Brain Research 883 (2000) 87–97www.elsevier.com/ locate /bres

Research report

Expression of vascular endothelial growth factor by reactive astrocytesand associated neoangiogenesis

a a,b a a cBodour Salhia , Lilyana Angelov , Luba Roncari , Xiaoli Wu , Patrick Shannon ,a,b ,*Abhijit Guha

aLabatts Brain Tumor Center, Hospital for Sick Children, Toronto, Ontario, CanadabDivision of Neurosurgery, University Health Network, University of Toronto, Western Division, Toronto, Ontario, Canada

cDivision of Neuropathology, University Health Network, University of Toronto, Toronto, Ontario, Canada

Accepted 8 August 2000

Abstract

Injury to the central nervous system (CNS) invokes a reparative response known as astrogliosis, characterized largely by hypertrophy,proliferation and increased expression of glial fibrillary acidic protein (GFAP), resulting in reactive astrocytosis. Based on our priorobservation that peritumoral reactive astrocytes express Vascular Endothelial Growth Factor (VEGF), a highly potent and specificangiogenic growth factor, we have hypothesized that reactive astrocytosis also contributes to the neovascularization associated withastrogliosis. To evaluate this hypothesis we evaluated human surgical /autopsy specimens from a variety of CNS disorders that induceastrogliosis and an experimental CNS needle injury model in wild type and GFAP:Green Fluorescent Protein (GFP) transgenic mice.Using computer image semi-quantitative analysis we evaluated the number of GFAP-positive reactive astrocytes, degree of VEGFexpression by these astrocytes, associated Factor VIII-positive microvascular density (MVD) and Ki-67 proliferating endothelial cells. Thedegree of reactive astrocytosis correlated to levels of VEGF immunoreactivity and MVD in the neuropathological specimens. Themouse–needle–stick brain injury model demonstrated this correlation was temporally and spatially related and maximal after 1 week.These results, involving both human pathology specimens augmented by experimental animal data, supports our hypothesis that theneoangiogenesis associated with reactive astrogliosis is correlated to increased reactive astrocytosis and associated VEGF expression. 2000 Elsevier Science B.V. All rights reserved.

Theme: Cellular and molecular biology

Topic: Neuroglia

Keywords: Astrogliosis; GFAP; Neoangiogenesis; Microvascular density; Reactive astrocytes; VEGF

1. Introduction cell processes and bodies with increased expression of theastrocyte specific cytoskeletal intermediate filament Glial

Reactive astrogliosis is a normal reparative mechanism Fibrillary Acidic Protein(GFAP) [7,11,14,35,41,50,64,65].induced in many CNS disease processes such as ischemia, This reactive astrocytic response has been noted to occurinfection, trauma, degenerative diseases, epilepsy and brain within 2 days after experimental brain injury in rodentstumors [29,31,49,50,63]. Amongst the various cellular and approximately 1 week in humans and may persist forresponses of reactive gliosis, there is a prominent as- several weeks with progressive decline over months totrocytic response [31,58] comprised of normally quiescent years [43,47,50,63]. The function of this florid reactiveastrocytes proliferating, undergoing hypertrophy of their astrocytosis is not fully deciphered, but is postulated to

play a role in the healing phase after insult to the CNS bymonitoring and controlling the molecular and ionic con-

*Corresponding author. Division of Neurosurgery, University Health tents of the CNS extracellular space. This is likelyNetwork, University of Toronto, Western Division, 399 Bathurst Street,

achieved via multiple membrane receptors for neurotrans-Toronto, Ontario, Canada M5T 2S8. Tel.: 11-416-603-5740; fax: 11-mitter peptides, growth factors and specific ion channels416-603-5298.

E-mail address: ab.guha@utoronto.ca (A. Guha). that are known to be expressed by normal and reactive

0006-8993/00/$ – see front matter 2000 Elsevier Science B.V. All rights reserved.PI I : S0006-8993( 00 )02825-0

88 B. Salhia et al. / Brain Research 883 (2000) 87 –97

astrocytes [31,50,63]. However, it is hypothesized that trocytosis, neoangiogenesis and VEGF expression from aeven moderate reactive astrocytosis may interfere with spectrum of conditions including Alzheimer’s disease,eventual reconnection of functional neuronal circuitry by brain abscess, peritumoral brain from metastatic car-inhibiting axonal regeneration, preventing remyelination or cinoma, head injury and cerebral infarcts. Further, topromoting abnormal neuronal connections with increased evaluate the temporal relationship of reactive astrocytosisseizure foci [57]. and neoangiogenesis, a stereotactic brain injury model was

There is currently little understanding of the molecular utilized. In order to obtain the proliferative index ([ Ki67mechanisms regulating reactive astrocytosis. Several ob- positive cells /1000 endothelial cells3100) of endothelialservations suggest that reactive astrocytosis is a regulated cells, surgical specimens from brain abscess and metastaticprocess involving cytokines, inflammatory mediators, carcinoma were stained for Ki-67. In addition, a CNSgrowth factors and physiological stimuli such as hypoxia injury model was utilized in transgenic mice which express[45,50,69]. For example, axonal growth and regeneration Green Fluorescent Protein (GFP) under the control of theafter trauma and cerebral infarction are retarded by inhib- astrocyte specific GFAP promoter [5], to co-localize VEGFitory proteins associated with the intense astrogliotic scar expression in reactive astrocytes.[6,45,50]. Second, reactive astrocytosis although mostintense immediately surrounding the area of injury, isfound at much removed distances suggestive of soluble 2. Methodstrophic factors acting in a paracrine fashion [50]. While avariety of these trophic factors have been studied in 2.1. Evaluation of human neuropathological diseasesastrocyte cell culture experiments, their role and interac-tions in the astrogliotic response in vivo is poorly under- In order to evaluate neoangiogenesis and VEGF expres-stood. Specifically, polypeptide growth factors such as sion in a variety of reactive astrogliotic specimens, 23basic Fibroblast Growth Factor (bFGF), Epidermal Growth cases of neurosurgical and autopsy specimens obtainedFactor (EGF), Transforming Growth Factor (TGF-a), from our neuropathology department were examined.Platelet Derived Growth Factor (PDGF), Cilliary Neuro- These cases represented a spectrum of benign and malig-trophic Factor (CNTF) and cytokines such as Interleukin-1 nant diseases that are known to induce reactive astrocytosis(IL1) and Tumor Necrosis Factor (TNF-a), have all been in the surrounding brain and included: four Alzheimer’simplicated in the astrogliotic response [3,13,32,37, Disease (AD), three bacterial abscesses, five carcinomas,49,50,59,68,69]. three head injuries and seven infarcts (two acute: ,7 days,

