Key facts
- What it is
- A highly selective barrier between the blood and the brain tissue
- Formed by
- The tightly sealed cells lining the brain's capillaries
- Held together by
- Tight junctions, supported by astrocyte end-feet and pericytes
- Lets in
- Glucose, oxygen, and small lipid-soluble molecules, via transporters or diffusion
- Keeps out
- Most toxins, pathogens, large molecules, and many drugs
- Diffusion cut-off
- Roughly under 400 to 500 daltons, with few hydrogen-bond donors
- Second line of defence
- ATP-driven efflux pumps, above all P-glycoprotein, which eject lipophilic molecules that got in
What the barrier is
The blood-brain barrier is not a single membrane or wall but a property of the blood vessels that thread through the brain. Everywhere else in the body, the walls of the smallest vessels, the capillaries, are somewhat leaky, letting fluid and small molecules pass freely between the blood and the surrounding tissue. In the brain, those same capillary walls are sealed almost watertight. The result is a border that the brain controls with great precision, admitting some substances and refusing others.
The purpose of this seal is protection and stability. Neurons signal by finely tuned shifts in ions and chemicals, and that signalling would be thrown into chaos if the brain's chemistry rose and fell with every meal or every fluctuation in the blood. The barrier holds the brain's internal environment steady, shields it from circulating toxins and infections, and keeps out molecules that would interfere with signalling. It is, in effect, a customs checkpoint around the most sensitive tissue in the body.
The scale of the interface is easy to underestimate. The brain is threaded with a dense mesh of capillaries so fine that almost no neuron sits far from one, giving a total vessel surface measured in many square metres folded into a small volume. Every point on that vast surface is sealed. The barrier is therefore not a single gate at the edge of the brain but a property spread across the entire vascular network within it, which is part of why it is so effective and so hard to circumvent.
How it is built
The barrier is the work of several cell types acting together, a partnership often called the neurovascular unit. Three players do most of the job.
Endothelial cells
The cells lining the inside of the capillary. In the brain they are joined by tight junctions and carry few of the pores and vesicles that make vessels leaky elsewhere, so they form a nearly continuous seal.
Tight junctions
Belts of protein that stitch neighbouring endothelial cells together, closing the gaps between them. They force any molecule entering the brain to go through the cells rather than slip between them.
Astrocytes and pericytes
Astrocyte end-feet wrap around the vessel and secrete the signals that instruct the endothelium to build and hold the barrier, while pericytes embedded in the vessel wall regulate blood flow and contribute their own signals to the same programme. They do not form the seal. They order it.
Endothelial cell: one of the flat cells that line the inside of every blood vessel, forming the surface the blood actually touches. Pericyte: a contractile cell wrapped around the outside of a capillary, embedded in its wall, which helps regulate blood flow and helps hold the seal. Neurovascular unit: the working partnership of endothelial cells, pericytes, astrocyte end-feet, the basement membrane between them, and the neurons they serve, treated as one functional structure because none of them maintains the barrier alone.
Tight junction: a sealed contact between two adjacent cells that blocks substances from passing through the gap between them. In the brain's capillaries, tight junctions are what close off the usual leaky route, so nothing crosses without passing through the endothelial cells themselves. They are the structural heart of the barrier.
Why tightness is the same thing as selectivity
The two most-repeated facts about this barrier, that it is tightly sealed and that it is highly selective, are usually presented as separate virtues. They are not separate. The second is a consequence of the first, and the argument takes three sentences.
Consider the two possible routes a molecule can take out of a capillary and into the tissue around it.
The paracellular route: between the cells
Paracellular means "beside the cell": through the gaps between neighbouring endothelial cells, without entering any of them. Everywhere else in the body this route is open, and it is the main way small molecules and fluid leave the blood. It is also completely unselective. A gap does not inspect what goes through it, and the cell beside the gap has no say in the matter. In the brain, tight junctions weld this route shut.
The transcellular route: through the cells
Transcellular means "across the cell": in through the membrane on the blood side, across the cytoplasm, out through the membrane on the brain side. With the paracellular route closed, this is the only way in. And a molecule that must pass through a cell can only do so on terms the cell sets, because the cell owns both membranes it has to cross and decides what proteins to put in them.
This is what converts a wall into a customs post. Sealing the gaps does not merely make the barrier stronger, as if selectivity were an extra feature bolted on afterwards. It forces every crossing through a checkpoint that the endothelial cell controls. By expressing a transporter, the cell admits a molecule. By not expressing one, it refuses. Glucose gets in because the cell chose to build a glucose transporter; a similarly sized water-soluble drug does not, because no cell ever built a transporter for it. The barrier's selectivity is a consequence of its tightness, not a separate property alongside it.
