Zachary Anita K. Stannard, Rohit Khurana, Ian M. Evans, Vassiliki ...

ZacharyAnita K. Stannard, Rohit Khurana, Ian M. Evans, Vassiliki Sofra, David I.R. Holmes and Ian

αTumor Necrosis Factor-Vascular Endothelial Growth Factor Synergistically Enhances Induction of E-Selectin by

Print ISSN: 1079-5642. Online ISSN: 1524-4636 Copyright © 2006 American Heart Association, Inc. All rights reserved.

Greenville Avenue, Dallas, TX 75231is published by the American Heart Association, 7272Arteriosclerosis, Thrombosis, and Vascular Biology

doi: 10.1161/01.ATV.0000255309.38699.6c2006;

2007;27:494-502; originally published online December 14,Arterioscler Thromb Vasc Biol.

http://atvb.ahajournals.org/content/27/3/494World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://atvb.ahajournals.org/content/suppl/2006/12/15/01.ATV.0000255309.38699.6c.DC1.htmlData Supplement (unedited) at:

http://atvb.ahajournals.org//subscriptions/

at: is onlineArteriosclerosis, Thrombosis, and Vascular Biology Information about subscribing to Subscriptions:

http://www.lww.com/reprints

Information about reprints can be found online at: Reprints:

document. Question and AnswerPermissions and Rightspage under Services. Further information about this process is available in the

which permission is being requested is located, click Request Permissions in the middle column of the WebCopyright Clearance Center, not the Editorial Office. Once the online version of the published article for

can be obtained via RightsLink, a service of theArteriosclerosis, Thrombosis, and Vascular Biologyin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on February 23, 2013http://atvb.ahajournals.org/Downloaded from

http://atvb.ahajournals.org/content/27/3/494

http://atvb.ahajournals.org/content/suppl/2006/12/15/01.ATV.0000255309.38699.6c.DC1.html

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://atvb.ahajournals.org//subscriptions/

http://atvb.ahajournals.org/

Vascular Endothelial Growth Factor SynergisticallyEnhances Induction of E-Selectin by Tumor

Necrosis Factor-�

Anita K. Stannard, Rohit Khurana, Ian M. Evans, Vassiliki Sofra, David I.R. Holmes, Ian Zachary

Objective—The regulation of endothelial cell adhesion molecules (CAMs) by vascular endothelial growth factor (VEGF)was investigated in cell cultures and in a rabbit model of atherogenic neointima formation.

Methods and Results—VEGF regulation of vascular CAM-1 (vascular cell adhesion molecule), intercellular CAM-1(intercellular adhesion molecule), and E-selectin were investigated in human umbilical vein endothelial cells usingquantitative polymerase chain reaction, enzyme-linked immunosorbent assay, and flow cytometry, and in the rabbitcollar model of atherogenic macrophage accumulation by immunostaining. VEGF alone caused no significant inductionof vascular cell adhesion molecule-1, intercellular adhesion molecule-1, or E-selectin compared with tumor necrosisfactor-�. In both hypercholesterolemic and normal rabbits, adenoviral VEGF-A165 expression caused no increase inendothelial vascular cell adhesion molecule-1 or E-selectin. In contrast, pretreatment of human umbilical veinendothelial cells with VEGF significantly increased E-selectin expression induced by tumor necrosis factor-�, comparedwith tumor necrosis factor-� alone, whereas vascular cell adhesion molecule-1 and intercellular adhesion molecule-1were unaffected. VEGF similarly enhanced IL-1�–induced E-selectin upregulation. VEGF also synergistically increasedtumor necrosis factor-�–induced E-selectin mRNA and shedding of soluble E-selectin. Synergistic upregulation ofE-selectin expression by VEGF was mediated via VEGF receptor-2 and calcineurin signaling.

Conclusions—VEGF alone does not activate endothelium to induce CAM expression; instead, VEGF “primes” endothelialcells, sensitizing them to cytokines leading to heightened selective pro-inflammatory responses, including upregulationof E-selectin. (Arterioscler Thromb Vasc Biol. 2007;27:494-502.)

Key Words: cell adhesion molecules � endothelial cells � IL-1�

The endothelium has several important functions thatinclude providing a nonadhesive, nonthrombotic barrier

between the blood and the underlying tissues. In atheroscle-rosis, or in response to injury or inflammatory cytokines suchas tumor necrosis factor-� (TNF-�), the endothelium be-comes activated and cell adhesion molecules (CAMs) arerapidly induced.1,2 In particular, members of the immuno-globulin superfamily of CAMs, such as intercellular celladhesion molecule-1 (ICAM-1) and vascular cell adhesionmolecule-1 (VCAM-1), as well as the selectin family mem-bers, E-selectin and P-selectin, have crucial roles in theadhesion and migration of monocyte/macrophage infiltrationinto atherosclerotic lesions during the early and subsequentstages of atherosclerosis in a variety of animal models.1–7

Recent findings suggest that the angiogenic cytokine,vascular endothelial growth factor (VEGF or VEGF-A) mayact as a proinflammatory cytokine by increasing ICAM-1,VCAM-1, and E-selectin mRNA in cultured endothelial cellsthrough activation of nuclear factor-�B (NF-�B),8,9 an im-

portant transcription factor mediating CAM gene expression.So far, however, the relevance of VEGF-induced surfaceCAM expression for endothelial function has not been clar-ified. Given that VEGF is both a major focus of interest as atherapeutic angiogenic cytokine for coronary artery disease,and, paradoxically, has been reported to enhance atheroscle-rotic lesion formation,10–13 elucidation of the effects of VEGFon endothelial CAM expression has important practical, aswell as biological, implications.

VEGF elicits an array of biological effects on endothelialcells in vivo and in vitro including survival, proliferation andmigration, nitric oxide and prostacyclin (PGI2) production,and increased vascular permeability.14,15 VEGF exerts itsactions by binding two cell surface protein kinase receptors,VEGF receptor (VEGFR)2/KDR/Flk-1 and VEGFR1/Flt-1.Most biological responses triggered by VEGF in endothelialcells are mediated primarily by VEGFR2 and activation ofmultiple early signaling cascades, including phosphatidylino-sitol 3�-kinase (PI3K)-dependent Akt/PKB, phospholipase

Original received August 2, 2006; final version accepted November 23, 2006.From BHF Laboratories, Department of Medicine, University College London, United Kingdom.Correspondence to Ian Zachary, BHF Laboratories, Department of Medicine, The Rayne Building, University College London, 5 University Street,

London WC1E 6JJ, United Kingdom. E-mail [email protected]© 2007 American Heart Association, Inc.

Arterioscler Thromb Vasc Biol. is available at http://www.atvbaha.org DOI: 10.1161/01.ATV.0000255309.38699.6c

494 by guest on February 23, 2013http://atvb.ahajournals.org/Downloaded from


C-�, mitogen-activated protein/extracellular signal-regulatedkinase and calcineurin/nuclear factor of activated T-cellspathways.13–17

In the present study, we investigated the ability of VEGF tostimulate CAM expression in human umbilical vein endothe-lial cells (HUVECs) and in vivo. VEGF induced minorchanges in CAM mRNA, which were insufficient to increaseCAM expression at the cell surface as determined by sensi-tive flow cytometry, although TNF-� strongly induced CAMmRNA and cell surface expression. Adenoviral overexpres-sion of VEGF also had no effect on endothelial VCAM-1 andE-selectin expression in a rabbit model of neointima forma-tion and neointimal macrophage accumulation. Surprisingly,when HUVECs were pre-incubated with VEGF beforeTNF-� treatment, E-selectin mRNA and cell surface expres-sion were synergistically increased as compared with TNF-�alone, whereas other CAMs were unaffected. VEGF alsosynergistically enhanced TNF-�–induced shedding of solubleE-selectin. These findings demonstrate that VEGF is not ableto stimulate endothelial CAM expression alone, but interactswith TNF-� to synergistically and selectively enhanceE-selectin expression.

Experimental ProceduresMaterials and MethodsFor sources of cytokines, antibodies, and other reagents see supple-mental experimental procedures (http://atvb.ahajournals.org).

