Renal protection from ischemia mediated byA2A adenosine receptors on bone marrow–derived cells

Yuan-Ji Day, … , Joel Linden, Mark D. Okusa

J Clin Invest. 2003;112(6):883-891. https://doi.org/10.1172/JCI15483.

Activation of A2A adenosine receptors (A2ARs) protects kidneys from ischemia-reperfusioninjury (IRI). A2ARs are expressed on bone marrow–derived (BM-derived) cells and renalsmooth muscle, epithelial, and endothelial cells. To measure the contribution of A2ARs onBM-derived cells in suppressing renal IRI, we examined the effects of a selective agonist ofA2ARs, ATL146e, in chimeric mice in which BM was ablated by lethal radiation andreconstituted with donor BM cells derived from GFP, A2AR-KO, or WT mice to produceGFP®WT, A2A-KO®WT, or WT®WT mouse chimera. We found little or no repopulation ofrenal vascular endothelial cells by donor BM with or without renal IRI. ATL146e had noeffect on IRI in A2A-KO mice or A2A-KO®WT chimera, but reduced the rise in plasmacreatinine from IRI by 75% in WT mice and by 60% in WT®WT chimera. ATL146e reducedthe induction of IL-6, IL-1b, IL-1ra, and TGF-a mRNA in WT®WT mice but not in A2A-KO®WT mice. Plasma creatinine was significantly greater in A2A-KO than in WT mice afterIRI, suggesting some renal protection by endogenous adenosine. We conclude thatprotection from renal IRI by A2AR agonists or endogenous adenosine requires activation ofreceptors expressed on BM-derived cells.

Article Hematology

Find the latest version:

http://jci.me/15483-pdf

The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6 883

IntroductionBone marrow–derived (BM-derived) cells such as neu-trophils (1, 2) and T lymphocytes (3–5) are thought toplay prominent roles in the pathogenesis of ischemia-reperfusion injury (IRI). Activation of A2A adenosinereceptors (A2ARs) is known to reduce inflammation andhas been shown to produce tissue protection from IRI(6, 7). We previously demonstrated that 4-{3-[6-Amino-

9-(5-ethylcarbamoyl-3,4-dihydroxy-tetrahydro-furan-2-yl)-9H-purin-2-yl]-prop-2-ynyl}-cyclohexanecarboxylicacid methyl ester (ATL146e), a specific A2AR agonistwith 50 times higher affinity for A2ARs than the com-monly used A2A agonist CGS21680, attenuated plasmacreatinine levels after renal injury by 70–80% whenadministered before or immediately after the onset ofreperfusion and throughout the reperfusion period (8).The cellular target through which ATL146e mediatesrenal tissue protection is not known.

A2AR stimulation vasodilates the outer medullarydescending vasa recta (9), an effect that could reduceischemic damage. Inhibition of endothelial cell activationmay contribute to renal protection by A2AR agonists.ATL146e reduces the expression of the endothelial adhe-sion molecules P-selectin and ICAM-1 (10). BM-derivedcells such as neutrophils, macrophages, platelets, and Tlymphocytes (T cells) may also be the target of protectionby A2AR agonists. A2ARs are expressed on most hematopoi-etic cells, and activation of A2ARs on isolated cells has beenshown to have direct anti-inflammatory effects (11).

In this study, we investigated the contribution of A2ARsexpressed on BM-derived cells to renal tissue protectionfrom IRI. We created chimeric mice in which BM wasablated from WT mice and reconstituted with BM fromcongenic mice in which the A2AR gene had been deleted(A2A-KO). This method permits discrimination betweenA2A agonist effects on BM-derived cells and possible effects

Renal protection from ischemia mediated by A2A adenosinereceptors on bone marrow–derived cells

Yuan-Ji Day,1 Liping Huang,2 Marcia J. McDuffie,3 Diane L. Rosin,4 Hong Ye,2

Jiang-Fan Chen,5 Michael A. Schwarzschild,5 J. Stephen Fink,5 Joel Linden,1,2

and Mark D. Okusa2

1Department of Molecular Physiology and Biological Physics,2Department of Medicine,3Department of Microbiology, and4Department of Pharmacology, University of Virginia, Charlottesville, Virginia, USA5Molecular Neurobiology Lab, Department of Neurology, Massachusetts General Hospital and Harvard Medical School,Boston, Massachusetts, USA

Activation of A2A adenosine receptors (A2ARs) protects kidneys from ischemia-reperfusion injury (IRI).A2ARs are expressed on bone marrow–derived (BM-derived) cells and renal smooth muscle, epithelial,and endothelial cells. To measure the contribution of A2ARs on BM-derived cells in suppressing renal IRI,we examined the effects of a selective agonist of A2ARs, ATL146e, in chimeric mice in which BM was ablat-ed by lethal radiation and reconstituted with donor BM cells derived from GFP, A2AR-KO, or WT miceto produce GFP→WT, A2A-KO→WT, or WT→WT mouse chimera. We found little or no repopulationof renal vascular endothelial cells by donor BM with or without renal IRI. ATL146e had no effect on IRIin A2A-KO mice or A2A-KO→WT chimera, but reduced the rise in plasma creatinine from IRI by 75% inWT mice and by 60% in WT→WT chimera. ATL146e reduced the induction of IL-6, IL-1β, IL-1ra, andTGF-α mRNA in WT→WT mice but not in A2A-KO→WT mice. Plasma creatinine was significantlygreater in A2A-KO than in WT mice after IRI, suggesting some renal protection by endogenous adeno-sine. We conclude that protection from renal IRI by A2AR agonists or endogenous adenosine requiresactivation of receptors expressed on BM-derived cells.

J. Clin. Invest. 112:883–891 (2003). doi:10.1172/JCI200315483.

Received for publication March 19, 2002, and accepted in revised formJuly 8, 2003.

Address correspondence to: Mark D. Okusa, Division ofNephrology, Box 800133, University of Virginia Health System,Charlottesville, Virginia 22908, USA. Phone: (434) 924-2187; Fax: (434) 924-5848; E-mail: mdo7y@virginia.edu.Yuan-Ji Day’s present address is: Department of Anesthesiology,Chang Gung Memorial Hospital, Chang Gung University,Tauyuan,Taiwan.J. Stephen Fink is deceased.Conflict of interest: Mark D. Okusa, Joel Linden, and Marcia J.McDuffie have equity in Adenosine Therapeutics, which suppliedATL146e.Nonstandard abbreviations used: A2A adenosine receptors(A2ARs); ischemia-reperfusion injury (IRI); bone marrow (BM); 4-{3-[6-Amino-9-(5-ethylcarbamoyl-3,4-dihydroxy-tetrahydro-furan-2-yl)-9H-purin-2-yl]-prop-2-ynyl}-cyclohexanecarboxylic acidmethyl ester (ATL146e); periodate-lysine-paraformaldehyde (PLP);phosphodiesterase type IV (PDE 4); heart rate (HR); outermedulla (OM); macrophage inhibitory protein (MIP); IFN-γ–inducible protein 10 (IP-10); IL-1 receptor antagonist (IL-1ra);monocyte chemoattractant protein–1 (MCP-1).

