The term blood-brain barrier has a long history. Its current usage describes the structural, physiological, and molecular mechanisms that control the exchange (entry and exit) of molecules between the blood and the brain. The sum of these mechanisms results in the characteristically stable internal environment of the brain, both during development and in the adult. This has been an often confused and misunderstood field of neuroscience.

The main aim of this review is to explain what is known about brain barrier mechanisms and why understanding these mechanisms is fundamental to understanding normal brain development and normal brain function and how disorders of brain barrier mechanisms may contribute to a range of neuropathological conditions. We suggest that the neuroscience community should pay more attention to this topic and we advocate the need for new researchers to move into this intriguing and important field, as major advances in many fields have often come from an influx of new people unfettered by the prevailing dogmas.

The other focus of this review will be to consider the clinically important problem of developing ways to deliver drugs to the brain for treating neurological and psychiatric disorders. This has been a major effort in the blood-brain barrier field for the past 20–30 years but has yielded little of practical value. We list the diverse attempts that have been tried, we analyze some of the possible reasons why they have been unsuccessful, and we suggest some alternative/new approaches to the problem.

The use of the term “barrier” is in many ways unfortunate ¹ , as for those outside the field it disguises the multiplicity of mechanisms involved. Perhaps this also explains the almost exclusive focus of people interested in pathological conditions involving the blood-brain barrier on tests of its integrity, largely ignoring until recently the numerous cellular mechanisms at the various blood-brain interfaces that may be disrupted. Almost all of the early work on the blood-brain barrier involved the use of dyes, which could be visualized. This field has recently been reviewed with translations from key oft-cited papers published in their original languages showing that many of the citations were incorrect ² . To set the record straight, it was Lena Stern who was the first to coin the term blood-brain barrier (“barrière hémato-encéphalique” ³ ) and not, as often cited, Ehrlich ⁴ , Lewandowsky ⁵ , or Goldmann ⁶ . The current understanding of the term “blood-brain barrier” is that it covers a number of morphological entities and a plethora of cellular transport mechanisms both inward and outward, which we next describe briefly.

There are six interfaces to be considered. Figure 1 illustrates their sites and main morphological features. An essential component of all interfaces with barrier properties is the presence of specialized junctions between the cells of the interface. In most of the barriers, these junctions are tight junctions; they restrict the movement of molecules between the endothelial and the epithelial cells. As a direct consequence of this restriction, the intercellular junctions have the important functional effect of allowing the numerous transporters within individual cells to operate over the large surface of the barrier interfaces; without this permeability restriction, the inward and outward transporter mechanisms would be ineffective. In recent years, it has become increasingly apparent that there is a much greater complexity involved in the structural organization of the brain barriers; in the case of the blood-brain barrier itself, this includes astrocytes, pericytes, basement membrane, and extracellular matrix ( Figure 1). However, there is much to be learned about the precise role of individual morphological components of the brain barriers and their interactions in normal and pathological brains ^{7– 10} . We shall now consider each barrier interface in turn: (a) the meningeal barrier, shown in Figure 1(a), is structurally the most complex of all the brain barriers and is situated at the meninges (pia, arachnoid, and dura mater). The barrier-forming cells are the outer layer of the arachnoid membrane (the arachnoid barrier cells), which have tight junctions between adjacent cells forming a physical barrier between the outer cerebrospinal fluid (CSF) in the subarachnoid space and more superficial dural layers (dural border cells and the dura mater). The blood vessels in the subarachnoid space have tight junctions with similar barrier characteristics as cerebral blood vessels, although lacking the surrounding pericytes and astrocytic end-feet ^{11– 13} . In contrast, blood vessels within the dura mater are fenestrated; other important components of the barrier are the basement membrane and glia limitans. (b) The blood-brain barrier, shown in Figure 1(b), is situated at the level of cerebral blood vessels between the lumen of the vessel and brain parenchyma. Tight junctions are present between the endothelial cells restricting permeability of the paracellular cleft ( 11 and Text Box). A basement membrane and extracellular matrix ¹⁴ surround both the endothelial cells and the pericytes ^{15, 16} . End feet from astroglial cells progressively encircle cerebral blood vessels during development ¹⁷ . These cellular structures are known collectively as the neurovascular unit ¹⁸ . (c) The blood-CSF barrier, shown in Figure 1(c), is situated in the choroid plexus within each brain ventricle. In contrast to other cerebral blood vessels, the endothelial cells forming choroid plexus blood vessels are fenestrated and do not form a barrier. The barrier-forming cells are the epithelial cells, which have tight junctions ¹¹ at their apical (CSF) side. Choroid plexus cells have microvilli on their apical side, increasing their exchange surface to the internal CSF. (d) Circumventricular organs, shown in Figure 1(d). These include the median eminence, pineal gland, area postrema, and subfornical organ. The blood vessels have permeability characteristics similar to elsewhere in the body and have the functional property of allowing feedback penetration of peptide hormones controlled by the hypothalamic-pituitary axis. These peptides and other molecules are prevented from entering the CSF by tanycytes, the specialized ependymal cells of these brain areas, connected by tight junctions between their apices; entry into the rest of the brain is prevented by tight junctions between astroglial cells ^{19, 20} . (e) Ependyma in adult brain, shown in Figure 1(e). Apart from areas where there are specialized tanycytes, ependymal cells are linked by gap junctions that do not restrict exchange of even large molecules, such as proteins, between CSF and interstitial space of brain ^{11, 21} . (f) The embryonic CSF-brain barrier, shown in Figure 1(f). In the ventricular zone is a temporary barrier between the CSF and brain parenchyma ²¹ . In early brain development, strap junctions are present between adjacent neuroepithelial cells; these form a physical barrier restricting the movement of larger molecules, such as proteins, but not smaller molecules ^{22, 23} . At later stages of development and in the adult brain, these strap junctions are no longer present when this interface becomes ependyma.