Part of any reparative tissue process is neoangiogenesis, two subacute: 7–30 days, three old: .30 days). Normalvividly demonstrated in the collagen fibrovascular response autopsy brain tissue was used as a control. Standardassociated with wound healing and one that can be hematoxylin and eosin (H&E) staining was performed onspeculated to be an important component of reactive 5-mm tissue sections from paraffin embedded tissueastrocytosis in the CNS [10]. Amongst the regulators of blocks, which were independently reviewed by a neuro-neoangiogenesis, Vascular Endothelial Growth Factor pathologist (PS). Sequential sections were incubated over-(VEGF) is highly potent and specific by activating its night at 48C with a primary rabbit antibody against GFAPcognate receptors on the endothelial cells resulting in their (1:3000, Dako Corp, CA) to mark reactive astrocytes, aproliferation, migration and sprouting [17,27,61]. VEGF human anti-Factor VIII antibody (1:2000, Dako Corp) tohas been implicated in most physiological and pathological determine the microvascular density (MVD) or a polyclon-processes involving neovascularization including embryo- al rabbit anti-VEGF antibody (1:50, Santa Cruz Biotech-genesis, wound healing, tumor growth, myocardial is- nology Inc., CA) that recognizes all four VEGF isoforms.chemia, ocular neovascular diseases and chronic diseases Detection was through the ABC (Vector)-DAB (VectaStainsuch as rheumatoid arthritis [16,21]. In addition to its Elite, CA) system. The sections were analyzed using theangiogenic effects, VEGF is also a potent edemogenic MicroComputer Image Device (MCID; Imaging Researchagent, resulting in disruption of the blood–brain-barrier Inc., Canada), linked to a color CCD camera (Sony DXC(BBB), and hence was also known as Vascular Permeabili- 970 MD) mounted on a transmitted-light microscopety Factor (VPF). We observed that VEGF/VPF was co- (Zeiss Axioskop) fitted with a Ludl Biopoint motorizedexpressed with GFAP-positive reactive astrocytes in the stage. Microscope fields (4003magnification) were ac-peritumoral edematous brain around meningiomas, where quired and digitized for each of the sequential sectionsits expression was correlated to the vascularity and associ- stained for GFAP, Factor VIII and VEGF, with 4–6 high-ated edema induced by these CNS tumors [56]. This powered fields (HPF) analyzed in each tissue section.suggested that reactive astrocytes may express VEGF, These HPFs were chosen to be adjacent to the site of tissuemodulating neoangiogenesis and also the edema and break necrosis by the underlying pathology (i.e, adjacent to thedown of the blood brain barrier associated with reactive abscess wall, carcinoma, infarct, contusion) or in theastrogliosis. To address this hypothesis, immunohisto- mesial temporal lobes of the AD specimens. All raw valueschemical techniques were used to quantify reactive as- were reported as a mean6standard error of the means

B. Salhia et al. / Brain Research 883 (2000) 87 –97 89

(S.E.M.) and also represented as a percentage of the endogenous GFAP was performed by incubating with amaximum value obtained. The extent of astrocytosis was rabbit anti-GFAP antibody (1:10, Dako) or mouse mono-assessed by averaging the number of GFAP-positive clonal anti-VEGF antibody (1:10, UBI) for 1 h at roomreactive astrocytes /4–6 high-powered fields (HPF). MVD temperature. After three washes in PBS containing 1%counts were derived by averaging the number of Factor BSA and 0.9% NaCl, the sections were incubated withVIII-positive vessels /4–6 HPF and VEGF staining was either Cy5-conjugated goat anti-mouse IgG (1:50, Jacksonquantified by averaging the percentage VEGF immuno- ImmunoResearch Laboratories Inc.) or Cy3-conjugatedreactivity /4–6 HPF. goat anti-rabbit IgG (1:20, Jackson ImmunoResearch

In three cases of brain abscess and two cases of Laboratories Inc.) for 1 h at room temperature. Sectionsmetastatic carcinoma, sections were stained for Ki-67 were washed in PBS and mounted for immunofluorescent(MIB-1, 1:50, Immunotech) after microwave antigen re- microscopy. Sections were evaluated with the greentrieval and the proliferative index of the endothelium was immunofluorescence filter to detect expression of the GFPobtained. The proliferative index was determined by transgene under the GFAP promoter and for either endog-counting 1000 endothelial cell nuclei and determining the enous GFAP (Cy3) or VEGF (Cy5).percentage staining positive for Ki-67.

2.2. Wild type and transgenic GFAP:GFP mouse 3. Resultsstereotactic needle brain injury

3.1. Reactive astrocytosis, VEGF immunoreactivity andTo determine the temporal association between reactive neoangiogenesis in human CNS diseases

