At low drug doses the leukaemia dies through FLT3, not through HOXA9 - a second H3K79me2 pathway the dose-response hides
also called non-canonical H3K79me2 pathway · low-dose pinometostat FLT3 pathway · H3K79me2-FLT3-STAT5 axis
Everyone assumed a DOT1L drug works by switching off HOXA9 and MEIS1. But at doses far too low to do that, some KMT2A leukaemias still die - and the reason is that the H3K79 tag is also holding up the growth receptor FLT3. That means the drug has two separate dose ranges with two different mechanisms.
The same thing, for a scientist
Richter et al. found that some MLL-rearranged leukaemias are susceptible to pinometostat concentrations **far below** those that downregulate HOXA9 and MEIS1, and defined the alternative pathways [E1]. By calibrated internal-standard ChIP (**ICeChIP-seq**) H3K79me2 is markedly depleted at pinometostat-downregulated and MLL-fusion target genes, with **paradoxical increases of H3K4me3 and loss of H3K27me3**. Downregulation of Polycomb components accounts for part of the proliferation defect, but **transcriptional downregulation of FLT3 is the major pathway**: loss-of-FLT3-function recapitulates the cytotoxicity and expression consequences of low-dose pinometostat, and overexpression of constitutively active STAT5A - a target of FLT3-ITD signalling - largely rescues them. The pathway also depends on MLL1, leading the authors to propose combinations of DOT1L, MLL1 and FLT3 inhibitors for FLT3-mutant leukaemia. Three things follow for this graph. **The dose-response is not monotone in mechanism**, so any claim about what a DOT1L inhibitor does must state the exposure. **This is the strongest argument that the pinometostat trial was informative in the wrong direction** - a trial that never reached its maximum tolerated dose was operating in the range where the non-canonical pathway dominates. And **it is a ready-made combination rationale** with a drug class already approved in AML.
Why the calibrated assay matters as much as the finding
This is the only paper in this beat that uses calibrated internal-standard ChIP rather than conventional ChIP with a state-specific antibody 1. That matters because every other H3K79 measurement in the field - including the pharmacodynamic endpoint of the clinical trial - is a relative measurement whose absolute scale is unknown, and the states co-occur on the same residue so cross-reactivity is not a small correction.
The paradoxical results are the reason to care: H3K4me3 up and H3K27me3 down at the same loci where H3K79me2 falls. A relative-scale assay reporting 'H3K79 methylation reduced' cannot distinguish a true depletion from a normalisation artefact when other marks are moving in opposite directions, and this graph should treat uncalibrated H3K79 measurements accordingly.
The method is the reason exp-h3k79-state-resolved-icechip-dose-series is specified as ICeChIP rather than as ChIP-seq.
What this does to the pinometostat verdict
It cuts against the simplest reading of the trial in a specific way. If low exposures act through FLT3 and high exposures act through HOXA9/MEIS1, and the trial never reached its maximum tolerated dose, then the trial largely tested the non-canonical pathway - in an unselected population where FLT3 status was not the enrolment criterion.
That is not the same as saying the trial was under-dosed for the canonical mechanism, although it implies it. It is a stronger and more specific claim: the trial may have tested a mechanism that predicts response in FLT3-mutant disease, while enrolling on KMT2A rearrangement. The two complete remissions were in t(11;19)/KMT2A::MLLT1 and their FLT3 status is not reported in the abstract; whether responders were FLT3-mutant is a checkable question that nobody in this graph has checked.
The authors' own proposed combination - DOT1L plus MLL1 plus FLT3 inhibitors - is a prospectively testable version of the same idea, and FLT3 inhibitors are already approved in AML.
| mechanism class | drug mechanism / dose-dependent pathway switching |
|---|---|
| mark | mark-h3k79me2 |
| minor low exposure contributor | downregulation of Polycomb components |
| assay | ICeChIP-seq (calibrated internal-standard ChIP) - the only calibrated H3K79me2 measurement in this beat |
| unchecked fact | FLT3 mutation status of the two pinometostat complete responders |
low exposure pathway
{
"effector": "gene-flt3",
"downstream": "STAT5A",
"rescued_by": "constitutively active STAT5A overexpression",
"also_depends_on": "gene-kmt2a (MLL1)"
}high exposure pathway
{
"effector": "gene-hoxa9 and gene-meis1 downregulation"
}paradoxical observations
[ "H3K4me3 increased at the same loci", "H3K27me3 lost at the same loci" ]
combination proposed by authors
[ "DOT1L inhibitor", "MLL1 inhibitor", "FLT3 inhibitor" ]
- A drug with two mechanisms in two dose ranges will look inconsistent across trials that differ only in exposure, and the resulting failure would be attributed to the target rather than to the design.
- If the achievable exposure acts mainly through FLT3, then combining a DOT1L inhibitor with a FLT3 inhibitor is combining two agents on the same node, which predicts redundancy rather than synergy in FLT3-mutant disease.
