Keywords

1 Neural Signalling and the Importance of CNS Barrier Layers

The brain and spinal cord (central nervous system, CNS) are the control centres of the body, generating central programmes, coordinating sensory input and motor output and integrating many of the activities of peripheral organs and tissues. CNS neurons use chemical and electrical signals for communication, requiring precise ionic movements across their membranes. This is particularly critical at central synapses generating graded synaptic potentials and somewhat less so along axons signalling via all-or-none action potentials. Hence precise control (homeostasis) of the CNS microenvironment is crucial for reliable neural signalling and integration. It has been argued that this was one of the strongest evolutionary pressures driving the development of cellular barriers at the interfaces between the blood and the CNS, since animals with better CNS regulation would have more reliable, efficient and rapid neural signalling, giving selective advantage in finding and remembering food sources, catching prey and avoiding predators (Abbott 1992). These cellular barriers at the interfaces act as key regulatory sites, controlling ion and molecular flux into and out of the CNS, while the resident cells of the CNS including neurons and their associated glial cells, the macroglia (astrocytes, oligodendrocytes) and microglia contribute to local regulation of the composition of the interstitial (or extracellular) fluid (ISF, ECF) (for reviews see Abbott et al. 2010; Nicholson and Hrabětová 2017). The molecular flux control at CNS barriers includes delivering essential nutrients, removing waste products and severely restricting the entry of potentially toxic or neuroactive agents and pathogens. The barrier layers also act as the interface between the central and peripheral immune systems, exerting strong and selective control over access of leucocytes from the circulation (Engelhardt and Coisne 2011; Greenwood et al. 2011; Ransohoff and Engelhardt 2012; Engelhardt et al. 2017).

Three main barrier sites can be identified (Fig. 1.1): the endothelium of the brain microvessels (forming the blood-brain barrier, BBB) (Reese and Karnovsky 1967), the epithelium of the choroid plexus (specialised ependyma) secreting cerebrospinal fluid (CSF) into the cerebral ventricles (Becker et al. 1967) and the epithelium of the arachnoid mater covering the outer brain surface above the layer of subarachnoid CSF (Nabeshima et al. 1975); the choroid plexus and arachnoid form the blood-CSF barrier (BCSFB) (Abbott et al. 2010). The endothelium forms the largest interface (based on surface area) between blood and CNS and hence represents the major site for molecular exchange and the focus for drug delivery; the choroid plexus also plays a critical role, while the properties of the arachnoid membrane suggest it plays a relatively minor role in exchange. (Note that recent microanatomical studies clarify the role of arachnoid granulations in CSF drainage (Sokołowski et al. 2018; Kutomi and Takeda 2020).) At each of these sites (endothelium, choroid plexus and arachnoid), intercellular tight junctions (zonulae occludentes) restrict diffusion of polar solutes through the cleft between cells (paracellular pathway), forming the ‘physical barrier’. In brain endothelium, blood-arachnoid barrier (Uchida et al. 2020) and choroid plexus, solute carriers on the apical and basal membranes together with ecto- and endo-enzymes regulate small solute entry and efflux. In brain endothelium, mechanisms of adsorptive and receptor-mediated transcytosis allow restricted and regulated entry of certain large molecules (peptides, proteins) with growth factor and signalling roles within the CNS. Finally, the endothelial and choroid plexus barriers help regulate the innate immune response and the recruitment of leucocytes, contributing to the surveillance and the reactive functions of the central immune cell population. Thus, these interface layers work together as physical, transport, enzymatic (metabolic) and immunological barriers (for reviews see Abbott and Friedman 2012; Abbott 2013). The barrier functions are not fixed but dynamic, able to respond to a variety of regulatory signals from the blood and the brain side, and can be significantly disturbed in many CNS and systemic pathologies. This chapter will focus on the physical, transport and enzymatic barrier functions of the blood-brain barrier and the choroid plexus, as most relevant to CNS drug delivery. As this chapter is meant primarily to provide an introduction and overview, references to key reviews are interspersed with those to original findings; more detailed background may be obtained by consulting sources within the reviews cited.

