adenosine mediates hypoxic induction of vascular

Adenosine mediates hypoxic induction of vascular endothelial growth factor in retinal pericytes and endothelial cells

CitationTakagi, H., King, G.L., Robinson, G.S., Ferrara, N., Aiello, L.P. 1996. Adenosine Mediates Hypoxic Induction of Vascular Endothelial Growth Factor in Retinal Pericytes and Endothelial Cells. Invest. Ophthalmol. Vis. Sci. 37 (11): 2165-2176.

Permanent linkhttp://nrs.harvard.edu/urn-3:HUL.InstRepos:33776256

Terms of UseThis article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA

Share Your StoryThe Harvard community has made this article openly available.Please share how this access benefits you. Submit a story .

Accessibility

http://nrs.harvard.edu/urn-3:HUL.InstRepos:33776256

http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA

http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA

http://osc.hul.harvard.edu/dash/open-access-feedback?handle=&title=Adenosine%20mediates%20hypoxic%20induction%20of%20vascular%20endothelial%20growth%20factor%20in%20retinal%20pericytes%20and%20endothelial%20cells&community=1/4454685&collection=1/4454686&owningCollection1/4454686&harvardAuthors=8ebf93e5874781145f87dcbca5e93810&department

https://dash.harvard.edu/pages/accessibility

Adenosine Mediates Hypoxic Induction of VascularEndothelial Growth Factor in Retinal Pericytes andEndothelial Cells

Hitoshi Takagi* George L. King,*-fX Gregory S. Robinson,§ Napoleone Ferrara,\\and Lloyd Paul

Purpose. To determine the mechanistic role for adenosine and adenosine receptors in thehypoxic induction of vascular endothelial growth factor (VEGF) in retinal microvascular cells.Methods. Bovine retinal capillary endothelial cells and microvascular pericytes were studiedunder normoxic (95% air, 5% CO2) or hypoxic conditions (0% to 2% O2, 5% CO.,, 93% to95% Nj.) using a variety of well-characterized adenosine and adenosine receptor agonists andantagonists. Vascular endothelial growth factor mRNA expression was evaluated by Northernblot analysis, VEGF protein levels were determined by Western blot analysis, and cyclic adeno-sine monophosphate (cAMP) accumulation was measured by radioimmunoassay.Results. Inhibitors of oxidative respiration increased VEGF mRNA 5 ± 3 times (P < 0.001)after 3 hours. Adenosine A, receptor (A,R) agonist N('-cyclopentyl-adenosine did not increaseVEGF mRNA at A|R stimulatory concentrations; however, adenosine A., receptor (Av.R) ago-nists DPMA, NECA, and CGS21680 increased VEGF mRNA in a dose-dependent manner withelevations of 2 ± 0.3 (P < 0.001), 2.3 ± 0.5 (P = 0.016), and 2 ± 0.2 (P = 0.002) times,respectively. A;>R antagonist CSC and adenosine degradation by adenosine deaminase reducedhypoxic stimulation of VEGF mRNA 68% ± 18% (P = 0.038) and 37% ± 6% (P = 0.025),respectively, in a dose-dependent manner. A!R antagonists DPCPX and 8-PT had no significanteffect. Hypoxia and NECA increased VEGF protein secretion 4.7 times, whereas CSC inhibitedhypoxia-induced VEGF protein secretion by 96%. NECA and CGS21680 increased cAMPproduction within 10 minutes, and cAMP stimulation increased VEGF mRNA 4.8 ± 2.6 times(P = 0.034). CSC suppressed the hypoxic elevation of cAMP (P< 0.05). Inhibition of proteinkinase A using H-89 reduced hypoxia-induced VEGF expression 61 % ± 6.3% (P = 0.043) ina dose-dependent manner.

Conclusions. These data suggest that the hypoxia-induced accumulation of adenosine stimulatesVEGF gene expression through stimulation of adenosine A.J;1 receptor and subsequent activa-tion of the cAMP-dependent protein kinase A pathway in retinal vascular cells. Invest Ophthal-mol Vis Sci. 1996; 37:2165-2176.

Vascular endothelial growth factor (VEGF) is an en-dothelial cell mitogen1 2 and vasopermeability factor13

whose expression is increased dramatically in vivo andin many types of cultured cells, including retinal vascu-lar cells.'1"7 Vascular endothelial growth factor exerts

From tlu: * Research Division and the \Beetha:m Eye Institute, Joslin Diabetes Center,"[Harvard Medical School; %Defiartment of Medicine, Brigham & Women'sHospital, Boston; ^Research Division, Hybridan Inc., Worcester, Massachusetts; and\\Cenentech Inc., San Francisco, California.Supported in part l/y National Institutes of Health grants EY09178 (CLK) andEY10827 (I. JJA), Juvenile Diabetes Foundation International Research grant194102, and the V. Rasmussen Foundation (LPA). The Joslin Diabetes Center isthe. recipient of National Institutes of Health Diabetes and Endocrinology ResearchCenter Grant. 36836.Submitted for publication February 22, 1996; revised. May 22, 1996; accepted June21, 1996.'Proprietary interest, category: N.Reprint requests: Lloyd Paul Aiello, Beet ham Eye Institute, /oslin Diabetes Center,One Joslin Place, Boston, MA 02215.

its action on endothelial cells through high-affinity,endothelial-specific, autophosphorylating tyrosine ki-nase receptors.8 ' Elevated levels of VEGF have beenassociated with tumor neovascularization,7 develop-mental angiogenesis,10 and ischemia-induced prolifer-ative ocular diseases, such as diabetic retinopathy andcentral retinal vein occlusion.'1"

Recently accumulating evidence suggests that thehypoxic induction of VEGF gene expression appearsto be regulated by transcriptional activation and in-creased mRNA stability.12"15 Cis elements and transfactors are reported to affect the hypoxic regulationof VEGF. A second-messenger pathway, such as theprotein kinase C pathway,11' the cyclic adenosinemonophosphate (cAMP)-dependent pathway,"1 and

Investigative Ophthalmology & Visual Science, October 199(5, Vol. S7, No.Copyright © Association for Research in Vision and Ophthalmology 2165

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/Journals/IOVS/933414/ on 08/24/2016

2166 Investigative Ophthalmology & Visual Science, October 1996, Vol. 37, No. 11

the Ca2+ ion channel,17 can regulate VEGF gene ex-pression. Activation of c-Src has been reported to bea regulator of hypoxic VEGF induction.18 However,the mechanism by which hypoxia induces cells to in-crease VEGF expression is not well understood.

