Paracrine and Autocrine Functions of Brain-derived Neurotrophic Factor (BDNF) and Nerve Growth Factor (NGF) in Brain-derived Endothelial Cells*

, Brain-derived neurotrophic factor (BDNF) is expressed by endothelial cells. We investigated the char-acteristics of BDNF expression by brain-derived endothelial cells and tested the hypothesis that BDNF serves paracrine and autocrine functions affecting the vasculature of the central nervous system. In addition to expressing TrkB and p75 NTR and BDNF under normoxic conditions, these cells increased their expression of BDNF under hypoxia. While the expression of TrkB is unaffected by hypoxia, TrkB exhibits a base-line phosphorylation under normoxic conditions and an increased phosphorylation when BDNF is added. TrkB phosphorylation is decreased when endogenous BDNF is sequestered by soluble TrkB. Exogenous BDNF elicits robust angiogenesis and survival in three-dimensional cultures of these endothelial cells, while sequestration of endogenous BDNF caused significant apoptosis. The effects of BDNF engagement of TrkB appears to be mediated via the phosphatidylinositol (PI) 3-kinase-Akt pathway. Modulation of BDNF levels directly correlate with Akt phosphorylation and inhibitors of PI 3-kinase abrogate the BDNF responses. BDNF-mediated effects on endothelial Har-bor, antibody. research microscope equipped SPOT Photoshop Macintosh G4 FACS Analysis of Apoptosis— FACS analysis of cultured, transfected bEnd-WT cells was performed as described previously Propidium iodide and Annexin V (BD Biosciences) were used to assess apoptotic cells. BD FACStation Software Mac OSX was used to analyze the results, and Statview Software was used to determine significance.

esis has been shown to be regulated by factors secreted by neuronal and glial cell populations in an orderly, spatiotemporal fashion (2). In recent studies we and others have (3,4) shown that selected angiogenic factors, VEGF 1 in particular, are capable of not only affecting a variety of endothelial behaviors but also are capable of affecting neuronal behavior in a receptor-specific fashion. Interestingly, recent studies have demonstrated that neurotrophins expressed by endothelia and are capable of influencing several endothelial cell functions including endothelial cell survival and vessel stabilization (5)(6)(7) and that endothelial cells may express neurotrophin receptors (8).
Neurotrophins mediate their action on responsive neurons by binding to two classes of cell surface receptor (20). TrkA, TrkB, and TrkC selectively bind brain growth factor, BDNF, and NT-3 (21). In addition, the neurotrophins can interact with another low affinity neurotrophin receptor, p75 NTR , which has been shown to initiate an apoptotic signal in neurons when engaged by pro-NGF (22,23). TrkB and BDNF are expressed at high levels not only in central and peripheral nervous tissue (24 -26), but also in several nonneuronal tissues, including muscle, heart, and the vasculature at levels comparable with those of the brain (27)(28)(29)(30). In pathologic states, BDNF and TrkB expression are induced in neointimal vascular smooth muscle cells of the adult rodent and human aorta following vascular injury (31). These studies suggest that there may be a complex and dynamically regulated cross-talk between neuronal cells and endothelial cells during development, growth, and in response to pathological stimuli in the brain and prompted us to investigate these potential interactions.
In this report we have demonstrated the expression of BDNF by brain-derived endothelial cells and the expression and activation of the neurotrophin receptors TrkB and p75 NTR in these brain-derived endothelial cells. In addition, we have shown that engagement of either TrkB or p75 NTR (by BDNF and pro-NGF, respectively) results in distinct endothelial behaviors, survival and angiogenesis in the case of BDNF activation of TrkB and apoptosis in the case of pro-NGF activation of p75 NTR . Furthermore, the importance of these findings in the control of neurovascular development and responses to chronic sublethal hypoxic injury is discussed.

MATERIALS AND METHODS
Recombinant NGF and Pro-NGF-Cleavage-resistant pro-NGF was purified from the media of cells stably expressing the construct, using nickel chromatography and imidazole elution as described (23). Mature NGF or media from cells expressing the plasmid were used in parallel (23).
