Hypothesis

ANRIL isoform switching in neurons rewires the relationship between DNA methylation and CDKN2A/B transcription, converting methylation from a repressive signal into an activating one in Alzheimer's disease.

Mechanistic Model

In healthy aging, linear ANRIL isoforms recruit Polycomb repressive complex 2 (PRC2) and DNMT1 to the CDKN2A/B promoter, establishing a methylation‑dependent silencing loop that correlates with rising p16INK4a expression as cells enter senescence. Neuronal activity‑dependent splicing shifts the balance toward circular ANRIL (circANRIL) species that lack the PRC2‑binding domain but retain a motif capable of binding DNMT3A and the transcription factor YY1. This circANRIL‑DNMT3A‑YY1 complex methylates adjacent CpG sites while simultaneously stabilizing YY1‑mediated transcriptional initiation, producing a positive correlation between methylation and CDKN2A expression. The resulting anomalous methylation‑expression coupling suppresses the canonical senescence‑associated rise of p16INK4a, contributing to neuronal vulnerability in Alzheimer's pathology.

Predictions & Experimental Design

1. Isoform quantification – qRT‑PCR with isoform‑specific primers will show a significant increase in circANRIL/linear ANRIL ratio in post‑mortem Alzheimer's prefrontal cortex compared with age‑matched controls (prediction: >2‑fold increase). Falsifiable: if ratios are unchanged, the model fails.
2. Methylation‑expression coupling – Bisulfite sequencing of individual CpG sites paired with RNA‑FISH for p16INK4a will reveal a positive slope only when circANRIL is overexpressed in primary human neurons (prediction: slope >0). Falsifiable: overexpression of circANRIL does not alter the methylation‑expression slope.
3. Complex formation – RNA immunoprecipitation followed by mass spectrometry will detect DNMT3A and YY1 in circANRIL pull‑downs but not in linear ANRIL pull‑downs (prediction: enrichment >5‑fold). Falsifiable: no specific enrichment.
4. Functional rescue – Antisense oligonucleotides targeting the back‑splice junction of circANRIL will normalize methylation‑expression correlation and restore p16INK4a up‑regulation under oxidative stress (prediction: p16INK4a expression increases ≥1.5‑fold). Falsifiable: no change in p16INK4a levels.

Potential Implications

If validated, this mechanism explains why Alzheimer's brains show low CDKN2A expression despite epigenetic marks that typically activate the locus, and it identifies the ANRIL isoform ratio as a therapeutic entry point. Modulating circular ANRIL levels could re‑establish proper senescence signaling, potentially mitigating neuroinflammatory cascades linked to aging‑related dementia.