Adenosine is released by hypoxic tissues, and con-centrations of this naturally occurring nucleoside areelevated (>30 times) in hypoxic hepatocytes, primar-ily as a result of decreased recycling by adenosine ki-nase.19 In addition, adenosine is released from, andinduces the proliferation and migration of, chick as-trocytes, human astrocytoma cells,20 and human um-bilical vein endothelial cells.21 Erythropoietin, a mole-cule whose expression is increased by hypoxia22'23 andwhose regulation is analogous to that of VEGF,24 issimilarly stimulated by adenosine through A2R.25

Adenosine is a metabolic precursor for nucleicacid biosynthesis and a regulator of numerous mam-malian systems, including neurologic, cardiovascular,renal, respiratory, immunologic, gastrointestinal, andmetabolic.26 Physiologic responses include coronaryvasodilatation, renal vasoconstriction, inhibition ofplatelet aggregation, sedation, and reduced cardiacrate and contractile force.27 The majority of adenosineeffects are mediated through cell surface receptorsthat characteristically bind adenosine more than nu-cleotides, are selectively blocked by methylxanthines,couple to adenyl cyclase, and do not alter prostaglan-din synthesis.26 High concentrations of adenosine,however, can inhibit adenyl cyclase direcdy throughan intracellular "P-site." Adenosine receptors actthrough guanosine triphosphate binding proteins,coupling not only to adenyl cyclase but also to ionchannels and phospholipases.28 Adenosine receptorshave seven transmembrane helical structures29 and arecategorized as A|, A2a, A2b, A3, and possibly other A,subtypes by pharmacologic interaction with a varietyof agonists and antagonists.2627 Adenosine A,- and A3

receptor binding decreases cAMP levels, whereas A2

receptor binding increases these same levels.26'27 Fouradenosine receptor subtypes have been cloned,26 andnumerous species homologues have been identified.Recendy, adenosine has been shown to stimulateVEGF expression in U-937 mononuclear cells30 andpig cerebral endothelial cells,31 although the mecha-nism of stimulation and mediating receptors appearto differ.

Because adenosine is released from hypoxic tissueand can stimulate endothelial cell proliferation, weinvestigated the possibility that the hypoxia-inducedincrease in VEGF expression is mediated by adenosineinteraction with its receptors on retinal vascular cells.Our results demonstrate that hypoxia-induced adeno-sine release results in adenosine A2 receptor stimula-tion and activation of the cAMP-dependent proteinkinase A (PKA) pathway, which together account for

much of the hypoxic induction of VEGF in retinalmicrovascular endothelial cells and pericytes.

MATERIALS AND METHODS

Chemicals and Antibodies

Carbonyl cyanide p-(trifluoromethoxy) phenyl-hydra-zone (FCCP) and N6,2'-O-dibutyryladenosine 3':5'-cy-clic monophosphate were obtained from Sigma (St.Louis, MO). N6-[2-(3,5-Dimethoxypheynl)-2-(2-meth-ylphenyl)-ethyl] adenosine (DPMA),8-cyclopentyl-l,3-dipropylxanthine (DPCPX), 8-(3-chlorostyryl)-caffeine(CSC), 8-phenyltheophylline (8-PT), 5'-(N-ethylcar-boxamido)-adenosine (NECA), 2-^(2-carboxyethyl)-phenethyl-amino-5'-N-ethylcarbox-amidoadenosine(CGS21680), and N6-cyclopentyl-adenosine (CPA)were obtained from Research Biochemicals Interna-tional (Natick, MA). N-[2-((3(4-bromophenyl)-2-pro-penyl)-amino)-ethyl]-5-isoqinolinesulfonamide (H89)and RO-20-1724 were purchased from Calbiochem(San Diego, CA). Cyclic adenosine monophosphateradioimmunoassay kits were obtained from Dupont-New England Nuclear (Boston, MA), 32P-dATP wasacquired from Amersham (Arlington Heights, IL),and adenosine deaminase was obtained from Worthin-gton Biochemicals (Freehold, NJ). Recombinant hu-man vascular endothelial growth factor and humanVEGF cDNA was provided by Genentech (San Fran-cisco, CA).

Cell Cultures

Bovine retinal endothelial cells and retinal pericyteswere isolated from slaughterhouse eyes by homogeni-zation and a series of filtration steps as previously de-scribed.32 Primary endothelial cell cultures similarlywere obtained and grown on fibronectin (NYBen Re-agents; New York Blood Center, New York, NY)-coateddishes (Costar, Cambridge, MA) containing Dulbec-co's modified Eagle's medium (DMEM) with 5.5 mMglucose, 10% plasma- derived horse serum (Wheaton,Pipersville, PA), 50 mg/1 heparin, and 50 U/l endo-thelial cell growth factor. The cells were cultured in5% CO2 at 37°C, and media were changed every 3days. Endothelial cell homogeneity was confirmed byreactivity with anti-factor VIII antibodies. Retinal peri-cytes were isolated as described above and cultured inDMEM with 5.5 mM glucose and 20% fetal bovineserum (Hyclone, South, UT) on fibronectin-coateddishes. Homogeneity of retinal pericytes was con-firmed by reactivity with monoclonal antibody 3G5.32

Retinal endothelial cells and retinal pericytes wereevaluated in these studies, each yielding similar re-sults. Data from experiments with a single cell type,as described in the text, are presented.


Adenosine Mediates Hypoxic Retinal Induction of VEGF

1 \lM FCCP 100 |iM FCCP

1

m m

W

m

-< 28S ><VEGF •

-< 18S >

< 36B4 >* *

600

0 0.5 3 < hours > 0 1 3

FCCPJIuM) FCCP (100uM)

Hours

FIGURE l. Inhibition of electron transport stimulates vascularendothelial growth factor (VEGF) mRNA expression. Con-fluent monolayers of retinal endothelial cells were treatedwith 1 fiM or 100 /xM FCCP under normoxic conditions forthe duration indicated. Total RNA was isolated and North-ern blot analysis performed as described in Materials andMethods. Representative Nordiern blots using VEGF cDNAand control 36B4 probe {to})) and quantitation of multipleexperiments after normalization to the control signal areshown (bottom). Results are expressed as percent of controlVEGF mRNA expression ± standard error.

Hypoxia Studies

Confluent cell monolayers were exposed to oxygenconcentrations noted in the text using a Lab-Line Ad-vanced Computer Controlled Infrared Water-JacketedCO;; Incubator with reduced oxygen control (model480; Lab-Line, Melrose Park, IL). All cells were main-tained at 37°C in a constant 5% carbon dioxide atmo-sphere, with oxygen deficit induced by nitrogen re-placement. Cells under these conditions showed nomorphologic changes by light microscopy after expo-sure periods exceeding 72 hours, excluded TrypanBlue dye (>98%), and subsequently could be pas-saged normally. Cells incubated under standard cul-ture conditions from the same batch and passage were

2167

used as controls (95% air, 5% CO^). Throughout thisarticle, normoxic refers to standard cell culture condi-tions (95% air, 5% COj), whereas hypoxia refers tolower oxygen concentrations (2% and 0.5% in theseexperiments) than present under standard conditions(21% oxygen). It should be noted that "normoxic"culture conditions are hyperoxic compared to physio-logic conditions. However, the 2% and 0.5% oxygenconcentrations used in these studies are relatively hyp-oxic compared to control conditions, cultured retinalcells respond to hypoxia similarly below 5% oxygen,'1

and 0.5% oxygen corresponds to <20 mm Hg. whichis physiologically hypoxic.