Cell Culture; RBE4 and bEnd-WT Cell Culture-Transformed rat brain endothelial (RBE4) cells were obtained from F. Roux (Hospital F. Widal, Paris, France). The RBE4 cells were cultured from passages 16 -25 as described previously (32). Immortalized mouse brain endothelial cells (bEnd-WT) were obtained from Dr. Britta Engelhardt (Max-Planck-Institute for Vascular Biology, Mü nster, Germany) and were cultured and passaged as described (33). For three-dimensional culture experiments, acid-soluble calf dermis type I collagen (ASC I) was prepared and solubilized in 10 mM acetic acid (2.5 mg/ml) as described previously (32). RBE4 or bEnd-WT cells were added to the collagen to a final concentration of 2 ϫ 10 5 cells/ml. Droplets of the cell-collagen suspension were spotted onto Petri dishes. Following polymerization, the droplets were overlaid with media (␣-minimal essential medium and F-10 nutrient mixture with glutamine, basic fibroblast growth factor, Geneticin, and 10% fetal bovine serum) and incubated in 5% CO 2 at 37°C. For RBE4 and bEnd-WT culture experiments, recombinant BDNF at concentrations of 10 and 50 ng/ml, soluble, recombinant TrkB receptor bodies (R&D Systems, Minneapolis, MN) at a concentration of 2 g/ml; pro-NGF at concentrations of 1, 5, and 10 ng/ml; and mature NGF at a concentration of 50 ng/ml were added. Wortmannin, LY294002, and PD98059 were purchased from Sigma.
Cells were cultured for 6 days. Medium was changed, and recombinant proteins were added every 24 h.
Transfection of bEnd-WT Cells-bEnd-WT cells were infected with recombinant adenoviruses at ϳ90% confluence. Cells were infected with adenovirus containing HA-tagged dominant negative Akt (Akt-AAA) with a marker of green fluorescent protein (a generous gift of Dr. William Sessa, Yale University) in serum-free Dulbecco's modified Eagle's medium medium for 1 h and then incubated for 24 h in complete growth medium as described (34 -37) before the start of expereiments. Recombinant adenovirus encoding ␤-galactosidase was used as a control. Infection efficiency of bEnd-WT cells with recombinant adenoviruses at 40 multiplicity of infection was close to 100% as determined by the green fluorescent color observed in the cells and immunohistochemical staining of ␤-galactosidase. The expression and relative levels of endogenous and recombinant adenoviral Akt were confirmed by Western blotting.
Matrigel Assay-BD Matrigel TM matrix was used to coat tissue culture dishes according to the manufacturer's instructions (BD Biosciences). Cells were plated onto the matrix at a density of 5 ϫ 10 5 cells per 30-mm plate and allowed to grow for various times. At specific time points, light microscopy images were taken and analyzed, and cell lysates were prepared as described (38).
Vessel Counting-Vessel counts were performed with three samples per condition. Three random fields were photographed per sample. Random digital images of cultures were taken using an inverted research microscope (Olympus Co.) equipped with Nikon coolpix 995 digital camera using Photoshop 5.0 software on a Macintosh G4 computer. NIH Image 1.62 or IP LabSpectrum software was used to select, measure, and analyze the images to determine aggregate tube length. Data were expressed in terms of pixel change (NIH Image) or microns (IP LabSpectrum) compared with normoxic (5% CO 2 and room air (20% O 2 )) controls. Statistics (Student's t test and standard error) were calculated and graphically presented using Excel 2000 on a Macintosh G4 computer. Statistical significance was assumed for p values Ͻ 0.05.
FACS Analysis of Apoptosis-FACS analysis of cultured, transfected bEnd-WT cells was performed as described previously (33). Propidium iodide and Annexin V (BD Biosciences) were used to assess apoptotic cells. BD FACStation Software for Mac OSX was used to analyze the results, and Statview Software was used to determine significance.

