Core Hypothesis

During sleep, noradrenergic troughs activate a kinase cascade that phosphorylates astrocytic AQP4 at perivascular endfeet, stabilizing its polarized localization and enabling efficient glymphatic convective flow. This phosphorylated AQP4 state is required for the meningeal lymphatic system to recognize and clear senescent astrocytes that express VEGF‑C. Senescent astrocytes, when not cleared, secrete SASP factors that dephosphorylate AQP4, collapsing the glymphatic conduit and creating a feed‑forward loop of impaired waste removal and neuroinflammation.

Mechanistic Rationale

1. Sleep‑linked AQP4 phosphorylation – Phosphoproteomic studies show increased AQP4‑Ser111 phosphorylation during murine non‑REM sleep (see [5]). We propose that reduced norepinephrine signaling during sleep relieves inhibition of CK2 (casein kinase 2), which directly phosphorylates AQP4, boosting its water‑channel efficiency and perivascular anchoring.
2. VEGF‑C as a senescence tag – Senescent astrocytes up‑regulate VEGF‑C (see [3]), which binds VEGFR‑3 on meningeal lymphatic endothelial cells, promoting their chemotactic migration toward perivascular spaces. Phosphorylated AQP4 creates a permissive interstitial pressure gradient that facilitates VEGF‑C‑dependent docking of senescent astrocyte debris onto lymphatic uptake sites.
3. SASP‑mediated AQP4 dephosphorylation – The SASP includes proteases (e.g., MMP‑3) and reactive oxygen species that activate phosphatases (PP2A) targeting AQP4‑Ser111, causing its diffusion away from endfeet (see [4]). This reduces glymphatic inflow, diminishes meningeal lymphatic access, and allows senescent astrocytes to persist.
4. Feedback loop – Persistent senescent astrocytes perpetuate SASP release, further suppressing AQP4 phosphorylation, deepening glymphatic failure, and accelerating amyloid‑β/tau accumulation (see [1], [2], [9]).

Testable Predictions

• Prediction 1: Pharmacological enhancement of CK2 activity during sleep will increase AQP4‑Ser111 phosphorylation, raise CSF influx rates measured by intrathecal gadolinium‑based MRI, and reduce senescent astrocyte burden (p16^INK4a^+ cells) in aged mice.
• Prediction 2: Genetic ablation of VEGFR‑3 in meningeal lymphatic endothelium will block the sleep‑dependent clearance of VEGF‑C‑positive astrocytes despite normal AQP4 phosphorylation, leading to accumulated senescent astrocytes and worsened cognitive performance.
• Prediction 3: Inducing chronic sleep fragmentation will elevate PP2A activity in astrocytes, decrease AQP4‑Ser111 phosphorylation, and correlate with increased meningeal lymphatic permeability to inflammatory mediators (evidenced by heightened CCL2 levels in CSF).

Experimental Approach

1. Mouse models: Use AQP4‑S111A (non‑phosphorylatable) and AQP4‑S111D (phosphomimetic) knock‑in lines, combined with CK2 overexpression via AAV‑GFAP‑CK2. Assess glymphatic flow via two‑photon imaging of CSF‑tracer influx during induced sleep vs. wake.
2. Senescence tracking: Cross p16‑3MR mice to monitor senescent astrocytes by bioluminescence; quantify VEGF‑C expression via immunofluorescence.
3. Lymphatic blockade: Employ VEGFR‑3^fl/fl^; Prox1‑CreERT2 mice to delete VEGFR‑3 specifically in meningeal lymphatics after tamoxifen induction.
4. Readouts: MRI glymphatic index, PET for amyloid‑β/tau, flow cytometry for meningeal lymphocyte subsets, behavioral assays (Morris water maze, novel object recognition).
5. Human relevance: Correlate CSF AQP4 phosphorylation proxies (via exosome‑isolated phospho‑AQP4) with actigraphy‑measured sleep efficiency and p16^INK4a^‑positive extracellular vesicles in older adults enrolled in a sleep‑intervention trial.

Falsifiability

If enhancing AQP4 phosphorylation fails to improve glymphatic clearance or reduce senescent astrocyte load despite verified target engagement, the hypothesis that sleep‑dependent AQP4 phosphorylation is a necessary gate for meningeal lymphatic senescence clearance would be refuted. Conversely, if VEGFR‑3 loss does not impair senescent astrocyte removal despite intact AQP4 polarization, the VEGF‑C‑dependent lymphatic checkpoint model would be invalidated.