Cell CultureHUVECs were cultured as described online in the supplement.17

Collar Placement and Adenoviral Gene TransferAnimal experiments were conducted in accordance with the UKAnimals (Scientific Procedures) Act 1986 and the Animal Care andEthics Guidelines of University College London, UK. Maintenanceof New Zealand White rabbits on a high-cholesterol diet, collarplacement around the carotid artery, and delivery of adenoviruses(Ark Therapeutics Ltd, Kuopio, Finland) encoding VEGF-A165

(Ad.VEGF-A165) or LacZ (Ad.LacZ) were performed as described.11

RNA Isolation, Reverse-Transcription PolymeraseChain Reaction and Quantitative Real-TimePolymerase Chain ReactionTotal RNA isolation from rabbit carotid arteries and HUVECs,reverse-transcription polymerase chain reaction, and quantitativereal-time polymerase chain reaction were performed as described(supplemental Table I, available online at http://atvb.ahajournals.org).11,17,18

LacZ Staining and ImmunohistochemistryDetection of �-galactosidase and immunostaining of VCAM-1,E-selectin, macrophages, CD31, and VEGF were performed asdescribed.11

Morphometry and Image AnalysisIntima/media ratios, and staining of CD31 (endothelial neovascular-ization), RAM-11 (macrophages), VCAM-1, and E-selectin in col-lared rabbit carotid arteries were quantified as previously describedusing Image J.11,18

Enzyme-Linked Immunosorbent AssayAnalysis of total (cell surface and intracellular) VCAM-1 expressionwas performed as described.19 E-selectin, ICAM-1, and von Wille-brand factor were analyzed by substituting anti-VCAM-1 with

anti-E-selectin (5 �g IgG1/mL), anti-ICAM-1 (1 �g IgG1/mL), oranti- von Willebrand factor (1.2 �g IgG1/mL).

Flow CytometryCell surface VCAM-1, E-selectin, ICAM-1, and PECAM-1 wasmeasured in confluent HUVECs as described.19

Soluble E-selectin ImmunoassayHuman soluble E-selectin (sE-selectin) was measured by commercialimmunoassay in HUVEC culture supernatants.

Statistical AnalysisDifferences between different treatment groups in rabbits wereevaluated by ANOVA and Bon Ferroni correction (SPSS). Statisticalanalysis of cell culture experiments was performed using Student ttest or 1-way or 2-way ANOVA with post hoc analysis by FisherPLSD, when appropriate. Results are shown as means�SE ormean�SD, and P�0.05 was considered significant.

ResultsVEGF Does Not Upregulate Endothelial ProteinExpression of E-selectin, ICAM-1, or VCAM-1The effects of VEGF on CAM mRNA levels were measuredin HUVECs by quantitative real-time reverse-transcriptionpolymerase chain reaction and normalized to GAPDH ex-pression (Figure 1A). Tissue factor (TF) mRNA, previouslyshown to be markedly upregulated by VEGF,17 was increased�70-fold 90 minutes after addition of VEGF and returned tonear control levels by 24 hours. VEGF increased E-selectin,VCAM-1, and ICAM-1 mRNA expression to maximumlevels of, respectively, 7-fold after 45 minutes, 29-fold after 3hours, and 2-fold after 1.5 hours (Figure 1A). Levels ofPECAM-1 (CD31) mRNA, constitutively expressed byHUVECs, were unchanged by VEGF incubation (data notshown). To assess the biological relevance of VEGF-inducedincreases in CAM expression, we compared the effects ofVEGF with those of the inflammatory cytokine, TNF-�. Theresults of this comparison showed that the modest inductionsof CAMs by VEGF were negligible compared with the stronginduction of E-selectin, ICAM-1, and VCAM-1 by TNF-�alone: �900-, 300-, and 800-fold increases in mRNA, respec-tively (Figure 1B).

Because VEGF has previously been reported to stimulateprotein expression of CAMs in endothelial cells,8,9 we exam-ined the possibility that VEGF might increase CAM expres-sion despite the relatively small effect of VEGF on mRNAinduction compared with TNF-�. Although TNF-� treatmentinduced increases in E-selectin (Figure 1C), ICAM-1 (Figure1D), and VCAM-1 (Figure 1E) of 6-fold, 5.6-fold, and 2-foldabove basal unstimulated levels, respectively, VEGF failed toinduce any significant increase in CAM expression after anyof the incubation times tested (Figure 1C through 1E). In eachexperiment, von Willebrand Factor, which is known to beincreased by VEGF,20 was significantly increased after 24-hour incubation with VEGF by 44.3�16.3% (P�0.05, n�3,results not shown). Measurement of cell surface CAMs usingflow cytometry confirmed the results of enzyme-linked im-munosorbent assays, and was unable to detected a subpopu-lation of cells with increased CAM expression in response toVEGF; fluorescence-activated-cell sorter profiles providedclear evidence that CAM expression on the cell surface was

Stannard et al VEGF Synergistically Upregulates E-Selectin 495



not increased by VEGF, whereas TNF-� strongly upregulatedCAM expression (data not shown).

Effect of VEGF on VCAM-1 and E-selectinExpression in a Model of AtherogenicMacrophage AccumulationBecause of the key role thought to be played by upregulationof endothelial VCAM-1 in neointimal macrophage infiltra-tion during the early stages of atherosclerosis, the effects ofhigh-efficiency adenoviral VEGF-A165 (Ad.VEGF) expres-sion on VCAM-1 expression were investigated in a model ofneointimal thickening and neointimal macrophage accumula-tion induced by placement of an inert silastic collar aroundthe carotid artery of rabbits fed a high-cholesterol diet.Staining for �-galactosidase in arteries transduced with Ad.LacZ revealed abundant strongly stained cells in the adven-titia consistent with a high efficiency of gene transfer (�5%to 10%) (supplemental Figure IA).

Expression of the VEGF transgene after periadventitialdelivery of Ad.VEGF to carotid arteries was confirmed byreverse-transcription polymerase chain reaction (supplemen-tal Figure IB). No VEGF transgene expression was detectedin Ad.LacZ-transduced arteries, in segments of the trans-duced carotid arteries distal to the collared arterial region, incontralateral noncollared control arteries, or in nontargetedtissues (results not shown), indicating that the perivascularcollar localized transgene expression to the collared region ofthe artery. Immunostaining of sections of transduced arterieswith a specific antibody to VEGF showed strong expressionof VEGF in the adventitia of Ad.VEGF-transduced arteries(supplemental Figure IC).

Consistent with previous findings,11 immunostaining withmacrophage-specific RAM11 and specific VCAM-1 antibod-ies showed that neither neointimal macrophage accumulationnor endothelial VCAM-1 expression occurred in the collaredarteries of rabbits fed a normal diet. Ad.VEGF delivery tocollared arteries in rabbits on a normal diet significantlyincreased neointima formation and adventitial neovascular-ization, but had no effect on either neointimal macrophageaccumulation or VCAM-1 expression (supplemental FigureID; results not shown). However, as compared with Ad.lacZ,Ad.VEGF delivery had no significant neointima-increasingeffect in the collared arteries of cholesterol-fed rabbits (Fig-ure 2A). Collar placement in hypercholesterolemic rabbitsinduced lesions containing RAM11-positive macrophagesand marked endothelial VCAM-1 and E-selectin expression(Figure 2B and 2C), but Ad.VEGF caused no significantincrease in either neointimal macrophage accumulation (Fig-ure 2B) or endothelial VCAM-1 or E-selectin (Figure 2C),although Ad.VEGF did significantly increase adventitialangiogenesis (supplemental Figure II).

hours (VCAM-1) with 25 ng/mL VEGF-A (V), or for 6 hours with100 U/mL TNF-� (T). VEGF failed to increase E-selectin (C),ICAM-1 (D), or VCAM-1 (E) protein levels, whereas TNF-� stimu-lated CAM expression in each case. Representative experimentsare shown and findings were confirmed in 3 independent assaysusing different batches of HUVECs.

0.0

0.1

0.2

0.3

0.4

0.5

0.6mn

09

4 ta e

cn

abr

os

bA 0 3 6 9 24 TNF-

VEGF treatment, hours

VCAM-1

E

bA

ec

na

bro

st a

mn

09

4

0.00.20.40.60.8

1.01.21.41.6

0 3 6 9 24 TNF-

ICAM-1

D


C1.2

0.0

0.2

0.4

0.6

0.8

1.0

mn

09

4 ta

en

ca

bro

sb

A

0 3 6 9 24 TNF-

E-selectin


A

B

Figure 1. VEGF does not increase CAM protein or mRNA levelsin HUVECs. HUVECs were treated in triplicate with 25 ng/mLVEGF for the times indicated, or with 100 U/mL TNF-� for 6hours. CAM mRNA level was measured by quantitative real timereverse-transcription polymerase chain reaction (A,B). CAM pro-tein expression was analyzed by cell-bound enzyme-linkedimmunosorbent assay (C to E). A, VCAM-1 (f, yellow) andE-selectin (F, red) mRNA levels were modestly increased by 3hours after addition of VEGF, whereas ICAM-1 (Œ, green) mRNAlevels were virtually unchanged. TF mRNA levels increased asexpected (�, blue). Reverse-transcription polymerase chainreaction results are expressed as fold-increases �SD above thecontrol level (in basal unstimulated cells) normalized to GAPDHexpression. B, Levels of VCAM-1 (open bars), ICAM-1 (graybars), and E-selectin (black bars) mRNA were measured inHUVECs treated for 1.5 hours (ICAM-1 and E-selectin) or 3

496 Arterioscler Thromb Vasc Biol. March 2007



VEGF and TNF-� Synergize to UpregulateE-Selectin ExpressionWe next investigated the possibility that VEGF could mod-ulate CAM expression induced by TNF-�. VEGF pretreat-ment of HUVECs for 24 hours before TNF-� stimulation forthe last 6 hours of incubation increased E-selectin expressionby 61.7�5.8% from the level induced by TNF-� alone(Figure 3A; P�0.004, n�3). In contrast, TNF-�–inducedICAM-1 expression was not enhanced with VEGF pretreat-ment (Figure 3A).

