Vascular Endothelial Growth Factor Promotes Pericyte Coverage of Brain Capillaries, Improves Cerebral Blood Flow During Subsequent Focal Cerebral Ischemia, and Preserves the Metabolic Penumbra
Background and Purpose—Therapeutic angiogenesis aims at improving cerebral blood flow by amplification of vascular sprouting, thus promoting tissue survival under conditions of subsequent ischemia. It remains unknown whether induced angiogenesis leads to the formation of functional vessels that indeed result in hemodynamic improvements. Observations of hemodynamic steal phenomena and disturbed neurovascular integrity after vascular endothelial growth factor delivery questioned the concept of therapeutic angiogenesis.
Methods—Mice were treated with recombinant human vascular endothelial growth factor (0.02 μg/d; intracerebroventricular) for 3 to 21 days and subsequently exposed to 90-minute middle cerebral artery occlusion. Angiogenesis, histological brain injury, IgG extravasation, cerebral blood flow, protein synthesis and energy state, and pericyte coverage on brain capillaries were evaluated in a multiparametric approach combining histochemical, autoradiographic, and regional bioluminescence techniques.
Results—Vascular endothelial growth factor increased brain capillary density within 10 days and reduced infarct volume and inflammation after subsequent middle cerebral artery occlusion, and, when delivered for prolonged periods of 21 days, enhanced postischemic blood–brain barrier integrity. Increased cerebral blood flow was noted in ischemic brain areas exhibiting enhanced angiogenesis and was associated with preservation of the metabolic penumbra, defined as brain tissue in which protein synthesis has been suppressed but ATP preserved. Vascular endothelial growth factor enhanced pericyte coverage of brain endothelial cells via mechanisms involving increased N-cadherin expression on cerebral microvessels.
Conclusions—That cerebral blood flow is increased during subsequent ischemic episodes, leading to the stabilization of cerebral energy state, fosters hope that by promoting new vessel formation brain tissue survival may be improved.
Patients with cerebrovascular diseases frequently exhibit chronic hemodynamic disturbances,1 predisposing them to severe and devastating strokes. There is a major need for revascularization strategies in such patients.1 Also in patients without hemodynamic deficits, vascular sprouting takes place after stroke.2,3 It has been proposed that subjects at vascular risk might benefit from the therapeutic amplification of angiogenesis.4 The underlying hypothesis is that vascular growth should increase blood flow and reduce the impact of future stroke events.3
Animal studies examining the effects of angiogenesis induced by vascular endothelial growth factor (VEGF) rapidly led to clinical trials in a variety of medical conditions, including coronary heart and peripheral occlusive artery disease.3,4 However, such trials have so far been of limited success.3,4 This raises questions about the viability of angiogenesis as a therapeutic approach. In animal models of ischemic stroke, VEGF-induced angiogenesis has been shown to result in structural neuroprotection and functional neurological recovery.5,6 Whether the preservation of ischemic tissue was a consequence of enhanced cerebral blood flow (CBF) or of neuroprotective effects of VEGF was unclear.3 VEGF promotes neuronal survival both directly via VEGF’s receptor VEGFR27 and indirectly by release of brain-derived neurotrophic factor.8
Although several studies have examined VEGF-induced angiogenesis in models of focal cerebral ischemia,3 only 2 have evaluated the extent to which VEGF influences regional CBF in the ischemic brain. In a MRI study, acute VEGF infusion was shown to induce a transient CBF increase in ischemic brain tissue lasting over 3 hours,5 which was interpreted as vasorelaxation induced by this growth factor. In mice expressing human VEGF chronically under a neuron-specific NSE promoter, increased CBF was observed in nonischemic brain areas alongside a reduction in CBF in ischemic brain areas,6 suggesting that the enhanced angiogenesis had induced a hemodynamic steal flow. The mice examined expressed human VEGF throughout the brain, resulting in globally increased vessel densities.6 CBF was scarcely altered in this mouse line under physiological conditions.9
Raising further questions whether newly formed blood vessels are functional, loss of pericyte coverage of endothelial cells has previously been noticed in Matrigel assays following VEGF treatment in a model of platelet-derived growth factor (PDGF)-BB–induced angiogenesis.10 Immunoprecipitation studies revealed a hitherto unknown VEGF-induced deactivation of PDGF-BB’s receptor PDGFRβ that is expressed on pericytes, which was mediated by the interaction of PDGFRβ with VEGF’s receptor VEGFR2.10 Inhibition of VEGFR2 prevented the formation of this receptor complex and restored pericyte coverage, thus stabilizing the newly formed blood vessels.10 Whether interactions between VEGFR2 and PDGFRβ take place in vivo in the brain and whether they are relevant for ischemic stroke were unknown.