This has been a major field of endeavor over the past 20–30 years. It has been largely unsuccessful in that few of the proposed methods appear to have been independently replicated, and we are not aware of any neuropharmaceutical drugs that have been identified using these methods. There are several comprehensive reviews of the methods that have been developed ^{58– 60} . Here, we provide only a list of these methods ( Table 1), with some observations on their limitations. We deal in more detail with methods designed to allow entry of drugs into the brain by disruption of the blood-brain barrier. This is the only method of drug delivery that has translated to clinical practice, albeit on a limited basis. Newer, more focal methods hold promise of significant advances using this approach.

The following drug delivery approaches have been tried:

(i) In vitro blood-brain or blood-CSF barrier models (see reviews in 61– 64)

The hallmarks of success of such systems are generally held to be a high transendothelial resistance (TEER) and limited permeability to barrier integrity markers such as ¹⁴C-sucrose ⁶⁴ . The only in vivo TEER values that have been measured are for pial blood vessels ⁶⁵ . It is unclear whether these reflect the properties of vessels within the brain. The type of endothelial cells isolated in preparation of the cultures is often not clear ⁶⁶ . But perhaps the biggest limitation of these systems is that few attempts have been made to characterize at the molecular level the barrier and transport properties in vitro compared to those in vivo. This would seem to be particularly important given the propensity for cells to transform in culture. Where attempts have been made, the extent to which the in vivo properties are retained is limited ⁶⁷ .

(ii) Receptor-mediated and adsorptive-mediated transcytosis (see Table 1)

Lajoie and Shusta ⁶⁰ review a number of more recent developments using alternative targets on cerebral endothelial cells, but it is too soon to tell whether these will be more successful than earlier developed methods.

(iii) Influx transporters

As indicated above, there are numerous influx transporters in brain endothelial cells. In the case of only a few, it has been possible to use these to achieve penetration of a therapeutic compound into the brain. The best known, and one of the earliest to be described, is L-DOPA for the treatment of Parkinson’s disease, e.g. 68. Its introduction transformed the treatment of this condition, but, after prolonged experience, it is clear that it has serious clinical limitations.

(iv) Inhibition of efflux transporters (see above, 69, and Table 1)

The number of ABC transporters that have been shown to be functionally effective at the blood-brain barrier is only a small proportion of the known total of 49. A huge number of drugs and other xenobiotics are excluded from the brain ⁷⁰ , which explains the lack of specificity of ABC transporters. Unless some way could be found to limit the effect of the inhibitor to cerebral endothelial cells, and preferably only those in the neurological target area of the brain, this is unlikely to be a viable method of promoting drug entry to the brain.

(v) Modulation of integrity of the blood-brain barrier

Three methods have been tried − osmotic opening, ultrasound, and electrical stimulation ( Table 2).

Reversible osmotic opening of the blood-brain barrier was first demonstrated in animals by Rapoport et al. ⁷¹ using a variety of hypertonic electrolyte and non-electrolyte solutions. Brightman et al. ⁷² showed that barrier opening to horseradish peroxidase was due to opening of cerebral vessel tight junctions. Since 1979, Neuwelt has pioneered the use of osmotic opening of the blood-brain barrier as a means of delivering chemotherapeutic agents to treat brain tumors ⁷³ . He has built up an impressive array of animal and patient imaging techniques, which allowed careful evaluation of the use of hypertonic solutions to open the barrier under well-controlled, carefully monitored conditions and to develop methods for mitigating some of the potentially devastating side effects ^{74, 75} . Reversible osmotic opening of the blood-brain barrier is the only technique for improving drug delivery to the brain that has successfully translated to the clinic. It is not widely used, probably because it requires repeated hospital admissions and general anesthesia, as well as being associated with increased risk of stroke and epileptic seizures ⁷⁶ and other surgical and neurological problems ⁷⁵ .

