The antiinflammatory mechanism of
methotrexate. Increased adenosine release at
inflamed sites diminishes leukocyte
accumulation in an in vivo model of
B N Cronstein, … , D Naime, E Ostad
J Clin Invest.
Methotrexate, a folate antagonist, is a potent antiinflammatory agent when used weekly in
low concentrations. We examined the hypothesis that the antiphlogistic effects of
methotrexate result from its capacity to promote intracellular accumulation of
5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) that, under conditions of cell injury,
increases local adenosine release. We now present the first evidence to establish this
mechanism of action in an in vivo model of inflammation, the murine air pouch model. Mice
were injected intraperitoneally with either methotrexate or saline for 3-4 wk during induction
of air pouches. Pharmacologically relevant doses of methotrexate increased splenocyte
AICAR content, raised adenosine concentrations in exudates from carrageenan-inflamed air
pouches, and markedly inhibited leukocyte accumulation in inflamed air pouches. The
methotrexate-mediated reduction in leukocyte accumulation was partially reversed by
injection of adenosine deaminase (ADA) into the air pouch, completely reversed by a
specific adenosine A2 receptor antagonist, 3,7-dimethyl-1-propargylxanthine (DMPX), but
not affected by an adenosine A1 receptor antagonist, 8-cyclopentyl-dipropylxanthine.
Neither ADA nor DMPX affected leukocyte accumulation in the inflamed pouches of
animals treated with either saline or the potent antiinflammatory steroid dexamethasone.
These results indicate that methotrexate is a nonsteroidal antiinflammatory agent, the
antiphlogistic action of which is due to increased adenosine release at inflamed sites.
Antiinflammatory Mechanism of Methotrexate
Increased Adenosine Release at Inflamed Sites Diminishes Leukocyte Accumulation inan InVivoModel of Inflammation
Bruce N. Cronstein, Dwight Naime, and Edward Ostad
DepartmentofMedicine, Division ofRheumatology, New York University Medical Center, New York 10016
Methotrexate, a folate antagonist, is a potent antiinflammatory agent when used weekly in low concentrations. We examined thehypothesis that the antiphlogistic effects of methotrexate result from its capacity to promote intracellular accumulation of5-aminoimidazole4-carboxamide ribonucleotide (AICAR)
that,underconditionsof cellinjury, increases local adenosine release. We now present the first evidence to establish this mechanism of action in an in vivo model of inflammation, the murine air pouch model. Mice were injected intraperitoneally with either methotrexate or saline for34wkduring induction of air pouches. Pharmacologically relevant doses of methotrex-ate increasedsplenocyte AICAR content, raised adenosine
con-centrations in exudates from carrageenan-inflamed air pouches, and markedly inhibited leukocyte accumulation in
in-flamed air pouches. Themethotrexate-mediated reduction in leukocyte accumulation was partially reversed by injection of
adenosine deaminase (ADA) into the air pouch, completely reversed by aspecific adenosineA2 receptor antagonist, 3,7-di-methyl-l-propargylxanthine (DMPX),but notaffectedby an adenosine
Alreceptor antagonist, 8-cyclopentyl-dipropylxan-thine. NeitherADAnorDMPXaffected leukocyte accumula-tion in theinflamedpouchesof animalstreated with either sa-line or the potent antiinflammatory steroid dexamethasone. Theseresults indicate that methotrexate isanonsteroidal
an-tiinflammatoryagent, theantiphlogisticactionofwhich is due toincreased adenosinereleaseatinflamed sites.(J. Clin. In-vest. 1993.92:2675-2682.) Keywords:leukocyte-
antagonistfirst developed forthe treat-mentof
widelyusedin thetreatmentof rheumatoid arthritis (1). Unlike its use in the treatment of
malignancies(pulses of 20-250
rheu-This materialwaspresented in abstract formattheannualmeetingof the American Federation for ClinicalResearch/Am.Soc. for Clinical Investigation/Am. Association ofPhysicians in Wash. DC, 2 May 1993.
AddresscorrespondencetoDr.Bruce N.Cronstein,Dept. of Medi-cine, Division of Rheumatology, New York University Medical Center, 550 First Avenue,NewYork,NY 10016.
Receivedfor publication6July1993 andinrevisedform26July 1993.
matoid arthritis and other inflammatory diseases (1). Al-though theoriginalrationale for the use of methotrexate in the treatmentofrheumatoid arthritis was "immunosuppression," themolecular mechanism by which methotrexate suppresses
inflammationis not well understood. It is unlikely that, in the dosesgiven,methotrexate diminishesproliferationof immune cells by inhibiting de novo purine and pyrimidine synthesis
sinceleukopeniaand mucosalulcerations,phenomena best
ex-plainedby theantiproliferative effects ofmethotrexate, are
con-sidered evidence ofdrugtoxicityand indications to decrease or stoptherapy. Otherproposed mechanisms includeadecrease inneutrophil (but not macrophage) leukotriene synthesis (2) andinhibition of transmethylation reactionsby inhibiting the
formationof S-adenosyl-methionine,amethyl donor required
We have recently proposed an alternative biochemical
mechanism of action ofmethotrexate; methotrexate promotes therelease ofthe antiinflammatory autocoid adenosineat
in-flamed sites (4). Previous studies havesuggestedthat metho-trexateaccumulates within cells and, both directly and
indi-rectly, inhibits 5-aminoimidazole-4-carboxamide ribonucleo-tide
(AICAR)'transformylase, resulting in the intracellular accumulation ofAICAR (Fig. 1; references 5-9). Increased
intracellular concentrations ofAICAR promote, byacomplex mechanism, theincreasedreleaseofthe potent
antiinflamma-tory autocoid adenosine ( 10, 11). Results ofinvitro studies
hypothesis(4);lowconcentrations (< 10nM) of
methotrexateorhigher concentrations ofthecell-soluble,
non-phosphorylated precursor of AICAR, AICARibonucleoside (acadesine), promote adenosine release from fibroblastsand
endothelial cells. Theincrease in extracellularadenosine
neutrophilstoadhereto themethotrexate-treated endothelialcells and
fibroblasts,inan in vitro model ofan
inflammatoryinteraction. Asako et al. (1.2)haveconfirmedthat methotrexate suppresses inflamma-tion by
usingthe hamstercheek pouchmodelofacuteinflammationbut
We report here thefirst evidence frominvivostudiesthat demonstrates that low-dose weekly methotrexate treatment causesintracellularaccumulation of
AICAR,increased adeno-sine releaseatsitesof
inflammation,andadenosine-dependent inhibitionof inflammation. Moreover,wehaveconfirmedthat inmethotrexate-treated mice adenosine diminishes inflamma-tion viaoccupancyof adenosine A2receptors. These data
pro-vide the first in vivo demonstration ofa novel biochemical mechanism ofactionformethotrexate.