astrocytosis, neoangiogenesis and VEGF expression, areproducible stereotactic needle stick injury model to the The mean age of patients with the different diseasesfrontal cerebral cortex of mice was used [42,47,50]. A total were 79 years — AD; 38 years — Abscess; 65 years —of 27 CD1 mice were anesthetized with 0.5 ml of 2.5% Peri-metastatic carcinoma brain; 52 years — Head Injury;Avertin (2,2,2-tribromoethanol and 2-methyl-2-butanol) 71 years — Cerebral infarcts. Reactive astrocytosis with(Sigma-Aldrich Chemical Company Inc., WI) by intra- increased numbers of large and darkly stained GFAP-peritoneal (i.p.) injection. The head was stabilized in a positive reactive astrocytes were a prominent feature of allstereotactic frame and a burr hole drilled 2 mm anterior to these disease processes compared to normal brain, al-the coronal suture and 2 mm lateral to the saggital suture. though the degree of reactive astrocytosis varied with theThe needle of a 5-ml Hamilton syringe was then inserted type of pathology, as demonstrated in Table 1, Fig. 1. Thethrough the burr hole to a depth of 2.5 mm for a duration most exuberant reactive astrocytosis was induced in theof 1 min, then withdrawn. The contra-lateral frontal lobe brain surrounding bacterial abscesses (Table 1, Fig. 1-Rowwas used as the non-injured control side in all evaluations. A), with an average raw value of 82628 GFAP-positiveThe track was made in cortical gray matter, not penetrating astrocytes /HPF. In comparison, diffuse head injury (Fig.sub-cortical areas that normally contain higher numbers of 1-Row C), induced the lowest amount of reactive as-GFAP-positive cells [8]. All mice tolerated the procedure trocytosis, with a raw value 2067 GFAP-positive as-well. Three mice were sacrificed on each of days 1–9 trocytes /HPF, though this result should be tempered due topost-injury by cervical dislocation and the brains fixed in our lack of knowledge of severity or the time of theformalin for 24 h and embedded in paraffin for histological autopsy after the head injury. The amount of VEGFprocessing. Axial, 6-mm thick paraffin sections at right immunoreactivity paralleled the degree of induced reactiveangles to the needle tract were cut, stained with anti-GFAP, astrocytosis, except for cerebral infarcts and peri-tumoralanti-Factor VIII and anti-VEGF antibodies and analyzed brain, where VEGF immunoreactivity was disproportion-with the computer assisted image analysis system as ately higher (Table 1). The number of Factor VIII-positiveabove. blood vessels (MVD), as a measure of neoangiogenesis,

Three FVB/N background GFAP:GFP transgenic mice was related to the degree of reactive astrocytosis and(a gift from Dr. A. Messing, Madison, WI, USA) under- VEGF immunoreactivity. This is best exemplified bywent a similar needle tract injury and were evaluated 7 MVD, raw counts of 2063 vessels /HPF, associated withdays after injury with immunofluorescence microscopy. the florid reactive astrocytosis in bacterial abscess, com-Mice were sacrificed by cervical dislocation. Brains were pared to only 661 vessels /HPF in brains with prior headremoved within 5 min of sacrifice, embedded in OCT injury, AD or infarcts. In these latter conditions, the degree(Tissue Tek) embedding medium on dry ice and stored at of reactive astrocytosis and associated neoangiogenesis2708C until further processing. Axial cryostat sections although relatively low compared to brain around bacterialwere fixed in ice cold acetone for 3 min at 2208C, washed abscess, was still significantly high relative to normal brainwith PBS for 10 min and blocked in a solution of PBS (Table 1). In all cases, astrocytosis,VEGF immunoreactivi-containing 1% BSA and 0.9% NaCl for 30 min at room ty and MVD was increased from normal brain (Fig. 1-Rowtemperature. Double immunofluourescence for VEGF and F). In addition, blood vessels associated with brain abscess

90 B. Salhia et al. / Brain Research 883 (2000) 87 –97

Table 1aSemi-quantitative analysis of immunohistochemical staining in a variety of human neuropathologies

Diagnosis n Reactive VEGF VascularityAstrocytosis Expression (No. Factor(No. GFAP- (VEGF VIII-stainedPositive immunoreactivity vessels /HPF)Cells /HPF) /HPF)

Normal 1 261 (2) 461 (9) 261 (10)Alzheimer’s disease 4 3962 (48) 2564 (54) 661 (30)Bacterial abscess 3 82628 (100) 661 (100) 2063 (100)Head injury 3 2067 (24) 1364 (28) 661 (30)Infarct 6 2263 (27) 2363 (50) 661 (30)Peritumoral brain 4 2861 (34) 3061 (65) 1163 (55)a Reactive astrocytosis, VEGF expression and vascularity in a spectrum of human neuropathological specimens. Raw mean values6S.E.M. are shown andvalues are reported as a percentage of the maximum in brackets. All slides were quantified at 4003magnification. VEGF, vascular endothelial growthfactor; S.E.M., standard error of the mean; GFAP, glial fibrillary acidic protein; HPF, High Power Field.

appeared larger and more robust than other CNS diseases parameters subsequently decreased towards normal levelsinvestigated and the blood vessels associated with head by day 9. While the peak days for each parameter analyzedinjury seem much smaller compared to other conditions. In do not occur on the same day in this experimentalthe astrogliotic response associated with AD and cerebral paradigm, the overall temporal correlation between them isinfarcts, the vascularity, although increased from normal clearly demonstrated. The GFAP:GFP transgenic micebrain, was relatively less compared to the degree of VEGF were used for co-localization experiments in a needle stickimmunoreactivity by reactive astrocytes. The proliferative injury model to confirm that the reactive astrocytes in-index of endothelial cells ranged from 8 to 12% in brain duced by the needle tract injury (* in Fig. 4) wereabscesses (n53) and from 2 to 6% (n52) near metastatic expressing VEGF. Green immunofluorescence denoted thetumor. These results clearly demonstrate active neoan- expression of the GFP transgene under regulation of thegiogenesis in these two conditions. Material taken from GFAP promoter (Fig. 4A). These same GFP expressingautopsy specimens was not used to determine proliferative cells were reactive astrocytes, as they also expressedindex because, in our experience, detection of this antigen endogenous GFAP as detected by double immunofluores-is not reliable in tissue undergoing prolonged fixation. cence, where the cells are yellow due to the combination of

green and the Cy3 secondary antibody used to detect3.2. Temporal and spatial expression of reactive endogenous GFAP (Fig. 4B). On adjacent sections, theastrocytosis, VEGF immunoreactivity and GFP reactive astrocytes are the same cells that are express-neoangiogenesis ing VEGF, where they appear red due to combination of

green and Cy5 immunofluorescence used to detect VEGFThe mouse stereotactic needle stick injury resulted in a (Fig. 4C). There were no significant reactive GFP-positive

