Keywords

1 Introduction

The blood brain barrier (BBB) is a physical and metabolic barrier between systemic blood circulation and the central nervous system (CNS). This concept originated over a hundred years ago when Paul Erlich and subsequently E. E. Goldman published studies demonstrating a limited penetration of certain substances (i.e., dyes) from blood to brain and from brain to blood (see review 21). Present notion about the CNS includes other barriers (i.e., blood-cerebrospinal fluid (BCSF) and blood-retinal barrier (BRB)). The morphologic localization of the barriers was advanced by light microscopy and electron microscopy using contrast media (i.e., horseradish peroxidase, etc.). The capillary endothelium in the brain and retina was demonstrated at the site of the BBB and BRB, whereas the choroid plexus represents the site of BCSF. Many additional studies showed that the passage of substances through the BBB depends on the size and lipid solubility. It has also been demonstrated that non-diffusible polar substances (such as nutrients) require specific transporters to cross this barrier. These transporters were characterized as facilitated or active (energy-linked to ATP) transporters dependent on the given substance’s intrinsic requirement namely it’s downhill or uphill concentration gradient. In addition, both in vivo and in situ studies showed the existence of a metabolic barrier (i.e., enzymatic) provided by endothelial cells that limits the passage of molecules from blood to brain and brain to blood. These observations were reinforced and advanced by the ability to separate the capillaries and microvessels from the rest of the brain parenchyma as well as the subsequent culturing of endothelial cells and other brain cells. It became evident that the cerebrovascular tree represents a separate and distinct biochemically active compartment. Moreover, the understanding of the BBB has been increased by the recently acquired molecular insight into its structural components. This function has been attributed to the endothelial cells, the main constituent of the BBB. Thus, these cells originally thought of as physical barrier were shown to have a variety of biological activities that were responsible for BBB properties and could be influenced by other cells. This brings us to take into consideration the likely contribution of cells adjacent to the endothelium and broadens the previous supposition regarding the just endothelial cells as BBB and to redefine it as the so-called Neurovascular Unit (NVU).

The great numbers of past and recent reviews of these barriers (1, 5, 17–19, 21, 23, 26) provide the basis for this communication. However, this report will present an abrogated review of the BBB. It will focus on past as well as the latest available studies that are relevant to the understanding of processes involved in the development and progression of ischemic brain edema.

2 The Current View of the BBB

2.1 The Cellular Components of the BBB

Morphological and functional characteristic of the BBB reside in the capillary and microvascular endothelium (including venules and arterioles), pericytes, astrocytes and neurons, which are collectively designated as the NVU (Fig. 1a). In the last decade the identification and demonstration of subcellular components of the capillary and microvascular endothelium as well as the surrounding cells expanded our comprehension of the BBB. Particularly the emerging knowledge indicating that the functional role of maintaining the BBB integrity is not limited to endothelial tight junctions; luminal and basement membrane along with the extracellular matrix (ECM) are also involved. In general, there is a functional interplay among all the cellular elements of the NVU that is responsible for the homeostasis of the brain environment. Under physiological condition the BBB safeguards a constant supply of nutrients (i.e., oxygen, glucose and other substances) for the brain and steers inflammatory responses to the changes in the neighborhood.

Fig. 1
figure 1

Blood-brain barrier and tight junctions. (a) Cross section displaying schematically the localization of cellular and sub cellular components of the NVU (b) Schematic representation of proteins associated with tight junctions (modified from Ballabh et al. (2004) (1)).

Full size image

Endothelium Endothelial cells (EC) of the BBB differ from the endothelial cells from the rest of the body by the presence of extensive tight junctions, a great number of mitochondria, an absence of fenestration and sparse pinocytotic vesicular transport. The tight junctions limit the flux of hydrophilic molecule across BBB while lipophylic substances (i.e., oxygen and CO2) freely diffuse through the plasma membrane along their concentration gradient. Brain endothelial cells produce agents and factors as well as express various receptors that may be involved in autocrine and paracrine regulation of the microvascular function of the brain. In general, the substances (e.g., prostacyclin, nitric oxide (NO), and adenosine) produced by these cells are considered to be cytoprotective. However, several other agents that are formed in endothelial cells (i.e., endothelin-1 (ET-1), angiotesin II, thromboxane, leukotrienes, platelet-activating factor, and superoxide radicals) impair perfusion, alter BBB permeability and/or mediate cellular injury when released in excess. Many of these substances which modulate the secreatory functions and reactivity’s of the vascular endothelium can also be released from adjacent cellular elements including other vascular cells, circulating blood cells and brain cells.