1.Abbreviations usedin thispaper:AICAR, 5-aminoimidazole-4-car-boxamideribonucleotide; DMPX,3,7-dimethyl-1-propargylxanthine.
DeNovo Purine Biosynthesis
MTXu AICAR ATP
DHF8u FAICAR ADP SAM
KIMP: AMP THF
Inosine Adenosine SAH
Figure1.Proposed molecular mechanism of action of methotrexate. Shown herearethemajorsteps inpurine synthesisanddegradation. Abbreviations: GAR,
fl-glycinamideribonucleotide;FGAR, a-N-for-mylglycinamide ribonucleotide; MTXSU, methotrexate polygluta-mate;DHFSIU, dihydrofolate polyglutamate; AICAR, 5-aminoimida-zole-4-carboxamide ribonucleotide;FAICAR,formyl-AICAR;IMP, inosinicacid;THF,tetrahydrofolate(reduced); SAM, S-adenosyl-methionine;SAH,S-adenosylhomocysteine.
Materials. 3,7-Dimethyl-l-propargylxanthine and 8-cyclopentyl-di-propylxanthine were obtained from Res. Biochem. Inc. (Natick, MA). Injectable methotrexate (US Pharmacopeia)waspurchased from
Le-derle Laboratories Division of American Cyanamid (Wayne, NJ). Adenosine deaminase (Type IV, calf intestinal), carrageenan, and dexamethasone wereobtainedfromSigmaChemical Co. (St. Louis, MO). All other reagents were the highest grade that could be obtained. Induction ofairpouchesandcarrageenan-induced inflammation. To induceanair pouch, mice (6-wk-old, female Balb/c mice, Taconic Farms Inc.,Germantown, NY) were injected subcutaneously with 3 cc ofair (on the back) three times per wk. After 3-4wk,theair pouchwas
injectedwith 1 mlofa2%(wt/vol) suspension of carrageenan and mice were returned to their cages, where they were allowed to run free for4h.Attheend of4h, the animalswerekilled, 2 cm3 of normal salinewasinjected into the pouch and the contents of the pouchwere
aspirated,diluted 1:2with normal saline, and samples were taken for cellcount.Smearsof the undiluted fluidwereprepared andanaliquot
wasstained (Wright-Giemsa), revealing that >95% of the exudate cellswere PMNs. In someexperimentstheair pouchwasdissected from the subcutaneous tissue, fixed in formalin, processed for histo-logic sections, and stained (hematoxylin and eosin and Giemsa stains) usingstandard methods ( 13). These studies werereviewed and ap-proved by the Institutional Animal Care and Use Committee of New YorkUniversity Medical Center.
Treatment ofmice with methotrexate. Mice were treated with weekly intraperitoneal injections ( 1 ml) of methotrexate (U.S.P., 0.05-0.5 mg/kg)orpyrogen-free(U.S.P.) normal saline (0.9%) for 3 or4
wk. Therewere noapparent adverse effects of either treatment that
could bedetected by visual inspection of the animals and there were no apparent differences between the animals treated with saline and those treatedwith methotrexate.
Preparationofadenosinedeaminase. Adenosine deaminase(50,l,
4000U/ml)wasdialyzedagainstPBSovernight (4°C) before dilution andinjectioninto the air pouch.Forsomeexperiments, dialyzed adeno-sinedeaminasewasincubated with deoxycoformycin(1 MM) for 30 minat roomtemperaturebefore dilution ( 1:2600) in PBS containing carrageenan( 14).
Injectionofadenosine deaminase andadenosine receptor
antago-nists. In someexperiments adenosine deaminase, previously inacti-vatedadenosine deaminase (see above), and adenosine receptor antag-onistswereadded to the carrageenan suspension to an appropriate final concentration. The final volume of the carrageenan suspension (with
inhibitors)didnotdiffer from that in control mice
Histologic analysis ofsections ofairpouches. Slides of stained (Giemsa) sections ofmouseairpouchwereexaminedmicroscopically usingaLeitz researchmicroscopetowhichwasattacheda high-resolu-tion videocamera.Videoimageswereprojected directlyonto a screen
foranalysisbyuseof JAVA software(JandelSci.,CorteMadera,CA)
runon aZenith 386 computer. Allimagesweredigitized directlyand enhanced forcontrastandbrightness usingPHOTOSTYLER software (Aldus, Inc.,Seattle, WA).
QuantitationofAICAR.In someexperimentsthespleenswere har-vested and the cellswereisolatedbyscraping throughgauze. The cells were washed and resuspendedat100X 106/mlin PBS. The cellswere
thenlysed and theproteinsweredenaturedbyaddition of 1 vol of trichloroaceticacid (10% vol/vol). The trichloroacetic acidwas
ex-tracted withfreon-octylamineand the supernatantswerecollected and stored at -80'C before analysis. Nucleotides were quantitated by HPLC by amodification of the methodofChenetal. (15). Briefly, nucleotideswereinjectedonto aPartisil-10SAX column(Whatman Inc.,Clifton,NJ), isocratic elution with 0.007 MNH4H2PO4,pH4.0, followedbyalinear gradientto0.25 MNH4H2PO4, pH 5,formedover
30 minat aflowrateof 2 ml/min. Absorbancewasmonitoredat250 and 260nmandconcentrationwascalculatedbycomparisonto
stan-dards.Preliminary studies showed that 85% of addedAICARibotide was recoveredusing this technique.