distinct tract that could be consistently visualized on the astrocytes in the non-injured contra-lateral hemisphereinjured side with a surrounding reactive astrogliotic re- (Fig. 4D).sponse (Fig. 2A). The injury induced GFAP-positiveastrocytes (Fig. 2B) to express VEGF (Fig. 2C) withaccompanying neoangiogenesis, as marked by Factor VIII- 4. Discussionpositive endothelial cells (Fig. 2D). The number of GFAP-positive reactive astrocytes reached a maximum on day-7 Astrogliosis and the formation of the glial scar, with apost-injury, which was taken as 100% for comparison with predominant reactive astrocytosis component, is an im-the other time points analyzed (Figs. 2, 3).VEGF immuno- portant and widespread neuropathological response toreactivity increased to its maximum level (100%) on day 6 diverse neurological insults, a feature that is well con-post-injury, when the reactive astrocytic response had served across different species. This prominent, rapid andreached 47614% of maximal day-7 levels (Fig. 3). The evolutionarily conserved reparative response, suggests thatnumber of Factor VIII-positive blood vessels (MVD) reactive astrocytosis fulfills an important function in thereached a maximal level (100%) on day 5 post-injury, CNS. However, the exact role of reactive astrocytosis andwhen the degree of reactive astrocytosis was 31614% and whether it is beneficial or ultimately detrimental to CNSVEGF immunoreactivity was 5967% of maximal levels recovery after injury remains unclear and controversial.on day 7 and day 6 respectively (Figs. 2, 3). Therefore, Reactive astrogliosis is a dynamic process that leads to athere was a progressive increase in reactive astrocytosis, densely interwoven glial scar composed of many cellsVEGF immunoreactivity and neoangiogenesis after injury, including reactive astrocytes, microglia, macrophages andwith maximal levels clustering around days 5–7. All three endothelial cells [12,24,26,31,53]. Reactive astrocytes

B. Salhia et al. / Brain Research 883 (2000) 87 –97 91

Fig. 1. Astrocytosis and neoangiogenesis in human CNS disease. Photomicrographs (scale bar depicts 50 mm in length) illustrating immunohistochemicalstaining of a spectrum of human neuropathological disease states with: Row A — Brain Abscess; Row B — Alzheimer’s Disease; Row C — Old Infarct;Row D — Peritumoral Brain; Row E — Head Injury; Row F — Normal Brain; Column 1 — anti-glial fibrillary acidic protein (GFAP) antibody; Column 2— anti-vascular endothelial growth factor (VEGF) antibody; Column 3 — anti-Factor VIII antibody. When compared to the staining observed in normalbrain, GFAP reactive astrocytes and neoangiogenesis as indicated by VEGF immunoreactivity and Factor VIII-stained vessels, is common and increased inall of the human pathologies (reddish /brown staining). GFAP positivity is noted in the reactive astrocytes and astrocytic foot processes around bloodvessels. The most intense staining for all three parameters is observed in the tissue taken from the brain surrounding the abscess.

92 B. Salhia et al. / Brain Research 883 (2000) 87 –97

Fig. 2. Visualization and characterization of mouse–needle–stick injury. Immunohistochemical sections of mouse brain stained for GFAP, VEGF andFactor VIII. (A) Coronal section through the injury site (scale bar depicts 2 mm in length) displaying GFAP immunoreactivity (brown staining) in reactiveastrocytes straddling the tract site (arrow). The contralateral non-injured site (N) is GFAP negative in the same cortical region. Mouse brain sections athigher power magnification at the site of the needle stick injury (day 7 post-injury) (scale bar depicts 50 mm in length) stained with (B) anti-glial fibrillaryacidic protein (GFAP) antibody; (C) anti-vascular endothelial growth factor (VEGF) antibody; (D) anti-Factor VIII antibody. Day 7 post-injury is themaximum GFAP response, also shown in Fig. 3. The injury tract (*) can be seen in the upper left corner of each section. Arrows in B and C are pointing topositively stained reactive astrocytes.

form scars by extending their long, thick cytoplasmic neoangiogenesis (MVD) and VEGF immunoreactivity in aprocesses toward the site of the lesion [29]. It is likely that variety of human CNS pathologies (see Table 1, Fig. 1).the glial scar can serve to separate normal brain from the Reactive astrocytosis, neoangiogenesis and increasedsurrounding damaged tissue, but its formation may also be VEGF was maximal in response to bacterial brain abscessharmful by essentially creating a mechanical barrier [50] and probably reflects the acuteness of the illness and thewhich may impede neuronal dendritic growth, remodeling type of response that it induces in the CNS. In an abscess,and axonal regeneration [25,29,50,63,71]. In addition, the in response to the released toxins, there is an inflammatoryglial scar may contribute to electrical instability in the cellular response, components of which themselves releaseregion and in turn promote seizure activity [50]. However, VEGF and other inflammatory cytokines modulators. Thismore recent evidence suggests that astrogliosis may actual- leads not only to a vigorous reactive astrogliotic responsely facilitate CNS recovery through neurotrophic factor but it is also one of the few CNS diseases that induces aproduction around the area of the lesion [2,34,67]. In collagen fibrovascular response.summary, like the common collagen fibrovascular wound The amount of VEGF immunoreactivity and MVDhealing process, reactive astrogliosis serves a beneficial exceeded the degree of reactive astrocytosis in the brainCNS reparative function, however, under certain circum- around carcinomas (Table 1), perhaps attributable to thestances it may be detrimental much like exuberant scar and additive expression of VEGF from the GFAP positivekeloid formation in other tissues. reactive astrocytes and the carcinoma cells themselves

This study demonstrates, using human specimens and [22]. In cerebral infarcts, irrespective of whether it wasexperimental models of CNS injury, that reactive as- acute, subacute or old, VEGF immunoreactivity was highertrocytosis is accompanied by a neoangiogenic response, than the degree of reactive astrocytosis or neoangiogenicsimilar to the collagen fibrovascular response elsewhere. response (see Table 1). This may reflect the potent hypoxicFurthermore, we speculate based on our results that a induction of VEGF expression by the surviving reactivepotential source of this angiogenic response lies in in- astrocytes [15], without a concomitant increase in thecreased VEGF secretion by the quiescent reactive as- number of astrocytes or the ability to recruit new bloodtrocytes. This conclusion is based on the overall consistent vessels due to the cerebral ischemia. This data is similar torelationship between GFAP-positive reactive astrocytes, results in a study by Plate et al., where in a middle cerebral

