Beneficial effects of CD39/ecto-nucleoside triphosphatediphosphohydrolase-1 in murine intestinal ischemia-reperfusion injuryOlaf Guckelberger1, 2, Xiao Feng Sun2, Jean Sévigny2, Masato Imai2, Elzbieta Kaczmarek2,Keiichi Enjyoji2, Jonathan B. Kruskal3, Simon C. Robson2
1Department of Visceral- and Transplantation-Surgery, Charité Campus Virchow-Clinic, Humboldt University, Berlin, Germany2Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, USA3Department of Radiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, USA
Thromb Haemost 2004; 91: 576–86
were used to study platelet-endothelial cell interactions anddetermine capillary leakage. In wild-type animals, ischemiareperfusion injury resulted in 60% mortality within 48 hours. Inmutant mice null or deficient for cd39, ischemia reperfusion-related death occurred in 80% of animals. Apyrase supplemen-tation protected all wild-type animals from death due to intes-tinal ischemia but did not fully protect cd39-null and cd39-hem-izygote mice. Adenosine/amrinone treatment failed to improvesurvival figures. In wild type mice, platelet adherence to postca-pillary venules was significantly decreased and vascular integritywas well preserved following apyrase administration. In cd39-null mice, ischemia-reperfusion induced marked albumin leak-age indicative of heightened vascular permaeability when com-pared to wild-type animals (p=0.04).Treatment with NTPDaseor adenosine supplementation abrogated the increased vascu-lar permeability in ischemic jejunal specimens of both wild-typemice and cd39-null. CD39 activity modulates platelet activationand vascular leak during intestinal ischemia reperfusion injury invivo. The potential of NTPDases to maintain vascular integritysuggests potential pharmacological benefit of these agents inmesenteric ischemic injury.
KeywordsIschemia reperfusion injury, small intestine, platelet activation,endothelial cell activation, videomicroscopy
SummaryCD39 (ecto-nucleoside triphosphate diphosphohydrolase-1;E-NTPDase-1), is highly expressed on quiescent vascular endo-thelial cells and efficiently hydrolyzes extracellular ATP and ADPto AMP and ultimately adenosine. This action blocks extracel-lular nucleotide-dependent platelet aggregation and abrogatesendothelial cell activation. However, CD39 enzymatic activity israpidly lost following exposure to oxidant stress. Modulation ofextracellular nucleotide levels may therefore play an importantrole in the pathogenesis of vascular injury. Acute ischemic injury of the bowel is a serious medical condition characterizedby high mortality rates with limited therapeutic options. Herewe evaluate the effects of cd39-deletion in mutant mice and theuse of supplemental NTPDase or adenosine in influencing theoutcomes of intestinal ischemia-reperfusion. Wild-type, cd39-null, or hemizygous cd39-deficient mice were subjected tointestinal ischemia. In selected animals, 0.2 U/g apyrase (solubleNTPDase) was administered prior to re-establishment ofblood-flow. In parallel experiments adenosine/amrinone wasinfused over 60 min during reperfusion periods. Survival rateswere determined, serum and tissue samples were taken.Intravital videomicroscopy and studies of vascular permeability
Correspondence to:Simon C. RobsonDepartment of MedicineBeth Israel Deaconess Medical CenterHarvard Medical School, 99 Brookline AvenueRoom 301, Boston, Massachusetts02215, USATel.: +1 617 632 0831, Fax: +1 617 632 1861E-mail: [email protected]
Received June 17, 2003Accepted after resubmission December 23, 2003
Grant support:SCR acknowledges support from grants: NIH ROI HL 57307 and HL 63972; JS wassupported by the American Liver Foundation and the Canadian Institutes of Health
Research; OG received a study grant from Deutsche Forschungsgemeinschaft(DFG Gu 490/1-2).
Prepublished online February 4, 2004 DOI: 10.1160/TH03-06-0373
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
577
IntroductionEndothelial cell (EC) activation may be considered a crucialmodulatory event in several inflammatory conditions, inclusiveof ischemia-reperfusion injury (IRI), acute graft dysfunction insmall bowel and solid organ transplantation, and inflammatorybowel disease (IBD). Acute inflammatory responses to IRI arecharacterized by neutrophil (PMN) activation and adhesionwith cytokine and free oxygen radical release (1-5). WhilePMN-EC interactions have been characterized in several reports(1, 2), little is known of the contributory role of platelets to thevascular injury observed in IRI (6, 7). Earlier work has suggest-ed that erythrocyte trapping in association with platelet aggre-gate formation may be directly correlated with cold ischemictimes in hepatic allograft preservation injury (8).
EC activation may be considered the key event in the settingof IRI that promotes platelet microthrombus formation (9). Thequiescent EC maintains an anticoagulant and antithromboticvascular phenotype, that is mediated by a number of thrombo-regulatory mechanisms (10). During EC activation, heightenedexpression of selectins and β2-integrins mediates PMN adhe-sion and migration (1, 2). The loss of thrombomodulin, heparansulfate, and nucleoside triphosphate diphosphohydrolase(NTPDase) from EC further facilitates the formation of a pro-thrombotic state (11).
CD39 is an endothelial cell NTPDase that hydrolyzes extra-cellular ATP and ADP to AMP. Further conversion of AMP toadenosine is mediated by endothelial associated 5’-nucleotidase(5’NT). ADP released by the endothelium or other vascular cellsinduces platelet recruitment and activation, while activatedplatelets release high concentrations of ADP that augment positive feedback mechanisms (12). Furthermore, decreasedadenosine generation abrogates the potential anti-aggregatoryeffects on platelets (13) and EC protection mediated by cyclicadenosine-5’-monophosphate (cAMP) (14). The capacity ofvascular NTPDase for enzymatic degradation of ATP and ADPto AMP interrupts the evolution of this process in vitro (11). Inaddition, endothelial release of nitric oxide is also influenced bynucleotides (15), and overexpression of CD39 ameliorates ECactivation and apoptosis in vitro. (16). Therefore, oxidant medi-ated loss of vascular NTPDase would be expected to have detrimental effects in vivo (17). Consequently, CD39 may beassigned a critical regulatory element in the control of inflam-matory responses, platelet recruitment and aggregation duringhemostasis and processes of vascular injury.
