Oncofusion

start / graph / mark-h3k4me3

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q41-prc2-arm

H3K4me3 - the mark KMT2A's own SET domain writes, that the fusion cannot write, that an interface in this graph already binds, and that switches PRC2 off

also called H3K4me3 · H3K4 trimethylation · histone H3 lysine 4 trimethylation

H3K4me3 is the 'this gene is on' tag sitting at the start of active genes. The healthy KMT2A protein writes it - but the leukaemia fusion has lost the part that does the writing, which is the central paradox of this disease: the enzyme's own product is not what the fusion delivers.

The same thing, for a scientist

H3K4me3 is written by the MLL/SET1 family (MLL1/KMT2A, MLL2/KMT2B, MLL3, MLL4, SET1A, SET1B) through C-terminal SET domains that require the WRAD module - WDR5, RBBP5, ASH2L, DPY30 - for activity on nucleosomes [E1]. **The fusion loses the SET domain.** KMT2A::AFF1 retains the AT-hooks, CXXC and menin-binding motifs and loses PHD1-3, the bromodomain, FYRN, the TAD and SET, so the leukaemic allele cannot deposit this mark. This graph already contains `interface-kmt2a-phd3-h3k4me3` - a KMT2A domain reading H3K4me3 - and until this entry, **the mark that interface binds was not a node.** Three things make it load-bearing for the Polycomb arm. **(i) It switches PRC2 off.** Active-transcription marks allosterically inhibit PRC2; the structural mechanism is solved for H3K36me2/3 and unmodified H3K36 in the EZH2-DNA interface [E2], and H3K4me3 is the other active mark in the same logic. **(ii) It is the readout of the bivalency this disease shows.** A subset of KMT2A oncofusion binding sites carry bivalent H3K4me3/H3K27me3 with single-cell heterogeneity, and **aberrant gene-body H3K4me3 is menin-inhibitor sensitive** [E3]. **(iii) It is the mark that moves when menin is inhibited.** A menin-independent function of KMT2A/B maintains H3K4me3 at bivalent genes and opposes polycomb there, and menin targeting releases KMT2A from active genes, shifting the balance [E4]. Counter to the simple picture, side-by-side profiling of four marks in MLL-AF9 leukaemia stem cells found H3K4me3 **not** aberrant at fusion targets [E5], and the literature is large - `H3K4me3 AND (KMT2A OR MLL) AND leukemia` returns **152** records [E6] - so this entry is a hub rather than a complete treatment.

Go deeper
Why this mark is written here, in the Polycomb beat

The dispatch brief asked this beat to decide about H3K4me3 on the evidence and say why. Written, for four reasons, and none of them is completeness.

  1. The graph had an interface binding a mark that was not a thing. interface-kmt2a-phd3-h3k4me3 existed with no mark-h3k4me3 behind it. That is a dangling forward reference in the ontology, not a small tidiness problem: nothing in the graph could answer 'what is the mark PHD3 binds, and what writes it'.
  2. It is the PRC2 off switch. Every claim in mark-h3k27me3, protein-prc2-complex and mechanism-menin-inhibition-shifts-the-kmt2a-polycomb-balance about PRC2 being excluded from active chromatin runs through the active marks. A Polycomb arm that does not carry its own inhibitors is half an arm.
  3. It is the measured readout of menin inhibition on chromatin. Aberrant gene-body H3K4me3 is menin-inhibitor sensitive 3, and it is the mark the bivalency-balance mechanism moves 4. It is therefore a candidate pharmacodynamic for the drug class this whole project is organised around.
  4. The fusion cannot write it, which makes it the sharpest available statement of what the fusion is not. This graph spends considerable effort on WRAD (gene-wdr5, gene-rbbp5, gene-ash2l, gene-dpy30, concept-rbbp5-what-is-established-for-wild-type-kmt2a-and-what-for-the-fusion) precisely because the WRAD-to-fusion relationship is contested; the mark those subunits exist to produce should be a node in the same graph.

What this entry deliberately does not attempt: the 152-record H3K4me3/KMT2A literature, the broad-H3K4me3-domain literature, the SET1A/SET1B division of labour, or the COMPASS mechanism, all of which belong to a dedicated beat. It records what H3K4me3 is, what writes it, what reads it in this graph, and how it bears on PRC2.

