The phosphatidylinositol-3 kinase/Akt pathway ... - The FASEB Journal

The FASEB Journal • FJ Express Summary

The phosphatidylinositol-3 kinase/Akt pathway mediates VEGF’s neuroprotective activity and induces blood brain barrier permeability after focal cerebral ischemia ¨ lkan Kilic,* Yaoming Wang,*,2 Claudio L. Bassetti,* Hugo H. Marti,† Ertugrul Kilic,*,1 U and Dirk M. Hermann* *Department of Neurology, University Hospital Zurich (USZ), Switzerland; and †Institute of Physiology and Pathophysiology, University of Heidelberg, Germany To read the full text of this article, go to http://www.fasebj.org/cgi/doi/10.1096/fj.05-4829fje SPECIFIC AIMS Based on its trophic influence on neurons and vascular cells, vascular endothelial growth factor (VEGF) is a promising candidate for stroke treatment. VEGF’s survival-promoting effects are purchased at the expense of an increased blood brain barrier (BBB) permeability, which potentially compromises tissue survival. The mechanisms via which VEGF protects the brain against ischemia and induces BBB permeability remained unknown. This study aimed to characterize signal transduction pathways responsible for VEGF’s neuroprotection and BBB disturbances in our transgenic mouse line V1 that expresses human VEGF165 in the brain under control of a neuron-specific enolase (NSE) promoter, which we subjected to 90 min of intraluminal middle cerebral artery (MCA) occlusion. PRINCIPAL FINDINGS 1. VEGF receptor-2 (VEGFR-2) is expressed on ischemic neurons and astrocytes and is activated by human VEGF To find out whether VEGFR-2 is expressed and activated in the ischemic brain, we combined immunohistochemical and immunoprecipitation studies. We show that VEGFR-2 is expressed on ischemic neurons and astrocytes at 24 h after reperfusion and that this receptor is phosphorylated (i.e., activated) by VEGF. 2. VEGF increases phosphorylated Akt and extracellular–regulated kinase (ERK)-1/-2 and reduces phosphorylated p38 and c-Jun NH2-terminal kinase (JNK)-1/-2 To elucidate how VEGF influences cell signaling, we prepared Western blots with tissue samples from ischemic brains. These blots revealed that phosphorylated 0892-6638/06/0020-1185 © FASEB

(but not total, nonphosphorylated) Akt and ERK-1/-2 levels were increased, whereas phosphorylated p38 and JNK-1/-2 levels were reduced by VEGF (Fig. 1). 3. VEGF down-regulates inducible NO synthase [inducible NOS (iNOS)] in the ischemic brain To evaluate whether VEGF influences the expression of iNOS, which contributes to free radical formation in the reperfused brain, we performed immunohistochemistries. Robust expression of iNOS was noticed in ischemic tissue of wild-type (WT) mice (24.7⫾6.7 cells/ 62⬘500 ␮m2 square; mainly NeuN⫹ neurons) that was decreased by VEGF (10.3⫾4.5 cells/ square, P⬍0.05). 4. The phosphatidyl inositol-3 kinase (PI3K)/Akt inhibitor Wortmannin reverses VEGF’s neuroprotective activity and restores BBB integrity To clarify the role of the PI3K/Akt pathway in VEGF’s effects, we applied the solvent dimethyl sulfoxide (DMSO) or DMSO containing 0.1 mM of the PI3K/Akt inhibitor Wortmannin into the intracerebroventricular (i.c.v.) space of WT and V1tg mice subjected to focal cerebral ischemia, and compared these animals with animals not receiving i.c.v. injections. Whereas cerebral laser Doppler flow (LDF) did not differ between groups (Fig. 2A), VEGF reduced infarct volume (Fig. 2B), improved neurological abnormalities (Fig. 2C), and increased BBB permeability (Fig. 2E) without influencing macroscopic brain swelling (Fig. 2D). Inhibition of PI3K/Akt completely abolished VEGF’s neuroprotec1

Correspondence: Department of Neurology, University Hospital Zurich, Frauenklinikstr. 26, Zurich CH-8091, Switzerland. E-mail: [email protected] 2 Present address: Buck Institute for Age Research, Novato, California, USA. doi: 10.1096/fj.05-4829fje 1185

Figure 1. Expression and phosphorylation state of Akt, ERK-1/-2, MAP kinase/p38, and JNK-1/-2 in WT and V1tg mice subjected to focal cerebral ischemia. Animals were either untreated or i.c.v. treated with 2 ␮l of the solvent DMSO or DMSO containing Wortmannin, a PI3K/Akt inhibitor. Tissue samples were taken from the ischemic cortex and underlying striatum. Note that human VEGF increases phosphorylated (but not total) Akt and ERK-1/-2 and reduces phosphorylated p38 and JNK-1/-2 levels. Furthermore, note the reversal of p38, but not of ERK-1/-2 or JNK-1/-2 phosphorylation in animals treated with Wortmannin, in which Akt phosphorylation was suppressed. Our data suggested that Akt was involved in VEGF’s neuroprotective function and BBB permeability. Values are mean ⫾ sd (n⫽3 samples/group), normalized with corresponding blots for ␤-actin. *P ⬍ 0.05 compared with WT mice; #P ⬍ 0.05 compared with V1tg mice receiving DMSO.

tion (Fig. 2B, C), at the same time reversing BBB leakage to below levels in wt mice (Fig. 2D, E). 5. Reversal of VEGF neuroprotection by Wortmannin is associated with restoration of p38, but not ERK-1/ -2 and JNK-1/-2 To analyze downstream effects of PI3K/Akt, we prepared additional Western blots for DMSO and DMSO/ Wortmannin treated mice. Our data confirmed that Wortmannin indeed inhibited Akt phosphorylation (Fig. 1). In contrast to phosphorylated ERK-1/-2 and JNK-1/-2, which were not influenced by the PI3K/Akt inhibitor, p38 phosphorylation was increased in Wortmannin treated animals that exhibited exacerbated brain infarcts (Fig. 1). Our results indicate that the regulation of p38 by VEGF closely depends on the PI3K/Akt activation state. 6. PI3K/Akt inhibition does not reverse iNOS levels in ischemic neurons To find out whether iNOS inhibition by VEGF occurs in a PI3K/Akt-dependent manner, we also examined 1186

Vol. 20

June 2006

DMSO and DMSO/Wortmannin treated animals by immunohistochemistry. In contrast to MAP kinase/ p38, which was activated when PI3K/Akt was inhibited, neuronal iNOS levels were not reversed by Wortmannin (9.9⫾3.0 vs. 8.9⫾4.4 cells/square in DMSO and DMSO/Wortmannin-treated mice, respectively), demonstrating that iNOS inhibition by VEGF does not depend on PI3K/Akt.

