Delineating multiple functions of VEGF-A in the adult brain

Abstract

Vascular endothelial growth factor-A (abbreviated throughout this review as VEGF) is mostly known for its angiogenic activity, for its activity as a vascular permeability factor, and for its vascular survival activity [1]. There is a growing body of evidence, however, that VEGF fulfills additional less ‘traditional’ functions in multiple organs, both during development, as well as homeostatic functions in fully developed organs. This review focuses on the multiple roles of VEGF in the adult brain and is less concerned with the roles played by VEGF during brain development, functions described elsewhere in this review series. Most functions of VEGF that are essential for proper brain development are, in fact, dispensable in the adult brain as was clearly demonstrated using a conditional brain-specific VEGF loss-of-function (LOF) approach. Thus, in contrast to VEGF LOF in the developing brain, a process which is detrimental for the growth and survival of blood vessels and leads to massive neuronal apoptosis [2–4], continued signaling by VEGF in the mature brain is no longer required for maintaining already established cerebral vasculature and its inhibition does not cause appreciable vessel regression, hypoxia or apoptosis [4–7]. Yet, VEGF continues to be expressed in the adult brain in a constitutive manner. Moreover, VEGF is expressed in the adult brain in a region-specific manner and in distinctive spatial patterns incompatible with an angiogenic role (see below), strongly suggesting angiogenesis-independent and possibly also perfusion-independent functions. Here we review current knowledge on some of these ‘non-traditional’, often unexpected homeostatic VEGF functions, including those unrelated to its effects on the brain vasculature. These effects could be mediated directly (on non-vascular cells expressing cognate VEGF receptors) or indirectly (via the endothelium). Experimental approaches aimed at distinguishing between these possibilities for each particular VEGF function will be described. This review is only concerned with homeostatic functions of VEGF in the normal, non-injured brain. The reader is referred elsewhere in this series for a review on VEGF actions in response to various forms of brain injury and/or brain pathology.

VEGF-A (abbreviated throughout this review as VEGF) is mostly known for its angiogenic activity, for its activity as a vascular permeability factor and for its vascular survival activity [1]. There is a growing body of evidence, however, that VEGF fulfills additional less ‘traditional’ functions in multiple organs, both during development, as well as homeostatic functions in fully developed organs. This review focuses on the multiple roles of VEGF in the adult brain and is less concerned with the roles played by VEGF during brain development, functions described elsewhere in this review series. Most functions of VEGF that are essential for proper brain development are, in fact, dispensable in the adult brain as was clearly demonstrated using a conditional brain-specific VEGF loss-of-function (LOF) approach. Thus, in contrast to VEGF LOF in the developing brain, a process which is detrimental for the growth and survival of blood vessels and leads to massive neuronal apoptosis [2–4], continued signaling by VEGF in the mature brain is no longer required for maintaining already established cerebral vasculature and its inhibition does not cause appreciable vessel regression, hypoxia or apoptosis [4–7]. Yet, VEGF continues to be expressed in the adult brain in a constitutive manner. Moreover, VEGF is expressed in the adult brain in a region-specific manner and in distinctive spatial patterns incompatible with an angiogenic role (see below), strongly suggesting angiogenesis-independent and possibly also perfusion-independent functions. Here we review the current knowledge on some of these ‘non-traditional’, often unexpected homeostatic VEGF functions, including those unrelated to its effects on the brain vasculature. These effects could be mediated directly (on non-vascular cells expressing cognate VEGF receptors) or indirectly (via the endothelium). Experimental approaches aimed at distinguishing between these possibilities for each particular VEGF function will be described. This review is only concerned with homeostatic functions of VEGF in the normal, non-injured brain. The reader is referred elsewhere in this series for a review on VEGF actions in response to various forms of brain injury and/or brain pathology.

VEGF and vascular homeostasis in the non-injured adult brain

Generally speaking, the adult brain is angiogenically quiescent. Yet, like in any other organ a natural vascular turnover necessitates a mechanism for replacing lost vessels with new ones in order to secure adequate perfusion. Moreover, increased oxygen consumption due to increased neuronal activity may pose a requirement for increasing microvascular density (MVD). VEGF, by virtue of its hypoxia- and hypoglycemic responsiveness, is very likely the angiogenic factor responsible for constantly matching MVD to increased metabolic demands (reviewed by [8]). Rearing mice in an enriched environment or voluntary running was shown to increase hippocampal vascular density and surface area, respectively [9, 10] and rearing rats under different visual stimuli during the critical period shows a correlation between the complexity of visual experience, the density of capillary networks and expression of VEGF in the visual cortex [11].