To evaluate the concept of VEGF-induced therapeutic angiogenesis, we exposed mice that had been treated intracerebroventricularly with VEGF to focal cerebral ischemia. In a multiparametric imaging strategy, combining CBF and cerebral protein synthesis (CPS) autoradiography, regional ATP bioluminescence imaging and histochemical techniques, we then assessed the effects of VEGF-induced angiogenesis on regional CBF, brain metabolism, and vascular integrity.
Materials and Methods
Experiments were performed with government approval according to the National Institutes of Health guidelines for the care and use of laboratory animals. In the first set of studies, male C57BL6/j mice (20–25 g) were randomly assigned to 6 groups treated with vehicle (normal saline) or recombinant human VEGF165 (rhVEGF165) for 3, 10, or 21 days (6–7 animals per group). In these animals, no ischemia was induced. Brain capillary and arteriole densities and diameters were evaluated by immunohistochemistry or ex vivo angiography. In the second set of studies, male C57BL6/j mice were randomly assigned to 6 groups treated with vehicle (normal saline) or recombinant human VEGF165 (rhVEGF165). Three, 10, or 21 days later, 90 minutes of middle cerebral artery (MCA) was induced, followed by 24 hours of reperfusion (6–7 animals per group). These animals were used for histochemical analysis of angiogenesis, ischemic injury, and molecular biological studies. A third set of male C57BL6/j mice was treated with vehicle or rhVEGF165 for 21 days, followed by 90-minute MCA occlusion and 60-minute reperfusion. These animals were used for CBF and CPS double autoradiography and regional ATP bioluminescence imaging.
Delivery of Recombinant Human VEGF165
Cannulae linked to miniosmotic pumps (Alzet 2004; Palo Alto, CA) were implanted into the left lateral ventricle for administration of vehicle or rhVEGF165 (Peprotech, Hamburg, Germany; 0.02 μg/d), as described.11 For details, see Methods in the online-only Data Supplement.
Evaluation of Brain Microvessel Density
In 20-μm cryostat sections obtained from the rostrocaudal level of the midstriatum, that is, the site of maximum extension of the MCA territory, brain microvessels were evaluated by fluorescence immunohistochemistry using a rat anti-CD31 antibody (BD Biosciences, San Diego, CA).11 The densities of capillaries (diameter, <10 μm) and arterioles (≥10 μm) were analyzed in 7 regions of interest (3 in parietal cortex and 4 in lateral striatum) in the MCA territory both ipsilateral and contralateral to the stroke using the Cell-F software (Olympus, Hamburg, Germany). For details, see Methods in the online-only Data Supplement.
Ex Vivo Angiography of Vessel Lumina
Microvessel lumina were visualized by injection of a mixture of carbon black dyes through the left ventricle.12 Photographs were taken from the ventral aspect of these brains, on which the mean diameter of the MCA was measured. Subsequently, cryostat sections were obtained that were stained using a rabbit anti-collagen-IV (1:500; Millipore, Billerica, MA) or rat anti-CD31 (BD Biosciences) antibody.11,12 The diameters of capillary and arteriole lumina were quantified in 7 regions of interest (see above) of the MCA territory both ipsilateral and contralateral to the stroke. For details, see Methods in the online-only Data Supplement.