Markedly decreased NTPDase activity has been observed inrat kidneys subjected to ischemia-reperfusion injury (17). Wehave also recently shown that specific NTPDase activity isdecreased in the very early phase of graft reperfusion in a smallanimal heart transplantation model (18). Administration of apy-rase (a soluble NTPDase) in a comparable model prolonged dis-cordant xenograft survival in both hyperacute and delayed xen-
ograft rejection while markedly attenuating the onset and extentof platelet aggregation (19).
Recently, cd39-null mice have been developed, that exhibitloss of vascular thromboprotective mechanisms (20). Thesemutant mice provide an additional model to study vascularresponses to IRI in the primary absence of cd39.
The studies presented here further elucidate the potential ofvascular NTPDases in the maintenance of vascular integrity inIRI in vivo. The cd39-null mice demonstrated excessive ECactivation following IRI, while supplementation of apyraseprior to re-establishment of blood-flow not only attenuatedplatelet-EC interactions, but also significantly improved survi-val and other outcomes. In this experimental model, combinedadministration of adenosine and amrinone, a selective phospho-diesterase III inhibitor shown to increase cAMP levels and aug-ment adenosine effects (21), provided only partial protectionfrom IRI.
Materials and methodsAnimalsThe cd39-null (n=48), hemizygous cd39-deficient (n=10), andcontrol wild-type mice (C57BL/6x129Svj, n=43) were housedand bred in our own facility and have previously been charac-terized in detail (20). Additional male C57BL/6x129Svj miceweighing 17-25 g (n=57) were obtained from Taconic(Germantown, NY). All experimental animal protocols wereapproved by the Beth Israel Deaconess Medical Center AnimalCare and Use Program.
Animals were housed in a pathogen-free facility accreditedby the American Association for Accreditation of LaboratoryAnimal Care and compliant with the requirements of humaneanimal care as stipulated by the United States Department ofAgriculture and the Department of Health and Human Services.Animals were maintained on a 12-hour light/dark cycle and pro-vided with commercially available rodent chow and tap waterad libitum.
Surgical procedurePrior to surgical intervention, all experimental animals werefasted overnight with unrestricted water access. Mice wereanesthetized with methoxyflurane inhalation (Metofane,Mallinckrodt, Mundelein, IL) in a supine position and a midlinelaparotomy was performed. The superior mesenteric artery(SMA) was isolated at its origin and occluded using a nontrau-matic microvascular clamp (RS-5431, Roboz, Rockville, MD).Loss of SMA pulsation was verified with a stereoscopic zoommicroscope (SMZ-U, Nikon, Melville, NY). During the ischem-ic period, a running suture (Silk 4-0, Ethicon, Somerville, NJ)was used for closure of the abdominal incision to prevent exten-sive heat and fluid loss, and animals were allowed to awakefrom anesthesia. Mice were re-anesthetized five minutes prior
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
578
to re-establishment of intestinal blood-flow and intravenouslyinjected with the respective test solutions. After intestinal reper-fusion, animals in survival studies were resuscitated by intraper-itoneal installation of 2 ml saline, and the abdominal wall wasclosed with two layers of running sutures (Dermalon 4-0,Sherwood-Davis & Geck, St. Louis, MO). Sham operated con-trol animals underwent the same procedure except for closingthe clamp around the SMA.
Experimental groups
NTPDase treatmentAnimals randomly assigned to NTPDase treatment groupsreceived a single intravenous injection of soluble NTPDase (0.2units/g bodyweight of a 20 units/ml stock solution in saline,apyrase grade VII, Sigma, St. Louis, MO) prior to blood-flowrelease. Controls were injected with an equal volume of saline.One group of NTPDase treated cd39-null mice received addi-tional apyrase injections of 0.2 units/g bodyweight i.p. everyeight hours for up to 48 hours post reperfusion.
Adenosine/amrinone treatmentIn additional treatment groups, animals were continuouslyinfused via the left renal vein with 1 µmol/kg/min adenosine(Sigma, St. Louis, MO) and 0.5 µmol/kg/min amrinone (Sigma,St. Louis, MO) over 60 min using an infusion pump (CompactInfusion Pump Model 975, Harvard Apparatus, Holliston, MA,total volume 0.576 ml), starting 5 min prior to re-establishmentof blood-flow. The open peritoneal cavity was covered withsaline soaked gauze to prevent excessive fluid loss.
Survival studiesMice were subjected to 60 min intestinal ischemia as describedabove and observed for seven days. Additional analgesia wasgiven (0.01 mg buprenorphine per ml drinking water, Buprenex,Reckitt & Colman, Richmond, VA). Study groups consisted ofeither untreated, NTPDase treated or adenosine/amrinoneinfused animals from all three genotypes (wild-type, cd39+/+,n=5; cd39-hemizygotes, cd39+/-, n=5; cd39-null, cd39-/-, n=5).
Preliminary experiments with varying durations of intestinalischemia revealed approximately 50% survival of control mice(untreated, wild-type) after 60 min of intestinal ischemia(LD50).
Intravital videomicroscopy for platelet-endothelial cell interactionsAfter induction of 60 min intestinal ischemia, wild-type micewere injected via the penile vein with either fluorescent labeledplatelets alone (controls, n=5; sham controls, n=3) or in a com-bination with soluble NTPDase (treatment group, n=5).Prolonged anesthesia was induced by intraperitoneal pentobar-bital installation (70 µg/g bodyweight, Nembutal, Abbott, North
Chicago, IL). Once blood-flow was re-established, intravitalvideomicroscopy (IVM) of mesenteric venules and arterioleswas performed.
Cd39-null mice subjected to 60 min intestinal ischemia pre-sented with insufficient circulation during IVM and were there-fore excluded from this set of experiments.