To determine the degree of cerebral vessel dependence on continued VEGF signaling, different methods of VEGF inhibition were utilized. Blood vessels in the adult brain were found to be relatively independent of VEGF when compared to vascular networks in other organs [12, 13]. A notable exception is blood vessels residing in the choroid plexus (CP). This vascular bed is composed of highly fenestrated capillaries and surrounding choroid epithelial cells. Fenestrae are made to secure proper production of cerebro-spinal fluid (CSF) [7, 14–16]. The CP vasculature significantly regresses following inhibition of VEGF signaling [7, 16]. This is almost certainly attributable to the fact that CP vessels are the only cerebral vascular bed composed of fenestrated endothelium in conjunction with findings that fenestrated endothelial cells are exceptionally dependent on VEGF [16], shown to be required not only for fenestrae formation but also for keeping fenestrations open [13].

The different ways by which VEGF may disrupt the blood–brain barrier (BBB), including through formation of inter-endothelial gaps and modulation of junctional protein expression, are beyond the scope of this review.

Multiple functions of VEGF are suggested by distinctive patterns of its cerebral expression

Initial clues on possible functions of VEGF in the normal adult brain have been provided by in situ analyses of its endogenous patterns of expression. Most informative in this regard has been the complementary approach of using a transgenic reporter knock-in onto the VEGF locus where a LacZ cassette was inserted into the 3′-untranslated region of the VEGF gene [17] to overcome embryonic lethality associated with even a heterozygous VEGF loss of function [18]. Some of the major sites of constitutive VEGF expression highlighted with the aid of this methodology are shown in Fig. 1. The most prominent site of VEGF expression in the adult brain is in the choroid plexus [19]. The high expression level observed is in accordance with its important role in CP endothelial cell survival and fenestrae formation described above.

Fig. 1.

Endogenous VEGF expression in the adult mouse brain. X-gal staining of transgenic mouse expressing LacZ (blue) under the authentic VEGF promoter. Astrocytes were visualized using GFAP staining (in all images except bottom right image) to highlight VEGF-expressing astrocytes. In the DG, GFAP staining of NSC in the SGZ did not co-localize with X-gal. In the CA1, arrows point to x-gal- positive pyramidal cells. In the cerebellum, arrows point to Purkinje cell layer

In the brain parenchyma, co-expression with astrocyte-specific markers (e.g., GFAP) has shown that major producers of VEGF throughout the adult brain are astrocytes [4, 5, 11, 20, 21])(see also Fig. 1). Findings show that astrocytes play a major role in the upregulation of HIF1 target genes in response to cerebral hypoxia [22]. It is conceivable that astrocytic expression of VEGF, a direct HIF1 target, reflects a situation in which astrocytes function as more efficient hypoxia sensors than other cell types in the brain. Upon pathological conditions, the expression of VEGF may alter. VEGF was elevated in the dentate gyrus (DG) of the hippocampus after electroconvulsive shock [23]. Under severe hypoxia, VEGF is potentially upregulated in all cell types within the brain [24].

In several brain regions, VEGF is produced by cells other than astrocytes. In the olfactory bulb, for example, a high level of expression by mitral and periglomerular neurons is observed [4, 19]. In the hippocampus, VEGF is expressed by CA1 pyramidal neurons [25] and its production is augmented upon cell depolarization [26]. In the cortex, VEGF is expressed by some pyramidal neurons [25] and in the cerebellum, it is expressed by Purkinje cells [19, 27]. It is noteworthy that all of these sites of constitutive VEGF expression are angiogenically quiescent, thus suggesting non-angiogenic roles of VEGF.

