate cyclase(Pierce et al., 1992). As expected, MPE1 antibody, which isdirected against an extracellular epitope of A2BR was able todetect A2BR molecules in either nonpermeabilized or

permeabilized cells (Fig. 2B,C). Due to the specificity of MPI1towards an intracellular region of the receptor, this antibodyonly detected A2BRs in permeabilized cells (Fig. 2C). Liver isa tissue where the level of A2BR expression is very low (Stehleet al., 1992). For this reason, the human hepatocellularcarcinoma-derived cell line, HEP-G2, was used to test whetherMPE1 led to some cell surface labeling. In these cells whereA2BRs were barely detectable by pharmacological techniques,MPE1 did not lead to specific labeling, thus indicating that theantibodies did not recognize molecules different from A2BR(Fig. 2D). Moreover this antibody did not recognize A1adenosine receptors expressed in neurons (not shown). Thehigh degree of colocalization of the label in permeabilizedtransfected CHO cells stained using MPE1 or MPI1 indicatethat these two antibodies recognize the same molecule.Whereas MPE1 did not recognize A2ARs receptorsoverexpressed in CHO cells (Fig. 2E), MPI1 lead to some faintlabeling in permeabilized cells overexpressing A2ARs (Fig.2F). None of these antibodies did detect by immunoblottingany band, which could be readily considered specific forA2BRs. All these results indicate that MPE1 antibody works

M. Mirabet and others

Fig. 4. Immunodetection ofA2BR in resting peripheral bloodlymphocytes (A) and Jurkat Tcells (B). Cells were fixed andlabeled with MPE1 antibodyand a FITC-conjugatedsecondary antibody as describedin Materials and Methods.Fluorescence of labeled cellswas analyzed by flow cytometry(specific label is in bold solidline) and confocal microscopy.

497A2B adenosine receptors and T cell activation

well for immunocytochemistry and is specific for A2BRswhereas MPI1 antibody presents some cross-reactivity withA2ARs. All subsequent immunocytochemical studies wereperformed with the MPE1 antibody.

Expression of A 2BR in resting lymphocytesThe MPE1 antibody, which specifically recognizes anextracellular epitope of A2BR, was used to analyze theexpression of these receptors in lymphocytes. The percentageof peripheral blood lymphocytes expressing A2BRs was 58±13(mean ± s.d., n=13), with individual values ranging from 29 to74. Double labeling experiments were performed to analyzethe expression of the receptor in different subsets of T cells.According to Fig. 3 no significant changes in the percentageof A2BR+ cells were found among CD4+ or CD8+ T-cellsubpopulations. In general terms, the intensity of the labelingin A2BR+ cells was moderate (Fig. 4A). In the T cell line Jurkat,the number of cells expressing the receptor (>95%) and thelevel of expression was higher than that found in bloodlymphocytes (Fig. 4B).

A2BRs in peripheral blood lymphocytes are functional sincethe agonist NECA led to increases in cAMP levels (Fig. 5). Therelative potency of the different agonists at a concentration of25 µM (NECA»R-PIA>CGS21680) is consistent with thepharmacology of A2BR. Due to the fact that NECA, via A2BR,leads to intracellular calcium mobilization in Jurkat cells(Mirabet et al., 1997), the possible effect of NECA upon

calcium levels was investigated in peripheral bloodlymphocytes. The compound neither had any effect by itself norit altered the effect exerted by OKT3, a mAb directed againstthe T cell receptor/CD3 complex which mobilizes calcium viainositol phosphate-sensitive stores (Fig. 6). The same failure todetect NECA-induced changes in calcium levels was evidenced

CONTROL NECA R-PIA CGS 21680

5

4

3

2

1

0

pm

ol c

AM

P/1

06 c

ells

Fig. 5.Cyclic AMP production via A2B adenosine receptors. PBLwere incubated with the cAMP phosphodiesterase inhibitor RO-20-1724 for 10 minutes. Then 25 µM of each adenosine receptor agonistwas added and cells were stimulated for 10 minutes at 37°C. cAMPlevels were determined as described in Materials and Methods.

