Oncofusion

start / graph / hyp-evicted-fusion-is-degraded-not-displaced

Hypothesisspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q4b-localization

Hypothesis: menin inhibition does not merely displace KMT2A::AFF1 from chromatin, it destabilises it — the evicted fusion is degraded, not parked

also called fusion degradation on menin inhibition · evicted-and-destroyed · CHIP Hsp70 fusion turnover

Menin inhibitors knock the fusion protein off DNA, and everyone assumes it then sits in the nucleus waiting to come back. This proposes the opposite: stripped of its chromatin anchor the fusion becomes unstable and the cell destroys it, which would mean the drug is a partial degrader and that intermittent dosing could work.

The same thing, for a scientist

Premise chain. (i) Menin inhibitors evict menin and MLL1 from chromatin, and in NUP98-rearranged AML the fusion protein itself is evicted, all measured by chromatin occupancy. (ii) CORRECTED 2026-09-19: this previously read 'No study has measured the abundance, half-life or location of a KMT2A fusion protein after menin inhibition.' That was wrong, and refuted by a paper this graph already cites elsewhere. Abundance HAS been measured once: in a KMT2A-rearranged sample that lineage-switched under revumenib, the oncoprotein and menin were both reduced to UNDETECTABLE levels [E_oncoprotein_abundance_measured]. Half-life, degradation route and subcellular location remain unmeasured, and n=1 in a switching sample cannot separate protein destruction from loss of the expressing population - so the hypothesis stands, but its premise is now 'measured once, in the predicted direction' rather than 'never measured'. (iii) The one protein-fate measurement in this literature shows that menin-MLL inhibitors reduce **menin** protein level and half-life without changing mRNA, via inhibitor-promoted menin-CHIP (Hsp70-associated ubiquitin ligase) interaction and increased ubiquitination, rescued by proteasome or E1 inhibition. (iv) Large chromatin proteins are typically stabilised by their chromatin and complex partners; an evicted KMT2A fusion is a 240-kDa protein with exposed interaction surfaces and no anchor, which is a canonical chaperone-triage substrate. **Hypothesis:** the same CHIP/Hsp70 triage that destroys evicted menin also acts on the evicted fusion, so menin inhibition reduces KMT2A::AFF1 protein half-life measurably (predicted >30% reduction in half-life within 24 h of saturating exposure), and the transcriptional and differentiation response is partly attributable to loss of fusion protein rather than only to loss of fusion occupancy. Prior probability raised 0.3 -> 0.45 on this correction, because the one existing measurement fell in the predicted direction: the chaperone logic is sound and the menin precedent is direct, but large scaffold proteins are often surprisingly stable and the fusion could equally be stabilised by AFF1/SEC binding independent of chromatin.

Go deeper
Why this is worth stating as a hypothesis rather than assuming either way

The graph already contains a cluster of questions whose answers hinge on this one and which are currently being reasoned about without it: whether pulsed menin dosing can work (hyp-agency-revocation-hysteresis-pulsed-menin, exp-menin-pulse-chase-hysteresis), whether residual occupancy explains incomplete responses (hyp-apparent-authority-residual-fusion-occupancy), and whether the fusion is continuously required (hyp-fusion-continuously-required-not-hit-and-run).

If the fusion is degraded on eviction, recovery after drug withdrawal is gated by resynthesis, which is slow, and hysteresis is expected. If it is merely displaced, recovery is gated by rebinding, which is fast, and hysteresis is not expected. The same experiment therefore arbitrates three existing entries.

Predictions, in order of how cheaply they can be checked
  1. Cycloheximide chase in SEM or MV4;11 with and without revumenib: fusion half-life shortens by at least 30% at 24 h. Western blot with an antibody that resolves the 240-kDa species from wild-type AFF1 by size, which the 1998 antibody work shows is feasible.
  2. Proteasome inhibition (bortezomib or MG132, short exposure) partially rescues fusion protein level under menin inhibition and partially rescues target-gene suppression, mirroring the published menin result.
  3. Co-immunoprecipitation shows increased fusion-CHIP/STUB1 or fusion-HSPA8 association under menin inhibition.
  4. Ubiquitin-remnant proteomics shows increased ubiquitination of peptides in the retained KMT2A N-terminal region under drug.
  5. In the imaging time course, total nuclear fusion signal declines rather than redistributing.

A composite negative on 1, 2 and 5 kills the hypothesis outright.