CONCLUSIONS Using our transgenic mouse line V1 that expresses human VEGF in the brain, we demonstrate that VEGF protects against focal cerebral ischemia and also induces BBB permeability via the PI3K/Akt pathway. Our data were obtained using: 1) in vivo experiments, in which brain injury was evaluated following 90 min of intraluminal MCA occlusions (24 h reperfusion); 2) immunohistochemical and immunoprecipitation studies using brain tissue samples, in which we examined the expression and phosphorylation (i.e., activation) of VEGF-receptor-2, Akt, MAP kinases (ERK-1/-2, p38,

The FASEB Journal

KILIC ET AL.

Figure 3. VEGF protects against ischemia and induces BBB leakage by activating the PI3K/Akt pathway. In ischemic brains of WT mice, both Akt and ERK-1/-2 activities are low, whereas p38 and JNK-1/-2 activities are high. As a consequence, ischemic neurons die and BBB remains functional. In mice constitutively expressing human VEGF (V1tg), on the other hand, Akt and ERK-1/-2 pathways are activated, whereas p38 and JNK-1/-2 pathways are inhibited. Thereby, ischemic neurons remain viable, at the expense of an enhanced BBB permeability. Inhibition of the PI3K/Akt pathway with Wortmannin completely abolishes the tissue protection induced by VEGF, at the same time reversing BBB function.

JNK-1/-2), and iNOS, as well as 3) additional in vivo experiments, in which we applied the pharmacological PI3K/Akt inhibitor Wortmannin into the i.c.v. space and evaluated effects of VEGF on ischemic injury and BBB permeability. By showing that VEGF: 1) activates VEGFR-2, which as we demonstrate is expressed on ischemic neurons and astrocytes; 2) stimulates Akt and ERK-1/-2; 3) inhibits p38 and JNK-1/-2; and 4) downregulates iNOS, we identify a complex cellular signaling scenario, which VEGF activates in order to exert its neuroprotective function. Based on our finding that the PI3K/Akt pathway is responsible for VEGF’s neuroprotection and its BBB permeability, we predict that it may not easily be possible to make use of VEGF’s tissue survival without accepting its unfavorable consequence, the increased BBB leakage.

Figure 2. The PI3K/Akt pathway mediates VEGF’s neuroprotective function and BBB permeability. Data from wt and V1tg mice that were either untreated or i.c.v. treated with DMSO or DMSO/Wortmannin. Note that LDF during ischemia does not differ between groups (A). Assessments of infarct vol. (B) and neurological deficits (C) reveal that human VEGF significantly reduces brain injury and ameliorates postischemic recovery. Brain swelling (D) is not influenced by human VEGF, whereas IgG extravasation (E) is moderately increased, indicating that VEGF expression promotes BBB leakage. Note that the neuroprotective effects of VEGF are reversed by

PI3K/Akt inhibition (B, C), while brain swelling (D) and BBB permeability (E) are reduced, despite the exacerbation of brain infarcts. These data demonstrate that the PI3K/Akt pathway is responsible for VEGF’s neuroprotective function and vascular leakage. Values are means ⫾ sd (n⫽5–11 animals/ group). *P ⬍ 0.05 compared with WT mice; # P ⬍ 0.05 compared with V1tg mice receiving DMSO.

PI3K/AKT MEDIATES VEGF NEUROPROTECTION AND INDUCES BBB PERMEABILITY

1187

The FASEB Journal • FJ Express Full-Length Article

The phosphatidylinositol-3 kinase/Akt pathway mediates VEGF’s neuroprotective activity and induces blood brain barrier permeability after focal cerebral ischemia ¨ lkan Kilic,* Yaoming Wang,*,2 Claudio L. Bassetti,* Hugo H. Marti,† Ertugrul Kilic,*,1 U and Dirk M. Hermann* *Department of Neurology, University Hospital Zurich (USZ), Zurich, Switzerland; and †Institute of Physiology and Pathophysiology, University of Heidelberg, Heidelberg, Germany Based on its trophic influence on neurons and vascular cells, vascular endothelial growth factor (VEGF) is a promising candidate for stroke treatment. VEGF’s survival-promoting effects are purchased at the expense of an increased blood brain barrier permeability, which potentially compromises tissue survival. The mechanisms via which VEGF protects the brain against ischemia remained unknown. We examined signaling pathways underlying VEGF’s neuroprotective activity in our transgenic mouse line, which expresses human VEGF165 under a neuron-specific enolase (NSE) promoter. We show that VEGF receptor-2 (Flk-1) is expressed on ischemic neurons and astrocytes and is activated by VEGF. Following 90-min episodes of middle cerebral artery occlusion, VEGF increased phosphorylated (but not total) Akt and ERK1/-2 and reduced phosphorylated mitogen activated protein kinase/p38 and c-Jun NH2-terminal kinase (JNK)-1/-2 levels, at the same time decreasing inducible NO synthase expression in ischemic neurons. Inhibition of Akt with Wortmannin reversed VEGF’s neuroprotective properties, diminished brain swelling, and restored the vascular permeability induced by VEGF to below levels in WT animals. The aggravation of brain injury by Wortmannin was associated with the restitution of p38, but not of JNK-1/-2, ERK-1/-2, or inducible NOS (iNOS). Our data demonstrate that VEGF mediates both neuroprotection and blood brain barrier permeability via the phosphatidylinositol-3 kinase (PI3K)/Akt pathway. Based on our observation that VEGF neuroprotection and vascular leakage depend on PI3K/Akt, which is putatively regulated by VEGF receptor-2, we predict that it may not easily be possible to make use of VEGF’s neuroprotective function without accepting its unfavorable consequence, the increased ¨ ., Wang, Y., vascular permeability.—Kilic, E., Kilic, U Bassetti, C. L., Marti, H. H., Hermann, D. M. The phosphatidylinositol-3 kinase/Akt pathway mediates VEGF’s neuroprotective activity and induces blood brain barrier permeability after focal cerebral ischemia. FASEB J. 20, E307–E314 (2006)