VEGF and adult neurogenesis

Adult neurogenesis takes place constitutively and in a continuous manner in two defined areas of the adult brain, namely, in the DG of the hippocampus and in the subventricular zone (SVZ) placed in the lateral wall of the lateral ventricles. The hippocampus is composed of several sub-regions known as the CA1, CA3 and the DG (see Fig. 1) and plays a key role in spatial learning and short-term memory consolidation. Neural stem cells reside within the sub-granular zone of the DG and give rise to dentate granule neurons that eventually integrate in the existing dentate network. These newly added neurons differ, however, from ‘older’ neurons by virtue of being more excitable [28–30]. At the second neurogenic site, neuronal stem cells residing in the SVZ constantly produce neuroblasts destined to become olfactory bulb (OB) interneurons. Newly made neuroblasts leave the SVZ and migrate rostrally (tangential migration) along the rostral migratory stream (RMS). Upon reaching the OB, cells migrate radially and, upon reaching their destinations at the granule cell layer or glomerular layer, differentiate into granule cells (GC) or periglomerular cells (PG) respectively, becoming fully functional interneurons.

Studies from several laboratories using different modes of VEGF overexpression in the respective neurogenic sites, and various readouts for neuroblast proliferation, have clearly established that VEGF is indeed a ‘neurogenic factor’ in the sense of significantly augmenting the basal rate of cell proliferation, neuroblast production and neuronal differentiation in the hippocampus [5, 31–33]. Likewise, VEGF was shown to play a similar role in SVZ neurogenesis in several in vivo and in SVZ-derived neurosphere studies [32–36].

While studies outlined above clearly established that VEGF may enhance adult neurogenesis overall, a particular role(s) of VEGF in the particular sub-processes of this multi-step process is only partially understood. The following is an account of studies aimed at delineating specific roles of VEGF in the particular cellular events that compose the highly complex process of adult neurogenesis. While both DG and SVZ neurogenesis are discussed in parallel, the latter appears more suitable for discerning successive stages of the overall process in which VEGF might play a role, due to the fact that neuronal birth, migration, differentiation and integration are spatially and temporally well separated. For uncovering a natural role of VEGF, it is also essential to use a VEGF loss-of-function approach in the respective neurogenic niches. This was achieved using several approaches [4, 5, 31, 37, 38] and will be further elaborated.

VEGF and the neurogenic stem cell niches

Stem cells (SC), in general, are known to reside in specialized cellular niches believed to be required for proper SC function. The ways in which SC properties and performance are controlled by the niche, however, are poorly understood. The respective organ vasculature is thought to constitute an integral component of the SC niche. Taking into consideration that VEGF is the factor promoting and orchestrating most, if not all angiogenic responses, as well as having a profound effect on so-called ‘angiocrine factors’ elaborated by the endothelium, it is possible that VEGF’s impact on overall neurogenesis may include affecting NSCs-vascular interactions taking place at the niche.

A prerequisite for the presumed control of blood vessels on NSC activity is a physical proximity of SCs and vessels. An intimate association of SCs and blood vessels was indeed demonstrated for both neurogenic niches within the adult brain.

The SVZ has a complex architecture, composed of stem cells (type B cells), transient amplifying progenitors (type C cells) and neuroblasts (type A cells) (for a review on the SVZ architecture see [39]). GFAP+ type B cells were visualized sending a long cellular process towards a blood vessel [40]. Transient amplifying type C cells were shown to proliferate near blood vessels [41, 42], making a direct contact with capillary endothelium without intervening mural cells [42]. Proper contacts between neural stem/progenitor cells and endothelium are likely mediated by α6β1 integrin on neural cells to laminin on the endothelium, as evidenced by detachment of neuroblast clusters from the ventricular wall upon β1 integrin blockade [43] or by decreased attachment of SVZ neurospheres to an endothelial monolayer in the presence of α6β1 neutralizing antibodies [41]. Further, β1 integrin expression coincides with activated (type C) cells whereas type B cells show no β1 expression [43].