Fig. 1.1
figure 1

Location of barrier sites in the CNS. Blood enters the brain via surface arteries (red arrow, top). Barriers between blood and neural tissue are present at three main sites: (1) the brain endothelium forming the blood-brain barrier (BBB), (2) the choroid plexus epithelium which secretes cerebrospinal fluid (CSF) and (3) the arachnoid epithelium forming the middle layer of the meninges. At each site, the physical barrier results from tight junctions that reduce the permeability of the paracellular pathway (intercellular cleft). In circumventricular organs (CVO), containing neurons specialised for neurosecretion and/or chemosensitivity, the endothelium is leaky. This allows tissue-blood exchange, but as these sites are separated from the rest of the brain by an external glial barrier and from CSF by a barrier at the ependyma, CVOs do not form a leak across the BBB. ISF (ECF): interstitial or extracellular fluid. Figure based on Segal MB and Zlokovic BV 1990 Fig. 1.1, p2 in ‘The Blood-Brain Barrier, Amino Acids and Peptides’ (Kluwer), modified by A Reichel. Reproduced from Abbott et al. 2003 Lupus 12:908, and with permission of Springer

Full size image

2 The Brain Endothelium and the Neurovascular Unit

The brain capillaries supply blood in close proximity to neurons (maximum diffusion distances typically 8–25 μm); hence the activities of the BBB are key to brain homeostasis. The brain endothelium of the BBB acts within a cellular complex, the neurovascular unit (NVU) (Fig. 1.2) (Abbott et al. 2010; Muoio et al. 2014; Iadecola 2017), composed at a local level of grey matter of the segment of capillary, its associated pericytes, perivascular astrocytes, basement membranes and microglial cells, the resident immune cells of the CNS (Ransohoff and Perry 2009; Mäe et al. 2011; Bohlen et al. 2019). Together this cellular complex supports a small number of neurons within that NVU module (Iadecola and Nedergaard 2007; Abbott et al. 2010). Recent advances in understanding show that the NVU concept should be expanded to a ‘multidimensional’ network of signalling between local capillary networks ‘in which mediators released from multiple cells engage distinct signalling pathways and effector systems across the entire cerebrovascular network in a highly orchestrated manner’ (Iadecola 2017). There are also implications for development and pathology (Dalkara and Alarcon-Martinez 2015).

Fig. 1.2
figure 2

The neurovascular unit (NVU). The NVU is composed of several cell types in close association, working together to maintain an optimal neuronal microenvironment. Cerebral endothelial cells forming the BBB make tight junctions which restrict the paracellular pathway. Pericytes partially envelope the endothelial cells and share a common basal lamina with them. Astrocytes ensheath the microvessel wall. Pericytes and astrocytes are important in barrier induction and maintenance, and astrocytes provide links to neurons. Microglia are CNS-resident immune cells with highly motile cellular processes, some of which can contact the astrocyte basal lamina. By S Yusof and NJ Abbott, from Abbott (2013) with permission

Full size image

Several functions of the BBB can be identified and their roles in CNS homeostasis highlighted (Abbott et al. 2010; Abbott 2013). By regulating ionic and molecular traffic and keeping out toxins, the barrier contributes to neuronal longevity and the health and integrity of neural network connectivity (Iadecola 2017). Ionic homeostasis is essential for normal neural signalling. Restricting protein entry limits the innate immune response of the brain and the proliferative potential of the CNS microenvironment. Separating the neurotransmitter pools of the peripheral nervous system (PNS) and CNS minimises interference between signalling networks using the same transmitters while allowing ‘non-synaptic’ signalling by agents able to move within the protected interstitial fluid (ISF) compartment. Regulating entry of leucocytes allows immune surveillance with minimal inflammation and cellular damage. Finally, the system is well organised for endogenous protection and ‘running repairs’ (Liu et al. 2010; Tian et al. 2011; Ransohoff and Brown 2012; Daneman 2012; Posada-Duque and Cardona-Gómez 2020). The other cells of the NVU, especially the astrocytes, pericytes and microglia, together with components of the extracellular matrix (ECM), contribute to these activities (Errede et al. 2021).

Given the key role of circulating leucocytes in patrolling, surveillance and repair of the CNS, it has been proposed that these cells, plus the glycocalyx at the endothelial surface (Haqqani et al. 2011; Okada et al. 2020), should be included in an ‘extended NVU’ (Neuwelt et al. 2011) (Fig. 1.3). Current research on the cell/cell interactions involved is revealing further details of the complexity of the NVU and its critical role in maintaining a healthy BBB. Damage to the endothelial glycocalyx in lungs and other organs including the brain during the Covid-19 (SARS-CoV-2) pandemic (2020) may have contributed to severity of this pandemic disease (Okada et al. 2020).

Fig. 1.3
figure 3

The ‘extended’ NVU at the level of the microvessel wall, incorporating the glycocalyx and circulating cells. Recent work has highlighted the importance of the glycocalyx on the luminal endothelial surface for endothelial function and the role of circulating leucocytes in monitoring and interacting with this surface. By S Yusof and NJ Abbott, modified from Abbott et al. (2010) with permission

Full size image

3 Nature and Organisation of the Membranes of the Barrier Layers

Many powerful techniques are being applied to increase molecular understanding of barrier function (Redzic 2011; Pottiez et al. 2011; Daneman 2012; Saunders et al. 2013), including biophysical investigation of the lipid membranes, quantitative proteomics, imaging at close to the level of individual molecules and use of genetic mutants and siRNA to test the roles of individual components.