Fig. 6.Lack of calcium mobilization in PBL via A2BR. A fraction of PBL was maintained in the culture conditions described in Materials andMethods until the Ca2+ determination assay (cells in A). Another fraction was activated with 1 µg/ml PHA for 72 hours, washed and maintainedfor an additional 24 hour-period in complete medium containing 10 units/ml IL-2 (cells in B). Then resting (A) and activated (B) PBL wereloaded with Indo-1 and the effect of 100 µM NECA on intracellular calcium levels was analyzed, both in the absence of a previous Ca2+-mobilizing stimulus or after the addition of 1 µg/ml OKT3. Traces correspond to a representative experiment performed with PBL obtainedfrom a donor.

NE

CA

400

300

200

100

0

[Ca2+

] i (

nM

)

400

300

200

100

0

600

500

400

300

200

100

0 1 min

NE

CA

OK

T3A

B 600

500

400

300

200

100

0

NE

CA

NE

CA

OK

T3

OK

T3

OK

T3

[Ca2+

] i (

nM

)

[Ca2+

] i (

nM

)[C

a2+] i

(n

M)

1 min

498

irrespective of the blood donor. On the other hand, NECA wasunable to mobilize intracellular calcium in T cells expanded bychronic activation with phytohemagglutinin (PHA) followed byIL-2 treatment (Fig. 6).

Expression and function of A 2BR in T cell activationeventsThe expression of A2BR was analyzed in activated cells. Two

activation signals were used, a relatively unspecific, triggeredby PHA and one specific, triggered by the anti-T cellreceptor/CD3 complex mAb, OKT3. Irrespective of thestimulus, the percentage of lymphocytes expressing A2BRs wassignificantly higher (72±10 with PHA and 79±13 with OKT3)than that found in resting cells.

When OKT3-activated cells were analyzed by forward andside scatter criteria, two different populations (B1 and B2)

M. Mirabet and others

Fig. 7.Expression of A2BR on the cell surface of activated lymphocytes.Isolated mononuclear cells were incubated in flasks precoated or not withpurified OKT3 mAb. (A) Dot plot representing the side (properties ofcytoplasmic and nuclear components) versus forward scatter (cell size)distribution of activated lymphocytes. B1, in red, represents the gatedpopulation remaining in the same region than nonstimulated lymphocytes(not shown), whereas B2, in green, represents the lymphocyte populationwith higher size and cellular complexity generated upon activation.(B) Dot plots representing the forward scatter versusthe fluorescenceintensity of cell surface A2BR in the above described two populations, B1(red) and B2 (green). The patterns for unstimulated and stimulated cellsare represented in the upper part and the lower part of the panel,respectively. Unspecific labeling is shown in the images on the left. Notethat although in this experiment there are no major differences onfluorescence intensity between stimulated and unstimulated cells in theB1 population, we found significant differences in 5 out of 9 experimentsperformed with blood from different donors.

499A2B adenosine receptors and T cell activation

appeared as described elsewhere (Martín et al., 1995). B1population contains cells structurally similar to thenonstimulated controls, whereas larger, more complex cells areincluded within B2 population (Fig. 7A). The percentage ofcells expressing A2BR in B1 population was similar (Fig. 7B)or slightly higher than in resting cells. Following activationwith OKT3, the majority of cells (>91±7%, n=9, range: 81-99%) in the B2 population expressed enhanced levels of A2BRs(Fig. 7B). Variations in the intensity of labeling weremonitored by flow cytometry (Fig. 7B) and by confocalmicroscopy (Fig. 8). The level of cell surface A2BR expressionin small cells, which correspond to the B1 population ofOKT3-activated cells, was slightly higher than that found inresting cells (see legend to Fig. 7), whereas the expression ofthe receptor in cells from the B2 population (cells of highersize) was markedly enhanced. Similar results were obtainedwhen cells were activated with PHA although the degree oflabeling enhancement was not as marked as in the case oflymphocytes activated via the T cell receptor (results notshown). These results provide evidence that the expression ofA2BR increases upon lymphocyte activation.

