Your brain is an energy hog. It weighs less than 2 percent of your total weight yet consumes one fifth of your body's energy. The brain draws its fuel—oxygen and glucose—from blood delivered by a whopping 400 miles of blood vessels. Lined up end to end, all that vasculature would extend from New York City to Montreal.

These blood vessels are astonishingly dynamic. They tune the flow of blood to respond to the brain's needs from moment to moment. When certain brain areas work hard at something, more blood flows to those regions to help them refuel. Vessels do this by dilating near the spots that need a supply boost. This widening coaxes blood to reroute, much as customers in a busy store redistribute themselves whenever a new checkout line opens.

Fuel for neurons is limited, so your blood vessels must carefully choreograph every instant to sustain your brain. Yet what if the brain's blood vessels fall out of sync with their neurons? If the vasculature fails to deliver more blood when neurons need it, those cells might starve. In the short term, cognition could suffer. Longer term, entire networks of brain cells could wither away.

Historically, neuroscientists have seen blood vessels in the brain as mundane roadways, irrelevant to the neurons they support. Yet a city needs its roads. More than a simple conduit for noisy cars, transport infrastructure profoundly alters how we function. When Hurricane Sandy hit New York, for example, the rising water levels and power outages disrupted distribution networks for people, food and supplies, bringing the city to a standstill.

Blood flow is equally vital to brain function, and there are compelling reasons to think that dysfunction in one could impair the other. Brain scans have shown us that the brains of healthy individuals behave differently from those of people with Alzheimer's disease, attention-deficit/hyperactivity disorder, schizophrenia, depression, autism or multiple sclerosis, to name just a few conditions. The standard interpretation is that neuronal activity has deviated from a typical state.

There is a catch, though. Functional magnetic resonance imaging (fMRI), the technique most widely used for imaging brain activity, measures changes in blood flow as a proxy for neuronal activity. If the relation between blood flow and neurons has gone off the rails, fMRI scans will still deviate from the norm no matter what the neurons are up to. Scientists are left in the dark as to which brain disorders might solely affect neurons and which might also disturb cerebral blood flow.

To get to the bottom of this, my laboratory has embarked on a mission to uncover how and when blood vessels and neurons might fall into discord. The evidence we have found suggests that this relation can indeed go awry and that it could contribute to—or even cause—neurological or psychiatric disorders. Fortunately, we might already possess the tools we need to correct the patterns of blood flow in the brain.

Written in Blood

Twenty years ago fMRI revolutionized the way that researchers study the human mind. This imaging technology can produce snapshots of the brain responding to a stimulus—a sound, a picture, a suggestion. Its magic comes from the unique properties of hemoglobin, the iron-rich protein in red blood cells that carries oxygen around the body.

Oxygen is invisible to an MRI machine as long as it is bound to hemoglobin. Once the oxygen detaches, the deoxygenated hemoglobin left behind acquires a magnetic property that an MRI machine can detect. Oxygen is an essential fuel for neurons, and the general idea is that busy neurons need to replenish their energy supply by pulling oxygen from the blood vessels lacing through the brain. The process is the neuronal equivalent of nibbling on a sandwich to keep your energy levels up. But things get a little tricky because fMRI scans do not actually show oxygen getting consumed.

When activity in a certain brain region revs up, fresh oxygenated blood rushes in and flushes out the deoxygenated hemoglobin. This new overabundance of oxygen is interpreted by fMRI as a sign of neuronal activity. Rather than noticing that your neurons are nibbling on their sandwich, fMRI detects the delivery truck that is bringing you a huge portion of cake and ice cream for dessert.

Scientists assume that these blood flow changes reflect what nearby neurons are up to. Those inferences come with large caveats. First, the blood flow response is slow—neurons fire within milliseconds, but a corresponding increase in blood flow peaks three to five seconds after the event. So neurons clearly do not need their cake and ice cream to activate.

A second major concern is that we do not yet know how neurons communicate with blood vessels. Researchers have only a rough understanding of how variation in neurons' signaling—differences in the amplitude, frequency and duration of their activity—might tweak blood flow. Third, neuroscientists have barely considered what might happen if the brain's blood flow does not march in lockstep with the demands of neurons.