Tight Junctions The junctional complexes of endothelial plasma membranes consist of tight and adherent junctions (TJ and AJ, respectively) (Fig. 1b). The TJ are composed of three integral membrane proteins, namely claudins, occludins and junctional adhesion molecules as well as cytoplasmic accessory molecules (i.e. ZO1, ZO2, ZO3, cingulin and others). The cytoplasmic protein links the membrane protein to actin, the primary cytoskeleton protein that maintains the structural functional integrity of EC.

Claudin 1, 2 and 3 (among about 20 known members of this family) are 22-kDa phosphoproteins. They represent the main components of TJ and are localized exclusively at TJ. Claudins bind homotypically to other claudins on adjacent EC to form a primary seal. The carboxyl terminal of claudins binds to cytoplasmic proteins such as ZO-1 ZO2 and ZO3.

Occludin, a 65-kDa phosphoprotein, has four transmembrane domains, with two cytoplasmic domains (a long COOH-terminal and a short NH2-terminal. The cytoplasmic domain of occludin is directly associated with ZO proteins. The two extracellular loops of occludins and claudins originating from neighboring cells form the ‘paracellular’ barrier of TJ. Occludin appears to be a regulatory protein that can alter paracellular permeability. Occludin phosphorylation occurs at serine and threonine residues that control the intracellular distribution of this TJ junction protein and subsequent barrier properties. The decreased phosphorylation status of occludin (by inhibition of protein kinase C) correlates with a rapid fall in transendothelial resistance. On the other hand, TJ phosphorylation is associated with increased BBB permeability, which is induced by vascular endothelial growth factor (VEGF); a similar observation has been made with monocyte chemotactic protein (MCP-1).

Junctional adhesion molecules (JAM) is a 40 kDa protein belonging to the immunoglobulin superfamily that binds directly to ZO-1. It has a single transmembraneous domain and the extracellular part has two immunoglobulin groups that are formed by disulfide bonds. JAM-1 and -2 but not JAM-3 exist in brain blood vessels. JAM-1 localizes with actin at the cell-to-cell contacts and is involved in cell-to-cell adhesions. JAMs are also expressed by circulating blood cells (i.e., monocytes, neutrophils, subsets of lymphocytes and platelets). Although their function is not clear, they are reportedly (in vitro studies) involved in transmigration of monocytes through EC.

Cytoplasmic accessory proteins include zonula occluden proteins (ZO-1, -2 and -3), cingulin, and others. The zonula occluden proteins have similar amino acid sequences and belong to the family of proteins known as membrane associated guanylate kinase-like proteins (MAGUL). They function as protein binding molecules and play a role in organizing proteins at the plasma membrane. Many signaling pathways and proteins have been implicated in regulating the tight junction assembly (i.e., Ca2+, protein kinase A and C, G-protein, calmodulin, cAMP, and phospholipase). In addition, small G-proteins and GTPase have been suggested to play a regulatory role of the structure and function of tight junction.

Adherens junction (AJ) consists of membrane protein, cadherin, which joins to the actin cytoskeleton via intermediary proteins know as catenins to form adhesive contact between cells. AJ are known to interact with ZO-1and catenins and influence the TJ assembly.

2.2 Endothelial Membranes

Both luminal and basal membranes are known to be sites of enzyme receptors and transporters. The advance of specific immuno-markers and the ability to separate these membranes enable better localization of their components.

2.3 Examples of Luminal Membrane Constituents at the BBB

Glycocalyx is a negatively charged surface coating the luminal EC membrane and consisting of proteoglycans, glycosaminoglycan and absorbed plasma proteins. It participates in maintaining vascular permeability. It contains many enzymes that might be involved in vasoactive processes (i.e., extracellular superoxide dismutase (SOD) that converts oxygen radicals to H2O2) and is bound to heparin sulfate proteoglycans within glycocalyx. Damage leads to a pro-oxidant state (i.e., ischemia and/or trauma) and the shearing of the glycocalyx permitting attachment of leukocytes (i.e., inflammation).