Adenosine determination. Aliquots of pouch exudates were added
to asimilar volume oftrichloroaceticacid ( 10%vol/vol),followedby extraction of theorganic acid,asdescribed above. The adenosine
con-centration ofthe supernatantswasdetermined by reverse-phase HPLC,
as wehavepreviouslydescribed( 14).Briefly, sampleswereappliedto a
CI8gBondapackcolumn (Waters Chromatography Div., Milford, MA) and eluted witha0-40%-linear gradient(formedover60min)of 0.01 M ammoniumphosphate (pH 5.5) andmethanol, with a 1.5 ml/minflowrate.Adenosinewasidentified by retention time and the characteristic UV ratio of absorbanceat250/260,and the
concentra-tionwascalculatedbycomparisontostandards. Insomeexperiments theadenosine peakwasdigested bytreatmentwith adenosine
deami-nase(0.15 IU/ml, 30 min at 37°C)toconfirm that the peakso identi-fied containedonly adenosine (16). Preliminarystudies demonstrated that 90% of added adenosinewasrecoveredusingthistechnique.
Digitizationofchromatograms. Chromatograms were digitized us-ingaHewlett-Packard(Palo Alto, CA) Scanjet apparatus and the
re-sultingimageswereenhancedforcontrastandbrightnessusing PHO-TOSTYLER softwarerunon aZenith 386 personal computer.
Statisticalanalysis. The data wereanalyzed by the appropriate levelof ANOVAperformedby EXCEL 4.0software (Microsoft,Inc., Bothell,WA).
Low-dose weeklymethotrexate markedly inhibits leukocyte ac-cumulation inthe airpouch in response to carrageenan. Low-dose weekly methotrexate isa potent form of antiinflamma-torytherapy in patients suffering fromrheumatoid arthritis. To
confirm that the murine airpouch model of inflammation was a reasonable model in which to study the antiinflammatory
effects ofmethotrexate, we determined the effect of various dosesoflow-doseweekly methotrexate (administered
intraperi-toneally) on accumulation ofleukocytes in the murine air
pouch after injection ofcarrageenan. Methotrexate
dimin-ished, inadose-dependentmanner, the number of leukocytes that accumulated in carrageenan-treated air pouches by as
muchas60% (IC50=0.08 mg/kg perwk,P < 0.0002 vs. saline-treatedanimals; Fig.2). Moreover, the doses of methotrexate
requiredtoachieveamaximalantiinflammatoryeffect in this
animalmodel aresimilarto those required for the treatment of
rheumatoid arthritis(the equivalent of 10-15 mg/wk in a
metho-6- Figure 2. Weekly injec-tionof low-dose metho-trexateis antiinflamma-tory in the airpouch model. Methotrexate wasgivento the mice
(2 \by intraperitoneal
injec-tionattheindicated dosesfor 3 to4wk
dur-2 inginduction of the air
.05 .1 5
[Methotrexatel (mg\kg\wk) wasinjected with
carra-geenan(2% wt/vol), exudateswere harvested4hlater, and the cells were counted. Each point represents themean(±SEM)of cell counts from three mice. Analysis of variance demonstrates that the exudate cell count varied significantly with the dose of methotrexate (P<0.0002).
trexate-mediated inhibition ofleukocyte accumulation was
similarevenwheninflammation wasinducedupto 6 dafter the lastdoseof methotrexate (datanotshown).
Low-dose weekly methotrexate treatment increases intra-cellular concentrations ofAICAR. We have proposed that the antiinflammatory actions ofmethotrexate result, bothdirectly
andindirectly, from theinhibition ofAICARtransformylase
(4).If thismechanism iscorrect,specific inhibitionofAICAR
transformylaseshould result inhigher intracellular
concentra-tions ofAICAR. Wedirectlytested thevalidity of this
hypothe-sis byexamining AICAR concentrations in splenocytes from
saline- andmethotrexate-treated mice (0.5 mg/kg by weekly
intraperitoneal injection for4 wk) by HPLC. We found that
splenocytes from micetreatedwithmethotrexatecontained
sig-nificantlymoreAICARthan those treatedwith saline (Table I,
Fig. 3). These results are consistent with the hypothesis that
Low-dose weekly methotrexate treatment increases
inflammatoryexudates. We have previ-ouslyshown thattreatmentofcells inculture witheither meth-otrexate orAICARibonucleoside (acadesine),a
nonphosphor-ylated, cell-soluble precursorofAICAR, promotes release of
adenosine into thesupernate and thatadenosinerelease was
TableLMethotrexate (0.5mg/kgperwk) Treatment IncreasesIntracellular AICAR and Extracellular Adenosine
Exudate adenosine AICARconcentration concentration Condition (pmol/106splenocytes±SEM) (1M, ±SEM)
Control 26.5±10 0.57±0.09
(0.5 mg/kg per wk) 72.4±16* 1. 1 1±0.1 9t
Miceweretreatedwith aweeklyintraperitoneal injectionof sterile salineormethotrexate for4wkduringwhich timeanairpouchwas
inducedonthebacks of thesemice,asdescribed. After4wkthe air poucheswereinjectedwith carrageenan(2%wt/vol),thesplenocyte lysates andinflammatoryexudateswerecollected andanalyzed by HPLC,asdescribed. *P<0.02vs.control, Student'st test. *P
Figure3. Intracellular concentration of AICARishigherin
spleno-cytesfrom animalstreatedwith methotrexate(0.5 mg/kgperwk).
Miceweretreated with methotrexate for4wkduring induction of
the airpouch.Afterthe animalswerekilled,and-the airpouch
exu-datewasharvested,thespleenswerecollected,and the cellswere
col-lected. The cellnumberwasadjusted, the cellswerelysed, and the
AICAR-concentration analyzed by reverse-phase, ion exchange HPLC and detectedatOD260.Shown isarepresentative
chromato-gramof six ofsplenocyteAICAR from(A)acontrol(22.6 pmol/106
enhanced under conditions of "stress" (4). To determine whether low-dose weekly methotrexate treatment also
pro-motesadenosinereleaseinvivowequantitated the adenosine
concentration in inflammatory exudates taken from air pouches in saline- and methotrexate-treated (0.5 mg/kgper
wk) mice. We found that methotrexatetreatmentledtoa
signif-icantly higher adenosine concentration in the pouch exudate (Table I, Fig. 4). Thus, low-dose, intermittent methotrexate
therapypromotesadenosine releaseataninflamed site.
