Oncofusion

start / graph / mark-h3k36me2-3

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

H3K36me2/3 - the mark that physically switches PRC2 off, that LEDGF reads to reach chromatin, and that a comparator fusion uses to keep EZH2 off HOXA

also called H3K36me2 · H3K36me3 · H3K36 dimethylation · H3K36 trimethylation · histone H3 lysine 36 methylation

H3K36 methylation marks the body of a gene that is being read. It is also a physical off-switch for the Polycomb machine: where this mark is, the repressive machine cannot work. One leukaemia fusion exploits exactly that to keep its target genes switched on.

The same thing, for a scientist

H3K36 di- and tri-methylation marks transcribed chromatin, and it is the only **structurally solved** antagonist of PRC2. Cryo-EM of PRC2 on dinucleosomes shows unmodified H3K36 occupying a critical position in the EZH2-DNA interface; H3K36R or H3K36A mutation inhibits PRC2 methylation of H3K27 in vitro, and *Drosophila* H3K36A/R mutants show reduced H3K27me3 and defective Polycomb repression of *HOX* genes [E1]. **That relationship is not theoretical in this disease family.** NUP98-NSD1 binds elements adjacent to *HoxA7*/*HoxA9*, **maintains H3K36 methylation and histone acetylation there, and thereby prevents EZH2-mediated repression of the Hox-A locus during differentiation**; inactivating NSD1's H3K36 methyltransferase activity abolishes Hox-A activation and progenitor immortalisation [E2]. **This is the trithorax-versus-Polycomb-at-HOXA mechanism the graph was missing, and it turns out to be written in H3K36, not in H3K79 or H3K4.** In the index family the mark is also load-bearing: ASH1L writes H3K36me2, **LEDGF preferentially binds it in vivo** and colocalises with MLL, ASH1L and H3K36me2 genome-wide, ASH1L recruits LEDGF and wild-type MLL to key leukaemia target genes and is required for MLL-dependent transcription and leukaemic transformation - while **KDM2A, an H3K36me2 demethylase and Polycomb-group silencing protein, antagonises MLL-associated leukaemogenesis** [E3]. This graph already contained `interface-psip1-pwwp-h3k36me-nucleosome`, an interface binding a mark that was not a node; and KMT2A::AFF1 spreading into gene bodies is accompanied by **disrupted H3K36me3** [E4]. The literature is small by symbol - `H3K36me2 AND (KMT2A OR MLL) AND leukemia` returns 7 and H3K36me3 returns 16 [E5] - which is why this was the mark most at risk of being declined.

Go deeper
Why this mark was written rather than declined

The dispatch brief asked this beat to decide about H3K36me2/3 on the evidence and say why. The case against writing it was real: the counts are small (7 and 16 records), the mark belongs to the LEDGF/PSIP1 and NUP98 comparative beats as much as to this one, and a thin entry is worse than none.

Three findings during the beat made declining untenable.

  1. It is the only structurally solved PRC2 antagonist. Every statement in mark-h3k27me3, protein-prc2-complex and mechanism-menin-inhibition-shifts-the-kmt2a-polycomb-balance about active chromatin excluding PRC2 rests on 1, and the Polycomb arm cannot cite a mark it has not defined.
  2. A fusion in this graph already uses that antagonism at HOXA, and the graph's own fusion-nup98-nsd1 entry records it correctly - NUP98-NSD1 maintains H3K36 methylation at HoxA7/A9 and thereby prevents EZH2-mediated repression 2. The brief asked for 'a mechanism entry on trithorax-Polycomb antagonism at HOXA in this disease'. The best-evidenced such mechanism in this graph was already present and was written in H3K36. Not writing the mark would have left it unreachable from the Polycomb arm.
  3. The index-family route runs through LEDGF, which is one of this project's central objects. ASH1L writes H3K36me2, LEDGF reads it, and the H3K36me2 demethylase KDM2A antagonises MLL leukaemogenesis 3 - a writer, a reader and an eraser, all in one paper, all in MLL leukaemia.

