Hypothesis

We propose that the glymphatic system’s nocturnal triage is not a bulk clearance of debris but an active selection process guided by activity‑dependent ubiquitin tagging of synapses acquired during prior wakefulness. This tagging flags synapses for either preservation or removal, and its fidelity relies on NAD+-sirtuin regulation of dephosphorylation enzymes that reset tags before sleep. When NAD+ declines with age, sirtuin activity drops, tags persist aberrantly, and the glymphatic flow either removes too many tagged synapses or fails to clear detrimental ones, leading to maladaptive plasticity and cognitive decline.

Mechanistic Reasoning

During wakefulness, neuronal activity elevates intracellular Ca2+, activating CaMKII which phosphorylates postsynaptic proteins such as PSD‑95 and GluA1. Phosphorylated residues create a recognition site for E3 ubiquitin ligases (e.g., Ube3A) that attach K48‑linked polyubiquitin chains, marking the synapse for potential glymphatic‑mediated removal. Simultaneously, sleep‑associated NAD+ rise activates SIRT1 and SIRT2, which deacetylate and activate phosphatases (e.g., PP1) that strip the phosphate groups, thereby weakening ubiquitin ligase binding and shielding the synapse from clearance. This creates a bistable switch: high phosphorylation/ubiquitination = earmarked for removal; low phosphorylation = protected.

Age‑related NAD+ depletion reduces sirtuin deacetylase activity, leaving phosphates on synaptic proteins. Consequently, ubiquitin tags persist through sleep, biasing the glymphatic system toward excess removal of potentiated synapses. Conversely, if microglial phagocytosis is impaired (as seen with AQP4 mislocalization), tagged synapses accumulate, fostering hyperexcitability and network noise.

Testable Predictions

1. Phospho‑Ubiquitin Synaptic Signature – In mice, wakefulness will increase K48‑linked ubiquitination of PSD‑95 in cortical synaptosomes; sleep deprivation will prolong this elevation, while normal sleep will reduce it by ~40% within 2 h of sleep onset (quantified by immunoblot).
2. NAD+ Rescue – Administration of NAD+ precursor (nicotinamide riboside, 300 mg/kg) during the dark phase will restore sirtuin activity, accelerate dephosphorylation of PSD‑95, and normalize ubiquitin tag levels despite sleep deprivation, measured by phospho‑specific antibodies.
3. Causal Link to Memory Specificity – Optogenetic inhibition of CaMKII during wakefulness will prevent ubiquitin tagging, resulting in indiscriminate glymphatic clearance (detected by live‑two‑photon imaging of CSF‑ISF tracer flux) and impaired discrimination in a fear‑conditioning task, whereas chemogenetic activation of SIRT1 during sleep will rescue discrimination performance.
4. Human Correlation – PET ligands for synaptic vesicle glycoprotein 2A (SV2A) combined with CSF biomarkers of ubiquitin‑conjugated proteins will show an inverse relationship in older adults: higher ubiquitin‑conjugate burden predicts lower SV2A binding in prefrontal cortex, mediated by reduced NAD+ metabolomics scores.

Falsifiability

If enhancing NAD+ or activating sirtuins fails to modify synaptic ubiquitin tag dynamics or does not improve memory specificity under sleep loss, the hypothesis would be refuted. Likewise, if blocking CaMKII does not alter ubiquitination patterns or glymphatic flux, the activity‑dependent tagging premise would be invalid.

Implications

This framework integrates glymphatic clearance, mitophagy, and epigenetic regulation into a unified sleep‑dependent quality‑control system where the brain actively edits its connectome each night. Interventions targeting the NAD+-sirtuin phosphatase axis could preserve adaptive synapses while permitting removal of truly deleterious circuitry, offering a precise strategy to counteract age‑related cognitive decline.

• Key References: AQP4-mediated glymphatic decline [1]; autophagy/mitophagy and cGAS/STING suppression [2]; mitochondrial NAD+‑sirtuin‑histone axis [3]; NAD+ precursor activation of sirtuins [4]; exercise‑enhanced glymphatic flow and meningeal lymphatics [5].