a

sdpc(bpbbi

LUj

Microvascular Research 59, 265–277 (2000)doi:10.1006/mvre.1999.2225, available online at http://www.idealibrary.com on

Effect of Vascular Endothelial Growth Factoron Cultured Endothelial Cell MonolayerTransport Properties

Yong S. Chang, Lance L. Munn,* Mechteld V. Hillsley, Randal O. Dull,*,†Jin Yuan,* Sunitha Lakshminarayanan, Thomas W. Gardner,‡Rakesh K. Jain,* and John M. Tarbell1

The Pennsylvania State University, Departments of Physiology and Chemical Engineering, PhysiologicalTransport Studies Laboratory, University Park, Pennsylvania 16802; ‡Department of Ophthalmology, Collegeof Medicine, Hershey, Pennsylvania 17033; Massachusetts General Hospital/ Harvard Medical School,*Edwin L. Steele Laboratory for Tumor Biology, Department of Radiation Oncology, and†Department of Anaesthesia and Critical Care, Boston, Massachusetts 02114

Received September 24, 1999

Vascular endothelial growth factor (VEGF) is a potentenhancer of microvascular permeability in vivo. To date,its effects on hydraulic conductivity (Lp) and diffusivelbumin permeability (Pe) of endothelial monolayers

have not been thoroughly assessed in vitro. We hypothe-ized that VEGF affects endothelial transport propertiesifferently depending on vessel location and endothelialhenotype. Using three well-established endothelial cellulture models—human umbilical vein endothelial cellsHUVECs), bovine aortic endothelial cells (BAECs), andovine retinal microvascular cells (BRECs)—grown onorous, polycarbonate filters we were able to produceaseline transport properties characteristic of restrictivearriers. Our results show 3.1-fold and 5.7-fold increasesn endothelial Lp for BAEC and BREC monolayers, re-

spectively, at the end of 3 h of VEGF (100 ng/ml) expo-sure. HUVECs, however, showed no significant alter-ation in Lp after 3 h (100 ng/ml) or 24 h (25 ng/ml) of

1 To whom correspondence should be addressed at 155 Fenske

ab, Department of Chemical Engineering, The Pennsylvania Stateniversity, University Park, PA 16802. Fax: (814) 865-7846. E-mail:

mt@psu.edu.

0026-2862/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

incubation with VEGF even though they were responsiveto the inflammatory mediators, thrombin (1 U/ml; 27-fold increase in Lp in 25 min) and bradykinin (10 mM;4-fold increase in Lp in 20 min). Protein kinase C (PKC)and nitric oxide (NO) are downstream effectors of VEGFsignaling. BAEC Lp was responsive to activation of NO(SNAP) and PKC (PMA), whereas these agents had noeffect in altering HUVEC Lp. Moreover, BAECs exposedto the PKC inhibitor, staurosporine (50 ng/ml), exhibitedsignificant attenuation of VEGF-induced increase in Lp,but inhibition of nitric oxide synthase (NOS) with L-NMMA (100 mM) had no effect in altering the VEGF-induced increase in Lp. These data provide strong evi-dence that in BAECs, the VEGF-induced increase in Lp ismediated by a PKC-dependent mechanism. Regardingdiffusive albumin Pe, at the end of 3 h, BAECs and BRECsshowed 6.0-fold and 9.9-fold increases in Pe in responseto VEGF (100 ng/ml), whereas VEGF had no significanteffect after 3 h (100 ng/ml) or 24 h (25 ng/ml) in changing

HUVEC Pe. In summary, these data indicate that VEGFaffects endothelial transport properties differently de-pending on the vessel type and that differences in cell

265

abatie1

a

v

b

m(br

signaling pathways underlie the differences in VEGFresponsiveness. © 2000 Academic Press

Key Words: VEGF; hydraulic conductivity; albuminpermeability; endothelial cells.

INTRODUCTION

Vascular endothelial growth factor (VEGF) is a mul-tifunctional cytokine that acts as an important regula-tor of angiogenesis, proliferation and migration ofendothelial cells, expression of adhesion molecules,and vascular permeability (Brown et al., 1997; Ferrarand Davis-Smyth, 1997; Melder et al., 1996). VEGF iselieved to play an important role in the hyperperme-bility of microvessels in tumors, the leakage of pro-eins in diabetic retinopathy, and the pathologies seenn other angiogenesis and permeability-related dis-ases (Brown et al., 1997; Ferrara and Davis-Smyth,997; Fukumura et al., 1997; Hobbs et al., 1998; Jain et

al., 1998; Roberts and Palade, 1995, 1997; Yuan et al.,1996). Due to alternative splicing, there are four dom-inant isoforms of VEGF: VEGF121, VEGF165, VEGF189,nd VEGF206 (Houck et al., 1991; Tischer et al., 1991), of

which VEGF165 is the most abundant form found inivo. VEGF189 and VEGF206 are bound to the cell or the

extracellular matrix, whereas VEGF121 is secreted insoluble form (Park et al., 1993), and VEGF165 is partly

ound and partly secreted (Houck et al., 1992). VEGFacts preferentially on endothelial cells through twohigh-affinity receptor tyrosine kinases, Flt-1 (fms-liketyrosine kinase; also known as VEGF-R1) (de Vries etal., 1992; Shibuya et al., 1990) and KDR/Flk-1 (kinasedomain region/ fetal liver kinase-1; also known asVEGF-R2) (Terman et al., 1992), which initiate signaltransduction pathways within cells.

