Functional and morphological consequences of removing astrocytes from the in vivo blood–brain barrier

Our purpose was to attempt to determine the role of astrocytes in the maintenance and repair of the blood–brain barrier in adult F344 rats. Systemic administration of the glyceraldehyde‐3‐phosphate dehydrogenase inhibitor S‐alpha‐chlorohydrin produces selective destruction of astrocytes within specific brain stem nuclei (Cavanagh et al. Neuropath. Appl. Neurobiol. 19, 1993). At appropriate dose levels (140–160 mg/kg i.p.) other brain areas are undamaged, which allows the full time course of the response of sensitive areas to be studied in rats showing relatively little functional impairment. All experiments were carried out in accordance with UK Home Office requirements. Lesion development was followed by in vivo Magnetic Resonance Imaging, by post mortem immunohistochemistry and confocal microscopy, or electron microscopy. Barrier function was studied by the permeation of fluorescent conjugated dextrans of 3–500 KDa molecular weight given intravenously 1 h before perfusion with fixative, and visualised by confocal microscopy (Guérin et al. Neurosci. 103, 2001). Astrocytic death was seen throughout areas such as the inferior colliculi and vestibular nuclei over 6–24 h and was followed by secondary neuronal death and a microglial reaction from 48 h. After a delay of about 12 h following the loss of astrocytes, endothelial cells become permeable to dextran for up to 4 d and then recovered. This recovery was seen in the absence of direct contact with astrocytes, as determined by immunohistochemistry and electron microscopy. Since chlorohydrin does not increase endothelial monolayer permeability in vitro (Romero et al. Neurotoxicol. 18, 1997), we hypothesise that this transient loss of barrier function in vivo represents a partial dedifferentiation in response to loss of astrocytic contact. The subsequent restoration of endothelial barrier function in areas devoid of astrocytes represents the first in vivo demonstration that direct astrocytic contact is not needed for establishment of the blood–brain barrier.