The normal blood-brain barrier
The cerebral capillary endothelial cells that constitute the BBB differ from those of the peripheral circulation in that they form tight, impermeable junctions between adjoining cells. In addition, they are devoid of fenestrations and transcapillary channels and express a very low rate of pinocytotic activity. This continuous barrier means that, for all but a few compounds, passage into the brain from the circulation is determined by the solutes' lipid solubility. However, for the brain-essential nutrients such as glucose and some amino acids this is not the case, as they enter the CNS at rates greater than would be predicted by simple diffusion. In order to meet the high metabolic demand of the brain these substances are transported across the BBB on specific transport systems that facilitate their entry into the brain. Although for convenience the BBB is often considered a single membrane it actually consists of the luminal plasma membrane, the cytosol and the abluminal membrane of the endothelial cell. BBB carrier systems, which are saturable and may be competetively inhibited, have been described for glucose, amino acids, ketone bodies, nucleotides, choline and thiamin. This strict regulation in the entry of water-soluble molecules enables the vasculature to control carefully the composition of the brain extracellular fluid (ECF) that bathes the cells of the CNS. Maintenance of a controlled environment is especially important in the CNS, where small fluctuations in ECF composition may, amongst other deleterious actions, seriously effect the sensitivity of synaptic signalling [3].
For nonpolar substances the general rule applies that the greater the lipophilicity the greater the brain uptake. A notable exception to this rule, however, is the lipid soluble immunosupressant, cyclosporin, which fails to enter the brain [4] despite the fact that it readily appears to cross the blood-retinal barrier. Lipophilic substances may also be excluded from the brain parenchyma by the presence of a membrane glycoprotein, P-glycoprotein (Pgp) situated at the BBB and other blood-tissue barrier sites [5].This glycoprotein is thought to have evolved as an energy-dependent effiux pump and is responsible for multidrug-resistant properties found in some cultured cells and tu-mours. It has been suggested that the brain is protected from the entry of some harmful nonpolar molecules, such as certain xenobiotics, by Pgp transporting substances out of the endothelial cell and back into the circulation.
Enzymes within the endothelial cytosol that are capable of metabolising lipophilic molecules have also recently been reported [6]. In view of such findings it is clear that the traditional role of the BBB as a barrier to hydrophilic molecules must be redefined to include a limited capacity to restrict the entry of some lipophilic molecules.
Although most macromolecules are prevented from entering the brain by the BBB, receptor-mediated processes have been described for transferrin, insulin, insulin-like growth factor, and low density lipoproteins. The rate of uptake of these substances is low and to date there is scant evidence for transcytosis of these compounds across the cerebral endothelium.
The blood-brain barrier and disorder of the CNS
In many diseases that affect the brain, the cerebral endo-thelium plays an active part in the disease process with the BBB becoming disrupted, or modified, in such a way that there is a dramatic increase in vascular permeability. A widely reported consequence of this phenomenon is the propagation of vasogenic brain oedema [7, 8] and the possible development of secondary brain damage [9]. Diseases in which increases in BBB permeability have been reported include neoplasia, ischaemia, hypertension, dementia, epilepsy, infection, multiple sclerosis, experimental allergic encephalomyelitis and trauma. Irrespective of whether BBB dysfunction is the cause or a consequence of a particular disease process our understanding of the cellular mechanisms that lead to a disrupted BBB is limited.
Control of blood-brain barrier permeability
There is increasing evidence that the cerebral endothelium lS under direct/influence from astrocytes, and that factors released by, or present on these cells are in part responsible for the unique characteristics oftheBBB [10-12]. In certain pathological conditions it is possible that the influence of the astrocyte and/or its factor is lost, giving rise to structural changes in the endothelium and an increase in permeability. Indeed, in some tumours and following ischaemia both intrinsic and newly formed capillaries are often devoid of BBB characteristics and resemble endothelium from peripheral tissue [13, 14]. This phenomenon may also occur in some types of cerebral graft [15].
There is increasing evidence to suggest, however, that in many diseases of the CNS, induction of barrier dysfunction may be brought about by the release or activation of mediator substances from damaged or activated cells. These vasoactive compounds may originate from a number of different sources including the blood, the cells of the CNS or from the endothelial cell itself.
