Oncofusion

start / graph / hyp-kmt2a-aff1-alters-h3-3-deposition-at-fusion-targets

Hypothesisspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q37-histone-chaperones

Does KMT2A::AFF1 change histone turnover at its target genes, or does it read a chromatin state it did not make?

also called fusion writes the clock vs fusion reads the clock · H3.3 turnover at HOXA under a KMT2A fusion

A KMT2A fusion forces genes to be read continuously, and reading a gene knocks its histones out. So the fusion may be speeding up histone replacement at the genes it controls - or it may just be sitting on genes that were already turning over fast. The two possibilities predict different things about how quickly a drug's effect should appear and disappear.

The same thing, for a scientist

**Statement.** KMT2A::AFF1 increases replication-independent nucleosome turnover at its occupied target genes, so H3.3 deposition and H3K79 methylation valency at those loci are consequences of fusion-driven elongation rather than pre-existing properties of the chromatin. **The rival.** The fusion is recruited to loci that were already transcribed and already high-turnover (CpG islands, active gene bodies, `gene-psip1`-marked H3K36me chromatin), and it alters the mark without altering the turnover. **Why the answer matters, concretely.** `mechanism-h3k79-turnover-and-dilution` and `exp-h3k79-state-resolved-icechip-dose-series` build a dosing argument on the clearance clock. If the fusion **sets** the turnover rate, then the clock is a property of the oncogene and the same drug schedule applies to every fusion-driven locus. If the fusion **inherits** it, the clock is a property of the cell state, and it will differ between a cycling blast and `cell-state-quiescence` / `cell-state-relapse-initiating-persister` by an amount nobody has measured. **Prior probability: 0.35** that the fusion measurably increases turnover at its targets relative to matched loci. Low because the evidence for it is entirely mechanistic reasoning, and because `mechanism-nucleosome-turnover-clearance-clock-machinery` records that the only in-vivo chaperone datum that bears on it - a histone half-life of ~2 weeks in a postmitotic heart - points to turnover being substantial everywhere rather than fusion-specific.

Go deeper
What the graph already contains that bears on this

For the hypothesis:

  • Forced elongation evicts histones; H3.3 is deposited at transcribed gene bodies (gene-hira E1, protein-histone-h3-3 E1).
  • H3K79 valency tracks turnover rate genome-wide and turnover explains the asymmetric distribution of H3K79me toward the transcriptional unit (mechanism-nucleosome-turnover-clearance-clock-machinery E2).
  • H3.3K79me1 is higher than H3.1/H3.2K79me1 at an equal methylation rate (protein-histone-h3-3 E3), which is what a transcription-coupled variant accumulating the mark would look like.

Against the hypothesis:

  • Graded weakening of the DOT1L-AF9 interaction changes H3K79me2/3 at target genes without any reported change in nucleosome turnover (mechanism-nucleosome-turnover-clearance-clock-machinery E2 counter-evidence, PMID 25921540) - enzyme tethering alone is sufficient to move the mark.
  • No chaperone in this beat is a selective dependency of KMT2A-rearranged lines once lineage and pan-essentiality are accounted for (ana-histone-chaperone-dependency-lineage-split). If the fusion were driving turnover harder than other leukaemias, a turnover-machinery dependency would be the expected signature and it is absent - indeed FACT and the H2A.Z machinery are less required.
  • "nucleosome turnover"[tiab] AND (MLL OR KMT2A) returns 0 records: nobody has looked, so the hypothesis has neither support nor refutation in this disease.
Facts and figures
statementKMT2A::AFF1 increases replication-independent nucleosome turnover at its occupied target genes, so H3.3 deposition and H3K79 valency there are consequences of fusion-driven elongation rather than inherited properties of the chromatin.
originliterature-gap
prior probability0.35
predictions
[
  "Newly deposited H3.3 accumulates faster at fusion-occupied loci than at expression-matched, non-occupied loci in the same cell.",
  "Acute degradation of the fusion reduces H3.3 deposition rate at those loci within hours, before any change in cell cycle.",
  "H3K79me2 clears faster at high-turnover fusion targets than at low-turnover ones under a fixed dose of DOT1L inhibitor.",
  "In a quiescent KMT2A-rearranged cell, H3K79me2 clearance is slower than in a cycling one but is NOT absent, because replication-independent exchange continues.",
  "A DOT1L degrader clears H3K79me2 more slowly than a SAM-competitive inhibitor at long transcribed genes, if DOT1L's yeast chaperone activity transfers to human."
]
experiments
[
  "exp-h3-3-turnover-at-fusion-targets-under-dot1l-inhibition",
  "exp-does-kmt2a-aff1-bind-set-taf1",
  "exp-h3k79-state-resolved-icechip-dose-series"
]
analyses
[
  "ana-histone-chaperone-dependency-lineage-split"
]
falsifiers
[
  "Newly deposited H3.3 accumulates at the same rate at fusion-occupied loci as at expression-matched non-occupied loci. The hypothesis is dead: the fusion reads the state rather than making it.",
  "Acute fusion degradation changes H3K79me2 without changing H3.3 deposition rate at the same loci within the same window. Recruitment, not turnover, sets the mark.",
  "H3K79me2 clearance rate under DOT1L inhibition shows no relationship to measured nucleosome turnover rate across loci. The clearance-clock framing itself is wrong, not just this version of it.",
  "Quiescent KMT2A-rearranged cells show no measurable histone exchange over two weeks. The original quiescence bite-back is right and this entry's qualification of it is wrong."
]
How this could bite us back
  • If the fusion sets the turnover rate, then removing the fusion - which is what a menin inhibitor partly does - would SLOW clearance of H3K79 methylation at the same loci, so combining a menin inhibitor with a DOT1L inhibitor could be antagonistic on this axis even while being synergistic on transcription.
  • If the fusion does not set the turnover rate, then the clearance clock is a cell-state property and every dosing argument in this graph becomes a statement about the patient's proliferative fraction rather than about the drug.
  • Either answer makes the quiescent relapse-initiating compartment the hardest to treat, and neither gives a way to speed clearance up - only ways to slow it down.
Still open
  1. What is the right expression-matched control locus set for a KMT2A-fusion ChIP comparison? Without it neither arm of this hypothesis is testable, and no published KMT2A-fusion dataset supplies one.
  2. Can turnover be measured at locus resolution in a primary infant KMT2A::AFF1 blast, or only in cell lines? The cell lines are the ones this graph already distrusts for this disease.
  3. If the fusion does set the clock, does a menin inhibitor reset it - and on what timescale relative to its transcriptional effect?
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Nucleosome turnover rates determine the valency of H3K79 methylation, turnover simulations predict propagation of the mark across the genome, and turnover accounts for the asymmetric distribution of H3K79me toward the transcriptional unit.