RNA Isolation and Northern Blot Analysis

Total RNA was extracted from individual P-100 tissueculture plates using guanidium thiocyanate." Radio-active probes were generated using AmershamMultiprime labeling kits and ;v~P-dATP. Northern blotanalysis was performed using ICN Biochemicals (Ir-vine, CA) Biotrans nylon membranes, ultraviolet cross-linking (UV Stratalinker 2400; Stratagene, La Jolla,CA), and a rotating hybridization oven (model 400;Robbins Scientific, Sunnyvale, CA). Analysis was per-formed with a Molecular Dynamics Computing Phos-phorlmager (Mountain View, CA). Lane-loading dif-ferences were normalized using 36B4 control cDNA/'1

Graphs of Northern blot quantitation represent re-sults from at least three experiments, with each pointwithin each experiment performed in triplicate. Eachlane was normalized to its own 36B4 control signal.

Cyclic Adenosine MonophosphateAccumulation

Cells grown in six-well cluster dishes were washed twicewith phosphate-buffered saline and were incubatedwith adenosine receptor subtype-specific ligands at theindicated concentration in 1 ml of serum-free DMEMin the presence of 300 fM RO-20-1724 for 10 minutes.The incubation was terminated by aspirating the me-dia and the addition of 1 ml ice-cold 6% trichloroace-tic acid. After a 20-minute incubation on ice, die cellswere scraped off the plate, sonicated, and stored at—20°C. The sample was thawed and centrifuged at2000g-for 15 minutes at 4°C, the supernatant was ex-tracted five times with 2 vol water-saturated ethylether,and the remaining ether was evaporated to dryness.The dried samples were resuspended in 50 mM so-dium acetate (pH 6.2) and assayed for cAMP by radio-immunoassay using a cAMP assay kit from New En-gland Nuclear.

Vascular Endothelial Growth Factor ProteinSecretion

Either CSC (10 ^M), NECA (1 ^uM), or phosphate-buffered saline was added to 90% confluent cultures



<28S< VEGF

<28SVEGF

36B4

0 0.01 0.1 1.0 10 50 <CPA(uM)

FIGURE 2. Stimulation of adenosine A( receptor does not induce vascular endothelial growthfactor (VEGF) mRNA expression. Normoxic bovine retinal pericytes were exposed to increas-ing concentrations of CPA and VEGF mRNA levels evaluated by Northern blot analysis after4.5 hours (A). A[R are significantly more sensitive to CPA than are A2R (Ki = 1 nM and680 nM, respectively). The time course of VEGF expression in retinal pericytes exposed to50 fjM CPA is presented in B. Representative Northern blots and quantitation of multipleexperiments after normalization to the control signal are shown at the top and bottom ofeach panel, respectively.

of bovine retinal pericytes for 1 hour. The culturemedia then were replaced with 5 ml of labeling media(3.75 ml DMEM minus methionine and cysteine, 1.25ml DMEM with 20% fetal bovine serum, 200 fid MS-Translabel [ICN, Costa Mesa, CA]) supplementedwith CSC or NECA as described above. CSC-treatedand -untreated cells were exposed to hypoxic condi-tions, whereas NECA-treated and -untreated cells wereexposed to normoxic conditions for 17 hours beforecollection and centrifugation of the media. Equal me-dia protein (as determined by equal trichloroaceticacid precipitable counts) were immunoprecipitatedovernight at 4°C using a polyclonal rabbit anti-humanVEGF antibody (a generous gift from Kevin Claffey,Department of Pathology, Beth Israel Hospital, Bos-ton, MA). The antibody-VEGF complexes were pel-leted using protein A sepharose (Pharmacia, Uppsala,Sweden), washed 3X (10 mM Tris, pH 8, 140 mMNaCI, 0.1% Triton X-100, 0.1% bovine serum albumin,

and 0.01% sodium azide), and resuspended in 2Xsodium dodecyl sulfate-polyacrylamide gel electro-phoresis (SDS-PAGE) loading buffer containing 7mM dithiothreitol. Samples were separated on a5.5%-12.5% SDS polyacrylamide gel, enhanced usingEntensify solution (New England Nuclear), and dried.Results were quantitated using a phosphoimager (Mo-lecular Dynamics, Mountainview, CA).

Statistical AnalysisAll determinations were performed in triplicate, andexperiments were repeated at least three times. Re-sults are expressed as the mean ± standard deviation.Statistical analysis used the paired Student's /-test tocompare quantitative data populations with normaldistributions and equal variance. Data were analyzedusing the Mann-Whitney Rank Sum Test for popula-tions with nonnormal distributions or unequal vari-ance. Data within multiple sets were compared to a


Adenosine Mediates Hypoxic Retinal Induction of VEGF 2169

11 11 1IMP*'

•

1If

•< 18S

< JGB4

10 100 <DPMA(nM.

VEGF

36B4

NECA

- 7 «

i

DPMA (nM)

I1

1•

i VEGF*

y36B4>- • »

0.01 0.1 10 CGS21680(MM)

i 300

NECA (nM)

FIGURE 3. Adenosine A2 receptor agonists stimulate vascular endothelial growth factor(VEGF) mRNA expression. Retinal pericytes were exposed to the indicated concentrationsof DPMA (A), NECA (B), and CGS21680 (C) for 4.5 hours under normoxic conditions,and Northern blot analysis was performed. A representative Northern blot {top) and quantita-tion of multiple experiments after normalization to the control signal {bottom) are shown.

0.01 0.) 10

CCS2l680(nM)

single control group. Validity of the statistical analysiswas confirmed with one-way analysis of variance(Kruskal-Wallis/Dunn's method for nonnormal dataand Tukey test for normal data). All statistical conclu-sions remained valid.

RESULTS

Inhibition of Electron Transport StimulatesVascular Endothelial Growth Factor mRNAExpression

To determine whether the mechanism of hypoxia-in-duced VEGF mRNA induction was dependent on theabsence of oxygen itself, we examined whether artifi-cial inhibition of oxidative phosphorylation in a nor-moxic environment could mimic the VEGF mRNAregulation observed under hypoxic conditions. FCCPinhibits mitochondrial oxidative phosphorylation atits most distal site, preventing the electron transportfrom cytochrome aa3 to oxygen.35 Thus, this com-pound more closely mimics an oxygen deficit than doproximal transport inhibitors such as rotenone andantimycin A. When confluent normoxic retinal endo-thelial cells were exposed to either 1 /xM or 100 fiMFCCP, VEGF mRNA levels increased 2.4 ± 0.8 times(P = 0.024) and 5 ± 0.3 times (P < 0.001) within 3hours, respectively (Fig. 1). These data suggest thatthe mechanism of hypoxia-induced VEGF mRNA in-

duction is not dependent on the absence of oxygenitself but might involve byproducts of an inactive mito-chondrial respiratory system or effects from depletedenergy stores within the cell.