ABSTRACT

0892-6638/06/0020-0307 © FASEB

Key Words: vascular endothelial growth factor 䡠 neuroprotection 䡠 vascular permeability 䡠 mitogen activated protein kinase

VEGF is a homodimeric growth factor that is constitutively expressed in the brain (1, 2). VEGF is upregulated during hypoxia and ischemia as part of an adaptive response aiming at protecting the tissue from injury (3– 4). In vitro studies have demonstrated that VEGF protects neurons from death under neurodegenerative conditions (5–7). VEGF’s neuroprotective activity was recently shown in vivo after stroke, where VEGF reduced infarct size, ameliorated neurological deficits and prevented delayed neuronal injury by inhibiting caspase-3, when it was supplied locally by intracerebroventricular (i.c.v.) delivery (8) or constitutively expressed in transgenic mice (9). In view of its trophic actions on neurons and vascular cells, VEGF is a promising candidate for stroke treatment. Yet, VEGF’s survival-promoting effects are purchased at the expense of an increased blood brain barrier (BBB) permeability (10), which potentially compromises tissue survival. Indeed, rodent studies have shown that systemic delivery of VEGF in the early postischemic phase may exacerbate vasogenic edema thus increasing brain damage, while VEGF antagonisation with the high-affinity VEGF-binding protein mFlt(1–3)-IgG reduced brain swelling and infarct size (11, 12). In face of the detrimental consequences of malignant edema, making use of VEGF’s protective effects but avoiding its vascular permeability would be desirable, when VEGF is therapeutically used. The mechanisms via which VEGF protects the brain against ischemia and induces BBB permeability remained largely unknown. To elucidate effects of VEGF 1

Correspondence: Department of Neurology, University Hospital Zurich, Frauenklinikstr. 26, CH-8091 Zurich, Switzerland. E-mail: [email protected] 2 Present address: Buck Institute for Age Research, Novato, California, USA. doi: 10.1096/fj.05-4829fje E307

in the brain, we have recently established transgenic mice expressing human VEGF165 in the central nervous system (CNS) under control of a neuron-specific enolase (NSE) promoter (so-called V1tg mouse line; 13). When submitted to intraluminal middle cerebral artery (MCA) occlusion, V1tg animals exhibited smaller brain infarcts than wild-type (WT) mice, despite an increased vascular permeability (9). To characterize the signaling mechanisms responsible for VEGF’s neuroprotective properties, we now studied the expression and phosphorylation (i.e., activation) of VEGF receptor-2 (VEGFR-2, Flk-1), the phosphatidylinositol-3 kinase (PI3K)-dependent factor Akt and of mitogen activated protein (MAP) kinases (extracellular-regulated kinase (ERK)-1/-2, p38 and Jun kinase (JNK)-1/-2) and analyzed inducible NO synthase (iNOS) expression in WT and V1tg mice submitted to focal cerebral ischemia. We then modulated the activity of one of the signal transduction factors, Akt, by delivering a pharmacological inhibitor, Wortmannin, investigating the role of the PI3K/ Akt pathway in VEGF’s neuroprotective function and BBB permeability.

MATERIALS AND METHODS Animals All animal experiments were carried out with governmental approval according to National Institutes of Health guidelines for care and use of laboratory animals. The generation and characterization of the NSE-VEGF transgenic mouse line V1, which were backcrossed on a C57Bl/6 background, have been previously described in detail (9).

Induction of intraluminal MCA occlusions Adult V1tg mice and their nontransgenic littermates (21–28 g) were anesthetized with 1% halothane (30% O2, remainder N2O) (n⫽11 animals/ group). Rectal temperature was maintained between 36.5 and 37.0°C by using a feedback-controlled heating system. Focal ischemia was induced using an intraluminal technique (9, 14) using a 8 – 0 silicon coated (Xantopren: Bayer Dental, Osaka, Japan) nylon monofilament (Ethilon; Ethicon, Norderstedt, Germany). During the experiments, laser Doppler flow (LDF) was monitored using a flexible 0.5 mm fiberoptic probe (Perimed, Stockholm, Sweden) attached to the intact skull overlying the MCA territory. LDF changes were measured during 90 min of MCA occlusion and up to 30 min after reperfusion onset. At that time, wounds were carefully sutured and anesthesia was discontinued. After 24 h of reperfusion, neurological deficits were evaluated using a five-point neurological deficit score ranging from 0 ⫽ normal function to 4 ⫽ absence of spontaneous motor activity (15, 9). Animals were then reanesthetized with halothane and sacrificed. Brains were removed, frozen on dry ice, and cut on a cryostat into coronal 18 ␮m sections that were used for cresyl violet stainings and for immunohistochemistry for extravasated IgG, VEGFR-2, and iNOS. In addition, tissue samples were taken from the ischemic brain (striatum and overlying cortex) at the level of the bregma for immunoprecipitation studies and Western blots. Histological injury was evaluated on cresyl violet sections taken from five equidistant brain levels, 2 mm apart, on which infarct borders between infarcted and noninfarcted E308

Vol. 20

June 2006

tissue were outlined using an image analysis system (Image J). Thus, infarct volume and brain swelling were determined.

Analysis of serum IgG extravasation With gentle stirring, brain sections were rinsed for 10 min at room temperature in 0.1 M phosphate buffered saline (PBS), in order to remove intravascular IgG, and were fixed in 4% paraformaldehyde (16). Following blocking of endogenous peroxidase with methanol/ 0.3% H2O2 and immersion in 0.1 M PBS containing 5% bovine serum albumin (BSA) and normal swine serum (1:1000), sections were incubated for 1 h in biotinylated anti-mouse IgG (Santa Cruz, Nunningen, Switzerland) and stained with an avidin peroxidase kit (Vectastain Elite; Vector Labs, Burlingame, CA) and diaminobenzidine (Sigma, Deisenhofen, Germany). For reasons of data comparability, all sections were processed in parallel. Sections were scanned and densitometrically analyzed (n⫽5– 6 animals/group). IgG extravasation was analyzed by subtracting optical densities in the contralateral nonischemic from that in the ischemic cortex and underlying striatum.

