In mammals, the terminal CNS vascularization occurs postnatally via active angiogenesis, and thereafter this vasculature remains essentially quiescent under basal, non pathological conditions. In the present study we show that 1) the vascularization of hypothalamic magnocellular nuclei can be modified throughout adulthood via local angiogenesis induced by hyperosmotic stimuli and 2) such local angiogenic events are related to the expression of high levels of VEGF by magnocellular VP and OT neurons.
Osmotic stimuli induce proliferation of SON capillary endothelial cells
More than 28 years ago, it has been established that prolonged stimulation of hypothalamic VP and OT neurons by hyperosmotic stimuli induced substantial proliferation of glial and endothelial cells within the supraoptic nucleus . Although they fully confirm these previous data, our present data indicate that the cell proliferation occurring locally within hypothalamic magnocellular nuclei mostly involve endothelial cells, whereas proliferation of both endothelial cells and astrocytes was previously reported. A first explanation for such discrepant data is that the proliferation rate of astrocytes has been underestimated in the present study. Proliferative astrocytes were identified here as those cells immunostained for the astrocytic marker GFAP that exhibited a BrdU-labeled nucleus. Since newly formed astrocytes only express very low levels of GFAP[31, 32], it is thus possible that proliferative astrocytes appeared GFAP-negative. In all the sections examined however, BrdU-labeled nuclei were preferentially located within the core of the SON, whereas the SON astrocytic cell bodies are localized to the ventral border of the nucleus. The differential evaluation of astrocytic proliferation may rather be related to marked differences in the methods used to label proliferative cells. In the present study, BrdU was injected 5 hours before the fixation of the rats (therefore labeling the cells that have undergone a process of cell division during this short period), while 3H-thymidine was previously injected twice daily during the whole duration of the hyperosmotic stimulation (14 days). It is thus very likely that under the present conditions, only those cells exhibiting high rate of proliferation were BrdU-labeled, whereas all the cells that have proliferated all along the period of osmotic stimulation were 3H-thymidine-labeled in the previous study. This clearly indicates that, in the SON of osmotically stimulated rats, the rate of proliferation of SON endothelial cells highly surpasses that of astrocytes.
We also show that within the SON of osmotically stimulated rats, high rate proliferation of capillary endothelial cells is accompanied by lower proliferation rate of NG2-labeled cells. Interestingly, the proteoglycan NG2 has been found to be expressed by vascular pericytes surrounding the nascent vessels [33–35]. Although the numerous NG2-labeled cells dispersed throughout the adult rat brain were initially identified as oligodendrocyte precusors , it is now generally admitted that these cells represent multipotent progenitor cells . It can thus be assumed that at least part of these cells differentiate into vascular pericytes that participate in the formation and/or stabilization of new capillaries [36–38].
SON angiogenesis reversibly modifies the local vasculature
The proliferation of endothelial cells, occurring either under physiological or pathological conditions, is generally associated with angiogenesis that leads to the formation of new capillary vessels. The idea that the proliferative response detected within the SON corresponds to local angiogenesis is strongly supported by the present data demonstrating that prolonged osmotic stimulation induced 1) pronounced modifications of the anatomical organization of SON capillary vessels, with both an expansion of a capillary network throughout the increased SON volume and an increase in the network density, and 2) an increased expression by SON capillaries of nestin and vimentin, two intermediate filament proteins highly expressed by newly formed endothelial cells [19, 20, 22, 23]. Since prolonged osmotic stimuli induce strong activation of the functional activity of VP and OT magnocellular hypothalamic neurons , it can reasonably be assumed that the formation of new vessels within the SON will increase the circulating metabolic support to these neurons. A particularity of magnocellular neurons is that the size of their cell body increases during prolonged osmotic stimulation, leading to a progressive expansion of the volume occupied by the corresponding hypothalamic nuclei . Newly formed vessels may thus also contribute to the vasularization of the expanded nucleus.