884 The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6

on vascular or renal parenchymal cells. We also showed byusing BM from GFP mice that few if any renal endothelialor smooth muscle cells are derived from donor BM, withor without renal IRI. Our findings demonstrate that acti-vation of A2ARs on BM-derived cells is primarily responsi-ble for protection of the kidney from IRI. Furthermore,these studies provide evidence for a direct causal relationbetween activation of inflammatory cells and IRI.

MethodsGenotype and phenotype of congenic A2A-KO mice (C57BL/6).Mice with functionally disrupted A2ARs were initiallygenerated on a mixed genetic background after severalgenerations of inbreeding from (129/SvJae, A2AR–/– ×C57BL/6) F1 founder mice (12). These mice were typedwith a skeleton panel of 96 informative microsatelliteloci covering every chromosome (except Y) at an averagedistance of 15 centimorgan. A marker-assisted breedingstrategy was used to effectively move the mutantAdora2a allele onto the C57BL/6 background (99.3%homozygosity) by generation 3. The only detectable 129contamination surrounds the Adora2a locus, which wemapped during gene transfer onto chromosome 10, ina region of conserved synteny with the human locus.

DNA was obtained from the tail of A2A-KO mice andsubjected to a PCR genotyping strategy using the follow-ing oligonucleotides to amplify neomycin resistanceinserted into the A2A-KO gene: NEO-F, 5′-AGACAATCG-GCTGCTCTGAT; NEO-R, 5′-CAAGCTCTTCAGCAATAT-CACG; and a portion of the WT gene deleted from theA2A-KO construct, WT-F, 5′-AGCCAGGGGTTACATCTGTG,and WT-R, 5′-AGACAATCGGCTGCTCTGAT. PCR prod-ucts were separated by 1% agarose gel electrophoresis andvisualized by staining DNA with ethidium bromide. Tis-sue from WT C57BL/6 or A2A-KO mice was prepared forimmunohistochemistry, and A2AR immunoreactivity wasdetected as described previously (13).

Generation of chimeric mice. C57BL/6 male mice (HilltopLab Animals, Scottsdale, Pennsylvania, USA), congenicA2A-KO mice, and GFP mice (Jackson Laboratory, Bar Har-bor, Maine, USA) were used for BM transplantation.Donor mice (12 weeks of age, 25–28 g) were anesthetizedwith Nembutal (0.02 mg/g) and sacrificed by cervical dis-location. The marrow from the tibia and femur was har-vested under sterile conditions. Bones were flushed withRPMI (Life Technologies, Grand Island, New York, USA)plus 10% FCS. The marrow was passed sequentiallythrough a 22-gauge needle followed by three passagesthrough a 25-gauge needle in order to obtain single-cellsuspensions of BM cells. Cells were washed and resus-pended, and viable cells were counted. The yield wasapproximately 50 × 106 nucleated BM cells per mouse.Recipient mice (8–10 weeks of age, 22–25 g) were lethallyirradiated with two exposures to 6 Gy 4 hours apart.Immediately after irradiation, 3 × 106 BM cells were inject-ed through the tail vein. The resulting chimeric mice werehoused in microisolators for 6–8 weeks before experi-mentation and fed autoclaved food and water containing5 mM sulfamethoxazole and 0.86 mM trimethoprim.

To evaluate the efficiency of reconstitution, a mutatedmouse strain, B6.SJL-Ptprca Pep3b/BoyJ, was used as thesource of donor BM. The CD45.1 epitope, absent in cellsof recipient mice, was detected by immunofluorescentcell sorting 6 weeks after BM transplantation. PBMCsfrom transplanted recipients were purified from 200 µlof tail vein blood using the Ficoll (density, 1.077) gradi-ent method with centrifugation at 500 g for 10 minutes.Harvested PBMCs were placed in 1.5 mM NH4Cl to lysered blood cells and then resuspended into 200 µl of RPMIwith 5% fetal calf serum. After blocking nonspecific Fcbinding with anti-mouse CD16/CD32 (BD Pharmingen,San Diego, California, USA), PBMCs were incubated withR-phycoerythrin conjugated anti-mouse CD45.1 mono-clonal antibody (BD Pharmingen) and either allophyco-cyanin-conjugated rat anti-mouse CD11b (Mac-1 β chain)monoclonal antibody (BD Pharmingen), fluoresceinisothiocyanate–conjugated rat anti-mouse CD8a mono-clonal antibody (BD Pharmingen), or allophycocyanin-conjugated rat anti-mouse CD4 monoclonal antibody(BD Pharmingen) on ice for 30 minutes. CD11b-, CD8a-, and CD4-positive cells were sorted by flow cytometry,and the percentage of cells expressing CD45.1 was deter-mined in each population of cells (FACScan, CellQuest,Becton Dickinson, San Jose, California, USA).

GFP mouse chimera and anti-GFP immunohistochemistry.GFP chimeras were created by transplanting BM fromGFP mice into lethally irradiated C57Bl/6 controls(GFP→WT) as described above. Immunohistochemistrywas performed as described previously (14). Spleens orkidneys were harvested, fixed in 4% periodate-lysine-paraformaldehyde (PLP) and embedded in either paraf-fin (spleens) or epoxy resin (kidneys). Spleens and kidneyswere cut to yield 4 and 0.5 µm sections, respectively.Epoxy resin was then removed (15). GFP was detectedimmunohistochemically, since direct detection of GFPfluorescence was found to be unreliable because of non-specific green fluorescence in WT animals, particularly intissues subjected to IRI. Sections were incubated with rab-bit anti-GFP antibody (1:1000; Novus Biologicals Inc., Lit-tleton, Colorado, USA) followed by goat anti-rat rabbitsecondary antibody conjugated to FITC secondary (Vec-tor Laboratories, Burlingame, California, USA). In somecases, kidney sections were subjected to dual labeling byincubating with rabbit anti-GFP antibody and mouse-anti Mac-2 (1:500; Accurate Chemicals, Westbury, NewJersey, USA) followed by goat anti-rabbit secondary anti-body conjugated to FITC (1:1000) and goat anti-mousesecondary antibody conjugated to rhodamine (1:1000).Stained cells or tissue sections were cover slipped with anaqueous-based mounting solution consisting of p-phe-nylenediamine (1 mg/ml) and 70% glycerol. Sections wereviewed under a Zeiss AxioSkop photofluorescence micro-scope. Photographs were taken with a SPOT RT Camera(software version 3.3, Diagnostic Instruments Inc., Ster-ling Heights, Michigan, USA) and imported into AdobePhotoshop (3.0) for brightness/contrast adjustment.

