Bivalent and primed chromatin at fusion target loci — and why the H3K4/H3K27 story is weaker than usually told
also called bivalency in KMT2A-r · H3K4me3/H3K27me3 · primed chromatin · broad H3K4me3 domains
Some genes carry both an 'on' and an 'off' chemical tag at once, poised to go either way. A subset of the fusion's target genes are like this — but the evidence that this poised state is the main story is weaker than the textbooks suggest.
The same thing, for a scientist
Bivalency (co-occurrence of H3K4me3 and H3K27me3) is documented at a subset of KMT2A oncofusion binding sites by single-cell CUT&Tag, where those sites show cell-to-cell heterogeneity suggestive of lineage plasticity. However the better-powered studies point elsewhere: the fusion's distinguishing lesion is broad/extended active-mark domains (H3K4me3 in gene bodies, H3K79me2 spreading) with H3K27me3 that is reduced rather than co-gained. In calibrated ICeChIP-seq, loss of H3K79me2 on DOT1L inhibition was accompanied by paradoxical increases in H3K4me3 and loss of H3K27me3 — i.e. the two marks move in opposite directions, the anti-bivalent pattern. The defensible position is that bivalency is a real minority feature linked to plasticity, not the central mechanism.
What is actually established2 deeper
Broad active domains: established. KMT2A::AFF1 occupies developmental regulatory genes important for haematopoietic stem cell identity and self-renewal, and those regions have grossly altered chromatin structure, with trithorax-group and DOT1 histone modifications extending across large domains 1. This 'breadth' phenotype — active marks spread over unusually long stretches — is the most reproducible chromatin signature of the fusion, and it connects directly to the spreading phenotype in mechanism-fusion-dna-targeting-cxxc-athooks.
Bivalency: real but a subset. Automated CUT&Tag profiling of diverse KMT2A oncofusion binding sites in cell lines and patient samples found that a subset are marked by bivalent (H3K4me3 + H3K27me3) signatures, and single-cell CUT&Tag showed those sites display cell-to-cell heterogeneity suggestive of lineage plasticity. The same study found that aberrant enrichment of H3K4me3 in gene bodies is sensitive to menin inhibitors 2.
That last point is the most useful thing in this node: gene-body H3K4me3 is a menin-inhibitor-responsive chromatin readout, which is exactly the kind of pharmacodynamic marker that HOXA transcripts fail to be (see mechanism-hoxa-high-vs-hoxa-low).
The counter-evidence, which is stronger than the claim
Two independent studies argue against a bivalency-centric model.
Profiling H3K79me2, H3K4me3, H3K27me3 and H3K36me3 side by side in haematopoietic progenitors and MLL-AF9 leukemia stem cells found abnormal profiles only for H3K79me2 at fusion target loci. H3K4me3 and H3K27me3 profiles were not aberrant 3. If bivalency were the mechanism, this is the experiment that should have found it, and it did not.
Quantitative, internally calibrated ICeChIP-seq then measured the marks on the same scale. At pinometostat-downregulated fusion target loci, H3K79me2 was markedly depleted with paradoxical increases of H3K4me3 and loss of H3K27me3 4. The two marks moved in opposite directions. That is the signature of a switch, not of bivalency.
A methodological note worth internalising: most bivalency claims in this literature rest on population-average ChIP, which cannot distinguish a truly bivalent nucleosome from a mixture of H3K4me3-marked and H3K27me3-marked cells or alleles. The CUT&Tag work 2 is better because it is single-cell — and notably, what it found single-cell was heterogeneity, which is exactly the mixture explanation. Read carefully, the best bivalency evidence in KMT2A-r is arguably evidence for cellular heterogeneity rather than for nucleosomal bivalency.
'Primed' is better supported than 'bivalent' — but the priming mark is H3K79me
There is a genuine priming phenomenon at fusion target loci; it just is not the H3K4/H3K27 kind. At HOX loci, H3K79me2 is normally converted to H3K79me1 as haematopoietic cells mature, coinciding with declining HOX expression, and AF10 inactivation reverses the leukemia-associated profile and precludes abnormal HOXA expression 5.