Adenosine mediates the antiinflammatory effect
ofmetho-trexateinthe airpouch. Todetermine whether the methotrex-ate-induced increase in adenosine concentration observed in pouch fluid exudateswasrelatedtothe antiinflammatory ef-fects ofmethotrexate,westudied the effect of adenosine
deami-nase onleukocyte accumulationinmethotrexate-treatedmice. Adenosine deaminase irreversibly deaminates extracellular adenosine to its inactive metabolite, inosine, and thereby renders it inactiveat adenosine receptors. Adenosine
deami-nase(0.15 IU/ml)did notsignificantly affect the number of
leukocytesrecovered from pouchesof saline-treated animals, but partially reversed the antiinflammatory effect of metho-trexate treatment(Fig. 5). Histologic examination oftheair
pouch tissuerevealed that,similarto its effectson leukocyte
Control ADA DMPX MTX MTX+ MTX+
18 9 9 18 9 8
I5 20 25
Figure 4. Theconcentrationofadenosine ishigherinexudates of
mice treated with methotrexate (0.5 mg/kgperwk). Micewere
treated with methotrexate for4 wkduringinductionof the airpouch. Aftertheanimalswerekilled theairpouchexudatewasharvested
and soluble adenosinewasextracted aftertreatmentof the exudates
with10%trichloroacetic acid.Theadenosineconcentration of
OD260.Shown isarepresentative chromatogramof
16 of pouchexudateadenosinefrom(A)acontrol(0.39,uM)and
filtration into thepouch tissue (38±2 vs. 106±14 cells/ 160X
field, methotrexatevs.control,P<0.01),andadenosine deam-inase reversed the antiinflammatory effect of methotrexate (88±3 cells/ 160x field,P<0.01 vs.methotrexatealone)
with-outaffecting leukocyte infiltrationincontrol mice(91±6 cells/ 160X field, P=NSvs.control,Fig. 6). Adenosine
deaminase-mediated reversal of the antiinflammatory effect of
methotrex-atetreatmentwasspecific since adenosine deaminase didnot
reverse the antiinflammatory effects of dexamethasone ( 1.5
mg/kg, injected intraperitoneally 1 h beforeinjection of the pouch with carrageenan, Fig. 7). Moreover, conversion of adenosinetoinactive metaboliteswasresponsible for reversal
of the antiinflammatory effect since adenosine deaminase whichwasinactivated by prior incubation with its tight-bind-ing, irreversible inhibitor deoxycoformycin (1 MM), did not affect the antiinflammatory capacity of methotrexate treat-ment(datanotshown). We conclude from these experiments that the increase inextracellular adenosineinthe methotrex-ate-treatedanimals isresponsible,atleastinpart,for the antiin-flammatory effects of methotrexate.
Theantiinflammatory effect of adenosineis mediatedvia
adenosineA2receptors.Thereareatleasttwomajor subtypes of adenosinereceptor,
A,andA2, that canbedifferentiated, in
part,onthebasis ofagonist and antagonist specificity ( 17, 18). Since extracellular adenosineappearedtomediatethe antiin-flammatory effects of methotrexate, wesought to determine whether the antiinflammatory actions of adenosinewere me-diatedbyoccupancyofaspecific adenosinereceptor.We
there-Figure5. Adenosine deaminase (ADA,0.15 lU/ml)and DMPX
(mg/kg)reversetheantiinflammatoryeffects of methotrexate treat-ment(0.5 mg/kgperwk). Miceweretreated with saline(control)
ormethotrexate for 3to 4 wk beforeinflammationwasinduced in
the airpouch. Shownarethemeans(±SEM)of the numberof cells
that accumulatedinthepouchexudates. Methotrexatesignificantly
inhibited the accumulation ofleukocytesinthepouchexudate
(4.0±0.4vs. 1.5±0.1 X 106cells/pouch,controlvs.methotrexate,P
<3X 10-6). NeitherADA(4.8±0.5X 106cells/pouch)norDMPX
(4.8±0.4X 106cells/pouch)significantlyaffectedthenumber of cells
that accumulatedinthe controlairpouches,but both ADA(2.3±0.8
X 106cells/pouch) andDMPX(3.8±0.5X 106cells/pouch)
signifi-cantlyreversed theantiinflammatoryeffectof methotrexate(P
<0.006 and P<0.001 vs.methotrexatealone,respectively).
fore injected receptor-specific adenosinereceptorantagonists into the airpouch with the inflammatory stimulus. The adeno-sine
A,receptor antagonist 8-cyclopentyl-dipropylxanthine (0.2 mg/kg) didnotaffect leukocyte accumulationintheair pouch in either control animals ormethotrexate-treated
ani-mals (Fig. 8). Because of its poorsolubility in aqueous
me-dium, higher concentrations of 8-cyclopentyl-dipropylxan-thine could not be utilized forstudy. In contrast, a specific
adenosine A2 receptor antagonist, 3,7-dimethyl-1-propargyl-xanthine (DMPX), completely reversed the antiinflammatory effect of methotrexatetreatment(IC50= 0.2 mg/kg, P<0.01;
Fig. 9) without affecting accumulation of leukocytes in either control animals (Fig. 5) or dexamethasone-treated animals
(Fig. 7). We conclude from these experiments that the in-creased adenosine found at inflamed sites in methotrexate-treated animalsmediates theantiinflammatory effects of meth-otrexatebyengaging adenosine A2receptors.
