Trithorax against Polycomb at HOXA: the textbook antagonism, and the two ways this disease breaks it
also called trithorax-Polycomb antagonism · menin-EZH2 axis · KMT2A-polycomb balance · bivalent gene activation by menin inhibition
Textbooks say the KMT2A family switches genes on and Polycomb switches them off, and that they fight over the same genes. In this leukaemia that picture fails twice over: Polycomb and the fusion's anchor protein actually work together to keep differentiation genes off, and blocking that anchor turns bivalent genes on by a completely different route.
The same thing, for a scientist
KMT2A is trithorax-group and PRC2 is its canonical antagonist, and the antagonism is real in outline - human PRC2 and MLL1/2-Trithorax complexes share response elements that fall into three distinct classes [E1], and PRC2 is allosterically switched off by the active marks that accompany transcription [E2]. **But in KMT2A-rearranged leukaemia the relationship at the loci that matter is cooperation, not competition, and there are two distinct mechanisms, which this entry keeps apart because merging them is the obvious mistake.** **Route 1, menin builds PRC2's output.** Menin binds the *Ezh2* promoter and promotes its expression; EZH2 then interacts with C/EBPalpha and represses its targets, and menin depletion lowers EZH2, lowers H3K27 methylation at C/EBPalpha targets, and induces pro-differentiation genes [E3]. The index-disease counterpart has the same shape with a B-lineage effector: EZH2 with KMT2A::AFF1 silences *HDAC7* in t(4;11) infant pro-B ALL, and the menin inhibitor MI-538 restores it [E4]. **Route 2, menin inhibition redistributes KMT2A itself.** A whole-genome CRISPR screen on a bivalent reporter identifies MTF2-PRC2.1, PCGF1-PRC1.1 and Menin-KMT2A/B as the bivalency maintainers; menin loss or inhibition **phenocopies polycomb disruption**, and the mechanism is that a **menin-independent** function of KMT2A/B maintains H3K4me3 and opposes polycomb at bivalent genes, so releasing KMT2A from active genes shifts the balance and activates them [E5]. **Both routes end at de-repression after a menin inhibitor and they are not the same claim.** Route 1 says a menin inhibitor is partly a PRC2 inhibitor. Route 2 says a menin inhibitor is partly a PRC2 *potentiator* displaced by relocated KMT2A. The graph must carry both, because a menin-inhibitor pharmacodynamic built on H3K27me3 would be read in opposite directions under them.
Why this entry is not the PRC2 mirror of mechanism-dot1l-prc1-antagonism
This graph holds mechanism-dot1l-prc1-antagonism: DOT1L provides transcriptional memory through PRC1.1 antagonism. The dispatch brief asked for the PRC2 equivalent. It does not exist, and the reason is informative.
- The DOT1L/PRC1 relationship is antagonistic and was measured directly in this disease.
- The PRC2 relationship in this disease is cooperative at specific loci 34 and indirect at bivalent loci 5.
- The only mechanistically established PRC2 antagonism is with the active marks rather than with any KMT2A complex: H3K36me2/3 allosterically inhibits PRC2 and unmodified H3K36 sits in the EZH2-DNA interface 2. H3K4me3 is the other side of the same logic.
- There is no published allosteric or physical relationship between H3K79 methylation and PRC2. The anticorrelation exists in the data - calibrated ICeChIP in MV4;11 finds H3K79me2 depletion at pinometostat-downregulated targets accompanied by paradoxical H3K4me3 gain and H3K27me3 loss (this graph's
mechanism-bivalent-primed-chromatin-fusion-targets) - but no mechanism connects them, and writing one would be inventing it.
So the honest statement is: PRC1 is antagonised by DOT1L; PRC2 is not antagonised by the fusion machinery at all - it is switched off by transcription and switched on by menin, at different genes. The asymmetry between the PRC1 and PRC2 arms is a finding, not a gap.