Enzyme-linked immunosorbent assay data were confirmedby flow cytometry. Figure 3B shows a representative2-dimensional fluorescence-activated cell sorter overlay his-

togram for cell surface E-selectin expression in HUVECs.The fluorescein isothiocyanate fluorescence profile showedthat basal E-selectin expression in unstimulated cells wasnegligible (Figure 3B), similar to the profile of cells incu-bated with VEGF alone or to isotype-treated controls (datanot shown). TNF-� treatment upregulated cell surfaceE-selectin, whereas VEGF pretreatment before TNF-� incu-bation further increased cell surface E-selectin expression by77.2�15.0% (Figure 3C, P�0.005, n�3). In parallel, VEGFpreincubation did not significantly increase either ICAM-1 orVCAM-1 above TNF-�–induced levels (Figure 3C). Consti-tutive levels of PECAM-1 (ie, without TNF-� stimulation)were not altered significantly by VEGF (data not shown).

Synergistic upregulation of E-selectin expression by VEGFand TNF-� also occurred at 10 U/mL (0.5 ng/mL) TNF-�(Figure 3D). Indeed, pretreatment with VEGF before additionof 10 U/mL TNF-� induced a greater synergistic effect than100 U/mL TNF-� (143% versus 71%, respectively, above thecorresponding TNF-� controls), as shown by enzyme-linkedimmunosorbent assay.

IL-1� and TNF-� stimulated E-selectin protein expressionto a similar extent, �25-fold above the level in controlunstimulated cells (data not shown). IL-1� synergized withVEGF to enhance E-selectin expression (48.2�10.5% in-crease from level of IL-1� control, P�0.005, n�3; Figure3E), although in parallel cell cultures, TNF-� caused a greatersynergistic effect with VEGF than IL-1� (77.2�15.0%enhancement).

VEGF and TNF-� Synergize to UpregulateE-Selectin mRNA LevelsQuantitative real-time reverse-transcription polymerase chainreaction showed that VEGF and TNF-� synergize to increaseE-selectin transcription (Figure 3F). The induction inE-selectin mRNA increased from �830-fold above the con-trol unstimulated level with TNF-� alone to 1500-fold withVEGF and TNF-� (81% increase, P�0.05, n�3). NeitherICAM-1 nor VCAM-1 mRNAs were significantly enhancedby the combination of VEGF and TNF-� above the levelsinduced by TNF-� alone (data not shown). Consistent withprevious findings,21,22 VEGF and TNF-� synergized to up-regulate TF mRNA expression (P�0.05), whereas VEGF/TNF-� caused no synergistic upregulation of COX-2 (sup-plemental Figure III). Although VEGF induces rapidupregulation of COX-2, COX-2 expression declines to basallevels after 6 hours and is not detectable after 24 hours.16,17

VEGF/TNF-� Synergy Requires PreincubationWith VEGFHUVECs were preincubated with VEGF for various times (0to 24 hours) before TNF-� stimulation and then analyzed forE-selectin expression by flow cytometry to find the minimumpreincubation time needed with VEGF for synergy to occur.The maximum synergistic effect of VEGF/TNF-� onE-selectin required at least 4 hours pretreatment with VEGF(Figure 4A).

The possibility that VEGF might enhance E-selectinexpression via secretion of a soluble factor was tested bytransferring media from VEGF-treated cells to another set

Ad.Ad.LacZLacZ

IELIEL

Ad.VEGFAd.VEGF

IELIEL

LacZ0

0.1

0.2

0.3

Iintima/Media

VEGF

B

0100200300400500600

MØ/m

m2

LacZ VEGF

01020304050

%VCAM-1

LacZ VEGF

C

A

Ad.LacZ Ad.VEGF

0

5

10

15

%E-selectin

LacZ VEGF

Figure 2. VEGF does not induce endothelial VCAM-1 orE-selectin expression in hypercholesterolemic rabbits. A, Collarplacement and gene transfer were performed in cholesterol-fedrabbits and intima/media ratios were determined in collaredcarotid arteries infected with Ad.lacZ (n�10) or Ad.VEGF-A165

(n�10) as described in Experimental Procedures. Intimal thick-ening was not significantly increased in Ad.VEGF-A165-transduced cholesterol-fed rabbits. Representative H&E sectionsindicating the position of the IEL are shown. B, Collared carotidarteries in rabbits on a high-cholesterol diet were transducedwith Ad.lacZ or Ad.VEGF-A165 and macrophage content wasdetermined by immunostaining with RAM11. Macrophage den-sity is expressed as neointimal RAM11-positive cells�SEper mm2 total neointimal area. Arrowheads indicate areas ofpositive staining. C, Immunohistochemical staining of VCAM-1and E-selectin in sections from collared arteries transduced withAd.lacZ or Ad.VEGF-A165. Quantification of endothelial immuno-staining is shown to the right: values represent means�SEVCAM-1 and E-selectin staining on sections taken from 3 pointsin each treated artery (n�8 arteries each for Ad.lac Z andAd.VEGF-A165 for VCAM-1 and n�3 each for E-selectin). No sig-nificant differences were found between Ad.lacZ and Ad.VEGF-A165 groups.




of cultures that were then immediately incubated withTNF-� for 6 hours. The results showed that media fromVEGF-treated cells did not enhance E-selectin expression(Figure 4B), suggesting that VEGF does not induce pro-duction of a diffusible secreted factor that mediates syn-ergy with TNF-�. In the wells where the VEGF-containing

media had been removed after incubation, addition of freshmedia without VEGF but containing TNF-� still causedsynergy (Figure 4B). These experiments confirmed thatpre-incubation with VEGF is needed for synergy withTNF-�, although VEGF does not need to be present whencells are stimulated with TNF-�.