The contradiction the graph must hold: aberrant or not aberrant?

Two well-powered results in this disease disagree about whether H3K4me3 is abnormal at fusion targets, and the disagreement is informative rather than resolvable here.

  • Not aberrant. Side-by-side profiling of H3K79me2, H3K4me3, H3K27me3 and H3K36me3 in haematopoietic progenitors and MLL-AF9 leukaemia stem cells found abnormal profiles only for H3K79me2 5.
  • Aberrant. Automated CUT&Tag across diverse KMT2A oncofusions finds aberrant enrichment of H3K4me3 in gene bodies, sensitive to menin inhibitors 3. And this graph already records that calibrated ICeChIP in MV4;11 shows H3K4me3 increasing at pinometostat-downregulated fusion targets (mechanism-bivalent-primed-chromatin-fusion-targets).

The likely reconciliation is technical and worth stating: the 2011 study is uncalibrated ChIP, which normalises away global changes by construction and compares promoter profiles; the CUT&Tag result is about gene bodies, which is a different genomic compartment, and the ICeChIP result is spike-in calibrated and is measuring a change under drug rather than a baseline difference. 'H3K4me3 is normal at fusion targets' and 'H3K4me3 is aberrantly in gene bodies' are compatible statements about different things. This graph should not resolve them by preferring one.

Facts and figures
markH3K4me3
histoneH3
residueK4, N-terminal tail
writer requirementWRAD module - gene-wdr5, gene-rbbp5, gene-ash2l, gene-dpy30
written by the fusionfalse
fusion noteKMT2A::AFF1 loses the SET domain, so the leukaemic allele cannot deposit H3K4me3; whatever deposits the aberrant gene-body H3K4me3 at fusion sites is not the fusion
relationship to prc2active marks allosterically restrain PRC2; the solved structural case is H3K36me2/3 (PMID 33211010), and the H3K4me3 arm is asserted here only qualitatively
scope notewritten as a hub node for the Polycomb arm; the full H3K4me3/COMPASS literature in this disease is out of scope and is named as a forward reference
writers
[
  "KMT2A/MLL1",
  "KMT2B/MLL2",
  "KMT2C/MLL3",
  "KMT2D/MLL4",
  "SETD1A",
  "SETD1B"
]
readers in this graph
[
  "interface-kmt2a-phd3-h3k4me3"
]
behaviour in this disease
{
  "promoter_profile_mll_af9_lsc": "not aberrant among four marks profiled (PMID 21741597)",
  "gene_body": "aberrantly enriched at oncofusion sites and menin-inhibitor sensitive (PMID 34663924)",
  "under_dot1l_inhibition": "paradoxically INCREASES at pinometostat-downregulated fusion targets (calibrated ICeChIP; see mechanism-bivalent-primed-chromatin-fusion-targets)",
  "under_menin_inhibition": "falls at active genes, rises at bivalent genes (PMID 36635503, outside leukaemia)"
}
literature size
{
  "query": "H3K4me3 AND (KMT2A OR MLL) AND leukemia",
  "records": 152,
  "title_checked": false,
  "date": "2026-09-20"
}
How this could bite us back
  • An intervention that raises H3K4me3 - which menin inhibition does at bivalent genes - would be expected to exclude PRC2 further at those loci. If any of the leukaemia's dependence on PRC2 is at bivalent genes, a menin inhibitor and a PRC2 inhibitor could be acting on the same axis twice, and the combination would be redundant rather than synergistic.
  • H3K4me3 is the product of the normal KMT2A enzyme, which is required for normal haematopoiesis. Any therapeutic reasoning that treats 'lowering H3K4me3' as an objective is proposing to disable the wild-type allele the patient still needs.
Still open
  1. What deposits the aberrant gene-body H3K4me3 at KMT2A oncofusion sites, given the fusion has no SET domain? The candidates are the untranslocated wild-type KMT2A allele, KMT2B, or a SET1-family enzyme travelling with elongating polymerase, and distinguishing them is a degron experiment away.
  2. Is H3K4me3 a usable pharmacodynamic for menin inhibition? PMID 34663924 reports gene-body H3K4me3 is menin-inhibitor sensitive and PMID 36635503 reports it moves in opposite directions at active and bivalent genes - so a bulk measurement would cancel and only a locus-resolved one would report.
  3. Does H3K4me3 inhibit PRC2 in this disease as the general mechanism predicts? The structural case is made for H3K36 and the H3K4 case was not verified with a primary citation in this beat; if it holds, the fusion's ectopic active domains should be H3K27me3-free by construction, which is testable in one calibrated experiment.
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Methyltransferases of the MLL family - MLL1, MLL2, MLL3, MLL4, SET1A and SET1B - implement methylation of histone H3 on lysine 4 through C-terminal SET domains associated with a common set of regulatory factors, WDR5, RBBP5, ASH2L and DPY30; cryo-EM of human MLL1 and MLL3 catalytic modules on nucleosome core particles shows extensive contacts with the histone-fold and DNA regions that give access to the H3 tail, that H2B-conjugated ubiquitin binds directly to RBBP5 to orient the complex on the nucleosome, and that MLL1 and MLL3 differ at the WDR5-RBBP5 interface, accounting for the opposite roles of WDR5 in regulating the two enzymes.

