Hypothesis

The global, non‑selective loss of H3K27me3 observed in aged neurons stems from constitutive low catalytic activity of KDM6A/B demethylases driven by a declining α‑ketoglutarate/succinate ratio, whereas in cancer cells oncogenic signaling pathways (e.g., mTORC1‑4E‑BP‑EZH2 axis) locally phosphorylate and activate KDM6A/B, producing focused H3K27me3 removal at bivalent promoters of metastasis‑related genes.

Mechanistic basis

1. Co‑factor dependence – KDM6A/B are Fe(II)/α‑KG‑dependent dioxygenases; their activity falls when intracellular succinate accumulates or α‑KG declines. Aging neurons exhibit mitochondrial dysfunction and elevated succinate, predicting a basal reduction in demethylase velocity across the genome.
2. Signal‑dependent phosphorylation – Recent work shows mTORC1 enhances EZH2 translation via 4E‑BP, but mTORC1 also sustains S6K1‑mediated phosphorylation of KDM6B at Ser^xxx (hypothetical site). This modification increases KDM6B’s Vmax without altering protein levels, creating a permissive state for rapid demethylation only where mTORC1 is hyperactive, such as at oncogene promoters.
3. Chromatin accessibility feedback – H3K27me3 loss facilitates binding of pioneer factors (e.g., FOXM1) that further recruit mTORC1 complexes, reinforcing a local loop that sharpens the demethylation signal in cancer but is absent in aged tissue where chromatin remains relatively closed.

Testable predictions

• Prediction 1: Nuclear extracts from middle‑aged mouse neurons will show ~30‑40 % lower KDM6A/B demethylase activity (measured by succinate‑sensitive HPLC assay) compared with young adults, while total KDM6A/B protein levels remain unchanged.
• Prediction 2: Pharmacological elevation of α‑KG (cell‑permeable dimethyl‑α‑KG) in aged neurons will rescue H3K27me3 levels at developmental gene promoters without affecting global H3K4me3.
• Prediction 3: In glioblastoma stem cells, CRISPR‑mediated mutation of the putative S6K1 phosphorylation site on KDM6B will diminish H3K27me3 loss at oncogenic promoters (PAK5, MMP9) and reduce invasive capacity, whereas EZH2 inhibition will produce a similar phenotype.
• Prediction 4: Proximity ligation assays will detect increased KDM6B‑S6K1 interaction in HER2+ breast cancer cells but not in senescent fibroblasts, correlating with local H3K27me3 depletion at ER‑specific bivalent loci.

Experimental outline

1. Enzymology – Isolate nuclei from young (3 mo), middle‑aged (12 mo), and old (24 mo) mouse cortex; perform in‑vitro demethylation assays with recombinant nucleosomes and varying α‑KG/succinate ratios; quantify product formation by mass spectrometry.
2. Metabolite rescue – Treat aged neuronal cultures with dimethyl‑α‑KG (5 mM) for 48 h; perform ChIP‑seq for H3K27me3 and assess restoration at non‑bivalent developmental genes.
3. Phospho‑mutant cancer models – Generate KDM6B S→A mutants in U87‑MG and MDA‑453 lines; conduct ChIP‑seq for H3K27me3 and RNA‑seq to evaluate changes at oncogenic bivalent promoters; assay migration/invasion.
4. Signaling blockade – Apply rapamycin (mTORC1 inhibitor) to cancer cells; monitor KDM6B phosphorylation (phospho‑specific antibody) and H3K27me3 levels at PAK5 promoter via CUT&RUN.

Falsifiability

If any of the following observations occur, the hypothesis is refuted:

• KDM6A/B catalytic activity in aged neurons is equal to or greater than that in young neurons despite metabolite measurements.
• Elevating α‑KG fails to alter H3K27me3 distribution in aged tissue.
• Phospho‑site mutation of KDM6B does not affect promoter‑specific H3K27me3 loss or cancer phenotypes.
• mTORC1 inhibition does not reduce KDM6B phosphorylation or alter focal H3K27me3 dynamics in cancer cells.

By linking metabolic state, signal‑dependent enzyme modification, and chromatin context, this model provides a concrete, testable explanation for why H3K27me3 erosion is genome‑wide and non‑selective in aging yet tightly targeted to oncogenic pathways in cancer.