A

020406080

100120140160180,

noi

ss

erp

xe

MA

Cl

o rtn

oc f

o %

ICAM-1E-selectin

**

C

10 0 10 1 10 2 10 3

Log FITC fluorescence

0

25

slle

c fo r

eb

mu

N

B

D

0.00.10.20.30.40.50.60.70.80.91.0

1 10 100TNF- concentration, U/ml

mn

09

4 ta

ec

na

bro

sb

A

TNF-

VEGF + TNF-

020406080

100120140160180l

ortn

oc f

o % ,

nitc

ele

s-E

**

Il-1 VEGF+ Il-1

E

F

VEGF+ TNF-

0

500

1000

1500

2000

basal VEGF TNF-

HD

PA

G/nit

cel

es-

E

*

020406080

100120140160180200

E-selectin ICAM-1 VCAM-1

**,n

ois

ser

px

e M

AC

lort

no

c fo

%

Figure 3. VEGF and TNF-� synergize to upregulate E-selectin expression. HUVECs were pretreated in triplicate with 25 ng/mL VEGFfor 24 hours and then with 100 U/mL TNF-� for 6 hours (gray bars) or treated with TNF-� alone (black bar). A, E-selectin, determinedby specific enzyme-linked immunosorbent assay, was increased by 61.7�5.8% (**P�0.01). Results show values for CAM expressionabove the basal unstimulated level normalized to the corresponding TNF-� control (both 100�4.2%, black bars), obtained from 3 inde-pendent experiments performed in triplicate (means�SE) using different cell batches. B, Anti-E-selectin binding to HUVECs wasdetected using a secondary fluorescein isothiocyanate-conjugated antibody before fixation in 1% paraformaldehyde and flow cytomet-ric analysis. The 2-dimensional overlay histogram shows a 3-decade log10 fluorescence scale against the number of gated intact cellswith corresponding fluorescein isothiocyanate fluorescence values from a representative experiment. Basal E-selectin expression wasnegligible (solid gray line), TNF-� (6 hours) upregulated cell surface E-selectin (dashed gray line), and VEGF preincubaton before TNF-�further upregulated E-selectin expression (solid black line). C, Cell surface E-selectin determined by fluorescence-activated cell sorterwas increased by 67.2�15.0% (**P�0.01, n�3) when cells were treated with VEGF before TNF-� (gray bars), as compared with theTNF-� control (black bars). Results show CAM expression above the basal level normalized to the corresponding TNF-� control (all100�2.0%, black bars) expressed as means�SE of three or 4 independent experiments each performed in triplicate with different cellbatches. Total cellular protein content was unaltered by any of the incubations described, as determined by Bradford assay (data notshown). D, HUVECs were incubated overnight with VEGF before stimulation with 1, 10, or 100 U/mL TNF-�, and E-selectin expressionwas measured by enzyme-linked immunosorbent assay. Enzyme-linked immunosorbent assay data are from an experiment using tripli-cate wells expressed as means�SD, and are representative of 2 independent experiments each yielding very similar results. E,HUVECs were preincubated with (gray bar) or without (black bar) VEGF for 24 hours, before stimulation with recombinant human Il-1�for a further 6 hours (100 U/mL or 1 ng/mL), and fluorescence-activated cell sorter analysis of E-selectin expression. VEGF pretreat-ment increased E-selectin 48.2�10.5% above the effect of IL-1� alone (gray bar; **P�0.01 vs IL-1�). Results show E-selectin expres-sion above the basal level normalized to the IL-1� control expressed as means�SE of 3 independent experiments performed in tripli-cate with different cell batches. F, HUVECs were pretreated for 24 hours either with (VEGF, VEGF�TNF-�) or without (basal, TNF-�) 25ng/mL VEGF, and then incubated with 100 U/mL TNF-� for a further 6 hours. Expression of mRNA for E-selectin was then determinedby quantitative real-time reverse-transcription polymerase chain reaction. E-selectin mRNA was increased by 81% with VEGF/TNF-�(P�0.05, n�3; gray bar) compared with TNF-� alone (black bar).




VEGF/TNF-� Synergy Is Mediated by VEGFR2and the Calcineurin/Nuclear Factor of ActivatedT-Cells Signaling PathwayWe examined the receptor and second messenger mecha-nisms mediating the VEGF/TNF-� synergistic effect onE-selectin. Placental growth factor (PlGF), a ligand forVEGFR1 but not for VEGFR2, failed to synergize withTNF-� to enhance E-selectin expression, and PlGF had noeffect on the VEGF/TNF-� synergy when HUVECs werepreincubated with a combination of PlGF and VEGF (Figure5A), as shown by flow cytometry. Preincubation with thespecific VEGFR2 inhibitor, SU5614, almost completelyblocked the synergistic effect on E-selectin expression(84.6�8.1% inhibition, P�0.005), indicating that VEGFR2mediated this effect (Figure 5B).

Cyclosporin A (0.2 �mol/L), an inhibitor of the proteinphosphatase activity of calcineurin/protein phosphatase 2B,

partially blocked the VEGF/TNF-� synergism implicatingthe involvement of the calcineurin/nuclear factor of activatedT-cells signaling pathway (59.3�0.8% inhibition P�0.03,n�3; Figure 5C). Inhibition of either the extracellular signal-regulated kinase 1/2 pathway using the specific MAP kinasekinase (mitogen-activated protein/extracellular signal regu-lated kinase) inhibitor, U0126 (10 �mol/L), or PI3K-

B

0

20

40

60

80

100

120

140ec

ne

cs

ero

ulf n

aid

eM

Basal TNF- VEGF+TNF-

VEGFremoval

Media transfer(from VEGF-primed cells)

Ae

cn

ec

ser

oulf

na i

de

M 0

20

40

60

80

100

120

Basal TNF- 4h VEGF+TNF-

VEGF+TNF- , 6h

2h VEGF+TNF-

24h VEGF+TNF-

VEGF pre-treatment

Figure 4. VEGF/TNF-� synergy requires preincubation ofHUVECs with VEGF. A, HUVECs were either preincubated with25 ng/mL VEGF for the times indicated before stimulation with100 U/mL TNF-� for a further 6 hours (gray bars), or weretreated with TNF-� alone for 6 hours (black bar), or were coin-cubated with VEGF and TNF-� for 6 hours. E-selectin expres-sion was then determined by flow cytometry. Synergisticupregulation of E-selectin expression required a minimum of 4hours VEGF preincubation with HUVECs. B, HUVECs were pre-incubated with VEGF for 16 hours and then the media fromthese cells were transferred to fresh cells that were then alsotreated with 100 U/mL TNF-� for 6 hours (hatched bar). In par-allel, VEGF-treated cells from media had been removed werealso treated with 100 U/mL TNF-� for 6 hours (open bar). Theeffects of TNF-� treatment for 6 hours with (gray bar) or without(black bar) pretreatment with 25 ng/mL VEGF for 16 hours areshown for comparison. After all treatments, E-selectin expres-sion was measured by flow cytometry. There was no synergywith TNF-� in the wells exposed to cell media conditioned byVEGF pretreatment for 16 hours (hatched bar), whereas in thecells where VEGF had been removed, TNF-� caused a synergis-tic upregulation of E-selectin (open bar). Fluorescence-activatedcell sorter data are from an experiment using triplicate wellsexpressed as means�SD and representative of 2 independentexperiments producing very similar results.

A

020406080

100120140160180

T PT VT PVT

**

nitc

ele

s-E

)lort

no

c fo

%(

B

020406080

100120140160180

T ST VT SVT

** **

C

020406080

100120140160180

T CT VT CVT

** *

D

020406080

100120140160180200

T UT VT UVT

** *

nitc

ele

s-E

)lort

no

c fo

%( nit

cel

es-

E)l

o rtn

oc f

o %(

nitc

ele

s-E

) lort

no

c fo

%(

Figure 5. Synergistic E-selectin upregulation by VEGF andTNF-� is mediated by VEGFR2 and calcineurin signaling. A,HUVECs were incubated with PlGF (PT, hatched bar), VEGF (VT,gray bar), or a combination of PlGF and VEGF (PVT, open bar)before stimulation with TNF-� and subsequent E-selectin analy-sis by flow cytometry. TNF-� control (T) is black bar(100�2.8%). B to D, HUVECs were incubated with either5 �mol/L SU5614 (B), or 0.2 �mol/L cyclosporin A (C), or10 �mol/L U0126 (D), or an equal volume of solvent for 1 hour,before either treatment with 100 U/mL TNF-� (T) alone for 6hours (hatched and black bars), or preincubation for 16 hourswith 25 ng/mL VEGF and subsequent treatment with 100 U/mLTNF-� (VT, gray and open bars) for 6 hours (all incubation stepsin the continuous presence of inhibitors; open bars). SU5614 (S)reduced the VEGF/TNF synergy by 84.6�8.1% (**P�0.01 forSVT versus VT), but had no significant effect on the TNF-� con-trol (ST, hashed bar). Treatment with cyclosporin A (C) inhibitedthe VEGF/TNF synergy by 59.3�0.9% (*P�0.05 for CVT vs VT).U0126 (U) did not inhibit the VEGF/TNF synergy but significantlyenhanced the effect (*P�0.05 for UVT vs VT, gray bar). Allresults show E-selectin expression above the basal level nor-malized to the corresponding TNF-� expressed as means�SEfrom 3 independent experiments performed in triplicate with dif-ferent cell batches (*P�0.05, **P�0.01).




dependent Akt activation using LY294002, did not signifi-cantly inhibit the synergistic cooperation between VEGF andTNF-� (Figure 5D and results not shown).

VEGF/TNF-� Synergism IncreasesE-Selectin SheddingAfter endothelial activation, surface E-selectin undergoesshedding to release soluble isoforms.23 To test the effect ofVEGF on TNF-�–induced shedding, culture media wereremoved after VEGF pretreatment and subsequent incubationwith TNF-� for 6 hours and replaced with fresh media notcontaining VEGF or TNF-�. Soluble E-selectin (sE-selectin)shed into the media was measured using a sensitive sandwichenzyme-linked immunosorbent assay (Figure 6). TNF-�–induced sE-selectin expression in HUVECs was detectableafter 2 hours and increased up to 24 hours (Figure 6). WhileVEGF alone did not cause any detectable increase in the levelof sE-selectin produced by HUVECs, VEGF preincubationsynergistically increased TNF-�–induced shedding of sE-selectin at all time points, an effect that was statisticallysignificant (P�0.05) after 6, 20, and 24 hours (Figure 6).