So the loci are pre-marked with H3K79me2 in immature cells, and maturation normally resolves that mark. The fusion prevents resolution; it does not create the mark de novo. That is a priming model with real evidence behind it, and it reframes the therapeutic question from 'remove an aberrant mark' to 'permit a developmental transition that is being blocked' — the same reframing that mechanism-self-renewal-differentiation-block arrives at from the phenotype side.
| chain | mechanism-kmt2a-fusion-causal-chain |
|---|---|
| bivalency status | real at a subset of oncofusion binding sites; not the central mechanism |
| dominant chromatin signature | broad/extended active domains — gene-body H3K4me3 and H3K79me2 spreading — with reduced H3K27me3 |
| h3k4 h3k27 relationship | anti-correlated on DOT1L inhibition (H3K4me3 up, H3K27me3 down), not co-gained |
| priming mark | H3K79me2 pre-existing at HOX loci in immature cells; normally resolved to H3K79me1 on maturation |
| menin inhibitor responsive readout | aberrant gene-body H3K4me3 enrichment |
| method caveat | population ChIP cannot distinguish a bivalent nucleosome from a mixture of cells or alleles |
marks
[ "H3K4me3", "H3K27me3", "H3K79me2/3", "H3K36me3" ]
- If bivalency reflects lineage plasticity, effective therapy could select for the plastic subpopulation and drive lineage switch rather than eradication.
- Marks that move in opposite directions mean single-mark pharmacodynamic assays can mislead: DOT1L inhibition raises H3K4me3 at the very loci it is meant to shut down, which an H3K4me3-only readout would score as activation.
- A bivalency-centric model has driven interest in EZH2 inhibitors in this disease; since H3K27me3 is reduced rather than gained at fusion targets, that rationale is weak and the PRC2 requirement runs through core PRC2 rather than EZH2 (see mechanism-polycomb-prc1-prc2-cbx8).
- Is bivalency at fusion target loci nucleosome-level or a cell-mixture artefact? Sequential/dual-mark single-molecule CUT&Tag would settle a question the field has been asserting for fifteen years without resolving.
- Does bivalency at a subset of sites predict lineage switching under immunotherapy? The single-cell heterogeneity finding makes this the most clinically consequential untested prediction in this node.
- Is gene-body H3K4me3 a usable clinical pharmacodynamic marker of menin-inhibitor target engagement, including in HOXA-low disease? It is menin-inhibitor-sensitive and measurable by low-input CUT&Tag on a marrow aspirate.
- Was the chromatin state at fusion target loci already permissive in the fetal cell of origin before the translocation? Nobody has profiled the pre-leukemic fetal progenitor and the matched leukemia.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
KMT2A::AFF1 occupies developmental regulatory genes important for haematopoietic stem cell identity and self-renewal, and these regions have grossly altered chromatin structure with trithorax-group and DOT1-catalysed histone modifications extending across large domains.
Aberrant chromatin at genes encoding stem cell regulators in human mixed-lineage leukemia · Guenther MG, Lawton LN, Rozovskaia T, et al. (Young RA, senior author) · Genes & Development 22(24):3403-3408 2008What this evidence cannot show
Human ALL cell lines, ChIP-chip era resolution; the abstract makes no bivalency claim and does not address H3K27me3, so this supports breadth of active marks only.
Evidence that argues against this (1)
- "Grossly altered chromatin structure" over large domains at genes the fusion occupies is compatible with the marks being a CONSEQUENCE of sustained transcription rather than a primed substrate the fusion selects. Broad H3K79me2 domains follow elongation, so their presence at occupied loci cannot order the two events. eLife 2021
- E2strong 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 showed that a subset are marked by bivalent H3K4me3/H3K27me3 signatures with single-cell heterogeneity suggestive of lineage plasticity, and that aberrant gene-body H3K4me3 enrichment is sensitive to menin inhibitors.
Automated CUT&Tag profiling of chromatin heterogeneity in mixed-lineage leukemia · Janssens DH, Meers MP, Wu SJ, et al. (Ahmad K, Henikoff S, senior authors) · Nature Genetics 53(11):1586-1596 2021What this evidence cannot show
Bivalency is reported at a subset of sites and is inferred from separate-mark profiling plus single-cell data, not from allele-resolved dual-mark measurement on the same nucleosome; the single-cell heterogeneity finding is itself compatible with a cell-mixture rather than a bivalent-nucleosome interpretation. No test of whether bivalency precedes the fusion.