VEGF was originally described in the context of itsability to increase permeability of microvessels in vivoand termed vascular permeability factor (VPF) (Sengeret al., 1983). Local injection of VEGF acts within min-utes, and it has been estimated, using the Miles assay,that VEGF is some 50,000 times more potent than

266

histamine (Dvorak et al., 1995; Senger et al., 1983) inenhancing transvascular transport. The vascular per-meablilizing effects of VEGF have been described in a

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

variety of tissue preparations in vivo including theskin, subcutaneous tissue, peritoneal wall, mesentery,and tumors (Brown et al., 1997; Dvorak et al., 1995;Ferrara and Davis-Smyth, 1997; Monsky et al., 1999;Nagy et al., 1995; Senger et al., 1983). VEGF has beenshown to induce fenestrations and endothelial gaps innormal and tumor vessels in vivo (Roberts and Palade,1995, 1997). Neutralizing VEGF by anti-VEGF anti-body or down-regulating by hormone withdrawalfrom a hormone-dependent tumor has been shown tolower permeability in vivo (Hobbs et al., 1998; Jain etal., 1998; Yuan et al., 1996). However, the relationshipbetween VEGF exposure and vascular permeabilityappears to be host organ dependent (Fukumura et al.,1997; Hobbs et al., 1998; Monsky et al., 1999; Yuan et al.,1994). Thus definitive conclusions regarding the directeffects of VEGF on endothelial transport propertiescannot be drawn from these in vivo studies, sincesecondary mediators within the tissue may be in-volved.

Cell culture models of endothelial monolayers limitthe source of secondary mediators and allow a moredirect investigation of the effects of VEGF on transportproperties. In our laboratory we have previously re-ported that VEGF caused a dramatic, time-dependentincrease in hydraulic conductivity (L p) (18-fold in 90

in) in bovine retinal microvascular endothelial cellsBRECs) (Yaccino et al., 1997). Wang et al. (1996) usingovine brain microvascular endothelial cells (BMECs)eported that VEGF increased [14C]sucrose transport

1.5- to 3-fold after 10–26 h of incubation. Most re-cently, Hippenstiel et al. (1998) showed that humanumbilical vein endothelial cells (HUVECs) respondwith a significant increase in L p after a 100 pg/ml doseof VEGF. This observation, however, is not consistentwith the results of the present study. A more detaileddiscussion of this point will be presented in the dis-cussion section.

The basic hypothesis of this study was that VEGFaffects endothelial transport properties differently de-pending on vessel location and endothelial pheno-type. To test this, we determined to what extent thetransport properties of three well-established endothe-

Chang et al.

lial cell culture models (human umbilical vein endo-thelial cells, HUVECs; bovine aortic endothelial cells,BAECs; and bovine retinal microvascular cells,

aaalcaw5

BRECs) responded to VEGF. We used methods de-scribed previously in our laboratories to determine theacute changes in hydraulic conductivity (L p) and al-bumin permeability (P e) of HUVEC, BAEC, and BRECmonolayers grown on porous, polycarbonate filters inresponse to VEGF. In addition, we attempted to elu-cidate some of the mechanisms by which VEGF acts toelicit its effects, either through the nitric oxide (NO) orprotein kinase C (PKC)-dependent pathways whichhave been shown to be relevant in other studies (Wu etal., 1996; Xia et al., 1996). We conclude that VEGFaffects endothelial transport properties differently de-pending on the vessel origin and cell-specific signalingpathways.

MATERIALS AND METHODS

Reagents

The following chemicals were obtained from SigmaChemical Co. (St. Louis, MO): bovine serum albumin(BSA, Fraction V, 30% solution), minimal essentialmedia (MEM), fetal bovine serum (FBS), gelatin, fi-bronectin, NG-monomethyl-l-arginine (l-NMMA),bradykinin (BK), thrombin, phorbol 12-myristate 13-acetate (PMA), and staurosporine. S-nitroso-N-acetylpenacillamine (SNAP) was obtained from Re-search Biochemicals International (Natick, MA).Polycarbonate filters (Transwell Chambers, 0.4-mmpore size, 24.5-mm diameter) were obtained fromCorning Costar (Cambridge, MA). VEGF165 was pur-chased from R&D Systems (Minneapolis, MN).

Cell Culture

HUVECs. Human umbilical vein endothelial cells(from the laboratory of Dr. Michael A. Gimbrone, Jr)were cultured in endothelial growth medium (EGM;Clonetics) which contains 2% fetal bovine serum, bo-vine brain extract (BBE) with heparin, 10 ng/ml hu-man epidermal growth factor (hEGF), and 1 mg/ml

VEGF on Cultured EC Transport

hydrocortisone. This was supplemented with 10%FBS. HUVECs were seeded at a density of 1.0 3 105

cells/cm2 onto Transwell polycarbonate filters

(0.4-mm pore size) that had been precoated with gel-atin (5 mg/ml) and fibronectin (30 mg/ml). Cells be-tween passages 4 and 6, at 3 to 5 days postplating,were used in experiments.

BAECs. Bovine aortic endothelial cells were har-vested from thoracic aortas and maintained as de-scribed previously (Sill et al., 1992). Briefly, cells werecultured and then monocloned with MEM supple-mented with 10% FBS. Cells between passages 7 and12 were plated at a density of 2.5 3 105 cells/cm2 onpretreated Transwell filters as described above. Exper-iments were performed 7 to 9 days postplating.

BRECs. Isolated bovine retinal microvascular en-dothelial cells were cultured in MEM supplementedwith 20% FBS, 10 mg/ml endothelial cell growth sup-plement (ECGS), and 5 mg/ml heparin as describedpreviously (Wong et al., 1987; Yaccino et al., 1997).BRECs between passages 6 and 12 were plated atdensity of 1.0 3 105 cells/cm2 on pretreated Transwellfilters, and experiments were carried out 5 to 8 dayspostplating.

Measurement of Lp and Pe

The experimental apparatus used to measure waterand albumin flux has been described elsewhere(Chang et al., 2000; Sill et al., 1995). Briefly, the entirepparatus was housed within a Plexiglas box and keptt an ambient air temperature of 37°C. A polycarbon-te membrane Transwell filter containing the endothe-ial monolayer was sealed between two pieces of poly-arbonate assembly separating the luminal andbluminal compartments. The luminal compartmentas continuously flushed with a positive pressure of

% CO2–95% air to maintain the pH of the medium at7.4. The abluminal compartment contained a magneticstir bar to provide uniform mixing of the chamber formacromolecule concentration determination. Twoports were available on the side of the abluminalchamber: one covered with a rubber septum for sam-pling of abluminal media and the other connected to areservoir of media replacing the volume taken duringsampling. In addition, the bottom of the abluminal

267

chamber was connected to a second, external reservoirthrough a series of tygon and glass tubing. This res-ervoir could be lowered to a desired height to create

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

mbedtatitcmppsw(5flrc

wit

w

wa

the hydrostatic pressure gradient required to drivewater flux across the cell monolayer. To eliminate theoncotic pressure gradient, the same medium(MEM-1% BSA) was added to both the luminal andthe abluminal compartments.