Adenosine Agonists Stimulate VascularEndothelial Growth Factor mRNA Expression

It has been reported that hypoxia reduces adenosinekinase-mediated recycling of adenosine to adenosinemonophosphate, resulting in dramatic elevations of in-tracellular adenosine.10 These levels of adenosine arecapable of mediating physiologic effects through bind-ing to various adenosine receptors. Because hypoxic tis-sues release adenosine and adenosine induces prolifera-tion and migration of endothelial cells,21 we investigatedwhether adenosine stimulation could increase the ex-pression of VEGF mRNA. Stable adenosine receptor ago-nists were used to test his hypothesis because adenosineitself degrades rapidly in culture, and a steady mediaconcentration cannot be assured. CPA is a stable adeno-sine Ai receptor (A,R) agonist at low concentrations (Ki= 1 nM) and both an A,R and an A«R (Ki = 680 nM)agonist at high concentrations/*017 Stimulation of nor-moxic retinal pericytes with CPA did not affect VEGFmRNA expression at levels up to 100 nM (Fig. 2A) butdid increase VEGF mRNA expression at higher concen-trations (P= 0.005). High-dose CPA (50 fiM) increasedVEGF mRNA levels 2 ± 0.1 times {P < 0.001), with


2170 Investigative Ophthalmology &; Visual Science, October 1996, Vol. 37, No. 11

* P < 0.05** P < 0.01

***P< 0.001

10 100

0 5 1DPCPX 8PT

Ligand Concentration (uM)CSC

FIGURE4. Adenosine antagonists and adenosine A2 receptorantagonists inhibit hypoxia-induced vascular endothelialgrowth factor (VEGF) mRNA expression. Immediately be-fore the initiation of hypoxia (2% O2, 5% CO2, 93% Na),retinal pericytes were treated with the indicated dose ofeach compound. After 12 hours, Northern blot analysis wasperformed. Quantitation of multiple experiments after nor-malization to the control signal are shown. All doses areexpressed in //M except those for adenosine deaminase(AdD), which is expressed in U/ml.

consistent induction of expression noted within 3 hours(Fig. 2B).

To determine whether adenosine induced VEGFexpression through specific effects on A^R, we stimu-lated normoxic retinal endothelial cells with the AaRagonists DPMA, NECA, and CGS21680. Each com-pound increased VEGF mRNA expression in a dose-dependent manner at concentrations of 10 nM andhigher (Fig. 3). Vascular endothelial growth factormRNA expression was increased 2 ± 0.3 (P < 0.001),2.3 ± 1.1 (P = 0.016), and 2 ± 0.2 times (P = 0.002)after stimulation with DPMA (100 nM), NECA (1 fM),and CGS21680 (1 |iM) respectively. Vascular endothe-lial growth factor stimulation by the A2;iR specific ago-nist CGS21680 was not statistically different from thestimulation observed with the nonsubtype specific AjRagonist NECA.

Adenosine Antagonists Inhibit Hypoxia-inducedVascular Endothelial Growth Factor mRNAExpression

To determine whether the hypoxic VEGF responsecould be reduced by blocking adenosine receptor acti-vation, we studied the effects of A|R antagonistsDPCPX and 8-phenyltheophylline (8-PT), the A2R an-tagonist CSC, and the adenosine-degrading enzyme

FIGURE 5. Adenosine agonists stimulate cyclic adenosinemonophosphate (cAMP) synthesis in retinal cells. cAMP ac-cumulation was measured by radioimmunoassay in bovineretinal pericytes after the application of adenosine agonistsat the indicated concentrations for 10 minutes in the pres-ence of the cAMP phosphodiesterase inhibitor RO-20-1724.cAMP accumulation is expressed as increase over unstimu-lated cells. Each point represents the mean ± SD of threetriplicate experiments.

cO"55 2 -

NA

E>

t of

cc

E§LL. <S

800

700

600

500

400

300

200

100

P = 0.034

1ft1<VEGF

O6B4

0 6hr

0Hours

FIGURE 6. Cyclic adenosine monophosphate (cAMP) analogsstimulate vascular endothelial growth factor (VEGF) mRNAexpression. Confluent retinal pericytes were exposed to 50pM dibutyryl cAMP for 6 hours, and VEGF mRNA expres-sion was evaluated by Northern blot analysis. Quantitationof five experiments after normalization to the control signalis expressed as the mean ± SD. Northern analysis is shownin the insert.



adenosine deaminase. Both DPCPX and 8-PT had lit-tle effect on hypoxia-induced (2% O2, 5% CO2, 93%N2) expression of VEGF at A)R inhibitory concentra-tions (Fig. 4). Adenosine deaminase reduced VEGFexpression 37% ± 6% (P = 0.025). Blocking A2;>R(ICr)0 = 54 nM) with CSC inhibited hypoxia-inducedVEGF expression by 68% ± 18% (P = 0.038) in a dose-dependent manner at levels not expected to affect AiR(IC.,() = 28AIM).35-36

Adenosine Agonists Stimulate Cyclic AdenosineMonophosphate Synthesis

If adenosine mediates hypoxia-induced VEGF produc-tion, then retinal pericytes and endothelial cellsshould express adenosine receptors. As previouslymentioned, A,R stimulation decreases cAMP whereasA2R stimulation increases it.2h'27 We measured cAMPaccumulation in normoxic bovine retinal pericytesafter the application of adenosine agonists for 10 min-utes in the presence of the cAMP phosphodiesteraseinhibitor RO-20-1724. Both A2R agonists NECA andCGS21680 increased cAMP accumulation in a dose-dependent manner, with consistent increases noted atconcentrations greater than 1 to 10 nM (Fig. 5). NECAand CGS 21680 increased cAMP levels by 2.4 ± 0.9(P = 0.008) and 2.9 ± 0.5 times (P < 0.001) at aconcentration of 100 nM. At low (A,R effective) con-centrations, stimulation of AtR with CPA did not re-duce cAMP accumulation after cAMP stimulation withforskolin (data not shown).

Cyclic Adenosine Monophosphate AnalogsStimulate Vascular Endothelial Growth FactormRNA Expression

Because adenosine receptor agonists increased cAMPlevels and VEGF mRNA expression in retinal cells, wedetermined whether cAMP could induce VEGF ex-pression. Confluent normoxic retinal pericytes wereexposed to 50 (JM of the cAMP analog dibutyryl cAMP,and VEGF mRNA expression was analyzed by North-ern blot analysis. VEGF mRNA levels increased 4.8 ±2.6 times (P = 0.034) after 6 hours (Fig. 6).

