Hypothesis: Chronic metabolic stress in neurons redirects ANRIL splicing toward a short isoform that preferentially recruits Jmjd3 over PRC2, erasing H3K27me3 and suppressing CDKN2A/B transcription despite aging‑associated DNA methylation loss. This isoform switch explains the paradoxical downregulation of p16INK4a in Alzheimer's disease brains while peripheral blood retains the age‑related increase.

It's plausible that neuronal metabolic stress reshapes ANRIL splicing. The CDKN2A/B locus is regulated by a tug‑of‑war between repressive PRC2 complexes guided by the long non‑coding RNA ANRIL and activating Jmjd3 demethylase that removes H3K27me3 marks【ANRIL recruits PRC2](https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2018.00405/full)】【Jmjd3 antagonizes silencing](https://pmc.ncbi.nlm.nih.gov/articles/PMC4919535/)】. In healthy aging, blood CDKN2A mRNA rises with age, mirroring senescence accumulation【Blood CDKN2A mRNA expression increases with age](https://pmc.ncbi.nlm.nih.gov/articles/PMC8461666/)】, while DNA methylation at the promoter inversely correlates with age and metabolic traits such as ApoE and Lp(a)【DNA methylation inversely correlates with age](https://pmc.ncbi.nlm.nih.gov/articles/PMC6743916/)】. Genetic variants at 9p21.3 link the locus to cancer, atherosclerosis and type 2 diabetes【GWAS associations](https://pmc.ncbi.nlm.nih.gov/articles/PMC3090043/)】, yet its role in late‑onset AD remains unclear.

We propose that neuronal exposure to metabolic stressors—elevated Aβ oligomers, palmitate‑induced lipotoxicity, or altered ApoE lipidation—activates MAPK/ERK signaling that phosphorylates splicing regulators SRSF1 and hnRNPA1. This shifts ANRIL transcription toward an exon‑skipped isoform (ANRIL‑S) lacking the PRC2‑binding hairpin but preserving a conserved Jmjd3‑interacting motif. ANRIL‑S thus serves as a scaffold that recruits Jmjd3 to the CDKN2A/B promoter, accelerating H3K27me3 demethylation. The resulting chromatin becomes more accessible to DNMT3A, which is upregulated in AD‑associated inflammation, leading to de novo CpG methylation that sterically hinders transcription factor binding (E2F1, RB1) and ultimately silences p16INK4a expression despite global hypomethylation trends observed in aging blood.

We don't expect this mechanism to operate in peripheral leukocytes, where inflammatory cues favor the long ANRIL isoform and sustain PRC2‑mediated repression, allowing the age‑related rise in p16INK4a to persist.

To test this, we will: (1) Isoform‑specific qPCR and RNA‑seq on prefrontal cortex from AD versus control donors to quantify ANRIL‑S/ANRIL‑L ratios; (2) Perform RNA immunoprecipitation followed by qPCR for Jmjd3 and EZH2 using antibodies that discriminate ANRIL isoforms; (3) Map H3K27me3 and CpG methylation at CDKN2A/B promoters via ChIP‑seq and bisulfite sequencing in the same samples; (4) Treat primary human iPSC‑derived neurons with Aβ42 or palmitate, measure ANRIL‑S induction, Jmjd3 recruitment, p16INK4a mRNA and protein, and senescence markers; (5) Deploy splice‑switching antisense oligonucleotides targeting the ANRIL‑S splice site to restore the long isoform and assess whether p16INK4a re‑expression rescues neuronal viability under stress.

A falsifiable outcome would be the absence of a significant increase in ANRIL‑S in AD cortex, or a lack of correlated Jmjd3 enrichment and H3K27me3 loss at the CDKN2A/B promoter. Conversely, demonstration that blocking ANRIL‑S splicing restores repressive marks, elevates p16INK4a, and mitigates stress‑induced neuronal death would support the hypothesis. If proven, this mechanism would directly connect metabolic‑epigenetic crosstalk at 9p21.3 to the cell‑cycle deregulation observed in neurodegenerative disease, offering a unifying explanation for why GWAS hits at this locus predispose to metabolic disorders and atherosclerosis while showing inconsistent links to longevity and AD.