The role of A2BRs present in activated lymphocytes wasinvestigated by measuring interleukin-2 production in cellsactivated using OKT3. The presence of NECA, at aconcentration which activates A2BRs (5 µM), led to asignificant reduction of interleukin-2 levels produced byactivated T cells. In contrast, similar concentrations of eitherR-PIA or CGS21680 did not exert any effect (Fig. 9). Theseresults indicate that adenosine, via A2BRs, mediates a reductionin the production of interleukin-2. This probably constitutes afeed-back mechanism for lymphocyte deactivation.

DISCUSSION

The results presented in this paper indicate that the antipeptideantibody MPE1 is specific for human A2BRs inimmunolabeling assays. The specificity is based on the resultsobtained with wild-type and A2BR-transfected CHO cells, onthe lack of label when using a hepatoma-derived human cellline in which A2BRs are not detectable by pharmacologicalprocedures, and on the failure to recognize other adenosinereceptor subtypes such as the A1 or A2A receptors.

0

200

400

600

800

IL-2

(p

g/m

l ) *

2.5 µg/ml OKT3 - + - + - + - +

Untreated 5 µM NECA 5 µM CGS 21680 5 µM R-PIA

Fig. 9.Effect of adenosine analogs on interleukin-2production. PBLs were activated using 2.5 µg OKT3 inthe presence or the absence of NECA, R-PIA orCGS21680. Interleukin-2 secreted to the medium in 72hours (37ºC), was measured as indicated in Materials andMethods. Data are the mean ± s.e.m. from a representativeexperiment in triplicates. *P<0.01.

Fig. 8.Confocal microscopy images of resting and OKT3-activatedPBL labeled with MPE1 antibody. Note that the intensity of labelingin activated cells is enhanced in cells of higher size. Small cells (B1population) are marked with an arrowhead. Bar, 10 µm.

500

This is the first report showing direct evidence of A2BRexpression on the cell surface of blood cells. Based upon thebinding of the nonselective agonist NECA to human Tlymphocytes, Schultz et al. (1988) suggested that A2 but notA1, receptors were present in membranes from these cells. Asimilar conclusion was obtained by the agonist-inducedincreases in cAMP levels observed in human T lymphocytesand thymocytes (Dinjens et al., 1986). The establishment ofwhich of the two A2 receptor subtypes, A2AR or A2BR, isoperating in human T lymphocytes has remained elusive. Thefact that adenosine and NECA are low-potency stimulators ofcAMP production, has led to the suggestion that A2B receptorsare those mainly found in the different T lymphocytes and Tcell lines assayed (Fredholm et al., 1991). The results of flowcytometry (Figs 3 and 4) indicate that the A2BR is expressed,but at a relatively low level, in a percentage of restinglymphocytes (58±13). In comparative terms, the level of A2BRson the cell surface of Jurkat T cells was higher than inperipheral blood cells (Fig. 4). The functionality of A2BRs inthese samples isolated from the blood of human donors isproven by the agonist-induced increase in cAMP levels. Thedifferent capacity for producing cAMP exerted by NECA,CGS21680 and R-PIA (Fig. 5) is consistent with a stimulationof adenylate cyclase activity via A2BR and not via A2AR.Working with lymphocyte membranes we have failed to detectspecific [3H]CGS21680 binding (data not shown), which issuggestive of the absence of A2ARs in these blood cells. Theabsence of functional A2ARs and the presence of A2BRscoupled to adenylate cyclase has been shown in Jurkat cells(Mirabet et al., 1997). It should be noted that activation ofA2BRs in Jurkat cells leads to calcium mobilization by inositolphosphate-insensitive stores (Mirabet et al., 1997).Interestingly, A2BRs in either resting or activated humanlymphocytes are not coupled to calcium channels (Fig. 6) incontrast to what happens in the leukemia-derived T cell line.Other cAMP-raising agents such as prostaglandins, alsoproduce increases in intracellular calcium by inositolphosphate-independent pools in Jurkat cells (Kelley et al.,1990) but not in peripheral blood lymphocytes (Chouaib et al.,1987). This suggests that different G protein-coupled receptor-mediated signaling pathways operate depending upon the stageof T cell development.

