Oncofusion

start / graph / mechanism-q136-menin-displacement-is-drug-class-diagnostic-even-though-the-mll1-channel-is-not

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q136-menin-vs-dot1l-inhibitors

Resolving cycle 83's threat: the MLL1 channel moves under both drug classes and is therefore not diagnostic, but the MEN1 channel separates them completely - so 'menin displacement' remains a specific readout of menin-interface engagement and the corpus must stop treating the two channels as one

also called the channel distinction · cycle 83 adjudication

A worry had been hanging over this project for months: if a drug that does not touch menin moves the same signal as a drug that does, then the whole menin idea might be non-specific. It turns out the worry applies to one particular measurement, not to menin itself - menin's own signal separates the two drugs perfectly.

The same thing, for a scientist

**The threat, precisely as stated.** At MLL1's own top sites in murine MLL-AF9, the MLL1-N share ratio falls to **0.3012 under SGC0946** (a DOT1L inhibitor with no described action at the menin-KMT2A interface) and **0.3298 under VTP-50469** [E1]. A related measurement on the FLAG-tagged fusion gives the DOT1L inhibitor **about 70%** of the menin inhibitor's effect at the fusion's own top bins [E2]. The inference drawn was that *neither drug's effect on this channel is diagnostic of the mechanism it was chosen to test*. **Three channels are being conflated across those entries and they behave differently.** | channel | what the antibody sees | behaviour under a menin inhibitor | behaviour under a DOT1L inhibitor | |---|---|---|---| | **MEN1** | menin itself | **cleared from 19/19 canonical target peaks, p = 0.0003** [E3] | **retained at 14/19, 14/19, 8/19** [E3] | | **DOT1L** | the enzyme | cleared from 5/5 target peaks [E3] | retained at 5/5 [E3] | | **MLL1-N / tagged fusion** | wild-type MLL1 AND the fusion | falls to ~0.33 / ~0.18 [E1][E2] | falls to ~0.30 / ~0.30 [E1][E2] | **So the non-specificity is a property of the MLL1 channel, not of the drug class.** Two readings of why are available and this project cannot yet choose between them: 1. **The shared-epitope problem.** An MLL1-N antibody reports wild-type MLL1 and the fusion together. Both drug classes perturb the wild-type MLL1/menin/WRAD machinery - menin inhibitors directly, DOT1L inhibitors through the transcriptional collapse of the programme those complexes sit on - so a mixed channel can move under both without either drug acting on the same species. 2. **Genuine two-way dependence.** Fusion residency may be partly downstream of H3K79 methylation, which is what [E2] concluded. If so the MLL1 channel is reporting real biology and it is simply not a *specificity* assay. **Either way the operational conclusion is the same and it is the point of this entry: menin displacement at pre-specified target loci discriminates the drug classes and the MLL1 channel does not.** A pharmacodynamic claim, a mechanism claim or a falsifier phrased on 'the MLL1 channel' cannot be transferred to 'the menin channel', and this corpus has done that transfer at least once. **The independent prediction this confirms.** Krivtsov et al. report that VTP-50469 causes **global loss of menin binding** while MLL1 is not globally displaced and specifically **retains occupancy at HOXA, MYB and GAPDH** [E4]. This analysis finds menin **absent** at HOXA9/HOXA10 and MYB under VTP-50469 in a different deposit, a different timepoint and a different peak caller. The two channels part company at exactly the loci the pivotal paper named, which is the cleanest statement in the corpus that they are not interchangeable.

Go deeper
What would overturn this

A MEN1 ChIP under a DOT1L inhibitor showing loss comparable to VTP-50469 at pre-specified target loci in any system. That is a single experiment and this deposit is the only place it has ever been run; three DOT1L inhibitors at a target-engaging dose failed to produce it. A second, independent instance - ideally in KMT2A::AFF1 - is what would make this safe to build on.

Scope, stated plainly

KMT2A::MLLT3 (MOLM13), 4 d, n = 1 per condition, within-sample contrasts only, one menin antibody. Nothing here is shown in KMT2A::AFF1. The MLL1-channel measurements this adjudicates are in murine MLL-AF9. So the adjudication is between two channels measured in the same fusion family, and it says nothing about the index fusion directly.