Some clues on how nearby blood vessels may affect neural stem cells come from identifying factors secreted by SVZ endothelial cells. Thus, endothelially produced CXCL12 (SDF1) induces type B cell activation accompanied by their translocation from near the ependyma to more lateral blood vessels [44]. Further, fluorescently tagged transplanted neural progenitors home to blood vessels after integration, a process mediated by interaction of endothelial SDF1 and cognate CXCR4 receptors expressed by neuroblasts [44]. Betacellulin, a member of the EGF family, was recently shown to be secreted by SVZ endothelial cells and nearby choroid plexus and to greatly increase neurogenesis. Its inhibition by blocking antibodies reduced neuronal cell proliferation within the SVZ, suggesting a natural neurogenic role for betacellulin [47]. Endothelial derived brain derived neurotrophic factor (BDNF) was shown to support neuronal proliferation of SVZ explants [46] and its in vivo injection enhanced SVZ neurogenesis [47]. A particular neurogenic process where the role of VEGF–BDNF axis was further analyzed was in the seasonal addition of new neurons to the high vocal center (HVC) of male songbirds [35]. Here, testosterone-induced VEGF was found to stimulate nearby endothelial cells to secrete BDNF that, in turn, promotes recruitment of neurons from the HCV ventricular zone. Finally, Pigmented Epithelium-Derived Factor (PEDF) was shown to be naturally expressed by ependymal and endothelial cells of the SVZ and infusion of PEDF or, conversely, administration of a dominant negative variant, respectively enhanced or inhibited activation of type B cells [48]. Taken together, these findings argue for a paracrine role of multiple endothelium-derived factors in controlling NSC activity. A more rigorous approach of endothelial-specific ablation of candidate angiocrine factors is required, however, to scrutinize the vascular origin of any factor operating within the niche, similar to the study proving the restricted vascular origin of stem cell factor (SCF), required for proper function of hematopoietic stem cells [49].

VEGF was shown to be expressed by astrocytes in the vicinity of the SVZ (Fig. 1) but CP-derived VEGF may also play a role in the nearby ventricular wall. SVZ-derived neurospheres were shown to expand upon exposure to VEGF [32, 33]. In vivo, ICV infusion or AAV-mediated administration of VEGF to mice enhanced BrdU uptake by SVZ cells [32, 34, 36]. GFAP+ cells were shown to express Flt1 and doublecortin+ SVZ cells were shown to express Flk1 and to respond to VEGF infusion by Flk1 phosphorylation [50] but, in contrast, its infusion did not change the levels of BrdU+ cells [33, 50]. In accordance with these studies, when signaling of the endogenously produced VEGF was precluded through transgenic induction of its decoy receptor sFlt1, no apparent change in SVZ neurogenesis was evident [4].

In the hippocampus, proximity of NSCs residing in the subgranular zone of the DG to blood vessels has been demonstrated over a decade ago [51]. Yet, the functional significance of this anatomical arrangement is unclear and the notion that blood vessels are required for NSC function in a way other than providing proper perfusion is currently missing. An interventional approach to manipulate blood vessels in the presumed niche (for this matter, at both neurogenic sites) is, therefore, required to advance beyond the current status of ‘guilt by association’.

While studies have demonstrated that VEGF-induced DG angiogenesis is accompanied by increased DG neurogenesis [5, 31–33], it remains unclear whether the apparent neurogenic activity of VEGF is attributed to its direct affect on neurons (or glia) or, alternatively mediated by its effect on the niche vasculature. In the case of the latter, it is further questioned whether increased neurogenesis requires ongoing VEGF vascular signaling or, alternatively, whether VEGF-instructed increase of the niche microvascular density is sufficient by itself to maintain an elevated neurogenic level. To address these questions, we have used a strategy of conditionally inducing VEGF in the DG followed by its de-induction after the newly added DG vasculature became refractory to ectopic VEGF withdrawal. Results clearly showed that mere expansion of the niche vasculature is sufficient to dramatically increase DG neurogenesis and that the level of enhanced neurogenesis is directly proportional to the increase in microvascular density at the DG niche [5] and our unpublished results). However, AAV-mediated administration of Placental-derived growth factor (PlGF), a Flt1-specific ligand, resulted in increased angiogenesis without affecting neurogenesis, thus supporting an independent role for VEGF in neurogenesis [31]. VEGF loss-of-function does not impair the basal level of DG neurogenesis as evidenced by findings that it was not affected by sFlt1 overexpression [5, 37], by treatment with the Flk1 inhibitors PTK787/ZK222584 and SU5416 [38, 52], or by VEGF inhibition via AAV-mediated delivery of interfering shRNA [31]. On the other hand, VEGF inhibition abrogated the increase in hippocampal neurogenesis gained by physical exercise or by rearing animals in an enriched environment [31, 37]. Together, these studies suggest that endogenous VEGF is dispensable for maintaining basal neurogenesis, but is essential for environmentally induced neurogenesis, possibly through its angiogenic activity.