Induction of Focal Cerebral Ischemia
Intraluminal MCA occlusion was induced during 1% isoflurane anesthesia using a silicon-coated microfilament.6,13 Animals were euthanized 24 hours later by transcardial perfusion with normal saline or kept under anesthesia for delivery of radioactive tracers. For details, see Methods in the online-only Data Supplement.
Cryostat sections 1 mm apart were stained with cresyl violet.6 The border between infarcted and healthy tissue was outlined using image analysis software (Image J; National Institutes of Health, Bethesda, MD) and the infarct volume quantified.6
Serum IgG Extravasation Studies
Brain sections obtained from the midstriatum were processed for serum IgG immunohistochemistry.13 Stained sections were scanned, converted into gray values, and densitometrically analyzed within the core of the MCA territory.
Protein lysates were obtained from extracted crude microvessels using samples collected from the ischemic and contralateral nonischemic MCA territory.14 Polyvinylidene fluoride membranes dissolved by SDS-PAGE were incubated with rabbit anti-N-cadherin antibody (4061; Millipore, Schwalbach, Germany). After secondary antibody exposure, membranes were exposed to photoluminescence solution. Protein loading was controlled using a mouse anti-β-actin antibody (4967; Millipore). Blots were repeated 3× to confirm reproducibility. Protein levels were densitometrically analyzed. For details, see Methods in the online-only Data Supplement.
Evaluation of Pericyte Coverage of Brain Capillaries
To evaluate endothelial proliferation and the pericyte coverage of brain capillaries, brain sections were stained with rat anti-CD31 (BD Biosciences; endothelial marker), rabbit anti-desmin (Abcam, Cambridge, UK; pericyte marker), goat-anti-PDGFRβ (R&D Systems; pericyte marker), rabbit anti-Ki67 (ab15580, Abcam, proliferation marker), goat anti-VEGFR2 (AF644; R&D Systems), rat anti-VEGFR1 (MAB471; R&D Systems), or rabbit anticleaved caspase-3 (AB3623, Millipore) antibody,15 which in some experiments was counterstained with terminal transferase dUTP nick-end labeling (kit 11684795910; Roche, Basel, Switzerland). In CD31/desmin, CD31/PDGFRβ, and CD31/Ki67 doublestainings, the percentage of pericyte+(desmin, PDGFRβ) or proliferating (Ki67) microvessels was examined in 7 regions of interest (see above) ipsilateral and contralateral to the stroke, out of which mean values were formed. Confocal 3-dimensional (3D) stacks were also obtained using a laser scanning microscope (LSM 510; Carl Zeiss MicroImaging, Jena, Germany). In these stacks, CD31+capillaries and surrounding desmin+pericytes were outlined, allowing for the evaluation of capillary and pericyte volumes, respectively, which were integrated for all 2-dimensional images, resulting in capillary and pericyte volumes, from which volume ratios were formed. For details, see Methods in the online-only Data Supplement.
CBF and CPS Double Autoradiography
A total of 150 μCi L-[4,5-3H] leucine (specific activity 151 Ci/mmol; Amersham, Braunschweig, Germany) were administered intraperitoneally, followed by an intraperitoneal injection of 10 μCi 4-iodo-N-methyl-[14C] antipyrine (Amersham) 43 minutes later. After another 2 minutes, animals were instantly frozen in liquid nitrogen. Regional CBF and CPS were calculated as described previously.6,16,17 For details, see Methods in the online-only Data Supplement.
Regional ATP Bioluminescence Imaging
For ATP measurement, frozen sections were coated with a layer of frozen reaction mix, and light emissions were recorded using a CCD camera. The metabolic penumbra was calculated by subtracting the ATP-preserved area obtained from ATP bioluminescence images and the CPS-deficient area determined from CPS autoradiography.18,19 For details, see Methods in the online-only Data Supplement.