Compound administration. Mice were anesthetized withvaporized halothane (Halothan Vapor 19.1, North

The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6 885

American Drager, Telford, Pennsylvania, USA) beforesubcutaneous implantation of osmotic minipumps(model 1003D, ALZA Corp., Palo Alto, California, USA).The pumps released either vehicle (0.01% DMSO inPBS), ATL146e (10 ng/kg per minute) or rolipram (0.1ng/kg per minute), a type IV phosphodiesterase (PDE 4)inhibitor. Doses of ATL146e and rolipram were chosenon the basis of previous dose-ranging studies to pro-duce a 60–70% reduction of plasma creatinine levels 24hours after renal IRI (8). Administration of vehicle orcompounds began 5 hours before ischemia and contin-ued throughout the 24-hour reperfusion period.

Procedure for producing renal IRI. Mice were anesthetizedwith ketamine (100 mg/kg intraperitoneally), xylazine(10 mg/kg intraperitoneally), and acepromazine (1mg/kg intramuscularly) and subjected to bilateral flankincisions as previously described (6). Both renal pedicleswere cross-clamped for 32 minutes. Surgical wounds wereclosed with metal staples, and mice were returned tocages for 24 hours. After 24 hours of reperfusion, animalswere reanesthetized, blood was obtained by cardiac punc-ture, and kidneys were removed for various analyses.

Plasma and urine chemistry. Plasma creatinine concen-trations were determined using a colorimetric assayaccording to the manufacturer’s protocol (Sigma-

Aldrich, St. Louis, Missouri, USA). Urine sodium andpotassium concentrations were measured by an auto-mated autoanalyzer (Massachusetts General HospitalClinical Laboratory, Boston, Massachusetts, USA).

Tail cuff blood pressure measurement. Mean arterial pres-sure was measured by using a photoelectric sensor forpulse detection in mouse tail (IITC Model 179, IITCInc./Life Science Instruments, Woodland Hills, Cali-fornia, USA). Mice were allowed to rest for 10 minutesin a chamber with the temperature controlled at 26°C.Blood pressures were measured twice and averaged.

Leukocyte isolation from spleen. Spleens were removedfrom mice, gently dispersed, and filtered throughgauze. Leukocytes were isolated by Ficoll density cen-trifugation, counted using a hemocytometer, andsubjected to flow cytometry.

Multiprobe RNase protection assays. Total kidney RNA wasextracted from homogenized tissue with RNAzol B (LeedoMedical Laboratories, Houston, Texas, USA) and analyzedby 1.5% agarose gel electrophoresis to assess the integrityof RNA before solution hybridization. Cytokine mRNAexpression was assessed by the RiboQuant multiprobeRNase protection system (BD Pharmingen) according tothe manufacturer’s protocol. In brief, mRNA-specificRNA probes were labeled with 32[P]UTP using multiprobe

Figure 1Genotyping and phenotyping mice for the A2AR gene. (a) Ethidium bromide–stained PCR amplification products of tail-clip DNA from WT(A2A

+/+) and A2A-KO (A2A–/–) mice. The A2A-KO allele was identified by PCR of the inserted neomycin resistance cassette using oligonucleotide

primers (see Methods) NEO-F and NEO-R and yielding a 618-bp product. The A2A WT allele was identified by PCR of a portion of the WTA2AR gene that was deleted from the KO construct using primers WT-F and WT-R and yielding a 163-bp band. +/+, WT; +/–, heterozygous;–/–, homozygous A2A-KO. Each lane represents PCR products from individual mice. (b) Immunohistochemical localization of A2AR in mouseforebrain sections. In the WT brain sections (top), dense A2AR-like immunoreactivity is present in the striatum (caudate putamen) and extendsthrough the cell bridges of the striatum (arrow) to the olfactory tubercles on the ventral surface of the brain. In A2A-KO brain (bottom), spe-cific immunoreactivity is completely absent. Scale bar, 1 mm. CPu, caudate putamen; Tu, tubercles; cc, corpus callosum. (c) Effect ofATL146e in A2A-KO mice. A2A-KO mice were subjected to IRI (see Methods). The rise in plasma creatinine after ATL146e was similar to thatafter vehicle administration (n = 6, not significant). Values are means ± SE.

886 The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6

template sets (mCK1, mCK2b, mCK3, and mCK5; BDPharmingen) for cytokine/chemokine genes. Kidney totalRNA was subjected to solution hybridization with eachprobe set. Hybridization was performed at 56°C beforeRNase treatment. After RNase treatment, protected frag-ments were separated by gel electrophoresis on 5% poly-acrylamide gels and exposed to Kodak X-Omat AR film at–70°C with a single intensifying screen.

Statistical analysis. Unpaired Student’s t-test was usedfor all comparisons. A P value of less than 0.05 was usedto define statistical significance.

ResultsGenotype and phenotype of A2A-KO mice. The genotype ofmice was determined by PCR using primers to amplify a618-bp fragment of the neomycin cassette present in theKO allele and a 163-bp fragment present in WT sequencethat was deleted from the KO allele (Figure 1a). Mousephenotypes were assessed immunohistochemicallyusing a selective A2AR monoclonal antibody (13). Con-sistent with our previous results in rats (13), immuno-histochemical labeling of brain sections revealed theexpected pattern of dense A2AR-like immunoreactivity inthe striatum of WT mice (Figure 1b, top) and theabsence of immunoreactivity in the striatum of A2A-KOmice (Figure 1b, bottom). Taken together, these resultsconfirm disruption of the A2AR gene and the loss ofexpression of A2AR protein in A2A-KO mice.

No significant differences were detected between WTand A2A-KO mice in mean arterial pressure (MAP),heart rate (HR), pH, pCO2, pO2, or core body tempera-ture (12), and the kidneys of A2A-KO mice were histo-logically indistinguishable from those of C57BL/6 mice(data not shown). Baseline plasma creatinine levels were0.128 ± 0.013 (n = 8) and 0.111 ± 0.014 mg/dl (n = 14, notsignificant) for WT and A2A-KO mice, respectively.Baseline 24-hour urine sodium excretion levels were

0.096 ± 0.020 (n = 9) and 0.099 ± 0.021 µmol per 24hours (n = 8, not significant), and 24-hour urine potas-sium excretion levels were 0.144 ± 0.029 (n = 9) and0.138 ± 0.019 µmol per 24 hours (n = 8, not significant)for WT and A2A-KO mice, respectively.