In this report we also provide evidence that the expressionof A2BR in lymphocytes changes upon activation. Activationof T cells via PHA or TCR/CD3 leads to a moderate butsignificant increase in the percentage of cells expressing A2BR.Moreover, according to flow cytometric data and micrographsshown in Figs 7 and 8, there was a marked increase influorescence intensity, thus indicating that the number ofmolecules on the plasma membrane of activated T cellsincreased significantly. This behavior is similar to thatobserved for CD26, which is expressed at relatively lowintensity in about half of resting lymphocytes, whereas itsexpression is markedly enhanced upon activation (Martín et al.,1995). CD26, which is considered a T cell activation antigen,is able to bind ecto-adenosine deaminase, the extracellularadenosine-consuming enzyme, to the surface of lymphocytes.The lack of functional A2AR (at least in resting lymphocytes),the demonstration of A2BRs in human lymphocytes and the up-regulation of A2BR in T cell activation strongly indicates thatreceptor-mediated adenosine effects are elicited in these cells

by engagement of A2BRs. In fact, a reduction in interleukin-2production is observed when OKT3-activated T lymphocyteswere treated with NECA, at a concentration that ensuresactivation of A2BRs. Other adenosine analogs, R-PIA orCGS21680, which are more selective for A2A receptors, did nothave any effect. This indicates that upregulation of A2BRs is amechanism for lymphocyte deactivation. A similar role hasbeen devised for A2BR expressed in murine macrophages. Inthese cells A2BR are upregulated in response to interferon-γand they mediate an inhibition of the interferon-γ-inducedexpression of either MHC class II genes, nitric oxide synthaseor pro-inflammatory cytokines (Xaus et al., 1999). Therefore,up-regulation of A2BR expression induced by activation signalscan be a feed-back mechanism for deactivation. Thismechanism seems to operate in a variety of immunocompetentcells and can be important for the design of newimmunoregulatory and antiinflammatory drugs.

Using murine lymphocytes Huang et al. (1997) have shownthat agonists of A2ARs modulate the T cell receptor-mediatedCD25 upregulation. In a murine lymphoblastoid cell line theA2BR is also expressed (data not shown) and, therefore, A2BRare probably expressed on activated murine T cells. This wouldsuggest that in the murine model adenosine likely regulates Tcell activation events via both A2A and A2B receptors. Besidesmurine lymphocytes, a variety of cell types from differentsources coexpress A2A and A2B receptors. Activation of A2ARand A2BR in the same cell may lead to synergy in cAMPproduction. In addition, adenosine may deliver divergent signalsif each of these receptors subtypes are linked to differenteffector molecules. In nonlymphoid cell systems (Windscheif,1996) and in Jurkat (Mirabet et al., 1997) or CEM T cells (datanot shown) adenosine receptors can be linked to a variety ofsignaling pathways in addition to that of adenylate cyclase.

The expression analysis in the two structurally different cellpopulations found in T lymphocytes triggered via the T cellreceptor/CD3 complex (B1 and B2) suggests that the up-regulation of A2BR is higher in the population of cellsresponding better to the stimulus (B2), which present a higherintracellular complexity (Martín et al., 1995). In the case ofmitogen-stimulated cells it is also found that the population ofstructurally more complex cells express more surface A2BR. Ina previous study we have analyzed the pattern of expression ofecto-adenosine deaminase and CD26 in B1 and B2 populationsfrom activated T cells (Martín et al., 1995). The pattern foundwas very similar to that presented here for A2BR, which indicatesthat expression of the three cell surface molecules, A2BR, ecto-adenosine deaminase and CD26, is analogously regulated inactivated T lymphocytes. It is difficult to reconcile thecoordinated expression ecto-adenosine deaminase/adenosinereceptor with the fact that the enzyme degrades the ligand of thereceptor. This apparent contradiction can be circumvented ifsome kind of regulation of the enzyme activity occurs or ifadenosine deaminase has an extraenzymatic role. In a cell modelin which A1 adenosine receptors and ecto-adenosine deaminaseare expressed, it has been demonstrated that both proteinsinteract on the cell surface (Ciruela et al., 1996) and, in responseto receptor agonists, they are internalized together using thesame endocytic pathway (Saura et al., 1998). Interestingly, insuch a cellular model, the interaction ecto-adenosinedeaminase/A1 adenosine receptor is necessary for efficientsignaling via A1 receptors. Actually, catalytically inactive ADA