Hypoxia-induced Cyclic AdenosineMonophosphate Synthesis Is Suppressed byAdenosine A2 Receptor Inhibition

Because A2R stimulation increases cAMP, we studiedwhether hypoxia could stimulate cAMP synthesis inretinal cells. In addition, to determine whether theobserved hypoxia-induced cAMP increase resultedfrom the stimulation of A2;1R, we studied the effect ofan A2;|R antagonist. Bovine retinal endothelial cellswere incubated in 0.5% or 21% oxygen for 4 hourswith or without 10 //M of the A2;,R antagonist CSC.Cyclic adenosine monophosphate levels were in-creased nearly 50% (P < 0.01) in cells exposed to

30

20

10

0

CSC (10nM)

21% O2 0.5% O2

FIGURE 7. Adenosine Aj receptor antagonist inhibits hyp-oxia-stimulated cAMP synthesis. Bovine retinal endothelialcells were incubated in 0.5% or 21% oxygen for 4 hours withor withovit 10 iM CSC, and the cAMP level was measuredby radioimmunoassay. cAMP accumulation is expressed aspmol/mg protein, and each point represents the mean ±SD of three triplicate experiments.

hypoxic conditions compared to those exposed to nor-moxic conditions (Fig. 7). This hypoxia-inducedcAMP elevation was inhibited 69% by CSC (P < 0.05);CSC had no significant effect on cells incubated undernormoxic conditions.

VEGF Protein Secretion Is Stimulated byAdenosine Receptor Agonists and Hypoxia-induced VEGF Protein Expression Is Inhibitedby A2aR AntagonistsTo determine whether VEGF protein levels respondedto adenosine in a manner similar to VEGF mRNA,VEGF protein was immunoprecipitated from mediaconditioned by retinal pericytes exposed to H5S-methi-onine under normoxic or hypoxic (0.5% O2, 5% CO2,94.5% N2) conditions for 17 hours, and SDS-PAGEwas performed (Fig. 8). In the media, VEGF proteinwas increased 4.6 times by 17 hours of hypoxia. Expo-sure to 1 /iM NECA under normoxic conditions alsoincreased VEGF protein concentrations 4.7 times.However, pretreatment with the A2aR antagonist CSCfor 1 hour before 17 hours of hypoxia inhibited 96%of the hypoxia-induced increase in VEGF protein.Protein Kinase A Inhibition SuppressesHypoxic Induction of VEGF mRNABecause cAMP has been well documented to stimulatePKA and because we have demonstrated that cAMP



Hypoxia (17 hr) + - +NECA (1 uM) . +CSC (10 uM) - - _ +

FIGURE 8. Adenosine agonists stimulate vascular endothelialgrowth factor (VEGF) protein expression, and A^R antago-nists prevent hypoxia-induced VEGF protein expression.Compounds were added to confluent cultures of retinal per-icytes in the presence of ;ir>S-methionine for 17 hours asindicated. Cells receiving CSC were pretreated with the drugfor 1 hour before induction of hypoxia. The media werethen immunoprecipitated using anti-VEGF antibody, run onSDS-PAGE, and analyzed by phosphoimager. Equal proteinloading was confirmed using trichloroacetic acid precipita-ble counts. Relative trichloroacetic acid-normalized VEGFexpression was 1, 4.6, 4.7, and 1.1 for left to right lanes,respectively. Similar results have been obtained from threeexperiments.

analogs stimulate VEGF mRNA expression, we investi-gated whether PKA is involved in the hypoxic responseof VEGF. Bovine retinal pericytes and endothelial cellswere exposed to 21% or 0.5% oxygen for 12 hours inthe presence of various concentrations of PKA inhibi-tor H-89. H-89 inhibited the hypoxic stimulation ofVEGF mRNA in a dose-dependent manner, achievinga 61% ± 6.3% (P = 0.043) suppression at 20 //M(Fig. 9).

DISCUSSION

Recent studies have demonstrated that the expressionof the angiogenic protein VEGF is increased greatlyin ocular cells on exposure to hypoxic conditions invitro4-38 and in vivo/><3-"-3t> Retinal microvascular endo-thelial cells and pericytes are among the cell typeswith strong VEGF induction that would experiencehypoxia in ischemic retinal disorders.4 In addition,the intraocular concentration of VEGF in patients withdiabetes mellitus is correlated closely with the pres-ence of active proliferative diabetic retinopathy,h andthe inhibition of VEGF reduces ischemia-induced reti-nal40 and iris" neovascularization in animals. Themechanism by which hypoxia induces VEGF expres-sion, however, remains incompletely understood.

We have investigated the mechanism of hypoxicVEGF induction in retinal microvascular endothelialcells and pericytes because of their known respon-siveness and their key roles in intraocular neovascu-larization. Although it is not known which ocular cell

types contribute most to physiologically relevant VEGFexpression, endothelial cells and the intimately endo-thelial cell-associated pericyte could have a substantialimpact on endothelial responses caused by either au-tocrine mechanisms or high local concentrations. Ourinvestigations demonstrate that retinal vascular cellsincrease VEGF mRNA and protein levels after adeno-sine receptor stimulation under normoxic conditions.Conversely, adenosine receptor inhibition reduces theinduction of VEGF mRNA and protein expressionwhen cells are exposed to hypoxic conditions. Indeed,the magnitude of VEGF protein stimulation producedby hypoxia can be achieved by physiologically relevantconcentrations of adenosine, and adenosine receptorinhibition prevents nearly all the hypoxia-induced in-

VEGF

36B4;

Hypoxia >•

H89 (|iM) *• 20+0

+0.1

+5

+20

0.1 20

H89

FIGURE 9. Protein kinase A inhibition suppresses hypoxic-induction of vascular endothelial growth factor (VEGF)mRNA. The PKA inhibitor H-89 was added to confluentmonolayers of retinal bovine pericytes at the indicated con-centrations, and its effect on hypoxic stimulation (0.5% oxy-gen for 12 hours) of VEGF mRNA expression was evaluatedby Northern blot analysis. Representative Northern blotsusing VEGF cDNA and control 36B4 probe {top) and quanti-tation of multiple experiments after normalization to thecontrol signal are shown (bottom). Values are expressed asthe percent of observed stimulation compared to untreatedcells, and they represent the mean ± SD of three triplicateexperiments.



crease in VEGF protein. These data suggest that themajority of hypoxia-induced VEGF expression may bemediated by the increased concentrations of adeno-sine present under hypoxic conditions.1'1

In the current study, we have demonstrated thatVEGF gene expression in retinal pericytes and endo-thelial cells is stimulated by the inhibition of electrontransport at its terminal step and does not require theabsence of oxygen itself. Similar findings in cardiacmyocytes were reported by Levy et al17 using amobarbi-tal and rotenone, which act at complex I of the elec-tron transport chain. Surprisingly, in Levy's study,FCCP, which blocks more distally in the electron trans-port chain, did not induce VEGF gene expression.The differences in the FCCP response may result fromthe differences in cell types, dose, or duration usedin these experiments. Further studies are necessary todetermine whether a specific portion of the electrontransport chain is particularly involved in hypoxicVEGF stimulation.

