The blood-brain barrier

This chapter does not have any specific reference point within the official CICM primary exam curriculum, as the 2017 CICM Primary Syllabus does not have any entry listing the blood-brain barrier as a subject of interest. However, that does not mean the examiners are not interested in it - in fact, they clearly are, as it has come up in at least three past paper questions:

• Question 3 from the first paper of 2015
• Question 12 from the second paper of 2012
• Question 8 from the second paper of 2009

These questions were worded fairly predictably, focusing on structure and function, and the implications of the barrier for the CNS penetration of drugs. That was plenty to work with. 

This important topic has enjoyed a lot of publisher interest over the years, and the trainee looking for papers to read would be spoiled for choice. There's barely any reason to pick up a textbook. Quality reviews are available from Bechmann et al (2007), Ballabh et al (2004), Abbott et al (2010) Daneman & Prat (2015), and that's just on the first page of a Google Scholar search. If you are pressed for time and need to choose between these resources, the best option would be Ballabh et al, mainly because of their structured approach and readability. On the other hand, if time is not the issue and one really can't get enough of this topic, The Blood-Brain Barrier in Health and Disease by Katerina Dorovini-Zis (2015) will fill that hole with about eight hundred pages, over two volumes.

Definition of the blood brain barrier

As most exam questions start with the stem "describe the blood brain barrier", one feels as if one ought to have a short and pithy blurb to headline the answer, defining this structure and mellowing the examiner. There's probably nothing better than the opening line of Ballabh et al, which starts with:

"The blood – brain barrier (BBB) is a diffusion barrier, which impedes influx of most compounds from blood to brain".

There's certainly no official definition, and the reader is invited to craft their own. Even though some authors might do it, for the purposes of the CICM exam answer one would not be expected to expand on this short statement with anything more about cellular components, selectivity, transport function or metabolic role, because the rest of the answer will fill that in. 

Ultrastructure of the blood-brain barrier 

The blood-brain barrier is present along most of the microvasculature of the central nervous system, and is therefore topographically a massive labyrinthine structure of cyclopean complexity. Perusing textbooks, the reader will often be pelted with factoids like "each gram of brain tissue contains capillaries with a total surface area of 240m²". That one probably comes from Crone (1963) who arrived at this figure on the basis of the measured permeability coefficients for various molecules. With the average brain weight being about 1400g, the total surface area of the BBB should therefore be something like 350,000 m², which is clearly an insane number and which should not be quoted anywhere. Other textbooks give different numbers, and other authors shamelessly requote them, and none of it means anything.

This structure has common cellular and ultrastructural elements which are present variably throughout the CNS. The specific organisation and permeability characteristics of this thing are very dynamic - it does not cover all the capillaries, and where it covers them can change, and it may have discontinuities along specific vascular beds for various reasons. In short, it certainly does not behave like some kind of static impermeable wall. Rather, it behaves like the borders of a small European nation state during a war - geographically flexible, porous and with rules constantly changing depending on the prevailing conditions. The term used to describe this interdependence of neuronal glial and vascular functional and structural components is "neurovascular unit", a term used to describe the "intimate relation of BBB and CNS cells", to borrow a turn of phrase from Banks & Erikson (2010).  

In short, there's only a few elements that need to be mentioned in a written exam answer, which are:

• Endotheial cells, which are not fenestrated
• Tight junctions between these endothelial cells
• Basal lamina
• Pericytes
• Astrocyte foot processes
• Perivascular potential space

Or, to mangle an image from Anderson et al (2011):

It would be tempting to edify the reader with some genuine electron microscopy images of blood-brain barrier structures, but this is one of those rare instances where a diagram is better, mainly because the blood-brain barrier is not especially photogenic. For example, here are some transmission electron microscopy images available in the 1967 paper by Reece & Karnovsky, and they leave much to be desired from the perspective of the educator. Firstly, the authors filled vessels with horseradish peroxidase, which is the mess of mottled black garbage in the capillary lumen. To make matters worse, as you can see the image is dominated by the massive erythrocytes trapped inside the sliced capillary, the astrocyte foot processes require significant imagination to appreciate, and by comparison, the little blue pericyte is a mere sliver. The basal membrane can not be visualised at all (even though the magnification is quoted as ×210,000).

Endothelial cells of the BBB are morphologically unlike the endothelial cells in normal microcirculatory beds, as they have no fenestrations. The upshot of this is that the endothelial cells act as a continuous sheet of lipid bilayer, which limits the transfer of large molecules and water-soluble substances. Anything that wants to get through his bilayer would have to ask permission from the transport proteins of these endothelial cells.

