VEGF Increases ParacellularPermeability in Brain Endothelial Cells

via Upregulation of EphA2ZIWEI MIAO, YANBIN DONG, WENGANG FANG, DESHU SHANG, DONGXIN LIU,

KE ZHANG, BO LI, AND YU-HUA CHEN*Department of Developmental Cell Biology, Key Laboratory of Cell Biology, Ministry of

Public Health, and Key Laboratory of Medical Cell Biology, Ministry of Education, ChinaMedical University, Heping District, Shenyang, China

ABSTRACTNeurological disorders are associated with an increase in the perme-

ability of human brain microvascular endothelial cells (HBMEC). Our pre-vious findings have indicated that EphA2 could increase the permeabilityof HBMEC. Recent evidence has linked EphA2 and vascular endothelialgrowth factor (VEGF) to abnormalities in the vascular response. However,it is unclear whether EphA2 is involved in the VEGF-induced changes inthe permeability of HBMEC. Here, changes in permeability were deter-mined by measuring transendothelial electrical resistance (TEER) and theflux of FITC-dextran. We found that knockdown of EphA2 in HBMEC abol-ished the VEGF-induced reduction in TEER and increase in flux of fluores-cent dextran. Moreover, VEGF-induced redistribution of ZO-1 and therecruitment of detergent-soluble occludin and claudin-5 were also pre-vented. Further results showed that VEGF increased EphA2 expression ina time- and dose-dependent manner, which was inhibited by a neutralizingantibody against VEGFR2 or SU1498. VEGF-induced EphA2 expressionwas suppressed in the brain endothelium following treatments with thePI3K inhibitor LY294002, Akt inhibitor or transfection with the dominant-negative PI3K mutants (Dp110). Similar results were obtained when ERK1/2 activation was inhibited by PD98059 or ERK1/2 siRNA transfection. Ourdata suggest that VEGF upregulates the expression of EphA2 in HBMECthrough binding to VEGFR2 and subsequently activating the intracellularPI3K/Akt and ERK1/2 signaling pathways, which contribute to an increasein paracellular permeability. These data reveal a novel role for VEGF as aregulator of EphA2 expression in the brain endothelial cells and provideinsights into the molecular mechanisms of VEGF-mediated changes in par-acellular permeability. Anat Rec, 297:964–972, 2014. VC 2014 WileyPeriodicals, Inc.

Key words: EphA2; VEGF; human brain microvascular endo-thelial cell; PI3K/Akt; ERK1/2

Abnormalities in the blood–brain barrier (BBB) per-meability are associated with serious pathological eventsin a variety of neurological disorders, including stroke,bacterial meningitis, Alzheimer’s disease, multiple scle-rosis, and brain tumors (Zlokovic, 2008). It is well knownthat vascular endothelial growth factor (VEGF), alsoknown as the vascular permeability factor, plays a pri-mary role in disrupting the BBB, which consists ofmicrovascular endothelial cells lining the brain micro-

Grant sponsor: National Natural Science Foundation of China;Grant numbers: 30970120, 31171291, 81171537, 81101225.

*Correspondence to: Yu-Hua Chen, Department of Develop-mental Cell Biology, Key Lab of Cell Biology, Ministry of PublicHealth of China, China Medical University, Shenyang 110001,China. Fax: 186-24-23260246. E-mail: yhchen@mail.cmu.edu.cn

Received 16 October 2013; Accepted 13 December 2013.

DOI 10.1002/ar.22878Published online 24 January 2014 in Wiley Online Library(wileyonlinelibrary.com).

THE ANATOMICAL RECORD 297:964–972 (2014)

VVC 2014 WILEY PERIODICALS, INC.

vessels (Ma et al., 2012). VEGF has been demonstratedto increase paracellular permeability by downregulatingthe expression of tight junction (TJ) proteins, such asclaudin-5, occludin, and zonula occludens-1 (ZO-1), inthe brain endothelial cells (Argaw et al., 2009; Morin-Brureau et al., 2011). Conversely, VEGF has been shownto have a negative influence on BBB permeability afterintracerebral hemorrhage, which is closely related to anincrease in AQP4 expression induced by VEGF (Chuet al., 2013). Thus, it is possible that VEGF may triggerthe expression of different molecules in the brain endo-thelial cells to regulate paracellular permeability.