The outer cell membranes (plasmalemma) of the barrier layers, like other mammalian cell membranes, consist of a lipid bilayer with embedded protein, the ‘fluid mosaic’ model of the membrane. The membrane lipids include glycerophospholipids, sterols and sphingolipids. The hydrophilic polar heads of phospholipids form a continuous layer at the outer and inner leaflets of the membrane, with hydrophobic chains extending into the core of the membrane; the outer leaflet contains mainly zwitterionic phosphatidylcholine (PC) and phosphatidylethanolamine (PE), while the inner leaflet contains mainly negatively charged phosphatidylserine. PC and PE are the main phospholipids in brain endothelium at 20% and 30%, respectively, with cholesterol at ~20% (Krämer et al. 2002). Under physiological conditions, the lipid bilayer is in a liquid crystalline state. The high percentage of PE and cholesterol in brain endothelium helps to increase its packing density (Gatlik-Landwojtowicz et al. 2006; Seelig 2007) which affects the way molecules partition into and diffuse through the membrane. At the molecular level, there is continual motion of the phospholipid tails within the membrane, creating transient gaps that permit flux of small gaseous molecules (oxygen, CO2) and small amounts of water (Abbott 2004; Dolman et al. 2005; MacAulay and Zeuthen 2010). Many lipophilic agents including drugs permeate well through the lipid bilayer (Bodor and Buchwald 2003) (Fig. 1.4). However, the tight lipid packing restricts permeation of certain hydrophobic molecules including many drugs and regulates access to particular membrane transport proteins such as the ABC (ATP-binding cassette) efflux transporters, P-glycoprotein (Pgp) (Aänismaa et al. 2008) and breast cancer resistance protein (BCRP) (Fig. 1.4).

Fig. 1.4
figure 4

Routes across the brain endothelium. Diagram of brain endothelium showing (numbered red circles) the tight junctions (1) and cell membranes (2) forming the ‘physical barrier’, transporters (3) and vesicular mechanisms (4) (forming the ‘transport barrier’), enzymes forming the ‘enzymatic barrier’ (5) and regulated leucocyte traffic (6) the ‘immunologic barrier’. (a) Solutes may passively diffuse through the cell membrane and cross the endothelium; a higher lipid solubility and several other physicochemical factors favour this process. (b) Active efflux carriers (ABC transporters) may intercept some of these passively penetrating solutes and pump them out. Pgp and BCRP are strategically placed in the luminal membrane of the BBB endothelium. MRPs 1–5 are inserted into either luminal or abluminal membranes, with some species differences in the polarity and the MRP isoforms expressed. (c) Carrier-mediated influx via solute carriers (SLCs) may be passive or primarily or secondarily active and can transport many essential polar molecules such as glucose, amino acids and nucleosides into the CNS. The solute carriers (black numbers) may be bidirectional, the direction of net transport being determined by the substrate concentration gradient (1), unidirectional either into or out of the cell (2/3), or involve an exchange of one substrate for another or be driven by an ion gradient (4). In this last case, the direction of transport is also reversible depending on electrochemical gradients. (d) RMT requires receptor binding of ligand and can transport a variety of macromolecules such as peptides and proteins across the cerebral endothelium (transcytosis). AMT appears to be induced in a non-specific manner by positively charged macromolecules and can also transport across the endothelium. Both RMT and AMT appear to be vesicular-based systems which carry their macromolecule content across the endothelial cells. (e) Leucocyte entry is strictly regulated; under some conditions leucocytes may cross the endothelium by diapedesis either through the endothelial cells or via modified tight junctions. Tight junction modulation can result from signals from cells associated with the NVU or be induced pharmacologically. Modified from Abbott et al. (2010), with permission

Full size image

In certain regions of cell membranes, zones enriched in cholesterol and sphingolipids form dynamic microdomains termed ‘lipid rafts’; these 10–200 nm heterogeneous structures are associated with a variety of proteins and play roles in cell polarisation, endocytosis, signal transduction, adhesion, migration and links to the cytoskeleton, among others. In brain endothelium, such rafts (Cayrol et al. 2011) have documented functions in leucocyte adhesion and trafficking, junctional molecular architecture and localisation and function of transporters (Dodelier-Devilliers et al. 2009). A subset of rafts form caveolae, with high expression of caveolin-1, and can be further classified by function in scaffolding for junctional proteins and adhesion to basal lamina, immune cell adhesion and recruitment and transendothelial transport. Certain environmental pollutants such as polychlorinated biphenyl (PCB) induce disruption of BBB tight junction occludin and endothelial barrier function; activation of matrix metalloprotease MMP-2 in lipid rafts is involved in the reduction of occludin levels by BCB (Eum et al. 2015).