M. Mirabet and others

501A2B adenosine receptors and T cell activation

can deliver signals through its interaction with A1 adenosinereceptors in nonlymphoid cells (Ciruela et al., 1996; Saura et al.,1996, 1998) or with CD26 in T lymphocytes (Martín et al.,1995). Thus, coordinated upregulation of A2BRs and ecto-adenosine deaminase in T cell activation events may reflect afunctional interaction not necessarily related to the catalyticactivity of the enzyme. This possibility may lead to new insightsinto the molecular basis of the pathophysiology of adenosinedeaminase-related SCID.

We acknowledge the technical help received from Jaume Comasand Mª del Rosario González (flow cytometry section), from SusanaCastel and Elisenda Coll (confocal microscopy section), in the ServeisCientifico Tècnics de la Universitat de Barcelona, and from CatalinaRelaño from the Servei de Cultius Cellulars. We thank the advicereceived from Dr Ampar Castell from Biokit company (Lliçàd’Amunt, Barcelona), which facilitated production of antipeptideantibodies. We are indebted to the Fundación María Francisca deRoviralta for financial support in the acquisition of thespectrofluorimeter needed for the measurement of intracellularcalcium levels. We thank Robin Rycroft from the Serveid’Assessorament Lingüístic de la Universitat de Barcelona for theexcellent technical assistance in the preparation of the manuscript. M.Mirabet is the recipient of a fellowship from ComissióInterdepartamental de Recerca i Innovació Tecnològica (CIRIT).Supported by Grants from CICYT (PB94/0941 and PB97/0984) andfrom CICYT’s I+D program Salud y Farmacia (SAF97/0066).

REFERENCES

Antonysamy, M. A., Moticka, E. J. and Ramkumar, V. (1995). Adenosineacts as an endogenous modulator of IL-2 dependent proliferation ofcytotoxic T lymphocytes. J. Immunol. 155, 2813-2821.

Apasov, S. G., Koshiba, M., Chused, T. M. and Sitkovsky, M. V. (1997).Effects of extracellular ATP and adenosine on different thymocyte subsets:Possible role of ATP-gated channels and G protein-coupled purinergicreceptor. J. Immunol. 158, 5095-5105.

Boyum, A. (1968). Separation of leukocytes from blood and bone marrow.Scand. J. Clin. Invest. 21, 51-62.

Chouaib, S., Robb, R. J., Welte, K. and DuPont, B. (1987). Analysis ofprostaglandin E2 effect on T lymphocyte activation. Abrogation ofprostaglandin E2 inhibitory effect by the tumor promotor 12. 0 tetradecanoylphorbol-13 acetate. J. Clin. Invest. 80, 11-25.

Ciruela, F., Saura, C., Canela, E. I., Mallol, J., Lluis, C. and Franco, R.(1996). Adenosine deaminase affects ligand-induced signalling byinteracting with adenosine receptors. FEBS Lett. 380, 219-223.

Daly, J. W., Butts-Lamb, P. and Padgett, W. (1983). Subclasses of adenosinereceptors in the central nervous system: interactions with caffeine andrelated methylxanthines. Cell. Mol. Neurobiol. 3, 69-80.

Dinjens, W. N. M., van Doorn, R., van Laarhoven, J. P. R. M., Roos, R.,Zeijlemaker, W. P. and de Bruijn, C. H. M. M. (1986). Adenosinereceptors on human T lymphocytes and human thymocytes. Adv. Exp. Med.Biol. 195, 1-6.