Endothelial cells may dynamically (and reversibly) switch from a state of quiescence to different forms of an activated state. This may determine, in turn, the composition of the endothelial secretome and its resultant influence on SC behavior. Whether the niche vasculature must be in a particular state in general, or in a VEGF-activated state in particular in order to secure proper SC function is currently unknown. Because VEGF also controls vascular permeability and the presumed influence of vascular permeability on the content of blood-borne factors released by the niche vasculature, as was demonstrated for the SVZ [42], it is possible that VEGF levels might impact SC function also via this mode. Hemodynamic changes in SVZ vasculature might potentially also affect NSC behavior. A recent study has shown that ICV injection of bFGF and HGF leads to a rapid 50 % increase in blood flow, specifically in SVZ blood vessels (but not in nearby striatum vessels) and a concomitant 60 % increase in EdU⁺ cells within the SVZ [53]. It is currently unknown whether these concurrent effects are causally related nor whether the SVZ vasculature is more prone to hemodynamic influences.

VEGF and neuroblast migration

The neurogenic process, whether in developmental or adult neurogenesis, usually involves a stage of newborn neuroblasts migrating from their birth place to their remote site of integration and VEGF was shown to play an important role in neuroblast migration during brain development. Notable examples include controlling correct migration of facial nerve somata [54] and an essential role in granule cell migration during cerebellum development at postnatal life [27]. Here, VEGF attracted the growth cone of granule cells in vitro and the migration of these cells was disturbed in both VEGF δ/δ hypomorphic mice in vivo, and under ectopic expression of VEGF in granule cells ex vivo. Knockout of Flk1 in cerebellar granule cells shows a direct role for VEGF on the migration of these cells [27]. This migration was also found to be dependent on direct phosphorylation of several NMDA receptor subunits by Flk1 [55].

Focusing on the adult brain, however, a possible role of VEGF in neuroblast migration was mostly addressed with regard to SVZ neurogenesis. This migratory process is distinguished by long-distance neuroblast migration along a well-characterized route, namely, the RMS. Neuroblasts migrating along the RMS were shown to utilize blood vessels as a migratory scaffold [56, 57] and because VEGF produced by astrocytes was recently shown to play a key role in forming this migration-promoting vascular scaffold in the postnatal brain, an indirect role of VEGF in the process could be argued [57]. At early postnatal stages of brain development, VEGF inhibition results in disoriented RMS vasculature and a subsequent abnormal migration. At a later postnatal stage (p14) and consistent with diminished VEGF production by RMS astrocytes, VEGF inhibition no longer results in aberrant vascular scaffold or impaired migration [57]. A similar conclusion was drawn from our study showing that conditional VEGF knock-down in the embryonic RMS results in vascular collapse. These animals, in adulthood, display uninterrupted migration up to the point of vascular collapse and progressive accumulation of neuroblasts therein [4]. In contrast, exercising VEGF loss-of-function in the fully mature brain has excluded a natural role for VEGF in RMS migration as evidenced by an unaltered number of neuroblasts reaching and populating the OB [4]. Interestingly, enforced VEGF overexpression (directly or through the use of VEGF-R1 TK⁻/⁻ mice [58]) resulted in less newborn neuroblasts along the RMS and their higher abundance in the OB, suggesting that VEGF might accelerate migration [50]. Consistent with this interpretation are findings that migration of SVZ-derived cells is enhanced upon exposure to VEGF [59].

Which other factors may promote migration along vessels? BDNF was shown to be expressed uniquely on the endothelium of the RMS and not in the nearby striatum or cortex [60]. Upon secretion, it is attached to migrating neuroblasts expressing p75 receptors and alternatively, following induction of GABA signaling, to astrocytes of the RMS expressing the high affinity receptor. Thus, GABA signaling attenuates BDNF-mediated migration [60]. Ectopic expression in the striatum leads to de-routing of neuroblasts from the RMS to the striatum, again using endothelial cells as scaffold. Endothelial BDNF knock-down or p75 knock-down in the RMS also resulted in abnormal migration to the OB [60]. Since VEGF promotes endothelial BDNF expression in songbirds [35], it may potentially do so also in the RMS.