Recent evidence has linked VEGF and EphA2 to neo-vascularization and vasopermeability abnormalities.EphA2, a member of the largest known family of proteintyrosine kinase receptors, has been implicated in theendothelial cells migration and vascular assembly(Brantley-Sieders et al., 2004). In bovine retinal endo-thelial cells, EphA2 receptor stimulation by its ligandephrinA1 has been shown to inhibit VEGF-induced phos-phorylation of VEGF receptor 2 (VEGFR2), leading to areduction in the VEGF-induced angiogenic cell activity(Ojima et al., 2006). The alterations in paracellular per-meability induced by EphA2-ephrinA1 have been shownto be similar to the changes induced by VEGF in thelung vascular endothelial cells (Larson et al. 2008). Ourprevious findings indicated that EphA2 was involved inmodulating paracellular permeability in human brainmicrovascular endothelial cells (HBMEC) (Zhou et al.,2011). However, it is still unclear whether EphA2 areinvolved in VEGF-induced paracellular permeability inthe brain endothelial cells.

In the present study, we provide evidence to demon-strate that VEGF upregulates the expression of EphA2,leading to an increase in the permeability of HBMEC.Furthermore, VEGFR2, the principal VEGF receptor, andits associated downstream signaling cascades, includingphosphoinositide 3-kinase (PI3K)/Akt and ERK1/2, con-tribute to VEGF-induced EphA2 expression.

MATERIALS AND METHODS

Chemicals and Antibodies

Recombinant human VEGF and neutralizing antibodiesagainst VEGFR1 and VEGFR2 were purchased from R&DSystems (Minneapolis, MN). The inhibitors, PD98059,SP600125, LY294002, wortmannin, Y27632, G€o6976,G€o6983, PP1, PP2, PP3, and SU1498 were obtained fromCalbiochem (San Diego, CA). Akt1/2 kinase inhibitor andFITC-dextran were obtained from Sigma-Aldrich (St.Louis, MO). Antibodies against VEGFR2, occludin, andFITC-conjugated ZO-1 were purchased from Invitrogen(Grand Island, NY). The anti-claudin-5 antibody was pur-chased from Abcam (Cambridge, MA). The anti-VEGFR1antibody was obtained from Santa Cruz Biotechnology(Santa Cruz, CA). The anti-EphA2 antibody was pur-chased from Millipore (Billerica, MA). Antibodies againstERK, phospho-ERK, Akt, and phospho-Akt were obtainedfrom Cell Signaling Technology (Danvers, MA).

Cell Culture

The HBMEC were a generous gift from Dr K.S. Kim(Johns Hopkins University School of Medicine). HBMECwere cultured in RPMI 1640 medium supplemented with

10% fetal bovine serum (Life Technologies, Grand Island,NY), 10% Nu-serum (BD Biosciences, Franklin Lakes,NJ), 2 mM glutamine, 1 mM sodium pyruvate, 13 non-essential amino acids and 13 MEM vitamin. A stableHBMEC cell line transfected with a dominant-negativePI3K (p110gD948–981) was constructed as described pre-viously (Li et al., 2009).

RNA Interference

The small interfering RNA (siRNA) sequences targetingEphA2, VEGFR2, and nonsilencing control were synthe-sized and purified by GenePharma (Shanghai, China).The siRNA sequences are listed in Table 1. The siRNAsequences for VEGFR1 and ERK1/2 were used asdescribed previously (Li et al., 2013). Transient transfec-tion procedures were performed using Lipofectamine2000reagent according to the manufacturer’s instructions(Invitrogen Corporation, Carlsbad, CA). The transfectedcells were harvested for experiments 48–72 hr aftertransfection.

Real-Time Reverse Transcription (RT)-PCR

Total RNA from cells was extracted with TRIzol Rea-gent (Sigma-Aldrich, St. Louis, MO) and then wasreverse-transcribed using M-MLV reverse transcriptase(Promega, Madison, WI). Real-time PCR was performedon the ABI 7500 real-time PCR system (Applied Biosys-tems, Foster City, CA) using a SYBR green premix ExTaq kit (Takara Biotechnology, Tokyo, Japan), accordingto the manufacturer’s instructions. The relative quantifi-cation was calculated by the 2-DDCT method and wasnormalized against the internal control (GAPDH). Theprimer sequences for human EphA2 were 50-CCAACCAGGCTGTGTTGAAGT-30 and 50-CCCTTGTACACCTCCCCAAA-30. The primer sequences for human GAPDHwere 50-GAAGGTGAAGGTCGGAGTC-30 and 50-GAAGATGGTGATGGGATTTC-30.

Cell Fractionation and Western Blot

Cell fractionation experiments were performed asdescribed previously (Li et al., 2006). The protein con-centrations for each sample were quantified by using aBCA protein assay reagent kit (Pierce, Rockford, IL).Following protein quantification, the fractions were ana-lyzed by Western blot. Equal amounts of samples wereseparated by SDS-PAGE and transferred to PVDF mem-branes (Millipore, Billerica, MA). The membrane wasthen blocked and incubated with primary antibodiesovernight at 4�C. The blots were incubated with anHRP-conjugated secondary antibody (Santa Cruz Bio-technology, Santa Cruz, CA) and visualized by SuperSig-nal West Pico Chemiluminescent Substrate (Pierce,Rockford, IL) using a LAS 3000 mini (Fuji Film, Tokyo,Japan). Quantification of band density was performedusing Image J software.