Among molecules involved in transcytoplasmic route for blood borne proteins are clathrin and caveolae (23). The former takes part in receptor-mediated endocytoses and the latter represents clathrin-independent endocytotic pathway.

Caveolae and Caveolin The caveolae consist of microdomains of invaginations in plasma membranes that are involved in receptor-mediated uptake and regulation of many dynamic events (i.e., signal transduction, cell growth, apoptosis and lipid metabolism). Their functional activities are due to caveolins, a family of integral membrane proteins (isoforms 1, 2 and 3). In the brain, caveolin-1 expression was demonstrated on endothelial cells, pericytes and perivascular astrocytes, while caveolin-3 was muscle specific. The expression of caveolin was co-localized with endothelial nitric oxide synthase (eNOS) and plasma glycoprotein (P-gp) in the endothelial caveolae (8). Reports also indicate that caveolin-1 has the ability to negatively regulate NO production and proangiogenetic factors (7). Caveolin deficiency increases cerebral ischemic injury as demonstrated in caveolin-1 knockout mice, which also showed increased levels of eNOS and apoptosis along with cerebral volume of infarction (7). Studies in vitro also demonstrated that caveolin-1 regulates expression of tight junction-associated protein, occludin (20).

P-glycoprotein (P-gp) is a major representative of the ATP-dependent efflux transport family and multidrug resistant-associated proteins (MRP). Cationic and zwitterionic compounds are preferentially transported by the P-gp whereas ionic compounds are transported by MRP. The disputed localization of P-gp was recently clarified by demonstration of its expression on capillary endothelial cells comprising the BBB as well as pericytes and astrocytes; this transporter was colocalized with caveolin 1. P-gp expression was demonstrated along the nuclear envelope in caveolae, cytoplasmic vesicles of the Golgi’s complex and endoplasmic reticulum. Increased intestinal expression of P-gp reduces the absorption of drugs and it has been suggested that P-gp on EC comprising the BBB might likewise regulate drug transport processes in CNS at both cellular and subcellular levels. The interaction of caveolin and P-gp also negatively affects the function of P-gp; a reduced interplay between the two proteins separately increases the transport activity of P-gp.

2.4 Examples of Basement Membrane Components at the BBB

The role of the basement membrane is multifold in that it includes maintenance of capillary and microvascular morphology, cell adhesion interactions and prevention of vascular plasma protein leakage. It is constructed from extracellular matrix proteins (i.e., collagen type 4, laminin) and is regulated by matrix metalloproteinases (MMP) and plasma.

Matrix metalloproteinases (MMP) are members of zinc-dependent endopeptidases that are active at neutral pH and are inhibited by proteins known as tissue inhibitors of metalloproteinases (TIMPs) (18, 19). The blocking activity of TIMPs occurs through high affinity noncovalant binding to MMP catalytic domains. Both MMPs and TIMPs are implicated in many pathological processes (i.e., ischemia and trauma). They are also important in recovery by facilitating angiogenesis and neurogenesis.

Extracellular matrix (ECM) of basal laminin separates the endothelium from the astrocyte. It provides an anchor for the EC through interactions of laminin, collagen type 4, fibronectin and other matrix proteins with endothelial integrin receptors. The capillary endothelial cells, microvascular smooth muscle cells and pericytes with surrounding astrocytes cooperate to generate and maintain the basal laminal membrane and the unique function of endothelial cells (4, 13).

Matrix adhesion receptors such as integrins and dystrophan represent two types of ECM receptor families that are associated with the microvasculature in the CNS. Both are present on the opposite side of endothelium and astrocytes (4, 13). Their position contributes to endothelial cell matrix adhesions and maintenance of the connection of the astrocytic endfeet to the abluminal endothelial surface. Selective integrins are also expressed on neurons, microglia and oligodendroglia.

Integrins are cell surface transmembranous, noncovalently linked, α/β heterodimers that recognize specific matrix ligands. Functionally they connect the matrix with the cytoskeleton and are implicated in regulating cellular responses by transmitting extracellular stimuli to intracellular signals and generating increased receptor specificity.