Intravital videomicroscopy for capillary leakageTo secure sufficient circulation during IVM in cd39-null mice,animals were either subjected to 45 min intestinal ischemia(n=5 for each group; wild-type, cd39-null) or underwent shamsurgery (n=3 for control groups). Additionally, in all these ani-mals the right jugular vein was catheterized and an infusion portwas implanted (Micro Implantable Infusion Port, HarvardApparatus, Holliston, MA). Prior to re-establishment of blood-flow, prolonged anesthesia was induced and fluorescent labeledalbumin was injected through the infusion port (4 µl/g body-weight of a 6.25 mg/ml stock solution in phosphate bufferedsaline, FITC-albumin, Sigma, St. Louis, MO) followed by 120µl saline to flush the port chamber. Thereafter, IVM was per-formed immediately.
Intestinal vascular permeabilityThe cd39-null or wild-type mice (n=3 for each group) were sub-jected to 45 min of intestinal ischemia. After induction of pro-longed anesthesia and 10 min prior to blood-flow release, injec-tion of 8 µl/g bodyweight of 0.5% Evans Blue in phosphate buf-fered saline (Sigma, St. Louis, MO) via the penile vein was per-formed. All treatment solutions (vehicle, soluble NTPDase, oradenosine/amrinone) were continuously administered via theright iliac vein over 60 min using the infusion pump.Concentrations of treatment solutions were adjusted to body-weight, the total volume injected was not changed (0.576 ml).Before harvesting, the portal vein was cut open and perfusionwith pre-warmed saline via the left cardiac ventricle was per-formed until portal outflow was clear. Sham operation was per-formed as described above. For additional baseline measure-ments of vascular permeability, animals were injected withEvans Blue without performing a sham operation. Unlike shamprocedures, the peritoneal cavity was not open prior to harvest-ing.
Blood and tissue analysisTreated or untreated ischemic wild-type animals or sham oper-ated controls were euthanized prior to re-establishment of intes-tinal blood-flow or one hour after onset of reperfusion (n=3 foreach group). Blood was drawn from inferior vena cava andserum samples were aliquoted and stored for further analysis.Jejunal tissue samples were rinsed with 10 ml pre-warmedsaline and either fixed in formalin or snap frozen with and with-out tissue freezing medium (TBS, American Master*Tech
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
579
Scientific, Lodi, CA). Additionally, mesenteric tissues contain-ing major vascular structures supplying the small intestine wereharvested distal to pancreatic structures and snap frozen. Jejunalspecimens from naive wild-type and cd39-null mice were alsoprocessed to evaluate differences in intestinal NTPDase activ-ity.
Platelet labelingPlatelet labeling was performed as previously described (20).Briefly, blood samples from two wild-type donor mice wereobtained to prepare labeled platelets for each experimental ani-mal and collected in tubes containing 0.1 volume anticoagulant(38 mM citric acid / 75 mM trisodium citrate / 100 mM dex-trose). Platelet-rich plasma (PRP) was prepared by two subse-quent centrifugation steps at 280g for six and four minutes,respectively. Purified platelets for calcein AM labeling wereachieved by filtration through a PIPES buffer (25 mM Pipes /137 mM NaCl / 4 mM KCl / 0.1% dextrose / pH 7.0) equilibrat-ed Sepharose 2B column (25 ml ‘dry’ volume, Sigma, St. Louis,MO). Final platelet concentrations were determined utilizing anautomated particle counter (Z1, Coulter, Miami, FL) and stan-dardized to 3x108/ml. Filtered platelets were incubated with1µg/ml calcein AM for 15 min (Molecular Probes, Eugene, OR)in the dark and checked for fluorescence prior to injection. Allsteps involving platelets were carried out at room temperature.Recipient mice were injected with 5x106 platelets/g bodyweightin 200 to 400 µl PIPES buffer.
Intravital videomicroscopy
Platelet-endothelial cell interactionsExperimental animals were subjected to intestinal IRI andinjected with fluorescent labeled platelets as described above.
Immediately thereafter, mice were placed on a custom mademicroscope stage and an upper jejunal loop was exteriorizedand fixed in the superfusion chamber. The exposed jejunal loopwas constantly superfused with prewarmed lactated Ringer’ssolution (Baxter, Deerfield, IL), and successful re-establishmentof blood-flow was confirmed by conventional light microscopy.
Groups of 10 segments of submucosal arterioles and postca-pillary venules were randomly chosen and visualized for 30 susing epi-illumination through a 40x water immersion objec-tive. The microscope (Optiphot, Nikon, Melville, NY) wasequipped with a 100W mercury lamp (HB-1010 AF, Nikon) anda suitable excitation emission filter set. Microscopic imageswere captured by a CCD video camera (CCD 72, Dage-MTI,Michigan City, IN), displayed on a 12” screen (HR120, Dage-MTI), and taped on a S-VHS recorder (PV-S7670, Panasonic,Secaucus, NJ). For off-line evaluation, taped sequences wereeither reviewed with the above described analog video equip-ment or copied to digital video, captured on a Macintosh computer and analyzed using the public domain NIH Imageprogram (developed at the U.S. National Institutes of Healthand available on the Internet at http://rsb.info.nih.gov/nih-image/).
Labeled platelets interacting with the vasculature were clas-sified according to definitions given earlier by Massberg, et al.(7). In brief, depending on their interaction with the endothelialcell lining platelets were either classified as free floating, roll-ing or adherent (Fig. 1). Adherent platelets were firmly attachedto the vessel wall during the whole observation period, and didnot move at all. Platelets that detached at least once during thisperiod were considered rolling, which otherwise was defined asplatelets crossing any possible plane vertical to blood-flow at avelocity significantly lower than centerline blood velocity.Adherent platelets were assessed as number of cells per square
Figure 1: Adherent and rolling platelets inintravital videomicroscopy.Fluorescent labeled donor platelets wereinjected intravenously and considered eitherflowing, rolling or adherent. Images depict thesame jejunal postcapillary venule at differenttime points with its blood-flow from upperright to the lower left corner (gray arrow).This venule measures 86 µm in diameter(white bar in upper row).Upper row: An adherent platelet (openarrow) opposed to the vessel wall through-out the observation period (A: start ofobservation, B: end of observation). Lowerrow C, D: At a higher magnification, a rollingplatelet (filled arrow) travels with reducedspeed along the vessel wall. In-between sec-onds four and five of the observation period,a distance of 37 µm is covered (white bars).