TABLE 1. The siRNA sequences for RNA interference

Target gene SiRNA sequence

EphA2 50-CCAUCAAGAUGCAGCAGUAUA-30

VEGFR2 50-CCGGGAUAUUUAUAAAGAUTT-30

Nonsilencing control 50-UUCUCCGAACGUGUCACGUTT-30

VEGF INDUCES EphA2 EXPRESSION FOR PERMEABILITY 965

Transendothelial Electrical ResistanceMeasurement

The HBMEC transfected with EphA2 siRNA or nonsilenc-ing control siRNA were seeded onto 0.4-lm pore size Trans-wells (Corning, Lowell, MA) in 24-well plates. Experimentswere conducted when transendothelial electrical resistance(TEER) was>200 Ohm�cm2. The cells were stimulated with100 ng/mLVEGF for 24 hr and TEER was measured using aMillicell-ERS (Milipore, Billerica, MA) according to the man-ufacturer’s protocol. The final TEER values were calculatedas Ohm�cm2 by multiplying the obtained value with the sur-face area of the monolayer. The background was determinedas the TEER of the wells containing only culture mediumand was subtracted from all values.

FITC-Dextran Permeability Assay

The HBMEC cells were prepared according to theTEER protocol as described above. When TEER was>200 Ohm�cm2, the permeability assay was performed.The FITC-dextran (1 mg/mL, average MW 40,000) wasadded to the culture medium in the upper chamber.After 1 hr incubation, the culture medium in the lowerchamber were collected and the amount of FITC-dextranwas measured at an excitation wavelength of 485 nmand an emission wavelength of 535 nm using a fluores-cence plate reader (Infinite M200 Pro, Tecan Group,Mannerdorf). The FITC-dextran flux was expressed asnanogram passed per centimeter surface area per hour.

Immunofluorescence

The HBMEC transfected with nonsilencing control orEphA2 siRNA were seeded on glass coverslips. The conflu-ent cells were stimulated with VEGF (100 ng/mL) for 24hr. The cells were washed in PBS and fixed in 4% parafor-maldehyde for 30 min at room temperature. Then, thecells were permeabilized using 0.1% Triton X-100 for 5min and blocked with 5% BSA in PBS for 30 min. Thecells were incubated with a FITC-conjugated antibodyagainst ZO-1 for 2 hr, and subsequently incubated withDAPI for 5 min to visualize the nuclei. The coverslipswere mounted with 90% glycerol and visualized using afluorescent microscope (Olympus BX51, Tokyo, Japan).

Luciferase Reporter Assay

A plasmid containing the human EphA2 promoter(fragment spanning nucleotides 21362 to 21) was pur-chased from GeneCopoeia (Guangzhou, China). Then, thepromoter was inserted into a pGL3-Basic vector (Promega,Madison, WI). The HBMEC were transiently transfectedwith each indicated reporter plasmid. After 24 hr, the cellswere treated with VEGF in the absence or presence ofAkt inhibitor, PD98059, SU1498, or a neutralization anti-body against VEGFR2. The cells were then harvestedusing a Dual-Luciferase Reporter Assay System (Promega,Madison, WI) according to the manufacturer’s instruc-tions. Firefly and Renilla luciferase activities were meas-ured using Lumat LB 9507 (Berthold Technologies).

Statistical Analysis

Statistical analysis was performed using one-wayANOVA and Student t test. One-way ANOVA was used

to compare multiple groups and Student t test was usedfor comparisons between two groups. All experimentswere repeated at least three times. Data were presentedas means of determinants (mean 6 SD) and P < 0.05were considered statistically significant.

RESULTS

The Role of EphA2 in VEGF-Induced Paracellu-lar Permeability in HBMEC

It is well known that VEGF can increase the perme-ability of the BBB in some neurological disorders. Toascertain whether EphA2 is required for this biologicalprocess, specific siRNAs targeting EphA2 were used toknockdown the expression of EphA2 in HBMEC. Asshown in Figure 1A, EphA2 siRNA effectively sup-pressed the expression of EphA2 in HBMEC comparedto the nonsilencing control siRNA. The EphA2 siRNA-transfected HBMEC were plated as monolayer on theTranswell insert, and the FITC-dextran permeabilityassay and TEER measurements were performed. Theresults showed that knockdown of EphA2 in the brainendothelial cells significantly abolished the VEGF-induced increase in FITC-dextran flux and a reductionin TEER (Fig. 1B,C). Because TJ integrity was mainlyresponsible for the brain endothelial cells permeability(Liu et al., 2012), we next evaluated the alterations inthe distribution of TJ proteins during this process.Immunofluorescence analysis showed that knockdown ofEphA2 inhibited VEGF-induced changes in ZO-1 distri-bution compared to the control cells (Fig. 1D). Thedetergent-insoluble TJ proteins, such as occludin andclaudin-5 have been shown to be an indicator of the TJintegrity. Once TJ is weakened, these proteins were relo-cated to the detergent-soluble fraction in cells (Li et al.,2013). As shown in Figure 1E,F, there was also anobvious shift in the occludin and claudin-5 distributionfrom insoluble to soluble fractions prepared from theHBMEC cells treated with VEGF, which was signifi-cantly inhibited when the levels of EphA2 were knock-down by EphA2 siRNA. These findings suggested thatEphA2 expression is necessary for VEGF-induced perme-ability in HBMEC.