4 Tight Junctions in Brain Endothelium and Barrier Epithelia: Structure and Restrictive Properties

The tight junctions of the CNS barrier layers forming the ‘physical’ barrier (Fig. 1.4) involve a complex 3-D organisation of transmembrane proteins (claudins, occludin), spanning the cleft to create the diffusional restriction and coupling on the cytoplasmic side to an array of adaptor and regulatory proteins linking to the cytoskeleton (Cording et al. 2013). Adherens junctions, while not themselves restricting paracellular permeability, are important in formation and stabilisation of tight junctions (Paolinelli et al. 2011; Daneman 2012).

The brain endothelial tight junctions are capable of restricting paracellular ionic flux to give high transendothelial electrical resistance (TEER) in vivo of >1000 ohm.cm2, while choroid plexus tight junctions are leakier, although the complex frond-like morphology of the in vivo mammalian plexus makes TEER harder to measure. TEER of ~150 ohm.cm2 has been recorded across the simpler bullfrog choroid plexus. The brain endothelium shows high expression of the ‘barrier-forming’ claudin 5, together with claudin 3 and 12, while in choroid plexus the ‘pore-forming’ claudin 1 dominates, with detectable claudin 2, 3 and 11 (Strazielle and Ghersi-Egea 2013). Metastatic cells may migrate through the cerebral endothelium either through the tight junctions (paracellular) or across the cells (transcellular) initiated by filopodia extended from the cells (Herman et al. 2019).

For the arachnoid epithelium the situation is less clear; the arachnoid barrier layer is closely apposed to the dura and difficult to isolate intact. It has recently proved possible to culture arachnoid cells in vitro, which express claudin 1 and generate a TEER of ~160 ohm.cm2 with restriction of larger solute permeation (Lam et al. 2011, 2012; Janson et al. 2011). The perineurium forming part of the outer sheath of peripheral nerves is a continuation of the arachnoid layer of the spinal meninges and easier to study than the arachnoid; a TEER of ~480 ohm.cm2 (Weerasuriya et al. 1984) and expression of claudin 1 have been observed (Hackel et al. 2012). The pattern of barrier properties is consistent with the brain endothelium exerting the most stringent effect on paracellular permeability, while the choroid plexus with a major role in secreting CSF is leakier; the arachnoid epithelium appears to create a barrier of intermediate tightness.

Several junctional proteins, especially occludin and ZO-1, show considerable dynamic activity (half times 100–200 s) (Shen et al. 2008) while maintaining overall junctional integrity and selectivity. Many modulators from both the blood and the brain side can cause junctional opening, some via identified receptor-mediated processes (Abbott et al. 2006; Fraser 2011), possibly aiding repair and removal of debris, but in healthy conditions this is local and transient and does not significantly disturb the homeostatic function of the barrier. Indeed, the presence of endogenous ‘protective’ molecules and mechanisms able to tighten the barriers is increasingly recognised as important in protection and maintenance at the barrier sites (Bazan et al. 2012; Cristante et al. 2013). Recent studies have highlighted the possible role of microRNAs in barrier protection (Reijerkerk et al. 2013), and astrocyte-derived fatty-acid-binding protein 7 protects BBB integrity through a caveolin-1/MMP signalling pathway following traumatic brain injury (Rui et al. 2019).

5 Small Solute Transport at the Barrier Layers

Many BBB solute carriers (SLCs) with relatively tight substrate specificities have been described (Abbott et al. 2010; Redzic 2011; Neuwelt et al. 2011; Parkinson et al. 2011; Zaragoza 2020), mediating entry of major nutrients such as glucose, amino acids, nucleosides, monocarboxylates and organic anions and cations and efflux from the brain of some metabolites (Fig. 1.4). Among the group of ABC (efflux) transporters, Pgp (ABCB1) and BCRP (ABCG2) are the dominant players on the apical (blood-facing) membrane, especially Pgp in rodents and BCRP in primates, but the expression levels, localisation and roles of the multidrug-resistant associated proteins (MRPs, ABCC group) are less clear (Shawahna et al. 2011) (Fig. 1.4). ABC transporters have broader substrate specificity than the SLCs, making analysis of their structure-activity relationship (SAR) difficult (Demel et al. 2009). Synergistic activity between Pgp and BCRP has been observed (Kodaira et al. 2010), and ABC transporters and cytochrome P450 (CYP) enzymes together generate an active metabolic barrier within the NVU (Declèves et al. 2011). Differences between species and between in vitro models may make it difficult to draw firm conclusions in comparative studies (Shawahna et al. 2013; Breuss et al. 2020).