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
580
millimeter of endothelial surface, assuming cylindrical geo-metry in the observed segment of the vessel. Rolling plateletswere quantified by number per millimeter of vessel diameter persecond.
Capillary leakageAnimals selected for this series of experiments were subjectedto 45 min intestinal ischemia. Following induction of prolongedanesthesia, injection of FITC-albumin through a venous portwas performed prior to release of blood-flow. Mice were placedon another custom made microscope stage, an upper jejunalloop was exteriorized and fixed in a superfusion chamber.Under constant superfusion, through light microscopy was usedto choose a suitable area of interest (AOI). Thereafter, the sameAOI was visualized using epi-illuminescence microscopy witha 40 x objective for 10 s every 10 min over one hour. The invert-ed microscope (Anxiovert 100, Zeiss, Thornwood, NY) wasequipped with an HBO100 mercury lamp (Zeiss) and a suitableexcitation emission filter set (Zeiss). Microscopic images werecaptured using a CCD video camera with gain and black levelcontrols (RC 300, Dage-MTI, Michigan City, IN), displayed ona 13” screen (PVM-137, Sony, San Jose, CA), and taped on ana-log S-VHS (SVO-9500 MD, Sony). For the first tapedsequence, gain and black level controls were manually adjustedand remained for the whole experiment unchanged. For evalu-ation, images were converted to digital video (DVMC-DA2,Sony), and again analyzed using the NIH Image program. Themean increase of light intensity over time in a perivascularlocated AOI (maximum distance 20 µm, Ip), measured in grayscale levels (arbitrary units, 0-255), related to mean light inten-sity in corresponding postcapillary venules (Iv) was considereda parameter of capillary albumin leakage (Ip / Iv).
Intestinal vascular permeabilityThe accumulation of Evans Blue dye in tissue specimens hasbeen closely linked to the transport of albumin across endothe-lial surfaces (22), and therefore considered a measure of vascu-lar permeability. Intestinal vascular permeability was quantifiedby intravenous administration of Evans Blue as described pre-viously (23). In brief, weight of snap frozen jejunal specimenswas determined (approximately 50 mg) and formamide (1.5 ml,Fisher Scientific, Fairlawn, NJ) extraction was performed for 2hours at 55°C. Light adsorption was read in triplicate at a wave-length of 610 nm and subtraction of reference adsorption at 450nm was performed. Data was expressed as light adsorption inarbitrary units per g of tissue weight.
Serum cytokine and VEGF levelsCommercially available ELISA kits for mouse IL-1α, IL-6,TNF-α, and VEGF were obtained from R&D Systems(Minneapolis, MN) and tests were performed according tomanufacturer’s instructions.
Tissue MPO activityTissue myeloperoxidase (MPO) activity has been considered ameasure of PMN accumulation in ischemic tissue and the MPOassay has been performed as described previously (24). Briefly,weights of snap frozen murine intestinal specimens were deter-mined (approximately 100 mg) and samples then homogenizedin 4 ml of 0.5% hexadecyltrimethylammonium bromide on ice.Homogenates were than centrifuged for 10 min (3000 rpm) at4°C. Supernatants have been transferred to fresh tubes and 100mM KH2PO4 were added to achieve a final concentration ofapproximately 1mg/ml. 10 µl of each sample were transferred toa 96-well-plate and 200 µl of Tetramethylbenzidine LiquidSubstrate System (TMB, Sigma, St. Louis, MO) were added tostart the reaction. After 30 min, reaction was stopped by adding100 µl of 0.5 M H2SO4 to each well. Readings were done at450nm. One unit of MPO activity was defined to degrade 18µmol of H2O2 per minute at 25°C (pH 6.0). All samples wereread in triplicates.
Tissue NTPDase activityHomogenates of snap frozen murine tissues were prepared andenzyme activity was determined as previously described (25,26). In brief, enzyme activity was tested in 8 mM CaCl2, 50 mMTris, and 50 mM imidazole pH 7.5. Buffer and protein samplewere preincubated at 37°C for 3 min. Thereafter, reaction wasstarted with addition of substrate (final concentration 0.3 mMATP or ADP as indicated) and terminated at 5 to 15 min with0.25 ml of malachite green reagent. To determine specific activ-ities, protein concentrations of homogenates was measuredusing the technique of Bradford (27).
H&E stainingSmall intestine specimens were fixed in neutral-buffered forma-line, and paraffin embedded. Sections for light microscopy werestained with hematoxylin and eosin.
Statistical analysisAll data are expressed as mean ± SEM. Calculations were doneusing the SPSS software package (SPSS Inc., Chicago, IL). Forstatistical analysis Mann-Whitney-U-Test, Jonckheere-Terpstra-Test, and Wilcoxon-Test were used as appropriate. Survivalrates were calculated according to the Kaplan-Meier-methodand compared using the log-rank-test.
A p-value of <0.05 was considered statistically significant.
ResultsTissue NTPDase activity and serum VEGFlevelsFollowing induction of intestinal ischemia in untreated wild-type mice, NTPDase activity in the mesenteric vasculaturedecreased to approximately half that of baseline activity
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
581
(p=0.049), but then returned to basal levels within 60 min ofreperfusion (Fig. 2).
Measurements of basal NTPDase activity in jejunal homog-enates revealed decreased activity in cd39-null mice (41.70nmol Pi/mg/min for ADP) when compared to wild-type animals(90.21 nmol Pi/mg/min, p=0.016).
Basal serum VEGF levels did not demonstrate any signifi-cant differences between cd39-null and wild-type mice (98.3 ±23.7 pg/ml vs. 74.3 ± 3.9 pg/ml, p=0.293).
Survival dataThree out of 5 wild-type mice (60%) died within 24 hours after60 min of intestinal ischemia (Fig. 3). The cd39-null mice dem-onstrated immediate onset of severe intestinal hemorrhage, and4 out of 5 of these mice (80%) had died within 48 hours after re-establishment of blood-flow.