Effect of ATL46e or rolipram on IRI injury in A2A-KO mice.The primary goal of the current study was to determinethe contribution of BM-derived cells to A2AR ago-nist–mediated renal protection from IRI. For this pur-pose, we generated chimeric mice in which BM of WTmice was replaced with BM-derived cells from A2A-KOmice. This strategy rests on the assumption that theprotective effect of ATL146e is due to specific effects onA2ARs. Preliminary experiments examined the effects ofATL146e on renal function in WT and A2A-KO mice fol-lowing IRI. In WT mice, IRI produced a rise in plasmacreatinine levels that was reduced 78% by administra-tion of ATL146e to WT mice. Plasma creatinine levels inmice treated with vehicle or ATL146e were 1.41 ± 0.25and 0.35 ± 0.12 mg/dl (n = 3, P < 0.02). This is similar toa 62% reduction (n = 9, P < 0.001) that we reported pre-viously (8). In A2A-KO mice, the rise in plasma creatinineafter IRI was not affected by ATL146e administration(Figure 1c). Plasma creatinine levels were 1.72 ± 0.17 and1.77 ± 0.16 mg/dl for vehicle and ATL146e, respectively(n = 6, not significant). This lack of effect of ATL146e inreducing renal injury in A2A-KO mice is consistent withour previous demonstration that the selective competi-tive A2AR antagonist, ZM243185, inhibits ATL146e-mediated renal protection (6).

It is possible that deletion of the A2AR gene mightcause mice to be nonspecifically insensitive to protec-tion by ATL146e — for example, by enhancing the mag-nitude of IRI. To test this possibility, we examined theeffects of the PDE 4 inhibitor, rolipram, which we haveshown previously protects WT mice from renal IRI (8).Both A2AR activation and PDE inhibition are thought

Figure 2Effect of renal IRI and ATL146e on plasma creatinine in chimeric mice. (a) Reconstitution efficacy of granulocytes and lymphocytes at 6weeks after BM transplantation. PBMCs from transplanted recipients were purified and incubated with anti-CD45.1 as a marker of donorcells and either anti-CD11b, anti-CD4, or anti-CD8a monoclonal antibodies (see Methods). Labeled cells were enumerated by flow cytom-etry. The number at the top of each panel indicates the percentage of cells that are positive for both CD45.1 and each of the other markerfluorescent antibodies. Data are representative of 12 independent experiments. (b) Effect of renal IRI and ATL146e on plasma creatinine inchimeric mice. Chimeric mice were prepared by transferring to lethally irradiated WT recipients BM from either WT (WT→WT) or A2A-KO(A2A-KO→WT) donor animals. Chimeric animals were treated with vehicle or ATL146e (10 ng/kg per minute) beginning 5 hours before 32minutes of ischemia and continuing for 24 hours of reperfusion. n = 11 for each group. Values are means ± SE. *P < 0.0001 versus vehicle.

The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6 887

to elicit tissue protection through a cAMP/PKA path-way (8). We reasoned that rolipram can increase cAMPin the absence of A2ARs. Infusion of rolipram (0.1 ng/kgper minute) into A2A-KO mice significantly reducedplasma creatinine from 2.36 ± 0.16 (n = 4) to 1.52 ± 0.20mg/dl (n = 5, P < 0.02). These results suggest that fail-ure of ATL146e to prevent renal IRI in A2AR KO mice isdue specifically to deletion of receptors.

BM reconstitution. To determine the principal cellular tar-gets of A2A action, we generated chimera in which BM wasablated from WT mice and reconstituted with BM cellslacking the A2AR gene (A2A-KO→WT) or with BM fromWT mice (WT→WT). In preliminary studies, chimeraprepared from different mouse strains were used to eval-uate the extent of reconstitution of various BM-derivedcells. Figure 2a shows representative examples demon-strating the percentage of cells expressing the donormarker, CD45.1, in PBMCs that are positive for CD11b(98.4 ± 0.4), CD4 (83.8 ± 2.0), and CD8 (84.5 ± 2.3) by flowcytometry. These results demonstrate efficient repopula-tion of BM-derived cells in chimeric animals.

Effect of ATL146e on renal IRI in chimeric mice. We nextsought to determine the contribution of BM-derived cellsto the protective effects of A2AR activation. WT→WT andA2A-KO→WT chimeric mice were subjected to IRI, andplasma creatinine was used as a measure of renal injury(Figure 2b). In WT→WT chimera, the rise in the plasmacreatinine level after IRI was reduced significantly inATL146e-treated mice (n = 11; vehicle, 1.75 ± 0.09 mg/dl;ATL146e, 0.75 ± 0.06 mg/dl; P < 0.0001). The reductionin plasma creatinine by ATL146e in WT→WT chimericmice (43%) was similar to that determined previously inWT mice (38% of control) (8). Plasma creatinine levelsafter IRI in vehicle-treated A2A-KO→WT chimeric mice(1.99 ± 0.08 mg/dl, n = 12) were unaffected by ATL146einfusion (1.96 ± 0.13 mg/dl, n = 11, not significant).

There was no significant effect of IRI on MAP or HRin chimeric mice. During IRI with ATL146e, HRdecreased in the WT→WT group (n = 6, ∆HR = –39.9 ± 17)but increased in A2A-KO→WT mice (n = 6, ∆HR = 42.5± 17). These results demonstrate that the protectiveeffect of ATL146e in kidney IRI requires activation ofA2ARs on BM-derived cells.

Evidence against transdifferentiation of transplanted BM cellsinto endothelial dells. If renal endothelium is repopulated byBM-derived cells, then the deletion of A2AR on endothe-lial cell precursors could account for the absence of pro-tection by ATL146e in A2A-KO→WT chimera. To deter-mine the degree to which transdifferentiation ofdonor-derived endothelial progenitor cells to renal vas-cular endothelium occurs in chimera subjected to IRI,GFP→WT chimera mice were subjected to IRI. Cellsexpressing GFP were detected by anti-GFP antibody. Fig-ure 3a shows diffuse GFP immunoreactivity in variouscells of the outer medulla (OM) of kidney from a GFPmouse. In particular, endothelial cells are stained bright-ly and have a characteristic elongated morphology.Spleens from GFP→WT chimera contained abundantGFP immunoreactivity (data not shown). After IRI in

GFP→WT chimera, scattered GFP-immunoreactive cellsthat are morphologically distinct from endothelium arevisible in kidney under low magnification (Figure 3b);these may be inflammatory cells that infiltrate the kidneyafter IRI. In dual-labeling studies, high-magnificationimages of the OM demonstrate GFP-positive cells in theperitubular capillaries (Figure 3c) that are identified asmacrophages (Figure 3d). Occasional macrophages donot contain GFP immunoreactivity and may be of recip-ient mouse origin (arrowhead, Figure 3d). In renal sec-tions from four GFP→WT chimeric mice, we could notdetect any GFP-positive cells with an endothelial mor-phology. We conclude that 6 weeks after BM transplan-tation and 24 hours after IRI, there are few if any trans-differentiated endothelial cells in chimeric animals.