BDNF Expression Is Induced by Hypoxia in Vivo in the
Microvasculature of the Central Nervous System and Is Expressed by RBE4 Cells and Induced in These Cells by Hypoxia in Vitro-Imunohistochemical staining of cortical sections of pups reared in normoxia (Nx) and hypoxia (Hx) revealed increased expression of BDNF protein, primarily in the microvasculature Fig. 1, A-D). We then performed immunofluorescence microscopy to assess the expression of BDNF on RBE4 cells. We also performed Western blotting on RBE4 cell lysates. We determined that RBE4 cells expressed BDNF under baseline (normoxic (Nx)) culture conditions. Interestingly, under hypoxic conditions (hypoxic (Hx)) BDNF expression was found to be increased in both RBE4 cells and astrocytes (data not shown) using both immunofluorescence and Western blotting methods (n ϭ 5; p Ͻ 0.03) (Fig. 1, E-H). BDNF was localized in essentially all RBE4 cells, and its expression was noted to be significantly increased under hypoxic culture conditions.

RBE4 Cells Form Tubes and Sprouts in Collagen Gels, Which Are Enhanced by Recombinant BDNF but Blocked by Soluble, Recombinant
TrkB-RBE4 cells placed in three-dimensional matrices of collagen type I gel and cultured for 6 days cluster to form multicellular cysts, from which elongated tube-like processes extend (32). Addition of exogenous recombinant BDNF stimulated increased tube formation in cultures (Fig. 2, compare A and B), while addition of recombinant, soluble TrkB (which sequesters BDNF) acted to inhibit the cyst and tube formation of RBE4 cells (Fig. 2, compare C with A and B).
When cultured under hypoxic conditions, tube formation of RBE4 cells was reduced compared with that of normoxic cultures (Fig. 2, compare D and A). As noted in normoxic cultures, treatment with exogenous BDNF stimulated tube formation TrkB Is Expressed by and Activated by BDNF in RBE4 Cells-To determine whether TrkB is expressed on RBE4 cells, we performed Western blot analyses of lysates of RBE4 cells derived from normoxic and hypoxic cultures. Western blotting illustrated the presence of TrkB in RBE4 cell lysates, and overall, protein levels of TrkB remained unchanged in response to hypoxic stimulation (Fig. 3A).
To determine whether the TrkB present in these brain-derive endothelial cells is activated, we performed Western bolts using anti-pTrkB followed by Western blotting analysis using anti-TrkB. We found that a fraction of TrkB was tyrosinephosphorylated under base-line culture conditions, and phosphorylated TrkB levels were increased following the addition of exogenous BDNF (Fig. 3B). Additionally, treatment of RBE4 cultures with recombinant, soluble TrkB resulted in a significant reduction of phosphorylated TrkB compared with the control (normoxic) cultures. Similar results were observed under hypoxic conditions. These results suggest that TrkB expressed on RBE4 cells is activated by endogenous and exogenous BDNF.
Akt, but Not Mitogen-activated Protein Kinase, Activation Is Associated with BDNF-mediated RBE4 Cell Survival in Vitro-To elucidate the mechanisms involved in this BDNFmediated endothelial cell survival and tube formation in this culture system, we assessed the activation states of Akt and ERK1/2, members of two signaling pathways known to be involved in mediating endothelial survival and tube formation (32,40). The level of phosphorylated ERK was significantly increased by hypoxia but not following treatment of exogenous BDNF or soluble, recombinant TrkB (Fig. 4B). In contrast, the level of serine-phosphorylated Akt was significantly increased following treatment with BDNF under normoxic and hypoxic conditions (Fig. 4A). In addition, treatment with soluble, recombinant TrkB reduced the levels of phosphorylated Akt in normoxic and hypoxic cultures (Fig. 4A). These results suggest that Akt is activated following BDNF engagement and activation of TrkB and this pathway may be in part responsible for mediating the survival of these endothelial cells.
To confirm the role of Akt activation in mediating brainderived endothelial survival bEnd-WT, cells were infected with a dominant negative HA-tagged Akt construct (Akt-AAA) or a ␤-galactosidase-containing vector and apoptotic levels determined following normal culture conditions, serum starvation, and BDNF treatment. While a base-line low apoptotic level was noted in cells infected with the ␤-galactosidase containing vector with and without addition of exogenous BDNF (approximately 10.6% Ϯ 1.3%), high apoptotic levels were observed in the ␤-galactosidase vector-infected cells cultured under serum starvation (17.3%). In contrast, cells infected with the dominant negative Akt-AAA construct exhibited high apoptotic rates in the absence (approximately 17.3% Ϯ 2%) and presence (approximately 20.5% Ϯ 2%) of exogenous BDNF (Fig. 4C). These results lend additional support to the concept that Akt is activated following BDNF engagement, and activation of TrkB and this pathway may be in part responsible for mediating the survival of these endothelial cells.