VEGF Upregulates the Expression of EphA2 inHBMEC

EphA2 expression has been detected in human brainendothelium, suggesting that this receptor may have animportant role in regulating the brain vessels (Zhouet al., 2011). EphA2 has been shown to be upregulatedby hypoxia in the lung endothelial cells (Cercone et al.,2009). However, little is known about the regulation ofEphA2 in the brain endothelial cells. Here, we examinedwhether VEGF affects the expression of EphA2. TheHBMEC were treated with VEGF (100 ng/mL) for vari-ous amounts of time (0, 1, 2, 4, 8, 12, 24, 48 hr) or withvarious concentrations of VEGF (0, 5, 10, 50, 100 ng/mL)for 24 hr. We found that VEGF could significantly inducean increase in the EphA2 mRNA levels in a time- anddose-dependent manner (Fig. 2A,C). Likewise, Westernblot analysis showed that stimulation with VEGF couldincrease the levels of EphA2 in HBMEC in a time- anddose-dependent manner (Fig. 2B,D). The highest level ofEphA2 was detected in HBMEC treated with 100 ng/mL

966 MIAO ET AL.

of VEGF for 24 hr. Therefore, this condition was usedfor the subsequent experiments. To further confirm theeffect of VEGF on EphA2 transcriptional regulation, theluciferase reporter assays were performed. Compared tothe control, VEGF induced significantly higher luciferaseactivity in HBMEC transfected with a construct contain-ing the EphA2 promoter (fragment spanning nucleotides21362 to 21) (Fig. 2E). Taken together, these dataclearly indicate that VEGF upregulates EphA2 expres-sion in HBMEC at the transcriptional level.

VEGF-Induced EphA2 Expression is Mediatedby VEGFR2 in HBMEC

VEGFR2 is considered to be the principle mediator ofseveral physiological and pathological effects induced byVEGF in endothelial cells, including proliferation, sur-vival, migration and permeability (Nowacka and Obu-chowicz, 2012). To determine whether VEGFR2 isrequired for VEGF-induced EphA2 expression, severalinhibition experiments were performed. As shown in Fig-

ure 3A, SU1498, a potent and specific inhibitor forVEGFR2, significantly blocked the effect of VEGF onEphA2 receptor expression in a dose-dependent manner.Likewise, a neutralizing antibody against VEGFR2 alsoeffectively abrogated VEGF-induced EphA2 expression(Fig. 3B). Similar results were obtained when VEGFR2expression was knocked down using VEGFR2-specificsiRNA (Fig. 3C). In addition to VEGFR2, VEGF alsobinds to VEGFR1 on the cell surface of endothelial cells.To determine whether VEGFR1 is involved in regulatingEphA2 expression, VEGFR1-specific siRNA and a neu-tralizing antibody against VEGFR1 were used. In con-trast to VEGFR2, inhibition of VEGFR1 did not affectVEGF-induced EphA2 expression (Fig. 3B,D). Thesefindings suggested that VEGFR2, but not VEGFR1, wasinvolved in VEGF-induced EphA2 expression. Addition-ally, the luciferase reporter assays were performed tofurther confirm the above results. As shown in Figure3E, treatment with VEGF led to an increase in the lucif-erase activity in HBMEC transfected with the EphA2promoter (fragment spanning nucleotides 21362 to 21).