To begin an experiment, the polycarbonate filtercontaining the endothelial monolayer was removedfrom the culture medium (MEM-10% FBS) and rinsedtwice with experimental medium, MEM-1% BSA.Then, 2 ml of MEM-1% BSA was pipetted onto thecell surface (luminal surface). The abluminal compart-ment was filled with approximately 50 ml of MEM-1%BSA. The monolayer was inserted into the chamberwithout applying hydrostatic pressure or cylindricalstrain to the endothelial cells. The compression ringwas then tightened to adjoin the upper and lowerassemblies.

For experiments in which albumin permeability (P e)was measured, 1 ml of 0.3 mg/ml (to achieve finalconcentration of 0.01%, or 0.1 mg/ml) fluorescent-labeled albumin was added to the luminal chamber.The system was equilibrated for 1 h with the height of

edium in the abluminal and luminal compartmentsalanced to eliminate any hydrostatic pressure gradi-nt. Note that there is negligible oncotic pressure gra-ient since MEM-1% BSA fills both compartments and

he labeled albumin is in trace concentration. FITC-lbumin samples were collected at 15-min intervals forhe first h and 30-min intervals for the next 3 h bynserting a needle attached to a 1-ml syringe throughhe septum-covered side port into the abluminalhamber. The removed volume was replenished si-ultaneously by having the media-filled reservoir

ort open while samples were drawn; this eliminatedressure perturbations across the monolayer duringampling. Fluorescent quantification of the samplesas determined with a fluorescent spectrophotometer

Perkin Elmer LS-5) using 490 nm for excitation and30 nm for emission. The transendothelial albuminux ( J s) was determined for each monolayer from theate of change of the fluorescent-labeled albumin con-entration in the abluminal medium, DCA/Dt, as fol-

268

lows,

Js 5 ~DCA/Dt!~VA!, (1)

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

where J s is the transmural albumin flow rate and V A isthe volume of the abluminal medium. The permeabil-ity coefficient was then determined by

Pe 5 Js/SDC, (2)

here S is the surface area of the monolayer, and DCs the concentration differential of FITC-BSA acrosshe monolayer (taken as C L, the constant luminal con-

centration which is much greater than the abluminalconcentration).

For experiments in which L p was measured, the sideports were closed and the bottom port was opened.After an hour of equilibration, a bubble was insertedinto the borosilicate glass tubing and the external res-ervoir was lowered to produce the desired hydrostaticpressure gradient (usually 10 cm H2O). The bubblemovement representing volume flux was tracked witha spectrophotometer mounted on a threaded rodwhich was driven by a stepper motor. This travelingspectrophotometer was interfaced to a computer andthe bubble position was displayed as a function oftime on the computer screen. The bubble displacementwas converted to fluid volume flow rate ( J v) by thefollowing formula,

Jv 5 ~Dd/Dt!~F!, (3)

here Dd/Dt is the bubble displacement per unit timeand F represents the volume of fluid contained in aknown length of tubing. Since there was no oncoticpressure differential across the monolayer, L p could becalculated as follows,

Lp 5 Jv/SDP, (4)

here DP is the hydrostatic pressure differentialcross the monolayer (10 cm H2O). There may have

been a slight excess of protein near the luminal surfacebecause of concentration polarization driven by thevolume flux, but this is expected to have a negligibleoncotic effect at a protein concentration of 1% (Lever etal., 1992).

Data Presentation and Statistical Analysis

Chang et al.

As described previously (Chang et al., 2000; Sill et al.,1995), upon application of the 10 cm H2O pressure

m

ta

ort ba

head there was an initial decrease in L p which stabi-lized over a period of 30 to 50 min. Therefore, a 1-hperiod (prior to VEGF addition) was allotted to estab-lish a baseline L p before intervention. The baselinereported in Table 1 is the L p value obtained 60 minafter application of the pressure head. Five-minuteaverage L p values were calculated, normalized withrespect to the established baseline L p, and presented as

ean 6 standard error of the mean (SEM).For P e experiments, FITC-albumin samples were

collected at 15-min time intervals for the first hour(prior to VEGF addition) and 30-min intervals for thenext 3 h. The four P e values determined in the firsthour were averaged and defined as the baseline P e. AllP e values were normalized with respect to the baselinevalue. Data for P e are presented as mean 6 SEM.

Tests of significant differences between group/treat-ment means were performed every 30 min after estab-lishment of baseline L p or P e and analyzed by ananalysis of variance (ANOVA). Planned pair-wisecomparisons were performed using a Bonferroni cor-rection. P , 0.05 was used as the significance level forhe statistical analysis.

RESULTS

Baseline Endothelial Transport Properties

First, we determined the basic transport responses(hydraulic conductivity and diffusive albumin perme-ability) of three well-established endothelial cell cul-

TABLE 1

Baseline Endothelial Transport Properties of the Three Cell CultureBovine Aortic Endothelial Cells (BAECs), and Bovine Retinal Micro

Cell type nBaseline EC L p Mean

(3 1027 cm/s/cm

HUVECs 14 3.55 6 0.39BAECs 16 3.58 6 0.18BRECs 17 2.79 6 0.33

Note. All three cell culture models exhibited physiological transp

VEGF on Cultured EC Transport

ture models to VEGF. The cell culture systems differedwith respect to culture duration, or days postplatingrequired to achieve minimal transport rates: HUVECs

required only 3 to 5 days whereas BRECs needed 5 to8 days. BAECs required higher plating density (2.5 3105 cells/cm2 compared to 1 3 105 cells/cm2 for HU-VECs and BRECs) as well as more time (7 to 9 days) inculture to achieve maximal barrier properties. How-ever, all cell types had similar baseline transport rates(Table 1), and these rates were consistent with physi-ological transport barrier properties reported previ-ously from our laboratory and observed in selectedintact vessels (Chang et al., 2000; Dull et al., 1991; Jo etal., 1991; Sill et al., 1992, 1995; Wu et al., 1996; Yaccinoet al., 1997; Yuan et al., 1992).