B. Salhia et al. / Brain Research 883 (2000) 87 –97 93

Fig. 3. Temporal association of astrocytosis, VEGF immunoreactivity and neoangiogenesis in the needle stick injury model. Histogram (depictingpercentage of the maximum mean6S.E.M.) demonstrating the temporal association in the expression patterns of GFAP, VEGF and MVD followingmouse–needle–stick injury. The maximal number of GFAP-positive astrocytes, VEGF expression and neoangiogenesis was temporally correlated andoccurred between days 5–7 post-injury supporting a relationship in the time course of the astrogliotic and neoangiogenic responses.

artery occlusion in rats, VEGF mRNA expression was positive reactive astrocytes. Our results confirm previousconsistently greater across different time points than the findings that reactive astrocytes after a stab wound injurynumber of CD31 stained vessels [54]. In neurodegenera- are VEGF positive [46,52] and also concur with thetive CNS diseases like AD, characterized by b-amyloid observation that VEGF mRNA was overexpressed indeposits, neuronal death is accompanied by proliferation of retinal astrocytes of diabetic rats [46].astrocytes, which react in an attempt to detoxify neuronal Although our study focuses on increased expression ofdebris, toxins and produce various trophic substances VEGF in reactive astrocytes, it is evident that there are[33,63]. Furthermore, decreased cerebral perfusion and several cell types capable of producing VEGF, that are alsoglucose metabolism found in AD may also induce the prevalent in reactive astrocytosis. These include, platelets,reactive astrocytes to up-regulate VEGF in an attempt to neurons, macrophages /microglia, smooth muscles aroundincrease neoangiogenesis (Fig. 1) [33,63]. vessel walls, megakaryocytes and polymorphonuclear

The overall results of the mouse–needle–stick injury leukocytes [36,44,46,52,54,66,72]. There is some con-model are supportive of a temporal correlation between troversy as to which of these cell types are the pre-reactive astrocytosis, neoangiogenesis and VEGF expres- dominant expressors of VEGF following damage to thesion. All three reached maximal levels between days 5–7, CNS. Plate et al. identified cells of the microglial /macro-with subsequent return towards basal levels by day 9 (Fig. phage lineage to be the major source of VEGF upregula-3). What is not obvious is why the MVD peaked first (day tion following a middle cerebral artery occlusion (MCAO)5), followed by VEGF (day 6) and then GFAP-positive and reported virtually no VEGF production in neurons orreactive astrocytes (day 7) (Fig. 3). Perhaps the initial reactive astrocytes [54]. Lennmyr et al. demonstrated thewave of neoangiogenesis on day 5 reflects the response to most striking changes in VEGF production followinginflammatory cells induced by the injury in these immuno- MCAO to be in neurons, with only weak-to-moderatecompetent CD1 mice, with release of other angiogenic VEGF immunoreactivity in monocytes and astrocytes [36].stimulants in addition to VEGF [52]. The GFAP:GFP The reasons for these differences are unclear but may betransgenic mice subjected to a similar needle stick injury, attributed to differences in methods and use of reagents.clearly demonstrated co-labeling of VEGF to the GFAP Interestingly however, using the same reagents and con-

94 B. Salhia et al. / Brain Research 883 (2000) 87 –97

Fig. 4. Co-localization of VEGF in reactive astrocytes of GFAP:GFP transgenic mice. Photomicrograph (scale bar depicts 20 mm in length) demonstratingthe co-localization of reactive astrocytes and VEGF expression in GFAP:GFP transgenic mouse brains post-stereotactic needle stick injury. (*) denotes thearea of the tract injury. (A) Endogenous reactive astrocytes (arrows) emit intense green fluorescence. Typical astrocyte morphology with cell bodies andnumerous processes can be seen. (B) Double labeling on the same section with anti-GFAP antibody and endogenous GFP immunofluorescence. Thereactive astrocytes are stained yellow (arrows) due to the combination of the GFP green fluorescence and the anti-GFAP antibody detected via aCy3-conjugated secondary antibody. (C) Double labeling on an adjacent section showing the co-localization of GFP-positive reactive astrocytes and VEGFprotein expression as demonstrated by the red immunofluorescence staining (arrows). (D) The non-injured contra-lateral hemisphere of the brain in thesame area demonstrated no significant GFP-positive green astrocytes.

ditions, we were able to detect increased VEGF immuno- angiogenic phenotype is activated when the normal bal-reactivity but unable to co-localize VEGF to reactive ance that exists between inducers and inhibitors of angio-astroctyes in GFAP:GFP transgenic and CD1 mice with a genesis is disrupted [4,21,51]. This induction results inMCAO (data not shown). This data suggests that there may angiogenesis through chemotactic and mitogenic effects onbe differences in VEGF immunolocalization in the stab endothelial cells which line blood vessels [21]. A numberinjury and focal ischemia models [46]. Further evidence of peptides are known to be angiogenic in vivo [23].indicative of differences in VEGF immunolocalization in However, VEGF, a homodimeric glycoprotein is a mitogencerebral infarct is seen in our human data, where despite that specifically acts on endothelial cells and is postulatedincreased levels of VEGF immunoreactivity, the correla- to be the main positive regulator of angiogenesis in vivotion to reactive astrocytosis was not as strong as was the [55,60,70]. There are many regulators of VEGF, withcorrelation in other neuropathologies including head injury hypoxia being the most potent physiological inducer ofand brain abscess. However, this is only speculation and VEGF [19,20,28,38,62]. The factors responsible for in-requires further studies with the appropriate controls. creased VEGF production by astrocytes after damage to

Angiogenesis is the sprouting of capillaries from pre- the CNS are unknown. However, hypoxia may be theexisting blood vessels and during embryonic development common regulator of VEGF expression by reactive as-is a fundamental process in the formation of the vascular trocytes in response to a variety of CNS insults that createsystem. In the adult, angiogenesis plays a crucial role in a relative hypoxic gradient in the surrounding brain.normal and pathological processes such as wound healing, Whether modulation of angiogenic factors such as VEGFdiabetic retinopathy and tumorigenesis [1,10,23]. The would favor repair in various non-neoplastic CNS con-