Tight junctions between these endothelial cells are made as impenetrable as possible to water and other molecules. The cells actually overlap and interdigitate to make the connection even tighter. Here, a good TEM micrograph of these structures is "borrowed" from Sukriti Nag (2003) to illustrate this concept:

Without going into too much detail here, it will suffice to say that these tight junctions are kept tight by the actions of all sorts of proteins with sticky names like "occludin" and "adherin".  Specifically, the claudin family of proteins is responsible for most of the integrity of these junctions.  For the intensivist, these proteins only become interesting when they disintegrate in conditions like stroke or inflammation. 

The basement membrane lies under the endothelial cells, and is similar to the basal lamina elsewhere. Considering that this is just a 3-40nm layer of collagen and laminins, it is hard for it to be particularly clever, and all the magic of the blood brain barrier resides in other ultrastructural elements. Of course it still plays a vitally necessary scaffold and maintenance structure role to help maintain the health and integrity of the BBB. If even one of the proteins or heparans in this membrane is knocked out of the genome, the mice die in utero at some early embryonic stage as their telencephalon fails to develop.

Pericytes  are supporting cells derived from mesenchymal (i.e. connective tissue) lineage which lurk inside the basement membrane and play a background supporting role. Their duties can be broadly described as garbo-janitorial. Brown et al (2019) attributes them with  "regulation of cerebral blood flow, maintenance of the blood-brain barrier (BBB), and control of vascular development and angiogenesis". 

Astrocyte foot processes and glial cells were once thought to make up some of the BBB, or ar least to contribute to it, but these days this idea has been debunked by studies which have demonstrated that they are relatively non-essential. For example, when Krum et al (1997) used immunotoxins to destroy all neurons and microglia in their rats, the remaining vascular and pericyte population were able to maintain a workable blood-brain barrier in the aftermath. Even 28 days after the initial insult the microvasculature of those rat brains remained intact (i.e. impermeable to albumin, still performing its glucose transporter role) in spite of basically no remaining living astrocytes or neurons. Moreover, all the studies with various tracer substances (eg. horseradish peroxidase, a 40 kD molecule) have demonstrated on countless occasions that the capillary endothelium is the site of the blood-brain barrier, not the astrocyte foot processes.

However, textbooks do tend to mention astrocytes in their discussion of the BBB. More importantly, the college examiners chided exam candidates for "not mentioning the presence of astrocyte foot processes" in their answer to Question 8 from the second paper of 2009. Astrocytes are a major part of the glial milieu and clearly play an important role in maintaining the health of neurons. A part of playing that role appears to require that they completely surround the vascular tubes of the capillaries (for some reason, the specific term repeatedly used in the literature is "ensheath").

These cells clearly have some sort of BBB-maintaining vascular regulatory function. As an example, when Janzer & Raff (1987) injected cultured astrocyte cells into the eyes of adult rats, the vascular cells there formed tight junctions and basically turned into an extracranial patch of protected blood-brain barrier-like vessels (which did not stain with Evans' Blue, for example, demonstrating their molecular impermeability). In short, though the presence of healthy astrocytes might not be mandatory to the maintenance of the BBB, they are required for its development, and they have a clear role to play in its normal function.

Barrier function of the blood-brain barrier

This thing was originally discovered when Paul Ehrlich injected dye into the bloodstream of mice, and noted that it stained everything except the brain and spinal cord. Ehrlich theorised that clearly some sort of Bluthirnschranke must be excluding the dye molecules from the central nervous system. This barrier mechanism has several facets:

Physical barrier function is mediated by the tight junctions between endothelial cells, and by the lack of their fenestrations. This cuts off the popular paracellular route of solute penetration. Molecules would need to be able to pass through several lipid bilayers on their way to a neuron, which is a major imposition for those that are lipid-soluble.

Transport barrier is mediated by the active transport mechanisms, or the lack thereof. In short, it is impossible to enter the brain without a ticket. The endothelial cells facilitate the transport of specific  molecules of interest, some of which get the red carpet treatment (for example, glucose transporters are highly prevalent on these membranes). In general, these mechanisms of transcytosis fall into a couple of groups, and it would probably be wasteful to describe them here in any massive detail. It will suffice to say that they are mainly saturable, ATP-powered and pinocytosis-like, including one interesting process where huge vesicles ("caveolae") of plasma fluid are entrained into endothelial cells and transported into the brain with no further chemical modification. In case one really feels the need to submerge in this deep water, Barar et al (2016) would offer a good drug-focused starting point.