VEGF Affects Hydraulic Conductivity DifferentlyDepending on Endothelial Cell Type

Figure 1 illustrates endothelial L p in response toVEGF. Upon application of 10 cm H2O pressure atTime 0, HUVEC L p (Fig. 1a) continuously decreasedover the 4-h period. Baseline L p for 0 ng/ml (control,n 5 5), 25 ng/ml VEGF (n 5 4), and 100 ng/ml (n 55) VEGF were 3.42 6 0.72 3 1027, 4.43 6 0.20 3 1027,nd 2.98 6 0.81 3 1027 cm/s/cm H2O, respectively.

Surprisingly, addition of VEGF (25 or 100 ng/ml) tothese monolayers had no significant effect on endo-thelial L p. At the end of 3 h, the L p of HUVECs treatedwith 0 (control), 25, and 100 ng/ml VEGF decreasedby 54, 80, and 76%, respectively. The decrease inHUVEC L p at the end of 3 h for 25 and 100 ng/mlVEGF was not significantly different from the de-crease in the control group (P . 0.15).

BAECs (Fig. 1b) and BRECs (Fig. 1c), on the other

ls: Human Umbilical Vein Endothelial Cells (HUVECs),ar Endothelial Cells (BRECs)

nBaseline EC P e Mean 6 SEM

(3 1026 cm/s)

12 1.93 6 0.3910 2.96 6 0.4116 2.99 6 0.43

rrier properties similar to those reported in the literature.

269

Modevascul

6 SEMH2O)

hand, showed significant increases in response toVEGF (100 ng/ml). BAEC monolayers exhibited abaseline L p of 3.27 6 0.31 3 1027 cm/s/cm H2O (n 5

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

B

(

r BAEld, resata ar

6) that increased rapidly for the first 90 min afterVEGF addition (2.84 6 0.87-fold above baseline).

AEC L p then stabilized and remained elevated forthe next 90 min. After 3 h of exposure to VEGF, BAECL p had increased by 3.05 6 0.24-fold over baselineP , 0.002). In the same time period, BREC L p (base-

line of 2.62 6 0.35 3 1027 cm/s/cm H2O (n 5 6))increased by a factor of 5.71 6 0.89 (P , 0.001).

FIG. 1. Responses of HUVEC (a), BAEC (b), and BREC (c) monolaVEGF was added. VEGF induced a time-dependent increase in L p foBREC (n 5 6) L p increased by 3.05 6 0.24-fold and 5.71 6 0.89-foHUVEC L p. * P , 0.05 for VEGF compared to stationary control. D

270

Interestingly, the increases in L p for BAECs andBRECs were similar but the times at which these in-creases occurred were different: BAEC monolayers

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

demonstrated a more rapid response to VEGF (15-minlag period before L p began to rise above the controlvalue) compared to BRECs which exhibited a 60-minlag period, indicating that different mechanisms maybe responsible for the observed responses.

VEGF-Induced Albumin Diffusion across MonolayersVaries with Cell Type

to VEGF. After 60 min, during which a baseline L p was established,C and BREC monolayers. After 3 h of exposure, BAEC (n 5 6) andpectively. Surprisingly, however, VEGF had no effect on alteringe presented as means 6 SEM.

Chang et al.

yer L p

Figure 2 depicts the effect of VEGF on endothelialmonolayer P e. Baseline P e values for 0 ng/ml (control;

ba

1l

s1

C andd of 3 h

B ompare

n 5 3), 25 ng/ml (n 5 4), and 100 ng/ml (n 5 5)VEGF were 3.09 6 1.27 3 1026, 1.21 6 0.21 3 1026, and1.81 6 0.40 3 1026 cm/s, respectively, for HUVECs.Similar to the L p response, VEGF had no influence onalbumin permeability across HUVEC monolayers

FIG. 2. Effect of VEGF on HUVEC (a), BAEC (b), and BREC (c) monon altering HUVEC P e. VEGF, however, significantly increased BAE6) P e began to increase significantly in response to VEGF. At the en

AEC and BREC monolayers, respectively. * P , 0.05 for VEGF c

VEGF on Cultured EC Transport

(Fig. 2a) (P . 0.6). However, both BAEC (100 ng/ml;aseline P e of 3.28 6 0.62 3 1026 cm/s (n 5 6); Fig. 2b)nd BREC (100 ng/ml; baseline P e of 2.86 6 0.55 3

otH

026 cm/s (n 5 6); Fig. 2c) monolayers displayed a 1-hag period in which P e did not change, after which a

significant increase in diffusive albumin P e was ob-erved. At 3 h, 6.02 6 1.09-fold (P , 0.01) and 9.86 6.53-fold (P , 0.001) increases above baseline were

P e. Similar to the L p response, VEGF (25 or 100 ng/ml) had no effectBREC P e. Following a 1-h lag period, BAEC (n 5 6) and BREC (n 5

, 6.02 6 1.09-fold and 9.86 6 1.53-fold increases were observed ford to stationary control. Data are presented as means 6 SEM.

271

olayer

bserved for BAEC and BREC monolayers, respec-ively. In additional experiments, preincubation of

UVECs with VEGF (25 ng/ml) for 24 h did not lead

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

b(

sp

to a significant deviation of L p (Fig. 3a) or P e (Fig. 3b)from baseline (P . 0.25).