Dos Reis, G. A., Nóbrega, A. F. and Paes de Carvalho, R. (1986). Purinergicmodulation of T-lymphocyte activation: differential susceptibility of distinctactivation steps and correlation with intracellular 3′-5′-cyclic adenosinemonophosphate accumulation. Cell. Immunol. 101, 213-231.

Franco, R., Valenzuela, A., Lluis, C. and Blanco, J. (1998). Enzymatic andextraenzymatic role of ecto-adenosine deaminase in lymphocytes. Immunol.Rev. 161, 27-42.

Fredholm, B. B., Ahlberg, S., Altiok, N., Gerwins, P., Parkinson, F., vander Ploegh, I., Kvanta, A., Nordstedt, C. and Dunwiddie, T. (1991).Interactions between adenosine receptors and other membrane receptors. InRole of Adenosine and Adenine Nucleotides in the Biological System(ed. S.Imai and K. Nakazawa), pp. 173-182. Elsevier Science, Amsterdam.

Fredholm, B. B. (1995). Adenosine, adenosine receptors and the actions ofcaffeine. Pharmacol. Toxicol. 76, 93-101.

Grynkiewicz, G., Poenie, M. and Tsien, R. Y. (1985). A new generation of

Ca2+ indicators with greatly improved fluorescence properties. J. Biol.Chem. 260, 3440-3450.

Hershfield, M. S. and Mitchell, B. S. (1989). Immunodeficiency diseasescaused by adenosine deaminase deficiency and purine nucleosidephosphorylase deficiency. In The Metabolic and Molecular Bases ofInherited Disease, 7th edition (ed. C. R. Scriver, A. L. Beaudet, W. S. Slyand D. Valle), pp. 1725-1768. McGraw-Hill, New York.

Huang, S., Apasov, S., Koshiba, M. and Sitkovsky, M. V. (1997). Role of A2aextracellular adenosine receptor-mediated signaling in adenosine-mediatedinhibition of T-cell activation and expansion. Blood90, 1600-1610.

Kelley, L. L., Blackmore, P. F., Graber, S. E. and Stewart S. T. (1990).Agents that raise cAMP in human T lymphocytes release an intracellularpool of calcium in the absence of inositol phosphate production. J. Biol.Chem. 265, 17657-17664.

Kitagawa, T. and Aikawa, T. (1976). Enzyme coupled immunoassay ofinsulin using a novel coupling reagent. J. Biochem. 79, 233-236.

Kyte, J. and Doolitte, R. F. (1982). A simple method for displaying thehydropathic character of a protein. J. Mol. Biol. 157, 105-132.

Linden, J., Taylor, H. E., Robeva, A. S., Tucker, A,L., Stehle, J. H., Rivkees,S. A., Fink, J. S. and Reppert, S. M. (1993). Molecular cloning andfunctional expression of a sheep A3 adenosine receptor with widespreadtissue distribution. Mol. Pharmacol. 44, 524-532.

Linden, J. (1994). Purinergic systems. In Basic Neurochemistry: Molecular,Cellular and Medical Aspects, 5th edition (ed. G. J. Siegel), pp. 401-416.Raven Press, New York.

Londos, C., Cooper, D. M. F. and Wolff, J. (1980). Subclasses of externaladenosine receptors. Proc. Nat. Acad. Sci. USA77, 2551-2554.

Marone, G., Plaut, M. and Lichtenstein, L. M. (1978). Characterization ofa specific adenosine receptor on human lymphocytes. J. Immunol. 11, 2153-2159.

Martín, M., Huguet, J., Centelles, J. J. and Franco, R. (1995). Expressionof ecto-adenosine deaminase and CD26 in human T cells triggered by theTCR-CD3 complex: Possible role of adenosine deaminase as costimulatorymolecule. J. Immunol. 155, 4630-4643.

Mirabet, M., Mallol, J., Lluis, C. and Franco, R. (1997). Calciummobilization in Jurkat cells via A2b adenosine receptors. Br. J. Pharmacol.122, 1075-1082.