B. Salhia et al. / Brain Research 883 (2000) 87 –97 95

[3] F. Besnard, F. Perraud, M. Sensenbrenner, G. Labourdette, Platelet-ditions is uncertain. For example, experiments whichderived growth factor is a mitogen for glial but not for neuronal ratmodulate VEGF in the zone of cerebral infarction as tobrain cells in vitro, Neurosci. Lett. 73 (1987) 287–292.

rescue the penumbra by neoangiogenesis [33], similar to [4] N. Bouk, Tumour Angiogenesis: the role of oncogenes and tumourexperiments with VEGF gene therapy of ischemic car- suppressor genes, Cancer Cells 6 (6) (1990) 179–185.diomyocytes [39], are under current study in our labora- [5] M. Brenner, A. Messing, GFAP transgenic mice, Methods 10 (1996)

351–364.tory.[6] J.A. Brodkey, E.D. Laywell, T.F. O’Brien, A. Faissner, K. Stefan-This study has demonstrated using several experimental

sson, U. Derries, M. Schachner, D.A. Steindler, Focal brain injuryand neuropathological models that reactive astrocytosis isand upregulation of a developmentally regulated extracellular matrix

accompanied by neoangiogenesis, part of which we can protein, J. Neurosurg. 82 (1995) 106–112.postulate is based on VEGF expression by the reactive [7] J.L. Calvo, A.L. Carbonell, J. Boya, Co-expression of glial fibrillary

acidic protein and vimentin in reactive astrocytes following brainastrocytes. The process of reactive gliosis involves manyinjury in rats, Brain Res. 566 (1991) 333–336.different cell types (with reactive astrocytes being a major

[8] P. Carmeliet, V. Ferrara, G. Breirer, S. Pollefeyt, L. Kieckens, M.component), several of which are capable of secretingGertsenstein, M. Fahrig, A. Vandenhoeck, K. Harpal, C. Eberhardt,

VEGF. Therefore, the functional importance of reactive Abnormal blood vessel development and lethality in embryosastrocytes and VEGF towards the neoangiogenesis accom- lacking a single VEGF allele, Nature 380 (1996) 435–439.

[9] R.A. Clark, Wound Repair, Curr. Opin. Cell Biol. 1 (1989) 1000–panying reactive gliosis requires further experimentation.1008.Presently hypomorph (under-expressing) and hypermorph

[10] E. Connolly Jr., C. Winfree, D. Stern, R. Solomon, D. Pinsky,(over-expressing) VEGF transgenic animals are being usedProcedural and strain-related variables significantly affect outcome

to address this issue, since both the homozygous and in a murine model of focal cerebral ischemia, Neurosurgery 38heterozygous VEGF knockouts are embryonically lethal (1996) 523–532.[9,17]. Another ongoing strategy utilizes small molecule [11] A. da Cunha, J.J. Jefferson, W.R. Tyor, J.D. Glass, F.S. Jannotta, L.

Vitkovic, Gliosis in human brain: relationship to size but not otherinhibitors of the VEGF receptors expressed by the endo-properties of astrocytes, Brain Res. 600 (1993) 161–165.thelial cells to inhibit the VEGF mediated neoangiogenic

[12] S. David, C. Bouchard, O. Tsatas, N. Giftochristos, Macrophagesresponse in reactive astrocytosis. These and other potential can modify the nonpermissive nature of the adult mammalian centralstrategies will help to further our molecular understanding nervous system, Neuron 5 (1990) 463–469.of reactive astrocytosis and associated neoangiogenesis and [13] F. Eclancher, F. Perraud, J. Faltin, G. Labourdette, M. Sensenbren-

ner, Reactive astrogliosis after basic fibroblast growth factor (bFGF)may provide therapeutic reagents to modulate this processinjection in injured neonatal rat brain, Glia 3 (1990) 502–509.favorably where indicated in the future.

[14] L.F. Eng, A. Yu, Y.L. Lee, Astrocytic response to injury, ProgressBrain Res. 94 (1992) 353–365.

[15] M.M. Feldkamp, N. Lau, J. Rak, R.S. Kerbel, A. Guha, Normoxicand hypoxic regulation of vascular endothelial growth factorAcknowledgements(VEGF) by astrocytoma cells is mediated by ras, Int. J. Cancer 81(1999) 118–124.

The authors would like to thank Dr. A. Messing [16] N. Ferrara, Vascular endothelial growth factor, Eur. J. Cancer 32A(Madison, Wisconsin) for the GFAP:GFP mice. We wish (1996) 2413–2422.

[17] N. Ferrara, K. Carver-Moore, H. Chen, M. Dowd, L. Lu, K.S.also to thank Ms. T. Nicklee (Toronto, ON) for herO’Shea, L. Powell-Braxton, K.J. Hillan, M.W. Moore, Heterozygousassistance with the image analysis system. LA was sup-embryonic lethality induced by targeted inactivation of the VEGFported by a National Cancer Institute of Canada and angene, Nature 380 (1996) 439–442.

American Association of Neurological Surgeons Research [19] G. Finkenzeller, A. Technau, D. Marme, Hypoxia induced transcrip-Fellowship. BS was supported by the Lunenfeld Research tion of the vascular endothelial growth factor gene is independent of

functional AP-1 transcription factor, Biochem. Biophys. Res. Com-Institute summer studentship. AG is a Medical Researchmun. 208 (1995) 432–439.Council of Canada Clinician Scientist and this work was

[20] J. Folkman, Toward an understanding of angiogenesis: search andsupported by operating grants to AG from Physiciandiscovery, Perspect. Bio. Med. 29 (1) (1985) 10–37.

Scientist Inc. and Heart & Stroke Fund of Ontario. [21] J. Folkman, Angiogenesis in cancer, vascular, rheumatoid and otherdiseases, Nature Med. 1 (1995) 27–31.

[22] J. Folkman, M. Klagsbrun, Angiogenic factors, Science 235 (1987)442–447.

References [23] J. Folkman, Y. Shing, Angiogenesis, J. Biol. Chem. 267 (1992)10931–10934.