DiscussionThe role of VEGF in atherosclerosis is contentious becausestudies have reached differing conclusions as to whetherVEGF can either inhibit or promote intimal thickening andatherosclerotic lesion formation in a variety of animal mod-els. A proatherogenic role of VEGF is supported by reportsthat VEGF upregulates expression of inflammatory CAMs inendothelial cells, including VCAM-1, ICAM-1, andE-selectin.8.9 Upregulation of endothelial CAMs, particularlyVCAM-1, is recognized to play a key role in the activation ofendothelium in inflammation, and a key mediator of mono-

cyte/macrophage adhesion and infiltration in early and lateratherosclerotic lesions responsible for the excessive accumu-lation of macrophages in the atherosclerotic plaque. Whetheror how VEGF regulates endothelial CAM expression istherefore a question of central importance for the pathophys-iology of this cytokine, which also has ramifications for thetherapeutic potential of VEGF in ischemic heart disease.

The first major finding of this article is that VEGF alone isunable to induce protein expression of endothelial VCAM-1,ICAM-1, and E-selectin in endothelial cells expressing highlevels of VEGFR2/KDR. In addition, adenoviral overexpres-sion of VEGF in a model of atherogenic neointima formationand macrophage accumulation also had no significant effecton endothelial VCAM-1 and E-selectin expression in vivo.Our previous findings showed that liposome-mediated plas-mid VEGF gene delivery to the collared arteries of hypercho-lesterolemic rabbits, reduced neointima formation, neointimalmacrophage accumulation, and endothelial VCAM-1 expres-sion,11 but this study did not examine the effects of high-efficiency adenoviral VEGF gene delivery. Importantly, nei-ther low- nor high-efficiency VEGF expression caused anyincrease in VCAM-1 expression. The different effects ofplasmid and adenoviral VEGF on intimal thickening, macro-phage accumulation, and VCAM-1 expression in the collarmodel may be attributable to an arterioprotective effect oflow intra-arterial VEGF concentrations that is impaired at thehigher local concentrations of VEGF produced by adenoviralexpression.11 Interestingly, Ad.PlGF gene delivery in thesame model in parallel with the study of Ad.VEGF presentedhere, increased neointimal thickening, macrophage accumu-lation and endothelial VCAM-1 expression in collared arter-ies in hypercholesterolemic rabbits.24 The present resultstherefore indicate a striking contrast in the biological effectsof VEGF and PlGF in atherogenic lesion formation in rabbits.

Whereas VEGF significantly increased VCAM-1,ICAM-1, and E-selectin mRNA expression in endothelialcells, these effects were extremely small compared with theeffect of TNF-�. Kim et al reported that VEGF increasedVCAM-1, ICAM-1, and E-selectin protein expression, asdetermined by Western blot.8,9 In contrast, we found thatVEGF had no significant effect on CAM protein expressionas judged by sensitive and quantitative enzyme-linked immu-nosorbent assays of total cellular CAM expression or mea-surement of cell surface CAM expression by flow cytometry.The lack of CAM upregulation as determined by flowcytometry also precluded the possibility that there was asubpopulation of cells that could upregulate CAM expressionin response to VEGF. Previous studies examined effects ofVEGF in freshly isolated HUVECs cultured in M199 mediumwith serum,8,9 whereas our studies used commercially avail-able HUVECs cultured in supplemented EBM. However, weconclude that differences in the source and culture ofHUVECs are unlikely to account for our failure to observe aneffect of VEGF alone on CAM expression. TNF-� and IL-1�strongly upregulated CAM expression at protein and mRNAlevel indicating that the cellular mechanisms mediating tran-scriptional upregulation of CAMs are not impaired in com-mercial HUVECs. We show here and in previous studies thatVEGF strongly induces an array of signaling and cellular

*

/g

p ,nit

cel

es-

Es

µµni

etor

p g

Time after cytokine treatment, hours

TNF-

TNF-+ VEGF

0

5

10

15

20

0 5 10 15 20 25

VEGF

**

Figure 6. VEGF/TNF-� synergism increases E-selectin shed-ding. HUVECs were pretreated with nothing (�, solid line) orwith VEGF (f, dashed line) for 24 hours and subsequentlytreated either with 100 U/mL TNF-� (�, f) or with no addition(Œ) for a further 6 hours. After VEGF and TNF-� treatments, cul-ture media was removed and replaced with fresh media (withoutcontaining VEGF or TNF-�) and any sE-selectin shed into themedia was measured after the times indicated using a sensitivesandwich enzyme-linked immunosorbent assay. After 2 hourssE-selectin was detectable in the media of TNF-�–treated cellsand this level increased for up to 24 hours. VEGF treatmentalone (Œ) caused no detectable increase in sE-selectin produc-tion. The results shown are representative of 4 independentexperiments and are expressed as means�SE from 3 indepen-dent experiments using different cell batches each performed intriplicate. Two-way ANOVA indicated significant overall time andtreatment effects on sE-selectin production (P�0.0005 in bothcases). Post-hoc analysis by Fisher PLSD indicated that VEGFpreincubation significantly increased TNF-�–induced sheddingof sE-selectin at 6, 20, and 24 hours after treatments by,respectively, 76%, 62%, and 48% (*P�0.05, n�3) comparedwith TNF-� alone.




responses in commercially obtained HUVECs, includingupregulation of multiple genes.17,25 Furthermore, we haveconfirmed that several key signaling and biological responsesto VEGF are preserved in low-passage number freshlyisolated and in commercial HUVECs.20,25 Moreover, ourprevious findings together with the results of adenoviralVEGF overexpression presented here also indicate thatVEGF is unable to significantly enhance endothelialVCAM-1 and E-selectin expression in vivo, even underinflammatory hypercholesterolemic conditions favoringCAM upregulation. Zhang and Issekutz also showed thatVEGF, in contrast to TNF-�, had no effect on ICAM-1,VCAM-1, and E-selectin expression in HUVECs as deter-mined by enzyme-linked immunosorbent assay, although thisstudy examined effects of a 3-day treatment.26 We provision-ally conclude that differences between our findings and thoseof other researchers may be attributable to the differentapproaches used to measure CAM expression, although morework is needed to establish more precisely the reasons forthese differences.

The second major finding of this paper is that whereasVEGF alone had no significant effect on CAM expression, itsynergistically enhanced the induction of E-selectin mRNAand cell surface protein expression by TNF-�. This was aselective synergistic interaction because VEGF did not en-hance TNF-�–induced expression of VCAM-1 or ICAM-1.VEGF also synergized with TNF-� to enhance shedding ofthe soluble extracellular domain of E-selectin. VEGF andTNF-� have previously been shown to synergistically in-crease functional TF expression in endothelial cells, but it hasbeen unclear whether cooperative interactions between these2 cytokines are relevant for other endothelial responses.Synergistic enhancement of E-selectin expression by VEGFwas mediated by VEGFR2/KDR; in addition, the inhibitionof the effect by cyclosporin A, a specific inhibitor ofcalcineurin, indicates that a calcineurin-dependent pathwayalso plays a major role in this response. It has been reportedthat mitogen-activated protein/extracellular signal-regulatedkinase/extracellular signal-regulated kinase is the conver-gence point of the VEGF and TNF-� signaling for thesynergistic upregulation of TF, which leads to activation ofthe transcription factor early growth response-1.27 However,inhibition of mitogen-activated protein/extracellular signalregulated kinase did not block the synergistic enhancement ofE-selectin. Early growth response-1 elements have beenfound in some CAM promoters but are unknown forE-selectin. The nuclear factor of activated T-cells/calcineurinpathway has been implicated in expression of E-selectin28 andVEGF-induced upregulation of TF,29 and Down syndromecritical region protein 1.17 Cyclosporin A has been reported tosuppress E-selectin, but not VCAM-1, induction in HUVECsby TNF-�, even though the E-selectin promoter is activatedby NF-�B rather than nuclear factor of activated T-cells.28

E-selectin has been implicated in angiogenesis in vitro30,31

and in vivo,32 although the underlying mechanisms remain tobe defined. Synergistic augmentation of TNF-�–regulatedE-selectin expression by VEGF might therefore be relevantfor neovascularization in an inflammatory milieu, as forexample in pathophysiological settings such as atherosclero-

sis, cancer, and rheumatoid arthritis. E-selectin is proteolyti-cally shed by an unknown mechanism with the resultingsE-selectin, lacking transmembrane and cytoplasmic do-mains, being biologically active and playing a role in angio-genesis through the Src-PI3K pathway.33,34 VEGF-dependentaugmentation of sE-selectin generation induced by inflamma-tory cytokines may contribute to the pathogenesis of cardio-vascular disease, possibly through enhanced angiogenesiswithin atherosclerotic lesions, a phenomenon that has beenhypothesized to promote intraplaque hemorrhage and desta-bilization leading to rupture. However, E-selectin–deficientmice develop normally and exhibited no impairment inexperimentally induced angiogenesis,35 indicating thatVEGF-enhanced shedding of sE-selectin is unlikely to play akey role in developmental angiogenesis, although it is notprecluded that it may contribute to pathophysiological angio-genesis in disease-specific settings. The role of sE-selectin inatherosclerosis is less clear because, whereas it has proin-flammatory effects on neutrophil function,36 induces mono-cyte chemotaxis,34 and is significantly increased in patientswith coronary artery disease, the increases observed havebeen small and, in several studies of human populations,sE-selectin levels have not emerged as a strong predictor ofcardiovascular disease.23

The importance of synergistic interactions between cyto-kines is becoming increasingly appreciated.37 However, thisis the first report to our knowledge of enhancement of CAMexpression by VEGF/TNF-� synergy. Furthermore, VEGFalso synergized with IL-1� to upregulate E-selectin. BecauseTNF-� and IL-1� share many of the same signaling path-ways, these findings suggest that VEGF-triggered signaltransduction mechanisms can cooperate with a commonpathway activated by inflammatory cytokines. VEGF may“prime” endothelial cells so they are capable of responding tolower levels of TNF-�. Consistent with this notion, weobserved relatively more synergy with a lower concentrationof TNF-�.