The results of the experiments reported here provide the first in vivo demonstration ofamolecular mechanism for the
antiphlo-gistic actions of methotrexate. Methotrexate, either acting di-rectlyorbypromoting the intracellular accumulation of
dihy-drofolatepolyglutamate, increases intracellularcontentof Al-CAR. The increase in intracellular AICAR concentration is associated with(and probably leads to)anincrease in
extracel-lular adenosine ininflammatory exudates. The increase in lo-cal adenosine concentrations at sites of inflammation sup-pressesinflammationviaoccupancyof adenosine A2receptors
Figure 6. Adenosine deaminase (ADA, 0.15 IU/ml)reversestheantiinflammatoryeffects of methotrexatetreatment(0.5 mg/kgperwk).Mice weretreated with saline (control) or methotrexate for 3wkbefore inflammationwasinduced in the airpouch. The air pouchesweredissected
outof theanimals,andfixed andprepared by standardhistopathological techniquesforphotomicroscopy.Thephotographic imageswere
digi-tizeddirectly using JAVA software and theimagesshownwereadjusted onlyforbrightnessandcontrast.Shownarerepresentativefields(of10 examined) fromonesection fromoneoftwoanimals studied under eachcondition.
Theobservation that low-dose weekly methotrexate ther-apy promotes the intracellular accumulation of AICAR in
splenocytes indicatesthat the"folate antagonism" oflow-dose weekly methotrexate is highlyspecific. Viainhibition of
dihy-drofolate reductase, high concentrations ofmethotrexate di-minishthe cellular contentofthe methyl donors required for
synthesis of purines and pyrimidines(6). In addition to the
synthesis of formyl-AICAR from AICAR (Fig. 1),reduced
fo-lateis required forthesynthesis ofa-N-formylglycinamide
3-glycinamideribonucleotide, precursors of
AICAR. Thus, under the conditions studied, if
inhibited folate-dependent reactions nonspecifically, then we would have expected either no change or a decrease in cellular AICAR content. We found the opposite, a net increase in
cellu-lar AICARcontent, an observation thatindicates that
selec-tive inhibition of AICAR
productionof AICAR. The
selective effect oflowconcentrations of methotrexateon pur-ine biosynthesis mostlikely follows from themetabolism of methotrexatetoits
directlyinhibits several steps in the
synthesisand metabolism of purines and
pyrimidines (5, 7-9). In
particular, polyglutamatedmetho-trexateis apotent direct inhibitor ofAICAR
(7). Moreover, inhibition of
bymeth-otrexate (and methotrexate
polyglutamate)leadstothe intra-cellular accumulation of
known and potentinhibitor of AICAR
2-Control Dex Dex Dex
(1.5mg\kg) +ADA +DMPX
Figure 7. Neither adenosine deaminase(ADA,0.15lU/ml)nor
dexametha-sone treatment(1.5mg/kg).Airpoucheswereinducedonmice for 3wk. 1hbeforeinjection ofcarrageenaninto the airpouch,themice receivedanintraperitonealinjectionof dexamethasone(1.5mg/kg)
orsaline. The exudateswereharvested 4h afterinjectionof
carra-geenanand the cell numberwasquantitated.Dexamethasone signifi-cantly diminished the number of cells that accumulated in theair pouch and neitherADAnorDMPXsignificantlyaltered the number of cellsthataccumulated in the airpouchof animals treated with dexamethasone.
mates(7)arerequiredtoinhibit AICARtransformylase, it is morelikelythatdihydrofolate polyglutamates areresponsible
for the intracellular accumulation of AICAR. Nonetheless, treatment with methotrexate may leadtoinhibition ofAICAR
transformylase (andaccumulationofAICAR) bytwodifferent butcomplementarymechanisms.
accu-mulation of AICAR increases adenosine release from some, butnotall,cell types(11).Barankiewiczetal. have shownthat treatmentofB-lymphoblastswithhighconcentrationsof Al-CARibonucleoside diminishes adenosine uptakeand utiliza-tion,resultinginincreased releaseof adenosine into the
degrada-tion(11,19).In contrast to T
lymphoblasts,which release little
Control DPCPX MTX MTX+
Figure 8. 8-Cyclopentyl-dipropylxanthine (DPCPX, 0.2 mg/kg) does
notreversetheantiinflammatory effect ofmethotrexate (0.5 mg/kg perwk). Miceweretreatedwith methotrexate for3 to 4 wkbefore
inflammationwasinduced in the air pouch. Shownare themeans
(±SEM) of thenumberofcells that accumulated in the pouch
exu-dates fromsix miceinthepresenceof the indicatedconcentrations
-- 0 4)
.05 .1 .5
Figure9. DMPX(mg/kg)reversestheantiinflammatoryeffect of methotrexate(0.5 mg/kgperwk). Miceweretreatedwith
metho-trexatefor 3to4wk beforeinflammationwasinduced in the air pouch. Shownarethemeans(±SEM) of the number of cells that
ac-cumulated in thepouch exudates from three mice in thepresenceof theindicated concentrations of DMPX. Analysis of variance
demon-stratesthat the numberof cells in the pouch exudate varies signifi-cantly with thedoseof DMPX (P<0.01).
adenosine under any condition, B lymphoblasts possess in-creased AMP-S'-nucleotidase activity (adenosine formation) and relatively littleadenosine kinase or adenosinedeaminase activity (adenosineutilization[ 1]). Thus, Barankiewiczetal. postulated that, since AICARibonucleoside does not affect adenosine production or transport, intracellular accumula-tions of AICAR must inhibit adenosine kinase or adenosine deaminaseactivityinordertopromotetheincreasein extracel-lularadenosine observed (11, 19). AICARibonucleoside may also lead to anincreaseinadenosine releaseatsites of "stress," such asreperfusion after ischemic insult totheheart, andthe increased extracellular adenosine that accumulates in ischemic tissueprotectstheaffected tissue from leukocyte-mediated in-jury (10). Ourdata suggest that treatment invivo with low-dose methotrexate similarly increases intracellular AICAR contentand,moreimportantly, promotes adenosine releaseat inflamed sites.