Facts and figures
statementMenin inhibition destabilises the KMT2A fusion protein, so a measurable fraction of the therapeutic effect is loss of fusion protein rather than loss of fusion chromatin occupancy alone.
originliterature-gap
prior probability0.3
predictions
[
  "Fusion protein half-life falls by at least 30% within 24 h of saturating menin-inhibitor exposure in a cycloheximide chase",
  "Proteasome or E1 inhibition partially rescues both fusion protein level and target-gene suppression under menin inhibition",
  "Fusion-STUB1/CHIP or fusion-HSPA8 co-immunoprecipitation increases under menin inhibition",
  "Total nuclear fusion fluorescence declines rather than redistributing in a live-imaging time course",
  "Recovery of target-gene expression after drug washout is gated by protein resynthesis rather than rebinding, producing measurable hysteresis"
]
experiments
[
  "exp-fusion-fate-imaging-menin-timecourse"
]
analyses
[
  "ana-fusion-halflife-degron-vs-menin"
]
falsifiers
[
  "Cycloheximide chase shows no change in fusion protein half-life under saturating menin inhibition",
  "Proteasome inhibition fails to rescue fusion protein level while rescuing menin level in the same lysate — the internal positive control for the assay working",
  "Live imaging shows total nuclear fusion signal unchanged with redistribution to a different compartment, which supports a competing entry rather than this one",
  "Target-gene expression recovers after washout with kinetics matching rebinding rather than resynthesis"
]
How this could bite us back
  • If menin inhibitors are partial degraders, then combining them with an HSP90 inhibitor or a proteasome inhibitor — both plausible combination partners in AML — could antagonise the very mechanism being relied on.
  • If the effect is real, cells can resist by stabilising the fusion (chaperone upregulation, E3 loss) with entirely wild-type MEN1, a resistance class invisible to the MEN1 sequencing that is currently the standard.
Still open
  1. If the fusion is degraded, is it via CHIP/STUB1 as menin is, or via a different E3? The answer determines which resistance genes to sequence in relapse samples.
  2. Does the reciprocal AFF1::KMT2A product, where expressed, behave the same way? It has no menin-binding motif, so it should be unaffected — which makes it a built-in internal control.
  3. Is the effect shared across the menin-inhibitor class (revumenib, ziftomenib, bleximenib, enzomenib) or compound-specific? Class-wide would make fusion protein level a general pharmacodynamic biomarker.
  4. Does the undetectable oncoprotein in the revumenib-treated switching sample reflect protein destruction or replacement of the fusion-expressing population? A clonality-resolved protein measurement separates them and is the difference between this hypothesis and its rival.
Every source behind this entry (5)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Menin-MLL inhibitors reduce menin protein levels and shorten menin half-life in MLL-fusion-expressing leukemia cells without changing mRNA; proteasome or E1 ligase inhibition rescues the degradation, and the inhibitors promote menin interaction with the Hsp70-associated ubiquitin ligase CHIP with increased menin ubiquitination.

    Disruption of the menin-MLL interaction triggers menin protein degradation via ubiquitin-proteasome pathway · Wu Y, Doepner M, Hojnacki T, Feng Z, Katona BW, He X, Ma J, Cao Y, Busino L, Zhou F, Hua X · American Journal of Cancer Research 2019
    PMID 31497350
    What this evidence cannot show

    Measures menin, not the fusion; single-group, cell lines only, modest journal, not widely replicated. It establishes the mechanism exists for one partner in this complex under this drug class, which is the basis for the hypothesis and not evidence for it.

    Evidence that argues against this (1)
    • Direct counter-evidence, and it lands on the retained fragment itself. Liu, Cheng and Hsieh show that wild-type MLL is degraded in two defined cell-cycle windows, by SCF(Skp2) at S phase and APC(Cdc20) at late M; that degradation initiates within the N-terminal approximately 1400 amino acids - the exact segment every KMT2A fusion retains; and that the prevalent MLL fusions they tested are RESISTANT to that degradation. They propose impaired degradation of MLL fusions as a universal feature of MLL leukemias, i.e. abnormal stability is part of what makes the fusion oncogenic. This hypothesis assumes the retained N-terminus becomes a chaperone-triage substrate once unanchored from chromatin; the fusion has already escaped the ubiquitin-proteasome pathway that reads that same region. It does not formally exclude a distinct CHIP/Hsp70 route engaged only on eviction, but it removes the 'a large unanchored scaffold must be unstable' intuition on which the prior rests, and argues the stated prior probability of 0.3 is too high - 0.1-0.15 is better supported. Read: abstract only (PMC full text was empty on retrieval). Liu H et al. 2007
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Menin inhibition evicts MLL1 and, in NUP98-rearranged AML, the NUP98 fusion protein itself from chromatin at genes required for the malignant phenotype.

    The menin-MLL1 interaction is a molecular dependency in NUP98-rearranged AML · Heikamp EB, Henrich JA, Perner F, Wong EM, Hatton C, Wen Y, Barwe SP, Gopalakrishnapillai A, Xu H, Uckelmann HJ, Takao S, Kazansky Y, Pikman Y, McGeehan GM, Kolb EA, Kentsis A, Armstrong SA · Blood 2022
    PMID 34582559DOI
    What this evidence cannot show

    Chromatin-occupancy readout only; establishes that eviction of a fusion protein happens but says nothing about its subsequent fate, which is the content of the hypothesis.