Fig. 1. EphA2 contributes to VEGF-induced paracellular permeabilityin HBMEC. (A) Knockdown of EphA2 in HBMEC by EphA2-specificsiRNA. The levels of EphA2 in transfected HBMEC were examined byWestern blot. The nonsilencing control siRNA served as the control. Theresults are presented as the amount of EphA2 normalized to GAPDHlevels. HBMEC with different transfectants were cultured on the upperchamber of a Transwell plate. The amount of FITC-dextran that diffusedto the lower chamber (B) and TEER (C) were measured in the presenceor absence of VEGF, respectively. (D) Following treatment with VEGF,the distribution of ZO-1 in transfected HBMEC was visualized by immu-

nofluorescence. Cell nuclei were stained with DAPI (blue). Arrows indi-cate the discontinuous distribution of ZO-1. Scale bar: 40 lm. (E)Differentially transfected HBMEC were treated with or without VEGF.The soluble and insoluble fractions of occludin proteins were detectedby Western blot. (F) Differentially transfected HBMEC were treated withor without VEGF. The soluble and insoluble fractions of claudin-5 pro-teins were detected by Western blot. Quantification of band densitywas performed using Image J software. Data was shown as the ratio ofsoluble to insoluble fractions. All data were presented as the mean 6 SDfor three independent experiments. *P< 0.05; **P< 0.01.

VEGF INDUCES EphA2 EXPRESSION FOR PERMEABILITY 967

Moreover, a significant reduction in the luciferase activ-ity was observed when SU1498 or VEGFR2 neutralizingantibody was used. Taken together, our results providedevidence to demonstrate that VEGFR2 mediates VEGF-induced EphA2 expression in HBMEC.

VEGF-Induced EphA2 Expression is Mediatedby Activation of PI3K/Akt and ERK1/2 Signalingin HBMEC

Because our results suggested that VEGF could modu-late EphA2 transcription via binding to VEGFR2, wedecided to further investigate the molecular mechanismsof this process. A growing body of evidence indicates thatVEGF binding to the extracellular domain of VEGFR2leads to the activation of different intracellular signalingpathways, such as Rho/ROCK, PI3K/Akt, protein kinaseC, MAPK/ERK1/2, and c-Src kinases. To determine theinvolvement of signaling molecules in VEGF-inducedEphA2 expression, HBMEC cells were pretreated with thespecific inhibitors for the above signaling molecule andthe levels of EphA2 were detected by real-time PCR andWestern blot. The results showed that PI3K inhibitors(LY294002 and wortmannin), Akt inhibitor and ERK1/2inhibitors (PD98059 and SP600125) significantly blockedVEGF-induced EphA2 expression (Fig. 4A–C), and other

inhibitors had no effects in this process. These resultssuggested that activation of PI3K/Akt and ERK1/2 signal-ing pathways is required for VEGF-induced EphA2expression. To further confirm the above results, a stableHBMEC cell line transfected with dominant-negativePI3K (Dp110) was used to analyze the effect of VEGF onEphA2 expression. The results showed that overexpres-sion of dominant-negative PI3K in HBMEC significantlyblocked VEGF-induced EphA2 expression (Fig. 4D). Tofurther confirm the role of ERK1/2 in this process, the lev-els of ERK1/2 were knocked down in HBMEC by siRNA.Our results showed that depletion of ERK1/2 expressionin HBMEC also significantly blocked VEGF-inducedEphA2 expression (Fig. 4E).

Based on these results, we next asked whether therewas a crosstalk between the PI3K/Akt and ERK1/2 sig-naling pathways. To address this question, Akt inhibitorand PD98059 were used together in an inhibition experi-ment. The results showed that simultaneous inhibitionof PI3K/Akt and ERK1/2 pathways could significantlyabrogate the increase of VEGF-induced EphA2 expres-sion in HBMEC (Fig. 5A). In addition, we also examinedthe activation status of PI3K/Akt and ERK1/2 signalingin HBMEC using specific antibodies against phosphoryl-ated Akt and ERK1/2 following VEGF stimulation. West-ern blot analysis revealed that phosphorylated Akt and

Fig. 2. VEGF induces the upregulation of EphA2 expression in HBMEC.HBMEC were treated with VEGF (100 ng/mL) for the indicated time inter-vals (A, B) or with the indicated concentrations of VEGF for 24 hr (C, D).Following treatment, the levels of EphA2 were analyzed by real-time PCRand western blot. EphA2 expression was quantified and normalized to

GAPDH. (E) HBMEC were transfected with empty vector (pGL3-Basic) orEphA2 promoter reporter (fragment spanning nucleotides 21362 to 21)and treated with vehicle or VEGF (100 ng/mL) for 24 hr. The ratio of firelyto Renilla luciferase activity was normalized to control. All data were pre-sented as the mean6 SD for three independent experiments. *P< 0.05.

968 MIAO ET AL.

ERK1/2 increased in a time-dependent manner. Moreinterestingly, both Akt and ERK1/2 activation wereobserved in HBMEC after 15 min of VEGF treatment(Fig. 5B). Thus, we hypothesized that PI3K/Akt andERK1/2 signaling might independently contribute to theVEGF-induced EphA2 expression. To address this ques-tion, further inhibition experiments were performed.Phosphorylation of ERK1/2 was detected in VEGF-treated HBMEC following the inhibition of PI3K/Akt byLY294002, wortmannin, or Akt inhibitor or throughtransfection with PI3K mutants (Dp110). The resultsshowed that inhibition of PI3K/Akt had no effect onVEGF-induced ERK1/2 activation in HBMEC (Fig.5C,D). Likewise, inhibition of ERK1/2 activity by

PD98059 or transfection with ERK1/2 siRNA also had anegligible impact on PI3K/Akt activation in HBMECstimulated with VEGF (Fig. 5E,F).

