A new look at cerebrospinal fluid circulation

Cerebrospinal fluid, the clear liquid that bathes the brain and spinal cord, has a textbook story. The choroid plexus, a gland-like structure tucked inside the brain's ventricles, secretes most of the fluid. That fluid flows in one direction through the ventricular cavities, into the cisterns and subarachnoid space, and drains into the venous blood through structures called arachnoid villi. Harvey Cushing called it the "third circulation" in nineteen twenty-six, and for most of the century that followed, the model felt complete. Brinker and colleagues set out to show why it isn't. The numbers that built the classical picture are real and reproducible. Human cerebrospinal fluid, or CSF, forms at about 0.3 to 0.4 milliliters per minute, with a total volume in adults of ninety to one hundred and fifty milliliters. Historical techniques confirmed these figures repeatedly, including lumbar drainage, ventriculo-cisternal perfusion, and neuroradiological washout studies. Phase-contrast magnetic resonance imaging later measured aqueductal flow at an average of 0.77 milliliters per minute in healthy adults. The classical model absorbed all of this without difficulty because all these measurements seemed consistent with a single directed stream flowing from choroid plexus to arachnoid villi. What that model missed was the nature of the flow itself. Brinker and colleagues describe cerebrospinal fluid movement not as a steady current, but as a pulsatile, to-and-fro motion that reaches throughout the entire brain, driven, beat by beat, by the cardiac cycle. Phase-contrast magnetic resonance imaging synchronized to the heartbeat revealed this directly: aqueductal flow is bidirectional within each cardiac cycle, and in some populations, it reverses entirely. Children under two years showed a mean reversed aqueductal flow of 0.41 milliliters per minute. Patients with normal-pressure hydrocephalus, a condition involving abnormal fluid dynamics, showed mean aqueductal stroke volumes of negative 63.2 microliters per cardiac cycle going in one direction, compared to positive 30.1 microliters per cycle in healthy controls going the other way. But the pulsatile flow is only part of what changes the picture. The bigger shift involves the blood-brain barrier. The classical model treats the barrier as a gate, controlling what enters the brain from the blood. Brinker and colleagues point out that this barrier is also the site of continuous, bidirectional fluid exchange between blood and the brain's interstitial fluid, or ISF — the fluid surrounding brain cells in the tissue itself. That exchange produces flow rates that exceed the choroid plexus's entire cerebrospinal fluid production by a substantial margin. The dominant fluid movement in the brain, in other words, is not the one the textbooks describe. It is happening at the barrier, continuously, in both directions. This is where the molecular biology becomes essential because the mechanisms driving that exchange have now been identified. The main actors are aquaporins, protein channels embedded in cell membranes that allow rapid, selective passage of water, and astrocytes, the glial cells whose end-feet wrap around virtually every capillary in the brain. Two aquaporins matter most here. Aquaporin one sits in the apical membrane of the choroid plexus epithelium, facing the ventricles, and participates in cerebrospinal fluid secretion. Knockout experiments showed that aquaporin one-null mice produced about twenty percent less cerebrospinal fluid, 0.30 versus 0.38 microliters per minute. That is a meaningful reduction, but it confirms that the choroid plexus is not the whole story. Aquaporin four is the more structurally significant player. It concentrates in the end-feet of astrocytes exactly where those cells contact capillaries and the brain surface — at the blood-brain barrier, the glia limitans, and around the Virchow-Robin spaces that accompany blood vessels into brain tissue. Aquaporin four sits at every interface where blood, interstitial fluid, and cerebrospinal fluid meet. The functional consequences of removing aquaporin four are large. Deletion produces a sevenfold reduction in cell membrane water permeability in cultured astrocytes, and a tenfold reduction in blood-brain barrier water permeability in mouse brain. Reducing aquaporin four protein expression by just twenty-seven percent cut the apparent diffusion coefficient, a measure of water mobility in tissue, by fifty percent. When the protein complex that anchors aquaporin four at the perivascular membrane is disrupted, roughly ninety percent of perivascular aquaporin four disappears. These channels are not minor contributors to brain water handling. They are the molecular valves governing the large, continuous exchanges that underlie macroscopic fluid dynamics. Now follow those flows into the tissue itself, along the structures called Virchow-Robin