- The paradoxical rise in H3K4me3 on DOT1L inhibition means the drug does not simply subtract an activating mark; it redistributes the active-chromatin landscape, and the consequences of that redistribution in normal cells are unmeasured.
- Were the two pinometostat complete responders FLT3-mutant? The trial enrolled on KMT2A rearrangement, and if the active mechanism at achievable exposures is the FLT3 pathway, then FLT3 status is the stratifier that was never applied.
- How prevalent is the low-dose-sensitive phenotype across KMT2A fusion partners and across lineages? The source describes it as 'some MLL-rearranged leukemias' without a denominator.
- Is the relationship between H3K79me2 loss and FLT3 downregulation direct - is FLT3 itself an H3K79me2-marked gene whose body loses the mark - or indirect through the Polycomb shift described in the 2026 account?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Some MLL-rearranged leukaemias are susceptible to pinometostat concentrations far below those that downregulate HOXA9 and MEIS1; by ICeChIP-seq H3K79me2 is markedly depleted at pinometostat-downregulated and MLL-fusion target genes with paradoxical increases of H3K4me3 and loss of H3K27me3; downregulation of Polycomb components accounts for some of the proliferation defect but transcriptional downregulation of FLT3 is the major pathway; loss of FLT3 function recapitulates the cytotoxicity and gene-expression consequences of low-dose pinometostat, overexpression of constitutively active STAT5A largely rescues them, and the pathway depends on MLL1.
Non-canonical H3K79me2-dependent pathways promote the survival of MLL-rearranged leukemia. · Richter WF, Shah RN, Ruthenburg AJ · Elife 2021What this evidence cannot show
Abstract read from the eutils record; full text not read. Cell-line work; the abstract says 'some MLL-rearranged leukemias' without stating how many lines or which fusions, so the prevalence of the low-dose-sensitive phenotype is unknown. STAT5A rescue is an overexpression experiment. The proposed triple combination is a suggestion, not a tested regimen. The FLT3 pathway's dominance is established in the low-dose regime specifically and is not claimed for the canonical regime.
Evidence that argues against this (1)
- A 2026 account places the dominant consequence of H3K79me loss elsewhere: enhanced non-canonical PRC1.1 activity and a genome-wide increase in H2AK119ub arising directly from progressive loss of the mark, independent of fusion displacement or transcriptional repression. Polycomb was the minor pathway in the FLT3 study and is the central mechanism in the newer one. Neville D et al. 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The canonical account of DOT1L dependence in this disease is that Dot1l inactivation downregulates direct MLL-AF9 targets and the MLL-translocation gene signature while leaving global gene expression largely unaffected, with abnormal H3K79me2 profiles only at fusion target loci in leukaemia stem cells.
MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L. · Bernt KM, Zhu N, Sinha AU, et al. · Cancer Cell 2011What this evidence cannot show
Abstract read from the eutils record; full text not read. Genetic Dot1l inactivation in a mouse MLL-AF9 model - a complete loss, therefore by definition a high-dose-equivalent regime, which is exactly why it cannot see the low-exposure pathway. Cited here as the canonical model this entry qualifies, not as a competing result.
Evidence that argues against this (1)
- Because genetic inactivation is a maximal perturbation, the canonical model was derived entirely in the regime where HOXA9/MEIS1 collapse dominates, and could not in principle have detected a distinct low-exposure mechanism. The two results are not in conflict; they describe different parts of the same dose-response and were mistaken for the whole of it. Richter WF et al. 2021
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In the adult phase 1, pinometostat reduced H3K79 methylation and produced 2 complete remissions in 51 patients, both in t(11;19).
The DOT1L inhibitor pinometostat reduces H3K79 methylation and has modest clinical activity in adult acute leukemia. · Stein EM, Garcia-Manero G, Rizzieri DA, et al. · Blood 2018What this evidence cannot show
Abstract read from the eutils record; full text not read. The t(11;19) attribution of both remissions is carried in this graph's existing entries and in the trial report; the FLT3 mutation status of responders is not stated in the abstract, which is the fact this entry would most want. The maximum tolerated dose was never reached.
Evidence that argues against this (1)
- Attributing the trial's exposure range to the non-canonical regime requires knowing the intracellular concentrations achieved relative to the in vitro low-dose window, which nobody has related across the two studies. The inference is plausible and unquantified, and this entry records it as such. Richter WF et al. 2021
H3K79me2 loss downregulates FLT3 at exposures below the HOXA9 threshold
FLT3 downregulation is the major low-dose pathway
- related to →Pinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
two mechanisms in two dose ranges; the trial sat in the lower one
Polycomb was the minor pathway here and the central one in the 2026 account
- related to →HOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it
the canonical pathway this one sits beneath in the dose-response
The locus-level epigenetic picture assembled here is the upstream half of that mechanism: FLT3's hypomethylated, H3K79me2-marked, HOXA9-enhancer-driven state is what low-dose pinometostat dismantles when FLT3 downregulation becomes the major killing pathway.