The two routes, side by side, and what would distinguish them
| Route 1: menin -> EZH2 | Route 2: menin -> KMT2A redistribution | |
|---|---|---|
| source | PMID 23349306 3, PMID 40722046 4 | PMID 36635503 5 |
| system | mouse MLL-AF9; human t(4;11) infant pro-B ALL | cancer cell lines and pluripotent stem cells; wild-type KMT2A/B |
| what menin does | transcriptionally drives Ezh2 | anchors KMT2A/B at active genes |
| effect of menin inhibition on PRC2 | less PRC2 activity at target loci | polycomb-like phenotype from relocated KMT2A, polycomb unchanged |
| predicted H3K27me3 after a menin inhibitor | down at C/EBPalpha targets / HDAC7 | unchanged or context-dependent at bivalent genes |
| predicted H3K4me3 after a menin inhibitor | not predicted | up at bivalent genes, down at active genes |
| genes that come on | myeloid differentiation genes; HDAC7 | bivalent genes including MHC class I |
The distinguishing experiment is cheap and has not been done in this disease: CUT&RUN for H3K27me3, H3K4me3 and EZH2 in an index-fusion line over a menin-inhibitor timecourse, with HDAC7 and a bivalent reporter locus as the two anchors. Route 1 predicts EZH2 occupancy and H3K27me3 fall at HDAC7; Route 2 predicts H3K4me3 rises at bivalent genes with PRC2 occupancy roughly intact. See exp-menin-inhibitor-h3k27me3-timecourse-index-fusion.
A third possibility the graph should hold open: both are true, in different cell types, and the MHC class I result has an immunological consequence - menin inhibition de-repressing MHC class I would make menin inhibitors combinable with immunotherapy, which nobody in this field has proposed.
| textbook claim | trithorax (KMT2A) activates, Polycomb (PRC2) represses, and they compete at HOX loci |
|---|---|
| status in this disease | false as stated; two non-competitive mechanisms replace it |
| not established | any allosteric or physical relationship between H3K79 methylation and PRC2 |
| asymmetry with prc1 | DOT1L antagonises PRC1.1 (mechanism-dot1l-prc1-antagonism); no equivalent DOT1L-PRC2 antagonism is published |
| distinguishing experiment | exp-menin-inhibitor-h3k27me3-timecourse-index-fusion |
routes
[
{
"route": "menin -> EZH2 -> repression of differentiation genes",
"pmids": [
"23349306",
"40722046"
],
"effectors": [
"CEBPA targets (mouse, myeloid)",
"HDAC7 (human, t(4;11) infant pro-B)"
],
"prediction_after_menin_inhibitor": "H3K27me3 and EZH2 occupancy fall at those loci"
},
{
"route": "menin inhibition releases KMT2A/B from active genes, shifting the KMT2A/polycomb balance at bivalent genes",
"pmids": [
"36635503"
],
"effectors": [
"bivalent genes including MHC class I"
],
"prediction_after_menin_inhibitor": "H3K4me3 rises at bivalent genes, falls at active genes; PRC2 occupancy roughly intact"
}
]established prc2 antagonists
[ "H3K36me2/3 (allosteric, PMID 33211010)", "H3K4me3" ]
- If menin inhibition works partly by lowering EZH2 (Route 1), then adding a PRC2 inhibitor to a menin inhibitor is redundant at best; if it works by shifting KMT2A away from active genes (Route 2), the combination could be antagonistic, because Route 2's activation of bivalent genes depends on polycomb still being there to be opposed. The combination therefore cannot be designed until the routes are separated.
- Route 2 predicts that menin inhibition activates bivalent genes generally, not selectively - in normal cells as well as blasts, and in pluripotent stem cells as demonstrated. De-repressing a broad bivalent programme in an infant is a developmental intervention whose consequences are not measurable in a response-rate endpoint.
- Both routes end in de-repression, so clinical de-repression of a marker gene after a menin inhibitor confirms neither mechanism. A pharmacodynamic panel that measures only expression will be over-interpreted.
- Does menin inhibition lower H3K27me3 in an index-fusion system, raise H3K4me3 at bivalent genes, or both? No trial or preclinical study of revumenib, ziftomenib or MI-538 has reported an H3K27me3 measurement, and the two published mechanisms predict different answers.
- In a fusion-driven blast, is the KMT2A that redistributes after menin inhibition the fusion or the untranslocated wild-type allele? PMID 36635503 is about wild-type KMT2A/B; the fusion lacks the SET domain and cannot deposit H3K4me3, so the menin-independent function described there may belong entirely to the wild-type allele.
- Does menin inhibition de-repress MHC class I in KMT2A-rearranged blasts? If the bivalency result transfers, menin inhibitors would increase antigen presentation, which is an immunotherapy-combination rationale nobody in this field has stated and which is cheap to test by flow cytometry.
- Is HOXA actually a site of PRC2/KMT2A competition in this disease, or only in Drosophila and in development? The fly H3K36A/R HOX phenotype is the clearest evidence for competition at HOX and it is not mammalian; no ChIP in a KMT2A-rearranged blast has shown PRC2 at HOXA being displaced.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Human PRC2 and MLL1/2-Trithorax complexes recognise response elements that fall into three distinct classes, establishing that the two complexes engage shared regulatory DNA in mammals rather than only in Drosophila.