    Structural basis of nucleosome recognition and modification by MLL methyltransferases. · Xue H, Yao T, Cao M, Zhu G, Li Y, Yuan G, et al. · Nature 2019
    PMID 31485071DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. Reconstituted human catalytic modules on defined nucleosome core particles in vitro. No residue numbers or PDB spans are asserted here; per the QUEEN_BRIEF structural rules a span must be derived by aligning the fetched entity sequence to a same-day UniProt sequence, which this beat did not do. The constructs are catalytic modules, not full-length MLL1, and none is a fusion protein.

    Evidence that argues against this (1)
    • This describes wild-type MLL1. The leukaemic allele KMT2A::AFF1 loses the SET domain entirely, so none of this catalytic architecture is present on the fusion - and this graph's own wrad-adjudication arm records that the canonical WRAD-versus-fusion negative is an inference from a deletion mutant rather than a measured fusion co-IP. Nothing about wild-type MLL1 catalysis should be read onto the fusion. Janssens DH et al. 2021
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    At transcriptionally active genes, di- and tri-methylated H3K36 inhibit PRC2 and unmodified H3K36 occupies a critical position in the EZH2-DNA interface, establishing that active-transcription histone marks allosterically restrain H3K27 methylation.

    Structural basis for PRC2 decoding of active histone methylation marks H3K36me2/3. · Finogenova K, Bonnet J, Poepsel S, Schäfer IB, Finkl K, Schmid K, et al. · Elife 2020
    PMID 33211010DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. **This paper establishes the mechanism for H3K36, not for H3K4.** It is cited here to support the general statement that active marks restrain PRC2; the specific H3K4me3-PRC2 inhibition is long-reported in the field but was NOT independently verified with a fetched primary citation in this beat, and no claim in this entry depends on it quantitatively.

    Evidence that argues against this (1)
    • Because the structural work is on H3K36 and not H3K4, the H3K4me3 arm of the 'active marks inhibit PRC2' statement is the weaker half here. A future beat should fetch the primary H3K4me3-PRC2 inhibition work before any entry leans on it, and this caveat is recorded so that does not happen silently. Du J et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Automated CUT&Tag profiling of diverse KMT2A oncofusion binding sites in cell lines and patient samples shows a subset marked by bivalent H3K4me3/H3K27me3 signatures with single-cell heterogeneity suggestive of lineage plasticity, and that aberrant enrichment of H3K4me3 in gene bodies is sensitive to menin inhibitors.

    Automated CUT&Tag profiling of chromatin heterogeneity in mixed-lineage leukemia. · Janssens DH, Meers MP, Wu SJ, Babaeva E, Meshinchi S, Sarthy JF, et al. · Nat Genet 2021
    PMID 34663924DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. 'A subset' is unquantified. Bivalency called from bulk co-occurrence can arise from two populations of cells each carrying one mark; co-occurrence at a single nucleosome is not established by this design. The abstract does not say which menin inhibitor, at what exposure, or whether the H3K4me3 decrease tracks the transcriptional response.

