Hypothesis

During slow‑wave sleep the glymphatic system does more than flush soluble waste; it actively samples synaptic protein composition and tags under‑performing circuits for removal. This triage depends on activity‑dependent phosphorylation of postsynaptic densities, which marks synapses for microglial phagocytosis when CSF influx reveals low synaptic vesicle turnover. Chronic sleep disruption therefore shifts the balance from circuit refinement to maladaptive retention of weak synapses, accelerating network noise and cognitive decline.

Mechanistic Basis

• CSF‑Synapse Interaction: The perivascular CSF flow that peaks in NREM sleep creates a transient shear stress on astrocytic end‑feet. This mechanical cue activates TRPV4 channels, raising intracellular Ca2+ in astrocytes and triggering local ATP release. ATP hydrolyzes to adenosine, which binds A1 receptors on presynaptic terminals, reducing vesicle release probability proportionally to prior activity.
• Synaptic Tagging: Neurons that fired strongly during wakefulness leave a trace of phosphorylated CaMKIIα at the PSD. This tag stabilizes PSD‑95 and resists dephosphorylation by PP1, protecting the synapse from the adenosine‑mediated weakening signal. Conversely, synapses lacking the tag experience increased PP1 activity, leading to AMPA‑receptor internalization and earmarking for complement‑mediated microglial engulfment.
• Glymphatic Modulation: AQP4 polarization on astrocytes directs CSF preferentially toward cortical layers with high synaptic density. When AQP4 mislocalizes with age, CSF sampling becomes uneven, causing certain regions to receive insufficient synaptic interrogation and others to be over‑cleared.

Predictions & Tests

1. Phosphoproteomic Shift: Sleep‑deprived mice will show a global decrease in PSD‑CaMKIIα phosphorylation and a rise in PP1 activity measured by western blot of synaptosomal fractions. Test: Quantify pCaMKIIα/total CaMKIIα ratio after 6 h wakefulness vs. normal sleep.
2. CSF Synaptic Biomarkers: CSF from sleeping subjects will contain elevated levels of soluble PSD‑95 and neurogranin compared to wakefulness, reflecting active synaptic sampling. Test: Lumbar CSF collection across the sleep cycle followed by ELISA.
3. Region‑Specific Clearance: In aged AQP4‑knockout mice, glymphatic MRI tracer influx will correlate poorly with regional synaptic protein turnover, whereas young wild‑type mice will show tight coupling. Test: Two‑photon imaging of dendritic spine stability alongside intrathecal tracer kinetics.
4. Microglial Engagement: Blocking A1 receptors during sleep should reduce microglial uptake of tagged synapses without affecting bulk amyloid‑β clearance. Test: Pharmacological A1 antagonism paired with CX3CR1‑GFP microglial imaging and synaptosome phagocytosis assays.

Implications

If the glymphatic system performs a synaptic triage, then sleep quality metrics should predict not only waste load but also the fidelity of cortical map preservation. Interventions that enhance adenosinergic signaling or bolster CaMKIIα tagging—such as timed low‑dose caffeine withdrawal or phosphoproteostabilizing peptides—could rescue circuit selectivity in neurodegenerative models where sleep is fragmented. This reframes sleep loss from a simple accumulation toxin problem to a failure of the brain’s nightly editorial board, with direct relevance to aging, schizophrenia, and Alzheimer’s disease where synaptic dysconnectivity precedes overt neurodegeneration.