Three classes of response elements for human PRC2 and MLL1/2-Trithorax complexes. · Du J, Kirk B, Zeng J, Ma J, Wang Q, et al. · Nucleic Acids Res 2018What this evidence cannot show
Abstract read from the eutils record; full text not read. This is a genomic/computational classification of response elements and is not performed in leukaemia or with a KMT2A fusion. 'Three classes' is a claim about element architecture, not about what happens at any particular locus in a blast. The title and DOI were re-fetched before hand-back after an earlier draft of this beat recorded both incorrectly from recall.
Evidence that argues against this (1)
- Shared response elements would predict competition at those elements, but in this disease the strongest locus-level evidence is cooperation - menin driving Ezh2 expression so that EZH2 represses C/EBPalpha targets. A shared-element model and the observed cooperation are not contradictory, but the element architecture does not by itself predict which way a given locus goes. Thiel AT et al. 2013
- 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; unmodified H3K36 occupies a critical position in the EZH2-DNA interface, H3K36R or H3K36A mutation inhibits H3K27 methylation by PRC2 on nucleosomes in vitro, and Drosophila H3K36A and H3K36R mutants show reduced 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 2020What this evidence cannot show
Abstract read from the eutils record; full text not read. Cryo-EM plus in vitro assays on reconstituted dinucleosomes with a Drosophila in vivo arm; no leukaemia, no fusion, no mammalian haematopoietic system, so the transfer is by mechanism and not by demonstration. The H3K36A/R mutants cannot separate loss of the inhibitory methyl mark from loss of the unmodified lysine's structural role.
Evidence that argues against this (1)
- If active marks excluded PRC2 completely, no fusion-bound locus would carry H3K27me3 - and a subset do. Automated CUT&Tag across diverse KMT2A oncofusions finds a subset of oncofusion-binding sites carrying bivalent H3K4me3/H3K27me3 signatures with single-cell heterogeneity, so exclusion is partial and site-specific. Janssens DH et al. 2021
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin binds the Ezh2 promoter and promotes its expression; EZH2 interacts with C/EBPalpha and represses C/EBPalpha target genes; menin depletion reduces MLL binding at the Ezh2 locus, EZH2 expression, and EZH2 binding and repressive H3K27 methylation at C/EBPalpha targets, thereby inducing pro-differentiation C/EBPalpha targets - so in contrast to its classical role antagonising trithorax function, EZH2 collaborates with trithorax-associated menin to block MLL-AF9 leukaemia cell differentiation.
The trithorax protein partner menin acts in tandem with EZH2 to suppress C/EBPα and differentiation in MLL-AF9 leukemia. · Thiel AT, Feng Z, Pant DK, Chodosh LA, Hua X · Haematologica 2013What this evidence cannot show
Abstract read from the eutils record; full text not read. Mouse MLL-AF9 with acute menin depletion. The effector is C/EBPalpha, a myeloid differentiation factor with no equivalent role in a pro-B blast, so the axis transfers to the index disease only by analogy plus the independent HDAC7 result. The relationship is transcriptional and indirect - no physical menin-PRC2 complex is reported - and whether EZH1 substitutes when menin-driven EZH2 falls is not addressed.
Evidence that argues against this (1)
- A whole-genome CRISPR screen reaches the opposite mechanism for the same intervention: menin loss or inhibition PHENOCOPIES polycomb disruption at bivalent genes, through a menin-INDEPENDENT KMT2A/B function that opposes polycomb and is unleashed when KMT2A is released from active genes. Both routes end in de-repression, so an experiment that only measures gene expression cannot tell them apart. Sparbier CE et al. 2023
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In t(4;11) infant B-ALL, HDAC7 is epigenetically silenced by EZH2 together with the KMT2A::AFF1 fusion protein; the menin inhibitor MI-538 restores HDAC7 expression, enhanced by the class I HDAC inhibitor chidamide, driving leukaemic pro-B cells toward a more differentiated state, impairing aberrant proliferation in an HDAC7-dependent manner, and increasing glucocorticoid sensitivity of PDX cells by repressing RUNX2.