The present article is consistent with the broader conclu-sion that the predominant effect of VEGF alone is notproinflammatory and does not cause endothelial cell activa-tion associated with cardiovascular disease. However, be-cause VEGF can selectively enhance E-selectin expressioninduced by TNF-� or IL-1�, we propose that the overallimpact of VEGF on endothelial function and, by extension,on vascular pathophysiology, may be modified by the localcytokine milieu. A fuller understanding of how VEGF inter-acts and synergizes with other endothelial cytokines will bekey to delineating its role as both a therapeutic and patho-genic factor in cardiovascular disease.

Sources of FundingThis work was supported by British Heart Foundation grantsRG/02/001 and BS/94001 to I.Z.

DisclosuresNone.

References1. Ross R. Atherosclerosis-an inflammatory disease. N Engl J Med. 1999;

340:115–126.




2. Chia MC. The role of adhesion molecules in atherosclerosis. Crit RevClin Lab Sci. 1998;35:573–602.

3. Nakashima Y, Raines EW, Plump AS, Breslow JL, and Ross R. Upregu-lation of VCAM-1 and ICAM-1 at atherosclerosis-prone sites on theendothelium in the apoE-deficient mouse. Arterioscler Thromb Vasc Biol.1998;18:842–851.

4. Truskey GA, Herrmann RA, Kait J, and Barber KM. Focal increases invascular cell adhesion molecule-1 and intimal macrophages at atheroscle-rosis-susceptible sites in the rabbit aorta after short-term cholesterolfeeding. Arterioscler Thromb Vasc Biol. 1999;19:393–401.

5. Iiyama K, Hajra L, Iiyama M, Li H, DiChiara M, Medoff BD, CybulskyMI. (1999). Patterns of vascular cell adhesion molecule-1 and inter-cellular adhesion molecule-1 expression in rabbit and mouse athero-sclerotic lesions and at sites predisposed to lesion formation. Circ Res.1999;85:199–200.

6. Dansky HM, Barlow CB, Lominska C, Sikes JL, Kao C, Weinsaft J,Cybulsky MI, Smith JD. Adhesion of monocytes to arterial endotheliumand initiation of atherosclerosis are critically dependent on vascular celladhesion molecule-1 gene dosage. Arterioscler Thromb Vasc Biol. 2001;21:1662–1667.

7. Cybulsky MI, Iiyama K, Li H, Zhu S, Chen M, Iiyama M, Davis V,Gutierrez-Ramos JC, Connelly PW, Milstone DS. A major role forVCAM-1, but not ICAM-1, in early atherosclerosis. J Clin Invest. 2001;107:1255–1262.

8. Kim I, Moon SO, Park SK, Chae SW, Koh GY. Angiopoietin-1 reducesVEGF-stimulated leukocyte adhesion to endothelial cells by reducingICAM-1, VCAM-1, and E-selectin expression. Circ Res. 2001;89:477–479.

9. Kim I, Moon SO, Kim SH, Kim HJ, Koh YS, Koh GY. Vascularendothelial growth factor expression of intercellular adhesion molecule 1(ICAM-1), vascular cell adhesion molecule 1 (VCAM-1), and E-selectinthrough nuclear factor-kappa B activation in endothelial cells. J BiolChem. 2001;276:7614–7620.

10. Simons M, Ware JA. Therapeutic angiogenesis in cardiovascular disease.Nat Rev Drug Discov. 2003;2:863–871.

11. Khurana R, Shafi S, Martin JF, and Zachary I. Vascular endothelialgrowth factor gene transfer inhibits neontimal macrophage accumulationin hypercholesterolaemic rabbits. Arterioscler Thromb Vasc Biol. 2004;24:1074–1080.

12. Celletti FL, Waugh JM, Amabile PG, Brendolan A, Hilfiker PR, DakeMD. Vascular endothelial growth factor enhances atherosclerotic plaqueprogression. Nat Med. 2001;7:425–429.

13. Ohtani K, Egashira K, Hiasa K, Zhao Q, Kitamoto S, Ishibashi M, UsuiM, Inoue S, Yonemitsu Y, Sueishi K, Sata M, Shibuya M, Sunagawa K.Blockade of vascular endothelial growth factor suppresses experimentalrestenosis after intraluminal injury by inhibiting recruitment of monocytelineage cells. Circulation. 2004;110:2444–2452.

14. Zachary I. VEGF signalling: integration and multitasking in endothelialcell biology. Biochem Soc Trans. 2003;31:1171–1177.

15. Neufeld G, Cohen T, Gengrinovitch S, and Poltorak Z. Vascular endo-thelial growth factor (VEGF) and its receptors. FASEB J. 1999;13:19–22.

16. Hernandez GL, Volpert OV, Iniguez MA, Lorenzo E, Martinez-MartinezS, Grau R, Fresno M, Redondo JM. Selective inhibition of vascularendothelial growth factor-mediated angiogenesis by cyclosporin A: rolesof the nuclear factor of activated T cells and cyclooxygenase 2. J ExpMed. 2001;193:607–620.

17. Liu D, Jia H, Holmes DI, Stannard A, Zachary I. Vascular endothelialgrowth factor-regulated gene expression in endothelial cells: KDR-mediated induction of Egr3 and the related nuclear receptors Nur77,Nurr1, and Nor1. Arterioscler Thromb Vasc Biol. 2003;23:2002–2007.

18. Khurana, R, Zhuang, Z, Bhardwaj, S, Murakami, M, De Muinck, E,Yla-Herttuala, S, Ferrara, N, Martin, J, Zachary, I and Simons M.Angiogenesis-Dependent and Independent Phases of Intimal Hyperplasia.Circulation. 2004;110:2436–2443.

19. Stannard AK, Riddell DR, Bradley NJ, Hassall DG, Graham A, and OwenJS. Apolipoprotein E and regulation of cytokine-induced cell adhesion

molecule expression in endothelial cells. Atherosclerosis. 1998;139:57–64.

20. Wheeler-Jones C, Abu-Ghazaleh R, Cospedal R, Houliston RA, Martin J,Zachary I. Vascular endothelial growth factor stimulates prostacyclinproduction and activation of cytosolic phospholipase A2 in endothelialcells via p42/p44 mitogen-activated protein kinase. FEBS Lett. 1997;420:28–32.

21. Clauss M, Grell M, Fangmann C, Fiers W, Scheurich P, Risau W.Synergistic induction of endothelial tissue factor by tumor necrosis factorand vascular endothelial growth factor: functional analysis of the tumornecrosis factor receptors. FEBS Lett. 1996;390:334–338.

22. Shen BQ, Lee DY, Cortopassi KM, Damico LA, Zioncheck TF. Vascularendothelial growth factor KDR receptor signaling potentiates tumornecrosis factor-induced tissue factor expression in endothelial cells. J BiolChem. 2001;276:5281–5286.

23. Roldan V, Marin F, Lip GY, Blann AD. Soluble E-selectin in cardiovas-cular disease and its risk factors. A review of the literature. ThrombHaemost. 2003;90:1007–1020.

24. Khurana R, Moons L, Shafi S, Luttun A, Collen D, Martin JF, CarmelietP, and Zachary IC. Placental Growth Factor (PlGF) Promotes Athero-sclerotic Intimal Thickening and Macrophage Accumulation. Circulation.2005;111:2828–2836.

25. Jia, H, Bagherzadeh A, Bicknell R, Duchen MR, Liu, D and Zachary I.Vascular Endothelial Growth Factor (VEGF)-D and VEGF-A Differen-tially Regulate KDR-mediated Signalling and Biological Function inVascular Endothelial Cells. J Biol Chem. 2004;279:36148–36157.

26. Zhang H, Issekutz AC. Down-modulation of monocyte transendothelialmigration and endothelial adhesion molecule expression by fibroblastgrowth factor: reversal by the anti-angiogenic agent SU6668. Am JPathol. 2002;160:2219–2230.

