Hypothesis

KDM5B-mediated removal of H3K4me3 at bivalent promoters primes KDM6A demethylase access, and age‑related decline in KDM5B activity locks H3K27me3 at tumor‑suppressor enhancers in hematopoietic stem cells, promoting leukemic priming.

Mechanistic rationale

In aged HSCs, ChIP‑seq shows a 4‑fold increase in bivalent H3K4me3/H3K27me3 domains at RUNX1/2/3 and HIF1A enhancers [1]. Neurons, by contrast, retain stable H3K4me3 while losing H3K27me3 globally [2]. This tissue‑specific divergence suggests that the enzyme setting the H3K4me3 landscape governs KDM6A accessibility. We propose that KDM5B, the predominant H3K4me2/3 demethylase in hematopoietic progenitors, normally erases H3K4me3 from nucleosomes flanking bivalent sites, creating a low‑affinity platform for the KDM6A/UTX complex. When KDM5B activity wanes—due to reduced NAD+‑dependent SIRT1 deacetylation or oxidative‑stress‑mediated phosphorylation—H3K4me3 persists, sterically hindering KDM6A binding and allowing PRC2 to maintain or spread H3K27me3. Consequently, tumor‑suppressor enhancers become locked in a repressive bivalent state, mirroring the cancer‑associated H3K27me3 gain at SLFN11 [4] and predisposing to leukemic transformation.

Testable predictions

1. Genetic – HSC‑specific knock‑down of Kdm5b in young mice will recapitulate the aged HSC bivalent profile (↑H3K4me3/H3K27me3 at RUNX1 enhancers) and increase clonal leukemic outburst upon serial transplantation.
2. Pharmacological – Treatment of aged HSCs with a SIRT1 activator (e.g., NAD+ booster) will restore KDM5B demethylase activity, reduce H3K4me3 at bivalent promoters, and decrease H3K27me3 occupancy as measured by CUT&RUN.
3. Biochemical – Recombinant KDM5B will show increased H3K4me3 demethylation only when pre‑incubated with SIRT1‑deacetylated KDM5B; acetyl‑mimetic mutants will fail to stimulate KDM6A nucleosome binding in vitro.
4. Epigenetic rescue – Overexpressing a catalytically active KDM5B transgene in aged HSCs will erase age‑acquired bivalent marks and suppress leukemia formation in a xenograft model, whereas a demethylase‑dead KDM5B will not.

Falsification

If KDM5B loss does not alter H3K4me3 levels at the specified enhancers, or if restoring KDM5B activity fails to diminish H3K27me3 despite verified enzyme activity, the hypothesis is refuted. Conversely, observing the predicted changes supports a model where KDM5B gates KDM6A function and links metabolic state to epigenetic aging in hematopoietic tissue.

Note: Citations correspond to: [1] PMID PMC7534803 [2] PMID PMC11353134 [4] PMID med.szu.edu.cn...