A rapid SVZ neurogenic wave is known to be elicited in response to a distant brain injury associated with re-directing neuroblast migration from the RMS route in the direction of the injury site and VEGF is thought to contribute to this altered migration cue. Multiple roles of VEGF in brain injuries are discussed elsewhere in this review series.

VEGF and neuronal maturation and differentiation

The fact that neuronal maturation and differentiation in the SVZ–RMS–OB pathway can be analyzed independently of preceding neurogenic stages due to their spatial and temporal separation was exploited to examine a putative role for VEGF in these particular sub-processes. Briefly, upon reaching the OB, cells migrate radially and, upon reaching their destination at the granule cell layer or glomerular layer, differentiate into GC or PG, respectively, becoming fully functional interneurons. Temporally, tangential migration is completed within 2–7 days from cell division, radial migration within 5–7 days, and differentiation to GCs or PGs within 2 or 4 weeks, respectively, from birth [61]. Changes in electrophysiological properties indicate several maturation steps [62]. Intriguingly, VEGF is constitutively highly expressed in OB cells (mitral and periglomerular neurons) residing at the destination sites of incoming neuroblasts, i.e., juxtaposed to their maturation and differentiation sites (see Fig. 1), thus suggesting a role for VEGF in these processes. Inhibition of VEGF signaling at these sites indeed impaired normal maturation of newly added neurons, as evidenced by reduced dendritic spine density of granule cells, significantly shorter and less branched dendrites in periglomerular neurons and, functionally, by impaired smelling. Importantly, already established OB neurons were fully refractory to VEGF inhibition [4].

Regarding a possible role of VEGF in directing neuroblast differentiation, it was shown that mice with a kinase dead Flt1 mutation have more OB periglomerular neurons with a higher proportion of dopaminergic, TH-expressing neurons [50].

VEGF, synaptic plasticity, and neuronal activity

Synaptic plasticity, i.e., the ability of the synapse to change in strength in response to either use or disuse of transmission over synaptic pathways, is considered one of the important neurochemical foundations of learning and memory. There are several underlying mechanisms that cooperate to achieve synaptic plasticity, including changes in the quantity of neurotransmitters released into a synapse, changes in how effectively cells respond to those neurotransmitters and changes in the number and density of synapses. A complex web of signaling pathways participate in modulating synaptic plasticity and the question arises of whether VEGF also contribute to the process.

Both inhibitory and stimulatory effects of VEGF on synaptic properties of hippocampal neurons have been reported: Bath application of VEGF to hippocampal slices reversibly reduced the amplitude of evoked responses in all hippocampal sub-regions and inhibited spontaneous discharges in slices from epileptic rats [63]. On the other hand, excitatory synaptic transmission (increasing the frequency of vesicular release and mEPSCs) in cultured hippocampal neurons was enhanced by HIF1α in a VEGF-dependent manner [64]. More informative studies addressing the issue of the effect of VEGF on synaptic plasticity were those using long-term plasticity (LTP) as a measure of dynamic plasticity changes, i.e., an increase in synaptic response following potentiating pulses of electrical stimuli that is sustained at a level above the baseline response for an hour or longer. Hippocampal slice preparations acutely treated with VEGF that enhance LTP response in CA1 pyramidal cells via CamkIIα activation and Ca⁺⁺ influx have been demonstrated [26]. Using in vivo LTP measurements, we have recently shown that VEGF dramatically enhances LTP responses in the perforant path (i.e., entorhinal cortex to DG) and, conversely, that VEGF inhibition completely abrogated LTP. Interestingly, even though comparable levels of VEGF (or sflt1) were induced in the CA1 and DG regions, only DG granule cells exhibited plasticity changes in response to the respective VEGF manipulation. Switching-off ectopic VEGF production resulted in return to a normal LTP response, indicating that ongoing VEGF signaling is required to maintain increased plasticity [5], uncoupled to its effects on neurogenesis and perfusion.