spaces. These are the fluid-filled sleeves that surround blood vessels as they penetrate from the subarachnoid space into brain tissue. For a long time, they were treated as passive anatomical features. Brinker and colleagues describe two active, recently emphasized functions. The first is waste clearance. Tracer studies in rabbits showed that only about five percent of albumin injected into cerebrospinal fluid reached the cervical lymph nodes after five hours, but separate estimates in sheep put the fraction of cranial cerebrospinal fluid volume absorbed by extracranial lymphatics at forty to forty-eight percent. Intravital microscopy during cisternal infusion demonstrated that this movement is respiratory-dependent: during inspiration, particles moved along drainage routes at ten to twenty millimeters per second; during expiration, no movement was observed. These perivascular pathways connect ultimately to cervical lymphatics, offering a route for soluble waste — including proteins like amyloid — to exit the brain. This is the architecture underlying what Nedergaard's group named the glymphatic system: periarterial cerebrospinal fluid entering the tissue from the subarachnoid space, exchanging with interstitial fluid via aquaporin four-mediated water transport at astrocyte end-feet, and draining out through perivenous channels. Two-photon laser scanning microscopy experiments showed fluorescent tracers arriving rapidly in periarterial Virchow-Robin spaces, entering the interstitium, and appearing later around venules. Large tracers stayed confined to the perivascular space while smaller tracers entered the tissue, a size-dependent distribution consistent with a specific pathway rather than simple diffusion. Aquaporin four-deficient mice showed significantly less perivascular and interstitial fluorescence, and reduced clearance of soluble amyloid from the interstitium. The second function of Virchow-Robin spaces is immunological. Blood-borne inflammatory cells are present in these spaces, and fluid movement through the perivascular compartment creates a site for exchange between the brain's immune environment and the systemic immune system. The brain was once considered immunologically isolated. The Virchow-Robin space is part of why that isolation is now understood to be incomplete. Brinker and colleagues are candid about the methodological debates surrounding the glymphatic evidence. The central two-photon studies used a scanning protocol that limits temporal resolution. Critics have raised the possibility that observed perivascular fluorescence partly reflects nonspecific binding of dextran tracers to arterial basement membranes. Sleep versus wakefulness comparisons may be confounded by unmeasured physiological variables like stress. The authors present these as substantive concerns requiring ultrastructural confirmation, higher-frequency imaging, and cross-species replication, not as reasons to dismiss the framework, but reasons to hold it carefully. Sleep is where the implications become most striking. Nedergaard's group showed that glymphatic clearance of interstitial waste is substantially more active during sleep than wakefulness, using in vivo two-photon microscopy and measurements of interstitial volume across awake, sleeping, and anesthetized animals. Aquaporin four-deficient mice showed reduced clearance even under sleep conditions. The suggestion is that sleep may be, in part, the brain running its waste-removal system at full capacity — a function that depends on the pulsatile perivascular flows, astrocytic water channels, and Virchow-Robin space drainage that the classical model did not account for. The clinical stakes are real, if still emerging. Impaired aquaporin four-dependent clearance has been associated with worse outcomes in experimental hydrocephalus. Reduced amyloid clearance in aquaporin four-null mice points toward potential links to neurodegenerative disease. And Brinker and colleagues highlight a therapeutic implication: contrary to the assumption that drugs injected into cerebrospinal fluid are rapidly washed out, perivascular pathways may carry intrathecally administered agents throughout brain tissue — a possible route for drug delivery that the classical model would not have predicted. What Brinker and colleagues ultimately argue is not that the old measurements were wrong. The choroid plexus does produce cerebrospinal fluid. The arachnoid villi do absorb it. Those numbers from the historical perfusion studies and phase-contrast magnetic resonance imaging are real. The revision is about what those measurements describe: one component of a far more dynamic system, dominated by pulsatile bidirectional flows, continuous barrier exchange, astrocyte-mediated water transport, and perivascular clearance pathways. A century of good data was built into a simpler model than the brain actually runs. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.