My colleagues and I believe that a mismatch might arise in numerous ways. The messages neurons send to alert blood vessels to their needs could get corrupted. The blood vessels themselves might lose some of their flexibility—constricting too easily, maybe, or dilating unduly. An acute brain injury, perhaps from a stroke or a head trauma, could also throw the vasculature and the affected region out of sync.

In short, the central premise of fMRI is that changes in the brain's blood flow are tightly coupled to specific neuronal events. If blood vessels fail to meet the demands of neurons, the implications could be enormous.

Strange Signals

To really be sure that it is the blood flow coupling that is broken, rather than the neurons themselves, you would need a way to measure both blood flow changes and the tiny electrical blips of neurons. This is easier said than done. If we had a robust way to map neuronal activity in the brain in real time, we would not need fMRI in the first place.

Electroencephalography (EEG) is one option. It can measure neuronal activity through electrodes placed on the scalp. Yet collecting clean EEG data inside the huge, electrically noisy magnet of an MRI machine is incredibly challenging. Researchers are still puzzling over the best ways to combine techniques to assess coupling in the human brain.

In my own work, we are using home-built microscopes that let us peer directly into the living rodent brain. In one study, published in 2013, we tackled a longstanding mystery in fMRI data. For almost as long as the imaging technique has been around, scans of the brains of infants and young children have looked very different from those of adults. In adults, an uptick in neuronal activity usually means an increase in the fMRI signal. In infants, many researchers saw a decrease in the signal.

My student Mariel G. Kozberg and I probed this oddity by observing blood flow in the brains of rodent pups during their first month of life. We took high-speed movies of the animals' exposed brain surfaces while we delivered mild zaps to their paws—comparable to a gentle touch. Our images showed us how this stimulus influenced blood flow in the brain area that corresponds to sensation in the paw. Subtle shifts in the color of blood also told us how much oxygen the blood was carrying.

We knew that neurons in the region were responding to the stimulation of the paw, yet in the youngest pups our videos revealed no increases in cerebral blood flow. In fact, we saw blood flow in the brain decrease. By observing rats of different ages, we discovered that over time the blood flow response gradually began to resemble that of an adult rat. This pattern suggested to us that the neurovascular system in a newborn brain has not yet synced up.

In more experiments, we found that when we delivered stronger stimuli, the blood pressure of the young rats spiked, equivalent to a startle reflex that makes a newborn cry or a sudden shock that gets your heart pounding. This increase in blood pressure caused blood to surge indiscriminately into the brains of the newborn pups. In a mature brain, a system known as autoregulation acts like a floodgate to protect against surges of blood. Our results suggest that this autoregulation system is also not mature in the developing brain.

This idea makes sense, of course—many aspects of the developing brain are in flux. We already know that after a baby is born new neurons continue to grow and that new connections form but also dissolve as the broad outlines of the brain's internal architecture take shape. Our group hypothesizes that the brain's mechanisms for deploying blood develop in tandem with these other processes.

That our neurovascular systems start out life incomplete raises a couple of concerns. First, fMRI might be blind during the early stages of brain development. Second, my colleagues and I have come to believe that some developmental problems might in fact stem from abnormalities in the way that neurovascular mechanisms mature. We are starting to explore this possibility by building new kid-friendly brain-imaging techniques to assess the emergence of vascular coupling in babies and young children.

Mind-Body Connections

Other important insights come from examining conditions known to affect both blood vessels and cognition. Take stroke, for example. During a stroke, a blocked or ruptured blood vessel causes a region of the brain to starve. If the vessels in that area cannot find a way to reroute the blood supply, neurons will begin to die.

Here again fMRI scans might be trying to tell us more than we once thought. Clinicians had hoped that brain imaging could help predict a stroke patient's recovery. The idea was that if brain scans show that the area affected by a stroke is responding to stimuli, the patient is on the road to recovery. Yet the data have turned out to be far too mixed to serve as an oracle. Some patients who showed promising fMRI activity did not recover well, whereas others whose brain scans were less encouraging ultimately regained function in that area.