Exogenous BDNF Blunted Activation of Caspase 3, while Soluble, Recombinant TrkB Induced Activation of Caspase 3 in RBE4 Cells-Additional studies revealed that BDNF modulated caspase 3 cleavage. Exogenous BDNF induced tube formation and rescued the cells from hypoxic insult. Under normoxic conditions, cleaved caspase 3 expression was decreased significantly by addition of exogenous BDNF and was increased following treatment with recombinant soluble TrkB (Fig. 5). Culture of RBE4 cells under hypoxic conditions induced activation of caspase 3; however, cleaved caspase 3 levels were significantly decreased following the addition of exogenous BDNF to hypoxic cultures. As noted above, soluble, recombi-FIG. 1. Cortical tissue, cortical microvasculature, and RBE4 cells express BDNF and exhibit induction of BDNF following hypoxia. A and B are representative sections of cortex from normoxia-reared pups revealing modest BDNF expression. C and D are representative sections of cortex from hypoxiareared pups revealing robust microvascular BDNF expression. E and F are representative immunofluorescence micrographs of cultured RBE4 cells stained with anti-BDNF illustrating modest expression in normoxic (Nx) and increased expression in hypoxic (Hx) conditions. G is a representative Western blot of lysates of RBE4 cells cultured in normoxic and hypoxic conditions for 6 days illustrating BDNF expression (13 kDa) (normalized for actin expression) in both culture conditions, being increased in hypoxic conditions. H represents the average of five Western blotting experiments, illustrating the expression of BDNF in RBE4 cells and its increased expression in hypoxic conditions. (n ϭ 5; * ϭ p Ͻ 0.05; vertical lines represent standard deviations).
nant TrkB further increased the activation of caspase 3 in hypoxic cultures (Fig. 5). These results suggest that BDNF modulates apoptosis in these brain-derived endothelial cells in part by regulating caspase 3 activity.

Modulation of Cleaved Caspase 3 Expression by PI 3-Kinase and MEK Inhibitors on BDNF-treated Endothelial Cells under
Hypoxia-To further determine the specific signaling pathways involved in the BDNF-induced inhibition of apoptosis and inhibition of caspase 3 activation, a pharmacological approach was taken. Namely, chemical inhibitors of either MEK (PD98059, 20 M) or PI 3-kinase (LY240002 and wortmannin, 20 M (data not shown)) were added daily for 6 days to RBE4 cultures. Under normoxic conditions, 20 M LY240002, but not PD98059, significantly increased the levels of cleaved caspase 3 in cultures treated with BDNF (Fig. 6). Under hypoxic conditions, these PI 3-kinase inhibitors, but not the MEK inhibitor, also increased the levels of cleaved caspase 3 significantly in cultures treated with BDNF (Fig. 6). These data suggest that the PI 3-kinase signaling pathway is involved in the BDNFmediated modulation of RBE4 cell caspase activation and survival behavior.
BDNF Treatment Increases VEGFR2 Expression on RBE4 Cells Cultured under Normoxic and Hypoxic Conditions-As we and others (3,32) have reported, VEGF is a potent angiogenic and survival factor in the central nervous system, affecting both endothelial cells and neurons. In previous studies we determined that chronic hypoxia induces VEGF expression in rodent cerebral cortex, specifically by neurons and glia and by astrocytes and cortical neurons in culture (2,3,41,42). Considering that BDNF has been reported to have a survival role in neurons and endothelial cells, we explored the possibility that there may be interactions between the BDNF and VEGF signaling pathways. To elucidate potential interactions between BDNF and VEGF signaling pathways, we performed Western blotting on lysates of RBE4 cells incubated with exogenous recombinant BDNF.