Laser Doppler flow recordings were evaluated by repeated measurement ANOVA with values determined at 15-minute intervals during MCA occlusion and at 5-minute intervals after reperfusion. Capillary density, infarct volume, IgG extravasation, and all other histochemical studies comparing 4 or 6 groups were evaluated by 2-way ANOVA followed by 2-tailed t tests. Changes in CBF, ATP depletion, and the metabolic penumbra were analyzed by 2-tailed t tests, as were histochemical studies with comparisons between 2 groups. All data are presented as mean±SD. P<0.05 was considered significant.
Effects of VEGF on Brain Microvessels in Nonischemic Mice
Intraventricular VEGF delivery increased the density and diameter of CD31+ brain capillaries in the left-sided hemisphere, in which VEGF was infused. Capillary density increased starting on day 10 of VEGF treatment (Figure 1A; Table I in the online-only Data Supplement). In the right-sided contralateral hemisphere, capillary density and diameter did not differ between groups (Figure 1B; Table I in the online-only Data Supplement). Endothelial proliferation, assessed by CD31/Ki67 doublestaining, was increased by VEGF at 3 days after initiation of VEGF treatment, more strongly in the left-sided hemisphere receiving the VEGF infusion than contralateral to it (Figure I in the online-only Data Supplement).
In contrast to brain capillaries, the density of arterioles, evaluated in nonischemic animals, was not influenced by VEGF (Table I in the online-only Data Supplement). Notably, VEGF slightly increased the diameter of arterioles. This increase was significant in the contralateral hemisphere after 21 days of VEGF exposure (Table I in the online-only Data Supplement). The diameter of the MCA at its origin was not influenced by VEGF (ipsilateral MCA: vehicle, 0.12±0.01 mm/VEGF, 0.12±0.05 mm; contralateral MCA: vehicle, 0.13±0.01 mm/VEGF, 0.11±0.02 mm).
Effects of VEGF on Brain Capillary Survival in Ischemic Brain Tissue
In animals exposed to MCA occlusion, the survival of CD31+ brain capillaries was increased in ischemic tissue of mice that had been treated with VEGF for 10 or 21 days before (Figure 1C). In the contralateral hemisphere, capillary density again did not differ between groups (Figure 1D).
Effects of VEGF on Ischemic Injury and Blood–Brain Barrier Integrity
Laser Doppler flow recordings taken during and after MCA occlusion to control the reproducibility of ischemias did not reveal any differences between groups (Figure II in the online-only Data Supplement). On MCA occlusion, laser Doppler flow decreased to 10% to 20% of baseline values in all groups. Reperfusion was associated with a rapid restoration of blood flow.
Infarct measurements on cresyl violet staining 24 hours after reperfusion revealed reduced infarct volumes in animals treated with VEGF for 10 or 21 days, but not for 3 days (Figure 2A). The attenuation of brain injury was associated with reduced IgG extravasation in animals receiving VEGF for 21 days, but not for 3 or 10 days (Figure 2B), demonstrating the formation of intact vessels not associated with blood–brain barrier leakage.
VEGF Increases Regional CBF in Nonischemic and Ischemic Brain Tissue
VEGF treatment for 21 days enhanced regional CBF in both hemispheres in nonischemic brain tissue (79.7±63.5 versus 181.9±114.5 mL/100 g per minute in left-sided striatum/96.0±84.9 versus 199.6±107.0 mL/100 g per minute in right-sided striatum of vehicle-treated and VEGF-treated mice, respectively), demonstrating that new vessel formation did indeed translate into a functional improvement of blood flow. When additional MCA occlusion was imposed, increased CBF was noted in ischemic brain tissue (Figure 3A). Interestingly, increased CBF values were again also observed in the contralateral brain (Figure 3B), albeit capillary density was not increased by VEGF. These data suggested that mechanisms other than capillary density contributed to CBF changes.
VEGF-Induced Angiogenesis Stabilizes the Metabolic Penumbra and Prevents ATP Depletion
A stabilization of the metabolic penumbra, defined as brain tissue in which CPS is suppressed but ATP preserved (Figure 3C), and a reduction in the area of tissue exhibiting ATP depletion (Figure 3D) were found in VEGF but not in vehicle-treated animals, thus indicating that blood flow stabilization resulted in a preservation of energy metabolism.