What is declined instead: mark-h3k27ac. H3K27ac is the reciprocal state on the same residue as H3K27me3 and rises whenever PRC2 is inhibited [see mark-h3k27me3], so it is arguably the next mark the graph needs. It is not written here because this beat did not read a primary H3K27ac dataset in this disease, and writing it from the three PRC2 papers that mention it in passing would be building an entry from other entries' asides. It is recorded as a forward reference in ana-prc2-arm-enumeration.

The two methyl states are different marks and this entry does not fully separate them

H3K36me2 and H3K36me3 have different writers, different genomic distributions and different readers, and an entry that treats them as one thing is doing what RBBP4/7 does in the review literature.

  • H3K36me2 is broad and intergenic, written in mammals largely by the NSD family (NSD1/2/3) and by ASH1L; it is what ASH1L writes and LEDGF reads in MLL leukaemia 3.
  • H3K36me3 is gene-body and co-transcriptional, written by SETD2; it is what is disrupted where KMT2A::AFF1 spreads into gene bodies 4.
  • Both inhibit PRC2 - the structural work tests di- and tri-methyl together 1.

This entry is filed as a single node because the PRC2-inhibitory property, which is why the Polycomb arm needs it, is shared. Where a claim depends on which state, this entry says so, and a dedicated SETD2/NSD beat should split it. The naming follows the existing interface, interface-psip1-pwwp-h3k36me-nucleosome, which is also state-agnostic.

Facts and figures
markH3K36me2 and H3K36me3
histoneH3
residueK36, N-terminal tail
relationship to prc2ALLOSTERIC INHIBITOR - the only structurally solved antagonist of PRC2; unmodified H3K36 is required in the EZH2-DNA interface (PMID 33211010)
role at hoxaNUP98-NSD1 maintains H3K36 methylation adjacent to HoxA7/HoxA9 and thereby prevents EZH2-mediated repression of the Hox-A locus during differentiation (PMID 17589499) - the best-evidenced trithorax-versus-Polycomb-at-HOXA mechanism in this graph, and it is written in H3K36 rather than H3K79 or H3K4
status in kmt2a aff1H3K36me3 patterns are DISRUPTED where the fusion spreads into gene bodies (PMID 19855078, title unverified in this beat) - the opposite arrangement from NUP98-NSD1, and its PRC2 consequence has never been measured
state separation noteme2 and me3 have different writers, distributions and readers; this node is filed jointly because the PRC2-inhibitory property is shared, and any claim depending on the state says so
writers me2
[
  "ASH1L",
  "NSD1",
  "NSD2/MMSET",
  "NSD3"
]
writers me3
[
  "SETD2"
]
erasers
[
  "KDM2A (H3K36me2 demethylase; antagonises MLL leukaemogenesis, PMID 27154821)"
]
readers in this graph
[
  "interface-psip1-pwwp-h3k36me-nucleosome (LEDGF/PSIP1 PWWP)"
]
literature size
{
  "H3K36me2_query": 7,
  "H3K36me3_query": 16,
  "title_checked": false,
  "note": "both decisive papers are INVISIBLE to these queries because they spell the mark out",
  "date": "2026-09-20"
}
How this could bite us back
  • Any intervention that lowers H3K36 methylation - including inhibiting ASH1L, which PMID 27154821 proposes as a target - would be expected to make chromatin MORE permissive to PRC2. In a disease where PRC2 helps the leukaemia repress differentiation genes, that is a mechanism by which an ASH1L inhibitor could deepen the differentiation block rather than relieve it, and nobody has looked.
  • Conversely, a PRC2 inhibitor's effect should depend on the H3K36 methylation state of the target loci, which differs between fusion partners: NUP98-NSD1 maintains H3K36 methylation at HOXA while KMT2A::AFF1 disrupts H3K36me3 where it spreads. A PRC2-directed agent is therefore not expected to behave the same way across fusion families, and no study has stratified by that.
Still open
  1. If KMT2A::AFF1 DISRUPTS H3K36me3 in the bodies of its own targets, does that make those bodies permissive to PRC2? That is the opposite of the NUP98-NSD1 arrangement, it is a direct prediction of the allosteric mechanism, and a single calibrated H3K27me3 CUT&RUN over spreading versus non-spreading targets would test it.
  2. Does ASH1L-deposited H3K36me2 recruit LEDGF to the FUSION, or only to wild-type MLL? PMID 27154821's recruitment claims are about wild-type MLL, and this graph's LEDGF arm already treats the fusion-versus-wild-type distinction as unresolved. The answer decides whether the H3K36me2 route is part of the fusion's targeting or part of the normal machinery it displaces.
  3. KDM2A is an H3K36me2 demethylase that is also a Polycomb-group silencing protein and antagonises MLL leukaemogenesis. It therefore sits at the junction of the two marks this beat cares about, and it has no entry in this graph.
Every source behind this entry (5)
  1. E1established 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; cryo-EM of PRC2 on dinucleosomes shows unmodified H3K36 occupying a critical position in the EZH2-DNA interface, mutation of H3K36 to arginine or alanine inhibits H3K27 methylation by PRC2 on nucleosomes in vitro, and Drosophila H3K36A and H3K36R mutants show reduced levels of H3K27me3 and defective Polycomb repression of HOX genes.