Comparing the effectiveness of the various adeno-sine receptor agonists and antagonists suggests that aparticular adenosine receptor subtype may mediatethe hypoxic VEGF response. CPA demonstrated littleeffect below 100 nM, whereas maximal VEGF mRNAstimulation occurred at 10 fiM. Because the Ki of CPAis approximately 1 nM for A,R and 680 nM for A2R,H<1 s7

the observed dose response suggests that VEGF induc-tion is most likely mediated through A2R. However,DPMA increased VEGF expression at concentrationsas low as 5 nM. Because the Ki of DPMA is 142 nMfor A|R and 4.4 nM for A2R,42 these data again suggestthat the VEGF response is mediated by the A2R. Simi-larly, NECA and CGS21680 increased VEGF mRNA atadenosine A2R-stimulating concentrations.'13 BecauseVEGF stimulation by the A2;,R specific agonistCGS21680 was not statistically different from the stim-ulation observed with the nonsubtype-specific A2R ag-onist NECA, a predominant role for A2;,R is suggested.However, because VEGF stimulation by NECA wasslightly (but nonsignificantly) greater than that ob-served for CGS21680, a contributing role for A2bR can-not be ruled out. The hypothesis of adenosine media-tion through A2;|R is supported further by the effective-ness of low-dose A2;|R antagonists, but not AiRantagonists, at inhibiting the hypoxia-induced eleva-tion of VEGF mRNA. A2hR-specific antagonists are notcurrently available.

Although theoretically the Ki derived in vitro foreach drug can be significantly different from that ob-served in cellular systems, the close correlation be-tween expected and observed values using multipleagents strongly suggests that the majority of the adeno-sine response is mediated through the A2R and proba-bly specifically through the A2;1R. Indeed, Hashimotoet al*0 have reported that hypoxic induction of VEGF

expression in mononuclear U-937 cells is mediatedthrough adenosine A2 receptor. However, they did notdifferentiate A2;|- from A2h receptor involvement. Incontrast with these studies, Fischer et alM reportedthat A, rather than A2 receptor is involved in the hyp-oxic VEGF induction in pig cerebral capillary endo-thelial cells. A| receptor-mediated expression, how-ever, appears not to be predominant in retinal vascu-lar cells because we could not see A! receptorstimulation by the inhibition of adenylate cyclase inthese cells. Again, different species and cell typesmight account for these differences.

Adenosine A| receptor binding decreases cAMP,whereas A2R receptor binding increases cAMP.2<)'27

Our finding that A2R agonists increase cAMP in retinalcells and that A,R stimulation does not reduce for-skolin-stimulated levels (data not shown) suggests thatfunctional A2R exists on retinal pericytes and endothe-lial cells, whereas the AiR subclass is either absent,present in low numbers, or relatively inactive biologi-cally. Our finding that hypoxia-induced cAMP accu-mulation is blocked by A2R antagonists and that cAMPcan increase VEGF mRNA levels further supports therole of A2R in mediating the hypoxic response. Similarincreases in cAMP production in response to hypoxiahave been observed in isolated guinea pig heart,'1'1 al-though bovine aortic and pulmonary artery endothe-lial cells have been reported to decrease cAMP levelsunder hypoxic conditions in vitro.'15

Although the manner in which cAMP increasesVEGF mRNA remains unknown, our data suggest thatcAMP stimulation of PKA may be involved becausePKA-specific blocker H89 inhibited hypoxic VEGF in-duction in a dose-dependent manner. Indeed, theVEGF promoter contains two potential AP-2 bindingsites at nucleotides -135 to -128 and -1875 to— 1868, which would indicate that VEGF expressioncould be increased by cAMP stimulation of PKA.'"' Infact, the 5' flanking region, which includes the latterAP-2 binding site, has been reported to contain anenhancer element for hypoxic VEGF induction.12 Incontrast, other reports suggest no involvement of this5' flanking region,1'1 and indeed this region is notconserved in ratgenomic sequences.13 It is still contro-versial whether the PKA pathway is involved in hypoxicVEGF induction. Blocking PKA in cardiac myocytesusing KT5720 failed to inhibit hypoxia-induced VEGFexpression.17 Forskolin and rolipram, which areknown to stimulate the cAMP-PKA pathway, did notincrease VEGF expression in one study,M whereasstimulation of the PKA pathway has increased VEGFexpression in others."'™ Some reports have suggesteda predominant role for the PKC pathway because hyp-oxia increases membrane-associated PKC'17 and thetranscriptional factors, fos and jun.'is AP-1 bindingsites, however, are reported not to be regulated by


2174 Investigative Ophthalmology 8c Visual Science, October 1996, Vol. 37, No. 11

hypoxia.13 These diverse findings demonstrate thecomplexity of the molecular mechanisms involved inthe hypoxic induction of VEGF. Thus, confirmationof PKA involvement awaits further investigation.

Hypoxic regulation of VEGF shares several similar-ities with other hypoxia-inducible proteins, such aserythropoietin (Epo)22~24 and GLUT-1.15 The steadystate mRNA expression of Epo and VEGF are in-creased by hypoxia and cobalt chloride, whereas it isinhibited by carbon monoxide. This suggests that thesensor regulating the expression of these moleculesmight be a heme protein. Epo has as elements andtrans factors involved in its hypoxic regulation. Indeed,VEGF has been reported to have a 5'-flanking en-hancer element homologous to the erythropoietinhypoxia-responsive enhancer.13 A homologous 3'flanking enhancer also is reported to be a cis-actingelement for hypoxic induction,12 although this regionis reported to be a 3' untranslated region by anothergroup.13 The other 5' flanking element (a potentialbinding site for the transcriptional factor, Spl) is re-ported to be a cis-acting element for hypoxic VEGFinduction.14 Thus, it is possible that hypoxic expres-sion of VEGF in retinal cells may be regulated by fac-tors that interact at several locations in response tothe adenosine-mediated stimulation of A2R. In addi-tion, the hypoxic induction of VEGF expression ap-pears to be regulated by mRNA stability, as is GLUT-l.13"15 The 3' untranslated region of these mRNA con-tain the destabilizing AUUUA motif in the context ofan AU rich region. It also is possible that binding ofregulatory proteins to these 3'UTR regions may occurin response to adenosine A2R stimulation and mayresult in stabilization of the VEGF mRNA.