    Nucleosome Turnover Regulates Histone Methylation Patterns over the Genome. · Chory EJ, Calarco JP, Hathaway NA, et al. · Molecular Cell 2019
    PMID 30472189DOI
    What this evidence cannot show

    Abstract read from efetch 2026-09-20 (identifiers re-fetched for this entry, not copied from the sibling entries that also carry this paper); full text not read. Chemical-induced proximity is an artificial recruitment regime and the propagation conclusions rest on simulation. Not a leukaemic system and no KMT2A fusion - this is the mechanistic premise of the hypothesis, not evidence for it in this disease.

    Evidence that argues against this (1)
    • Enzyme tethering alone moves the mark: graded weakening of the DOT1L-AF9 interaction changes H3K79 di- and tri-methylation at target genes and transformation potential, with no turnover change invoked. If recruitment sets the mark, the fusion need not touch turnover at all. Kuntimaddi A et al. 2015
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    SPT6, the histone chaperone that binds RNA polymerase II through elongation, is required for RNAPII processivity and for termination, and its acute removal causes readthrough transcription at thousands of genes - establishing that the chaperone layer and the elongation the fusion drives are mechanically coupled in human cells.

    Targeted protein degradation reveals a direct role of SPT6 in RNAPII elongation and termination. · Narain A, Bhandare P, Adhikari B, et al. · Molecular Cell 2021
    PMID 34233157DOI
    What this evidence cannot show

    Abstract read from efetch 2026-09-20; full text not read (PMC8354102 exists). Human cell line with an engineered degron; not leukaemic and no KMT2A fusion. Establishes coupling between elongation and a chaperone, which is the mechanistic premise; it says nothing about whether a fusion changes the coupling.

    Evidence that argues against this (1)
    • Coupling is a property of transcription everywhere, so demonstrating it does not distinguish a fusion-driven gene from any other highly transcribed gene - which is precisely the rival hypothesis. A control of matched-expression non-fusion-bound loci is required and no published dataset provides it. Oncofusion knowledge graph (internal analysis) 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    As of 2026-09-20 PubMed returns 0 records for '"nucleosome turnover"[tiab] AND (MLL OR KMT2A)', 0 for '(H3F3A OR H3F3B OR H3.3) AND (MLL OR KMT2A) AND leukemia' after title check, and 0 for 'HIRA AND (MLL OR KMT2A)' including the aliases TUPLE1 and DGCR1. The question this hypothesis asks has not been addressed in the published literature of this disease.

    What this evidence cannot show

    All query translations captured; none contained a [UID] term. **Bounded by its queries, and this beat demonstrated that bound failing**: the generic phrase 'histone chaperone' retrieved an MLL paper (PMID 15670842) that no gene-symbol query in the beat returned. Multi-word [tiab] terms were not run in both singular and plural in every case. A study that measured turnover using a method name rather than the phrase - CATCH-IT, SNAP-tag pulse-chase, calibrated ChIP of nascent histone - would not be caught.

    Evidence that argues against this (1)
    • A zero across several queries is still a zero across several queries. The correct next step before treating this as a real gap is a method-name sweep and a supplementary-table sweep, neither of which this beat ran. Shimoyama T et al. 2005
Connected to