Thrombin and Bradykinin Increase HUVECHydraulic Conductivity

FIG. 3. Effect of 24-h incubation of HUVEC monolayers withVEGF. Following a 24-h incubation with 25 ng/ml VEGF, HUVECmonolayer L p (a) and P e (b) were measured for 4 h following theame protocol employed for short-term treatments. The value de-icted in this figure represent L p and P e at the end of the 4 h. Similar

to Figs. 1a and 2a, this challenge also failed to alter HUVEC endo-thelial transport properties. Data are presented as means 6 SEM.

272

Because a response to VEGF was not observed forHUVECs, we examined the functionality of HUVEC

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

monolayers by challenging the cells with the inflam-matory mediators, thrombin and bradykinin (BK) (Fig.4). After the establishment of a baseline L p of 2.38 60.74 3 1027 cm/s/cm H2O (n 5 4) for 60 min, throm-bin (1 U/ml) was added. Within minutes of exposure,HUVEC L p increased rapidly, and by 5 min L p waselevated by 2.43 6 0.51-fold. L p continued to climband reached a maximum level of 27.35 6 9.25-foldabove baseline at 25 min after the treatment. Thereaf-ter, L p decreased dramatically, at a faster rate thanrecorded for the rapid increase, returning close to thebaseline level within 35 min after initial exposure tothrombin. Bradykinin (10 mm) showed similar, butless dramatic effects on HUVECs, producing a 3.96 61.88-fold increase in L p within 20 min of treatment.

PKC and NO-Dependent Pathways Involvementin HUVEC and BAEC Hydraulic Conductivity

Since HUVECs were responsive to inflammatorymediators, we hypothesized that the unresponsive-ness of HUVECs to VEGF may be due to alternatesignaling pathways in HUVECs compared to BAECsand BRECs. To test this hypothesis, we challengedHUVECs and BAECs with signaling molecules thatare known to be activated by VEGF: PKC and NO.

FIG. 4. Response of HUVEC L p to inflammatory mediators, throm-in and BK. After establishment of a baseline L p, 1 U/ml thrombinn 5 4) or 10 mM BK (n 5 4) was added to the luminal chamber.

Chang et al.

Approximately 4- and 27-fold increases in L p were observed for BKand thrombin, respectively, within 25 min of exposure, indicating aphysiological response. Data are presented as means 6 SEM.

4

la

nsi

e

sd(

The effects of phorbol ester, phorbol 12-myristate13-acetate (PMA) (1 mM) a PKC agonist, on endothe-lial L p are shown in Fig. 5. Similar to VEGF, PMA (n 5

) had no effect on HUVEC L p, whereas following a30-min delay, BAEC L p increased significantly to aevel 2.31 6 0.14-fold above baseline by 70 min afterddition of PMA. The NO donor, S-nitroso-N-

acetylpenacillamine (SNAP) (500 mM; n 5 4), alsofailed to elicit a response from HUVEC monolayers,but produced a 4.82 6 0.66-fold increase in L p inBAECs at the end of 3 h (Fig. 6).

Next, to further elucidate the downstream mecha-nisms responsible for the VEGF-induced increases inBAEC L p, monolayers were exposed to VEGF (100

g/ml) in the presence of the PKC inhibitor, stauro-porine (50 ng/ml), or the nitric oxide synthase inhib-tor, l-NMMA. The NOS inhibitor concentration used

was high enough to completely block NO productionin BAECs exposed to 20 dyn/cm2 shear stress in arelated study (Chang et al., 2000). The inhibitors alonehad no effect on the baseline L p values over the 4-htime frame of experiments. In L p experiments, mono-

FIG. 5. Effect of PMA (1 mM) on HUVEC and BAEC L p. PMA wasadded at time 60 min. In contrast to HUVEC (n 5 4) L p which wasunresponsive to PMA, BAEC (n 5 4) L p increased 2.3-fold within 70min following the addition of this pharmacological agent. Data arepresented as means 6 SEM.

VEGF on Cultured EC Transport

layers challenged with VEGF alone (n 5 7) displayedan increase in L p of 3.38 6 0.87-fold after 3 h ofxposure, as usual; but the L p of monolayers treated L

with VEGF and staurosporine (n 5 7) increased byonly 1.32 6 0.18-fold at the end of 3 h. This representsa significant inhibition (P , 0.04) of the VEGF re-ponse by the PKC inhibitor. In contrast, l-NMMAid not antagonize the increase induced by VEGFP . 0.25) as treatment with both molecules pro-

duced an L p value of 5.09 6 0.28-fold at the end of 3 h.

DISCUSSION

Diversity in Endothelial Response to VEGF

In this study, we investigated the effect of VEGF onpressure-driven L p and diffusive (nonconvective) P e

transport across endothelial cell monolayers. BAECsand BRECs were quite responsive to VEGF whereasHUVECs were not. This differential response mayhave been related to the levels of expression of theKDR receptor which has been reported to be highest inBRECs, followed closely by BAECs and then HUVECs.Specifically, compared to HUVECs which exhibit arange from 2 6 1 to 24 6 5 3 103 receptors/cell,BRECs and BAECs express 154 6 41 3 103 and 54 629 3 103 receptors/cell, respectively (Olander et al.,1991; Thieme et al., 1995).

FIG. 6. Effect of NO donor, SNAP (500 mM), on HUVEC and

273

BAEC L p. Addition of SNAP at Time 60 min resulted in a time-dependent increase in BAEC (n 5 4) L p. Note that HUVEC (n 5 4)

p was unresponsive to SNAP. Data are presented as means 6 SEM.

