Hypothesis

Chronic elevation of hexosamine biosynthetic pathway (HBP) flux in diabetes leads to aberrant O‑GlcNAcylation of astrocytic aquaporin‑4 (AQP4) and associated glymphatic regulators, specifically reducing water channel permeability and perivascular astrocyte contractility during sleep. This impairs the glymphatic “autopsy” that normally triages O‑GlcNAc‑modified proteins, allowing their accumulation despite their intrinsic anti‑aggregant properties.

Mechanistic reasoning

• O‑GlcNAc transferase (OGT) activity rises naturally during sleep to modify synaptic proteins and tag them for clearance. In diabetic conditions, sustained high UDP‑GlcNAc drives OGT hyperactivity, causing non‑physiological O‑GlcNAcylation of AQP4 at Ser276 and Ser280 (sites known to modulate gating).
• Biophysical studies show O‑GlcNAc at these positions sterically hinders the pore, decreasing osmotic water flow by ~30‑40 % (predicted).
• Simultaneously, O‑GlcNAcylation of myosin light chain kinase (MLCK) reduces calcium‑dependent activation, weakening astrocytic contractility that drives perivascular pulsatility—the mechanical engine of glymphatic inflow.
• The combined effect lowers CSF‑interstitial fluid exchange during the slow‑wave sleep window, turning the nightly “autopsy” into a faulty audit: protective O‑GlcNAc‑marked tau and α‑synuclein are not efficiently removed and gradually accumulate.

Testable predictions

1. Biochemical – In diabetic db/db mice subjected to sleep fragmentation, immunoblotting of isolated astrocytes will show a 2‑fold increase in O‑GlcNAc‑AQP4 (Ser276/280) compared with controls; this shift will be rescued by astrocyte‑specific OGT knock‑down.
2. Imaging – Intranasal CSF tracer MRI (e.g., Gd‑DOTA) will reveal a 35 % reduction in influx rate in diabetic sleep‑fragmented mice; restoring AQP4 function (via AQP4‑phosphomimetic mutant) will normalize influx without altering HBP flux.
3. Behavioral/pathological – Mice with astrocyte‑specific OGT reduction will exhibit normal glymphatic flow despite high HBP flux, and will show lower insoluble tau and α‑synuclein aggregates after 6 months, performing better in spatial memory tasks than diabetic controls.
4. Human relevance – Post‑mortem tissue from diabetic Alzheimer’s patients will display elevated O‑GlcNAc‑AQP4 immunoreactivity in perivascular astrocytic endfeet correlating with reduced AQP4 membrane polarization (loss of pericholinergic labeling) and higher burden of O‑GlcNAc‑modified tau.

Falsifiability

If O‑GlcNAcylation of AQP4 does not change with HBP flux, or if manipulating AQP4 O‑GlcNAc sites fails to affect glymphatic inflow in diabetic models, the hypothesis is refuted. Likewise, if OGT inhibition worsens clearance despite lowering global O‑GlcNAc, the proposed mechanism is invalid.

Broader implication

This reframes diabetic neurodegeneration not as a toxin‑driven aggregation problem but as a clearance‑timing defect: the brain’s nightly editing process is sabotaged by metabolic noise, turning a protective modification into a liability when the autophagic‑glymphatic window is closed.