Hypothesis

In aging neurons, p16INK4a expression rises not to kill the cell but to flag synapses that consume ATP without contributing to network output. High p16INK4a tags these weak connections for microglial phagocytosis, preserving overall circuit efficiency under declining energy budgets.

Mechanistic Model

1. Energy sensor – p16INK4a promoter contains CpG sites whose methylation state reflects local ATP/AMP ratios (see perinatal methylation effects [3]). Low energy → hypomethylation → increased transcription.
2. Activity read‑out – Calcium‑dependent kinases phosphorylate p16INK4a, altering its affinity for the ubiquitin ligase SCF^β‑TrCP. Phosphorylated p16INK4a accumulates at postsynaptic densities of low‑firing synapses.
3. Tagging for clearance – p16INK4a binds the extracellular protein MFG‑E8, bridging the synapse to microglial MerTK receptors, triggering complement‑mediated phagocytosis.
4. Feedback – Removal of inefficient synapses reduces maladaptive excitatory drive, lowering oxidative stress and slowing further p16INK4a induction.

Testable Predictions

• Prediction 1: In aged mouse cortex, synapses with low miniature excitatory postsynaptic current (mEPSC) frequency will show higher p16INK4a immunoreactivity than high‑activity synapses.
• Prediction 2: Neuron‑specific knockout of p16INK4a will preserve low‑activity synapses, leading to increased energetic burden (measured by 2‑DG uptake) and accelerated cognitive decline in aged mice.
• Prediction 3: Pharmacological inhibition of the p16INK4a–MFG‑E8 interaction (using a peptide blocker) will reduce synapse loss without altering neuronal p16INK4a levels, rescuing network efficiency in aged animals.
• Prediction 4: In human post‑mortem tissue, cortical areas with Alzheimer’s pathology will show a dissociation: neuronal p16INK4a mRNA is low (as reported [2]), but p16INK4a protein is enriched at dendritic spines of low‑activity synapses, detectable by proximity ligation assay.

Experimental Approach

1. Synapse‑specific p16INK4a mapping – Combine transgenic mice expressing a p16INK4a‑HA tag under the endogenous promoter with array tomography or expansion microscopy to quantify p16INK4a at individual synapses correlated with electrophysiological read‑outs from patch‑clamp recordings in acute slices.
2. Loss‑ and gain‑of‑function – Use Cre‑dependent CRISPRi to suppress p16INK4a in excitatory neurons (Camk2a‑Cre) and assess synapse density, mitochondrial ATP levels, and behavior (Morris water maze) across 6, 12, 18‑month ages.
3. Intervention – Administer a cell‑permeable peptide mimicking the p16INK4a MFG‑E8 binding domain to aged wild‑type mice; measure microglial phagocytic activity (pHrodo‑labeled synaptosomes) and cognitive performance.
4. Human validation – Apply multiplexed immunofluorescence and RNAscope on donated cortex (control, AD, non‑AD demented) to correlate p16INK4a mRNA, protein, synaptic markers (Synapsin‑1, PSD‑95), and activity‑dependent phospho‑EBN.

Potential Confounds and Controls

• Off‑target effects of global p16INK4a loss – restrict manipulations to forebrain excitatory neurons and verify that glial p16INK4a remains unchanged.
• Compensatory upregulation of p21 – monitor Cdkn1a expression to ensure phenotypes are not masked.
• Age‑related blood‑brain barrier changes – include vascular marker staining to rule out peripheral immune cell infiltration contributing to synapse loss.

Implications

If p16INK4a functions as a selective synapse‑pruning factor, interventions that modulate its activity—rather than bluntly suppressing senescence—could preserve cognitive function while allowing the brain to continue its energy‑saving remodeling. This reframes the CDKN2A/B locus from a marker of irreversible damage to a tunable rheostat for synaptic economy.

References

• [1] https://pmc.ncbi.nlm.nih.gov/articles/PMC2987737/
• [2] https://pmc.ncbi.nlm.nih.gov/articles/PMC8461666/
• [3] https://pmc.ncbi.nlm.nih.gov/articles/PMC5528139/
• [4] https://diabetesjournals.org/diabetes/article/67/5/872/39919/CDKN2A-B-T2D-Genome-Wide-Association-Study-Risk