Effect of ATL146e on cytokine and chemokine mRNA expres-sion in kidneys of chimeric mice. We used chimeric mice todetermine the contribution of activation of A2ARs onBM-derived cells to renal cytokine mRNA expressionafter IRI. In kidneys from WT→WT mice, 32 minutesof ischemia and 8 or 24 hours of reperfusion causedinduction of multiple cytokine mRNAs that were notevident in sham mice. IL-6 and TGF-β were induced at8 and 24 hours after reperfusion (Figure 4a, lanes 3, 5,8, and 10). ATL146e led to a pronounced reduction inthe induction of IL-6 and TGF-β in kidneys ofWT→WT mice at 8 (lane 4) and 24 (lane 6) hours buthad no effect on mRNA induction in A2A-KO→WTmice at 8 (lane 9) or 24 (lane 11) hours of reperfusion. A separate probe set was used to examine the effect of IRI on expression of various other cytokine transcripts after reperfusion. IL-6 mRNA was increased by IRI

Figure 3Immunohistochemistry of GFP-positive cells in kidney of GFP miceand GFP→WT chimera. (a) Immunohistochemical localization ofGFP in outer medulla of kidney from GFP mice. Magnification, ×630.Endothelial cells are identified by arrows. (b) Outer medulla of kid-ney from GFP→WT chimeras after renal IRI. Donor BM-derived cellsare identified by arrows. Magnification, ×630. (c and d) Dual label-ing of kidney from GFP→WT chimeras after renal IRI. Labeling withanti-GFP antibody (c) reveals cells of donor origin (arrows, c) thatare identified as macrophages (arrows, d) by labeling with anti-macrophage antibody (d). Arrowhead indicates recipient macro-phage. Magnification, ×1000 in (c) through (d).

888 The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6

and reduced by ATL146e in WT→WT mice but not in A2A-KO→WT mice (Figure 4b). IL-1β and IL1 receptorantagonist (IL-1ra) mRNA were also found to beincreased after IRI, an effect blocked by ATL146e (Fig-ure 4c). RANTES, eotaxin, macrophage inhibitory pro-tein–1α (MIP-1α), MIP-1β, MIP-1β, IFN-γ–inducibleprotein 10 (IP-10), and monocyte chemoattractant pro-tein–1 (MCP-1) mRNA were induced at 24 hours afterIRI; however, ATL146e had no effect on their expression

(Figure 4d). At 24 hours after IRI, TNF-β, lymphoto-xin β, INF-β, IL-4, IL-5, IL-10, IL-9, IFN-γ, IL-12p35, IL-12p40, IL-18/glycosylation-inhibiting factor, andmacrophage migration–inhibiting factor (Figure 4, a–d)transcripts were not detected.

Effect of endogenous adenosine on renal IRI in kidneys fromC57BL/6, A2A-KO, and chimeric mice. Given the protectiveeffect observed with A2AR activation, we sought to deter-mine whether A2A-KO mice are more sensitive to the

Figure 4Cytokine and chemokine gene expression from kidneys of chimeric mice subjected to IRI. Kidneys were subjected to 32 minutes of ischemiaand either 8 or 24 hours of reperfusion in animals that were treated with either vehicle or ATL146e. V, vehicle; A, ATL146e. (a) RNase pro-tection assay (RPA) analysis of selected cytokines from kidney mRNA using mCK3 probes. Sham-operated chimeric mice (lanes 2 and 7) andchimeric mice subjected to IRI (lanes 3–6 and 8–11). WT→WT mice at 8 (lanes 3 and 4) or 24 (lanes 5 and 6) hours and A2A-KO→WT miceat 8 (lanes 8 and 9) or 24 (lanes 10 and 11) hours as indicated. Each lane represents solution hybridization with RNA derived from kidneysof separate mice. (b) RPA analysis using the mCK1 probes. (c) RPA analysis using mCK2b probes. WT→WT mice at 24 hours (lanes 1–5)and A2A-KO→WT mice at 24 hours (lanes 6–8). Mice were subjected to sham operation (S) (lanes 1 and 6) or IRI (lanes 2–5, 7, and 8). Rel-ative mobility of unprotected cytokine mRNA is illustrated on the right. S, sham operation. (d) RPA analysis using mCK5 probes. KidneyRNA was isolated from WT→WT mice at 24 hours (lanes 1–3) and from A2A-KO→WT mice at 24 hours (lanes 4–6). Mice were subjectedto sham operation (lanes 1 and 4) or IRI (lanes 2, 3, 5, and 6).

The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6 889

effects of IRI than WT animals, as might be expected ifendogenous adenosine exerts a protective action. We sub-jected kidneys of A2A-KO mice and of age- and sex-matched WT mice to IRI. IRI produced an increase in plas-ma creatinine levels in WT mice (1.36 ± 0.18 mg/dl) and asignificantly greater increase in congenic A2A-KO mice(2.91 ± 0.28 mg/dl) (n = 5, P < 0.002). These data suggestthat endogenous adenosine may protect kidneys. Todetermine the effect of endogenous adenosine on A2ARson BM-derived cells, we compared the effect of IRI onWT→WT and A2A-KO→WT chimera. Plasma creatininelevels after 32 minutes of ischemia and 24 hours of reper-fusion were 1.48 ± 0.076 (n = 7) and 1.78 ± 0.78 (n = 7) inWT→WT and A2A-KO→WT mice, respectively (P < 0.02).These data suggest that endogenous adenosine has adirect effect on BM-derived cells to reduce renal IRI injury.

DiscussionThe purpose of the current study was to determine thecontribution of A2ARs expressed on BM-derived cells torenal tissue protection from IRI. We found that whenWT mice had their BM ablated and reconstituted withA2A-KO BM cells, the protective effect of ATL146e wascompletely abolished. This indicates that activation ofA2ARs expressed on BM-derived cells is necessary forATL146e-mediated tissue protection. We also showthat A2AR activation on BM-derived cells reduces IRI-induced elevation of IL-6, IL-1β, and IL-ra and TGF-βmRNAs in kidney. These cytokines may participate inrenal damage in response to IRI. Lastly, endogenousadenosine, by acting on BM-derived A2ARs, is probablya factor in limiting renal damage after IRI.