In addition to its modulation of caspase activation (Figs. 5 and 6), BDNF was found to modulate the expression levels of VEGFR2 in RBE4 cells as well as in bEnd-Wt cells cultured under normoxic and hypoxic conditions (Fig. 7). Specifically, addition of exogenous recombinant BDNF elicited significant increases in VEGFR2 expression of 50 and 100% in normoxic and hypoxic cultures of RBE4 cells respectively (Fig. 7A). Sim- ilar results were obtained using bEnd-WT cells as illustrated in Fig. 7B. These results indicate that exogenous BDNF can modulate VEGF-mediated activities in these brain-derived endothelial cells by altering VEGF receptor expression.
bEnd-WT Cells Exhibit Increased VEGFR2 Phosphorylation, Proliferation, and Tube Formation in Response to VEGF following Pretreatment with rBDNF-The findings of increased VEGFR2 expression following BDNF treatment prompted us to determine VEGF-induced VEGFR2 phosphorylation levels following BDNF-induced VEGFR2 expression and the potential functional significance of increased VEGFR2 expression and activation. Determination of phospho-VEGFR2 revealed increased VEGF-induced phosphorylation of the receptor following BDNF pretreatment, and the fraction of VEGFR2 phosphorylated was also significantly increased compared with that determined in control and TrkB-treated cultures (Fig. 8A). bEnd-WT cell cultures pretreated with 10 ng/ml rBDNF for 72 h followed by treatment with 10 ng/ml VEGF for 24 h also exhibited a marked increase in tube length and number, consistent with the increase in VEGFR2 expression and phosphorylation (Fig. 8, B and C). Additionally, pretreatment with rTrkB resulted in a modest decrease in VEGFR2 expression (Fig. 7B) and both tube length and number compared with control cultures (Fig. 8, B and C).
RBE4 Cells Express p75 NTR , and When Engaged by pro-NGF, Apoptosis Is Induced-Interestingly, in addition to expressing TrkB, RBE4 cells were found to express p75 NTR (Fig.  9, A-C), a common low affinity receptor of proneurotrophins, including pro-NGF (20,22). As observed for TrkB expression (Fig. 3A), expression of p75 NTR was not altered by hypoxic culture conditions. Since neurotrophins can be secreted as propeptides, pro-NGF would be capable of binding to RBE4 p75 NTR and initiating signal transduction. Engagement and activation of p75 NTR has been demonstrated to induce apoptosis in neuronal, vascular, and smooth muscle cell populations that express p75 NTR (20,22). Thus, we hypothesized that pro-NGF may bind to p75 NTR on RBE4 cells and initiate a signal transduction pathway distinct from that noted following BDNF mediated TrkB stimulation. To determine the effects of pro-NGF on RBE4 cells, we incubated three-dimensional cultures of RBE4 cells with pro-NGF and mature NGF. Addition of pro-NGF elicited a significant apoptosis of RBE4 cells as evidenced by increased Annexin V staining (Fig. 10A) and reduction of endothelial cell sprout and cyst formation and maintenance (Fig. 10B). In contrast, the vehicle alone and mature NGF treatment groups experienced no appreciable apoptosis.
These results suggest a complex, receptor-mediated regulation of endothelial cell behavior by the expression and processing of neurotrophins and the expression of their cognate receptors.

DISCUSSION
Angiogenesis is a tightly controlled process that is dependent upon the finely integrated and orchestrated expression, availability, and activities of a variety of soluble factors, solid phase components (including several extracellular matrix proteins, glycoproteins, proteoglycans, and glycosaminoglycans), their cellular cognate receptors, and several proteases. Angiogenesis occurs in a multicellular environment in which there is direct contact as well as juxtacrine, paracrine, and endocrine interactions among a variety of cell types, dependent upon which tissue/organ is involved. In previous studies we have determined that astrocyte-endothelial cell interactions (via glial end-foot apposition and secretion of VEGF) and neuronal-endothelial cell interactions (via secretion of VEGF) are critical for normal neurovascular and neuronal development (2,3,20,41,42). However, given the variety of soluble factors known to be expressed during neuronal and neurovascular development and recent reports of specific neurotrophins playing roles in endothelial cell survival and vessel stabilization (6 -8, 31), we reasoned that particular neurotrophins might elicit specific receptor-mediated responses in endothelial cells derived from the brain.