Taken together, currently available data suggestthe following mechanism for the development of neo-vascularization in diabetic retinopathy and other isch-emic retinal disorders. The retinal capillary loss ornonperfusion characteristic of all these diseases resultsin relative hypoxia that causes an elevation of adeno-sine levels primarily caused by decreased recycling byadenosine kinase.19 Adenosine binds to adenosine re-ceptors located on hypoxia-responsive VEGF-produc-ing cells, such as retinal pericytes,4 endothelial cells,4'49

Miiller cells,5 and pigment epithelial cells.4'50 Bindingof adenosine, probably to the A^R, induces adenylcyclase-mediated cAMP elevation by G protein cou-pling.26 Cyclic AMP may then activate PKA, which in-duces VEGF expression through an as yet uncharacter-ized signal transduction pathway. The soluble isoformsof VEGF51 are then free to diffuse within the eye andbind to retinal endothelial cells through autophosph-orylating VEGF receptors.9'49 Vascular endothelialgrowth factor receptor stimulation leads to autophos-phorylation and possible phosphorylation of interme-diate signaling molecules,9 eventually resulting in in-

creased endothelial cell mitogenesis4'5'9'49 and perme-ability.1'3

In conclusion, prior evaluation of the adenosineinduction of VEGF expression has been performed inmononuclear and pig endothelial cells in which theinvolved mechanisms appear to differ. Our findingsprovide the first mechanistic information regardingthe hypoxic stimulation of VEGF in retinal vascularcells. In addition, because the inhibition of VEGF hasbeen shown to suppress VEGF-mediated retinal41 andiris angiogenesis42 in animals, these data suggest thatadenosine A2 receptor antagonists might prove effec-tive inhibitors of ischemia-induced intraocular neovas-cularization.

Key Words

adenosine, adenosine receptor, cyclic adenosine monophos-phate (cAMP)-dependent protein kinase A, hypoxia, isch-emia, neovascularization, vascular permeability factor (VPF)

Acknowledgments

The authors thank Dr. Lloyd M. Aiello, Joan C. Taylor, Dr.Jerry D. Cavallerano, Dr. Kevin Claffey, and Susan Rook fortheir assistance.

References

1. Berse B, Brown LF, Van dWL, Dvorak HF, SengerDR. Vascular permeability factor (vascular endothelialgrowth factor) gene is expressed differentially in nor-mal tissues, macrophages, and tumors. Mol Biol Cell.1992;3:211-220.

2. Leung DW, Cachianes G, Kuang WJ, Goeddel DV,Ferrara N. Vascular endothelial growth factor is a se-creted angiogenic mitogen. Science. 1989; 246:1306-1309.

3. Senger DR, Perruzzi CA, Feder J, Dvorak HF. A highlyconserved vascular permeability factor secreted by avariety of human and rodent tumor cell lines. CancerRes. 1986; 46:5629-5632.

4. Aiello LP, Northrup JM, Keyt BA, Takagi H, IwamotoMA. Hypoxic regulation of vascular endothelialgrowth factor in retinal cells. Arch Ophthalmol. 1995;113:1538-1544.

5. Pierce EA, Avery RL, Foley ED, Aiello LP, Smith LEH.Vascular endothelial growth factor/vascular perme-ability factor expression in a mouse model of retinalneovascularization. Proc Natl Acad Sci USA. 1995;92:905-909.

6. Aiello LP, Avery RL, Arrigg PG, et al. Vascular endo-thelial growth factor in ocular fluid of patients withdiabetic retinopathy and other retinal disorders [seecomments]. NEnglJMed. 1994;331:1480-1487.

7. Shweiki D, Itin A, Softer D, Keshet E. Vascular endo-thelial growth factor induced by hypoxia may mediatehypoxia-initiated angiogenesis. Nature. 1992; 359:843-845.

8. Ferrara N, Houck KA, Jakeman LB, Leung DW. Molec-ular and biological properties of the vascular endothe-



lial growth factor family of proteins. Endocr Rev.1992; 13:18-32.

9. Thieme H, Aiello LP, Takagi H, Ferrara N, King GL.Comparative analysis of vascular endothelial growthfactor receptors on retinal and aortic vascular endo-thelial cells. Diabetes. 199544:98-103.

10. Millauer B, Wizigmann VS, Schnurch H, Martinez R,Moller NP, Risau W, Ullrich A. High affinity VEGFbinding and developmental expression suggest Flk-1as a major regulator of vasculogenesis and angiogene-sis. Cell. 1993; 72:835-846.

11. Adamis AP, Miller JW, Bernal MT, et al. Increasedvascular endothelial growth factor levels in the vitre-ous of eyes with proliferative diabetic retinopathy. AmJ Ophthalmol. 1994; 118:445-450.

12. Minchenko A, Salceda S, Bauer T, Caro J. Hypoxiaregulatory elements of the human vascular endothe-lial growth factor gene. Cell MolBiol Res. 1994; 40:35-39.

13. Levy AP, Levy NS, Wegner S, Goldberg MA. Transcrip-tional regulation of the rat vascular endothelialgrowth factor gene by hypoxia. J Biol Chem. 1995;270:13333-13340.

14. Ikeda E, Achen MG, Breier G, Risau W. Hypoxia-in-duced transcriptional activation and increased mRNAstability of vascular endothelial growth factor in C6glioma ceWs.JBiol Chem. 1995;270:19761-19766.

15. Stein I, Neeman M. Stabilization of vascular endothe-lial growth factor mRNA by hypoxia and hypoglycemiaand coregulation with other ischemia-induced genes.Mol Cell Biol. 1995; 15:5363-5368.

16. Claffey KP, Wilkison WO, Spiegelman BM. Vascularendothelial growth factor: Regulation by cell differen-tiation and activated second messenger pathways. JBiol Chem. 1992;267:16317-16322.

17. Levy AP, Levy NS, Loscalzo J, et al. Regulation of vascu-lar endothelial growth factor in cardiac myocytes. CircRes. 1995; 76:758-766.

18. Mukhopadhyay D, Tsiokas L, Zhou XM, Foster D,Brugge JS, Sukhatme VP. Hypoxic induction of hu-man vascular endothelial growth factor expressionthrough c-Src activation. Nature. 1995; 375:577-581.

19. Bontemps F, Vincent MF, Van den Berghe G. Mecha-nisms of elevation of adenosine levels in anoxic hepa-tocytes. J Biochem. 1993; 290:671-677.

20. Rathbone MP, Middlemiss PJ, Kim JK, et al. Adenosineand its nucleotides stimulate proliferation of chickastrocytes and human astrocytoma cells. Nucleosci Res.1992;13:1-17.

21. Ethier MF, Chander V, DobsonJG. Adenosine stimu-lates proliferation of human endothelial cells in cul-ture. AmJ Physiol. 1993;265:H131-H138.