There are many differences between the transporters and enzymes expressed in the different barrier layers, suggesting they play different but complementary roles in regulation of molecular flux (Strazielle and Ghersi-Egea 2013; Saunders et al. 2013; Yasuda et al. 2013; Zaragoza 2020). The transporters present include considerable overlap in function/apparent redundancy at each site, reflecting their evolutionary history (Dean and Annilo 2005) and ensuring maintained function in case of loss or defect of a single transporter.

6 Vesicular Transport and Transcytosis

Classification of types of vesicular transport by cells is complex, but it is clear that certain features of endocytosis and transcytosis in the highly polarised brain endothelium are different from those of less polarised endothelia such as that of skeletal muscle. Non-specific fluid-phase endocytosis and transcytosis are downregulated in the brain compared with non-brain endothelium. However, for certain endogenous peptides and proteins, two main types of vesicle-mediated transfer have been documented in the BBB: receptor-mediated transcytosis (RMT) and adsorptive mediated transcytosis (AMT) (Abbott et al. 2010) (Fig. 1.4). There appears to be some overlap in function between caveolar and clathrin-mediated vesicular routes and likely involvement of other types of molecular entrapment, engulfment and transendothelial movement that are less well characterised (Mayor and Pagano 2007; Strazielle and Ghersi-Egea 2013). Recent studies using manufactured non-ionic surfactant vesicles (NISVs) show that decoration with glucosamine can enhance delivery across the BBB in vivo and in vitro (Woods et al. 2020); these NISVs hold promise for drug delivery.

Electron microscopy of the choroid plexus shows a variety of vesicular and tubular profiles, but the epithelium appears to be specialised for secretion rather than transcytosis (Strazielle and Ghersi-Egea 2013).

7 Routes for Permeation Across Barrier Layers and Influence on Drug Delivery

Many of these routes for permeation across the brain endothelium (Fig. 1.4) can be used for drug delivery; several classical CNS drugs are sufficiently lipid-soluble to diffuse through the endothelial cell membranes to reach the brain ISF (Bodor and Buchwald 2003). However, for less lipophilic agents with slower permeation and hence longer dwell time in the lipid bilayer, activity of ABC efflux transporters can significantly reduce CNS access (Seelig 2007; Turunen et al. 2008; Aänismaa et al. 2008). As barrier tightness, transporter expression/activity and vesicular mechanisms can be altered in pathology, it is difficult to predict CNS distribution and pharmacokinetics of drugs in individual patients, particularly where barrier dysfunction may change both regionally and in time during the course of pathologies such as epilepsy, stroke and cancer (Stanimirovic and Friedman 2012).

8 Development, Induction, Maintenance and Heterogeneity of the BBB

Study of BBB evolution, development and maintenance gives valuable insights into both normal physiology and the changes that can occur in pathology. Studies in invertebrates and lower vertebrates especially archaic fish provide strong evidence that the first barrier layers protecting the CNS were formed by specialised glial cells at the vascular-neural interface and that as the intracerebral vasculature became more complete and complex, the barrier was increasingly supported by pericytes and endothelium. Later there was a shift to the dominant modern vertebrate pattern, where the endothelium forms the principal barrier layer (Bundgaard and Abbott 2008). Interestingly, the pericytes and astrocytes still remain closely associated with the brain endothelium, reflecting their evolutionary history and contributing to the NVU.

In the development of the mammalian brain, the endothelium of the ingrowing vessel sprouts develops basic restrictive barrier properties under the influence of neural progenitor cells (NPCs) (Liebner et al. 2008; Daneman et al. 2009), with pericytes subsequently refining the phenotype by downregulating features characteristic of non-brain endothelium; later, astrocytes help upregulate the full differentiated BBB phenotype (Daneman et al. 2010; Armulik et al. 2010; Stebbins et al. 2019).