27. Mechtcheriakova D, Wlachos A, Holzmuller H, Binder BR, Hofer E.Vascular endothelial cell growth factor-induced tissue factor expressionin endothelial cells is mediated by EGR-1. Blood. 1999;93:3811–3823.

28. Cockerill GW, Bert AG, Ryan GR, Gamble JR, Vadas MA, Cockerill PN.Regulation of granulocyte-macrophage colony-stimulating factor andE-selectin expression in endothelial cells by cyclosporin A and the T-celltranscription factor NFAT. Blood. 1995;86:2689–2698.

29. Armesilla AL, Lorenzo E, Gomez del Arco P, Martinez-Martinez S,Alfranca A, Redondo JM. Vascular endothelial growth factor activatesnuclear factor of activated T cells in human endothelial cells: a role fortissue factor gene expression. Mol Cell Biol. 1999;19:2032–2043.

30. Nguyen M, Strubel NA, Bischoff J. A role for sialyl Lewis-X/A glyco-conjugates in capillary morphogenesis. Nature. 1993;365:267–269.

31. Aoki M, Kanamori M, Yudoh K, Ohmori K, Yasuda T, Kimura T. Effectsof vascular endothelial growth factor and E-selectin on angiogenesis inthe murine metastatic RCT sarcoma. Tumour Biol. 2001;22:239–246.

32. Nguyen M, Corless CL, Kraling BM, Tran C, Atha T, Bischoff J, BarskySH. Vascular expression of E-selectin is increased in estrogen-receptor-negative breast cancer: a role for tumor-cell-secreted interleukin-1 alpha.Am J Pathol. 1997;150:1307–1314.

33. Koch AE, Halloran MM, Haskell CJ, Shah MR, Polverini PJ. Angio-genesis mediated by soluble forms of E-selectin and vascular celladhesion molecule-1. Nature. 1995;376:517–519.

34. Kumar P, Hosaka S, Koch AE. Soluble E-selectin induces monocytechemotaxis through Src family tyrosine kinases. J Biol Chem. 2001;276:21039–21045.

35. Hartwell DW, Butterfield CE, Frenette PS, Kenyon BM, Hynes RO,Folkman J, Wagner DD. Angiogenesis in P- and E-selectin-deficientmice. Microcirculation. 1998;5:173–178.

36. Ruchaud-Sparagano MH, Walker TR, Rossi AG, Haslett C, Dransfield I.Soluble E-selectin acts in synergy with platelet-activating factor to acti-vate neutrophil beta 2-integrins. Role of tyrosine kinases and Ca2� mobi-lization. J Biol Chem. 2000;275:15758–15764.

37. Iademarco MF, Barks JL, Dean DC. Regulation of vascular cell adhesionmolecule-1 expression by IL-4 and TNF-� in cultured endothelial cells.J Clin Invest. 1995;95:264–271.




1

SUPLEMENTARY EXPERIMENTAL PROCEDURES

Materials–Recombinant human VEGF165 and Placental Growth Factor (PlGF) were

from R&D Systems, Abingdon, UK. Cyclosporin A (CsA), SU5614 and U0126 were from

Calbiochem (Nottingham, UK). Anti-von Willebrand Factor (vWF; clone F8/86),

fluorescein isothiocyanate (FITC)-conjugated anti-mouse antibodies and monoclonal anti-

platelet endothelial CAM-1 (PECAM-1 or CD31; clone JC/70A) were from Dako (Ely,

UK). Monoclonal antibodies: VCAM-1 (CD106; clone BBIG-V1), ICAM-1 (CD54; clone

BBIG-I1) and E-selectin (CD62E; BBIG-E4) were from R&D Systems. Recombinant

human TNF-a, IL-1b, and other reagents were from Sigma (Poole,UK), unless otherwise

specified.

Cell Culture–Human umbilical vein endothelial cells (HUVECs) were purchased from

TCS CellWorks (Buckingham, UK). Cells were cultured in endothelial basal medium

(EBM) supplemented with 10 % (v/v) fetal bovine serum (FBS), 10 ng/ml human

epidermal growth factor, 12 mg/ml bovine brain extract and 50 mg/ml gentamycin sulphate

(Cambrex Bio Science, Wokingham, UK). For experiments, 1 % (v/v) serum-containing

medium, in the absence of other supplements, was used. Plastic culture dishes were

precoated with 1 % (w/v) gelatin, and cells were used between passages 1-4.

Adenoviruses–Adenoviral constructs (E1,E3-deleted) encoding either VEGF-A165

(Ad.VEGF-A165) or LacZ (Ad.LacZ) were produced by Ark Therapeutics, Finland, as

described.11 Adenoviruses were desalted using G50 Sephadex columns (Roche Diagnostics

Ltd, Lewes, UK) immediately prior to their use in animal studies.

Collar Placement and Gene Transfer–All experiments were conducted in accordance

with the Animal Care and Ethics Guidelines of University College London, UK. New

Zealand White male rabbits (2.5-3.2 kg) were fed a normal diet supplemented with 1.5%



2

cholesterol for 1 week prior to collar placement and throughout the experiment. In

parallel, rabbits were maintained on normal chow without added cholesterol. Placement

of a biologically inert, silastic collar (Ark Therapeutics) around the right carotid artery

was performed in anesthetized rabbits as described.11 Five days later, the collared arteries

were exposed and 100ml of each adenoviral vector solution containing 5 x 109 pfu, was

placed within the space between collar and artery using a pipette, and the wound sutured.

Serum samples were taken for assay of total cholesterol, Low Density Lipoprotein- and

High Density Lipoprotein -cholesterol (Roche Diagnostics Ltd), 7 days before initiation of

the high-cholesterol diet, 1 day after initiating the diet and 9 days after gene transfer. Nine

days after gene transfer, animals were euthanized; collared and contralateral control

arteries were then excised. Arteries were flushed with ice-cold saline and divided into two

segments. The proximal part was immersion-fixed in 1 % paraformaldehyde/phosphate

buffered saline (PBS) pH 7.4, for 6 h, rinsed in 70 % ethanol, and embedded in paraffin.

The distal part was either immersion-fixed in 4 % paraformaldehyde/PBS pH 7.4, for 30

min, rinsed in PBS for 15 min, cryoprotected in sucrose, embedded in OCT compound

(Miles), and stored at –80˚C or snap frozen in liquid nitrogen and stored at –80˚C for

total RNA extraction.

RNA Isolation and Reverse-transcription (RT) PCR–Total RNA was isolated from

30-50 mg frozen tissue (pooled from 2 carotids or non-targeted organs) using an adapted

RNeasy spin-column method, as described11 (RNeasy Fibrous Tissue Mini Kit, QIAGEN,

Crawley, UK). Total RNA (500ng) was reverse transcribed using Superscript III RT

(Invitrogen, Paisley, UK) and random hexamers according to the manufacturer’s

instructions. PCR was performed using Platinum® Taq DNA Polymerase (Invitrogen)

using b-actin as a reference gene. Nested PCR was performed for the amplification of



3

VEGF-A165, using transgene specific primers (5’ primers selected from the CMV promoter

and 3’ primers from the coding region). Primers for PCR of b-actin and VEGF-A165 were

as described.18 PCR products were run on a 2 % agarose gel, alongside 50kb DNA markers

(Invitrogen).

LacZ Staining–Detection of b-galactosidase activity was performed by overnight

incubation of whole arterial segments or 6 mm sections in the dark at 37˚C in b-gal

staining solution as described.11

Immunohistochemistry–The following antibodies were used: mouse IgG1 to rabbit

VCAM-1 (Rb1/9; gift of MI Cybulsky) at a 1:100 dilution in frozen sections; E-selectin

(10 µg/ml; R & D Systems), macrophage-specific RAM11 (1:50, Dako); mouse anti-

human CD31 (1:500; Dako); smooth muscle cell-specific a-actin antibody (1:150; Dako);

anti-VEGF mouse mAb sc-7269 (1: 200; Santa Cruz Biotechnology Inc, CA). Primary

antibodies were diluted in Tris-buffered saline, pH 7.2. Staining was performed on

deparaffinized or frozen sections as described and visualised using a Vectastain Elite ABC

Kit (Vector Laboratories, Peterborough, UK).11 The 5 mm frozen sections were fixed for

10 min in acetone at –20˚C and then air-dried. Negative controls for all immunostainings

were performed by omitting the primary antibody.

Morphometry and Image Analysis–Images of sections at x5 and x40 were acquired

with a high resolution colour camera (Zeiss microscope, Jenoptik Camera) and analyzed

using automated image analysis software (Image J, National Institute of Health). Intimas

were defined as the regions between luminal endothelium and the internal elastic lamina.