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

sdt

em

cstwhr5

eadGH

The inability of VEGF to produce changes inHUVEC transport properties is perplexing, consider-ing that these same cells show other responses tosimilar concentrations of VEGF. For example, VEGF isknown to stimulate the tyrosine phosphorylation ofthe Flt-1 and KDR receptors in HUVECs (Abedi andZachary, 1997). In addition, HUVECs up-regulatemany adhesion molecules after 24 h of incubation withVEGF (Melder et al., 1996). They are also acutely sen-itive to thrombin and BK administration (Fig. 4) in-icating that at least some transport signal transduc-

ion pathways are functioning.In apparent contradiction to our study, Hippenstiel

t al. (1998) reported significant increases in HUVEConolayer L p in response to 100 pg/ml doses of

VEGF. There were, however, significant differencesbetween the methods of the present study and those ofHippenstiel et al. (1998) that may account for the dis-repancy. The baseline L p values reported by Hippen-tiel et al. (1998) were approximately 15 times higherhan those of the present study, and in a range that

ould not normally be considered physiologic. Theigher baselines obtained in their study may be theesult of several factors. First, they used filters with-mm pores compared to our 0.4-mm pores. The large

pores are known to promote growth of endothelialcells into the pore structure (Albelda et al., 1988), pro-ducing a nonphysiologic morphology that may affectbarrier formation. Second, they used cells at the thirdpassage whereas our cells were generally passages4–6. We did, however, conduct preliminary experi-ments with cells at passage 3 and did not obtain aresponse of L p to VEGF. Third, we used a growthmedia that contained growth factors (EGM, Clonetics)whereas they used M199 without growth factors. Wedid, however, perform experiments with cells ingrowth factor-deficient medium (MEM), but still didnot observe changes in L p in response to VEGF. Fi-nally, it should be noted that Hippenstiel et al. (1998)prepared their own VEGF in transfected insect cellswhereas we used commercially available material(R&D).

The mechanism by which VEGF enhances perme-

274

ability is still not completely understood. Studies haveshown that a number of proteins are phosphorylatedin response to VEGF including phospholipase Cg,

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

phosphatidylinositol 3-kinase, ras GTPase, PKC, andmitogen-activated protein kinase (D’Angelo et al.,1995; Guo et al., 1995; Xia et al., 1996). In addition, Wut al. (1996) showed that VEGF-induced hyperperme-bility to albumin in porcine coronary venules is me-iated by a NO-guanylate cyclase (GC)-protein kinase(PKG)-dependent mechanism. Our attempt to alter

UVEC L p by activating similar pathways with theaddition of phorbol ester, PMA, or the NO donor,SNAP, resulted in a reduction of L p, not an increase(Figs. 5 and 6).

These pharmacological agents, however, did en-hance BAEC L p (Figs. 5 and 6). Moreover, the increasesinduced by PMA were similar in time course andmagnitude to those observed with VEGF which couldbe inhibited/attenuated with a PKC inhibitor, staruo-sporine. The effects of PMA on L p in BAECs andHUVECs parallel those reported by Yamada and Ya-kota (1996) and Yamada et al. (1990) who found thatPMA induced significant increases in albumin perme-ability in BAECs, but reduced the albumin permeabil-ity of HUVECs. Taken together, these data providestrong evidence that in BAECs, the VEGF-inducedincrease in L p is mediated by a PKC-dependent mech-anism.

Transport Pathways Involved in VEGF-MediatedResponses

Several recent studies have suggested that in orderto observe the effects of VEGF on endothelial trans-port in vitro, a differentiated state must be establishedin which transport structures such as vesicles andvesiculo-vacuolar organelles (VVOs) are preserved.Vasile et al. (1999) reported that an increase in theproduction of VVOs in response to VEGF was ob-served in bovine adrenal cortex microvascular endo-thelial cells (BAMECs) cultured on floating Matrigel-collagen type I gels, but not in HUVECs. Similarly,Esser et al. (1998) reported endothelial fenestrations inBAMECs cocultured with VEGF-transfected epithelialcells or epithelial-derived basal lamina-type extracel-lular matrix. Under the same cocultured conditions,

Chang et al.

no fenestrations were observed in HUVECs.We observed increases in the transport properties of

BAECs and BRECs cultured in a traditional manner

A

B

B

C

D

D

d

D

D

E

F

F

G

without any special effort to up-regulate vesicles. Theresponse of L p to VEGF in BAECs and BRECs is mostlikely the result of an alteration in the paracellularpathway which is generally believed to be the pre-dominant pathway for water flux (Roberts and Palade,1995, 1997). However, the response of albumin P e toVEGF in BAECs and BRECs could be mediated, atleast in part, by vesicles. Evidence for this is apparentin our data that show a 1-h delay in the onset of P e

increase after VEGF application in BAECs and BRECs(Fig. 2). This delay for albumin transport suggests thatalbumin traverses the endothelium by a differentpathway than water. Recently, Bates (1998) reportedthat VEGF increased microvascular L p without affect-ing the oncotic reflection coefficient. Candidates forthis alternative pathway which are modulated byVEGF include “leaky junctions” or vesicles. The exis-tence of alternative pathways for water and albuminin BAECs was deduced in a previous study (Dull et al.,1991), and the presence of a significant population ofvesicles, at least in BAECs, is well established (Davieset al., 1984; Sundqvist and Liu, 1993).

It seems, therefore, that BAECs and BRECs, culturedin the manner that we have described, are usefulmodels for investigating the effects of VEGF on endo-thelial transport properties. Future studies of both thephysical and the bio-molecular pathways which con-trol water and solute transport in response to VEGF inthese in vitro systems will contribute to our under-standing of the physiological and pathophysiologicalfunctions of VEGF.

ACKNOWLEDGMENTS

The research was supported by the Whitaker Foundation, NIHTraining Grant T32-GM08619-01, NIH Grant HL-57093, NIH GrantEY-12021, and NIH Grant CA-56591.

REFERENCES

VEGF on Cultured EC Transport

Abedi, H., and Zachary, I. (1997). Vascular endothelial growth factorstimulates tyrosine phosphorylation and recruitment to new focal

H

adhesions of focal adhesion kinase and paxillin in endothelialcells. J. Biol. Chem. 272, 15442–15451.

lbelda, S. M., Sampson, P. M., Haselton, F. R., McNiff, J. M.,Mueller, S. N., Williams, S. K., Fishman, A. P., and Levine, E. M.(1988). Permeability characteristics of cultured endothelial cellmonolayers. J. Appl. Physiol. 64, 308–322.

ates, D. O. (1998). The chronic effect of vascular endothelial growthfactor on individually perfused frog mesenteric microvessels.J. Physiol. (London) 513, 225–233.

rown, L. F., Detmar, M., Claffey, K., Nagy, J. A., Feng, D., Dvorak,A. M., and Dvorak, H. F. (1997). Vascular permeability factor/vascular endothelial growth factor: A multifunctional angiogeniccytokine. Exs 79, 233–269.

hang, Y., Yaccino, J., Frangos, J., and Tarbell. (2000). Shear-inducedincrease in hydraulic conductivity in endothelial cells is mediatedby a nitric oxide-dependent mechanism. Arterioscler. Thromb.Vasc. Biol. in press.

