Hypothesis: the menin inhibitor's effect is inactivation in place, not eviction - the fusion keeps most of its sites and loses its output
also called inactivation in place · retention without function · the third reading
There is a third possibility between the fusion being thrown off the DNA and it moving somewhere new: it stays exactly where it was and simply stops working, because the machinery that made it productive has gone. The published data fit this better than they fit either of the other two.
The same thing, for a scientist
Two readings have been argued in this project - the fusion is evicted and degraded (the field's null), or it relocates onto unused unmethylated CpG. **A third reading fits the published data better than either, and nobody has stated it.** Under VTP-50469, MLL1n is not globally displaced and keeps occupancy at HOXA, MYB and GAPDH, while menin is lost globally, DOT1L is reduced but not lost, and H3K79me2 shows no detectable global change [E1]. Transcription of the programme nonetheless collapses within two days, faster than DOT1L inhibition does it [E1]. **Occupancy is largely preserved and output is not**, so the drug's proximate effect on the fusion is a loss of productive function at retained sites, not removal. **A mechanism is already documented for how that could work.** Menin bridges and stabilises fusion spreading into gene bodies, which is the configuration associated with high expression and DOT1L dependence [E2]; remove menin and the promoter-proximal anchor may survive while the spread collapses. Silencing is then actively imposed rather than passively following: menin inhibition induces **PRC1.1-dependent H2AK119ub deposition** at a subset of fusion targets, reported as independent of fusion displacement [E3], and in the same drug state **MLL3/4-UTX floods more than 13,000 new promoters** [E4]. In that picture the fusion is a bystander at its own address - present, unproductive, and being written over. **Prior probability: 0.4**, higher than relocation, because each of its components is separately published and only their conjunction is new.
Why this is not just a restatement of residual occupancy
This graph already holds hyp-apparent-authority-residual-fusion-occupancy, which predicts that a defined subset of sites survives saturating inhibition and that the subset is the resistance substrate. This hypothesis is a stronger and differently-shaped claim:
- the residual is the majority, not a subset - retention is the rule and loss is the exception;
- the interesting quantity is not which sites survive but why survival stops mattering, i.e. what the fusion loses at a site it never left;
- it predicts a decoupling in time: occupancy should be nearly flat while nascent transcription falls, so a time course with both readouts separates it from every other model in one experiment.
It also inherits the residual-occupancy hypothesis's clinical consequence and sharpens it: if the fusion is present everywhere it used to be, resistance needs only to restore a cofactor, not to re-recruit an oncoprotein.
| statement | Menin inhibition leaves the KMT2A fusion bound at the majority of its sites and abolishes its transcriptional output there, by removing menin-dependent spreading and cofactor stabilisation and by allowing active PRC1.1 and MLL3/4-UTX rewriting of the same promoters. |
|---|---|
| origin | data-pattern |
| prior probability | 0.4 |
| null hypothesis | The drug evicts the fusion and transcription falls because the fusion is no longer there; occupancy and output fall together. |
| shown for fusion | partially - the retention observation is in KMT2A::AFDN (ML2) and KMT2A::MLLT3/KMT2A::AFF1 with a shared-epitope antibody; the spreading mechanism is in KMT2A::AFF1 (SEM); the PRC1.1 and UTX arms are murine MLL-AF9 |
predictions
[ "spike-in calibrated fusion occupancy is largely preserved at 24-72 h while nascent transcription at the same loci falls", "the occupancy-to-transcription lag is short and one-directional: transcription falls first or simultaneously, occupancy does not lead it", "fusion-associated SEC and PAFc components are lost from retained sites while the fusion remains", "H2AK119ub and UTX gains occur preferentially at sites where the fusion is retained rather than at sites it vacates", "on drug washout, target-gene transcription recovers faster than fusion re-recruitment would allow, because the fusion never left" ]
experiments
[ "exp-calibrated-early-timecourse-fusion-profiling-under-menin-inhibition", "exp-resistant-versus-sensitive-fusion-fate-men1-interface-mutant" ]
analyses
[ "ana-reanalysis-of-ml2-mll1n-chip-for-gained-peaks-under-vtp-50469" ]
falsifiers
[ "Spike-in calibrated fusion occupancy falls in proportion to transcription at the same loci and timepoints, leaving no retained-but-silent population", "Sequential co-profiling shows the fusion and its SEC and PAFc partners leave retained sites together, so there is no stripped complex", "UTX and H2AK119ub gains occur only at sites the fusion has vacated, making rewriting a consequence of eviction rather than of inactivation in place", "Washout recovery of target-gene transcription is slow and matches the kinetics of fusion resynthesis, implying the fusion was gone" ]
- If the fusion is retained and inactive, response depends on the cell's own silencing machinery staying intact - which is exactly what PRC1.1 loss breaks, and PRC1.1 loss is already a documented resistance route. The drug would then be only as good as the host epigenome.
- Retention with preserved protein means no lag for resistance: restoring one cofactor at sites the fusion never left can reactivate the programme in a single cell division.
- This hypothesis predicts that deeper menin inhibition buys nothing once displacement is complete, which if true means dose escalation in children is adding toxicity without mechanism.