Immunohistochemistry of VEGFR-2 and inducible NO synthase (iNOS) Brain sections were fixed in ice-cold acetone (VEGFR-2) or 4% paraformaldehyde/0.1 M PBS (iNOS), washed, and immersed for 1 h in 0.1 M PBS containing 0.3% Triton X-100 (PBS-T)/10% normal goat serum. Sections were incubated overnight at 4°C with monoclonal rat antimouse VEGFR-2 (Flk-1; 9) or polyclonal rabbit anti-inducible NOS (NOS-2, sc-650; Santa Cruz) antibody (Ab), diluted 1:100 in PBS-T. Counterstainings were performed with a mouse Ab against the neuronal nuclear protein NeuN (MAB377, Chemicon, Lucerne, Switzerland; 1:500) or with a goat Ab against the astrocytic marker glial acidic fibrillary protein (GFAP) (Molecular Probes, Basle, Switzerland; 1:250). Sections were finally incubated with 4⬘,6⬘-diamidino-2-phenylidole (DAPI) and cover-slipped. Sections stained for VEGFR-2 were evaluated by counting the density of VEGFR-2(⫹) parenchymal (i.e., nonvascular) cells in rectangular fields of the cerebral (sensory) cortex close to the infarct border and in the nonischemic (cingulate) cortex located outside the MCA territory (area evaluated: 62⬘500 ␮m2). These VEGFR-2(⫹)cells could clearly be distinguished based on their size, which was bigger than that of newborn capillary endothelial cells, which are also VEGFR2(⫹). Sections stained for iNOS were analyzed by counting iNOS(⫹) cells in a total of four regions within the parietal cortex (62⬘500 ␮m2), for which mean values were calculated. The regions chosen for this assessment were located within the evolving infarct, thus ruling out that differences in cell densities were a consequence of the severity of brain injury that differed between groups.

Immunoprecipitation analysis of VEGFR-2 Frozen tissue samples were homogenized in lysis buffer (10 mM Tris-HCl, pH 7.6; 150 mM NaCl; 1 mM EDTA; 1 mM EGTA, pH 8.0; 0.2 mM PMSF; 0.2 mM sodium vanadate; 0.5% Nonidet P-40; 1% triton X-100; 1 ␮g/ml aprotinin). Homogenates were centrifuged for 15 min at ⫹ 4°C (14,000 rpm) and supernatants subsequently collected. Supernatants (200 ␮l) were incubated with protein A-sepharose CL-4B beads in binding buffer (20 mM sodium phosphate, pH 7.0), containing polyclonal goat Ab against VEGFR-2 (sc-6251; Santa Cruz), diluted 200 ␮g/ml, for 60 min at ⫹ 4°C. In negative

The FASEB Journal

KILIC ET AL.

controls, supernatants were incubated with protein A-sepharose CL-4B beads in binding buffer without adding VEGFR-2 Ab. The supernatants and pellet from these experiments were then mixed and incubated overnight at ⫹ 4°C. Immunoprecipitates were centrifuged at 14,000 rpm for 20 min at ⫹ 4°C, washed four times with lysis buffer and resuspended in sodium dodecylsulphate (SDS) sample buffer, boiled for 5 min, and analyzed by Western blotting. Protein concentration was determined using a Bio-Rad protein assay kit. 20 ␮g of immunoprecipitated proteins were electrophoresed on 7% SDS-PAGE and transferred onto PVDF membranes. Membranes were then blocked with 5% nonfat milk solution in 10 mM Tris-HCl, pH 7.4/150 mM NaCl and 0.5% Nonidet p-40 at room temperature for 1 h and incubated with mouse monoclonal antip-Tyr Ab (PY99; sc-7020; Santa Cruz Biotechnology), diluted 1:500. Membranes were stripped and reprobed with VEGFR-2 Ab (200 ␮g/ml) to ascertain equal loading of protein. The phosphorylation state of VEGFR-2 was analyzed densitometrically on the p-Tyr blots and normalized with the concentration of total VEGFR-2 expression.

Western blots Tissue samples harvested from the striatum and overlying cortex of ischemic WT and V1tg mice were complemented with lysis buffer, homogenized, and centrifuged. Supernatants were used for SDS-PAGE. Prior to processing, samples from animals belonging to the same experimental group (n⫽5– 6 animals/group) were pooled. After SDS-PAGE, proteins were transferred onto PVDF membranes. Membranes were dried, incubated in blocking solution and immersed with polyclonal rabbit antitotal (detecting both the phosphorylated and unphosphorylated forms) Akt (9272; Cell Signaling, Allschwil, Switzerland), rabbit antiphospho-Akt (9271; Cell Signaling), rabbit antitotal ERK-1/-2 (9102; Cell Signaling), mouse antiphospho-ERK-1/-2 (M8159; Sigma), rabbit antitotal p38 (9212, Cell Signaling), rabbit antiphospho-p38 (9211, Cell Signaling), rabbit antitotal JNK-1/-2 (JNK-2, sc572; Santa Cruz) or rabbit antiphospho-c-Jun NH2-terminal kinase-1/-2 (9251; Cell Signaling) Ab, each diluted 1:500 in 0.1% Tween 20/0.1 M Tris-buffered saline (TBS) (see also 17). Membranes were rinsed, incubated in peroxidase-coupled secondary antibodies, diluted 1:2000 in 0.1% Tween 20/0.1 M TBS, washed, immersed in enhanced chemoluminescence (ECL) solution, and exposed to ECL-Hyperfilm (Amersham, Braunschweig, Germany). Protein loading was controlled with a monoclonal mouse Ab against anti-␤-actin (A5316; Sigma). Blots were performed at least three times. Protein levels were analyzed densitometrically, corrected with values determined on anti-␤-actin blots, and expressed as relative values compared with WT mice.