Remarkably, NTPDase treatment resulted in survival of allwild-type mice subjected to IRI (p=0.038 vs. untreated wild-type). A single dose of NTPDase, as well as repeated injectionsevery 8 hours, failed to improve outcomes in cd39-null mice.
Combined adenosine/amrinone treatment did not augment sur-vival following intestinal IRI in either wild-type or cd39-null mice.
Four out of 5 hemizygote cd39+/–mice (80%) died within 24hours. Following NTPDase treatment there were trends toenhanced survival, but the end result was failure to improvelong-term survival.
Postmortem evaluations demonstrated major intestinal hem-orrhage only in untreated ischemic cd39-null mice. NTPDasetreatment as well as adenosine/amrinone infusion rescued cd39-null mice from significant intestinal blood loss. The pathologi-cal finding present in all fatalities was massive intestinal edema
also with evidence for distant vascular injury. Pulmonary vascu-lar injury and infarcts were found in all untreated ischemic hem-izygote cd39+/– mice, in 1 out of 4 NTPDase treated hemizy-gotes, and in 2 out of 3 cd39-null mice given a single dose ofNTPDase.
Tissue and serum markers of IRI in wild-typemiceMeasurement of serum cytokine levels (IL-1α, IL-6, and TNF-α) and tissue MPO activity failed to detect significant diffe-rences between apyrase treated and untreated ischemic wild-type mice. At 60 min post re-establishment of blood-flow,trends to increased inflammatory cytokine and MPO levelswere demonstrated for vehicle treated ischemic wild type mice(IL-1α: 148.19 ± 33.10 pg/ml, IL-6: 1889.03 ± 900.26 pg/ml,TNF-α: 125.9 ± 8.85 pg/ml, MPO: 123.2 ± 68.1 mU/g) whencompared to sham operated mice (IL-1α: 62.18 ± 3.89 pg/ml,IL-6: 189.5 ± 1.51 pg/ml, TNF-α: 61.29 ± 3.06 pg/ml, MPO:75.1 ± 32.1 mU/g). Apyrase supplementation tended to decreaselevels after IRI but these did not achieve statistical significance(IL-1α: 81.54 ± 15.48 pg/ml, IL-6: 962.13 ± 120.28 pg/ml,TNF-α: 96.22 ± 17.21 pg/ml, MPO : 109 ± 62.8 mU/g).
In terms of mucosal integrity, PMN infiltration, and intesti-nal hemorrhage, histopathological findings revealed markedlyreduced injury in jejunal specimens of NTPDase supplementedwild-type animals (Fig. 4).
Platelet-endothelial cell interactions in wild-type miceWhen injected into sham operated animals, fluorescent labeleddonor platelets demonstrated low figures of background adher-
Figure 2: NTPDase activity following intestinal ischemia.The mesenteric tissues of ischemic (IRI) or sham operated wild-type mice were harvested immediately or 60 min (+60’) after blood-flow release. NTPDase activity in tissue homogenates has been determined for substrates ADP (A) and ATP (B). In both cases, signifi-cantly decreased NTPDase activity (p=0.049) was detected immediately after ischemia, that recovers within 60 min of reperfusion.Values are given as mean and error bars demonstrate standard errors.
A) ADP B) ATP
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
582
ence to the EC surface of either intestinal arterioles or venules.A 10-fold increase could be observed in postcapillary venules ofuntreated ischemic mice, and treatment with soluble NTPDaseprior to blood-flow release prevented this increment (Fig. 5).Similar trends were observed for adherent platelets in arterioles.
Numbers of rolling platelets, however, did not significantly dif-fer in animals that underwent sham surgery or intestinal isch-emia, also apyrase treatment demonstrated no major effect (datanot shown).
Figure 3: Survival after intestinal ischemia.Survival data for wild-type (A), cd39-null (B), and cd39-hemizy-gote (C) mice after 60 min of intestinal ischemia is given.Continuous gray curves show survival figures for vehicle treatedanimals (IRI), while black dotted curves demonstrate survival inapyrase supplemented mice (IRI + apyrase). The gray dotted curvein (B) depicts the survival figure for cd39-null mice treated withrepeated administration of apyrase (IRI + rep. apyrase). Black brokencurves represent survival in adenosine/amrinone treated animals(IRI + adenosine). Each curve represents survival figure of fiveanimals. In wild-type mice, a single dose of apyrase prevented deathdue to intestinal ischemia in all mice, while 60% of vehicle treat-ed animals died within 24 hours (p=0.038). Overall survival inadenosine/amrinone treated mice did not differ from vehicle treat-ed controls, however, no animal died earlier than 24 hours postreperfusion. In cd39-null mice, neither single doses of apyrase norrepeated apyrase injections substantially altered survival figures.Adenosine/amrinone administration had no effects on survival post-IRI. In cd39-hemizygote mice, single doses of apyrase prolongedsurvival to some extent, but did not increase survival data overall.
Figure 4: Jejunal specimens after intestinalischemia-reperfusion injury.Representative jejunal sections demonstratedmajor impairments in vascular and mucosalintegrity in vehicle treated wild-type miceafter intestinal ischemia-reperfusion injury(IRI) with segmental loss of epithelium, andmajor hemorrhage. In apyrase treated ani-mals (IRI + apyrase), only minimal denudationof villi, and minor hemorrhage wereobserved. (H&E stains, magnification 250x)
IRI IRI + apyrase
A) wild-type B) cd39-null
C) cd39-hemizygote
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
583
Capillary leakage in cd39-null miceBaseline sequestration of fluorescent labeled albumin into theperivascular spaces did not differ substantially between shamoperated wild-type and cd39-null mice under conditions of con-stant superfusion with pre-warmed saline. Immediately after re-establishment of blood-flow, ischemic cd39-/- animals demon-strated only slightly increased accumulation of labeled albuminwhen compared to wild-type controls (p=0.08, Fig. 6). At 60min post blood-flow re-establishment, mean perivascular lightintensity had clearly increased in cd39-null mice (p=0.035 vs.cd39+/+) in keeping with active capillary leakage.