Facts and figures
cycle 83 threat statusSURVIVES AS STATED about the MLL1 channel - GSE149183 contains no MLL1 ChIP and cannot re-measure it. The INFERENCE drawn from it, that neither drug is diagnostic, is REFUTED on the MEN1 channel.
shown for fusionKMT2A::MLLT3 in human MOLM13 and murine MLL-AF9. Nothing in KMT2A::AFF1 - the index fusion has no MEN1 ChIP under any DOT1L inhibitor in any deposit this project has found.
channels
{
  "MEN1": "separates the drug classes completely (0/19 vs 14/19, 14/19, 8/19)",
  "DOT1L": "separates them (0/5 vs 5/5, 5/5, 5/5)",
  "MLL1_N_or_tagged_fusion": "does NOT separate them (~0.30 vs ~0.33, and 0.298 vs 0.179)"
}
systems
[
  {
    "line": "MOLM13",
    "fusion": "KMT2A::MLLT3",
    "channels": [
      "MEN1",
      "DOT1L"
    ],
    "geo": "GSE149183"
  },
  {
    "line": "murine MLL-AF9 F3_NT",
    "fusion": "KMT2A::MLLT3",
    "channels": [
      "MLL1-N",
      "FLAG-tagged fusion"
    ],
    "geo": "GSE260742"
  }
]
loci where menin and mll1 part company
[
  "HOXA9",
  "HOXA10",
  "MYB"
]
How this could bite us back
  • If menin displacement is the specific readout, it becomes the natural pharmacodynamic endpoint - but it saturates at 50 nM in this experiment, so it cannot report dose-response and a trial using it would learn nothing about exposure.
  • The clean separation of drug classes on the menin channel argues against the combination rationale that menin and DOT1L inhibitors converge on one chromatin endpoint. If they do not converge, combination benefit has to come from somewhere else and the preclinical case has to be made on that basis.
  • A resistance mutation that restores menin binding would restore the diagnostic readout while the drug is failing - MEN1 M327I and its relatives are exactly that - so menin occupancy as a PD marker is blind to the commonest clinical resistance mechanism.
Still open
  1. Does a DOT1L inhibitor at the 5 microM / 72 h exposure used in the MLL1-channel experiments displace MENIN? The GSE149183 arms are 100 nM for 4 d. If menin loss appears at the high exposure, the channel distinction collapses into a dose distinction and this entry needs rewriting.
  2. Is the MLL1 channel's mobility a shared-epitope artefact or real two-way dependence? A tagged-fusion ChIP alongside an MLL1-N ChIP in the same cells under both drugs separates them, and no deposit does it.
  3. Which falsifiers in this corpus are phrased on 'the MLL1 channel' but would be scored on a menin measurement, or vice versa? The two are not interchangeable and the vocabulary does not currently distinguish them.
Every source behind this entry (4)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    At MLL1's own top sites in murine MLL-AF9 (F3_NT), the MLL1-N share ratio is 0.3012 under 5 microM SGC0946 for 72 h and 0.3298 under 500 nM VTP-50469 - a DOT1L inhibitor moves this channel as much as a menin inhibitor does.

    What this evidence cannot show

    This is the corpus's own curated measurement (ana-q122-curated-the-mll1-channel-is-not-inert-at-its-own-sites-and-a-dot1l-inhibitor-moves-it-as-much-as-a-menin-inhibitor), computed by a beat that was killed mid-write and curated by the coordinator without re-derivation. It also lacks a regression-to-the-mean floor: bins selected for being highest in one sample fall in any second sample.

    Evidence that argues against this (1)
    • That entry's own open questions flag both the missing DMSO-versus-DMSO floor and the fact that three things changed at once between it and the measurement it contradicts (normalisation, site set, species). Its 0.30 may be partly selection rather than drug, which would make the MLL1 channel less mobile than stated and weaken the threat further - in the same direction as this entry. Oncofusion knowledge graph 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    In murine MLL-AF9, SGC0946 reduces FLAG-tagged fusion enrichment at the fusion's own top-0.1% bins from 15.9x to 6.1x over the track median (share ratio 0.298) against VTP-50469's 3.0x (0.179) - about 70% of the menin inhibitor's effect.

    What this evidence cannot show

    Murine, 5 microM SGC0946 for 48-72 h - an exposure far above any clinical DOT1L inhibitor concentration, as that entry itself notes. A within-sample share ratio, so the same share-versus-absolute caveat applies.

    Evidence that argues against this (1)
    • At 5 microM for 48-72 h a DOT1L inhibitor has collapsed the transcriptional programme, so fusion loss there may be a consequence of the cell state rather than a direct chromatin effect. The GSE149183 arms are at 100 nM for 4 d and show no menin loss at all, which is consistent with dose and duration mattering more than drug class for this channel. Oncofusion knowledge graph 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In GSE149183 (MOLM13, 4 d, hg38, one shared input), VTP-50469 at 50 nM leaves 0/19 DMSO MEN1 peaks and 0/5 DMSO DOT1L peaks at 15 pre-specified canonical target loci, while EPZ5676, compound 10 and compound 11 at 100 nM leave 14/19, 14/19 and 8/19 (MEN1) and 5/5 each (DOT1L).

    What this evidence cannot show

    n = 1 per condition; within-sample contrasts only; MOLM13 is KMT2A::MLLT3. Strength-matched permutation p = 0.0003 (MEN1) and 0.0432 (DOT1L), 20,000 draws.

    Evidence that argues against this (1)
    • On the whole DMSO peak set rather than the pre-specified loci, the same data would score the threat as surviving. That metric is shown to be unable to distinguish two different antibodies in untreated cells, which is why it is not used - but a reader who prefers it reaches the opposite conclusion and is entitled to see that stated. Oncofusion knowledge graph 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The pivotal menin-inhibitor paper reports global loss of menin binding under VTP-50469 with MLL1 not globally displaced, and names HOXA, MYB and GAPDH as loci where MLL1n occupancy is retained.

    PMID 31821784DOI
    What this evidence cannot show

    A different series, 3 d, and the MLL1n antibody's fusion-specificity is handled there by including ML2 rather than by a tag.

    Evidence that argues against this (1)
    • Krivtsov et al. did not report a DOT1L-inhibitor arm for menin occupancy, so their paper cannot itself discriminate the drug classes on the menin channel. That comparison exists only in GSE149183 and only in MOLM13. Krivtsov AV et al. 2019
Connected to