Hypothesis

Normal aging involves stable METTL3/14 expression in glutamatergic neurons, but reduced demethylase activity (ALKBH5/FTO) combined with altered IGF2BP reader function creates a specific pattern of transcript hypermethylation. This primarily affects mRNAs encoding negative regulators of autophagy—particularly mTOR pathway components and LC3 inhibitors—as well as proteasomal inhibitors. Meanwhile, proteostasis activator transcripts lose their m6A marks, becoming less stable. The result is a bidirectional proteostasis failure that precedes Alzheimer's disease pathology.

Mechanistic Reasoning

Aged glutamatergic neurons display 75% hypermethylated differentially methylated sites despite unchanged writer expression, pointing to demethylase insufficiency rather than increased methyltransferase activity. I propose that IGF2BP2/3 proteins—m6A readers that stabilize target transcripts—undergo age-dependent substrate switching: they gain function toward autophagy-inhibiting mRNAs (which become hypermethylated) while losing function on proteostasis-promoting transcripts. This mirrors the IGF2BP2-mediated MIS12 stabilization seen in cellular senescence, though with pathological specificity in neurons. The complete loss of m6A-GABRA1 correlation in AD brain supports reader dysfunction as a unifying mechanism.

The downstream consequences appear significant. Ubiquitin-mediated proteolysis genes showing hypomethylation and reduced expression in AD cortex likely represent what happens when proteasome component transcripts lose m6A-mediated stabilization—proteasomal activity drops, aggregated proteins accumulate, and the Unfolded Protein Response activates. Notably, quantitative measurements linking these events in vivo are completely absent.

Testable Predictions

1. Aged mouse glutamatergic neurons will show decreased ALKBH5/FTO expression and activity compared to young neurons.
2. IGF2BP2/3 will exhibit increased binding affinity for mTOR and autophagy suppressor mRNAs (enriched in hypermethylated DMRs) but decreased binding to proteasome subunit mRNAs in aged brain.
3. Autophagy flux assays (mCherry-GFP-LC3) in aged neurons will reveal impaired autolysosome formation that reverses with IGF2BP2 knockdown.
4. Ribosome profiling of aged glutamatergic neurons will show increased translation of autophagy inhibitor transcripts and decreased translation of ubiquitin-proteasome components.
5. Conditional ALKBH5 knockout in glutamatergic neurons will accelerate age-related proteostasis defects and memory decline.

Falsifiability

Several outcomes would falsify this model. If aged neurons show increased autophagy flux despite hypermethylation, or if METTL3/14 knockdown replicates the proteostasis phenotype, the hypothesis would be undermined. The model specifically requires that reader protein substrate specificity, not writer abundance, drives the methylation pattern. Similarly, if demethylase levels remain unchanged in aged neurons, an alternative mechanism must be invoked.