[1] L.P. Aiello, R.L. Avery, P.G. Arrigg, B.A. Keyt, H.D. Jampel, S.T. [24] A. Fontana, W. Fierz, The endothelium–astrocyte immune controlShah, L.R. Pasquale, H.T. Mami, A. Iwamoto, E.J. Park, Vascular system of the brain, Immunopath. 5 (1985) 57–70.endothelial growth factor in ocular fluid of patients with diabetic [25] R. Ghirnikar, A. Yu, L.F. Eng, Astrogliosis in culture: III. Effect ofretinopathy and other retinal disorders, New England J. Med. 331 recombinant retrovirus expressing antisense glial fibrillary acidic(1994) 1480–1520. protein RNA, J. Neurosci. Res. 38 (1994) 376–385.

[2] T. Arendt, M.K. Bruckner, T. Krell, S. Pagliusi, L. Kruska, R. [26] D. Giulian, J. Chen, J.E. Ingeman, J. George, M. Noponen, The roleHeumann, Degeneration of rat cholinergic basal forebrain neurons of mononuclear phagocytes in wound healing after traumatic injuryand reactive changes in nerve growth factor expression after chronic to adult mammalian brain, J. Neurosci. 9 (1989) 4416–4429.neurotoxic injury. II. Reactive expression of NGF in astrocytes, [27] M.A. Goldberg, T.J. Schneider, Similarities between the oxygen-Neurosci. 65 (1995) 647–659. sensing mechanisms regulating the expression of vascular endotheli-

96 B. Salhia et al. / Brain Research 883 (2000) 87 –97

al growth factor and erythropoietin, J. Biol. Chem. 269 (1994) [50] M.D. Norenberg, Astrocyte response to CNS injury, J. Neuropathol.4355–4359. Exp. Neurol. 53 (1994) 213–220.

[28] D. Gospodarwicz, J.A. Abraham, J. Schilling, Isolation and charac- [51] K. Norrby, Angiogenesis: new aspects relating to its initiation andterization of a vascular endothelial cell mitogen produced by control, APMIS 105 (1997) 417–437.pituitary-derived follicle stellate cells, Proc. Natl. Acad. Sci. USA [52] E. Papavassiliou, N. Gogate, M. Proescholdt, J.D. Heiss, S. Wal-86 (1989) 7311–7315. bridge, N.A. Edwards, E.H. Oldfield, M. Merrill, Vascular endotheli-

[29] V. Guenard, G. Frisch, P. Wood, Effects of axonal injury on astrocyte al growth factor (vascular permeability factor) expression in injuredproliferation and morphology in vitro: implications for astrogliosis, rat brain, J. Neurosci. Res. 49 (1997) 451–460.Exp. Neurol. 137 (1996) 175–190. [53] V.H. Perry, M.C. Brown, S. Gordon, The macrophage response to

[31] M.E. Hatten, R.K.H. Liem, M.L. Shelanski, C.A. Mason, Astroglia central and peripheral nerve injury, J. Exp. Med. 165 (1987)in CNS injury, Glia 4 (1991) 233–243. 1218–1223.

[32] M.A. Kahn, J.A. Ellison, G.J. Speight, J. de Vellis, CNTF regulation [54] K.H. Plate, H. Beck, S. Danner, P.R. Allegrini, C. Wiessner, Cellof astrogliosis and the activation of microglia in the developing rat type specific upregulation of vascular endothelial growth factor in ancentral nervous system, Brain Res. 685 (1995) 55–67. MCA-occlusion model of cerebral infarct, J. Neuropathol. Exp.

[33] R. Kalaria, D. Cohen, D. Premkumar, S. Nag, J. LaManna, W. Lust,Neurol. 58 (6) (1999) 654–666.

Vascular endothelial growth factor in Alzheimer’s disease and[55] K.H. Plate, G. Breirer, H.A. Weich, W. Risau, Vascular Endothelial

experimental cerebral ischemia, Mol. Brain Res. 62 (1998) 101–Growth Factor is a potential tumor angiogenesis factor in human

105.gliomas in vivo, Nature 359 (1992) 845–848.[34] S. Komoly, L.D. Hudson, H. de Webster, C.A. Bondy, Insulin-like

[56] J. Provius, K. Claffey, L. delAguila, N. Lau, M. Feldkamp, A. Guha,growth factor I gene expression is induced in astrocytes duringMeningiomas: the role of vascular endothelial growth factor /Vascu-experimental demyelination, Proc. Natl. Acad. Sci. USA 89 (1992)lar permeability factor (VEGF/VPF) in angiogenesis and peri-1894–1898.tumoral edema, Neurosurgery 40 (5) (1997) 1016–1026.[35] N. Latov, G. Nilaver, E.A. Zimmerman, W.G. Johnson, A.J. Silver-

[57] P.J. Reier, S. Fedooroff, A. Vernadakis, in: Gliosis Following CNSmanm, R. Defendini, L. Cote, Fibrillary astrocytes proliferate inInjury: The Anatomy of Astrocytic Scars and their Influences onresponse to brain injury. A study combining immunoperoxidaseAxonal Elongation, Vol. 3, Academic Press, Orlando, Florida, 1986,technique for glial fibrillary acidic protein and radioautography ofpp. 263–324.tritiated thymidine, Devel. Biol. 72 (1979) 381–384.