HDAC7 induction combined with standard-of-care chemotherapy provides a therapeutic advantage in t(4;11) infant B-cell acute lymphoblastic leukemia. · de Barrios O, Ocón-Gabarró I, Gusi-Vives M, Collazo O, Meler A, Romecín PA, et al. · Biomark Res 2025What this evidence cannot show
Abstract read from the eutils record; full text not read. The abstract does not state whether EZH2 occupancy or H3K27me3 was measured at HDAC7, so the EZH2 step is named as a conclusion of a multiomics analysis rather than shown as a chromatin measurement. Crucially, the intervention tested is menin inhibition plus an HDAC inhibitor - **no PRC2 inhibitor arm is reported** - so the paper establishes an association between EZH2 and HDAC7 silencing, not a pharmacological dependence.
Evidence that argues against this (1)
- MI-538 restoring HDAC7 is equally consistent with menin acting at HDAC7 directly and EZH2 being a correlate, since displacing the fusion changes the locus whatever PRC2 is doing. Distinguishing them needs an EZH1/EZH2 inhibitor arm, and the only such experiment in a KMT2A::AFF1 system is a PDX growth study that did not measure HDAC7. Ito J et al. 2023
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Coupling a cell-surface readout of bivalent MHC class I gene expression to whole-genome CRISPR-Cas9 screens identifies specific roles for MTF2-PRC2.1, PCGF1-PRC1.1 and Menin-KMT2A/B complexes in maintaining bivalency; genetic loss or pharmacological inhibition of menin phenocopies polycomb disruption and de-represses bivalent genes in cancer cells and pluripotent stem cells; while menin and KMT2A/B contribute to H3K4me3 at active genes, a separate menin-independent function of KMT2A/B maintains H3K4me3 and opposes polycomb-mediated repression at bivalent genes, and release of KMT2A from active genes after menin targeting alters the polycomb/KMT2A balance at bivalent genes to facilitate 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 2023What this evidence cannot show
Abstract read from the eutils record; full text not read. The system is cancer cell lines and pluripotent stem cells with MHC class I as the bivalent reporter - **not a KMT2A-rearranged leukaemia and not the index fusion** - and the KMT2A/B in question is wild-type, not a fusion protein. Whether the same redistribution occurs when the anchored species is a fusion lacking the SET domain is not addressed and is the single most important unknown for transferring this to the index disease. The PRC2.1 attribution rests on MTF2 scoring in a single-gene screen that under-scores its PHF19/PHF1 paralogues.
Evidence that argues against this (1)
- The fusion is not wild-type KMT2A: KMT2A::AFF1 retains the N-terminal menin-binding motifs and CXXC but LOSES the SET domain, so the menin-independent KMT2A/B function that maintains H3K4me3 at bivalent genes in this study may be a property of the wild-type allele only. In a fusion-driven blast the relevant redistribution could be of the untranslocated wild-type KMT2A allele rather than of the fusion, which is a different and testable claim. Janssens DH et al. 2021
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Aberrant enrichment of H3K4me3 in gene bodies at KMT2A oncofusion binding sites is sensitive to menin inhibitors, and a subset of those sites carry bivalent H3K4me3/H3K27me3 signatures with single-cell heterogeneity suggestive of lineage plasticity.
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 2021What 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 cell populations each carrying one mark, which is why the single-cell arm matters and why co-occurrence AT A NUCLEOSOME is not established here. Cell lines and patient samples across diverse oncofusions rather than the index fusion specifically.
Evidence that argues against this (1)
- The better-powered bulk comparison in this disease profiled H3K79me2, H3K4me3, H3K27me3 and H3K36me3 side by side in MLL-AF9 leukaemia stem cells and found abnormal profiles ONLY for H3K79me2 - H3K27me3 and H3K4me3 were not aberrant. Bivalency at fusion sites is therefore a minority-site or context-specific phenomenon rather than a general feature. Bernt KM et al. 2011
THE HYPOTHESIS THIS BEARS ON: the one existing datapoint cuts against it, weakly.
the MHC class I arm of the polycomb-balance shift; same paper, PMID 36635503
the complex the screen names is Menin-KMT2A/B; KMT2B's half of it was never measured on chromatin
two published routes from menin inhibition to de-repression, with opposite predictions for PRC2
- regulates →PRC2 - the complex, not the subunit, is the unit every experiment in this disease actually resolves
menin licenses PRC2's repressive output (Route 1) or is opposed by relocated KMT2A (Route 2)
- analogous to →H3K79 methylation is what keeps Polycomb off - and that, not gene silencing, is what a DOT1L inhibitor actually undoes
the PRC1 arm is antagonistic; the PRC2 arm is not - the asymmetry is the finding
no menin-inhibitor study reports an H3K27me3 measurement, in any system
part of a menin inhibitor's effect runs through wild-type KMT2A/B and the polycomb balance
adds the genetic-arm H3K27me3 result from the paper that entry already cites