Methotrexateinduced increasedadenosine concentrations ininflammatory exudates and was a potent antiinflammatory agentinthe air pouch model. To prove that the effects of meth-otrexate on purine metabolism and inflammation were causally related, we used two different approaches: elimination ofextracellular adenosine by adenosine deaminase and antago-nism of adenosine at its receptors with a specific antagonist (DMPX). Bothof these experimental maneuvers reversed the
antiinflammatory effect of methotrexate but did not reverse the antiinflammatory effect of dexamethasone in this same model. Dexamethasone is a potent agonist at glucocorticoid receptors thatdiminishes leukocyte accumulation at inflamma-tory sitesbyamechanism that is not related to purine metabo-lism(forreview see reference 20). Thus, our observation that bothspecific eliminationandantagonism of adenosine reverse theantiinflammatoryeffectsofmethotrexateis strong evidence that adenosine mediates the antiphlogistic effect of metho-trexate.
Wehavepreviously observed that methotrexate treatment, in vitro, promotes an increase in adenosine release at the ex-pense ofhypoxanthine and inosine release (4). In this study we were unable to detect inosine in most samples and the HPLC
othercompoundspresent inthesecomplexbiologic fluids.
Nev-ertheless, the adenosine concentration present in inflamma-tory exudates ofmethotrexate-treated animals ( 1. 1 ,uM) is more thansufficientto account for thediminished inflamma-tionobserved; maximal inhibition ofstimulated neutrophil ad-hesion and generation of superoxide anion and H202 is achieved with adenosine concentrations greater than or equal to 1 juM ( 14, 21). Indeed, the concentration of adenosine foundin exudatesfromcontrolanimalswas less than the
con-centrationofadenosine foundintransudatesfrom "stressed"
isolated rabbithearts (during hypoxia, 1225±300 nM;
refer-ence22). Although the adenosine concentration measured in the inflammatory exudate probably reflects the metabolic
changes in methotrexate-treated animals and is sufficient to
inhibittheproduction of toxicoxygenmetabolitesby the cells present inthe inflammatoryexudate, it islikely that the in-crease in extracellular adenosine responsible for diminished inflammation isthatwhichoccursinthesurrounding tissues,a lessreadily accessible site forsampling.
There are atleast twomajorsubclassesof adenosine recep-torthat can bedistinguished on pharmacologic grounds,
and A2 ( 17, 18). Adenosine
A,receptors arerelatively high-af-finityreceptorsthat arelinked to pertussis toxin-inhibitedG
proteins (23-33). Adenosine
A,receptors have been demon-strated on neutrophils and macrophages (but notperipheral
blood mononuclear cells) where theymediate,whenoccupied, enhanced chemotaxis and phagocytosis of
immunoglobulin-coated particles (34-38). AdenosineA2 receptors are low-af-finityreceptors linked toGassignal transduction proteinsin many cell types.AdenosineA2 receptors arepresent on
occu-pied, generally suppress the
inflammatoryor immune func-tions ofthese cells
(forreview see references 39-41).
relatively selective antagonistswe found that the
antiinflam-matory effects of adenosine in methotrexate-treated animals
weremediatedbyoccupancyof adenosineA2 receptors,results
that wereidentical tothoseobtained byAsakoetal.(12).In contrast,Schrieretal.(42)observed,
utilizingreceptor-specific agonists,that occupancyof adenosine
A2 receptors isantiinflammatoryin a rat modelof inflamma-tion.Thediscrepancymay be due tospecies differencesin
ago-nistsensitivityoradenosinereceptorexpression. Alternatively, the apparent difference in receptor
antiin-flammatoryeffects of adenosine results fromadifference inthe
Wefirst suggestedthatadenosinemightbe anendogenous
antiinflammatoryagent whenwe observed thatadenosine
in-hibitsthegeneration of toxicoxygenmetabolites by stimulated neutrophils ( 14). In subsequent studieswe have shownthat
adenosine,both addedexogenouslyorreleasedendogenously, diminishesendothelial cell
injurymediated by stimulated
neu-trophils (43). The
cytoprotectiveeffects of adenosine result
from inhibition ofthegeneration of toxicoxygen metabolites
endothelium (43).Inthe modelunderstudy,the apparent ef-fect of adenosine wastodiminish extravasation of leukocytes into an
inflammatoryexudate. There may be an additional beneficial effectof methotrexate therapy for the
methotrexate;theconcentration of adenosinepresent in the
inflammatorypouchexudatesismore than sufficient to inhibit generation of toxic oxygen
metabo-lites by stimulated leukocytes. Thus, treatment with metho-trexatemay bothdiminish the number ofleukocytes that accu-mulate in aninflammatoryexudate and inhibitthe destructive capacity of thoseleukocytesthatdo arrive at theinflamed site. Inthe model studied here,inflammation was acute and was characterized by the accumulation of a neutrophilic infiltrate in both the air pouch and the surrounding tissues. Although it
islikelythat theadenosine released is acting directlyon neutro-phil adenosine receptors, it is also possible that adenosine in-hibits the generation of cytokines or chemoattractants required for accumulation of the inflammatory exudate. Indeed,
adeno-sine,probably actingat an A2 receptor, inhibitssynthesis of
cytokines (TNFa) and other inflammatory proteins (comple-ment C2) by macrophages (44, 45). Moreover, adenosine,
act-ing at its receptor, inhibits lymphocyte proliferation and
in-duces suppressoractivityin culturedlymphocytes(39).Thus, theantiinflammatory effects ofmethotrexate(actingvia
adeno-sine)aremoregeneral thanthosestudiedinthis model ofacute
inflammation.Indeed,itis likelythat theeffectsof methotrex-ate, acting via adenosine, onlymphocyteormonocyte function play a greater roleindiminishingthechronicinflammation of rheumatoid arthritisthan the effects on acute inflammation observed inthis model.
We have demonstrated a novelbiochemicalmechanism of
action of methotrexate. Low-doseweekly methotrexate ther-apy leads tointracellular accumulation of AICAR, which pro-motesincreasedadenosinerelease (and/ordiminished
adeno-sine uptake)atsitesofinflammation. This increasein extracel-lular adenosine diminishes both the accumulation and
function of leukocytesininflamed sites.Thesefindingssuggest several novel approaches to the development ofnew agents that inhibitinflammation by increasing adenosinerelease: de-velopmentofdirectinhibitors ofAICAR transformylase, inhib-itorsofadenosine deaminaseandadenosinekinase,and
adeno-sine uptake inhibitors.