The dystroglycan is a single α–β heterodimeric transmembranous receptor that forms a physical link between the intracellular cytoskeleton and the ECM. α-dystrophan of the extracellular glycosylated subunit binds to ECM proteins laminin, perlecan and agrin. The intracellular carboxy-terminus of β-dystrophan binds to cytoskeleton proteins (dystrophan and utropin). The ‘dystroglycan complex’ shares laminin as a ligand with a number of integrin receptors (including α/β1, α3/β, α6/β1 and α6/β4).

2.5 Cell Components of the BBB

Smooth Muscle These cells surround the endothelium of arterioles showing tight junction morphology. They are known to play a role in the regulation of cerebral blood flow (CBF). However, to the best of our knowledge, the full and distinct contribution of smooth muscle (vascular mural cells) to the BBB is unknown. Most of the information regarding the presence of various receptors that could be or are involved in regulating the microvascular tone and/or CBF (i.e., adrenergic, prostaglandins, NPY and ET-1) was obtained from in situ or in vivo microvascular studies. Therefore it is worthwhile mentioning that in the mid-1980s our comparative studies of separately cultured endothelial cells, smooth muscle and glial cells derived from rat brain clearly demonstrated distinct differences in the reaction of these cells to vasoactive factors, etc. (i.e., cAMP, vasopeptides, and 5-HT). For example, the stimulation of 6-keto-PGF1α synthesis by adrenergic agonists was seen in EC but not in smooth muscle. On the other hand, angiotensin-1 and -2, as well as bradykinin, markedly enhanced the 6-keto-PGF1α in smooth muscle but had less effect on glia and little or no effect on EC (27). These studies suggested that EC and glia cells provide protection of the smooth muscle against the exposure to substances which could contribute to the dysfunction of CBF and alter BBB integrity.

Pericytes These cells are localized within the basal membrane of the capillaries and microvessels opposite the luminal face of the astrocytic endfeet. Pericytes have been implicated in providing structural support and vasodynamic capacity to the vessels (5). The emerging observation suggests that these cells regulate endothelial proliferation, migration, differentiation, and vascular branching. In vitro studies also suggest that pericytes can stimulate the astrocyte evoked BBB ‘tightness’ and in this way may contribute to the BBB integrity.

Astrocytes The astrocytic endfeet envelop the capillary and microvascular endothelial basement membrane. In this way, astrocytes provide a linkage between the endothelium and neurons. The astrocyte-endothelium interactions induce and modulate the development of BBB and its distinct phenotype. Reports indicate that both the physical proximity and substances released from the astrocyte may influence the integrity of BBB permeability.

The recent demonstrations of aquaporins (AQPs) (11, 16) are of special interest due to their localization and co-localization with BBB-related factors (i.e., P-gp) as well as connection to ECM and inwardly rectifying potassium channels (i.e., Kir4.1). Moreover, keeping in mind the above mentioned endothelial co-localization of caveolin-1 expression with those of P-gp and eNOS, this strongly suggests an intricate functional relationship between the endothelial cells and astrocytes. AQPs are a family of water-selective channels that provide a major pathway for osmotically driven water transport through cell membranes and in the choroid plexus. Aquaporin 4 is expressed in foot processes of astrocytes and ependymocytes facing capillaries and CSF interfaces. AQP 9 is localized in tanycytes and astrocytic processes (and more recently in neurons). The perivascular astrocytic endfeet containing AQP4 is connected to the dystrophin-dystroglycan complex mediating the contact to ECM-laminin. Anchoring the AQP4 to cell membrane and the localization of astrocytic foot processes occurs through binding to α-syntrophin, a member of the protein dystrophan protein complex.

Neurons The innervation of brain vasculature or microvasculature and its influence on BBB was disputable even though anatomical and physiological observations were indicative of such a possibility. However, biochemical and immunocytochemical demonstration of microvascular, endothelial and/or astrocytic innervation as well as evidence of functional receptors provides support for neurogenic regulation of the BBB.