VEGF was shown to act on several non-synaptic ion channels, also via either inhibition or stimulation. VEGF pretreatment of cultured neurons or slices acutely attenuated Ca⁺⁺ influx induced by application of KCl or glutamate in an acute, dose-dependent, and reversible fashion. This effect was blocked by the Flk1 inhibitor SU1498. [65]. In cultured hippocampal cells, VEGF inhibited potassium-channel currents [66] and induced tyrosine phosphorylation of the Kv1.2 potassium channel [67]. As mentioned before, during cerebellar granule cell migration, VEGF induces clustering and phosphorylation of non-synaptic NMDA receptors with Flk1 leading to enhanced NMDA-dependent Ca⁺⁺ influx [55]. Direct effect of VEGF on ion channels may explain some of its short-term effects which could not be explained by altered gene expression or vascularization.

VEGF in learning, memory, and behavior

A growing body of evidence employing different modes of VEGF administration or induced overexpression has shown that VEGF can enhance hippocampus-dependent learning and memory. Thus, AAV-mediated VEGF delivery enhanced memory in the associative passive avoidance task and the Morris water maze [31]. Stabilization of HIF1α, and a resultant upregulated VEGF expression enhanced fear conditioning memory [68] and transgenic induction of VEGF also enhanced memory in the fear conditioning paradigm [5]. A natural requirement for VEGF for proper hippocampal learning and memory was demonstrated in reciprocal VEGF loss-of-function experiments. Flk1 inhibition using a protein acting in a dominant-negative manner [31] or using a specific PTK787/ZK222584 inhibitor [38] impaired long-term memory in the Morris water maze or the passive avoidance task, respectively. Similarly, VEGF inhibition by transgenic sFlt1 impaired memory in the fear conditioning and eight-arm maze [5]. Considering that endogenous VEGF expression by hippocampal astrocytes is further induced by physiological stimuli such as environmental enrichment [31], exercise [69], memory challenge [31] and excitatory stimulation of hippocampal neurons [26], it is likely that enhanced memory by these stimuli is also mediated by VEGF.

It does appear that VEGF acts in the hippocampus in multiple ways capable of increasing the vascular density, neurogenesis and cognitive function. The question then arises of whether these effects of VEGF are causally related or, alternatively, independent of each other. Pertinent to the ongoing general debate regarding the relative contribution of adult neurogenesis to learning and memory on one hand, and synaptic plasticity on the other hand, we found that memory gain by VEGF gof and memory impairment by VEGF lof were already evident at early time points where newly added neurons could not have yet become functional [5]. Further support to the contention that cognitive gain by VEGF gof is attributed to increased synaptic plasticity and not to VEGF-promoted neurogenesis was its complete loss upon switching-off transgene expression, whereas newly added neurons have persisted this manipulation [5]. Together, these findings demonstrate that the mechanisms underlying VEGF-induced neurogenesis and VEGF-induced memory enhancement can be uncoupled.

Open questions and future directions

In light of newly emerging studies that show VEGF has many roles in multiple organs extending far beyond its traditional role as an angiogenic factor, recent studies have uncovered new, often unexpected roles of VEGF in the adult normal brain as described above and summarized in Table 1. It is likely that additional cerebral functions of VEGF are yet to be discovered.

Table 1.

VEGF functions in the healthy, adult brain

Role	Constitutive/induced	Brain region	Remarks	References
Vascular
Matching MVD to perfusion demands	Physiological response	Throughout	Also neovascularization in response to vascular injury	[8]
Control of vessel permeability	Constitutive	Throughout		[83]
Maintaining endothelial fenestrations	Constitutive	CP	A role in BBB regulation	[7]
Controlling the vascular stem cell niche	Induced	DG		[5, 31]
Neurogenesis
Basal neurogenesis	Experimental induction	SVZ, DG	Unaffected by VEGF LOF	[4, 5, 31, 32]
Activity-induced neurogenesis	Enrichment, exercise	DG	Inhibited by VEGF LOF	[31, 37]
Neuroblast migration	Constitutive	RMS, Postnatal cerebellum	Not required for RMS migration	[27, 57]
Neuronal maturation		OB		[4, 50]
Synaptic plasticity
LTP	Constitutive and induced	DG	Induced LTP requires ongoing VEGF	[5, 26]
Learning and memory	Constitutive and induced		Reversible	[5, 31]