Wewould liketo thankCampbell Bunce for his excellenttechnical assistance with this project. Wewouldalsoliketothank Dr. Gerald Weissmann for his discussions and advice and for reviewing this
This workwasmadepossible bygrantsfrom the Arthritis
Founda-tion,New Yorkchapter, The AmericanHeartAssociation,NewYork affiliate,and theNational Institutes of Health (AR-1 1949, HL-19721 ) and with assistance from the General Clinical Research Center (MOIRR00096) and the Kaplan Comprehensive Cancer Center (CA16087).
1.Furst,D.E., andJ. M. Kremer.1988. Methotrexate in rheumatoid arthritis. Arthritis Rheum. 31:305-314.
2.Sperling,R.I.,J.S.Coblyn,J. K.Larkin,A.I. Benincaso,K. F.Austen, and
M. E.Weinblatt.1990.Inhibition of leukotrieneB4synthesis inneutrophilsfrom
patientswithrheumatoid arthritisbyasingleoral doseof methotrexate. Arthritis Rheum. 33:1149-1155.
3.Nesher,G.,and T. M. Moore.1990. The in vitro effects of methotrexateon
peripheralblood mononuclear cells:modulationbymethyldonorsand
spermi-dine. Arthritis Rheum. 33:954-959.
4. Cronstein, B.N., M. A. Eberle,H. E. Gruber,andR. I. Levin. 1991.
Methotrexateinhibitsneutrophilfunctionbystimulatingadenosinereleasefrom connective tissuecells.Proc.Nati. Acad.Sci.USA. 88:2441-2445.
5.Zimmerman,C. L.,T.J.Franz,and J. T.Slattery.1984.Pharmacokinetics
6.Chabner,B.A., C.J.Allegra, G.A.Curt,N. J.Clendeninn,J.Baram,S.
Koizumi,J.C.Drake, and J. Jolivet. 1985. Polyglutamation of methotrexate. Is methotrexate a prodrug? J. Clin. Invest. 76:907-912.
7.Allegra, C. J., J. C. Drake, J. Jolivet, and B. A. Chabner. 1985. Inhibition of phosphoribosylaminoimidazolecarboxamide transformylase by methotrexate anddihydrofolic acid polyglutamates. Proc. Natl. Acad. Sci. USA. 82:4881-4885. 8.Allegra, C. J.,K. Hoang, G. C. Yeh, J. C.Drake, and J.Baram. 1987. Evidence for direct inhibition of de novo purine synthesis in human MCF-7 breast cells as aprincipalmode ofmetabolic inhibitionby methotrexate. J.Biol. Chem. 262:13520-13526.
9. Baggott, J. E., W. H.Vaughn, and B. B. Hudson. 1986. Inhibition of 5-aminoimidazole-4-carboxamide ribotide transformylase, adenosine deami-nase, and5'-adenylate deaminase by polyglutamates of methotrexate and oxi-dized folates and by 5-aminoimidazole-4-carboxamide riboside and ribotide.
10. Gruber, H. E., M. E. Hoffer, D. R. McAllister, P. K. Laikind, T. A. Lane, G.W.Schmid-Schoenbein,and R. L.Engler. 1989. Increased adenosine
concen-tration in bloodfrom ischemicmyocardiumby AICA riboside: effects on flow, granulocytes, andinjury. Circulation 80:1400-1411.
1 1. Barankiewicz, J., G. Ronlov, R. Jimenez, and H. E. Gruber. 1990. Selec-tiveadenosine release from humanBbutnot Tlymphoid cell line. J. Biol. Chem. 265:15738-15743.
12. Asako, H., R. E.Wolf, and D. N. Granger. 1993. Leukocyte adherence in ratmesenteric venules: effects of adenosine and methotrexate.Gastroenterology.
13.Zurier,R.B., S.Hoffstein,and G. Weissmann. 1973.Suppression ofacute andchronic inflammation inadrenalectomizedratsbypharmacologicamounts
of prostaglandins. Arthritis Rheum. 16:606-618.
14.Cronstein, B. N., S. B. Kramer, G. Weissmann, and R. Hirschhorn. 1983. Adenosine:aphysiological modulator of superoxide anion generation by human neutrophils.J. Exp. Med. 158:1160-1177.
15.Chen, S. C., P. R. Brown, and D. M. Rosie. 1977. Extraction procedures for usepriortoHPLC nucleotideanalysis using microparticle chemically bonded
16. Hirschhorn, R., V.Roegner-Maniscalco, L. Kuritsky, and F. S. Rosen. 1981. Bone marrowtransplantationonlypartiallyrestorespurinemetabolites to normal inadenosinedeaminase-deficientpatients. J.Clin.Invest.68:1387-1393.
17.Londos,C.,D.M. F.Cooper, and J. Wolff. 1980. Subclasses of external adenosine receptors. Proc. Natl. Acad. Sci. USA. 77:2551-2554.
18. vanCalker, D., M. Muller, and B. Hamprecht. 1979. Adenosine regulates, via twodifferent types of receptors, the accumulation of cyclic AMP in cultured brain cells. J.Neurochem. 33:999-1005.
19.Barankiewicz, J.,R.Jimenez, G. Ronlov,M.Magill, andH. E.Gruber. 1990.Alteration ofpurinemetabolism by AICA-riboside inhuman B lympho-blasts.Arch. Biochem.Biophys.282:377-385.
20.Cronstein,B.N., S. C.Kimmel,R.I.Levin,F.Martiniuk,and G. Weiss-mann. 1992.Amechanism for theantiinflammatoryeffectsof corticosteroids:
theglucocorticoidreceptorregulates leukocyteadhesiontoendothelial cells and
expressionofELAM-1andICAM-1.Proc.Natl. Acad.Sci. USA. 89:9991-9996.
21.Cronstein,B.N.,E. D.Rosenstein,S.B.Kramer, G.Weissmann,andR.