Taken together, these data strongly suggest thatVEGF-induced EphA2 expression is mediated by the acti-vation of PI3K/Akt and ERK1/2 pathways in HBMEC.

DISCUSSION

The Eph receptors and their ephrin ligands have beenshown to play several key roles in regulating vascularfunction, such as endothelial cells migration and adhe-sion (Pasquale, 2008). Recent evidence suggests thatEphA2 receptor has diverse biological functions in

Fig. 3. VEGFR2 mediates VEGF-induced EphA2 expression inHBMEC. (A) HBMEC were pretreated with the indicated concentra-tions of SU1498 for 30 min, followed by VEGF treatment. The expres-sion of EphA2 in HBMEC was examined by Western blot. (B) HBMECwere pretreated with neutralizing antibodies against VEGFR1 andVEGFR2 for 30 min followed by treatment with VEGF. The expressionof EphA2 in HBMEC was detected by Western blot. (C) Following 48hr of transfection with nonsilencing control or VEGFR2 siRNA, HBMECwere treated with or without VEGF. The expression of EphA2 inHBMEC was detected by Western blot. (D) HBMEC transfected with

nonsilencing control or VEGFR1 siRNA were treated with or withoutVEGF. The expression of EphA2 in HBMEC was detected by Westernblot. The above results were quantified and the expression of EphA2was normalized to GAPDH. (E) HBMEC transfected with an EphA2promoter reporter (fragment spanning nucleotides 21362 to 21) werepretreated with SU1498 (10 lM) or a VEGFR2 neutralizing antibody for30 min, followed by VEGF treatment. The ratio of firely to Renilla lucif-erase activity was determined and normalized to the control. All datawere presented as the mean 6 SD for three independent experiments.*P< 0.05; **P< 0.01.

VEGF INDUCES EphA2 EXPRESSION FOR PERMEABILITY 969

addition to regulating angiogenesis. In the presentstudy, we provide evidence to demonstrate that EphA2 isupregulated by VEGF in HBMEC, which contributed tothe observed increase in paracellular permeability.

It has been reported that EphA2 expression isincreased following lung injury, which contributes tovascular leak in the injured lung (Larson et al., 2008).EphA2 can phosphorylate the cytoplasmic tail ofclaudin-4 and enhance the paracellular permeability ofepithelial cells (Tanaka et al., 2005). We previouslyreported that EphA2 was involved in the modulation ofparacellular permeability in the brain endothelial cells.Here, we demonstrated that EphA2 was required forVEGF-induced changes in permeability in brain endo-thelial cells. Several lines of evidence indicate a func-tional link between EphA2 and VEGF in angiogenesis.Inhibition of EphA2 activity using an EphA2-Fc fusion

protein might suppress VEGF-mediated angiogenesis(Chen et al., 2006). In contrast, the interactions betweenephrinA1 and EphA2 might be necessary for maximalinduction of neovascularization by VEGF (Brantley-Sieders et al., 2006). Contrary to the above studies, ourresults showed that there is a functional link betweenthese two factors in the regulation of brain endothelialpermeability. VEGF is a key regulator of EphA2 expres-sion and can induce an increase in EphA2 transcription,which is required for the VEGF-induced increase in par-acellular permeability. Combined with prior findings, wehypothesized that exogenous VEGF may induce EphA2expression in endothelial cells to promote paracellularpermeability in pathological conditions.

Next, we addressed the question of how VEGF upre-gulates the expression of EphA2. In endothelial cells,VEGF exerted its vasopermeability effects primarily

Fig. 4. VEGF-induced EphA2 expression was mediated by the acti-vation of PI3K/Akt and ERK1/2 signaling pathways in HBMEC. (A)HBMEC were pretreated with LY294002 (25 lM), wortmannin (1 lM),Akt inhibitor (10 lM), PD98059 (10 lM), or SP600125 (200 nM) for 1hr, and then treated with VEGF. The levels of EphA2 in HBMEC wereexamined by Western blot. (B) HBMEC were incubated for 1 hr in theabsence (DMSO) or presence of wortmannin (1 lM), Akt inhibitor (10lM), or PD98059 (10 lM), and were subsequently treated with VEGF.EphA2 mRNA levels were determined by real-time PCR. (C) HBMECwere transfected with an EphA2 promoter reporter (fragment spanningnucleotides 21362 to 21) and were pretreated with an Akt inhibitor

(10 lM) or PD98059 (10 lM), before being treated with VEGF. The ratioof firely to Renilla luciferase activity was determined and normalized tothe control. (D) HBMEC were transfected with a dominant-negativePI3K (Dp110) or empty vector and then treated with or without VEGF.The expression of EphA2 in HBMEC was measured by Western blot.(E) HBMEC transfected with nonsilencing control or ERK siRNA weretreated with or without VEGF. The expression of EphA2 in HBMECwas detected by Western blot. All results were quantified and pre-sented as the amount of EphA2 normalized to GAPDH. All data werepresented as the mean 6 SD for three independent experiments.*P< 0.05; **P< 0.01.

