Oncofusion

start / graph / hyp-q141-half-the-menin-at-fusion-promoters-is-held-by-a-tether-that-is-not-the-fusion

Hypothesisspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q141-menin-at-fusion-targets

Half the menin at fusion-occupied promoters is held by a tether that is not the fusion

When the fusion protein is destroyed, half the menin at its genes stays behind. We propose that this leftover menin is being held by a different partner - most likely LEDGF, or ordinary MLL1 - and that this leftover pool is what a menin-blocking drug fails to reach.

The same thing, for a scientist

Statement: at KMT2A-fusion target promoters, chromatin-resident menin comprises at least two pools - a fusion-dependent pool that is lost within 3 h of fusion degradation, and a fusion-independent pool of comparable size that is not. Quantitatively, the fusion-dependent pool is ~50% of menin TSS signal at those promoters in HCB1 (KMT2A::MLLT3). Candidate tethers for the residual pool, in order of prior plausibility: **LEDGF/PSIP1** (physically links the menin-MLL1 module to chromatin; essential for MLL-fusion leukaemogenesis, dispensable for steady-state haematopoiesis), **wild-type MLL1** (present at the same promoters and demonstrably not redistributing), and a slow off-rate of a single pool (disfavoured, because the 4-day ratio of 0.437 is barely below the 3-hour 0.503). The hypothesis predicts that a menin inhibitor and a fusion degrader produce non-overlapping chromatin effects at the same promoters, and that PSIP1 loss removes the residual pool.

Go deeper
Predictions
  1. A PSIP1/LEDGF ChIP in the HCB1 degron line will show PSIP1 occupancy at fusion target promoters unchanged 3 h after degradation.
  2. PSIP1 knockdown or degradation in the same line will reduce menin occupancy at fusion target promoters by an amount comparable to the residual pool, and the two perturbations combined will approach zero.
  3. VTP-50469 (or revumenib) will reduce menin at those promoters below the level reached by fusion degradation alone, because it competes for the shared MBM interface used by both the fusion and wild-type MLL1. The 4-day data are consistent with this: VTP 0.315 vs degron 0.437.
  4. In a KMT2A::AFF1 line the ordering of pools will be the same and the magnitude will differ.
Why this is not just restating the published LEDGF model

The published model says menin bridges MLL to LEDGF. It does not say what fraction of menin at a fusion target promoter survives loss of the fusion, and that fraction is the therapeutically relevant quantity: it sets the floor that a fusion-directed therapy cannot reach and the reservoir a menin inhibitor must clear. This hypothesis is about the size of that pool and its drug sensitivity, which is measured here for the first time in this corpus.