’Angelo, G., Struman, I., Martial, J., and Weiner, R. I. (1995).Activation of mitogen-activated protein kinases by vascular en-dothelial growth factor and basic fibroblast growth factor incapillary endothelial cells is inhibited by the antiangiogenic factor16-kDa N- terminal fragment of prolactin. Proc. Natl. Acad. Sci.USA 92, 6374–6378.

avies, P. F., Dewey, C. F., Jr., Bussolari, S. R., Gordon, E. J., andGimbrone, M. A., Jr. (1984). Influence of hemodynamic forces onvascular endothelial function. In vitro studies of shear stress andpinocytosis in bovine aortic cells. J. Clin. Invest. 73, 1121–1129.

e Vries, C., Escobedo, J. A., Ueno, H., Houck, K., Ferrara, N., andWilliams, L. T. (1992). The fms-like tyrosine kinase, a receptor forvascular endothelial growth factor. Science 255, 989–991.

ull, R. O., Jo, H., Sill, H., Hollis, T. M., and Tarbell, J. M. (1991). Theeffect of varying albumin concentration and hydrostatic pressureon hydraulic conductivity and albumin permeability of culturedendothelial monolayers. Microvasc. Res. 41, 390–407.

vorak, H. F., Brown, L. F., Detmar, M., and Dvorak, A. M. (1995).Vascular permeability factor/vascular endothelial growth factor,microvascular hyperpermeability, and angiogenesis. Am. J.Pathol. 146, 1029–1039.

sser, S., Wolburg, K., Wolburg, H., Breier, G., Kurzchalia, T., andRisau, W. (1998). Vascular endothelial growth factor inducesendothelial fenestrations in vitro. J. Cell Biol. 140, 947–959.

errara, N., and Davis-Smyth, T. (1997). The biology of vascularendothelial growth factor. Endocr. Rev. 18, 4–25.

ukumura, D., Yuan, F., Monsky, W. L., Chen, Y., and Jain, R. K.(1997). Effect of host microenvironment on the microcirculationof human colon adenocarcinoma. Am. J. Pathol. 151, 679–688.

uo, D., Jia, Q., Song, H. Y., Warren, R. S., and Donner, D. B. (1995).Vascular endothelial cell growth factor promotes tyrosine phos-phorylation of mediators of signal transduction that contain SH2domains. Association with endothelial cell proliferation. J. Biol.Chem. 270, 6729–6733.

275

ippenstiel, S., Krull, M., Ikemann, A., Risau, W., Clauss, M., andSuttorp, N. (1998). VEGF induces hyperpermeability by a directaction on endothelial cells. Am. J. Physiol. 274, L678–L684.

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

R

W

W

X

Y

Hobbs, S. K., Monsky, W. L., Yuan, F., Roberts, W. G., Griffith, L.,Torchilin, V. P., and Jain, R. K. (1998). Regulation of transportpathways in tumor vessels: Role of tumor type and microenvi-ronment. Proc. Natl. Acad. Sci. USA 95, 4607–4612.

Houck, K. A., Ferrara, N., Winer, J., Cachianes, G., Li, B., and Leung,D. W. (1991). The vascular endothelial growth factor family:Identification of a fourth molecular species and characterizationof alternative splicing of RNA. Mol. Endocrinol. 5, 1806–1814.

Houck, K. A., Leung, D. W., Rowland, A. M., Winer, J., and Ferrara,N. (1992). Dual regulation of vascular endothelial growth factorbioavailability by genetic and proteolytic mechanisms. J. Biol.Chem. 267, 26031–26037.

Jain, R. K., Safabakhsh, N., Sckell, A., Chen, Y., Jiang, P., Benjamin,L., Yuan, F., and Keshet, E. (1998). Endothelial cell death, angio-genesis, and microvascular function after castration in an andro-gen-dependent tumor: Role of vascular endothelial growth factor.Proc. Natl. Acad. Sci. USA 95, 10820–10825.

Jo, H., Dull, R. O., Hollis, T. M., and Tarbell, J. M. (1991). Endothelialalbumin permeability is shear dependent, time dependent, andreversible. Am. J. Physiol. 260, H1992–H1996.

Lever, M. J., Tarbell, J. M., and Caro, C. G. (1992). The effect ofluminal flow in rabbit carotid artery on transmural fluid trans-port. Exp. Physiol. 77, 553–563.

Melder, R. J., Koenig, G. C., Witwer, B. P., Safabakhsh, N., Munn,L. L., and Jain, R. K. (1996). During angiogenesis, vascular endo-thelial growth factor and basic fibroblast growth factor regulatenatural killer cell adhesion to tumor endothelium [see com-ments]. Nature Med. 2, 992–997.

Monsky, W., Fukumura, D., Gohongi, T., Ancukiewcz, M., Weich,H., Torchilin, V., Yuan, F., and Jain, R. (1999). Augmentation oftransvascular transport of macromolecules and nanoparticles intumor using vascular endothelial growth factor and placentagrowth factor. Cancer Res. 59, 4129–4135.

Nagy, J. A., Masse, E. M., Herzberg, K. T., Meyers, M. S., Yeo, K. T.,Yeo, T. K., Sioussat, T. M., and Dvorak, H. F. (1995). Pathogenesisof ascites tumor growth: Vascular permeability factor, vascularhyperpermeability, and ascites fluid accumulation. Cancer Res. 55,360–368.