22. Beru N, McDonald J, Lacombe C, Goldwasser E. Ex-pression of the erythropoietin gene. Mol Cell Biol.1986; 6:2571-2575.

23. Bondurant MC, Koury MJ. Anemia induces accumula-tion of erythropoietin mRNA in the kidney and liver.Mol Cell Biol. 1986;6:2731-2733.

24. Goldberg MA, Schneider TJ. Similarities between theoxygen-sensing mechanisms regulating the expression

of vascular endothelial growth factor and erythropoie-tin./£?o/ Chem. 1994; 269:4355-4359.

25. NakashimaJ, Ohigashi T, BrookinsJW, Beckman BS,Agrawal KC, Fisher JW. Effects of 5'-N-ethylcarboxami-deadenosine (NECA) on erythropoietin production.Kidney Int. 1993; 44:734-740.

26. Tucker AL, Linden J. Cloned receptors and cardiovas-cular responses to adenosine. Cardiovasc Res. 1993;27:62-67.

27. Collis MG, Hourani SMO. Adenosine receptor sub-types. Trends Pharmacol Sci Rev. 1993; 14:360-366.

28. Linden J. Structure and function of Al adenosine re-ceptors. FASEBJ. 1991;5:2668-2676.

29. Libert F, Parmentier M, Lefort A. Selective amplifica-tion and cloning of four new members of the G-pro-tein coupled receptor family. Science. 1989; 244:569-572.

30. Hashimoto E, Kage K, Ogita T, Nakaoka T, MatsuokaR, Kira Y. Adenosine as an endogenous mediator ofhypoxia for induction of vascular endothelial growthfactor mRNA in U-937 cells. Biochem Biophys Res Corn-mun. 1994; 204:318-324.

31. Fischer S, Sharma HS, Karliczek GF, Schaper W. Ex-pression of vascular permeability factor/vascular en-dothelial growth factor in pig cerebral microvascularendothelial cells and its upregulation by adenosine.Brain Res Mol Brain Res. 1995; 28:141 -148.

32. Nayak RC, Berman AB, George KL, Eisenbarth GS,King GL. A monoclonal antibody (3G5)-defined gan-glioside antigen is expressed on the cell surface ofmicrovascular pericytes. / Exp Med. 1988; 167:1003-1015.

33. Chomczynski P, Sacchi N. Single-step method of RNAisolation by acid guanidinium thiocyanate-PhOH-chloroform extraction. Anal Biochem. 1987; 162:156-159.

34. Masiakowski P, Breathnach R, Bloch J, Gannon F,Krust A, Chambon P. Cloning of cDNA sequencesof hormone-regulated genes from the MCF-7 humanbreast cancer cell lines. Nud Acids Res. 1982; 10:7895-7903.

35. Sanders SP, Zweier JL, Kuppusamy SL, et al. Hyper-oxic sheep pulmonary microvascular endothelial cellsgenerate free radicals via mitochondrial electrontransport. / Clin Invest. 1993;91:46-52.

36. Bruns RF, Lu GH, Pugsley TA. Characterization of theA2 adenosine receptor labeled by [3H]NECA in ratstriatal membranes. Mol Pharmacol. 1986;29:331-346.

37. Research Biochemicals International 1995 Handbook.Available from RBI, Strathmore Road, Natick, MA01760-2447.

38. Adamis AP, Shima DT, Yeo KT, et al. Synthesis andsecretion of vascular permeability factor/vascular en-dothelial growth factor by human retinal pigment epi-thelial cells. Biochem Biophys Res Commun. 1993;193:631-638.

39. Miller JW, Adamis AP, Shima DT, et al. Vascular endo-thelial growth factor/vascular permeability factor istemporally and spatially correlated with ocular angio-genesis in a primate model. Am J Pathol. 1994;145:574-584.



40. Aiello LP, Pierce EA, Foley ED, et al. Inhibition ofvascular endothelial growth factor suppresses retinalneovascularization in vivo. Proc Natl Acad Sci USA.1995;92:10457-10461.

41. Adamis AP, Shima DT, Tolentino MJ, et al. Inhibitionof vascular endothelial growth factor prevents retinalischemia-associated iris neovascularization in a non-human primate. Arch Ophthalmoi 1996; 114:66-71.

42. Bridges AJ, Bruns RF, Ortwine DF, Priebe SR, SzotekDL, Trivedi BK. N6-[2-(3,5-dimethoxyphenyl)-2-(2-methylphenyl) ethyl] adenosine and its uronamide de-rivatives: Novel adenosine agonists with both high af-finity and high selectivity for the adenosine A2 recep-tor. JMed Chem. 1988; 31:1282-1285.

43. Jarvis MF, Schulz R, Hutchison AJ, Do UH, Sills MA,Williams M. [3H]CGS 21680, a selective A2 adenosinereceptor agonist directly labels A2 receptors in ratbrain. /Pharmacol Exp Ther. 1989;251:888-893.

44. Kroll K, Schrader J. Myocardial adenosine stimulatesrelease of cyclic adenosine monophosphate from cap-illary endothelial cells in guinea pig heart. PflugersArch. 1993; 423:330-337.

45. Ogawa S, Koga S, Kuwabara K, et al. Hypoxia-inducedincreased permeability of endothelial monolayers oc-curs through lowering of cellular cAMP levels. Am JPhysiol. 1992;262:C546-C554.

46. Tischer E, Mitchell R, Hartman T, et al. The humangene for vascular endothelial growth factor: Multipleprotein forms are encoded through alternative exonsplicing. JBiol Chem. 1991;266:11947-11954.

47. Wieloch T, Cardell M, Bingren H, Zivin J, Saitoh T.Changes in the activity of protein kinase C and thedifferential subcellular redistribution of its isozymesin the rat striatum during and following transient fore-brain ischemia. J Neurochem. 1991;56:1227-1235.

48. Webster KA, Discher DJ, Bishopric NH. Induction andnuclear accumulation of fos andjun proto-oncogenesin hypoxic cardiac myocytes. / Biol Chem. 1993;268:16852-16858.

49. Simorre PV, Guerrin M, Chollet P, et al. Vasculo-tropin-VEGF stimulates retinal capillary endothelialcells through an autocrine pathway. Invest OphthalmoiVis Set. 1994;35:3393-3400.

50. Jacobson KA, Nikodijevic C, Padgett WL, Gallo-Ro-driguez C, Maillard M. 8-(3-chlorostyryl)caffeine(CSC) is a selective A2-adenosine antagonist in vitroand in vivo. FEBS Lett. 1993;323:141-144.

51. Houck KA, Leung DW, Rowland AM, Winer J, FerraraN. Dual regulation of vascular endothelial growth fac-tor bioavailability by genetic and proteolytic mecha-nisms. JBiol Chem. 1992;267:26031-26037.


adenosine mediates hypoxic induction of vascular

Documents