2.6 Pathophysiology

Brain edema accompanies many diseases of the CNS (i.e., ischemia (stroke) and trauma). Originally, Klatzo classified brain edema into two categories: cytotoxic brain edema (BE) and vasogenic edema (VE), which has been helpful in evaluating the brain damage and treatment (9). Brain edema (in which the BBB remains intact) is due to derangements in cellular metabolism from malfunction of the sodium and potassium pumps in the cell membrane and is characterized by: (a) an increase of intracellular sodium and water; (b) Na+/K+-ATPase failure; (c) uptake of osmotically active solution and the edema fluid rich in electrolytes; and (d) cell swelling with decreased interstitial space. Vasogenic edema is due to a breakdown of endothelial tight junctions and is characterized by: (a) an increase BBB permeability; (b) edema fluid rich in protein; and (c) increase in interstitial fluid without cell swelling. However, in most cases ischemia leads to so-called mixed edema, which expresses components of both BE and VE, and depends on the severity and duration of the insult.

Cerebral ischemia (stroke) and trauma are among many CNS diseases manifested by altered BBB integrity and formation of edema. The variety of ischemic models performed in different animal species provided disputable but valuable observations. It concerns the time course of increased BBB permeability and mechanism (vesicular and/or alteration of tight junction) that are responsible for it. However, the restricted space for this report does not permit discussion of all aspects, including BCSF involvement. Nevertheless, few of the past studies are noteworthy to mention since their results contribute to the understanding of this topic (21) and references. Firstly, the alteration of BBB to proteins depended on the duration of ischemic insult and its appearance was inversely proportionate to the time of re-established blood circulation. The observed increase in BBB permeability (to Evans blue complex or horse radish peroxidase (HRP)) was seen early (within 30 min of reperfusion) after 6 h brain injury and even later after 1 h brain ischemia (after 3 h of reperfusion). It was manifested by endothelial vesicles of various sizes and progressive presence of HRP in neuropil including swelling astrocytic processes (25). The absence of this marker was conspicuous in the tight junction. Thus, the observed changes suggested an altered BBB permeability due to increased vesicular transport that was also inducible in normal animals by injection of 5-HT. Secondly, the biphasic increment of water content in the brain was observed in the same gerbil model of unilateral ischemia (1 h) with various times of re-circulation (up to 20 h). The initial increase in water content (first phase) preceded the augmented BBB permeability to small and large molecules (i.e., sucrose, Evans blue complex, RISA, radioactive dextran). Thus, the first increment in brain water content was observed prior to the increase of small molecular markers, whereas, the brain swelling (second phase) corresponded to the increased passage of radiolabled large molecular markers. Thirdly, in addition, the inhibition of the observed increases of radiolabled 2-deoxyglucose (2-DG) uptake in the presence of unlabeled 2-DG was incomplete suggesting BBB dysfunction of glucose analog transport. It should be added that brain ischemia or anoxia per se reduces the brain uptake of glucose or glucose analog (21, 22). These observations were confirmed by in vitro studies in microvessels isolated from ischemic brains. A similar reduction of 2-DG uptake into isolated microvessels exposed to anoxia was completely reversed by replacement of nitrogen by oxygen or prevented by the addition of free-fatty acid albumin (22). This protection was partially or fully abated by individually and mixed saturated and unsaturated fatty acids to the incubation medium. It is suggestive of albumin preventing shedding of the membrane glycocalyx layer (21). Taken together these studies (ex-vivo and in vitro) implicated ischemic and anoxic microvascular membrane changes with affects on the uptake of glucose analog (caused by free-fatty acid). This supposition is supported by the observed ability of arachidonic acid (AA) to modify the endothelial membrane “fluidity” without altering the permeability of the cultured EC (24). However the exposure of these cells to both AA and H2O2 altered the EC permeability (preventable by catalase) to a greater extent than by AA alone, as evidenced by greater formation of lipid peroxidation (malondialdehyde (MDA) assay).

Interestingly, an increase in caveolin-1and -2 expression prior to that of occludin and claudin 5 in association with fibronectin, a known marker of BBB damage, was seen in microvessels after cold injury (14). These observations support the previous reports demonstrating an increased ischemic vesicular BBB transport and/or “open” tight junction which most likely depended on duration and the model of brain injury. The above noted fibronectin extravasation is in agreement with the reported ischemic reduction of other matrix ligands (i.e., laminin-1, collagen-4) within basal membrane (4, 13). The time-dependent events correlate well with simultaneous generation of matrix proteases (i.e., MMP-2 and MMP-9) as well as their activators urokinase, in microvessels and neurons within the region of injury (13, 18, 19). The role of ECM components in ischemia/reperfusion was strengthened by concomitant findings of increased BBB permeability and MMP-2 levels which were blocked with MMP inhibitors. Furthermore, this supposition was confirmed in MMP-9 knock-out mice. In addition, ischemia reduced the expression of dystroglycan on microvessels as well as dystroglycan and integrin α6/β4 expression on the astrocytic endfeet, both in vivo and in vitro studies. These changes corresponded to the separation of astrocytes from the vascular matrix and to early cell swelling following MCA occlusion (3, 4).