Facts and figures
statementAt KMT2A-fusion target promoters, roughly half of chromatin-resident menin is held by a tether other than the fusion - most plausibly LEDGF/PSIP1 or wild-type MLL1 - and that pool is the reservoir a fusion-directed therapy cannot reach.
origindata-pattern
prior probability0.55
predictions
[
  "PSIP1 occupancy at fusion target promoters is unchanged 3 h after fusion degradation",
  "PSIP1 depletion removes the residual menin pool at those promoters",
  "A menin inhibitor reduces menin at those promoters below the level reached by fusion degradation alone",
  "Pool ordering transfers to KMT2A::AFF1 while magnitude does not"
]
experiments
[
  "exp-q141-spike-in-menin-cut-and-run-in-a-kmt2a-aff1-line-with-replicates",
  "exp-q139-menin-chip-after-fusion-degradation-with-a-psip1-arm-and-a-spike-in"
]
analyses
[
  "ana-q141-menin-loses-half-its-promoter-signal-where-the-fusion-sat-and-nothing-where-it-did-not",
  "ana-q141-the-menin-gap-survives-signal-matching-but-only-two-deciles-of-the-two-strata-overlap"
]
falsifiers
[
  "In an HCB1 (KMT2A::MLLT3) dTAG line, 3 h after 500 nM dTAG-VHL, spike-in-normalised menin CUT&RUN signal over the authors' published MLL::AF9 target-gene list falls BELOW 0.15 of DMSO with n>=3 per arm - i.e. menin tracks the fusion after all and there is no fusion-independent pool of the claimed size.",
  "PSIP1 depletion in the same line, at a depth sufficient to remove >80% of PSIP1 occupancy at those promoters, leaves menin occupancy at fusion target promoters within 10% of control - i.e. LEDGF is not the tether and prediction 2 fails.",
  "A menin inhibitor at a dose that fully occupies the MBM pocket leaves menin occupancy at fusion target promoters ABOVE the level reached by fusion degradation alone, which would mean the residual pool is not accessible through the MBM interface and neither candidate tether is correct."
]
How this could bite us back
  • If the residual pool is a reservoir, menin-inhibitor monotherapy is predicted to be incompletely cytoreductive, which is what the clinic shows.
  • If PSIP1 is the tether, PSIP1 becomes a combination target - and PSIP1 is also the HIV integrase tether, so any PSIP1-directed agent inherits an infectious-disease interaction profile.
  • If the residual pool is wild-type-MLL1-bound, then clearing it requires blocking the interface used by normal haematopoiesis, which is the on-target toxicity that limits the class.
Still open
  1. Is the residual pool transcriptionally competent, or inert?
  2. Does the residual fraction predict depth of response to a menin inhibitor across cell lines?
  3. Does a KMT2A::AFF1 line show the same 50% floor, or a different one? This is the single most important unmeasured quantity for this project.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Menin retains 50.3% of its signal at fusion-occupied promoters 3 h after the fusion is degraded to 1.0-2.8%, and 97.4% at non-occupied promoters.

    What this evidence cannot show

    n=1 per arm; housekeeping-normalised; KMT2A::MLLT3 in HCB1, not the index fusion.

    Evidence that argues against this (1)
    • If the residual signal were antibody background rather than menin, the stratum contrast would still appear, because background scales with signal; the decile-matching control partially but not fully addresses this. Olsen SN et al. 2022
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Menin physically links MLL proteins with LEDGF on cancer-associated target genes, and MEN1 mutations that abrogate LEDGF binding while preserving MLL binding compromise MLL/menin-dependent function.

    Menin critically links MLL proteins with LEDGF on cancer-associated target genes · Yokoyama A, Cleary ML · Cancer Cell 14(1):36-46 2008
    PMID 18598942DOI
    What this evidence cannot show

    Biochemistry, mutational analysis and transformation assays; human cell lines and mouse progenitors. Does not measure occupancy after acute fusion loss.

    Evidence that argues against this (1)
    • Those MEN1 mutants were characterised for LEDGF binding and transformation, not for chromatin residence after acute loss of an MLL fusion, so the model they support is about complex assembly rather than about the residual pool measured here. Yokoyama A et al. 2008
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    LEDGF/p75 conditional knockout decreases MLL occupancy at MLL target genes and is essential for MLL-mediated leukaemia initiation while dispensable for steady-state haematopoiesis.

    LEDGF/p75 is dispensable for hematopoiesis but essential for MLL-rearranged leukemogenesis · El Ashkar S, Schwaller J, Pieters T, Goossens S, Demeulemeester J, Christ F, et al., De Rijck J · Blood 131(1):95-107 2018
    PMID 29084774DOI
    What this evidence cannot show

    Mouse; the readout is MLL occupancy, not menin occupancy, and initiation rather than maintenance.

    Evidence that argues against this (1)
    • If LEDGF is required for MLL occupancy generally, the residual menin pool might be tethered to wild-type MLL1 via LEDGF rather than to LEDGF directly, which changes which perturbation should remove it. Yokoyama A et al. 2008
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Wild-type MLL1 measured with a C-terminal antibody does not preferentially leave fusion-occupied promoters after degradation (matched stratum gap -0.055), so it is a stable reservoir at those sites.

    What this evidence cannot show

    1,249 genes in the sheet; only 69 of 200 target genes survive the authors' floor.

    Evidence that argues against this (1)
    • If the C-terminal MLL1 ChIP is simply insensitive, a stable wild-type MLL1 reservoir would be indistinguishable from no wild-type MLL1 at all, and the third tether hypothesis would be untestable with these data. Olsen SN et al. 2022
Connected to