Inhibition of the PI3K/ Akt pathway By means of a glass microelectrode with a tip outer diameter of 50 ␮m, 2 ␮l of either: 1) 100% dimethyl sulfoxide (DMSO) or 2) the PI3K/Akt inhibitor Wortmannin (0.1 mM; Sigma), dissolved in 100% DMSO, was carefully injected i.c.v. in WT and V1tg mice (n⫽5– 6 animals/group). After 30 min, animals were submitted to MCA occlusions, according to the same procedure as described above. After 24 h, neurological deficits were evaluated. Then, animals were deeply re-anesthetized and sacrificed. Brains were cut in 18 ␮m cryostate sections, and tissue samples were also harvested from the MCA territory.

Statistics Measurements were performed by two investigators blinded for the experimental conditions. For statistical analysis, a standard software package [Statistical Packages for the Social Sciences (SPSS) for Windows 10.1] was used. Values are given as means ⫾ sd. Differences between groups were compared by using one-way ANOVA, followed by least significant differences tests. P-values less than 0.05 were considered significant.

RESULTS VEGF receptor-2 (VEGFR-2) is expressed on ischemic neurons and astrocytes and is activated by human VEGF VEGFR-2 has been made responsible for VEGF’s neuroprotective function in vitro (5). To find out whether VEGFR-2 is expressed in the V1tg mouse brain, which is protected against focal cerebral ischemia and also reveals an increased BBB permeability after stroke (9), we examined VEGFR-2 expression by immunohistochemistry. In the nonischemic cerebral cortex, VEGFR-2 was detected mainly on brain capillaries (9). In the ischemic cortex, however, a distinct signal of VEGFR-2 was also noted on NeuN⫹ neurons and GFAP⫹ astrocytes of WT and V1tg animals, particularly in the infarct borderzone (Fig. 1A, B). To test, whether human VEGF activates the VEGFR-2, we precipitated this receptor from tissue samples obtained from the ischemic MCA territory that we further analyzed with a phospho-tyrosine Ab. Our experiments revealed an increased phosphorylation state of VEGFR-2 in V1tg mice (Fig. 2), confirming that VEGFR-2 was functional. VEGF increases phosphorylated Akt and ERK-1/-2 and reduces phosphorylated p38 and JNK-1/-2 levels To elucidate how VEGF influences cytosolic cell signaling, we prepared Western blots with ischemic tissue samples. Total levels of Akt, but not ERK-1/-2, p38, and JNK-1/-2, were lower in ischemic V1tg than WT mice (Fig. 3), which, however, did not reflect the levels of the phosphorylated proteins. As such, phosphorylated Akt and ERK-1/-2 levels were increased, whereas phosphorylated p38 and JNK-1/-2 levels were reduced in V1tg brains (Fig. 3). Our data suggested an involvement of the PI3K/Akt pathway in VEGF’s effects on ischemic injury and BBB integrity. VEGF down-regulates inducible NO synthase (iNOS) in ischemic neurons To evaluate whether VEGF influences the expression of iNOS, which contributes to free radical formation in the reperfused brain and presumably exacerbates ischemic damage (18), we performed immunohistochemical stainings for iNOS. Robust expression of iNOS was noticed in ischemic brain regions of WT mice that was


E309

significantly diminished by human VEGF (Fig. 4A, B). Double-stainings for NeuN revealed that a large percentage of iNOS⫹ cells (⬎50%) were neurons (Fig. 4A). Our observations propose that neuroprotection by VEGF involves the down-regulation of iNOS. The PI3K/Akt inhibitor Wortmannin reverses VEGF’s neuroprotective properties and restores BBB integrity To clarify the role of the PI3K/Akt pathway in VEGF’s effects on neuronal survival and BBB permeability, we examined the influence of 2 ␮l of the solvant DMSO or of DMSO containing 0.1 mM Wortmannin in WT and V1tg mice submitted to focal cerebral ischemia. LDF

Figure 2. VEGFR-2 is activated by human VEGF. Immunoprecipitation analysis of VEGFR-2 phosphorylation with ischemic tissue samples obtained from WT and V1tg mice. Protein samples were precipitated with a polyclonal goat Ab against VEGFR-2, and Western blotting was subsequently performed with monoclonal mouse antiphospho-tyrosine (pTyr) Ab. Membranes were then stripped and reprobed with VEGFR-2 Ab to ascertain equal protein loading. Note the elevated phosphorylation concentration of VEGFR-2 in V1tg mice. VEGFR-2 phosphorylation was analyzed by measuring the densities of p-Tyr bands and normalizing them with corresponding blots for (total) VEGFR-2. Values are mean ⫾ sd values (n⫽3 different samples/ group). *P ⬍ 0.05 compared with WT mice.

measurements during ischemia did not reveal any differences between groups, neither when WT and V1tg animals were studied, nor when DMSO or DMSO/ Wortmannin were compared (Fig. 5A). Yet, PI3K/Akt inhibition with Wortmannin completely abolished VEGF’s neuroprotection, both infarct volume (Fig. 5B) and neurological deficits (Fig. 5C) being reversed to levels in WT mice. At the same time, brain swelling (Fig. 5D) and IgG extravasation (Fig. 5E) were attenuated by the PI3K/Akt blockade. Thus, brain edema and BBB permeability dropped below levels in WT animals (Fig. 5D, E). In WT mice, the delivery of Wortmannin did not have any effect (Fig. 5A–E). Our results confirm that the PI3K/Akt pathway is indeed responsible for both, VEGF’s neuroprotection and BBB leakage.

Figure 1. VEGF receptor-2 (VEGFR-2) is expressed on ischemic neurons and astrocytes after focal cerebral ischemia. Immunohistochemistries revealing VEGFR-2⫹/NeuN⫹ neurons and VEGFR-2⫹/GFAP⫹ astrocytes in the ischemic (sensory) cortex close to the infarct border of WT mice (A). Note that VEGFR-2 expression does not differ between WT and V1tg mice (B). VEGFR-2 was rarely expressed on parenchymal cells in the nonischemic brain. Graphs show parenchymal cells (i.e., neurons and astrocytes), evaluated in squares measuring 62⬘500 ␮m2, expressed as means ⫾ sd (n⫽7– 8 animals/ group). *P ⬍ 0.05 compared with nonischemic mice. Fluorescent dyes: VEGFR-2, Cy-3 (red); NeuN, FITC (green); GFAP, FITC (green); DAPI (blue). Bar, 60 ␮m. E310

Vol. 20

June 2006

Reversal of VEGF neuroprotection by Wortmannin associated with restoration of p38, but not ERK-1/-2 and JNK-1/-2 To analyze downstream effects of PI3K/Akt in VEGFneuroprotection mice, we also prepared Western blots with tissue samples obtained from DMSO and DMSO/ Wortmannin treated mice. In these blots, we confirmed that Wortmannin indeed inhibits Akt phosphorylation (Fig. 3). In contrast to phosphorylated ERK-1/-2 and JNK-1/-2, which were not influenced by the PI3K/Akt inhibitor, p38 phosphorylation was increased in Wortmannin-treated animals that exhibited exacerbated

The FASEB Journal

KILIC ET AL.