VEGF Increases Pericyte Coverage of Brain Capillaries
In a conventional immunohistochemical analysis, the vast majority of capillaries (≈90%) in the nonischemic brain tissue were surrounded by pericytes that were immunoreactive for desmin (Figure 4B). Thus, no effects of VEGF were noted (not shown). In the ischemic tissue, this value was lower in vehicle-treated mice (≈60%; Figure 4A). Interestingly, VEGF did not affect the percentage of pericyte+capillaries, when initiated 3 or 10 days before stroke, but markedly increased pericyte coverage of ischemic capillaries, when started 21 days before (Figure 4A), indicating that VEGF induces the formation of mature vessels.
In view of the ceiling effect of pericyte coverage in the contralateral nonischemic tissue (Figure 4B), where an increase in CBF (Figure 3B), but not capillary density (Figure 1B), was found, a confocal data analysis was also performed, in which desmin+ pericyte volumes determined in 3D stacks were related to the volumes of CD31+cerebral capillaries. In this analysis, VEGF increased pericyte coverage in both hemispheres, when initiated 21 days before the stroke (Figure 4C and 4D). These data provided an explanation why VEGF increased CBF in the contralateral nonischemic brain tissue, although no increase in capillary density was seen.
Immunohistochemistry for PDGFRβ, a second pericyte marker, exhibited lower pericyte coverage rates than desmin in nonischemic brain tissue (≈60%; Figure III in the online-only Data Supplement). Although VEGF did not affect the percentage of pericyte+capillaries when delivered over 3 or 10 days before stroke, there was a trend toward increased pericyte coverage of ischemic brain capillaries in animals that received VEGF over 21 days (45.5±16.5% versus 62.6±12.2% in vehicle-treated versus VEGF-treated animals; nonsignificant in 2-way ANOVA; Figure III in the online-only Data Supplement). Taken together, our data indicated that VEGF enhances pericyte coverage on brain capillaries.
VEGF Does Not Influence Apoptotic Death of Pericytes
The number of desmin+/terminal transferase dUTP nick-end labeling+ and desmin+/cleaved caspase-3+pericytes was not influenced by VEGF, indicating that VEGF neither influenced DNA fragmentation nor caspase-3 activation in pericytes (Figure IV in the online-only Data Supplement).
VEGFR2 and VEGFR1 Are Not Expressed on Mouse Brain Pericytes
Because VEGF has previously been shown to induce pericyte ablation in Matrigel assays, which was interpreted as consequence of VEGFR2/PDGFRβ interaction on pericytes,10 the question arose whether pericytes in C57BL6/j mice expressed VEGFR2 or VEGFR1. In immunostainings, we were unable to detect VEGFR2 and VEGFR1 on pericytes (Figure V in the online-only Data Supplement).
VEGF Increases N-Cadherin on Cerebral Microvessels
The alignment of pericytes to endothelial cells is mediated by the junctional protein N-cadherin.20 To evaluate whether N-cadherin was influenced by VEGF, Western blots were prepared with capillary extracts which revealed increased N-cadherin expression in microvessels that had been treated with VEGF for 21 days (Figure 5).
By applying prophylactic VEGF for up to 21 days, we have shown that VEGF induces the formation of mature, functional blood vessels, which enables the brain to cope better with subsequent ischemic strokes, thus enhancing regional CBF, stabilizing cerebral energy state, and reducing brain infarction. Increased brain capillary densities were noted within 10 days of initiation of VEGF treatment, which were accompanied by enhanced pericyte coverage of endothelialcells within 21 days. The induction of angiogenesis closely paralleled tissue survival, providing evidence of links among capillary formation, hemodynamic changes, and tissue preservation.