    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. Cryo-EM plus in vitro methyltransferase assays on reconstituted dinucleosomes; the in vivo arm is Drosophila. No leukaemia, no KMT2A fusion, no mammalian haematopoietic system - the transfer is by mechanism, not by demonstration. The H3K36A/R mutants remove the residue entirely and cannot separate loss of the inhibitory methyl mark from loss of the unmodified lysine's structural role in the EZH2-DNA interface, which the same work shows is also required.

    Evidence that argues against this (1)
    • The Drosophila HOX phenotype is in a system where Polycomb response elements are sequence-defined; the mammalian situation required a separate classification of three distinct response-element classes for human PRC2 and MLL1/2-Trithorax complexes. A fly HOX result therefore does not license a statement about HOXA in human leukaemia, and the mammalian HOXA evidence in this entry comes from the NUP98-NSD1 work instead. Du J et al. 2018
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    NUP98-NSD1 induces AML in vivo, sustains myeloid stem-cell self-renewal, and enforces HoxA7, HoxA9, HoxA10 and Meis1 expression; mechanistically it binds genomic elements adjacent to HoxA7 and HoxA9, maintains H3K36 methylation and histone acetylation there, and prevents EZH2-mediated transcriptional repression of the Hox-A locus during differentiation, with deletion of the NUP98 FG-repeat domain, inactivation of NSD1 H3K36 methyltransferase activity, or loss of Hox-A binding each abolishing Hox-A activation and progenitor immortalisation.

    NUP98-NSD1 links H3K36 methylation to Hox-A gene activation and leukaemogenesis. · Wang GG, Cai L, Pasillas MP, Kamps MP · Nat Cell Biol 2007
    PMID 17589499DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. **This is NUP98-NSD1, not a KMT2A fusion** - a comparator system in this graph, myeloid, murine retroviral expression at non-physiological levels. The EZH2-antagonism model is inferred from the marks present and from differentiation-associated repression rather than from a direct measurement of PRC2 recruitment; this graph's own `fusion-nup98-nsd1` entry records that caveat and it is repeated here rather than softened. 2007-era ChIP without spike-in calibration.

    Evidence that argues against this (1)
    • A fusion that keeps EZH2 off HOXA is the OPPOSITE arrangement from the one this beat found for KMT2A fusions, where menin drives EZH2 expression and EZH2 helps the leukaemia repress differentiation genes. Both can be true - they concern different loci and different fusions - but the NUP98-NSD1 model should not be imported into the KMT2A arm as 'the fusion excludes PRC2', because in the KMT2A arm PRC2 is recruited, not excluded. Thiel AT et al. 2013
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The H3K36me2 chromatin mark written by the ASH1L histone methyltransferase is preferentially bound in vivo by LEDGF, which colocalises with MLL, ASH1L and H3K36me2 on chromatin genome-wide; ASH1L facilitates recruitment of LEDGF and wild-type MLL proteins to chromatin at key leukaemia target genes and is a crucial regulator of MLL-dependent transcription and leukaemic transformation, while KDM2A - an H3K36me2 demethylase and Polycomb-group silencing protein - antagonises MLL-associated leukaemogenesis.

