Hypothesis

Aging‑associated decline of METTL3/14 does not globally erase m6A but rewrites the epitranscriptome by preferentially sparing synaptic transcripts while depriving autophagy‑ and proteasome‑related mRNAs of methylation. This selectivity arises because limited METTL3/14 complexes are competitively sequestered by abundant synaptic‑target RNAs and stress‑induced non‑coding RNAs that act as molecular sponges, leaving autophagy transcripts undermethylated.

Mechanistic rationale

• Synaptic 3′UTR transcripts are enriched in motifs (e.g., DRACH) and bound by RNA‑binding proteins such as YTHDF1 that enhance their affinity for METTL3/14 complexes.
• In aged neurons, upregulated lncRNAs (e.g., NEAT1, MALAT1) or circRNAs contain high‑density DRACH sites and act as decoys, pulling METTL3/14 away from autophagy mRNAs.
• Consequently, m6A on TFEB, ATG7, LAMP2A drops, reducing IGF2BP1‑mediated stabilization and impairing autophagic flux, while synaptic m6A remains high, preserving excitatory signaling.

Testable predictions

1. MeRIP‑seq/miCLIP of FACS‑sorted cortical neurons from 3‑month vs 24‑month mice will show significant loss of m6A peaks on autophagy/proteasome genes (TFEB, ATG7, LAMP2A, PSMB5) but no change or gain on synaptic genes (SYN1, PSD95, GRIN2B).
2. CRISPRa‑mediated METTL14 overexpression in aged primary neurons will restore m6A on autophagy transcripts, increase IGF2BP1 binding, and rescue LC3‑II turnover and p62 degradation without altering synaptic m6A levels.
3. Antisense knockdown of the candidate decoy lncRNA (e.g., NEAT1) in aged neurons will mimic METTL14 overexpression, increasing m6A on autophagy genes and improving survival under ER stress.
4. In 5XFAD mice, AAV‑METTL14 delivery to hippocampus will normalize autophagy‑gene m6A, reduce p62 aggregates, and ameliorate memory deficits; failure to observe these effects would falsify the hypothesis.

Falsifiability

If comprehensive MeRIP‑seq reveals a global, uniform reduction of m6A across all transcript classes in aged neurons, or if METTL14 restoration fails to alter m6A on autophagy genes despite rescuing ER stress, the selective‑substrate‑sequestration model is refuted.