Our observation that induced angiogenesis increases blood flow in the ischemic brain contrasts with earlier findings of our group using a transgenic mouse line expressing human VEGF under control of a neuron-specific NSE promoter.6 In that mouse line, increased regional CBF, evaluated using 4-iodo-N-methyl-[14C] antipyrine, was described only in nonischemic brain areas outside the MCA territory, whereas regional CBF within the ischemic MCA territory was reduced.6 These data were interpreted as hemodynamic steal phenomenon induced by enhanced angiogenesis.6 Indeed, human VEGF-transgenic mice exhibited an increased capillary density throughout the brain, which might explain why blood flow was redirected into nonischemic areas. Although capillary density was increased by up to 100% or even more in human VEGF-transgenic mice, only mildly increased blood flow values were noted in this mouse line during hypercapnia.9 These findings raised doubts whether microvascular networks are adopted to tissue needs in this mouse line.
The capability of blood vessels to respond to tissue needs depends on proper interactions between endothelial cells with pericytes,20 which, as our data suggest, were enhanced by VEGF. Indeed, an increased pericyte coverage was noticed that was associated with enhanced N-cadherin expression. N-cadherin promotes the alignment of pericytes to endothelial cells.20 Our results are in contrast to observations after VEGF treatment in a model of PDGF-BB–induced angiogenesis, where a loss of pericyte coverage was reported, which was mediated by the deactivation of PDGFRβ by VEGFR2.10 We have not been able to detect VEGFR2 on brain pericytes. Differences of the growth factors used, combined VEGF/PDGF-BB delivery versus VEGF delivery only, together with differences in the experimental systems, Matrigel assay versus focal cerebral ischemia, may explain diverging results. We still do not know how VEGF increased pericyte coverage of endothelial cells. Because pericytes did not express VEGFR2 or VEGFR1 in our mouse model, indirect signals between endothelial cells and pericytes are likely to mediate pericyte proliferation. PDGF-B, which is produced by endothelial cells and acts through PDGFRβ on pericytes, and TGF-β, which is also produced by endothelial cells and activates the transcriptional regulator Smad2 that stimulates the transcription of N-cadherin on endothelial cells and pericytes, represent possible candidates.20,21
A major concern regarding VEGF-induced angiogenesis is the increased blood–brain barrier permeability that is more pronounced after acute than prophylactic or postacute VEGF delivery, and more pronounced after intravenous than local exposure.3 In our study, blood–brain barrier integrity remained intact after VEGF delivery. A conceptual problem of therapeutic angiogenesis remains that it is impossible to predict where future ischemic episodes are likely to occur.6 Angiogenesis induced in 1 specific brain area may potentially have detrimental effects on brain hemodynamics, if subsequent ischemias occur elsewhere in the brain. Indicating that therapeutic angiogenesis may have beneficial effects beyond elevating capillary density, we observed an enhanced pericyte coverage after VEGF treatment also in areas not exhibiting vascular sprouting. In the contralateral striatum, at distance to the VEGF infusion, increased CBF values were noticed, despite lack of capillary growth. The diameter of arterioles was slightly but significantly increased, which might explain this finding. We herein evaluated therapeutic angiogenesis in hitherto healthy adolescent mice with intact vascular networks. The question remains whether the data obtained can be translated to atherosclerotic mice. Future studies will have to address this issue.
We thank Beate Karow for technical assistance.
Sources of Funding
This work was supported by the German Research Foundation (HE3173/2-1 and HE3173/3-1; to Dr Hermann), Dr Werner-Jackstädt Foundation (Dr Zechariah), and Heinz-Nixdorf Foundation (Dr Hermann).
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.000240/-/DC1.
- Received October 25, 2012.
- Revision received February 20, 2013.
- Accepted March 12, 2013.
- © 2013 American Heart Association, Inc.
- Wang Y,
- Kilic E,
- Kilic U,
- Weber B,
- Bassetti CL,
- Marti HH,
- et al
- Kilic U,
- Kilic E,
- Järve A,
- Guo Z,
- Spudich A,
- Bieber K,
- et al
- Zechariah A,
- ElAli A,
- Hermann DM
- Helfrich I,
- Scheffrahn I,
- Bartling S,
- Weis J,
- von Felbert V,
- Middleton M,
- et al
- Hermann DM,
- ElAli A