[58] The glial scar: Its bearing on axonal elongation and transplantation[36] F. Lennmyr, K. Ahmad, K. Funa, Y. Olsson, A. Terent, Expressionapproaches to CNS repair, in: P.J. Reier, J.D. Houle, S.G. Waxmanof vascular endothelial growth factor (VEGF) and its receptors(Eds.), Functional Recovery in Neurological Disease, Advances in(Flt-1 and Flk-1) following permanent and transient occlusion of theNeurology, Vol. 47, Raven Press, New York, 1988, pp. 87–137.middle cerebral artery in the rat, J. Neuropathol. Exp. Neurol. 57 (9)

[59] M. Rostworowski, V. Balasingam, S. Chabot, T. Owens, V.W. Yong,(1997) 874–882.[37] S. Levison, M. Ducceschi, G. Young, T. Wood, Acute exposure to Astrogliosis in the neonatal and adult murine brain post-trauma:

CNTF in vivo induces multiple components of reactive gliosis, Exp. elevation of inflammatory cytokines and the lack of requirement forNeurol. 141 (1996) 256–268. endogenous interferon-gamma, J. Neurosci. 17 (10) (1997) 3664–

[38] A.P. Levy, N.S. Levy, M.A. Goldberg, Post-transcriptional regula- 3674.tion of vascular endothelial growth factor by hypoxia, J. Biol. Chem. [60] K. Samoto, K. Ikezaki, M. Ono, T. Shono, K. Kimitoshi, M.271 (1996) 2746–2753. Kuwano, M. Fukui, Expression of vascular endothelial growth factor

[39] D.W. Losordo, P.R. Vale, J.F. Symes, Gene therapy for myocardial and its possible relation with neovascularization in human brainangiogenesis: Initial clinical results with direct myocardial injection tumors, Cancer Res. 55 (1995) 1189–1193.of phVEGF as a sole therapy for myocardial ischemia, Circulation [61] D.R. Senger, S.J. Galli, A.M. Dvorak, C.A. Perruzzi, V.S. Harvey,165

98 (25) (1998) 2500–2504. H.F. Dvorak, Tumor cells secrete a Vascular Permeability Factor that[41] A.J. Mathewson, M. Berry, Observations on the astrocyte response promotes accumulation of ascites fluid, Science 219 (1983) 983–

to a cerebral wound in adult rats, Brain Res. 327 (1985) 61–69. 985.[42] T. Miyake, T. Hattori, M. Fukuda, T. Kitamura, S. Fujita, Quantita- [62] D. Shweiki, A. Itin, D. Soffer, E. Keshet, Vascular Endothelial

tive studies on proliferative changes of reactive astrocytes in mouse Growth Factor induced by hypoxia may mediate hypoxia-initiatedcerebral cortex, Brain Res. 451 (1988) 133–138. angiogenesis, Nature 359 (1992) 843–845.

[43] T. Miyake, M. Okada, T. Kitamura, Reactive proliferation of [63] M. Tacconi, Neuronal Death: Is there a role for astrocytes?,astrocytes studies by immunohistochemistry for proliferating cell Neurochem. Res. 23 (5) (1998) 759–765.nuclear antigen, Brain Res. 590 (1992) 3300–3302. [64] Y. Takamiya, S. Kohsaka, S. Toya, M. Otani, Y. Tsukada, Immuno-

[44] R. Mohle, D. Green, M.A. Moore, R.L. Nachman, S. Rafii, histochemical studies on the proliferation of reactive astrocytes andConstitutive production and thrombin-induced release of vascular the expression of cytoskeletal proteins following brain injury in rats,endothelial growth factor by human megakaryocytes and platelets, Devel. Brain Res. 38 (1988) 201–210.Proc. Natl. Acad. Sci. USA 94 (1997) 663–668. [65] V.K. Vijayan, Y.L. Lee, L.F. Eng, Increase in glial fibrillary acidic

[45] M.K. McMillian, L. Thai, J.S. Hong, J.P. O’Callaghan, K.R. protein following neural trauma, Neuropathology 13 (1990) 111–Pennypacker, Brain injury in a dish: a model for reactive gliosis, 122.Trends Neurosci. 17 (4) (1994) 138–142. [66] U. Wartiovaara, P. Salven, H. Mikkola, R. Lassila, J. Kaukonen, V.

[46] S. Nag, J.L. Takahashi, D. Kilty, Role of vascular endothelial Joukov, A. Orpana, M. Heikinheimo, H. Joensuu, K. Alitalo, A.growth factor in blood brain barrier breakdown and angiogenesis in Palotie, Peripheral blood platelets express VEGF-C and VEGFbrain trauma, J. Neuropathol. Exp. Neurol. 56 (8) (1997) 912–921. which are released during platelet activation, Thromb. Haemost. 80

[47] M. Nieto-Sampedro, R.P. Saneto, J. de Vellis, C.W. Cotman, The (1998) 171–175.control of glial populations in brain: changes in astrocyte mitogenic [67] R. Wen, Y. Song, T. Cheng, M.T. Matthes, D. Yasumura, M.M.and morphogenic factors in response to injury, Brain Res. 343 Lavall, R.H. Steinberg, Injury-induced upregulation of bFGF and(1985) 320–328. CNTF mRNAs in the rat retina, J. Neurosci. 15 (1995) 7377–7385.

[49] M. Noble, K. Murray, P. Stroobant, M.D. Waterfield, P. Riddle, [68] C. Winter, Y. Saotomis, S. Levison, D. Hirsh, The role of cillaryPlatelet-derived growth factor promotes division and motility and neurotrophic factor as an inducer of reactive gliosis, the glialinhibits premature differentiation of the oligodendrocyte / type-1 response to central nervous system injury, Proc. Natl. Acad. Sci.astrocyte progenitor cell, Nature 333 (1988) 560–562. USA 92 (1995) 5865–5869.

B. Salhia et al. / Brain Research 883 (2000) 87 –97 97

[69] V.W. Yong, R. Moumdjian, F.P. Yong, T.C. Ruijs, M.S. Freedman, N. [71] A. Yu, Y.L. Lee, L.F. Eng, Astrogliosis in culture: I. The model andCashman, J.P. Antel, Gamma interferon promotes proliferation of the effect of antisense oligodendrocytes on glial fibrillary acidicadult human astrocytes in vitro and reactive gliosis in the adult protein synthesis, J. Neurosci. Res. 34 (1993) 295–303.mouse brain in vivo, Proc. Natl. Acad. Sci. USA 88 (1991) 7016– [72] M. Xiong, G. Elson, D. Legarda, S.J. Leibovich, Production of7020. vascular endothelial growth factor by murine macrophages, Am. J.

[70] H. Yoshiji, D.E. Gomez, M. Shibuya, U.P. Thorgeirsson, Expression Pathol. 153 (2) (1998) 587–598.of vascular endothelial growth factor. Its receptor and other an-giogenic factors in breast cancer, Cancer Res. 56 (1996) 2013–2016.