Focused ultrasound disruption of the blood-brain barrier in laboratory animals was first investigated in the 1950s ⁷⁷ . In that study and in subsequent ones, it was necessary to perform a craniectomy in order to achieve sufficient ultrasound energy to produce effects in the brain (e.g. 78). A major advance was to combine intravenous injection of gas bubbles, previously developed as a contrast agent for ultrasound imaging, with focused ultrasound ⁷⁹ ; this reduced the ultrasound power required to disrupt the blood-brain barrier and was shown to be effective through the intact skull in rabbits. Subsequent studies have evaluated the safety of the procedure ⁸⁰ , effects of different anesthetic agents ⁸⁰ , and feasibility in large animals ⁸¹ . The mechanism of the interaction between the micro-bubbles and the focused ultrasound beam is unclear. Several possibilities are discussed by Burgess and Hynynen ⁸² and by Timbie et al. ⁸³ . The advantages of this approach compared to osmotic disruption of the blood-brain barrier are that (a) it is non-invasive (does not require craniotomy), (b) it can be targeted to a specific lesion, e.g. tumor, or region of neurological disorder such as the basal ganglia in Parkinson’s disease, (c) it is transient, although the estimates of duration barrier opening have varied from 6 to 24 hours in different studies, and (d) under well-defined conditions of ultrasound parameters, there appears to be no evidence of ischemia, apoptosis, or cognitive dysfunction (tested in primates ⁸² ). Investigations so far have concentrated on the mechanical disruptive effects of the method, but studies are needed to investigate possible effects on cellular transport across cerebral endothelial cells ⁸⁴ . Also, it needs to be considered whether the effects might be different in pathological brains.

The concentration in the past 25 years on developing in vitro systems for testing barrier permeability to drugs and the various drug delivery methods outlined above has been at the expense of fundamental research aimed at better understanding of brain barrier mechanisms. We list here some major questions, the answers to which might aid the development of more effective drug delivery strategies as well as our understanding of the involvement of brain barrier mechanisms in a wide range of neuropsychiatric conditions.

(i) Different approaches to drug development

Given the overwhelming importance of efflux transporters in excluding drugs from the brain, we need better understanding of the molecular nature of their mechanism(s) of action, as this would allow the development of drugs that evade these mechanisms. However, there would still be a need to develop ways of targeting the drugs not just to the cerebral vasculature but also to specific brain regions. This might come from better knowledge of the molecular characteristics of cerebral endothelial and peripheral endothelial cells as well as identifying such differences in different brain regions, as suggested by Pachter’s work ^{66, 85} .

(ii) Rapid high-throughput screening of drugs with potential for neurotherapeutic treatments

In the past, most such drugs have been developed in vitro but failed as useful agents in vivo because of inability to cross the blood-brain barrier ⁸⁶ . Two plausible approaches, on which a start has been made, are (a) to use in initial screens organisms that are easily available in large numbers, e.g. flies ^{87, 88} and zebrafish ^{89, 90} . These organisms utilize ABC efflux transporters as in mammals, although the morphological sites at which they function are different in invertebrates ⁸⁷ and the actual ABC transporters that are functionally important may be different in different species. Also, (b) we require a better understanding of the ABC transporters in human brain barriers (adult and developing) and their level of function. It should then be possible to devise appropriate cell-based screens using fluorophore-tagged drugs and competitors ⁹¹ .

(iii) What is the risk to the developing brain of drugs administered to pregnant women?

As indicated above, several ABC transporters are expressed at high levels in embryonic brain both in blood vessels and in the choroid plexuses ^{28, 35} . However, it is uncertain if expression levels can be equated to functional exclusion of drugs from the brain. If this can be shown, then coupled with similar transporter activity in the placenta, this suggests that the developing brain may be much better protected than is implied by the discredited but still current dogma that the blood-brain barrier in the embryo is unformed or “leaky” (see Text Box). Nevertheless, the loss of the placental protection in prematurely born infants may mean that they are more vulnerable to the ill effects of drugs than their full-term counterparts.

(iv) Involvement of barrier mechanism in brain disorders

The literature on possible involvement of blood-brain barrier mechanisms is too extensive to cover in this review. Recent papers describing different aspects of the pathobiology of brain barrier mechanisms are 92, which is particularly comprehensive, and 69, 93– 95. For decades, the focus has been on dysfunction defined by supposed disruption of the blood-brain barrier often defined with unsuitable markers (see Text Box). Only comparatively recently has attention turned to the possibility that transporter dysfunction may be involved. It is usually unclear whether barrier dysfunction is a cause or a consequence of a particular neurological disorder. This is an area in which further research with modern technology is likely to be fruitful.