    Evidence that argues against this (1)
    • Searched and empty, recorded as a fact rather than an absence. PubMed searches for a study measuring KMT2A-fusion protein abundance, half-life or subnuclear position after menin inhibition (combinations of menin inhibitor / revumenib / VTP-50469 with fusion protein level, western blot, degradation, half-life, eviction, live imaging, single molecule) returned nothing. The chromatin-occupancy result therefore still stands alone: no published experiment distinguishes 'evicted and intact' from 'evicted and destroyed' for any KMT2A fusion, so the hypothesis is untested rather than unsupported. Searches run 2026-09-18 against PubMed; no full texts required. No citation is attached because there is nothing to cite: this note is a bounded absence recording searches that returned nothing, not a sourced fact. Uncited and unciteable look identical in this field, so it is said here explicitly (Q42 citation-integrity repair, 2026-09-20).
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The MLL-AF4 fusion protein is detectable as a 240-kDa species distinguishable by size from the 125 and 145 kDa wild-type AF4 proteins in t(4;11) cell lines, and is present at lower abundance than wild-type AF4.

    PMID 9808577
    What this evidence cannot show

    Cited for feasibility rather than for the hypothesis: it establishes that the fusion can be resolved and quantified by western blot, which is what makes the cycloheximide-chase prediction testable with standard reagents.

    Evidence that argues against this (1)
    • The feasibility claim is calibrated on one fusion configuration and does not generalise across the disease. Meyer et al. determined genomic breakpoints in 2,345 patients and found 135 distinct MLL rearrangements, 94 molecularly characterised partners, and an age- and lineage-dependent shift in the MLL-side breakpoint. Fusion protein mass therefore varies with junction, so a cycloheximide chase read out by size resolution needs the junction typed per model, and any cross-line comparison of fusion half-life is comparing different proteins. This does not weaken the hypothesis - it means the experiment has to be done per clone, with the tag rather than the size doing the identification. Meyer C et al. 2018
  4. E_igf2bp3_abundance_confoundmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    KMT2A::AFF1 protein abundance is actively regulated and is not a free readout. Model systems in t(4;11) show the RNA-binding proteins IGF2BP1-3 influence fusion mRNA and protein abundance; KMT2A::AFF1 (MA4) alone drives aberrantly high expression of only a few target genes while the reciprocal AFF1::KMT2A (A4M) COUNTERACTS that effect, and MA4 induces RNA-binding-protein genes as a counterregulatory response. This does not refute this entry's claim, which is specifically about abundance after MENIN INHIBITION and still stands - but it means the proposed measurement has confounders the hypothesis does not currently account for.

    Influence of IGF2BP3 on t(4;11) fusion mRNA and protein abundance · Hanewald T, Benz T, Marschalek R · Leukemia Research 168:108286 2026
    PMID 42537480DOI
    What this evidence cannot show

    Read as abstract only, fetched 2026-09-19. Model systems rather than patient material, and the abstract does not quantify the effect sizes. From the Marschalek laboratory, which this project has flagged as the single source behind much of its descriptive layer - the same caution applies here.

    Evidence that argues against this (1)
    • Read against this entry the paper is a warning about experimental design, not support. If fusion abundance is set by RNA-binding proteins and opposed by the reciprocal, then measuring fusion protein level after menin inhibition and attributing a change to degradation-versus-displacement would be underdetermined: an IGF2BP-mediated or reciprocal-mediated shift would look identical. The experiment this hypothesis proposes needs those arms measured in the same samples, and as written it does not have them. Hanewald T et al. 2026
  5. E_oncoprotein_abundance_measuredmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    KMT2A fusion protein abundance HAS been measured under menin inhibition in a human sample. In a KMT2A-rearranged leukaemia that lineage switched during revumenib treatment, the oncoprotein and menin were both reduced to undetectable levels, while the cofactor ENL persisted on oncoprotein-target loci. The direction matches this hypothesis's central prediction that the fusion protein itself is lost, not merely relocated.

    PMID 39472576DOI
    What this evidence cannot show

    Found 2026-09-19 by the stale-correction queue, which flagged premise (ii) as an un-struck absence claim; the refuting paper was already cited by resistance-enl-marked-epigenetic-memory. n=1, and a lineage-switching sample is the worst case for this inference: an undetectable oncoprotein is equally consistent with the fusion-expressing population being replaced by a switched one. Abundance is not half-life, and nothing here identifies a degradation route.

    Evidence that argues against this (1)
    • This measurement cannot distinguish the hypothesis from its main rival. If the oncoprotein is undetectable because the cells expressing it were outcompeted by a switched population, no protein was degraded at all and the premise chain never engages. The same paper's headline finding - that ENL persists at target loci after the oncoprotein is gone - also weakens the therapeutic payoff: if destroying the fusion leaves the programme running, whether it is degraded or parked matters less than this entry assumes. Nature Communications 2024
Connected to