The problem is that when the vessels themselves are injured, blood flow routes might be blocked or damaged, and any number of biological mechanisms can come into play to try to save what is left. It seems almost impossible that the usual blood flow responses to neuronal activity would continue amid such catastrophe.

Yet we need not despair. Although the vasculature might not be speaking its usual language, fMRI can still listen. If we can learn to translate its messages, we might be able to understand how the brain tries to heal itself after a stroke—and better predict and guide recovery.

In the rest of the body, several disorders that afflict vasculature likewise affect the brain. Diabetes damages blood vessels and impedes memory and attention, as well as raising the risk of dementia. Similar cognitive troubles are seen in patients with heart failure. Hypertension and inflammation—risk factors for cardiovascular disease—increase a person's likelihood of developing Alzheimer's.

Physicians have circled these clues and connections for a number of years. We know, for example, that Alzheimer's has a blood flow component to it. In some patients, the brain receives less blood flow overall than it would in a healthy state. Such a deficit can make it harder to synthesize proteins crucial for learning and memory. Treatments for this aspect of Alzheimer's have focused on boosting blood flow brain-wide, but to little avail.

Those therapies might have missed an important clue. Perhaps the real problem is that the brain is not responding properly to local demands for fresh blood. One potential mechanism suggested by research is that beta-amyloid protein fragments—which form the characteristic plaques in Alzheimer's—may collect along parts of the vasculature. Beta-amyloid can trigger a chain of events that decreases the amount of nitric oxide available in the brain. Nitric oxide is a vital signaling molecule that instructs vessels to dilate. If the coupling of blood flow and neurons has indeed gone awry in Alzheimer's, improving the tuning of the blood supply (as opposed to simply ensuring higher overall flow) could offer a new strategy for treating some forms of dementia.

Numerous studies have noted altered responses in the fMRI scans of patients with Alzheimer's, diabetes and some forms of heart disease. If these anomalies are a sign that blood flow regulation is faltering, it may be possible to intervene and rescue imperiled neurons before it is too late.

The Missing Link

A critical next step is to nail down how it is that neurons and blood vessels communicate. We have a lot to learn. For example, it is tempting to think that hungry neurons use up local oxygen supplies, triggering an increase in blood flow, but the reality is not quite that simple. Even when a rodent is inside a hyperbaric oxygen chamber, which saturates the brain with oxygen, the animal will still exhibit a surge of blood to an area where neurons are hard at work. The same thing happens when very high levels of glucose are available.

So the call for more blood involves something more than a simple “low fuel” alarm. The quest to identify this sequence of events has focused largely on the brain's support cells, in particular astrocytes and pericytes. Astrocytes, which are star-shaped cells, are interspersed with neurons and can be found clinging to blood vessels, wrapping their appendages around like ivy on an old drainpipe. Pericytes are known to spiral around the tiny capillary vessels of the brain. Recent studies suggest that pericytes may be able to squeeze and relax, giving them the power to fine-tune the flow of blood. Both cell types can alter blood flow in the brain, but the complete picture is not yet clear.

Recent work in my lab has focused on another component: the vessels themselves. The brain's vasculature is well organized, with large arteries and veins coursing along the surface. Smaller vessels branch off and dive into the brain tissue, routing blood through dense beds of tiny capillaries that weave among neurons. To observe exactly what series of events generates the signal seen in an fMRI scan, we used a high-speed camera to take pictures of the surface of a rat brain.

After stimulating the rat's paw, we first saw a rapid blush in the brain's sensory region as red blood cells became denser in the capillary beds below the surface. Within a fraction of a second, we saw small arteries dilate at the surface, followed by the larger arteries from which they branched.

Imaging more quickly, we spotted the wave of dilation traveling along the arteries, in many cases against the direction of blood flow. The wave moved faster and farther than could be explained by astrocytes and pericytes alone. With these data, we essentially had a baseball card with a list of performance stats. Could we scout a player (or team of players) who could run that fast and that far?