This layered design is what makes the barrier both strong and adjustable. The tight junctions provide the seal; the astrocytes and pericytes maintain and regulate it, and the next section explains how.
The barrier is taught, not inherited
The statement that the barrier weakens if you remove the astrocyte signals is easy to write and easy to skim past. What signals? The question matters, because the answer overturns the natural assumption that the barrier is a property of a special kind of blood vessel.
It is not. Astrocytes and the surrounding neural tissue secrete developmental signalling molecules, the same molecules that pattern tissues during embryonic development, and the ones repeatedly implicated are sonic hedgehog, the Wnt family, and retinoic acid. It is worth being precise about the source, because the shorthand overstates the astrocyte. Sonic hedgehog is the best-established genuinely astrocytic signal. Wnt and retinoic acid are also required for barrier induction, but they come chiefly from neural progenitors and radial glia rather than from mature astrocytes. These are instructions. They tell the endothelial cells sitting next to them what kind of endothelial cell to become: to assemble tight junctions, to install the transporters the brain needs, to shut down the leaky vesicle traffic that ordinary capillaries run, and to switch on the efflux pumps described further down this page. Pericytes contribute their own signals to the same programme, which is why animals lacking pericytes have leaky brain capillaries.
The barrier is not a property of blood vessels. It is a property that brain tissue teaches its blood vessels to have. The experimental demonstration is the kind that settles an argument: transplant a piece of blood vessel from outside the nervous system into brain tissue, and it acquires barrier properties it never had before. Transplant brain tissue somewhere else, and the vessels growing into it acquire them too. The instruction travels with the brain tissue, not with the vessel. This also explains the barrier's fragility. An instructed state has to be maintained, and anything that damages the astrocytes, as stroke, infection and chronic inflammation all do, degrades the instructions and, with them, the seal.
What it lets in and keeps out
The barrier is selective, not simply closed. A brain sealed off from everything would starve. Its task is to admit precisely what the brain needs while excluding what would harm it, and it does this molecule by molecule.
Let in
Oxygen and carbon dioxide, which cross freely. Glucose and amino acids, carried in by dedicated transporter proteins. Small, fat-soluble molecules, which dissolve straight through the cell membranes for the reason given in the section on fat-solubility. Water, within limits.
Kept out
Most bacteria and other pathogens. Many circulating toxins. Large molecules such as most proteins and antibodies. Water-soluble molecules that lack a transporter, including the great majority of drugs. And, actively pumped back out, many of the fat-soluble molecules that did get in.
The theme is that entry is a privilege the brain grants deliberately. Fuel such as glucose does not diffuse in; it is escorted across by specific carriers. This means the barrier does not merely block, it actively chooses, and that selectivity is exactly what makes it so difficult to smuggle a medicine past.
Glucose is worth singling out, because the brain is unusually dependent on it. Neurons burn glucose almost exclusively for energy and hold very little in reserve, so a steady supply across the barrier is not a convenience but a necessity. A dedicated glucose transporter, present in large numbers in the brain's endothelial cells, keeps that supply flowing. It is a neat illustration of the barrier's logic: the more vital a substance, the more likely the brain has built a specific channel to admit it, while everything unrequested is left outside.
How molecules cross
Given so tight a seal, how does anything get through at all? Every route is now transcellular, so every route must solve the same problem: getting across a cell membrane. There are four solutions, and which one a molecule can use depends on its size, its charge, and its solubility in fat.
Simple diffusion
Small, uncharged, fat-soluble molecules, including oxygen, carbon dioxide, and some drugs and anaesthetics, dissolve straight through the cell membranes. Being small and lipid-soluble is the classic passport into the brain, and the next section explains why.
Carrier transport
Essential nutrients that are not fat-soluble, such as glucose and certain amino acids, are ferried across by specific transporter proteins built into the endothelial cells. Each carrier is tuned to its cargo.
Receptor-mediated transport
Some large but vital molecules, such as insulin and iron-carrying transferrin, bind receptors that pull them across in vesicles. This route is a target for drug designers hoping to hitch a ride into the brain.
Active efflux
The barrier also works in reverse. Pumps such as P-glycoprotein grab certain molecules that have entered and eject them back into the blood, actively keeping many drugs out even after they slip in. This turns out to be the single most consequential fact on this page, and it has a section of its own below.
Why fat-solubility is the passport
"Small and lipid-soluble" is repeated in every account of the barrier, and it is almost never explained. It sounds like an arbitrary rule that has to be memorised. It is not arbitrary at all: it follows directly from what a membrane is made of, and once you see the mechanism you can predict the rule rather than remember it.