Figure 3. Human VEGF increases phosphorylated (but not total) Akt and ERK-1/-2 and reduces phosphorylated MAP kinase/ p38 and JNK-1/-2 levels after focal cerebral ischemia. Western blots with tissue samples obtained from WT and V1tg mice that were either untreated or i.c.v. treated with 2 ␮l of the solvant DMSO or DMSO containing Wortmannin (0.1 mM), a PI3K/Akt inhibitor. Tissue samples were taken from the ischemic cortex and underlying striatum. Note that delivery of Wortmannin blocks Akt phosphorylation and also reverses phosphorylated p38 to levels in WT mice, indicating that MAP kinase/ p38 phosphorylation closely depends on PI3K/Akt activity. Values are mean ⫾ sd (n⫽3 different samples/group), normalized with corresponding blots for ␤-actin. *P ⬍ 0.05 compared with WT mice; #P ⬍ 0.05 compared with V1tg mice receiving DMSO.

brain infarcts (Fig. 3), suggesting that p38 phosphorylation closely depends on the PI3K/Akt activation state. PI3K/Akt inhibition does not restore iNOS levels In order to find out whether iNOS inhibition by VEGF occurs in a PI3K/Akt-dependent manner, we further analyzed brain sections from DMSO and DMSO/Wortmannin-treated mice that we stained for iNOS. In contrast to MAP kinase/p38, which was activated when PI3K/ Akt was blocked, iNOS levels were not reversed by Wortmannin (Fig. 4), demonstrating that iNOS inhibition by VEGF does not depend on PI3K/Akt.

DISCUSSION We demonstrate that the PI3K/Akt pathway mediates VEGF’s neuroprotective activity and also induces BBB permeability after focal cerebral ischemia. Our data were obtained using the transgenic mouse line V1 that expresses human VEGF under a NSE promoter (13). We

submitted these mice to 90 min of intraluminal MCA occlusion and examined cell signaling factors involved in VEGF’s neuroprotective function and BBB integrity, by combining IgG extravasation studies with histochemical (Western blotting, immunocytochemistry, immunoprecipitation) and pharmacological (i.c.v. delivery of PI3K/ Akt inhibitor Wortmannin) experiments. We have previously shown in V1tg mice that human VEGF protects against ischemia in a caspase-3 dependent manner and ameliorates neurological deficits after stroke, despite an increased vascular permeability that was noticed in V1tg mice (9). The survival-promoting effects of VEGF were not due to hemodynamic improvements. Conversely, a reduction of cerebral blood flow was noticed in ischemic regions of V1tg mice by 14C-iodoantipyrine autoradiography, which could be attributed to a hemodynamic steal of blood from ischemic to nonischemic regions (9). VEGFR-2 is expressed on ischemic neurons and astrocytes and is activated by human VEGF If VEGF indeed has neuroprotective effects in the brain, as our earlier data indicated (9), the question


E311

arises of how VEGF exerts its neuroprotective function. By demonstrating that VEGFR-2 is expressed on ischemic neurons and astrocytes, VEGFR-2 being activated by human VEGF, we provide in vivo evidence that VEGF may directly promote neuronal survival. In vivo, VEGFR-2 has previously been shown on endothelial cells under physiological conditions, but not on neurons and glial cells (19). From in vitro studies it was already known that neurons can express VEGFR-2

Figure 4. Human VEGF inhibits inducible NO synthase (iNOS) in the ischemic cortex in a PI3K/Akt-independent manner. Immunohistochemical double-stainings for iNOS (Cy-3, red) and the neuronal marker protein NeuN (FITC, green) counterstained with DAPI (blue) (A). V1tg mice were either untreated or received i.c.v. injections of DMSO or DMSO containing of the PI3K/Akt inhibitor Wortmannin. Cell densities were evaluated in the parietal cortex, which reproducibly revealed brain injury (B). Note that the iNOS inhibition evoked by vascular endothelial growth factor is not reversed by Wortmannin. Values are mean ⫾ sd (n⫽5– 6 animals/group). *P ⬍ 0.05 compared with WT mice. Bar, 100 ␮m. ‹ Figure 5. The PI3K/ Akt pathway mediates VEGF’s neuroprotective function and BBB permeability. Laser Doppler flow (LDF) measurements (A) in WT and V1tg mice that were either untreated or received i.c.v. injections of DMSO or DMSO containing the PI3K/ Akt inhibitor Wortmannin. Note that cerebral blood flow during ischemia does not differ between groups (A). Assessments of infarct volume (B) and neurological deficits (C) reveal that human VEGF significantly reduces brain injury and ameliorates postischemic recovery. Brain swelling (D) is not influenced by human VEGF, whereas IgG extravasation (E) is moderately increased, indicating that VEGF expression promotes BBB leakage. Note that the neuroprotective effects of vascular endothelial growth factor are reversed by Wortmannin (B, C), whereas brain swelling (D) and BBB permeability (E) are reduced, despite the exacerbation of brain infarcts. In WT mice, the PI3K/Akt inhibitor does not have an effect (A–E). These data demonstrate that the PI3K/ Akt pathway is responsible for VEGF’s neuroprotective function and BBB leakage. Values are means ⫾ sd (n⫽5–11 animals/group). *P ⬍ 0.05 compared with WT mice; #P ⬍ 0.05 compared with V1tg mice receiving DMSO. E312

Vol. 20

June 2006

The FASEB Journal

KILIC ET AL.

