Background

Blood CDKN2A mRNA rises with age in healthy cohorts (r = 0.407, p = 0.005)[1], yet falls in Alzheimer’s disease (AD) brains where its promoter becomes hypermethylated and paradoxically correlates positively with transcript levels[1]. The locus also produces the antisense lncRNA ANRIL, which can recruit Polycomb complexes to silence CDKN2A/B in cis[2]. In healthy aging, declining Ezh2 and NSD2 activity reduces H3K27me3 and H3K36me2 marks, permitting gradual CDKN2A activation[3][4]. In AD, however, ANRIL may adopt a distinct conformation that favors activator complexes or disrupts repressive looping, leading to a maladaptive epigenetic state.

Hypothesis

We propose that ANRIL functions as a molecular switch whose allele‑specific RNA structure determines whether the CDKN2A/B locus adopts a repressive or active chromatin configuration in a tissue‑dependent manner. In peripheral blood of healthy older individuals, a prevalent ANRIL isoform recruits EZH2‑containing PRC2, maintaining low‑level H3K27me3 that buffers stochastic CDKN2A transcription, resulting in the observed age‑related increase when PRC2 wanes. In neurons susceptible to AD pathology, oxidative stress or amyloid‑β triggers a shift to an ANRIL isoform that binds preferentially to MLL/SET1 histone methyltransferases, depositing H3K4me3 at the promoter. This conversion creates a positive feedback loop where promoter methylation (likely mediated by DNMT3A recruited via the same ANRIL‑MLL complex) stabilizes an open chromatin state, thereby coupling methylation to increased transcription—a reversal of the canonical silencing relationship.

Predictions

1. Allele‑specific RNA‑structure probing (SHAPE‑MaP) of ANRIL from blood versus AD‑affected hippocampal tissue will reveal distinct secondary‑structure signatures correlating with EZH2 versus MLL binding.
2. CRISPR‑based isoform‑specific knockdown of the putative "activating" ANRIL variant in AD‑model neurons will reduce H3K4me3 and DNA methylation at the CDKN2A promoter, lower CDKN2A mRNA, and rescue senescence‑associated phenotypes without affecting PRC2 occupancy.
3. Overexpression of the "repressive" ANRIL isoform in peripheral blood mononuclear cells from young donors will increase EZH2 recruitment, elevate H3K27me3, delay age‑related CDKN2A up‑regulation, and extend proliferative capacity in vitro.
4. Longitudinal profiling of circulating ANRIL isoforms in a cohort of cognitively normal elders will predict subsequent conversion to AD‑like epigenetic signatures (inverse CDKN2A‑methylation correlation) years before clinical diagnosis.

Experimental Approach

• Obtain paired peripheral blood and post‑mortem brain samples from healthy controls, mild cognitive impairment, and AD cases (n≥30 per group). Perform native RNA‑structure mapping on ANRIL, followed by RNA‑immunoprecipitation for EZH2 and MLL.
• Use allele‑specific CRISPR‑Cas13 to knock down each ANRIL isoform in iPSC‑derived neurons and CD34+ hematopoietic stem cells; assay chromatin marks (ChIP‑seq for H3K27me3, H3K4me3), DNA methylation (bisulfite seq), and CDKN2A transcription (RT‑qPCR, RNA‑seq).
• In vivo, deliver AAV‑mediated isoform‑specific ANRIL expression to mouse models of aging and AD‑like pathology; monitor blood CDKN2A levels, brain senescence markers (p16^INK4a immunostaining), and cognitive performance over 12 months.
• Apply machine‑learning to longitudinal plasma ANRIL isoform ratios from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) to test predictive power for incident AD.

If the data show that ANRIL isoform switching dictates the direction of CDKN2A/Epigenetic coupling across tissues, the hypothesis will be supported; converse results would falsify the proposed switch mechanism.

References

[1] https://pmc.ncbi.nlm.nih.gov/articles/PMC8461666/ [2] https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2018.00405/full [3] https://pmc.ncbi.nlm.nih.gov/articles/PMC6743916/ [4] https://liebertpub.com/doi/full/10.1089/rej.2022.0059