Background

The seed idea frames neuronal loss in aging as an active eviction of inefficient cells rather than random damage. Supporting data show that p16INK4a expression rises with age in rodent brain compartments p16INK4a expression rises with age in rodent brain and senescent neurons accumulate in aging human brains senescent neurons accumulate in aging human brains. Yet blood CDKN2A mRNA correlates positively with age in healthy controls but declines in Alzheimer's disease (AD) patients blood CDKN2A mRNA correlates with age in controls but drops in AD, and AD neurons display an unusual positive correlation between exon‑2 CpG methylation and p16INK4a transcription methylation‑expression inversion in AD. These observations suggest that the CDKN2A/B locus is not simply a uniform aging signal but undergoes disease‑specific epigenetic rewiring that could alter its functional output.

Mechanistic proposition

We hypothesize that in normal aging, increased p16INK4a transcription triggers endoplasmic reticulum stress, leading to surface exposure of calreticulin (an “eat‑me” signal) on neurons. Microglial TREM2 recognizes this signal and phagocytoses the marked, metabolically inefficient neurons, thereby preserving circuit efficiency. In AD, aberrant methylation at CDKN2A exon 2 drives expression of a p16INK4a isoform lacking the ER‑stress‑inducing domain. Consequently, neurons fail to display calreticulin, the microglial eat‑me cue is absent, and senescent neurons accumulate despite high p16INK4a transcript levels. This explains the blood‑tissue discordance (peripheral cells retain the canonical isoform, showing age‑related increase) and the epigenetic inversion observed in AD brains.

Testable predictions

• Prediction 1: Neurons from aged wild‑type mice will show higher surface calreticulin levels that correlate with p16INK4a expression; neurons from AD model mice (e.g., APP/PS1) will have elevated p16INK4a mRNA but reduced surface calreticulin.
• Prediction 2: Pharmacological demethylation of CDKN2A exon 2 in AD model neurons will restore the canonical p16INK4a isoform, increase calreticulin exposure, and enhance microglial phagocytosis in co‑culture.
• Prediction 3: Microglial‑specific TREM2 knockout will blunt the age‑dependent clearance of p16INK4a‑positive neurons in wild‑type mice, leading to accumulation of senescent neurons and cognitive decline without altering p16INK4a transcription.
• Prediction 4: In human post‑mortem tissue, AD brains will exhibit a mismatch: high p16INK4a mRNA, low calreticulin signal, and increased microglial activation markers indicative of a frustrated phagocytic state.

Experimental approach

1. Isoform‑specific reporters: Generate knock‑in mice expressing fluorescent tags under the control of the canonical versus exon‑2‑methylated p16INK4a promoters to track isoform expression in situ.
2. Surface biotinylation assays: Isolate neuronal fractions from young, aged, and AD‑model brains; quantify surface calreticulin via streptavidin pull‑down and western blot.
3. Phagocytosis assays: Co‑culture primary microglia with fluorescently labeled neurons; measure neuronal uptake using flow cytometry and confocal microscopy, manipulating methylation with CRISPR‑dCas9‑TET1 or DNMT3A constructs.
4. In vivo clearance: Use two‑photon imaging in Thy1‑YFP mice to monitor neuronal loss over time after inducing p16INK4a expression via AAV, with or without microglial depletion (CSF1R inhibitor).
5. Human validation: Perform multiplex immunofluorescence on AD and control cortical sections for p16INK4a, calreticulin, IBA1, and phosphorylated tau; quantify colocalization and spatial relationships.

Potential implications

If confirmed, this hypothesis reframes CDKN2A/B not as a generic senescence marker but as a regulatory switch that couples neuronal metabolic fitness to microglial surveillance. Therapeutic strategies aiming to restore the canonical p16INK4a isoform or to enhance calreticulin‑dependent eat‑me signaling could re‑engage the brain’s intrinsic eviction mechanism, potentially slowing cognitive decline in AD while avoiding broad senolytic approaches that might impair necessary plasticity.