Metabolic barrier is mediated by endothelial cell and astrocyte metabolism of actively transported molecules, or even ones that are still along the luminal side. These mechanisms occasionally involve the biotransformation of a previously BBB-permeable substance into something water-soluble which cannot cross the BBB. One good example of this is levodopa, which gets converted into dopamine by endothelial dopa decarboxylase, and therefore cannot actually get into the brain unless it is co-administered with a dopa decarboxylase inhibitor.

Transport function of the blood-brain barrier

That's enough about the ways in which the blood-brain barrier obstructs the transport of molecules. What about the ways in which it helps their transit? To put it simply, the brain needs things, and the cellular components of the blood-brain barrier need to facilitate its access to those things. Again borrowing from the excellent paper by Barar et al (2016), the mechanisms of transport and their substrates are listed here:

• Paracellular transport 
  • Mainly limited to small hydrophilic substances, and to parts of the blood-brain barrier where the tight junctions are relaxed or broken.
  • Under normal circumstances, this really only refers to water itself.
• Transcellular passive diffusion
  • This is limited to small highly lipophilic substances which are not substrates for any intrinsic BBB enzyme systems
  • In other words, these molecules need to dissolve through several lipid bilayers and not get splatted by metabolic enzymes in the endothelial cells or astrocytes
• Active transport
  • Glut-1 and Glut-3 transporters (glucose)
  • MCT family of monocarboxylic amino acid transporters (lactate, short-chain fatty acids, biotin, salicylic acid, and coincidentally also valproate)
  • Amino acid transporters (aspartate, glutamate, lysin, arginine, L-ornithine)
  • Choline transporters (choline and thiamine)
  • Peptide transporters (for large oligopeptide)
  • Medium-chain fatty acid transporters
  • Nucleoside transporters
• Endocytosis / pinocytosis
  • Specific molecules, by binding to a BBB surface receptor, can trigger their own pinocytosis and vesicular transport across the barrier.
  • A good example of this would be antibodies, and by extension monoclonal antibody drugs

Which is a good segue to the discussion of...

Characteristics of drugs which can cross the blood brain barrier

In order to penetrate the blood-brain barrier, drugs would obviously need to have certain properties. In short, they would need to either be passively transported, or actively transported, and each mechanism has its own highly predictable requirements. If one sees this as a question in the exam, one should view it as some sort of gift.

Drugs which rely on passive transport into the brain would have to have the following characteristics:

• For all drugs:
  • Small molecular mass, to facilitate diffusion (eg. ethanol)
  • High lipid solubility (eg. propofol)
  • Low volume of distribution (to ensure a large amount of drug available in the bloodstream).
• For specifically hydrophilic drugs:
  • A high concentration in the blood stream, to rely on the concentration gradient to force the drug through the paracellular route (eg. ethanol)
  • Low protein binding, so that a large free fraction is available
• Sufficient molecular similarity with another actively transported substrate
  • Exampes include lithium (pretends to be sodium), valproate (pretends to be lactate) and monoclonal antibodies (pretend to be real proper antibodies)

The college answer to Question 12 from the second paper of 2012 also included "low potency" in the list of drug characteristics which favour BBB penetration, which is clearly incorrect, as it is a double inversion of the normal relationship between pharmacodynamic potency and pharmacokinetic effect site penetration. You are not a low potency drug and therefore better able to penetrate the BBB; you are a low potency drug because you are not able to penetrate the BBB, and therefore your effect is of a lower potency (i.e little effect in spite of a high concentration of drug). The only possible explanation for this statement is where it has a relationship to concentration. For example, alcohol is a low potency drug, and therefore present in the bloodstream in massive concentrations, which probably would play a major role in its ability to cross the blood-brain barrier, if only it weren't also a tiny highly lipophilic molecule.

Macrostructure of the blood-brain barrier

What is the geographical extent of the blood-brain barrier? The exam answer would have to include a statement to the effect that it extends to all parts of the central nervous system, except for the few bits of brain tissue that must necessarily dangle their toes into the bloodstream to sample its chemical properties. In fact, though we refer to it as the "blood-brain barrier" it makes you think that it is limited to the central nervous system, but in fact it extends to the rest of the nerve tissues, and a "blood-nerve barrier" is also seen along peripheral nerves and their axons. Endoneurial (intrafascicular) capillaries are also non-fenestrated (Reina et al, 2015), and glia-like cells which make up the perineurium play a similar role to that which is played by astrocytes, ...ensheathing the axons and acting as a second barrier.