Pierce, K. D., Furlong, T. J., Selbie, L. A. and Shine, J. (1992). Molecularcloning and expression of an adenosine A2b receptor from human brain.Biochem. Biophys. Res. Commun. 187, 86-93.

Rivkees, S. A., Lasbury, M. E. and Barbhaiya, H. (1995). Identification ofdomains of the human A1 adenosine receptor that are important for bindingreceptor subtype-selective ligands using chimeric A1/A2A adenosinereceptors.J. Biol. Chem.270, 20485-20490.

Resta, R. and Thompson, L. F. (1997). SCID: the role of adenosinedeaminase deficiency. Immunol. Today18, 371-374.

Salvatore, C. A., Jacobson, M. A., Taylor, H. E., Linden, J. and Johnson,R. G. (1993). Molecular cloning and characterization of the human A3adenosine receptor. Proc. Nat. Acad. Sci USA90, 10365-10369.

Stiles, G. L. (1992). Adenosine receptors. J. Biol. Chem. 267, 6451-6454. Saura, C., Ciruela, F., Casadó, V., Canela, E. I., Mallol, J., Lluis, C. and

Franco, R. (1996). Adenosine deaminase interacts with A1 adenosinereceptors in pig brain cortical membranes. J. Neurochem. 66, 1675-1682.

Saura, C., Mallol, J., Canela, E. I., Lluis, C. and Franco, R. (1998).Adenosine deaminase and A1 adenosine receptors internalize togetherfollowing agonist-induced receptor desensitization. J. Biol. Chem. 273,17610-17617.

Schubert, P., Rudolphi, K. A., Fredholm, B. B. and Nakamura, Y. (1994).Modulation of nerve and glial function by adenosine. Role in thedevelopment of ischemic damage. Int. J. Biochem. 26, 1227-1236.

Schultz, L. A., Kammer, G. M. and Rudolph, S. A. (1988). Characterizationof the human T lymphocyte adenosine receptor: comparison of normal andsystemic lupus erythematosus cells. FASEB J. 2, 244-250.

Sorensen, K. and Brodbeck, U. (1986). A sensitive protein assay using micro-titer plates. Experientia42, 161-162.

Stehle, J. H., Rivkees, S. A., Lee, J. J., Weaver, D. R., Deeds, J. D. andReppert, S. M. (1992). Molecular cloning and expression of the cDNA fora novel A2-adenosine receptor subtype. Mol. Endocrinol. 6, 384-393.

Tanaka, T., Slamon, D. J. and Cline, M. J. (1985). Efficient generation ofantibodies to oncoproteins by using synthetic peptide antigens. Proc. Nat.Acad. Sci. USA82, 3400-3404.

Tucker, A. L. and Linden, J. (1993). Cloned receptors and cardiovascularresponses to adenosine. Cardiovasc. Res. 27, 62-67.

Van Calker, D., Müller, M. and Hamprecht, B. (1979). Adenosine regulates

502

via two different types of receptors, the accumulation of cyclic AMP incultured brain cells. J. Neurochem. 33, 999-1005.

Windscheif, U. (1996). Purinoceptors: From history to recent progress. Areview. J. Pharm. Pharmacol. 48, 993-1011.

Wolberg, G., Zimmerman, T. P., Hiemstra, K., Winston, M. and Chi, L.-C. (1975). Adenosine inhibition of lymphocyte-mediated cytolysis: Possiblerole of cyclic adenosine monophosphate. Science187, 957-959.

Xaus, J., Mirabet, M., Lloberas, J., Soler, C., Lluis, C., Franco, R. andCelada, A. (1999). IFN-γ up-regulates the A2B adenosine receptorexpression in macrophages. A mechanism of macrophage deactivation.J.Immunol.(in press).

Zhou, Q.-Y., Li, C., Olah, M. E., Johnson, R. A., Stiles, G. L. and Civelli,O. (1992). Molecular cloning and characterization of an adenosine receptor:the A3 adenosine receptor. Proc. Nat. Acad. Sci. USA89, 7432-7436.

M. Mirabet and others