A first hint to this supposition is the existence of distinctive expression patterns of VEGF that are still unaccounted for, primarily its massive expression in cortical neurons. Recently, a new niche for stem cells in the adult cortex was described that responds to both pathological (stroke) and physiological (enrichment) stimuli [70]. It would be interesting to examine whether VEGF or the vasculature play a similar role in this neurogenic niche as it does in the other niches. Another intriguing question concerns region-specific differences in the type of cells that produce VEGF. For example, within the hippocampus, VEGF is produced by neurons in the CA1 region whereas in the nearby DG it is produced solely by astrocytes. In certain regions of the developing brain VEGF was shown to fulfill specific additional functions and, by way of extrapolation, it is possible that similar functions are also important for the adult brain. For example, in the developing hypothalamus VEGF is required for survival of neurons that specifically produce GNRH [71] and it is conceivable that VEGF is required to sustain specific classes of neurons also in the adult brain.

In most cases, it is not known whether the plethora of established and newly discovered effects of VEGF on neuronal activity is direct, taking place via VEGF binding to neuronally expressed VEGF receptors, or indirect, i.e. via inducing endothelial or glial cells to secrete a factor acting on neuronal cells. The notion of a direct response of neurons to VEGF is supported by reports on expression of the high affinity VEGF receptors FLK1 and FLT1 and of the auxiliary non-signaling receptor neuropilin on neuronal cells [32, 71–74]. Also consistent with a direct mechanism are findings showing that VEGF can induce neurite growth in cultured cortical neurons [75–77]. In vivo, however, a rigorous proof for a direct effect of VEGF on neurons may necessitate neuronal-specific ablation (or functional knock-down) of each VEGF receptor. Direct VEGF signaling in neurons is poorly understood. A newly described mechanism is a VEGF-promoted clustering of Flk1 with several NMDA receptor subunits inducing their phosphorylation and downstream signaling [27, 55]. While this mechanism was shown to operate during VEGF-induced granule cell migration, it is possible that interaction of VEGF receptors with neurotransmitter receptors and ion channels may also play a role in VEGF-induced modulations of neuronal plasticity and memory changes on which not much is currently known.

In addition to direct VEGF signaling in neuronal cells, mediation by different type of glial cells should also be considered. Astrocytes, for example, which are known to profoundly affect neuronal activity, [78] were shown to also express VEGF receptors [50]. Taking into account that VEGF affects monocytes in many ways, including inducing their recruitment, positioning within the target organ and re-programming therein for secretion of multiple factors [79], a possible mediation by microglia of the VEGF effects on different facets of neuronal activity should also be examined.

Findings that some actions of VEGF are mediated by angiocrine factors elaborated by the VEGF-activated endothelium in other organs [80, 81], has prompted a search for VEGF-induced angiocrine factors operating within the adult brain. Putative angiocrine factors are currently under intensive research (for a recent review see [82]) and assigning a role for particular angiocrine factors in any VEGF-induced phenotype remains a major challenge.

VEGF may impact the neurovascular unit through vascular modulations not involving angiocrine factors. Thus, VEGF may impact neuronal activity via its well-documented affects of VEGF on vascular permeability, on vascular tone and vasodilatation, as well as by altering extracellular matrix (ECM) composition. The vascular ECM composition, particularly of integrins and cognate ligands, greatly impacts NSC-niche interactions [43]. As described above some consequences of upregulated VEGF might be simply be due to its angiogenic activity and increased microvascular density, as shown for the hippocampal NSC niche [5]. Increased MVD may, in turn, increase perfusion and local distribution of blood-borne substances.

Finally, the highly heterogeneous nature of the brain may pose yet another level of complexity. It is conceivable, for example, that different neurons residing in different areas of the brain have different inherent properties which, in turn, might dictate differential neuronal responses to VEGF. Yet, the idea of region-specific heterogeneity of brain vasculature and differential vascular responses to VEGF and other factors is of great importance when studying VEGF effect on brain functions, and data on differences between endothelium in cerebral sub-regions has only recently started to accumulate [42, 53, 60].