Some of the signalling mechanisms involved in this induction are known, including the Wnt/β-catenin (Liebner et al. 2008) and sonic hedgehog pathways (Alvarez et al. 2011), and some of them may be involved in maintaining barrier integrity in the adult. It is clear that endothelial cells and pericytes are in turn involved in signalling to astrocytes, to regulate the expression of ion and water channels, receptors, transporters and enzymes on the astrocyte endfeet, so that mutual induction and maintenance is involved in sustaining the critical features of barrier and NVU function (Abbott et al. 2006). This regulation extends to the microanatomy and microenvironment of the perivascular space created by the extracellular matrix/basal lamina components of the endothelial-pericyte-astrocyte complex (Liebner et al. 2011; Stebbins et al. 2019). Microglial cell processes are found among the astrocyte endfeet (Mathiisen et al. 2010), suggesting roles in monitoring and influencing the local cellular organisation and function; indeed, microglial cells have been shown to regulate leucocyte traffic (reviewed in Daneman 2012). Specific perivascular nerve fibres associated with cerebral microvessels are involved in regulation of vascular tone (Hamel 2006). However, less is known about microglial and neuronal induction of barrier properties, and the signalling pathways involved in barrier maintenance on a minute-by-minute basis are relatively unexplored.

The NVU contains several mechanisms for protection of the BBB against minor damage such as local oxidative stress, e.g. by tightening the barrier (Abbott et al. 2006) and presence of detoxifying transporters and enzymes (Strazielle and Ghersi-Egea 2013), but this field is expanding with recognition that some of the ‘protectins’, protective agents identified in peripheral tissues, are also active in the brain (Bazan et al. 2012). Recently the protein annexin-A1/lipocortin has been shown to be involved in the anti-inflammatory and neuroprotective effects of microglia (McArthur et al. 2010) and to act as an endogenous BBB tightening agent (Cristante et al. 2013; Wang et al. 2017). Improved understanding of the mechanisms for ‘self-repair’ within the NVU to correct minor local damage is likely to prove critical in future development of therapies that treat CNS disorders at much earlier phases of the pathology than currently possible, with expected major gains in efficacy.

There are several phenotypic and functional differences between the endothelial cells of different segments of the cerebral microvasculature (reviewed in Ge et al. 2005; Patabendige et al. 2013). Compared with arteriolar or venular endothelium, cerebral capillary endothelium has a more complex pattern of tight junction strands in freeze-fracture images consistent with tighter tight junctions and higher expression of solute transporters including efflux transporters and of certain receptors involved in transcytosis. Arteriolar endothelium shows higher expression of certain enzymes and absence of P-glycoprotein and in a few regions shows bidirectional transcytosis of tracers such as horseradish peroxidase, creating a local protein ‘leak’. The post-capillary venule segment is specialised for regulation of leucocyte traffic and control of local inflammation. Some differences between the vascular beds of different brain regions have been observed at both micro- and macro-levels, but in general their significance is unclear.

9 Beyond the Barrier: The Fluid Compartments of the ISF and CSF

The cells of the brain, chiefly neurons and macroglia (astrocytes and oligodendrocytes) but also microglia, the resident immune cells of the brain, are bathed by an ionic medium similar to plasma, but containing very low protein and slightly more Mg2+, less K+ and Ca2+ (Somjen 2004). This extracellular or interstitial fluid (ECF, ISF) occupies around 20% of the brain volume (Sykova and Nicholson 2008). The ventricles and subarachnoid space contain cerebrospinal fluid (CSF), secreted by the choroid plexuses of the lateral, third and fourth ventricles, and with a daily turnover in humans of two to four times per day (Silverberg et al. 2003). The outflow pathways include arachnoid granulations and outpouchings of the arachnoid membrane into veins in the dura, but some CSF also drains along cranial nerves (especially olfactory) and blood vessel sheaths to the lymph nodes of the neck. Species differences have been reported in the relative importance of these drainage routes (Johanson et al. 2008).

The origin and dynamics of the ISF are less well understood. The brain microvessels have the ionic transport mechanisms and channels and low but sufficient water permeability to generate ISF as a secretion (Fig. 1.5), and calculations show that a proportion of ISF water may come from glucose metabolism of the brain, aided by aquaporin 4 (AQP4) water channels in the perivascular endfeet membranes of astrocytes (Abbott 2004; Dolman et al. 2005). Within the neuropil, the small blocks of tissue demarcated by the lattice of fine microvessels, in which neural communication occurs, the distances from the vessel to the furthest neuron are small, typically <30 μm, so that diffusion within the neuropil is an effective means of ionic and molecular movement. Indeed, many studies in isolated brain slices and in situ confirm the local diffusive behaviour of test molecules injected into the brain (Thorne and Nicholson 2006; Wolak and Thorne 2013). However, superimposed on this local diffusion is the possibility for flow of ISF over longer distances, especially along perivascular spaces (Abbott et al. 2018). Convective flow of ISF through the delicate synaptic networks of the brain would be potentially damaging and has not been convincingly demonstrated in spite of claims of the ‘glymphatic’ hypothesis (Taoka and Naganawa 2020).