Blood-borne mediators. It has recently been demonstrated that the blood-borne products of the kinin and complement system, bradykinin and complement C3a respectively, increase the permeability of the BBB in experimental animals [16, 17]. Similarly, the cells of the blood have also been shown to be a major source of vasoactive com-pounds; platelets, basophils, macrophages and active T-lymphocytes are known to release substances that disrupt the BBB, such as histamine [18], serotonin [19] and leukotrienes [20]. They may also release a number of other compounds including prostaglandins, tumor necrosis factor (TNF), interferon (IFN-o~ and IFN-~3) and interleukin-1 (IL-1) that act upon cerebral endothelium.
Tissue mediators. Damaged and perturbed cells of the CNS, including mast cells, are an alternative source of vasoactive compounds, releasing bradykinin, histamine and IL-1. The release of arachidonic acid and its meta- bolites, the eicosanoids, from pathological tissue, has stimulated considerable interest. This fatty acid is liberated from membrane phospholipids of damaged cells of the CNS by the activation of phospholipase A2 and C, possibly by the release of free radicals or a rise in intracellular free calcium ([Ca2+]i). The concentration of arachidonic acid in ECF increases substantially in cerebral lesions and, along with its metabolites the prostaglandins and leukotrienes, increases the permeability of the cerebralvasculature [8, 9,21, 22]. The presence of free radicals may also play an important part in damage to the BBB [23]. A further source of vasoactive substances is the cerebral endothelium itself, which has has been shown to liberate, amongst other relevant substances, arachidonic acid metabolites [24].
The release of a mediator of barrier disruption into the circulation following thrombosis of the middle cerebral artery has also recently been described [25, 26]. Although the source of the vasoactive substance is unclear it is likely to be derived from either endothelial cells or from a blood-borne component such as the platelets.
Cellular mechanisms of increased blood-brain barrier permeability
Our understanding of the cellular mechanisms that initiate changes in BBB permeability is limited. Disruption of the BBB following hypertonic shock and cryostatic injury has been reported to be mediated by a dramatic increase in ornithine decarboxylase (ODC) activity and polyamine syn-thesis [27-29]. It appears likely that arachidonic acid metabolism may also be involved as BBB dysfunction could be suppressed with dexamethasone (an inhibitor of phospholipase A2) and aspirin (acyclooxygenase inhibitor)[28]. Furthermore, alterations in ([Ca2+]0 have also been implicated in the induction of BBB permeability. Indeed, it has been suggested that one of the fundamental consequences of many of these chemical mediators is to increase Ca2+influx into the endothelial cell which stimulates cyclic nucleotide production, which in turn leads to pinocytosis and vesicular transport [8, 30, 31]. A rise in ([Ca2+]i) in endothelial cells, induced by receptor binding (e. g. by his-tamine or serotonin) or by other processes, appears to be a two-step process, calcium initially being released from intracellular stores, possibly mediated by the formation ofin-ositol triphosphate, followed by a slower influx of extracellular calcium via gated Ca 2+ channels, nonspecific cation channels, or Na +-Ca 2+ exchange.
Recently it has been demonstated that tight junctions can be induced in cultured brain endothelium if the cell concentration of cyclic AMP is increased with agents such as forskolin [32]. Furthermore, it was shown that increasing cyclic AMP did not incrase the amount of pinocytosis, contrary to previous reports [8, 30]. By decreasing the levels of cyclic AMP in vivo, Rubin et al. were also able to affect the integrity of tight junctions and increase the de-livery of compounds normally excluded from the brain.
Route of blood-brain barrier leakage
Cerebral microvessels from pathological tissue often re- semble nonbraln endothelium, with significant increases in the number of vesicular profiles, abnormal junctions and attenuated areas of cytoplasm. However, the ultrastructural correlate of BBB dysfunction remains a contentious issue. The extravasation of plasma proteins may occur via a number of different routes, namely, diffusion through altered tight junction, induction of fluid-phase or nonspecific pinocytosis and transcytosis, formation of transendothelial channels or by disruption of the endothelial cell membrane. Of course, none of these paths are mutually exclusive and they may occur in combination. However, the problems associated with their ultrastructural detection are many. Firstly, an increase in the number of microvesicular profiles does not necessarily indicate the existence of transcytosis, and secondly, the two-dimensional appearance of these structures may result from sections through caveoli, tube-like structures in continuity with the luminal or abluminal membrane. Indeed, if trans-cytosis does occur there is little evidence for the direction, if any, in which vesicles are transported. Furthermore, ultrastructural evidence of disrupted tight junctions is scarce, due to the improbability of finding, in a single section, electron-dense markers filling the entire length from the luminal to abluminal aspect of the cell. Finally, the presence of channels or canalicular systems at the pathological BBB is controversial.
