Hypothesis

Aging hematopoietic stem cells (HSCs) accumulate bivalent domains at tumor‑suppressor enhancers, but simultaneously up‑regulate the H3K4 demethylase KDM5B. This shift tips the balance toward H3K4 loss, collapsing bivalency into a repressive H3K27me3‑only state that silences tumor suppressors and fuels leukemogenesis. Pharmacological inhibition of KDM5B in aged HSCs restores H3K4me3 at these loci, destabilizing the bivalent architecture, generating R‑loops at nearby repetitive elements, and activating the cGAS/STING pathway. The resulting viral‑mimicry response eliminates pre‑leukemic clones, whereas neurons—which do not acquire bivalent domains or KDM5B elevation—remain unaffected.

Mechanistic Rationale

• In cancer, KDM5A/B overexpression removes H3K4me3 from oncogenes and DNA‑repair genes, driving therapy resistance [KDM5A/B overexpression drives oncogenesis and therapy resistance across multiple cancers].
• Aging neurons lose virtually no H3K4me3 but show global H3K27me3 depletion, with 98% of lost H3K27me3 regions lacking bivalency [aging neurons show stable H3K4me3 with only 1 peak lost while H3K27me3 globally depletes; 98% of lost H3K27me3 regions in neurons lack H3K4me3 co-occurrence]
• Aging HSCs gain ~4‑fold more bivalent chromatin at tumor‑suppressor enhancers (RUNX1/2/3, HIF1A) [aging HSCs show ~4‑fold bivalent domain increases at tumor suppressors]
• KDM5 inhibition creates R‑loops that trigger cGAS/STING‑dependent interferon signaling [KDM5 inhibition induces R-loops and activates cGAS/STING signaling]

We propose that age‑linked KDM5B up‑regulation in HSCs converts protective bivalency into a silent state, while its pharmacological reversal re‑instates an active chromatin configuration that is sensed as nucleic‑acid stress, thereby linking epigenetic drift to immune surveillance.

Predictions & Experimental Design

1. Expression assay: Quantify KDM5B mRNA and protein in young vs aged murine HSCs and neurons (flow cytometry, western blot). Expect ↑ KDM5B in aged HSCs, no change in neurons.
2. Chromatin profiling: Perform CUT&RUN for H3K4me3 and H3K27me3 in sorted HSCs before and after treatment with a selective KDM5 inhibitor (e.g., CPI‑455). Predict loss of H3K27me3‑only domains and gain of H3K4me3 at bivalent sites in aged HSCs only.
3. R‑loop detection: Use S9.6 immunofluorescence or DRIP‑seq to measure R‑loops post‑inhibitor. Anticipate ↑ R‑loops at repetitive elements (LINE, SINE) in aged HSCs.
4. cGAS/STING activation: Measure phosphorylated TBK1, IRF3, and IFN‑β mRNA. Expect heightened signaling only in inhibitor‑treated aged HSCs.
5. Functional outcome: Transplant treated aged HSCs into irradiated recipients and monitor leukemia latency. Predict delayed or prevented leukemogenesis compared with vehicle.
6. Neuron control: Repeat steps 2‑5 in cortical neurons; predict no change in bivalency, R‑loops, or cGAS/STING activation.

Falsifiable outcomes: If KDM5B does not increase with age in HSCs, or if KDM5 inhibition fails to produce R‑loops or cGAS/STING activation in aged HSCs, the hypothesis is refuted.

Potential Implications

• Demonstrates that age‑specific epigenetic lesions can be turned into therapeutic liabilities via demethylase modulation.
• Provides a mechanistic bridge between clonal hematopoiesis, inflamm‑aging, and leukemia predisposition.
• Suggests that intermittent KDM5B inhibition could serve as a prophylactic strategy in older adults with high clonal hematopoiesis risk, sparing post‑mitotic tissues such as neurons.