Intestinal vascular permeabilityWithin 60 min after injection, accumulations of the Evans Bluedye in baseline jejunal specimens were equivalent in wild-typeand cd39-null mice (21.6 OD/g vs. 25.8 OD/g, p=0.29, Fig. 7).Under experimental conditions with an open abdominal cavityduring reperfusion, sham operated cd39-null mice demonstrat-ed higher susceptibility to intestinal vascular leakage with sig-nificantly increased Evans Blue accumulation (39.7 OD/g)compared to baseline conditions (p=0.007). In wild-type mice,no comparable increases were detected. Following intestinalischemia, Evans Blue accumulation in intestinal tissue
increased significantly compared to baseline conditions ineither wild-type and cd39-null mice (p=0.002 and p=0.001,respectively). The vascular permeability detected in ischemiccd39-null mice (51.7 OD/g) was also significantly higher thanunder sham conditions (p=0.013) or in ischemic wild-type mice(44.6 OD/g, p=0.039). Measurements of Evans Blue dye inischemic intestinal tissue from either wild-type or cd39-nullmice following treatment with NTPDase or adenosine/amrinonedid not differ from baseline values. This observation indicatedthat these interventions blocked the increases in permeabilityseen in reperfusion injury.
DiscussionPreviously, we have reported on the beneficial effects of apyraseadministration in experimental kidney and cardiac transplanta-tion (17, 19). Protective effects of apyrase supplementation
Figure 5: Adherent platelets in intestinal ischemia-reperfusioninjury.Numbers of adherent platelets per mm2 of vascular surface havebeen determined utilizing intravital videomicroscopy of thejejunum. In postcapillary venules (light gray bars), a majorincrease of adherent platelets has been observed in ischemicanimals (IRI) when compared to sham operated wild-type mice(sham, p=0.025). Supplementation of apyrase (IRI + apyrase)completely prevented increased adherence of platelets inischemic venules (p=0.014 vs. IRI). Numbers of adherentplatelets in arterioles of the intestine (dark gray bars) did notdemonstrate any substantial increase after ischemia. Values aregiven as mean and error bars demonstrate standard errors.
Figure 6: Capillary leakage in intestinal ischemia-reperfusioninjury.Sequestration of fluorescent labeled albumin into perivascularspaces of the jejunum was quantified by intravital videomic-roscopy.The quotient of light intensity in perivascular tissue and postcapillary venules (Ip/Iv) provides a measure for albuminleakage. Baseline accumulation of albumin did not differ in shamoperated wild-type (wild-type sham, open squares) or cd39-nullmice (cd39-null sham, open triangle). Following moderateischemia-reperfusion injury, wild-type mice demonstrated no significant increase in albumin sequestration (wild-type IRI, solidsquare), while cd39-null mice developed major capillary leakage(cd39-null IRI, solid triangle, p=0.035 at 60 min post reperfusionvs. wild-type IRI (*)). Values are given as mean and error barsdemonstrate standard errors.
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
584
Figure 7: Vascular permeability in intestinal ischemia-reperfu-sion injury.The accumulation of Evans Blue dye in jejunal specimens har-vested 60 min following re-establishment of blood-flow wasdetermined. Following formamide extraction, light adsorptionwas read at 610 nm. Readings are expressed as optical density(OD) per gram of tissue weight. The cd39-null mice (dark graybars) demonstrated baseline levels equal to those in wild-typemice (light gray bars). Minor intestinal injury (sham operation)resulted in increased levels of Evans Blue accumulation in cd39-null mice (* p=0.007) compared to baseline measurements.Following ischemia/reperfusion and vehicle treatment, significant-ly increased levels of Evans Blue dye have been detected in bothcd39-null (* p=0.001, vs. baseline level; ~ p=0.013, vs. sham level)and wild-type mice (# p=0.002, vs. baseline level). Treatmentwith either apyrase or adenosine/amrinone prevented increasedEvans Blue levels in ischemic tissues of both murine genotypes.
have been suggested during the early phase of renal ischemia-reperfusion injury (18). In this series of experiments, we wereable to confirm improved outcomes after intestinal ischemia-reperfusion injury in apyrase treated wild-type animals,observed increased susceptibility to IRI in cd39-deficientmutant mice, and described pathophysiological findings withinthe first hour of intestinal IRI modulated by vascular NTPDasein vivo.
Microthrombus formation has been described a key featureof clinical and experimental IRI. Potential beneficial effects ofapyrase supplementation were supported by in vitro findings,demonstrating attenuated platelet aggregation when incubatedwith sera from apyrase treated animals or recombinant solubleCD39 (19, 28). In keeping with these data, apyrase treatmentprevented microthrombi formation as well as P-selectin expres-sion in cardiac xenografts (19). Similarly, our experiments uti-lizing intravital videomicroscopy revealed significantlydecreased numbers of adherent platelets in postcapillaryvenules of apyrase treated wild-type mice subjected to intestinalIRI when compared to saline treated ischemic animals.
In the same setting, survival figures have demonstrated acomplete protection from death due to intestinal IRI by a singledose of NTPDase given prior to blood-flow release in wild-typemice (survival in untreated animals: 2 out of 5). Although ben-eficial effects of adenosine administration on PMN accumula-tion and lipid peroxidation in intestinal IRI (29), as well as inischemia of other organs (30), have been reported, the experi-mental model reported here did not demonstrate any benefits ofadenosine on survival of ischemic wild-type mice.
Furthermore, no significant improvement in terms of survi-val could be determined when treating cd39-deficient mice (either cd39-null or cd39+/-) with NTPDase. In keeping withpreviously described platelet desensitization (20), untreatedcd39-null mice responded with major hemorrhage to intestinalIRI. NTPDase treatment may have resulted in platelet re-sensi-tization (31), and NTPDase treated mutant mice did not presentwith any significant hemorrhage, suggesting endothelial cellperturbation rather than platelet dysfunction. Adenosine/amri-none treatment prevented severe hemorrhage in cd39-null mice,but failed to improve survival in intestinal IRI. Findings of pul-monary vascular injury and infarction in ischemic cd39-nullmice supplemented with NTPDase or ischemic cd39-hemizy-gote mice supported the possibility that rebound platelet hyper-activity could have resulted in vascular injury. Additionalexperimental data elucidating the inhibitory role of NTPDaseon endothelial cell activation in vitro (16, 32) necessitated fur-ther study of IRI induced EC activation and changes in vascularpermeability in the presence and absence of NTPDase in vivo.