970 MIAO ET AL.

through VEGFR2, the principal receptor for VEGF. Thus,we speculated that VEGFR2 might be involved in VEGF-induced EphA2 expression. In contrast to VEGFR1, theVEGF-dependent increase in EphA2 expression was sup-pressed when VEGFR2 was blocked in HBMEC, whichdemonstrated that VEGFR2 specifically mediated VEGF-induced EphA2 expression. Therefore, we were intriguedto find the intracellular pathways elicited by VEGF, whichupregulates EphA2 expression in HBMEC. The resultsfrom the inhibitor experiments determined that bothPI3K/Akt and ERK1/2 pathways were required for thisprocess. Further results proved that the activation of

PI3K/Akt and ERK1/2 pathways independently contrib-uted to the VEGF-induced EphA2 expression. The PI3K/Akt and MEK/ERK signaling molecules have been impli-cated in the VEGF-induced permeability response in otherstudies (Aramoto et al., 2004). Therefore, we hypothesizedthat VEGF upregulated the expression of EphA2 in thebrain endothelial cells via the activation of PI3K/Akt andERK1/2 signaling pathways, which resulted in increasedparacellular permeability.

In summary, we demonstrated that VEGF binds toVEGFR2 and separately activates the PI3K/Akt andERK1/2 signaling pathways, resulting in the upregulation

Fig. 5. The role of PI3K/Akt and ERK1/2 pathways in VEGF-inducedEphA2 expression in HBMEC. (A) HBMEC were incubated with Aktinhibitor, PD98059 or both Akt inhibitor and PD98059 for 1 hr andwere then treated with VEGF. The expression of EphA2 in HBMECwas examined by Western blot. The levels of EphA2 were quantifiedand normalized to GAPDH. (B) HBMEC were treated with VEGF forthe indicated times and the phosphorylation of Akt and ERK1/2 inHBMEC were detected by Western blot. (C) HBMEC were pretreatedwith the indicated inhibitors for 1 hr and then treated with VEGF. Phos-phorylation of ERK1/2 in HBMEC was detected by Western blot. Theresults were quantified and presented as the amount of ERK1/2 phos-phorylation normalized to total ERK1/2. (D) HBMEC transfected with

dominant-negative PI3K (Dp110) or empty vector were treated with orwithout VEGF. Phosphorylation of ERK1/2 in HBMEC was detectedusing Western blot. (E) HBMEC were pretreated with PD98059 for 1 hrand then treated with VEGF. Phosphorylation of Akt in HBMEC wasdetected by Western blot. The levels of Akt phosphorylation werequantified and normalized to total Akt. (F) HBMEC transfected withnonsilencing control or ERK siRNA were treated with or without VEGF.The phosphorylation of Akt in HBMEC was measured by Western blot.The levels of GAPDH were used as an internal control for all experi-ments. All data were presented as the mean 6 SD for three independ-ent experiments. *P< 0.05; **P< 0.01.

VEGF INDUCES EphA2 EXPRESSION FOR PERMEABILITY 971

of EphA2 expression in the brain endothelial cells. Thus,our findings prove the novel role of VEGF in regulatingEphA2 expression, which functionally links VEGF andEphA2 in the regulation of paracellular permeability inthe brain endothelial cells. Further investigations areneeded to understand the mechanism by which VEGFinduces EphA2 transcriptional activity and to identify themolecules that interact with the EphA2 promoter.

ACKNOWLEDGMENTS

The authors sincerely thank Dr Monique Stins andKwang Sik Kim (Department of Pediatrics, John Hop-kins University School of Medicine) for providingHBMEC as a gift. We gratefully acknowledge Jian Gaofor contributing to the immunofluorescence.

LITERATURE CITED

Aramoto H, Breslin JW, Pappas PJ, Hobson RW 2nd, Duran WN.2004. Vascular endothelial growth factor stimulates differentialsignaling pathways in in vivo microcirculation. Am J PhysiolHeart Circ Physiol 287:H1590–H1598.