The knowledge of astrocytic role in development of cytotoxic or vasogenic edema has been advanced by studying AQP4 null mice (1,3,11,16). As previously mentioned, AQP4 controls water fluxes into and out of the brain. The induction of hyponatremia in AQP4-null mice by intraperitoneal injection of water reduced mortality which was associated with decreased BBB permeability to water and reduced water flow. These animals also protected the cytotoxic edema in other models. Dystrophan-deficient mice have normal levels of AQP 4 mRNA but their protein expression is mislocated rendering similar features to AQP4-deficient mice showing delayed onset of cellular edema. Deletion of α-syntrophin or other components of dystrophin complex leads also to mislocation and unstable AQP4 protein expression. Reports indicate that α-null mice were also protected from cytotoxic edema induced by focal ischemia. On the other hand, the observed AQP4 deletion aggravates vasogenic edema induced by freeze injury or tumor implantation. In addition, an increase of AQP4 expression in astrocytes corresponded with maximum brain swelling induced by focal ischemia (3). In patients with middle cerebral artery occlusion, a single nucleotide polymorphism at 3′end of AQP4 was found to be associated with severe brain edema (10).

Actually, the post-ischemic hypoperfusion (reduce CBF) greatly contributes to the ischemic events leading to the development of BBB dysfunction and brain edema. It has been shown that all of the above mentioned endothelial functions are involved in this process. Presently, there is no doubt that the mechanism responsible for the noted ischemic and traumatic sequela is multifactorial and primarily involves the endothelium (as schematically illustrated in Fig. 2). Free radical species formation during ischemic hypoperfusion are one of the first among many factors (mediators or modulators) produced by endothelium and other brain and blood cells that were implicated in alterations of BBB permeability and edema formation. It is not surprising since tissue damage in the brain or other organs leads to the formation of free radical species (H2O2, ·OHi−, ·O−), release of fatty acids from phospholipids, and lipid peroxidation of cellular membranes (6). The same processes have also been implicated in altering the BBB permeability and the development of brain edema (12). The involvement of free radical species in ischemic reduction of CBF, alteration of BBB permeability, and formation of edema was substantiated by anti-oxidant therapy (i.e., nitroxide, melatonin). In addition, 2-arachinonylglycerol (2-AG), which reduced traumatic and ischemic tissue damage (BBB permeability and edema) was mediated through CB1 receptors and was shown to have antioxidant properties (15). Taking into consideration the endothelial in vitro functional characteristics, we investigated the endothelium and its ability to influence the microvascular activity implicated in the regulation and dysregulation of vascular tone, CBF and BBB permeability (12 and unpublished observation). These studies demonstrated that the altered cellular permeability, increased calcium mobilization, activated phosphorylation of 44/42 kinase, and cytoskeleton arrangements (actin and vimentin) induced by H2O2 were all decreased or abolished by anti-oxidants (same as described above). The results of this study are in agreement with our previous investigations regarding interactions between ET-1 and NO or 2-AG and their signal transduction pathways, and further demonstrate that the endothelium has the capacity to respond to various endogenous and exogenous mediators (2). Such endothelial reactivity is important for interconnective processes with other vascular and brain cells that affect the BBB and CBF.

Fig. 2
figure 2

Proposed mechanism of antioxidant protection of brain injury.

Full size image

Taken together the findings described in this review indicate that ischemic processes leading to changes of the BBB permeability and formation of edema involve all the cellular and subcellular components of the NVU. In spite of the abundance of available data, it is still not clear which of the factors initiate the cascade of events. Nevertheless, it is suggestive that the functional endothelial alteration of the luminal membrane, induced by ischemic “stress” and the release of blood-born mediators, occur prior to all the subsequent sequela.

Conflict of interest statement We declare that we have no conflict of interest.