(5–7), in addition to VEGF receptor-1 (6, 7) and neuropilin-1 (5), another VEGF receptor. As VEGFR-2 but not VEGF receptor-1 antisense oligonucleotides reversed VEGF’s neuroprotective activity (6) and neuropilin-1 ligands were unable to mimic VEGF’s survival effects (5), it was concluded that VEGFR-2 was responsible for VEGF’s neuroprotection. Based on our present results, VEGFR-2 is expressed and activated on neurons and astrocytes in vivo following stroke, but not under physiological conditions. Role of the PI3K/Akt pathway in VEGF’s neuroprotective activity In accordance with in vitro findings indicating that VEGF induces neuroprotection via the PI3K/Akt and ERK-1/-2 pathways downstream of VEGFR-2 (5–7, 20) and in agreement with very recent data showing elevated phosphorylated Akt levels in the ischemic brain (21), we demonstrate that VEGF stimulates Akt and ERK-1/-2 phosphorylation in the stroke brain. Although the brain comprises various cell-types, which may respond differently to ischemia and other types of injury, Western blotting is a very useful tool that allows us to evaluate molecular actions of brain-protective molecules. Indeed, in pharmacological experiments using the PI3K/Akt inhibitor Wortmannin, we show that the PI3K/Akt pathway mediates VEGF’s survivalpromoting function. Our observation that VEGF induces neuroprotection via the PI3K/Akt pathway is noteworthy, as this pathway stabilizes mitochondrial function under conditions in which tissue oxygen levels are compromised. Via phosphorylation of Bad, activated Akt releases Bcl-XL inside the mitochondria, which in turn prevents formation of the mitochondrial permeability transition pore and preserves the mitochondrial membrane potential (e.g., see 15). Bcl-XL, on the other hand, once released from Bad, blocks the secretion of cytochrome c from injured mitochondria, thereby inhibiting caspase-9 and -3 and preventing DNA cleavage (15). Role of PI3K/ Akt in VEGF-induced BBB permeability That the PI3K/Akt pathway also mediates VEGF’s BBB permeability, besides its neuroprotective influence, is a further major result of our study. Despite in vitro reports, which already suggested previously that the PI3K/Akt pathway is involved in VEGF’s endothelial permeability (22), it has to our knowledge not been demonstrated that PI3K/Akt signaling mediates vascular permeability after stroke. This finding is of interest, as vascular permeability and infarct size were inversely related to each other. In endothelial culture, it was previously shown that VEGF’s vascular permeability is mediated via serine/threonine phosphorylation and redistribution of tight junction proteins, zona occludens-1, and occludin, which are controlled by PI3K/Akt (22, 23). It is noteworthy that Wortmannin did not

influence brain swelling and BBB integrity in WT mice. This points toward a specific role of the PI3K/Akt pathway in VEGF’s BBB permeability. Inhibition of MAP kinase/p38 but not JNK-1/-2 by VEGF depends on PI3K/ Akt activity Elevated levels of VEGF in our transgenic mice not only stimulated Akt and ERK-1/-2 but also inhibited MAP kinase/ p38 and JNK-1/-2. p38 and JNK-1/-2 are both stress-inducible kinases, which are strongly activated in injured neurons (for review, 25). We previously found that acute hypoxia-induced VEGF expression in the brain resulted in p38-mediated edema formation (24). It is known that acute increases in VEGF levels lead to edema formation, while delayed VEGF elevations do not (12). The difference may result from an attenuated p38 response when sustained elevated VEGF levels are present. Indeed, no signs of edema formation were found in the brain of V1tg mice, despite the presence of high VEGF levels. Interestingly, the inhibition of p38 but not of JNK-1/-2 in VEGF overexpressing mice was reversed by Wortmannin. This suggests that the regulation of MAP kinase/p38 by VEGF closely depends on PI3K/Akt. Inhibition of iNOS by VEGF does not depend on the PI3K/ Akt pathway The inhibition of iNOS levels by VEGF, which we found predominantly in neurons, is also new and may represent a hitherto unknown player also in the context of VEGF’s neuroprotection. Indeed, iNOS is supposed to exacerbate ischemic damage after stroke, as it contributes to free radical stress via formation of nitrite and nitrate (18). Based on our data, VEGF may protect the brain tissue by reducing tissue levels of these highly toxic compounds. Inhibition of iNOS after focal cerebral ischemia has recently been shown by us also for another hypoxia-inducible factor, erythropoietin, which is also protective after brain ischemia (14). It is noteworthy that both in our recent study (14) and the present experiments, iNOS inhibition was not reversed by Wortmannin, which indicates that iNOS inhibition occurs PI3K/Akt independently. Our data show that VEGF and erythropoietin share common signaling mechanisms that provide protection against free radical damage in the hypoxic-ischemic brain.

CONCLUSIONS By demonstrating that human VEGF activates VEGFR-2, Akt, and ERK-1/-2 and inhibits MAP kinase/p38, JNK1/-2 and iNOS in the ischemic brain, we provide insight into the signaling pathways underlying VEGF’s neuroprotective activity. We show that the PI3K/Akt pathway mediates VEGF’s neuroprotection and BBB permeability in mice submitted to focal cerebral ischemia, most likely in a VEGFR-2 dependent way. Based on our


E313

finding that the same PI3K/Akt pathway is responsible for VEGF’s survival promoting function and BBB leakage, we predict that it may not easily be possible to make use of VEGF’s neuroprotective action after stroke without accepting its unfavorable consequence, the increased BBB permeability. We thank A. Fendel for technical assistance in the histochemical experiments. Supported by grants from the Swiss National Science Foundation (3200B0 –100790), the NCCR “Neural plasticity and repair,” the Center of Integrative Human Physiology (CIHP), the Hartmann-Mu¨ller-Stiftung (all to D.M.H.), and by a grant from the Ministry of Science, Research and the Arts of Baden-Wu¨rttemberg (23-7532.22-2012/1 (to H.H.M.).

11.

12.

13. 14.

15. 16.

REFERENCES 17. 1.

2.

3.

4.

5. 6.

7.

8.

9.

10.