Wild-type or cd39-null mice were subjected to moderateintestinal IRI and fluorescent labeled albumin accumulation inperivascular tissue was quantified by intravital videomicros-copy. While baseline sequestration of albumin was comparablein both groups, mice lacking NTPDase activity demonstrated anoticeable perivascular accumulation within 60 min after blood-flow re-establishment indicating an increased capillary leakage.Baseline serum levels of VEGF, a potent promoter of vascularpermeability (33), were not different in wild-type and cd39-nullmice. In the setting of ischemia or hypoxia, elevated VEGFserum levels require transcription upregulation with increasedlevels described at around day 2 post injury (34). Therefore,immediate mechanisms of EC activation unrelated to VEGFhave to be evaluated more closely.
In our experimental model, intestinal vascular permeabilitysignificantly increased in cd39-null mice subjected to onlyminor injury (sham operation) compared to baseline conditions,whereas wild-type mice were less susceptible to injury.Moderate IRI, however, resulted in significantly increasedpermeability figures in both cd39-null and wild-type mice.Evans Blue sequestration was more pronounced in animalslacking NTPDase activity. Treatment with soluble NTPDaserestored permeability indices to baseline levels in both wild typeand mutant mice. The same effects were observed in adeno-
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
585
sine/amrinone treated animals. We suggest that NTPDase exertsits protective effects by not only generating high levels of aden-osine, attenuating EC activation immediately after re-establish-ment of blood-flow by A2A-receptor stimulation, but by alsoremoving proinflammatory nucleotides (14).
In the mesenteric vasculature of wild-type mice, NTPDaseactivity significantly decreased following IRI with reconstitu-tion of baseline levels within 60 min of reperfusion. We there-fore propose that pharmacological administered NTPDasebridges the initial lack of endogenous cd39. NTPDase activityfalls rapidly following the effective period of administered apy-rase in cd39-deficient mice, thus resulting in inferior outcome in terms of survival compared to wild-type animals. Thoughadministration of adenosine/amrinone mimics downstreameffects of apyrase, extracellular ATP/ADP levels may remainunchanged, accounting for failure in survival benefits.
We conclude that vascular NTPDase activity is essential formaintaining vascular integrity in the setting of IRI. The phar-macological action of NTPDase is likely exerted by preventingplatelet activation and also by attenuating EC activation.NTPDase supplementation during IRI removes intravascularATP/ADP, provides increasing amounts of adenosine. Recentreports on amrinone and adenosine receptor involvement inischemic preconditioning (35-37) or beneficial effects frompost-ischemic adenosine administration (38) provide additional
support to the possibility of NTPDase administration to attenu-ate IRI.
Evidence of increased 5’NT activity during reperfusionperiods have been published (39, 40). However, at this time IRIrelated dynamics of adenosine nucleotide/nucleoside shiftsrequire to be further explored to determine whether the failureto remove extracellular nucleotides is of more importance thanthe block in the generation of the respective nucleosides.Irrespectively, the ability of NTPDases to maintain vascularintegrity suggests pharmacological benefit of these or relatedagents in mesenteric ischemic injury.
AcknowledgementsThe authors gratefully thank Sean P. Colgan, Center for ExperimentalTherapeutics and Reperfusion Injury at Brigham and Women’s Hospital,Boston, MA for his valuable help and advice with the Evans Blue assay of vas-cular permeability.
References1. Thiagarajan RR, Winn RK, Harlan JM. The
role of leukocyte and endothelial adhesionmolecules in ischemia-reperfusion injury.Thromb Haemost 1997; 78: 310-4.
2. Milazzo VJ, Ferrante RJ, Sabido F, et al.. Timecourse of leukocyte adhesion to endotheliumin ischemia-reperfusion. J Surg Res 1996; 61:139-42.
3. Squadrito F, Altavilla D, Squadrito G, et al..The involvement of tumor necrosis factor-alpha in the protective effects of 17 betaoestradiol in splanchnic ischaemia-reperfusioninjury. Br J Pharmacol 1997; 121: 1782-8.
4. Deshmukh DR, Mirochnitchenko O, GholeVS, et al.. Intestinal ischemia and reperfusioninjury in transgenic mice overexpressing cop-per-zinc superoxide dismutase. Am J Physiol1997; 273: C1130-5.
5. Rose S, Floyd RA, Eneff K, et al.. Intestinalischemia: reperfusion mediated increase inhydroxyl free radical fromation as reported by salicylate hydroxylation. Shock 1994; 1:452-6.
6. Frenette PS, Moyna C, Hartwell DW, et al..Platelet-endothelial interactions in inflamedmesenteric venules. Blood 1998; 91: 1318-24.
7. Massberg S, Enders G, Leiderer R, et al..Platelet-endothelial cell interactions duringischemia/reperfusion: the role of P-selectin.Blood 1998; 92: 507-15
8. Cywes R, Packham MA, Tietze L, et al.. Therole of platelets in hepatic allograft preserva-tion injury in the rat. Hepatology 1993; 18:635-47.
9. Lefer AM, Lefer DJ. Pharmacology of theendothelium in ischemia-reperfusion and cir-culatory shock. Annu Rev Pharmacol Toxicol1993; 33: 71-90.
10. Marcus AJ, Broekman MJ, Drosopoulos JHF,et al.. Thromboregulation by endothelial cells- Significance for occlusive vascular diseases.Arterioscler Thromb Vasc Biol 2001; 21: 178-82.
11. Robson SC, Kaczmarek E, Siegel JB, et al..Loss of ATP diphosphohydrolase activity withendothelial cell activation. J Exp Med 1997;185: 153-63.
12. Dubiak GR, El-Moatassim C. Signal transduc-tion via P2-purinergic receptors for extracellu-lar ATP and other nucleotides. Am J Physiol1993; 265: C577.