Evidence that argues against this (1)
- This entry's own E3 is a direct contradiction and should be read as such. 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 at fusion target loci, with H3K4me3 and H3K27me3 profiles not aberrant - i.e. no bivalency signal at the loci that matter. Combined with E4's finding that under DOT1L inhibition H3K4me3 rises while H3K27me3 falls (the two marks moving in opposite directions rather than being co-regulated), the bivalency reading of KMT2A-fusion target chromatin is contested rather than established, and a CUT&Tag subset finding does not settle it. Bernt KM et al. 2011
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Side-by-side profiling of H3K79me2, H3K4me3, H3K27me3 and H3K36me3 in haematopoietic progenitors and MLL-AF9 leukemia stem cells found abnormal profiles only for H3K79me2 at fusion target loci; H3K4me3 and H3K27me3 profiles were not aberrant.
MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L · Bernt KM, Zhu N, Sinha AU, et al. (Armstrong SA, senior author) · Cancer Cell 20(1):66-78 2011What this evidence cannot show
Murine MLL-AF9 AML with array-based profiling; a subset-level bivalency of the kind later reported by single-cell CUT&Tag would plausibly be missed at this resolution and in population averages. Absence of a detected abnormality is not proof of absence.
Evidence that argues against this (1)
- This is the strongest evidence in the entry and it argues against the entry's title. If side-by-side profiling finds H3K79me2 abnormal at fusion targets while H3K4me3 and H3K27me3 profiles are NOT aberrant, then bivalency is not a distinguishing feature of those loci in this model, and the entry needs an account of why its own cited comparison came out negative. eLife 2021
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Internally calibrated ICeChIP-seq in MV4;11 shows that at pinometostat-downregulated KMT2A-fusion target loci, H3K79me2 is markedly depleted with paradoxical increases of H3K4me3 and loss of H3K27me3 — the two marks moving in opposite directions rather than being co-gained.
Non-canonical H3K79me2-dependent pathways promote the survival of MLL-rearranged leukemia · Richter WF, Shah RN, Ruthenburg AJ · eLife 10:e64960 2021What this evidence cannot show
Single cell line (MV4;11, FLT3-ITD-positive); this is a perturbation response to DOT1L inhibition rather than the steady-state chromatin state, so it describes how the marks are coupled dynamically rather than what the untreated locus looks like.
Evidence that argues against this (1)
- The calibrated-ChIP result is one experiment, and this beat cites it in three separate entries as if it were three observations. Internal calibration fixes the scale but not the biology of the spike-in: paradoxical mark movements can arise when the standard's occupancy differs between conditions, and a single cell line under one inhibitor for one duration cannot establish which direction marks move in the disease. It is also the wrong kind of evidence for bivalency: H3K4me3 rising while H3K27me3 falls is the two marks moving APART, which is a description of resolution rather than of a bivalent state. eLife 2021
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
At HOX loci H3K79me2 is normally converted to H3K79me1 as haematopoietic cells mature, coinciding with declining HOX expression; AF10 inactivation reverses leukemia-associated epigenetic profiles and precludes abnormal HOXA expression, supporting a model in which the fusion prevents resolution of a pre-existing mark rather than creating it.
AF10 regulates progressive H3K79 methylation and HOX gene expression in diverse AML subtypes · Deshpande AJ, Deshpande A, Sinha AU, et al. (Armstrong SA, senior author) · Cancer Cell 26(6):896-908 2014What this evidence cannot show
This concerns H3K79 methylation state, not H3K4me3/H3K27me3 bivalency, and should not be cited as bivalency evidence. Mouse models and myeloid fusions; the me2-to-me1 conversion is inferred from population profiling across maturation stages.
Evidence that argues against this (1)
- The H3K79me2-to-me1 conversion model comes from AF10 inactivation in mouse leukaemia, and the clinical test of the pathway it implies was negative: pinometostat reduced H3K79 methylation as intended in 51 adults and produced 2 complete remissions. H3K79me-dependence measured genetically overstates what removing the methylation achieves. Blood 2018
same topic from the mechanism beat; this node flags that canonical H3K4me3/H3K27me3 bivalency in KMT2A-r was NOT verifiable in cycle 1
bivalency at HOXA is the proposed molecular content of the permissive state
the H2Bub layer is part of the machinery for leaving the poised state
the active half of the bivalency claim that entry adjudicates
modifier, not a step - constraint on the substrate the chain runs on