Hirschhorn. 1985. Adenosine:aphysiologicmodulatorofsuperoxide anion gen-eration by humanneutrophils. AdenosineactsviaanA2 receptor on human
22.Matherne, G. P., J.P.Headrick, S. D. Coleman, and R. M. Berne. 1990. Interstitial transudatepurinesinnormoxic andhypoxicimmature and mature rabbit hearts.Pediatr.Res.28:348-353.
23.Dolphin,A.C., and S. A. Prestwich. 1985. Pertussis toxin reverses adeno-sine inhibition of neuronalglutamaterelease. Nature(Lond).316:148-150.
24.Ramkumar, V., and G. L.Stiles. 1988. Reciprocal modulation of agonist andantagonist bindingtoAladenosine receptors by guanine nucleotides is me-diated via a pertussis toxin-sensitive G protein. J. Pharmacol. Exp. Ther. 246:1194-1200.
25. Parsons,W.J.,V.Ramkumar, and G. L.Stiles. 1988. Isobutylmethylxan-thine stimulates adenylate cyclase by blockingtheinhibitory regulatory protein, Gi. Mol.Pharmacol.34:37-41.
26.Ramkumar, V.,and G.L.Stiles. 1988.Anovel site ofaction ofahigh affinityAl adenosinereceptorantagonist.Biochem. Biophys.Res. Commun. 153:939-944.
27.Monaco, L.,D. A.DeManno,M. W.Martin,and M. Conti. 1988.
Adeno-sineinhibition ofthe hormonalresponsein the Sertoli cell is reversedby pertussis toxin.Endocrinology. 122:2692-2698.
28.Trussell,L.O., andM. B. Jackson. 1987.Dependenceofan
adenosine-ac-tivated potassiumcurrent on aGTP-binding proteinin mammalian central
neu-rons.J. Neurosci. 7:3306-3316.
29.Arend,L.J., W. K.Sonnenburg,W.L.Smith, and W. S. Spielman. 1987.
A,and A2 adenosine receptors in rabbit corticalcollecting tubule cells.
Modula-tion of hormone-stimulated cAMP. J. Clin. Invest. 79:710-714.
30.Rossi,N.F.,P.C.Churchill, andM. C.Churchill. 1987.Pertussis toxin reverses adenosinereceptor-mediated inhibition of renin secretion in rat renal cortical slices. Life Sci. 40:481-487.
31. Parsons,W.J., and G.L.Stiles. 1987.Heterologous desensitization ofthe inhibitoryAladenosine receptor-adenylatecyclase system inratadipocytes.
Reg-ulation of both Ns and Ni. J. Biol. Chem. 262:841-847.
32. Berman, M. I., C. G. Thomas, Jr., and S. N. Nayfeh. 1986. Inhibition of thyrotropin-stimulated adenosine 3',5-monophosphate formationin ratthyroid cellsbyanadenosine analog. Evidence that the inhibition is mediated by the putative inhibitory guanine nucleotide regulatory protein. J. CyclicNucleotide ProteinPhosphorylationRes.11:99-111.
33. Garcia Sainz,J.A., andM. L. Torner.1985. Rat fat-cells have three types of adenosine receptors (Ra,Ri,andP).Differential effects ofpertussistoxin. Biochem. J. 232:439-443.
34. Rose, F.R., R. Hirschhorn, G. Weissmann, and B.N.Cronstein. 1988. Adenosine promotesneutrophil chemotaxis.J.Exp. Med. 167:1186-1194.
35.Salmon,J.E., andB. N.Cronstein. 1990. Fcgammareceptor-mediated
functions inneutrophilsaremodulatedby adenosine receptor occupancy: A,
receptorsarestimulatoryand A2 receptorsareinhibitory.J.Immunol. 145:2235-2240.
Wil-liams. 1990. Theadenosine/neutrophilparadox resolved. Humanneutrophils
possess bothA,and A2 receptors that promote chemotaxis and inhibit
02-genera-tion, respectively.J. Clin.Invest.85:1150-1157.
37.Eppell,B.A.,A.M.Newell, andE.J. Brown.1989.Adenosine receptors
areexpressedduringdifferentiation of monocytestomacrophagesinvitro. J.
38.Salmon,J.E., N. Brogle, C. Brownie, J. C. Edberg, R. P. Kimberly, B.-X.
Chen,andB. F.Erlanger.1993. Human mononuclear phagocytes express adeno-sine A, receptors:anovel mechanismfor differential regulation ofFcyreceptor function. J.Immunol.151:2765-2775.
39.Cronstein,B.N.,and R.Hirschhorn. 1990. Adenosineandhostdefense: modulation through metabolism and receptor-mediated mechanisms.In Adeno-sine and AdenoAdeno-sineReceptors.M.Williams,editor. The Humana Press,Clifton,
40.Priebe,T.S., and J.A.Nelson.1991.Adenosine and immune system function.InAdenosine and Adenine NucleotidesasRegulators ofCellular Func-tion. J. W.Phillis,editor. CRC Press,BocaRaton,FL. 141-154.
41.Cronstein,B. N. 1991. Purines andinflammation: Neutrophils possessP,
and P2 receptors.InAdenosine andAdenineNucleotidesasRegulators of Cellu-larFunction. J. W.Phillis,editor. CRC Press, Boca Raton, FL. 133-140.
42.Schrier,D.J., M. E. Lesch, C. D. Wright, and R. B. Gilbertsen. 1990. The
antiinflammatory effectsof adenosine receptor agonists on the carrageenan-in-ducedpleuralinflammatoryresponse inrats.J.Immunol.145:1874-1879.
43.Cronstein,B.N.,R. I.Levin,J.Belanoff,G.Weissmann,and R. Hirsch-horn. 1986. Adenosine:anendogenousinhibitor ofneutrophil-mediated injury
toendothelial cells. J. Clin. Invest. 78:760-770.
44.Parmely,M.J.,W.-W.Zhou, C.K.Edwards III,D. R.Borcherding, R.
Silverstein,andD.C.Morrison. 1993. Adenosine andarelatedcarbocyclic nu-cleoside analogueselectivelyinhibittumornecrosisfactor-alpha production and protect miceagainstendotoxin challenge. J.Immunol. 151:389-396.