    ASH1L Links Histone H3 Lysine 36 Dimethylation to MLL Leukemia. · Zhu L, Li Q, Wong SH, Huang M, Klein BJ, Shen J, et al. · Cancer Discov 2016
    PMID 27154821DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. The abstract's recruitment claims concern **wild-type MLL** proteins; whether ASH1L-deposited H3K36me2 recruits LEDGF to the FUSION is not stated in the abstract and this graph's LEDGF arm treats the fusion-versus-wild-type distinction as unresolved. 'Antagonises MLL-associated leukaemogenesis' for KDM2A is a directional statement whose experimental basis is not specified in the abstract. The system is MLL leukaemia broadly, not the index KMT2A::AFF1 infant disease.

    Evidence that argues against this (1)
    • This graph's LEDGF beat has repeatedly found the PWWP-to-chromatin arm weaker than the model implies - it carries entries titled 'is LEDGF dispensable for fusion chromatin retention' and 'does the fusion need the LEDGF PWWP arm', and PMID 27154821 is already cited in eleven of its entries. The H3K36me2-LEDGF axis should be read through those entries rather than taken at face value from this abstract. Zhu L et al. 2016
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MLL-AF4 binding spreads into the bodies of a subset of target genes together with spreading menin binding, increased transcription, increased H3K79me2/3, disruption of normal H3K36me3 patterns and unmethylated CpG regions in the gene body.

    PMID 28076791DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. Human MLL-AF4 leukaemia cells. 'Disruption of normal H3K36me3 patterns' is a description of the spreading regions and is not a quantified measurement in the abstract, and no PRC2 or H3K27me3 measurement is reported at those regions - so the Polycomb consequence drawn in the counter-evidence below is a prediction, not a finding. NOTE on provenance: an earlier draft of this entry attached this claim to PMID 19855078, which is the infant promoter-methylation paper - the wrong record, taken from a sibling entry's claim text rather than from PubMed. The correct identifier was recovered by re-fetching before hand-back, which is the exact failure mode and the exact remedy the QUEEN_BRIEF describes.

    Evidence that argues against this (1)
    • Disrupted H3K36me3 where the fusion spreads has a direct Polycomb consequence that nobody has followed up: if H3K36me2/3 is what keeps PRC2 off transcribed chromatin, then a fusion that DISRUPTS H3K36me3 in the bodies of its own targets should make those bodies more permissive to PRC2, not less. That is the opposite of the NUP98-NSD1 arrangement and it is testable. Finogenova K et al. 2020
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    PubMed returns 7 records for `H3K36me2 AND (KMT2A OR MLL) AND leukemia` and 16 for the H3K36me3 equivalent, against 35 for H3K27me3 and 152 for H3K4me3 - so H3K36 methylation is the least-studied of the four marks in this disease, which is the reason this entry was at risk of being declined and is written as a hub.

    What this evidence cannot show

    Raw counts, not title-checked. The construction misses the two most important papers for this entry: PMID 17589499 (NUP98-NSD1) says 'H3K36 methylation' rather than H3K36me2 or me3, and PMID 27154821 (ASH1L) is titled 'Histone H3 Lysine 36 Dimethylation' spelled out. **So the counts understate the literature and are recorded as an illustration of that failure, not as its size.** A mark-name count keyed to the me2/me3 notation is exactly the symbol-set problem this project logs for gene aliases.

    Evidence that argues against this (1)
    • Both decisive papers for this entry are invisible to the query that produced this count, because one writes 'H3K36 methylation' and the other writes 'Histone H3 Lysine 36 Dimethylation' in full. A counted absence on a histone mark must enumerate the spelled-out forms as well as the me2/me3 notation, which this beat learned here and did not apply to the H3K4me3 or H3K27me3 counts. Zhu L et al. 2016
Connected to