Olander, J. V., Connolly, D. T., and DeLarco, J. E. (1991). Specificbinding of vascular permeability factor to endothelial cells. Bio-chem. Biophys. Res. Commun. 175, 68–76.

Park, J. E., Keller, G. A., and Ferrara, N. (1993). The vascular endo-thelial growth factor (VEGF) isoforms: Differential depositioninto the subepithelial extracellular matrix and bioactivity of ex-tracellular matrix-bound VEGF. Mol. Biol. Cell. 4, 1317–1326.

oberts, W. G., and Palade, G. E. (1995). Increased microvascularpermeability and endothelial fenestration induced by vascularendothelial growth factor. J. Cell. Sci. 108, 2369–2379.

Roberts, W. G., and Palade, G. E. (1997). Neovasculature induced byvascular endothelial growth factor is fenestrated. Cancer Res. 57,

276

765–772.Senger, D. R., Galli, S. J., Dvorak, A. M., Perruzzi, C. A., Harvey,

V. S., and Dvorak, H. F. (1983). Tumor cells secrete a vascular

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

permeability factor that promotes accumulation of ascites fluid.Science 219, 983–985.

Shibuya, M., Yamaguchi, S., Yamane, A., Ikeda, T., Tojo, A., Ma-tsushime, H., and Sato, M. (1990). Nucleotide sequence and ex-pression of a novel human receptor-type tyrosine kinase gene (flt)closely related to the fms family. Oncogene 5, 519–524.

Sill, H., Butler, C., Hollis, T., and Tarbell, J. (1992). Albuminpermeability and electrical conductivity as means of assessingendothelial cell monolayer integrity. J. Tissue Culture Methods14, 253–258.

Sill, H. W., Chang, Y. S., Artman, J. R., Frangos, J. A., Hollis, T. M.,and Tarbell, J. M. (1995). Shear stress increases hydraulic conduc-tivity of cultured endothelial monolayers. Am. J. Physiol. 268,H535–H543.

Sundqvist, T., and Liu, S. M. (1993). Hydrogen peroxide stimulatesendocytosis in cultured bovine aortic endothelial cells. ActaPhysiol. Scand. 149, 127–131.

Terman, B. I., Dougher-Vermazen, M., Carrion, M. E., Dimitrov, D.,Armellino, D. C., Gospodarowicz, D., and Bohlen, P. (1992). Iden-tification of the KDR tyrosine kinase as a receptor for vascularendothelial cell growth factor. Biochem. Biophys. Res. Commun. 187,1579–1586.

Thieme, H., Aiello, L. P., Takagi, H., Ferrara, N., and King, G. L.(1995). Comparative analysis of vascular endothelial growth fac-tor receptors on retinal and aortic vascular endothelial cells.Diabetes 44, 98–103.

Tischer, E., Mitchell, R., Hartman, T., Silva, M., Gospodarowicz, D.,Fiddes, J. C., and Abraham, J. A. (1991). The human gene forvascular endothelial growth factor. Multiple protein forms areencoded through alternative exon splicing. J. Biol. Chem. 266,11947–11954.

Vasile, E., Qu, H., Dvorak, H. F., and Dvorak, A. M. (1999). Caveolaeand vesiculo-vacuolar organelles in bovine capillary endothelialcells cultured with VPF/VEGF on floating Matrigel-collagen gels.J. Histochem. Cytochem. 47, 159–167.

Wang, W., Merrill, M. J., and Borchardt, R. T. (1996). Vascularendothelial growth factor affects permeability of brain microves-sel endothelial cells in vitro. Am. J. Physiol. 271, C1973–C1980.ong, H. C., Boulton, M., Marshall, J., and Clark, P. (1987). Growthof retinal capillary endothelia using pericyte conditioned me-dium. Invest. Ophthalmol. Vis. Sci. 28, 1767–1775.u, H. M., Huang, Q., Yuan, Y., and Granger, H. J. (1996). VEGFinduces NO-dependent hyperpermeability in coronary venules.Am. J. Physiol. 271, H2735–H2739.

ia, P., Aiello, L. P., Ishii, H., Jiang, Z. Y., Park, D. J., Robinson, G. S.,Takagi, H., Newsome, W. P., Jirousek, M. R., and King, G. L.(1996). Characterization of vascular endothelial growth factor’seffect on the activation of protein kinase C, its isoforms, andendothelial cell growth. J. Clin. Invest. 98, 2018–2026.

accino, J. A., Chang, Y. S., Hollis, T. M., Gardner, T. W., and

Chang et al.

Tarbell, J. M. (1997). Physiological transport properties of cul-tured retinal microvascular endothelial cell monolayers. Curr. EyeRes. 16, 761–768.

Y

Y

Yamada, Y., Furumichi, T., Furui, H., Yokoi, T., Ito, T., Yamauchi,K., Yokota, M., Hayashi, H., and Saito, H. (1990). Roles of calcium,cyclic nucleotides, and protein kinase C in regulation of endothe-lial permeability. Arteriosclerosis 10, 410–420.

Yamada, Y., and Yokota, M. (1996). Enhancement of barrier functionof human aortic endothelial cells by activators of protein kinaseC. Biochem. Mol. Biol. Int. 39, 69–76.

VEGF on Cultured EC Transport

uan, F., Chen, Y., Dellian, M., Safabakhsh, N., Ferrara, N., andJain, R. K. (1996). Time-dependent vascular regression andpermeability changes in established human tumor xenografts

induced by an anti-vascular endothelial growth factor/vascu-lar permeability factor antibody. Proc. Natl. Acad. Sci. USA 93,14765–14770.

uan, F., Salehi, H. A., Boucher, Y., Vasthare, U. S., Tuma, R. F.,and Jain, R. K. (1994). Vascular permeability and microcircu-lation of gliomas and mammary carcinomas transplanted in ratand mouse cranial windows. Cancer Res. 54, 4564 – 4568.

277

Yuan, Y., Granger, H. J., Zawieja, D. C., and Chilian, W. M. (1992).Flow modulates coronary venular permeability by a nitric oxide-related mechanism. Am. J. Physiol. 263, H641–H646.

Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.