An intriguing question raised by the present findings concerns the fate of that SON new vessels formed during the period of osmotic stimulation. Our data clearly show that the density of the capillary network and the phenotypic expression of SON capillary vessels returned to control levels several days after the cessation of the osmotic stimulus. It is thus very likely that newly formed vessels are progressively deleted following rehydration. In control, normally hydrated rats, the SON vascularization however always remains particularly dense as compared with the surrounding regions, and the SON capillary vessels always express high levels of nestin. It can thus be assumed that, even under basal physiological conditions, new vessels are continuously added to the SON vasculature via sparse angiogenic events.
Neuronal VEGF is the major stimulus for SON angiogenesis
The fact that osmotic stimulus induces reversible angiogenesis within the hypothalamic magnocellular nuclei strongly suggests that local secretion of potent angiogenic factor(s) occurs within these nuclei. Although a variety of angiogenic factors have been identified in the CNS, VEGF is generally recognized as the major factor involved in the processes of angiogenesis [25, 39, 40]. That endogenous VEGF plays a major role in the osmotically induced angiogenesis was suggested by the present findings that 1) contrasting with most CNS neurons, hypothalamic magnocellular neurons continue to express VEGF throughout adulthood, 2) this expression is highly increased by osmotic stimulation, and 3) osmotically induced angiogenic events are impaired when VEGF is inhibited by dexamethasone or by a blocking antibody.
Dexamethasone has been shown to inhibit VEGF expression in a large variety of tissue [27–30]. We show here that dexamethasone treatment of osmotically stimulated rats dramatically decreases the VEGF immunostaining associated with magnocellular neurons and, to a lesser extent, astrocytes. Our data further indicate that such decreased VEGF immunostaining of hypothalamic magnocellular neurons correlates with both a potent inhibition of cell proliferation within these nuclei, and a strong decrease of the nestin immunostaining of capillary vessels (not shown). This obviously supports the idea that endogenous VEGF is at least partly responsible for the local angiogenic events occurring within the hypothalamic magnocellular nuclei of hyperosmotically stimulated rats. Since hypothalamic magnocellular neurons of the adult rat express the glucocorticoid receptor , it is very likely that the dexamethasone effect observed here directly results from a direct repression of the VEGF gene. Glucocorticoids are however known to repress a variety of genes , which may also interfere with the angiogenesis described here. In order to demonstrate that endogenous VEGF is directly responsible for SON angiogenesis, we lastly performed a local infusion of a neutralizing antibody to VEGF that has been successfully used in previous studies to block endogenous VEGF activity in the rodent, both in peripheral organs [43, 44]and within the brain . An advantage of this antibody raised in the goat is that its diffusion throughout the brain parenchyma can easily be controlled. In the present study we show that the proliferative response to hyperosmotic stimulus was strongly inhibited within those hypothalamic nuclei that were located within this diffusion zone, but was not affected within the nuclei contra-lateral to the implanted cannula. Moreover, numerous proliferating cells were always detected along the border of lesional cavity induced by the cannula implantation, strongly suggesting that the blocking antibody specifically abolishes the VEGF-dependent cell proliferations, but has only poor effect on the proliferation of glial cells induced by CNS injury [46, 47].
Although VEGF was also detected within SON astrocytes, our present observations suggest that the angiogenesis described here is preferentially related to the cytokine of neuronal origin. Within the SON of osmotically stimulated rats, BrdU-labeled proliferating cells were indeed always restricted to the core of the SON containing the VEGF-labeled neuronal cell bodies, and were scarce in the ventral portion of the nucleus containing the labeled astrocytes (Fig. 8 A–B). Moreover, osmotic stimulus was found to increase the expression of VEGF mRNA in the only SON neurons. In a series of previous studies, astrocytic VEGF has been associated with the angiogenesis occurring in various CNS pathological processes [48–50]. It could thus be assumed that, in the SON as in the other brain regions, neuronal VEGF constitutes the stimulus for physiological angiogenesis, whereas astrocytic VEGF is essentially involved in those angiogenic events occurring under pathological or traumatic conditions.