Fig. 1.5
figure 5

Proposed sites of generation of ISF and routes for ISF flow. A large fraction of ISF is proposed to be formed by brain capillary endothelium, driven by the ionic gradient set up by the abluminal Na, K, ATPase (* circle + arrow). Water follows passively either through the endothelial cell membranes or via the tight junctions (dashed arrows). Driven by this hydrostatic pressure gradient and with the addition of some CSF from the subarachnoid space, ISF moves by bulk flow through low resistance pathways formed by perivascular spaces (PVS, predominantly around larger vessels including arterioles and arteries, venules and veins), connecting with (a) glial-lined boundary zones between blocks of neuropil and (b) regions adjacent to axon tracts. The narrow spaces between cells within the neuropil appear to be too narrow to permit significant bulk flow. Not to scale. Modified by S Yusof from Abbott (2004), with permission

Full size image

There is considerable historical evidence for flowing ISF, capable of clearing waste products including large molecules such as β-amyloid from the interstitium moving via routes offering the least resistance to flow, along axon tracts and blood vessels (Abbott 2004; Weller et al. 2008, 2009). Careful studies of clearance of tracer molecules injected into the parenchyma give a figure for clearance half-time of 2–3 h (Groothuis et al. 2007), around ten times faster than reported earlier (Cserr et al. 1981). Most of this flow can be accounted for by fluid secretion across cerebral capillary endothelium (Abbott 2004) (Fig. 1.5), but recent studies add to a body of earlier evidence showing that a proportion of CSF from the subarachnoid space can flow into the brain along periarterial (Virchow-Robin) spaces, contributing to ISF, with return out along nerve fibre tracts and blood vessels (Abbott 2004; Iliff et al. 2012; Yang et al. 2013). However, there is some controversy over whether arteries (Weller et al. 2008, 2009) or veins (Iliff et al. 2012) are chiefly responsible for the ISF outflow route from the brain parenchyma. In any event, with the flow largely confined to major extracellular ‘highways’ in the tissue, the rate of turnover will be similar to that of CSF. Thus ISF and CSF can be regarded as parallel fluids maintaining a continuous flow through the low resistance pathways of the brain (ISF) and through the ventricles and subarachnoid space (CSF), capable of some mixing hence with some shared roles, but also many distinct and complementary functions. Between them, the CSF and ISF contribute to maintaining tissue buoyancy, waste removal, circulation of secretory products such as vitamins and hormones from choroid plexuses, non-synaptic or ‘distance’ signalling (‘volume transmission’) and providing routes for immune surveillance without disturbing neuronal networks (Strazielle and Ghersi-Egea 2013; Dalakas et al. 2020).

10 Changes in BBB and BCSFB in Pathology

The BBB is altered in many CNS pathologies, including stroke, vascular dementia, Alzheimer’s disease, Parkinson’s disease, multiple sclerosis, amyotrophic lateral sclerosis, hypoxia, ischaemia, diabetes mellitus and epilepsy (reviewed in Abbott et al. 2006, 2010; Friedman 2011; Abbott and Friedman 2012; Daneman 2012; Stanimirovic and Friedman 2012; Potschka 2012; Michalicova et al. 2020). Even mild traumatic brain injury (mTBI) may result in changes in BBB integrity and function (Rawlings et al. 2020). Changes can include upregulation of luminal adhesion molecules, increased adhesion and transmigration of leucocytes, increased leakiness of tight junctions, extravasation of plasma proteins via paracellular or transcellular routes and altered expression of drug transporters. Given the importance of the BBB in CNS homeostasis, it is clear that gross barrier dysfunction is likely to be associated with disturbance of neural signalling, in both the short and the long term (Abbott and Friedman 2012). In many pathologies, a combination or sequence of events may make the barrier vulnerable, including hypoxia, infection, activation of the clotting system and inflammation, components of the diet and environmental toxins, and genetic factors may also contribute (Shlosberg et al. 2010). Inflammation and free radicals are now recognised to play major roles in many or even most of the pathologies with BBB disturbance, but the aetiology and sequence of changes are generally unclear, and in many cases, it is not known whether changes occur simultaneously or as part of an inflammatory cascade (Friedman 2011; Kim et al. 2013). Certain brain regions are more often affected, including the hippocampus and cerebral cortex grey matter, but again the reasons are uncertain.

The Coronavirus (SARS-CoV-2, Covid-19), first reported in humans in Wuhan, China, in Dec 2019, subsequently developed into a worldwide pandemic. A number of clinical and laboratory studies have followed on its sites and modes of action; Covid-19 has serious effects on the vasculature in multiple organ systems including the cerebral vasculature. In vitro evidence suggests that the viral spike proteins S1 and S2 trigger a pro-inflammatory response in brain endothelial cells that may contribute to altered BBB function (Buzhdygan et al. 2020).