Neurotrophins (BDNF) have been shown to be expressed by some, but not all, endothelial cells in culture (6,8,43). Using brain-derived, immortalized endothelial cells (RBE4 cells) we determined that these cells indeed express BDNF in our threedimensional culture system and that expression is increased following hypoxic treatment (Fig. 1), consistent with our im-

FIG. 8. BDNF pretreatment induces increased VEGFR-2 phosphorylation and VEGF-induced angiogenesis in bEnd-WT cells.
A, determination of phospho-VEGFR2 revealed increased phosphorylation of the VEGFR-2 following BDNF pretreatment of bEnd-WT cells. When normalized to the amount of VEGFR-2 expressed, the fraction of phosphorylated VEGFR2 was also significantly increased compared with that determined in control and TrkB-treated cultures. The upper panel is a representative series of immunoblots used for the quantitation that is illustrated in the lower panel (n ϭ 6; * ϭ p Ͻ 0.05; vertical lines ϭ standard deviations). B and C, bEnd-WT cell cultures pretreated with 10 ng/ml rBDNF for 72 h followed by no treatment (open boxes) or treatment with 10 ng/ml VEGF (shaded boxes) for 24 h exhibited a marked increases in tube length (B) and aggregate tube number (C), consistent with the increases in VEGFR2 expression and phosphorylation following BDNF pretreatment. As expected pretreatment with rTrkB resulted in decreased tube length and number compared with control cultures (n ϭ 6; * ϭ p Ͻ 0.05; vertical lines ϭ standard deviations). munohistochemical localization data. In previous studies we have shown that these cells form cysts with linear angiogenic sprouts emanating from them when placed in a three-dimensional collagen type I gel (3,32). Upon treatment of these cultures with recombinant BDNF, a significant increase in angiogenic sprouts was noted. In contrast, treatment with re-combinant, soluble TrkB (which sequesters BDNF) resulted in a marked loss of both angiogenic sprouts and cysts (Fig. 2, A-C  and G). Interestingly, under hypoxic culture conditions addition of exogenous BDNF promoted cell survival and robust angiogenic sprout formation (Fig. 2, D-G), similar to that noted previously for VEGF (32). These data are consistent with the presence of a BDNF receptor on these cells. Indeed, Fig. 3A illustrates the presence of TrkB. Activation (tyrosine phosphorylation) of this receptor in response to its ligand (BDNF) is required if we are to ascribe the survival/angiogenic response to a receptor-mediated process. Fig. 3B illustrates the required increase in TrkB phosphorylation in response to exogenous BDNF and the reduction of TrkB phosphorylation following sequestration of endogenous BDNF. This BDNF response appears to signal changes in endothelial cell survival via a PI 3-kinase/Akt pathway as addition of BDNF results in increased Akt phosphorylation, and sequestration of BDNF causes a reduction of Akt phosphorylation, while ERK is not appreciably affected, and overexpression of a dominant negative Akt renders the cells insensitive to exogenous BDNF (Figs. 4 and 5).
Apoptosis can be assessed by determination of caspase activation (cleavage). We found a correlation between the level of RBE4 cell apoptosis, the level of BDNF, the phosphorylation state of Akt, and the level of cleaved caspase 3 suggesting that activation of TrkB results in signaling cell survival via the PI 3-kinase/Akt pathway (Fig. 5). This was confirmed using synthetic inhibitors of PI 3-kinase (LY294002 and wortmannin) and MEK (PD98059) (Fig. 6).
Endothelial apoptosis is known to be modulated by several soluble factors and engagement of their cognate receptors (32,44). Our data suggest that engagement of TrkB results in a survival signal. However, TrkB-induced endothelial cell survival may also be mediated via indirect signaling pathways. Since it is known that VEGF signaling appears to signal survival in these cells (32), we assessed the effects of TrkB activation on expression of VEGFR-2 expression. Interestingly, in both normoxic and hypoxic conditions, addition of BDNF resulted in increased expression of RBE4 VEGFR-2 (Fig. 7). These data are consistent with the notion that in addition to its direct effects on apoptosis, engagement, and activation of TrkB may exert some of its anti-apoptotic and angiogenic effects via up-regulation of VEGFR-2, a known major modulator of endothelial survival, proliferation, and angiogenesis (Figs. 7 and 8).
