Hypothesis

In Alzheimer’s disease (AD) neurons, increased DNA methylation at the CDKN2A exon 2 recruits the methyl‑CpG‑binding protein MeCP2, which in turn brings histone deacetylase 2 (HDAC2) to the promoter. This MeCP2‑HDAC2 complex overrides the typical activation‑linked methylation seen in peripheral tissues and silences p16INK4a transcription despite higher methylation levels. The neuron‑specific effect is modulated by ANRIL isoforms that preferentially sequester MeCP2 in glia, leaving AD neurons vulnerable to repression.

Mechanistic Rationale

• DNA methylation of CDKN2A exon 2 normally correlates positively with mRNA expression during aging [[https://pubmed.ncbi.nlm.nih.gov/34219731/]].
• In AD brains, CDKN2A expression is reduced while methylation is elevated [[https://pmc.ncbi.nlm.nih.gov/articles/PMC8461666/]], indicating a tissue‑specific inversion of the methylation‑expression relationship.
• MeCP2 binds methylated DNA and can recruit HDAC complexes to repress transcription, a pathway well established in neuronal gene regulation [[https://pubmed.ncbi.nlm.nih.gov/34219731/]] (though not previously linked to CDKN2A).
• ANRIL transcripts have been shown to modulate chromatin states and interferon responses [[https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2018.00405/full]]; certain ANRIL isoforms are enriched in glial cells and can act as molecular decoys for MeCP2.

Thus, the AD‑specific silencing of CDKN2A arises from a neuronal context where methylation serves as a docking site for repressive MeCP2‑HDAC2 rather than an activating signal.

Testable Predictions

1. MeCP2 occupancy at CDKN2A exon 2 will be significantly higher in AD neuronal nuclei compared with age‑matched control neurons, while remaining low in glial cells from the same tissue [[https://pubmed.ncbi.nlm.nih.gov/34219731/]].
2. HDAC2 enrichment will co‑localize with MeCP2 at the CDKN2A promoter in AD neurons, and pharmacological inhibition of HDAC2 (e.g., with vorinostat) will increase p16INK4a mRNA in cultured AD‑derived neurons [[https://pmc.ncbi.nlm.nih.gov/articles/PMC4919535/]].
3. Knockdown of MeCP2 in AD‑patient induced pluripotent stem cell‑derived neurons will restore CDKN2A expression and increase senescence‑associated β‑galactosidase activity, whereas overexpression of MeCP2 in control neurons will repress CDKN2A despite unmethylated promoter [[https://www.oaepublish.com/articles/jca.2022.06]].
4. ANRIL isoform shift: overexpressing a glial‑enriched ANRIL isoform in AD neurons will reduce MeCP2 binding to CDKN2A and rescue p16INK4a expression, while knocking down the same isoform in glia will increase MeCP2 availability and enhance repression [[https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2018.00405/full]].
5. Cross‑species validation: in aged mouse models of AD‑like pathology, neuronal CDKN2A methylation will correlate with MeCP2 binding and reduced p16INK4a, whereas peripheral tissues will show the canonical positive methylation‑expression correlation.

Falsifiability

If any of the above predictions fail consistently—specifically, if MeCP2 does not show increased neuronal binding in AD, if HDAC2 inhibition does not lift CDKN2A repression, or if ANRIL manipulation does not alter MeCP2‑CDKN2A interaction—the hypothesis would be refuted. Conversely, confirmation of these mechanistic links would establish a neuron‑specific methylation‑dependent repression model that explains the CDKN2A expression paradox in Alzheimer’s disease and offers a novel epigenetic therapeutic target.