The Convective Flush

Maiken Nedergaard and Glymphatic Interstitial Clearance in Rochester

Fluorescent tracers injected into the cerebrospinal fluid of living mice in 2012 exposed a macroscopic clearance circuit inside the mammalian brain that anatomists had overlooked for more than a century. At the University of Rochester Medical Center, Danish neuroscientist Maiken Nedergaard and her colleague Jeffrey Iliff observed the living cerebral cortex through a two-photon laser scanning microscope. They tracked how the brain disposes of its metabolic waste. Weighing roughly 1,400 grams, the human brain consumes twenty percent of the body's oxygen and glucose while generating grams of misfolded peptide debris—including amyloid-beta and tau—every month. Every other organ relies on a dense mesh of lymphatic capillaries to vacuum up interstitial proteins. The brain parenchyma contains zero lymphatic vessels.

For decades, neurophysiologists assumed that extracellular solutes escaped the brain by slow Brownian diffusion through the narrow, tortuous clefts between neurons and glia. Under passive diffusion, the characteristic transit time ($t_{\text{diff}}$) over distance $x$ scales with the square of distance:

Passive diffusion can clear small ions across a one-millimeter mouse cortex. Across the seventy-millimeter radius of the human cerebrum, however, a large protein such as amyloid-beta ($4.5\text{ kDa}$) requires more than one hundred hours to diffuse to the surface. That timescale is far too slow to prevent toxic plaque aggregation.

Nedergaard's two-photon imaging proved that cerebrospinal fluid (CSF) does not diffuse passively. Instead, it is pumped convectively through annular tunnels surrounding the penetrating blood vessels, known as the Virchow-Robin perivascular spaces. With every heartbeat, the mechanical expansion and recoil of the cerebral arteries acts as a peristaltic pump. That arterial pulse drives CSF from the subarachnoid space deep into the brain along periarterial channels. Encasing these perivascular tunnels is a continuous sleeve of astrocyte glial endfeet whose membranes are tiled with crystalline arrays of aquaporin-4 (AQP4) water channels.

Driven by arterial pulsatility and slow vasomotion waves, CSF jets through the AQP4 channels and sweeps convectively across the extracellular space between neurons. The fluid collects dissolved waste proteins and exits along perivenous spaces into the dural meningeal lymphatics (characterized by Jonathan Kipnis and Kari Alitalo in 2015) and the deep cervical lymph nodes. Nedergaard named this glial-dependent plumbing the glymphatic system. In fluid mechanics, the competition between bulk convective flow (velocity $v$ over characteristic length $L$) and molecular diffusion ($D$) is measured by the dimensionless Péclet number:

When $\text{Pe} \gg 1$, convective bulk flow dominates transport. When Nedergaard genetically deleted the AQP4 gene in mice, convective CSF transport dropped by sixty-five percent and amyloid-beta clearance fell by more than half.

In 2013, Nedergaard and Lulu Xie discovered why every animal species with a complex nervous system must periodically lose consciousness in sleep. Using real-time tetramethylammonium iontophoresis to measure extracellular volume in awake versus sleeping mice, they found that the glymphatic dishwasher is gated by state-dependent brain geometry. During waking consciousness, continuous norepinephrine firing from the locus coeruleus swells astrocyte cell bodies. That cellular swelling compresses the extracellular interstitial space to just 14 percent of brain volume ($\alpha_{\text{awake}} = 0.14$). Hydraulic resistance stays high and convective flow is throttled. When the brain transitions into non-REM slow-wave sleep, locus coeruleus noradrenergic tone drops quiet. Astrocyte volume shrinks, expanding the interstitial channels by sixty percent to $\alpha_{\text{sleep}} = 0.23$. Hydraulic resistance plummets, CSF rushes through the widened parenchyma, and amyloid-beta clearance accelerates twenty-fold.