We have focused our efforts on defining the cellular tar-get(s) through which A2AR activation affords protectionfrom renal injury. In our previous studies in rats andC57BL/6 mice, we showed that infusion of ATL146e (ahighly selective A2A agonist that was previously calledDWH-146e) at doses below the threshold necessary tochange blood pressure reduced elevated plasma creatinineseen after renal IRI (6, 8, 10). The mechanism by which A2A

activation reduces IRI could involve effects on endothelial,epithelial, smooth muscle, or hematopoietic cells. Severalgroups of investigators have demonstrated that adenosineacts on activated neutrophils (16) in vitro to reduce pro-duction of oxygen radicals and neutrophil adherence toendothelial cells (17). These in vitro studies demonstratethat the renal protection observed in vivo may be due todirect effects of A2AR activation in suppressing adhesionand activation of neutrophils and release of toxic media-tors from neutrophils. We demonstrated that neutrophilaccumulation in the outer medulla of kidneys subjectedto IRI is significantly reduced by ATL146e administration.The reduction of neutrophil accumulation may be relat-ed to a decrease in endothelial P-selectin and ICAM-1expression (10) in mice given ATL146e, suggesting a director indirect effect on endothelial cells. Other groups haveshown that the vasoconstrictive effects of angiotensin IIin the outer medullary descending vasa recta are blockedby A2AR activation (9). Lastly, oxidant injury of immor-

talized human proximal tubule cells was reduced uponactivation of A2AR with CGS21680 (18). These studiessuggest multiple potential targets for adenosine A2A ago-nist action in ameliorating IRI.

To determine directly the contribution of A2ARs onBM-derived cells, we generated A2A-KO→WT chimera.Our approach is based on the concept that repopulationof cells in recipient mice after BM ablation and trans-plantation with donor BM cells creates chimeras withreconstitution limited to BM-derived cells. Preliminarystudies indicated that blood granulocytes were success-fully reconstituted by approximately 98% and T cells byapproximately 85%. These results are in keeping withpreviously published studies using this method (19).

The results provide direct evidence that the protectiveeffect of ATL146e requires activation of A2ARs expressedon BM-derived cells. Several important conclusions canbe drawn from these results. Many studies of renaldamage using IRI models in rats and mice (1), includ-ing our own (10), have suggested a pathogenic role fromneutrophil infiltration of kidney tissue at various timesafter injury. Other studies either directly or indirectlyimplicate T cells or macrophages in IRI (4, 5, 20). In thecurrent study, the finding that injury in A2A-KO→WTmice is unabated with an A2A agonist sheds additionallight on the role of BM-derived cells in tissue injury.The possibility that leukocytes are important in IRI hasbeen controversial (2). Our studies clearly suggest thatBM-derived leukocytes are important in the pathogen-esis of IRI and that suppressing circulating leukocytefunction with ATL146e reduces injury.

Although our studies indicate that BM-derived cells arethe target of ATL146e in mediating tissue protection, onepossible explanation for the data is that donor-derivedendothelial progenitor cells repopulate recipientendothelium. If most renal endothelial cells were replacedas a result of BM transplantation, the lack of effect of A2A

agonists on renal IRI could be due to repopulation withendothelial cells bearing an A2A-KO phenotype. Reason-ing that donor BM cells from GFP mice can be detectedin recipient tissues, we made GFP→WT chimera to deter-mine the extent to which endothelial progenitor cellsrepopulate recipient endothelium. Although the renalendothelium of GFP mice was brightly immunofluores-cent, we observed only scattered cells of donor origin inGFP→WT chimeric mice subjected to IRI. The rareGFP-positive cells did not have endothelial morphology,and most were identified as macrophages. Althoughdonor-derived circulating endothelial progenitors havebeen found in mice after bone marrow transplantation,the sites and magnitude of repopulation of theseendothelial progenitors should be considered in the con-text of the present study. For example, Asahara et al. (21)have shown that tie2/lacz transcript (donor origin) canbe detected in various murine organs after BM trans-plantation, but this was detected by the extremely sensi-tive technique of RT-PCR and was not quantified. Themost prominent evidence of donor endothelial cellrepopulation has been found associated with vasculoge-

890 The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6

nesis or angiogenesis (21–23) after vascular injury (24).The outgrowth of neovessels is rare under normalphysiological conditions. Significant neovascularizationhas been demonstrated in BM-transplanted mice onlyafter various chronic injuries such as implantation oftumor cells (22), ocular neovascularization after surgery(23), murine ischemic hind limb (21, 23), and myocardialischemia (25). Importantly, a detectable and very low levelof donor BM cells in recipient endothelium is only report-ed in chronic neovascularization models (at least 1 week).Sata et al. performed BM transplantation in which irradi-ated WT mice received BM from ROSA26 mice, whichexpress β-galactosidase (LacZ) (26). After BM transplan-tation, femoral artery of uninjured vessels did not expressLacZ-positive cells. Wagers et al. demonstrated that BMtransplantation using hematopoietic stem cells from GFPmice while reconstituting hematopoietic cells of recipientmice only rarely repopulated nonhematopoietic tissue. Inthe BM transplantation kidney ischemia–reperfusionmodel reported in this study, we clamped the renal arteryfor 32 minutes and sacrificed the mice after 24 hours ofreperfusion. No angiogenesis occurs over this time frame.Our studies are consistent with the conclusion thatendothelial repopulation by donor-derived progenitorcells occurs only to a very limited extent after IRI for 24hours. Thus, it is likely that the absence of a protectiveeffect of ATL146e in A2AKO→WT chimeras is due princi-pally to the absence of A2A-ARs on leukocytes.

Although we demonstrate that BM-derived cells areprincipal targets of action of ATL146e, the specific celllineage is not certain. Because chimeric mice havehematopoietic cells from all cell lineages, tissue protec-tion mediated by ATL146e may be due to effects onmyeloid and/or lymphocytic lineages. Recent studiesindicate that A2ARs are expressed on T cell subsets (27),and in light of the foregoing discussion, the effect ofATL146e could be mediated by action on T cells. Furthersupport of the concept that T cell A2ARs could partici-pate in tissue protection from IRI comes from in vitrostudies. A2A activation is also known to inhibit cytokinessuch as TNF-α, IL-6, IL-8, and IL-12 from mononuclearcells (28, 29). Identification of the specific cell lineage tar-get of ATL146e will be the goal of future studies.