A cell membrane is a double layer of lipid molecules. Each lipid has a water-loving head and two fatty, water-hating tails, and the two layers are arranged tail-to-tail, so that the fatty tails all point inward and are hidden from the water on either side. The interior of a membrane, the region a molecule has to traverse, is therefore not a hole and not a gate. It is a thin sheet of oil.
Now ask what happens when a molecule tries to enter it.
The water-soluble molecule is turned back
A water-soluble molecule is water-soluble precisely because it can form hydrogen bonds with water. In the bloodstream it does not travel alone: it sits inside a shell of water molecules bonded to it. To enter the oily core of the membrane it would first have to strip that shell off, breaking every one of those hydrogen bonds, and the fatty interior offers it nothing in return, because there is nothing there to bond to. The energy it would have to spend is greater than the energy it would get back. So it does not go. It is not blocked by any structure. It simply will not pay.
The lipid-soluble molecule walks through
A lipid-soluble molecule carries no such shell, because it is not hydrogen-bonded to water in the first place. Dropping it into the fatty core of a membrane is chemically much like dropping it into oil, which is exactly the environment it is happiest in. It dissolves into the membrane on the blood side, diffuses across, and dissolves out of it on the brain side. No transporter, no receptor, no permission. It walks through the wall because, to a molecule like this, the wall is a solvent.
The rule in one line: solubility in the membrane is the passport, and the membrane is made of fat. Everything else follows. Oxygen and carbon dioxide cross freely because they are small, uncharged and fat-soluble. A charged molecule, an ion, essentially never crosses unaided, because a charge is the most tenacious water-attracting feature there is. Glucose, which is water-soluble, has to be carried in by a protein, because on its own it would never leave the blood.
The practical cut-off
Size then imposes a second condition, and this one comes with a number, which matters because a page that discusses drug delivery at length owes the reader the figure they would actually need to apply the rule.
As a working guideline, a molecule has a realistic chance of crossing the barrier by simple diffusion if it is roughly under 400 to 500 daltons in molecular mass and has few hydrogen-bond donors, conventionally taken as no more than about eight to ten hydrogen bonds in total. Each hydrogen-bond donor is one more bond to water that must be broken at the membrane, so every one of them raises the toll. The mass limit reflects the fact that even a lipid-soluble molecule must part the fatty tails to squeeze between them, and a large molecule has to part more of them.
These are not sharp thresholds and exceptions exist in both directions, but they are the numbers medicinal chemists actually design against, and they explain the shape of the drug problem at a glance. Most small-molecule drugs are already too large or too polar. Antibodies and other biological drugs, at tens or hundreds of kilodaltons, are not near the limit; they are two orders of magnitude past it, which is why practically none of them reach the brain unaided.
The upshot is a rule of thumb with one large catch. To enter the brain unaided, a molecule should be small, uncharged, and fat-soluble. Anything else needs a dedicated transporter or a clever trick. And even a molecule that satisfies every condition may still be thrown straight back out, for the reason set out next.
Active efflux: why lipophilic drugs still fail
Follow the logic of the previous section to its natural conclusion and you arrive at an obvious strategy: if fat-solubility is the passport, then make the drug more fat-soluble. Medicinal chemistry spent a long time trying exactly that, and it largely does not work. Understanding why is the most useful thing on this page.
P-glycoprotein (P-gp): an ATP-driven efflux pump embedded in the membrane of the brain's endothelial cells, facing the blood. It binds molecules that have entered the cell and pushes them back out into the bloodstream, burning ATP to do it. It is not a filter, which is passive. It is a pump, which is active, and it works against the direction a molecule wants to go.
Two features of P-gp make it devastating for drug delivery, and they compound each other.
Its substrate recognition is unusually promiscuous
Most transport proteins are exquisitely specific: a glucose transporter carries glucose and is not interested in anything else. P-gp is the opposite. It recognises an enormous and structurally diverse range of molecules, which is precisely what you would want of a general-purpose defence against xenobiotics, chemicals foreign to the body. It cannot know in advance what poison it will meet, so it evolved to grab a very wide class of things. A novel drug is, to P-gp, simply another unfamiliar foreign molecule.
It is biased towards exactly the molecules that could otherwise get in
This is the cruel part. P-gp preferentially handles lipophilic compounds, the very ones whose fat-solubility would let them diffuse across the membrane in the first place. Water-soluble molecules were never getting in anyway and need no pump to stop them. The pump is aimed at the only molecules that pose a threat, and so its substrate profile overlaps almost perfectly with the profile a drug designer is trying to hit.