Our search did not take long. In fact, the answer was right in front of us. We stumbled across a paper on how blood vessels in the body's muscles send signals along a vessel at very high speeds, almost like a wire conducting electricity. The key is the vessel's innermost layer, a sleek mosaic of endothelial cells. In an artery (or smaller arteriole), this inner layer is wrapped in a chunky jacket of smooth muscle cells. The muscle cells are capable of squeezing to constrict the vessel or relaxing to let it dilate. These tiny muscle cells can receive commands from cells around the outside of a blood vessel, or they can be directed from the inside by the endothelial cells. This inner layer of cells can whisper the instruction to dilate or constrict using a range of different signaling mechanisms.

The elegance of the system is that once one endothelial cell hears that blood is needed, it can broadcast this whisper far and wide along the length of the blood vessel. Each smooth muscle cell in the path of the signal will dutifully obey the order to dilate. The match between this picture and our baseball card stats was too close to ignore.

The best way to show that this secret pathway was at work in the brain was to attempt to break it. We used a remarkably simple yet precise technique. We injected a special dye into the bloodstream. When hit by bright blue light, this dye produces oxygen free radicals that damage only the endothelial cells.

We chose an arteriole on the surface of the brain and shone a very thin line of light from a laser to effectively slice the vessel's endothelium at a specific point. The outer layers of the vessel remained intact. When we stimulated the paw again, we found that the disruption of this tiny section of the endothelium was enough to stop the dilation from spreading along the vessel.

Next we shone the laser on a large area of the brain's surface, which prevented all the arteries on the surface from dilating. By applying chemicals that we know should cause smooth muscle cells to relax and dilate, we were able to confirm that the vessels were not themselves irreparably damaged. These experiments told us that signaling within the endothelium is a critical component in generating blood flow increases in the brain.

Our finding has some intriguing implications. First, if we can discover exactly which signals tell the endothelium to start a wave of dilation, we may finally be able to say for sure what fMRI is showing. Also important is that mechanisms already known to regulate blood flow elsewhere in the body are also involved in the brain. This outcome might sound obvious, but until now many researchers have assumed that blood flow in the brain follows different rules. Now it seems possible that diseases affecting the cardiovascular system throughout the body could have a direct effect on brain health. The same mechanisms that lead a diabetic patient to develop a foot ulcer might also explain cognitive symptoms.

Fixing Blood Flow

Often scientists and physicians express frustration at the difficulty of developing drugs for brain disorders, for a simple reason—unlike the rest of the body, the brain has a protective wall, called the blood-brain barrier, that shields it from most molecules in the bloodstream. The endothelium is the blood-brain barrier, however, so it is closely exposed to chemicals in the bloodstream. This direct contact could be good or bad news.

The good news is that the endothelium is an accessible target for drugs aiming to treat a mismatch between blood flow and the demands of neurons. Whether neurovascular dysfunction is a cause or a symptom, drugs that act on the brain's vascular regulation could offer up treatments for conditions previously thought to be intractable.

A great many drugs already exist that target the vasculature, including analgesics and anti-inflammatories such as aspirin and ibuprofen. Antihypertensives such as angiotensin-converting enzyme (ACE) inhibitors also fall in this category. Even drugs like Viagra affect blood vessels. The bad news is that we do not yet know which of these drugs might have positive or negative influences on the carefully tuned regulation of brain blood flow.

Fortunately, we already have an exceptional tool for studying blood flow in the human brain: fMRI. We can use fMRI to hunt for signs of neurovascular dysfunction by looking for deviations from the normal patterns of responses in different disease states. If we find reliable signatures, fMRI could become a valuable clinical tool for diagnosing and monitoring neurovascular disorders and could help guide us to new treatments.

Our work suggests that we can no longer ignore the brain's vasculature as if it were mundane infrastructure. It is a critical partner in normal brain function. Scrutinizing the brain's vasculature, learning its language, and understanding how it develops, ages and responds to injury could finally bring us closer to untangling the mysteries of the human brain.