The media was defined as the area between the internal and external elastic laminas, and

analysed blindly. Intima/media ratios (I/M) were determined in serial sections cut at 500

mm intervals and stained with haematoxylin and eosin; I/M values were averaged and



4

expressed as means ± SE Neovascularisation was quantified by counting the numbers of

CD31-positive vessels. CD31 staining was regarded as positive if a single cell or a vessel

with a lumen was present and numbers of CD31-positive cells and vessels were expressed

per mm2 total adventitial area. Total RAM-11 positive macrophages were counted in the

intima and expressed per mm2. VCAM-1 and E-selectin immunostaining was quantified

using Image J on high resolution (1300 x 1030 pixel) images captured using OpenLab 3.14

software (Improvision Ltd), and for VCAM-1 is expressed as the number of pixels

representing endothelial immunostaining as a percentage of the total endothelial pixel

count, and for E-selectin is expressed as the positively stained area as a percentage of total

tissue area in the sections.

ELISA–Analysis of total (cell-surface and intracellular) VCAM-1 expression was

performed using confluent HUVECs in 96 well plates (2 ¥ 104 cells/well in quadruplicate)

as described.19 E-selectin, ICAM-1 and vWF were analysed by substituting anti-VCAM-1

with anti-E-selectin (5 mg IgG1/ml), anti-ICAM-1 (1 mg IgG1/ml) or anti-vWF (1.2 mg

IgG1/ml). Cells were incubated with 25 ng/ml VEGF in 1 % serum-containing medium

for times indicated. In some experiments TNF-a was added directly into the media for the

last 6 h to a final concentration of 100 U/ml (5 ng/ml) (or other concentrations as

indicated) except for basal (unstimulated) controls. Primary antibody binding was detected

by StreptABComplex/HRP Duet kit and 0-phenylenediamine chromogenic substrate

(Dako). Absorbances were measured at 492 nm by a microtitre plate spectrophotometer

(SpectroMax 250). Cell protein per well was measured using the Bradford reagent (Bio-

Rad Laboratories, Hemel Hempstead, UK).

Flow Cytometry–Cell surface VCAM-1, E-selectin, ICAM-1 and PECAM-1 was

measured in confluent HUVECs (3 ¥ 105 cells/well in triplicate) as described.19 Cells were



5

incubated with factors as described for ELISA and in results. After washing with PBS,

plates were placed on ice and the cells gently removed using a cell lifter. Primary antibody

binding and negative control antibody binding (at equivalent IgG1 concentrations) were

detected using goat anti-mouse FITC-conjugated antibodies. Intact cells were gated on

forward scatter versus side scatter linear amplifications and 5 ¥ 103 cells were analysed for

FITC fluorescence per well by FACScan (BD Biosciences, Cowley, UK) using Cellquest

software.

Reverse Transcription and Quantitative Real-time PCR– After treatment of confluent

HUVECs with factors as described in results, total RNA was extracted by using RNeasy

Mini Kit (QIAGEN), and quantitative real-time PCR was performed as described17 using

primers for CAMs (MWG-BIOTECH Ltd, Milton Keynes, UK) to sequences in the 3'

untranslated region of the genes to ensure the specificity of the fragment, and ensuring that

amplified fragments were 200-300 base pairs in size (see supplementary Table I).

Amplification of fragments of the predicted size was verified by conventional RT-PCR

methodology. Real-time PCR was performed by using the LightCycler-FastStart DNA

Master SYBR Green I Kit and Lightcycler System with LightCycler3 Software (Roche

Diagnostics). Data were normalised to the reference gene, GAPDH, and are presented as

the mean fold changes compared with control.

Soluble E-selectin (sE-selectin) Immunoassay–Confluent HUVECs (3 ¥ 105 cells/well in

triplicate) were preincubated with factors as described. Media was then replaced with fresh

1 % serum-containing medium and cells incubated for the times indicated. Cell culture

supernatants were centrifuged (15 000 g, 10 min) to remove particulates, diluted 1:1 with

sample diluent and assayed by commercial immunoassay for human sE-selectin (R&D

Systems). The optical density was read at 450 nm with the wavelength correction set at



6

650 nm for increased accuracy. The lower limit of detection of sE-selection was typically

~0.5 ng/ml. The intra-assay and inter-assay coefficients of variation were 4.6 % and 13.2

%, respectively. After removal of media supernatants, HUVEC monolayers were lysed in

lysis buffer [50 mM Tris-HCL, 150 mM NaCl, 10 mM EDTA, 1 % (v/v) Triton X-100, 0.5

% (v/v) NP40 and protease inhibitor cocktail tablets (CompleteTM; Roche Diagnostics),

pH7.6] and total cellular protein assayed using the Bradford assay.

Statistical Analysis-–Differences in VCAM-1 expression and morphometric differences

between different treatment groups in rabbits were evaluated by ANOVA and Bon

Ferroni Correction (SPSS). Data were considered statistically significant at p<0.05. The

statistical analysis of independent cell culture experiments was performed using Student’s t-

test, or, alternatively, one-way or two-way ANOVA with post hoc analysis by Fisher’s

PLSD, where appropriate. Results are shown as means ± SE or ± SD, and p<0.05 was

considered significant.



7

FIGURE LEGENDS FOR SUPPLEMENTARY DATA

FIG. I. Effects of Ad.VEGF delivery on intimal thickening in hypercholesterolaemic

rabbits. Panel A - periadventitial Ad.lacZ delivery to collared carotid arteries resulted in

efficient b-galactosidase expression predominantly in the adventitia. Panel B - RNA was

extracted from collared transfected carotid arteries in cholesterol-fed rabbits, and VEGF-

A165 transgene expression was determined by RT-PCR in the presence (+) or absence (-) of

template using b-actin as a reference gene. A predicted PCR product of 547 bps

corresponding to VEGF-A165 was detected only in arteries transduced with the appropriate

adenovirus and was absent from Ad.lacZ-transduced arteries (lac), or the adenoviral

backbone (Ad), or in uninfected control contralateral arteries from the same rabbit (CA).

A product of 90 bps corresponding to a-actin was present at a similar level in all samples.

RT-PCR of Ad.PlGF2 transduced in parallel into rabbit carotid arteries was used as a

negative control, and Ad.VEGF-A165 expressed in HUVECs as a positive control (+ve).

Panel C - arterial VEGF-A expression was detected by immunostaining with a specific

antibody for VEGF-A. Ad.lacZ-infected and uninfected sham-operated (control) arteries

stained with VEGF-A antibody, showed little detectable expression. VEGF-A

immunostaining in Ad.VEGF-A165-transduced arteries was detected mainly in the

adventitia, with some neointimal staining. Panel D - collar placement and gene transfer

were performed in rabbits on a normal diet and I/M ratios were determined in collared

carotid arteries infected with Ad.lacZ (n = 10), or Ad.VEGF-A165 (n=10) as described in

Experimental Procedures. Intimal thickening was significantly increased in Ad.VEGF-A165-

transduced rabbit carotid arteries. Representative H&E sections indicating the position of

the internal elastic lamina (IEL) are shown.



8

FIG. II. Angiogenic response to Ad.VEGF in collared arteries

A. Sections of collared carotid arteries in rabbits on a high cholesterol diet infected

with either no adenovirus (control), or Ad.lacZ, or Ad.VEGF-A165 were

immunostained with CD31 antibody. CD31 staining is shown at x10 (upper and

lower left photomicrographs) and x40 magnification (lower right, magnification of

boxed area in lower left photomicrograph). An increase in CD31-positive

microvessels occurred in the adventitia of arteries transduced with Ad.VEGF-A165.

B. The numbers of adventitial CD31-positive microvessels and cells in rabbits fed

high cholesterol or normal diets were quantified and results are presented as CD31-

positive vessels/mm2; *p < 0.05 for Ad.VEGF-A165 versus Ad.lacZ.

FIG III. VEGF and TNF-a synergistically increase TF mRNA levels. HUVECs were

pre-treated for 24 hours either with (VEGF, VEGF+TNF-a) or without (basal, TNF-a) 25

ng/ml VEGF, and then incubated with 100 U/ml TNF-a for a further 6 hours. Expression

of mRNA for TF (A) and COX-2 (B) was then determined by quantitative real time RT-

PCR. TF was increased by VEGF/TNF-a (grey bar, p<0.05) compared with TNF-a

alone. COX-2 showed no synergistic upregulation with VEGF and TNF-a (C). Bars

show means ± SE from 3 independent experiments using different cell batches.



Zachary Anita K. Stannard, Rohit Khurana, Ian M. Evans, Vassiliki ...

Documents

Transcript of Zachary Anita K. Stannard, Rohit Khurana, Ian M. Evans, Vassiliki ...