We observed that renal IRI in A2A-KO mice was moresevere than in control C57BL/6 mice as measured by plas-ma creatinine. A similar observation by Ohta andSitkovsky showed that liver injury in response to toxins ismore severe in mice that lack A2ARs (7). Furthermore, ourdata suggest that BM-derived cells are important for theeffect of endogenous adenosine, because IRI in kidneys ofA2A-KO→WT mice was more severe than in WT→WTmice. These observations suggest that activation of A2ARsby endogenously produced adenosine may participate ina negative feedback mechanism to minimize tissueinflammation. The concept that endogenous adenosinecontributes to tissue protection through activation ofA2ARs is supported by our findings with the PDE 4inhibitor rolipram. In the current studies, we observedthat rolipram reduced renal IRI by 36%. The reduction of

injury is less than the 58% reduction observed whenrolipram was administered to WT mice (8). This mightoccur because rolipram enhances the actions of endoge-nous adenosine in WT but not A2A-KO mice.

The ability of endogenous adenosine to suppress tissueinjury is consistent with in vitro and in vivo studies. It iswell known that within 2 minutes of renal ischemia,interstitial adenosine concentration increases more thansixfold over the normal level of less than 1 µM (30).Adenosine per se has suppressive effects on activated neu-trophils or endothelial cells (17). It is likely that althoughactivation of A2ARs by endogenous adenosine in WT micemediates protection from injury, mechanisms that couldcontribute to the protective effects of endogenous adeno-sine and to the degree of injury associated with IRI maydiffer in WT as compared with A2A-KO mice. The magni-tude of the protective effect of endogenous adenosine isquite small as compared with the protection produced byATL146e. Possible reasons for this are that (a) endoge-nous adenosine levels return to baseline within 10 min-utes of reversal of ischemia due to washout (b) adenosinehas a very short half-life in blood and is probably notaccessible to receptors on circulating white blood cells (c)ATL146e has a higher affinity for A2ARs than adenosine,and (d) adenosine binds nonselectively to all four adeno-sine receptor subtypes, including A1Rs and A3Rs, whichmay promote inflammation. The net effect of increasedadenosine levels will then depend on the balance ofeffects on multiple receptor subtypes and duration ofaction. In summary, adenosine can serve as a naturallyoccurring anti-inflammatory molecule and reduce tissueIRI, but this effect is probably limited primarily to theinterstitial space of ischemic tissue.

We measured the effects of ATL146e on cytokine andchemokine mRNA expression in kidneys from chimericmice. Our results demonstrate that expression of 11 ofthe 24 cytokines and chemokines measured in kidney isenhanced after IRI. The recruitment and activation ofmonocytes, macrophages, neutrophils, and T lympho-cytes to the kidney that occurs after IRI (20) is probablydue in part to the enhanced expression of IL-6, TGF-β,MIP-2, MCP-1, IP-10, and RANTES. IL-6 is a majorproinflammatory cytokine that interacts with TNF-αand IL-1 to stimulate dendritic cell migration. IL-6 isinvolved in Th1 lymphocyte differentiation and activa-tion, production of IL-2 (31), a major clonal expansioncytokine, and induction of C-reactive protein and man-nan-binding lectin from liver (32). C-reactive protein andmannan-binding lectin may bind to antigens releasedfrom damaged cells and then activate complementopsonization and phagocytosis. In this context, it isplausible that ATL146e exerts tissue protection partly byblocking IL-6 production at an early stage. TGF-β is adual immune modulator that may play a role in theinflammation process at an early stage and then act asan immune suppressor of activated T lymphocytes dur-ing the regeneration stage. TGF-β produced by T lym-phocytes is significant in immune suppression for itsattenuation of the Th1 response. Recent findings also

The Journal of Clinical Investigation | September 2003 | Volume 112 | Number 6 891

indicate that TGF-β may play a proinflammatory role byincreasing chemotaxis of mononuclear cells, cytokineproduction (IL-1α, IL-1β, TNF-α, PDGF-BB, bFGF),expression of integrin receptors (LFA-1, ICAM-1, VLA-3),and activity of phagocytes (33).

Our data show that activation of A2ARs decreases theexpression of TGF-β, IL-1β, IL-1ra, and IL-6 transcriptsin the kidney in WT→WT mice but not in A2A-KO→WTkidney. Because early effects of IL-1β, TGF-β, and IL-6may promote inflammation, their early appearance maycontribute to damage observed in IRI.

Taken together, this study using genetically engineeredand chimeric mice sheds light on the sequence of eventsthat culminate in renal IRI and the mechanism by whichA2AR activation can ameliorate this injury. First, theresults suggest a causal link between BM-derived cells andthe component of the injury response that occurs duringthe reperfusion period after renal ischemia. Second, tis-sue-protective effects of A2AR agonists require activationof receptors expressed on BM-derived cells. These obser-vations are particularly interesting since A2ARs are notlimited in their distribution to BM-derived cells but arealso found on several other tissues, notably on vascularendothelial cells. Although these data demonstrate thatA2ARs on BM-derived cells are necessary for A2A ago-nist–induced protection from renal injury, they do notexclude a minor contribution from nonhematopoieticcells expressing A2ARs. For example, it is possible thatA2AR activation on both BM-derived cells and on cells inthe kidney is required to elicit significant renal protec-tion. A third conclusion of this study is that endogenousadenosine modulates tissue injury by activating A2ARs.

The participation of BM-derived cells in adenosine-mediated renal protection is consistent with in vitrostudies demonstrating that A2A agonists such asATL146e potently inhibit leukocyte adhesion, oxidativeburst, and lymphocyte activation. It will be interestingin future studies to determine if the target of A2A ago-nists that protect the kidney is primarily on T cells orcells of myeloid lineage.

AcknowledgmentsThe authors gratefully acknowledge Timothy Macdon-ald and Jayson Rieger (Department of Chemistry, Uni-versity of Virginia) for the generous gift of ATL146e, andMelissa Marshall (Department of Medicine, University ofVirginia), Tom Obrig’s laboratory (Department of Medi-cine, University of Virginia), and Jan Redick (ElectronMicroscopy Core Facility, University of Virginia) for tech-nical assistance. This work was supported in part by NIHgrants DK56223, DK58413, HL37942, and ES10804.

1. Lauriat, S., and Linas, S.L. 1998. The role of neutrophils in acute renalfailure. Semin. Nephrol. 18:498–504.

2. Thadhani, R., Pascual, M., and Bonventre, J.V. 1996. Acute renal failure.N. Engl. J. Med. 334:1448–1460.

3. Zwacka, R.M., et al. 1997. CD4+ T-lymphocytes mediate ischemia/reper-fusion-induced inflammatory responses in mouse liver. J. Clin. Invest.100:279–289.