The trap: the barrier lets the lipid-soluble molecule dissolve into the endothelial cell, and then throws it straight back out into the blood before it can reach the other side. Making a drug more lipophilic therefore does not simply raise its chance of getting in. It often raises, at the same time, its affinity for the pump that will eject it, so the two effects cancel and net brain penetration barely moves. This is why "just make it more lipid-soluble" is not a solution, and it is a major reason so many compounds that work beautifully in a dish fail in the body. They are not failing to cross. They are crossing, and being deported.
P-gp is also not alone. It is the best-known member of a family of ATP-driven efflux transporters at the barrier, and together they constitute a second line of defence sitting behind the tight junctions: the junctions decide the route, and the pumps police it. A complete picture of the barrier is therefore not a wall with a few doors. It is a wall with a few doors and a guard who searches everyone who came through and sends most of them back.
Why it matters for medicine
For all its value in protecting the brain, the barrier is a formidable problem for treating it. The same seal that keeps out toxins keeps out most medicines, and this is one of the great obstacles in neurology and psychiatry.
The drug delivery problem: the large majority of small-molecule drugs, and virtually all large biological drugs such as antibodies, cannot cross the barrier in useful amounts. A treatment can be effective against a brain tumour or infection in the laboratory yet fail in the body simply because it cannot reach its target. And as the previous section showed, the obvious fix is a dead end: pushing a molecule towards higher lipophilicity tends to hand it to P-glycoprotein, which pumps it back out. So the search for ways across has largely moved elsewhere: keeping molecules under the size and hydrogen-bond limits while designing them to be poor substrates for the efflux pumps; attaching a drug to a natural ligand such as transferrin so that a receptor carries it across; inhibiting the pumps themselves, which is difficult to do safely because they protect the brain from genuine toxins as well; or temporarily and locally loosening the barrier, for instance with focused ultrasound, so a drug can pass in one small region. Each strategy attacks the same puzzle: how to deliver a medicine to the one organ most carefully walled off from the blood.
The flip side is that the barrier also explains why some drugs have few brain effects by design. A medicine deliberately made unable to cross can act on the body while sparing the brain, avoiding drowsiness or other central effects. The barrier, then, is not only an obstacle but a tool that drug designers work both with and against.
Where the barrier is weaker
The barrier is not uniform. In a few small, specialised regions it is deliberately leaky, because those areas need direct contact with the blood to do their jobs. These are the circumventricular organs.
Circumventricular organ: one of several small regions around the brain's ventricles where the barrier is weak or absent. Here the brain is allowed a direct chemical window onto the blood, so it can sense the body's state or release hormones into the circulation.
These windows serve clear purposes. The area postrema, for instance, samples the blood for toxins and can trigger vomiting, a protective reflex that only works if that patch of brain can taste the bloodstream directly. Other such regions let the hypothalamus monitor hormones and signals of hunger, thirst, and salt balance, or release its own hormones into the blood. The exceptions, in other words, are not flaws but features: places where the brain needs to talk to the body chemically, and the barrier steps aside to let it.
When the barrier fails
Because the barrier is actively maintained, it can break down, and when it does, the consequences are serious. A leaky barrier lets in fluid, proteins, and cells that the brain normally excludes, and the results range from swelling to inflammation and neuronal injury.
After loss of blood flow
When part of the brain is deprived of blood, the barrier in that region is damaged. Fluid then leaks into the tissue, causing the swelling that often does much of the harm in the hours after a stroke.
Meningitis and encephalitis
Some pathogens breach the barrier to infect the brain, and the resulting inflammation loosens it further. This vicious circle lets more immune cells and fluid in, driving the swelling seen in serious brain infections.
In chronic disease
In conditions such as Alzheimer's and multiple sclerosis, the barrier grows leaky over time. This is increasingly seen not just as a symptom but as part of the disease process, letting harmful substances reach vulnerable neurons.
A breached barrier is therefore both a result and a cause. Injury or disease can damage it, and once damaged, it allows further injury, so protecting or repairing the barrier has become a goal of treatment in its own right. The gatekeeper, when it fails, becomes part of the danger it was built to hold back.
Sources
- Kandel ER, Koester JD, Mack SH, Siegelbaum SA. Principles of Neural Science. 6th ed. McGraw-Hill; 2021.
- Purves D, Augustine GJ, Fitzpatrick D, et al. Neuroscience. 6th ed. Oxford University Press; 2018.
- Abbott NJ, Patabendige AAK, Dolman DEM, Yusof SR, Begley DJ. Structure and function of the blood-brain barrier. Neurobiology of Disease. 2010;37(1):13-25.
This page is an educational reference. It is not medical advice and does not diagnose or treat any condition.