- It also predicts that intermittent dosing is a bad idea, because recovery requires no re-recruitment step.
- Is the retained fusion still in a complex? A co-profiling or proximity experiment at retained sites under drug would say whether what remains is a complex missing menin or a stripped N-terminal fragment on uCpG.
- Does retained occupancy predict relapse? If the residual fraction varies between patients, it is a candidate biomarker that can be measured on a single CUT&RUN from a marrow aspirate.
- How does this interact with the reciprocal allele and with KMT2A*, the 5'-truncated internal-promoter product? An N-terminal antibody sees the fusion and the reciprocal product differently, and neither has been separated under drug.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Under 3 days of VTP-50469 in three human KMT2A-rearranged lines, MLL1n was not globally displaced and retained occupancy at HOXA, MYB and GAPDH; menin was lost globally; DOT1L was reduced but not lost; H3K79me2 showed no detectable global change; and yet target-gene transcription was clearly suppressed by day 2, faster than with DOT1L inhibition.
A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia. · Krivtsov AV, Evans K, Gadrey JY, et al. · Cancer Cell 2019What this evidence cannot show
The occupancy and expression readouts are at different timepoints (3 days and 2-7 days) and in overlapping but not identical experiments, so the decoupling is assembled across panels rather than measured in one time course. TSS-anchored quantification without spike-in. Adult-derived lines, two of three myeloid.
Evidence that argues against this (1)
- An alternative account of the same panels is that the retained MLL1n signal is mostly wild-type MLL1 in MOLM13 and RS4;11, with the fusion genuinely evicted; the ML2 line is the control against that, but it is a single unreplicated sample of a different fusion partner. Krivtsov AV et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin co-stabilises fusion binding and is required for spreading of MLL-AF4 into gene bodies, the configuration associated with the highest expression and the greatest DOT1L dependence, and menin and ENL show identical spreading patterns to the fusion.
MLL-AF4 Spreading Identifies Binding Sites that Are Distinct from Super-Enhancers and that Govern Sensitivity to DOT1L Inhibition in Leukemia. · Kerry J, Godfrey L, Repapi E, et al. · Cell Rep 2017What this evidence cannot show
SEM cells only; spreading occurs at 149 of about 4,400 sites (3.4% of target isoforms), so the mechanism proposed here applies directly to a small minority of sites and is extrapolated to the rest. Knockdown rather than acute depletion.
Evidence that argues against this (1)
- If spreading affects only 3.4% of sites, collapse of spreading cannot by itself explain a programme-wide transcriptional collapse, and something else must account for the loss of output at promoter-restricted sites. Krivtsov AV et al. 2019
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin inhibition induces PRC1.1-dependent deposition of H2AK119ub to silence a subset of MLL-fusion targets, and the authors report this as independent of MLL-fusion displacement - an active silencing step that does not require the fusion to leave.
DOT1L provides transcriptional memory through PRC1.1 antagonism. · Neville D, Ferguson DT, Heikamp EB, et al. · Nat Cell Biol 2026What this evidence cannot show
The primary subject of that work is DOT1L-mediated transcriptional memory; the menin-inhibitor arm is a comparison. Independence from fusion displacement is the authors' statement, and how displacement was assayed in that arm was not read in full for this entry.
Evidence that argues against this (1)
- If PRC1.1 silencing is what turns the programme off, then fusion occupancy is not the pharmacodynamically relevant variable at all, and both this hypothesis and the relocation hypothesis are arguing about a quantity that does not determine response. Zhou X et al. 2024
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In the same drug state, MLL3/4-UTX gains more than 13,000 promoter-proximal binding sites and H3K27ac, H4K16ac and MOF increase where it binds - so the chromatin at fusion target promoters is actively rewritten under menin inhibition rather than merely vacated.
A Molecular Switch between Mammalian MLL Complexes Dictates Response to Menin-MLL Inhibition. · Soto-Feliciano YM, Sánchez-Rivera FJ, Perner F, et al. · Cancer Discov 2023What this evidence cannot show
Murine MLL-AF9 with MI-503; the gained UTX sites are not shown to be the same sites the fusion retains, and that intersection is the test this hypothesis actually needs.
Evidence that argues against this (1)
- UTX gain is presented by its authors as a consequence of the MLL1-menin complex being displaced from promoters, which presupposes displacement - the opposite of retention. Whether UTX gains at sites the fusion retains or at sites it vacates is the unresolved point and would discriminate the two. Krivtsov AV et al. 2019
a chromatin-bound MLL1-N complex retaining 83% of its share at menin-co-occupied sites while menin falls 5-fold is the quantitative form of this hypothesis in the index fusion
the quantitative version of that hypothesis in the human index fusion, with the discrepancy measured at 4.00-fold
the quantitative form of exactly this hypothesis: what stays has lost menin, lost DOT1L and lost a third of its H3K79me2
supplies the residual occupancy fraction that the inactivation-in-place reading needs
co-measures occupancy and nascent transcription, which is what separates this reading
the active silencing step this reading needs is already published
the UTX arrival that rewrites the same promoters
a stronger form of the same intuition: the residual is the majority, not a subset
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
proposed for the index fusion; assembled from data in KMT2A::MLLT3, AFF1 and AFDN backgrounds