4. Burne, M.J., et al. 2001. Identification of the CD4+ T cell as a major path-ogenic factor in ischemic acute renal failure. J. Clin. Invest.108:1283–1290. doi:10.1172/JCI200112080.

5. Rabb, H., et al. 2000. Pathophysiological role of T lymphocytes in renalischemia-reperfusion injury. Am. J. Physiol. Renal Physiol. 279:F525–F531.

6. Okusa, M.D., Linden, J., Macdonald, T., and Huang, L. 1999. SelectiveA2A-adenosine receptor activation during reperfusion reduces ischemia-reperfusion injury in rat kidney. Am. J. Physiol. 277:F404–F412.

7. Ohta, A., and Sitkovsky, M. 2001. The critical role of G-protein coupledadenosine receptors in down-regulation of inflammation and preven-tion of tissue damage in vivo. Nature. 414:916–919.

8. Okusa, M.D., et al. 2001. Enhanced protection from renal ischemia-reperfusion injury with A2A-adenosine receptor activation and PDE 4inhibition. Kidney Int. 59:2114–2125.

9. Silldorff, E.P., Kreisberg, M.S., and Pallone, T.L. 1996. Adenosine mod-ulates vasomotor tone in outer medullary descending vasa recta of therat. J. Clin. Invest. 98:18–23.

10. Okusa, M.D., et al. 2000. A2a-adenosine receptor mediated inhibition ofrenal injury and neutrophil adhesion. Am. J. Physiol. 279:F809–F818.

11. Gessi, S., Varani, K., Merighi, S., Ongini, E., and Borea, P.A. 2000. A2A-adeno-sine receptors in human peripheral blood cells. Br. J. Pharmacol. 129:2–11.

12. Chen, J.F., et al. 1999. A(2A) adenosine receptor deficiency attenuates braininjury induced by transient focal ischemia in mice. J. Neurosci. 19:9192–9200.

13. Rosin, D.L., Robeva, A., Woodard, R., Guyenet, P., and Linden, J. 1998.Immunohistochemical localization of adenosine A2A receptors in the ratcentral nervous system. J. Comp. Neurol. 401:163–186.

14. Huang, L., Wei, Y.Y., Momose-Hotokezaka, A., Dickey, J., and Okusa, M.D.1996. α2B-Adrenergic receptors: immunolocalization and regulation bypotassium depletion in rat kidney. Am. J. Physiol. 270:F1015–F1026.

15. Maxwell, M.H. 1978. Two rapid and simple methods used for theremoval from 1.0 micron thick epoxy sections. J. Microsc. 112:253–255.

16. Cronstein, B.N., Rosenstein, E.D., Kramer, S.B., Weissmann, G., andHirschhorn, R. 1985. Adenosine, a physiological modulator of superox-ide anion generation by human neutrophils. Adenosine acts via an A2

receptor on human neutrophils. J. Immunol. 135:1366–1371.17. Cronstein, B.N. 1994. Adenosine, an endogenous anti-inflammatory

agent. J. Appl. Physiol. 76:5–13.18. Lee, H.T., and Emala, C.W. 2002. Adenosine attenuates oxidant injury in

human proximal tubular cells via A1 and A2a adenosine receptors. Am. J.Physiol. Renal Physiol. 282:F844–F852.

19. Jung, U., and Ley, K. 1999. Mice lacking two or all three selectins demon-strate overlapping and distinct functions for each selectin. J. Immunol.162:6755–6762.

20. Takada, M., Nadeau, K.C., Shaw, G.D., Maruette, K.A., and Tilney, N.L.1997. The cytokine-adhesion molecule cascade in ischemia/reperfusioninjury of the rat kidney. J. Clin. Invest. 99:2682–2690.

21. Asahara, T., et al. 1999. Bone marrow origin of endothelial progenitorcells responsible for postnatal vasculogenesis in physiological andpathological neovascularization. Circ. Res. 85:221–228.

22. Carmeliet, P.L., Luttun, A. 2001. The emerging role of the bone mar-row–derived stem cells in (therapeutic) angiogenesis. Thromb. Haemost.86:289–297.

23. Takahashi, T., et al. 1999. Ischemia- and cytokine-induced mobilizationof bone marrow–derived endothelial progenitor cells for neovascular-ization. Nat. Med. 5:434–438.

24. Rookmaaker, M.B., et al. 2002. Bone-marrow-derived cells contribute toendothelial repair after thrombotic microangiopathy. Blood. 99:1095.

25. Ikpeazu, C., Davidson, M.K., Halteman, D., Browning, P.J., and Brandt, S.J.2000. Donor origin of circulating endothelial progenitors after allogene-ic bone marrow transplantation. Biol. Blood Marrow Transplant. 6:301–308.

26. Sata, M., et al. 2002. Hematopoietic stem cells differentiate into vascular cellsthat participate in the pathogenesis of atherosclerosis. Nat. Med. 8:403–409.

27. Koshiba, M., Rosin, D.L., Hayashi, N., Linden, J., and Sitkovsky, M.V. 1999.Patterns of A2A-extracellular adenosine receptor expression in differentfunctional subsets of human peripheral T cells. Flow cytometric studieswith anti-A2A-receptor monoclonal antibodies. Mol. Pharmacol. 55:614–624.

28. Sullivan, G.W., and Linden, J. 2002. The role of adenosine in tissue pro-tection during ischemia-reperfusion. In Sensing, signaling and cell adapta-tion. K.B. Storey and J.M. Storey, editors. Elsevier Science B.V. Amster-dam, The Netherlands. 47–59.

29. Hasko, G., et al. 2000. Adenosine inhibits IL-12 and TNF-α productionvia adenosine A2a receptor-dependent and independent mechanisms.FASEB J. 14:2065–2074.

30. Osswald, H., Schmitz, H.-J., and Kemper, R. 1977. Tissue content ofadenosine, inosine and hypoxanthine in rat kidney after ischemia andpostischemic recirculation. Pflügers Arch. 371:45–49.

31. Garman, R.D., Jacobs, K.A., Clark, S.C., and Raulet, D.H. 1987. B-cell-stimulatory factor 2 (beta 2 interferon) functions as a second signal forinterleukin 2 production by mature murine T cells. Proc. Natl. Acad. Sci.U. S. A. 84:7629–7633.

32. Castell, J.V., et al. 1988. Recombinant human interleukin-6 (IL-6/BSF-2/HSF) regulates the synthesis of acute phase proteins in humanhepatocytes. FEBS Lett. 232:347–350.

33. Letterio, J.J., and Roberts, A.B. 1998. Regulation of immune responsesby TGF-beta. Annu. Rev. Immunol. 16:137–161.