Neurotrophin expression is known to be up-regulated during injury and stress to the central nervous system (53), resulting in significant changes in the levels and states of the neurotrophins (45,46), presumably affecting extent of injury and subsequent repair (47)(48)(49)(50). Neurotrophins can be secreted as propeptides, which are cleaved by specific metalloproteinases, producing active, mature neurotrophins, which in turn are capable of binding to their cognate Trk receptors with high affinities (22). In contrast, the proneurotrophins can interact with another neurotrophin receptor, p75 NTR , which has been shown to initiate an apoptotic signal in neurons when engaged (22,51). Considering that neurotrophin levels and their state (pro versus mature) can initiate diverse signaling pathways (23), we investigated whether RBE4 cells also expressed p75 NTR . As illustrated in Fig. 9, RBE4 cells indeed do express p75 NTR and appear to respond to its engagement with pro-NGF. Fig. 10 illustrates the effects of engagement of p75 NTR on the survival of cultured RBE4 cells. Engagement of p75 NTR by pro-NGF resulted in a marked increase in Annexin V staining (Fig. 9A) and a dramatic loss of cell viability as evidenced by quantitation of the numbers of endothelial cysts and sprouts remaining after 24 and 48 h of treatment (Fig. 9B). As ex- FIG. 9. RBE4 cells express p75 NTR . A and B, representative immunofluorescence micrographs of cultured RBE4 cells stained with anti-p75 NTR illustrating unchanged expression in normoxic (Nx) and hypoxic (Hx) conditions. C, upper panel: representative Western blots for p75 NTR in lysates of normoxic (Nx) and hypoxic (Hx) RBE4 cultures in the presence or absence of 50 ng/ml BDNF normalized for ERK2 expression. C, lower panel: quantitation of p75 NTR expression in RBE4 cells as described above. Note that there are no statistically significant changes in p75 NTR expression in any of the conditions tested (n ϭ 5; p Ͼ 0.05; vertical lines ϭ standard deviations).
FIG. 10. Pro-NGF induces RBE4 cell apoptosis. A, representative immunofluorescence micrographs of RBE4 cells cultured in the presence of vehicle alone (5 ng/ml mature NGF (data not shown)), 50 ng/ml mature NGF, or 5 ng/ml pro-NGF and stained for Annexin V. Note the increased staining in the cultures treated with pro-NGF but not with mature NGF. B, quantitation of RBE4 cell survival and sprout formation/survival in normoxic conditions in the presence of vehicle alone, 50 ng/ml mature NGF, or 5 ng/ml pro-NGF. Note the decreased numbers of cysts in the cultures treated with pro-NGF (n ϭ 5; * ϭ p Ͻ 0.05; vertical lines ϭ standard deviations). pected, mature NGF and vehicle alone had no appreciable effects on cell viability.
Given our findings that "vascular" ligands and their receptors (VEGF and VEGFRs) (3,32) and neurotrophins and their receptors (BDNF, NGF, TrkB, and p75 NTR ) are differentially expressed in central nervous system-derived endothelial cells, glia, and neurons, it is likely that these cell types (as well as other cell types in the central nervous system) interact with each other via soluble factors, resulting in a dynamic modulation of cellular behaviors during normal development and maintenance of the central nervous system as well as in response to noxious stimuli. During chronic sublethal hypoxia in vivo (2,42,52), and as modeled in our culture system (2,3,32,42), increased glial and neuronal VEGF expression is noted, which in turn initiates and maintains cerebral angiogenesis and loss of permeability function and alters neuronal apoptosis and differentiation. These perturbations and others involving neurotrophins (BDNF, NGF), their proteolytic processing, and their receptors (TrkB and p75 NTR ) would then affect subsequent central nervous system cell behaviors including cell survival, proliferation, migration, and differentiation, which in turn would have dramatic effects on the genes involved with synaptic maturation, post-synaptic function, neurotransmission, glial maturation, and angiogenesis. Thus, a more complete understanding of these complex, dynamic interactions among these cells and their soluble factors appears warranted if we are to develop rational therapeutic agents to beneficially affect the responses of the brain to injury and reparative processes.