Oedema
In nearly all diseases in which the BBB is damaged the most notable consequence is cerebral oedema. The classical work of Klatzo in the 1960's led to the classification of
brain oedema into two major types, vasogenic and cytotoxic [7, 33]. The original definition of vasogenic oedema as extravasation of plasma proteins resulting from the
breakdown or absence of a BBB has remained essentially unchallenged. However, in recent years the mechanisms involved in the production of cytotoxic oedema have become better understood, leading to one suggestion that it be reclassified into ischaemic, osmotic and interstitial oedema [34]. The metabolic processes involved in the genesis of"cytotoxic" oedema may ultimately lead to damage of the cerebral endothelium, extravasation of plasma proteins and the production of vasogenic oedema. With both types of oedema frequently being present, it is often impossible to classify a particular case of oedema as being one or the other [8, 33].
Once established, vasogenic oedema can spread, under the influence of hydrostatic and oncotic pressure, into areas surrounding the immediate site of BBB disruption. The size of the BBB lesion and arterial pressure also play a critical role in the degree of spread of oedema fluid. Vasogenic oedema may ultimately resolve if the BBB ceases to leak plasma proteins. This may occur as a result of endothelial cell regeneration, repair of tight junctions and cessation of pinocytosis or by the surgical removal of abnormally permeable vessels in cerebral tumours.
Tumours
It is uncommon for the vasculature of cerebral turnouts to posseses those properties normally associated with the BBB. In many cases the capillary endothelium of tumour vessels is highly abnormal, expressing a high degree of fenestrated regions, vesicles, open junctions and fragmented basal lamina which leads to a considerable increase in the permeability of the tumour vascular bed. The lack of barrier properties may be due either to new vessels that have not taken on the properties of the normal BBB or to existing vessels changing their properties. Indeed, it has recently been shown that cultured neoplastic glial cells, but not normal glia, produce a permeability factor that is capable of opening the BBB in vivo [35].
When considering the permeability of the tiamour vascular bed it is essential to remember that there is an enormous degree of heterogeneity between different tumours types. Moreover, within a single tumour there may be a considerable difference in the extent of vessel leakage during its development and within different regions. Heterogeneity within tumours will often reflect the morphological diversity, ranging from necrotic regions, through solid turnout, to the invading front.
Despite the often extensive abnormalities of the vasculature of cerebral tumours the endothelium may still present a significant barrier to polar solutes. Indeed, hyperosmolar opening of the blood-tumour barrier (BTB) has been used extensively in the laboratory, and to a limited extent clinically, to improve the delivery of water-soluble therapeutic agents to neoplastic cells.
It is a widely held belief that steroids resolve vasogenic oedema associated with cerebral turnouts by reducing the permeability of the BTB. Although this may be true for a number of tumour types there is some question as to whether it holds for all tumours. In one study, despite striking clinical improvement of A15A5 glioma bearing rats following dexamethasone treatment, no alteration in the permeability of the BTB to [14C] mannitol was detected [36]. Other studies using different tumour models and markers have demonstrated changes in the BTB with steroid treatment [37, 38], a finding also seen in humans using positron emission tomography [39]. It is likely that these anomalies may be explained by differences in the methods used for assessing BTB permeability, in the tu-mour type and in the steroid used [40].
Ischaemia
Breakdown of the BBB following an ischaemic episode has been well documented although the extent and duration of opening may depend upon plasma glucose status, duration of ischaemia and degree of reperfusion. If reflow occurs opening of the BBB may be short-lived, recovering within 24 h, whereas following permanent occlusion dysfunction of the BBB surrounding the infarct may persist for weeks. Early leakage of the barrier may be due to disruption of the existing BBB whilst a more pronounced and long term breakdown may coincide with the growth of new blood vessels with immature tight junctions and increased transcellular transport. Contrast enhancement
observed clinically with computed tomography (CT) may often have a ring-shaped configuration, reflecting the pattern of immature capillary growth from the margin of the infarct to the centre [14]. The early phase may not be recorded by CT as transient reactive hyperaemia in the blood vessels adjacent to the infarct could restrict the extravasation of contrast medium.
Lactic acidosis and a drop in pH following ischaemia have been implicated in the pathogenesis of disruption of the BBB [41-43], although the mechanism has not been established. It has been suggested, however, that barrier dysfunction in the ischaemic brain is mediated in part by microvesicular transport, an active process [44, 45], as low pH-induced breakdown can be inhibited by depleting the brain of energy reserves[46].