    Evidence that argues against this (1)
    • Aberrant gene-body H3K4me3 falling under a menin inhibitor is compatible with the fusion being displaced from chromatin and the mark disappearing as a side-effect, rather than with menin controlling H3K4me3 deposition. And the fusion itself has no SET domain, so whatever deposits that gene-body H3K4me3 is not the fusion - most plausibly the wild-type KMT2A/B allele or a SET1-family enzyme travelling with elongating polymerase. Sparbier CE et al. 2023
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Menin and KMT2A/B contribute to H3K4me3 at active genes, while a separate menin-independent function of KMT2A/B maintains H3K4me3 and opposes polycomb-mediated repression at bivalent genes; release of KMT2A from active genes following menin targeting alters the balance of polycomb and KMT2A at bivalent genes and facilitates gene activation.

    Targeting Menin disrupts the KMT2A/B and polycomb balance to paradoxically activate bivalent genes. · Sparbier CE, Gillespie A, Gomez J, Kumari N, Motazedian A, Chan KL, et al. · Nat Cell Biol 2023
    PMID 36635503DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. Cancer cell lines and pluripotent stem cells with MHC class I as the bivalent readout - not KMT2A-rearranged leukaemia and not the index fusion. The KMT2A/B in question is wild-type and SET-competent; a fusion cannot perform the H3K4me3-maintaining function described, so transfer to a fusion-driven blast is not automatic and the redistributing species may be the untranslocated allele.

    Evidence that argues against this (1)
    • This mechanism and the menin-EZH2 axis predict opposite things about PRC2 after a menin inhibitor - roughly intact here, reduced there - while both predict de-repression of the reported genes. A pharmacodynamic panel measuring only gene expression cannot distinguish them, which is the reason mechanism-menin-inhibition-shifts-the-kmt2a-polycomb-balance keeps them as two routes. Thiel AT et al. 2013
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Epigenetic profiling of H3K79me2, H3K4me3, H3K27me3 and H3K36me3 in haematopoietic progenitors and MLL-AF9 leukaemia stem cells found abnormal profiles only for H3K79me2 at fusion target loci; H3K4me3 profiles were not aberrant.

    MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L. · Bernt KM, Zhu N, Sinha AU, Vempati S, Faber J, Krivtsov AV, et al. · Cancer Cell 2011
    PMID 21741597DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. 2011-era uncalibrated ChIP, which normalises away global changes by construction, so a uniform shift in H3K4me3 would not have been detected. Mouse MLL-AF9, myeloid. The negative is bounded by the four marks profiled and by the genomic compartment examined.

    Evidence that argues against this (1)
    • A later study using a different method and looking at a different compartment reports aberrant gene-body H3K4me3 at KMT2A oncofusion sites that is menin-inhibitor sensitive. These are compatible rather than contradictory - promoter profile normal, gene-body distribution abnormal - and the graph should not report either as the settled answer. Janssens DH et al. 2021
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    PubMed `H3K4me3 AND (KMT2A OR MLL) AND leukemia` returns 152 records, against 35 for H3K27me3, 16 for H3K36me3 and 7 for H3K36me2 under the same construction - so H3K4me3 is by an order of magnitude the best-studied histone mark in this disease after H3K79 methylation, and this entry is a hub rather than a treatment of that literature.

    What this evidence cannot show

    Raw counts, NOT title-checked - 152 records were not read and this figure is a size estimate rather than a curated literature. The construction 'MLL' matches free text and will include mixed-lineage-leukaemia mentions unrelated to a rearrangement. The counts are recorded to justify the scope decision this entry makes, not to support any biological claim.

    Evidence that argues against this (1)
    • An untitled-checked count is exactly the number this project has repeatedly found to be noise - a 16-record result on recombination signal sequences turned out to be transgenic sunflower and carrot interferon, and 'RD' for NELFE inflated a real count of 2 to 115. This figure should not be quoted anywhere as the size of a real literature without the titles being read. NCBI PubMed 2026
Connected to