E314

Monacci, W. T., Merrill, M. J., and Oldfield, E. H. (1993) Expression of vascular permeability factor/vascular endothelial growth factor in normal rat tissues. Am. J. Physiol. 264, C995– C1002 Marti, H. H., and Risau, W. (1998) Systemic hypoxia changes the organ-specific distribution of vascular endothelial growth factor and its receptors. Proc. Natl. Acad. Sci. U.S.A. 95, 15809 – 15814 Forsythe, J. A., Jiang, B.-H., Iyer, N.V., Agani, F., Leung, S. W., Koos, R. D., et al. (1996) Activation of vascular endothelial growth factor gene transcription by hypoxia-inducible factor 1. Mol. Cell. Biol. 16, 4604 – 4613 Marti, H. J. H., Bernaudin, M., Bellail, A., Schoch, H., Euler, M., Petit, E., et al. (2000) Hypoxia-induced vascular endothelial growth factor expression precedes neovascularization after cerebral ischemia. Am. J. Pathol. 156, 965–976 Jin, K. L., Mao, X. O., and Greenberg, D. A. (2000) Vascular endothelial growth factor: direct neuroprotective effect in in vitro ischemia. Proc. Natl. Acad. Sci. U.S.A. 97, 10242–10247 Matsuzaki, H., Tamatani, M., Yamaguchi, A., Namikawa, K., Kiyama, H., Vitek, M. P., et al. (2001) Vascular endothelial growth factor rescues hippocampal neurons from glutamate-induced toxicity: signal transduction cascades. FASEB J. 15, 1218 –1220 Wick, A., Wick, W., Waltenberger, J., Weller, M., Dichgans, J., and Schulz, J. B. (2002) Neuroprotection by hypoxic preconditioning requires sequential activation of vascular endothelial growth factor receptor and Akt. J. Neurosci. 22, 6401– 6407 Sun, Y., Jin, K., Xie, L., Childs, J., Mao, X. O., Logvinova, A., and Greenberg, D. A. (2003) VEGF-induced neuroprotection, neurogenesis, and angiogenesis after focal cerebral ischemia. J. Clin. Invest. 111, 1843–1851 ¨ ., Weber, B., Bassetti, C. L., Marti, Wang, Y., Kilic, E., Kilic, U H. H., and Hermann, D. M. (2005) VEGF overexpression induces post-ischaemic neuroprotection, but facilitates haemodynamic steal phenomena. Brain 128, 52– 63 Schoch, H. J., Fischer, S., and Marti, H. H. (2002) Hypoxiainduced vascular endothelial growth factor expression causes vascular leakage in the brain. Brain 125, 2549 –2557

Vol. 20

June 2006

18. 19. 20.

21.

22.

23.

24.

25.

van Bruggen, N., Thibodeaux, H., Palmer, J. T., Lee, W. P., Fu, L., Cairns, B., et al. (1999) VEGF antagonism reduces edema formation and tissue damage after ischemia/reperfusion injury in the mouse brain. J. Clin. Invest. 104, 1613–1620 Zhang, Z. G., Zhang, L., Jiang, Q., Zhang, R., Davies, K., Powers, C., et al. (2000) VEGF enhances angiogenesis and promotes blood-brain barrier leakage in the ischemic brain. J. Clin. Invest. 106, 829 – 838 Vogel, J., Gehrig, M., Kuschinsky, W., and Marti, H. H. (2004) Massive inborn angiogenesis in the brain scarcely raises cerebral blood flow. J. Cereb. Blood Flow Metab. 24, 849 – 859 ¨ ., Soliz, J., Bassetti, C. L., Gassmann, M., and Kilic, E., Kilic, U Hermann, D. M. (2005) Brain-derived erythropoietin protects from focal cerebral ischemia by dual activation of ERK-1/-2 and Akt pathways. FASEB J. 19, 2026 –2028 Kilic, E., Dietz, G. P., Hermann, D. M., and Ba¨hr, M. (2002) Intravenous TAT-Bcl-XL is protective after middle cerebral artery occlusion in mice. Ann. Neurol. 52, 617– 622 Hermann, D. M., Mies, G., and Hossmann, K. A. (1998) Effects of a traumatic neocortical lesion on cerebral metabolism and gene expression of rats. Neuroreport 9, 1917–1921 ¨ ., Kilic, E., Soliz, J., Bassetti, C. L., Gassmann, M., and Kilic, U Hermann, D. M. (2005) Erythropoietin protects from axotomyinduced degeneration of retinal ganglion cells by activating ERK-1/-2. FASEB J. 19, 249 –251 Dalkara, T., and Moskowitz, M. A. (1994) The complex role of nitric oxide in the pathophysiology of focal cerebral ischemia. Brain Pathol. 4, 49 –57 Kovacs, Z., Ikezaki, K., Samoto, K., Inamura, T., and Fukui, M. (1996) VEGF and flt. Expression time kinetics in rat brain infarct. Stroke 27, 1865–1872 Jin, K., Mao, X. O., Batteur, S. P., McEachron, E., Leahy, A., and Greenberg, D. A. (2001) Caspase-3 and the regulation of hypoxic neuronal death by vascular endothelial growth factor. Neuroscience 108, 351–358 ¨ zdemir, Y., Yemisci, M., Tuncer, N., Aktan, S., Kaya, D., Gursoy-O and Dalkara, T. (2005) VEGF protects against focal ischemia without increasing blood-brain barrier when delivered intracerebroventricularly. J. Cereb. Blood Flow Metab. 25, 1111–1118 Pedram, A., Razandi, M., and Levin, E. R. (2002) Deciphering vascular endothelial cell growth factor/vascular permeability factor signaling to vascular permeability. Inhibition by atrial natriuretic peptide. J. Biol. Chem. 277, 44385– 44398 Fischer, S., Wiesnet, M., Marti, H. H., Renz, D., and Schaper, W. (2004) Simultaneous activation of several second messengers in hypoxia-induced hyperpermeability of brain derived endothelial cells. J. Cell. Physiol. 198, 359 –369 Issbru¨cker, K., Marti H. H., Hippenstiel, S., Springmann, G., Voswinckel R., Gaumann, A., Breier, G., Drexler, H. C. A., Suttorp, N., and Clauss, M. (2003) p38 MAP kinase—a molecular switch between VEGF-induced angiogenesis and vascular hyperpermeability. FASEB J. 17, 262–264 Irving, E. A., and Bamford, M. (2002) Role of mitogen- and stress-activated kinases in ischemic injury. J. Cereb. Blood Flow Metab. 22, 631– 647

The FASEB Journal

Received for publication October 24, 2005. Accepted for publication January 27, 2006.

KILIC ET AL.