For minor damage, the cells of the NVU aided by recruitment of leucocytes may effect a repair, and short- and long-term changes in protective mechanisms including upregulation of efflux transporters and enzymes may be involved. Certainly several types of altered cell/cell interaction can be detected in pathology, particularly between endothelium and astrocytes, but also with powerful roles played by microglia, the tissue-resident macrophage, changing from a relatively quiescent and static process-bearing morphology to a more amoeboid and migratory form, secreting a different repertoire of cytokines and chemokines (Saijo and Glass 2011; Smith et al. 2012; Daneman 2012). In certain pathological neuroinflammatory and neurodegenerative conditions, there is unexpected phenotypic convergence between CNS microglia (the tissue-resident macrophages of the brain) and peripheral macrophages, suggesting that the two cell types act synergistically, boosting their mutual activities and therapeutic potential (Grassivaro et al. 2021).

Agents released from most of the cells of the NVU in pathology can modulate brain endothelial tight junctions, with several inflammatory mediators increasing barrier permeability and a few agents able to counter or reverse this (Abbott et al. 2006). Potentiating effects of several cytokines including IL-1β and TNFα on the ‘first line’ of inflammatory mediators (e.g. bradykinin) have been documented (Fraser 2011). At the molecular level, a great many signalling pathways can be identified, regulating both the expression and activity of barrier features, particularly well documented for the effects of xenobiotics, neurotransmitters and inflammation on Pgp (Miller 2010). Recent identification of a number of microRNAs (miRNAs) shown to influence angiogenesis (Caporali and Emanueli 2011) vascular functions (Hartmann and Thum 2011) and BBB physiology/pathology (Reijerkerk et al. 2013; Mishra and Singh 2013) adds a further level of complexity. Furthermore, new information on a whole family of secreted and information-carrying extracellular vesicles including exosomes (György et al. 2011; Haqqani et al. 2013) adds to the repertoire of ways in which a cell or group of cells can influence other cells nearby or further away. Indeed, the flow pathways allowing circulation of the brain ISF have suitable properties for this kind of non-neural communication (Abbott 2004; Abbott et al. 2018) and could also play an important part in the dissemination of CNS pathologies (multiple sclerosis, cancers) that start at a relatively restricted locus.

The choroid plexus and CSF/ISF flow system are also affected by ageing and by many pathologies, including tumours, infection, trauma, ischaemia, neurodegenerative disease and hydrocephalus (Johanson et al. 2008; Serot et al. 2012). Many of these affect the anatomy, connectivity and outflow routes of the fluid flow systems, but changes in the physiology of the choroid plexus and the resulting disturbance in generation and composition of CSF are also critical. Ageing is associated with a reduction in CSF production and in secretion of many choroid plexus-derived proteins, particularly important for the zones of neurogenesis close to the ventricular wall.

11 Implications for Drug Delivery

The anatomy and physiology of the CNS barriers and fluid systems described here have many implications for drug delivery, whether for agents designed to act in the CNS or for those with peripheral targets where the aim is to minimise CNS side effects. Clearly barrier changes in ageing and pathology will affect distribution and delivery of both CNS and peripheral drugs. Improved experimental methods and models, molecular and pharmacokinetic modelling and new developments in understanding barrier function help in measuring and predicting the concentration of drugs at the active site. The expanding field of ‘biologic’ therapeutics, large molecules with specific actions in the CNS, poses new challenges but is also giving novel insights into mechanisms and ways to improve CNS drug delivery of complex molecules. Many of these issues will be discussed in further chapters in this volume.

12 Points for Discussion

  • Why is it important to understand the different properties of the three main barrier layers (Sect. 1.1)? What kinds of technique can be used to establish the relative importance of each in determining CNS distribution of a particular drug?

  • Several ‘key functions’ of the BBB are listed (1.2). Is it possible to put these in order of importance for brain function?

  • Much of the BBB and choroid plexus literature is devoted to documenting changes in pathology. Why has maintenance of healthy function received less attention?

  • What models and techniques would you propose for a new study on cell/cell interaction within the NVU?

  • What is the glycocalyx of the brain endothelium, and what properties of the cells is it most likely to influence?

  • Why is it important to know about the organisation of the BBB lipid membrane in modelling drug permeation? Why are potential substrates for ABC transporters particularly affected by the membrane composition?

  • Why is it difficult to establish how water moves across the BBB and choroid plexus?

  • How does knowledge of BBB development help in understanding barrier function?

  • What is the significance of heterogeneity in function, e.g. between the capillary and post-capillary venule segments of the cerebral microvasculature?