13. Seligmann C, Kupatt C, Becker BF, et al..Adenosine endogenously released duringearly reperfusion mitigates postischemicmyocardial dysfunction by inhibiting plateletadhesion. J Cardiovasc Pharmacol 1998; 32:156-63.
14. Burnstock G. Purinergic signaling and vascu-lar cell proliferation and death. ArteriosclerThromb Vasc Biol 2002; 22: 364-73.
15. Body SC. Platelet activation and interactionswith the microvasculature. J CardiovascPharmacol 1996; 27 (Suppl 1): S13-25.
16. Goepfert C, Imai M, Brouard S, et al.. CD39modulates endothelial cell activation andapoptosis. Mol Med 2000; 6: 591-603.
17. Candinas D, Koyamada N, Miyatake T, et al..Loss of rat glomerular ATP diphosphohydro-lase activity during reperfusion injury is asso-ciated with oxidative stress reactions. ThrombHaemost 1996; 76: 807-12.
18. Imai M, Takigami K, Guckelberger O, et al..Modulation of nucleotide triphosphate diphosphohydrolase-1 (NTPDase-1) cd39 inxenograft rejection. Mol Med 1999; 5: 743-52.
19. Koyamada N, Miyatake T, Candinas D, et al..Apyrase administration prolongs discordantxenograft survival. Transplantation 1996; 62:1739-43.
20. Enjyoji K, Sevigny J, Lin Y, et al.. Targeteddisruption of cd39/ATP diphosphohydrolaseresults in disordered hemostasis and throm-boregulation. Nat Med 1999; 5: 1010-7.
21. Kauffman RF, Schenck KW, Utterback BG, etal.. In vitro vascular relaxation by newinotropic agents: relationship to phosphodi-esterase inhibition and cyclic nucleotides. JPharmacol Exp Ther 1987; 242: 864-72.
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
Guckelberger, et al.: NTPDases in intestinal ischemia-reperfusion
586
22. Fry DL, Mahley RW, Weisgraber KH, et al..Simultaneous accumulation of Evans blue dyeand albumin in the canine aortic wall. Am JPhysiol 1977; 233: H66-79.
23. Barone GW, Farley PC, Conerly JM, et al..Morphological and functional techniques forassessing endothelial integrity: the use ofEvans blue dye, silver stains, endothelialderived relaxing factor. J Card Surg 1989; 4:140-8.
24. Mullane KM, Kraemer R, Smith B.Myeloperoxidase activity as a quantitativeassessment of neutrophil infiltration intoischemic myocardium. J Pharmacol Methods1985; 14: 157-67.
25. Sévigny J, Levesque FP, Grondin G, et al..Purification of the blood vessel ATP diphos-phohydrolase, identification and localizationby immunological techniques. BiochimBiophys Acta 1997, 1334: 73-88.
26. Baykov AA, Evtushenko OA, Avaeva SM. Amalachite green procedure for orthophosphatedetermination and its use in alkaline phos-phatase-based enzyme immunoassay. AnalytBiochem 1988; 171: 266-70.
27. Bradford MM. A rapid and sensitive methodfor quantification of microgram quantities of protein utilizing the principle of protein-dye binding. Analyt Biochem 1976; 72: 248-54.
28. Gayle RB III, Maliszewski CR, Gimpel SD, etal.. Inhibition of platelet function by recombi-nant soluble ecto-ADPase/CD39. J Clin Invest1998; 101: 1851-9.
29. Unal S, Demirkan F, Arslan E, et al..Comparison of ischemic and chemical precon-ditioning in jejunal flaps in the rat. PlastReconstr Surg 2003; 112: 1024-31.
30. Linden J. Molecular approach to adenosinereceptors: receptor-mediated mechanisms oftissue protection. Annu Rev PharmacolToxicol 2001; 41: 775-87.
31. Robson SC, Sevigny J, Imai M, et al..Thromboregulatory potential of endothelialCD39/nucleoside triphosphate diphosphohy-drolase: modulation of purinergic signalling inplatelets. Emerging Therapeutic Targets 2000;4: 155-71.
32. Imai M, Goepfert C, Kaczmarek E, et al..CD39 modulates IL-1 release from activatedendothelial cells. Biochem Biophys ResCommun 2000; 270: 272-8.
33. Josko J, Gwozdz B, Jedrzejowska-SzypulkaH, et al.. Vascular endothelial growth factor(VEGF) and its effect on angiogenesis. MedSci Monit 2000; 6: 1047-52.
34. Xu F, Severinghaus JW. Rat brain VEGFexpression in alveolar hypoxia: possible rolein high-altitude cerebral edema. J ApplPhysiol 1998; 85: 53-7.
35. Nilsson B, Friman S, Wallin M, et al.. Theliver protective effect of ischemic precondi-tioning may be mediated by adenosine.Transpl Int 2000; 13 (Suppl 1): S558-561.
36. de Jonge R, de Jong JW, Giacometti D, et al..Role of adenosine and glycogen in ischemicpreconditioning of rat hearts. Eur J Pharmacol2001; 414: 55-62.
37. Lee HT, Emala CW. Protective effects of renalischemic preconditioning and adenosine pre-treatment: role of A1 and A3 receptors. Am JPhysiol Renal Physiol 2000; 278: F380-F387.
38. Sakata C, Tanaka H, Takemura S, et al.. Post-ischemic intraportal adenosine administrationprotects against reperfusion injury of canineliver. J Hepatobiliary Pancreat Surg 2000; 7:78-85.
39. Sakaguchi T, Sawa Y, Kitakaze M, et al.. Ecto-5’-nucleotidase plays a role in the cardiopro-tective effects of heat shock protein 72 inischemia-reperfusion injury in rat hearts.Cardiovasc Res 2000; 47: 74-80.
40. Frassetto SS, Schetinger MR, Schierholt R, etal.. Brain ischemia alters platelet ATP diphos-phohydrolase and 5’-nucleotidase activities innaïve and preconditioned rats. Braz J MedBiol Res 2000; 33: 1369-77.
For personal or educational use only. No other uses without permission. All rights reserved.Downloaded from www.thrombosis-online.com on 2014-12-30 | ID: 1000472022 | IP: 128.103.149.52