In the brain, the blood-brain barrier encompasses the neurons, glia, and some of the peripherally related spaces filled with fluid and connective tissue. While Reece & Karnovsky (1967) were filling rat microvasculature with peroxidase, they also managed to take some pictures of its distribution on a non-nanometre level. Here's a nice slice of rat brain with the pial surface at the top, illustrating the boundaries of the blood-brain barrier. Just as with Ehrlich's dye, the peroxidase did not penetrate into the parenchyma and outlined the microvessels perfectly. Particularly clear are the pial perforating arterioles, which enter the brain tissue at a perpendicular angle.

The pia mater covers the exterior of these vessels, and reflects around their point of entry into the brain.  Some excellent artwork from Krahn (1982) was borrowed and vandalised with crayons to illustrate this environment:

This is a convenient point to mention the perivascular space, otherwise known as the Virchow-Robin space.  This is a fluid-filled cavity that separates the astrocyte foot processes from the smooth muscle of arterioles as they course further down into the brain tissue. The scale of this space is variable in size, and can become large enough to show up on MRI in cases of cerebrovascular disease (Groeschel et al, 2006).

These perivascular spaces are supposed to be inside the blood brain barrier. When one's bloodstream is filled with intravenous contrast media, they are expected to remain free from it, unless the blood-brain barrier is broken - eg. by stoke, tumour, what have you (Wardlaw et al, 2020). Their purpose is unclear, but authoritative sources (i.e. the sort of people who get invited to write for Nature Reviews) seem to attribute to them the same sort of sewer-like sanitation role as one might expect from the lymphatics in the rest of the body. Metabolic wastes and cellular debris appear to be cleared by these channels, and these processes seem to increase during sleep. Logically, this means that there must be some route for these wastes to exit, and in fact there are pial fenestrations that allow communication between the subpial space and the arachnoid space. These pores are probably responsible for some of the effect of local anaesthetics injected into the subarachnoid space, as Reina et al (2004) had speculated. Moreover a fair amount of fluid seeps from the cerebral interstitium into the CSF through those pores, to the effect that up to 60% of the CSF could have originated there. They are not exactly few in number: the image below is from the same 2004 paper, a SEM microscope image of spinal pia mater at 220 magnification illustrating numerous fenestrations.

Anyway: we are veering far off course, as the CICM syllabus contains no mention of any of this, and historical exam questions have never asked for any discussion of pial fenestrations or perivascular spaces. The main point of this digression, if it had one, would have to be that the blood-brain barrier is not perfect, and there are parts of the CNS where it is intentionally discontinuous, allowing the interstitial fluid of the nervous system to communicate, albeit very indirectly, with the CSF or the bloodstream.

Areas of the CNS which do not have a blood-brain barrier

"What parts of the brain lie outside of this barrier (and why)" was an important part of the answer for whoever graded Question 8 from the second paper of 2009. There are several intentional gaps in the blood-brain barrier which are mainly characterised by the reappearance of capillary fenestrations and the disappearance of tight junctions. These areas are mainly deep central structures, predominantly in the midline, and clustering around the ventricles, which has given rise to them being grouped as "circumventricular organs". That's the string you should enter into a search engine to find all the better articles about these areas, and it turns up some pearls. They are discussed in the greatest detail by Duvernoy & Risold (2007), who presented them in the form of a histological atlas; their homeostatic function is detailed by Bennaroch (2011), endocrine function by Ganong (2000) and developmental origins by Kiecker (2018). None of these papers are free, and moreover some are basically unreadable in their original form, which means they had to be blended and strained to present the reader with a condensed readily digestible product.

In short, the list of brain areas lacking a blood-brain barrier is:

• Area postrema
• Choroid plexus
• Pineal gland
• Organum vasculosum lamina terminalis
• Subfornical organ
• Median eminence
• Posterior pituitary
• Preoptic recess

In case you are wondering where these are exactly (not that you're ever going to use that information) here is a nice diagram from (blows dust off, squinting at the embossed letters) Kahle's 1979 Atlas de poche d'anatomie (tome 3 : Système nerveux et organes des sens).