Argaw AT, Gurfein BT, Zhang Y, Zameer A, John GR. 2009. VEGF-mediated disruption of endothelial CLN-5 promotes blood-brainbarrier breakdown. Proc Natl Acad Sci USA 106:1977–1982.

Brantley-Sieders DM, Caughron J, Hicks D, Pozzi A, Ruiz JC, ChenJ. 2004. EphA2 receptor tyrosine kinase regulates endothelial cellmigration and vascular assembly through phosphoinositide 3-kinase-mediated Rac1 GTPase activation. J Cell Sci 117:2037–2049.

Brantley-Sieders DM, Fang WB, Hwang Y, Hicks D, Chen J. 2006.Ephrin-A1 facilitates mammary tumor metastasis through anangiogenesis-dependent mechanism mediated by EphA receptorand vascular endothelial growth factor in mice. Cancer Res 66:10315–10324.

Cercone MA, Schroeder W, Schomberg S, Carpenter TC. 2009.EphA2 receptor mediates increased vascular permeability in lunginjury due to viral infection and hypoxia. Am J Physiol Lung CellMol Physiol 297:L856–L863.

Chen J, Hicks D, Brantley-Sieders D, Cheng N, McCollum GW, Qi-Werdich X, Penn J. 2006. Inhibition of retinal neovascularizationby soluble EphA2 receptor. Exp Eye Res 82:664–673.

Chu H, Tang Y, Dong Q. 2013. Protection of vascular endothelialgrowth factor to brain edema following intracerebral hemorrhageand its involved mechanisms: effect of Aquaporin-4. PLoS One 8:e66051.

Larson J, Schomberg S, Schroeder W, Carpenter TC. 2008. Endothe-lial EphA receptor stimulation increases lung vascular permeabil-ity. Am J Physiol Lung Cell Mol Physiol 295:L431–L439.

Li B, Wang C, Zhang Y, Zhao XY, Huang B, Wu PF, Li Q, Li H, LiuYS, Cao LY, Dai WM, Fang WG, Shang DS, Cao L, Zhao WD,Chen YH. 2013. Elevated PLGF contributes to small-cell lungcancer brain metastasis. Oncogene 32:2952–2962.

Li B, Zhao WD, Tan ZM, Fang WG, Zhu L, Chen YH. 2006. Involve-ment of Rho/ROCK signalling in small cell lung cancer migrationthrough human brain microvascular endothelial cells. FEBS Lett580:4252–4260.

Li M, Shang DS, Zhao WD, Tian L, Li B, Fang WG, Zhu L, ManSM, Chen YH. 2009. Amyloid beta interaction with receptor foradvanced glycation end products up-regulates brain endothelialCCR5 expression and promotes T cells crossing the blood-brainbarrier. J Immunol 182:5778–5788.

Liu WY, Wang ZB, Zhang LC, Wei X, Li L. 2012. Tight junction inblood-brain barrier: an overview of structure, regulation, and reg-ulator substances. CNS Neurosci Ther 18:609–615.

Ma Y, Zechariah A, Qu Y, Hermann DM. 2012. Effects of vascularendothelial growth factor in ischemic stroke. J Neurosci Res 90:1873–1882.

Morin-Brureau M, Lebrun A, Rousset MC, Fagni L, Bockaert J, deBock F, Lerner-Natoli M. 2011. Epileptiform activity induces vas-cular remodeling and zonula occludens 1 downregulation in orga-notypic hippocampal cultures: role of VEGF signaling pathways.J Neurosci 31:10677–10688.

Nowacka MM, Obuchowicz E. 2012. Vascular endothelial growthfactor (VEGF) and its role in the central nervous system: a newelement in the neurotrophic hypothesis of antidepressant drugaction. Neuropeptides 46:1–10.

Ojima T, Takagi H, Suzuma K, Oh H, Suzuma I, Ohashi H,Watanabe D, Suganami E, Murakami T, Kurimoto M, Honda Y,Yoshimura N. 2006. EphrinA1 inhibits vascular endothelialgrowth factor-induced intracellular signaling and suppresses reti-nal neovascularization and blood-retinal barrier breakdown. Am JPathol 168:331–339.

Pasquale EB. 2008. Eph-ephrin bidirectional signaling in physiologyand disease, Cell 133:38–52.

Tanaka M, Kamata R, Sakai R. 2005. EphA2 phosphorylates thecytoplasmic tail of Claudin-4 and mediates paracellular perme-ability. J Biol Chem 280:42375–42382.

Zhou N, Zhao WD, Liu DX, Liang Y, Fang WG, Li B, Chen YH.2011. Inactivation of EphA2 promotes tight junction formationand impairs angiogenesis in brain endothelial cells. MicrovascRes 82:113–121.

Zlokovic BV. 2008. The blood-brain barrier in health and chronicneurodegenerative disorders. Neuron 57:178–201.

972 MIAO ET AL.