"And why" will be explained in the briefest possible way below:

The area postrema is the most geographically distant of all the circumventricular organs, sitting all the way out in the medulla, at the border between the brain and spine, on the floor of the fourth ventricle. Weirdly, wherever the CVOs are discussed, the area postrema has ended up being discussed last, as if each author has decided that they will go through these organs in some kind of front-to-back order. This makes no sense and we will not obey convention here. The area postrema is an excellent posterchild for a sensory area of the brain which needs to be outside the blood-brain barrier to fulfill its role.

Consider: one of the main roles of that barrier is to protect the delicate neurons from the toxic horror  chemicals of the blood stream. This would obviously be counterproductive for the area postrema, whose main job is to detect and characterise those chemicals. It is closely connected to the nucleus of the solitary tract and motor dorsal vagal nucleus, via which its sensory function is integrated with the autonomic nervous system. By detecting vasopressin and angiotensin-II, this sensor mediates a lot of the sympathetic and vagal responses to changes in circulating volume and blood pressure, for example by changing the sensitivity of the baroreflex to accommodate a lower range of pressures in response to vasopressin release (analogous to the intensivist changing the goal MAP to 60 mmHg when the patient's blood pressure just won't stay up without a trickle of noradrenaline). This, probably, is its most important regulatory role. In case more detail is required regarding these mechanisms, Hasser et al (2001) presents a solid overview.

The area postrema chemoreceptor trigger zone also has an important role to play in vomiting, and though this might be of trivial importance for the organism (in comparison to its autonomic role), it is of paramount importance to the trainee of intensive care or anaesthesia, because it is the target of antiemetic drugs. Specifically, it is thought to be the site of action of dopamine receptor antagonists and serotonin receptor antagonists.

The choroid plexus is not one spot but on fact several patches of CSF-secreting tissue in the posterior parts of the lateral ventricles and in the roof of the third and fourth ventricles. This secretory tissue contains many fenestrated capillaries which are separated from the CSF by a layer of cuboidal epithelial cells. The attentive reader will point out that their is no brain there, and so whatever barrier this would have been, you couldn't have called the "blood-brain barrier". The choroid plexus probably needs to be referred to as part of a "blood-CSF barrier", as it separates the blood and the CSF and actively modifies the composition of the CSF it secretes by various active and facilitated transport mechanisms. It is also usually not listed among the circumventricular organs, even though its anatomical position is by definition in and around the ventricles - probably because, unlike the other CVOs, it does not have a specific homeostatic controller role which relies on sampling the biochemistry of the blood.

The pineal gland is an endocrine organ tucked between two halves of the thalamus, secreting melatonin into the circulation. In order for this secretory function to be fulfilled, the pineal gland is supplied with fenestrated capillaries which allow the melatonin to enter the bloodstream directly. It does not appear to have any regulatory function related to sensing humoral mediators, and also often ends up excluded from the list of circumventricular organs.

Organum vasculosum lamina terminalis, or the vascular organ of the lamina terminalis for anybody who did not go to a private Catholic school, is a patch of brain tissue that serves a mainly sensory role, detecting changes in sodium concentration and serum osmolality (for which it obviously needs to have an open blood-brain barrier). It feeds this information to the hypothalamus and is essential to the process of osmoregulation. Its anatomical location is along the ventral part of the anterior wall of the third ventricle.

Subfornical organ, imaginatively named because it sits at the inferior surface of the fornix, is also a sensory organ, developmentally separated from the organum vasculosum only by some embryological gerrymandering, where the area between them ended up being rezoned as the anterior commissure. It also senses osmolality and sodium, as well as (apparently) angiotensin.

The median eminence is the area immediately posterior to the posterior pituitary, where the hypothalamus secretes its hormone-releasing hormones into the portal capillary circulation. The gap in the blood-brain barrier here is mainly meant to facilitate this local endocrine function, i.e. it is really a portal-brain barrier.

The posterior pituitary secretes oxytocin and vasopressin into the systemic circulation, and for this reason needs to have fenestrated capillaries. The anterior pituitary is not listed along with the posterior, even though it clearly also has a secretory endocrine role and fenestrated capillaries, but because it arises from oral ectoderm, it is not strictly speaking a brain structure, and so its lack of a blood-brain barrier is understandable (i.e. it is not brain). 

The preoptic recess of the anterior hypothalamus is involved in the coordination of thermoregulation and mating behaviour (two weirdly unrelated roles). The interruption of the blood-brain barrier here is presumably due to the need to sense and respond to sex hormones. According to McKinley et al (2015), it also has numerous other homeostatic functions which in many ways duplicate the activity of many other circumventricular organs.