Two constructs that would decide whether menin dependency is about the menin-binding motif or about HOXA level
also called MBM addback experiment · trans MBM deletion · cis/trans menin test
Two engineered versions of these leukaemia proteins would settle a question the whole field is arguing around: add the missing menin grip back to the one drug-resistant NUP98 fusion, and separately delete the grip from the cell's own normal copy in the drug-sensitive ones.
The same thing, for a scientist
A two-arm construct panel testing `hyp-menin-dependency-tracks-available-mbm`. **Arm 1 (cis addback):** express NUP98::KMT2A with KMT2A exon 1 restored, alongside the published exon-1-deleted fusion, and measure HOXA/MEIS1 output and menin-inhibitor dose-response. **Arm 2 (trans deletion):** in NUP98-NSD1 and NUP98-KDM5A models, replace endogenous *Kmt2a* with an MBM-deleted (Δ6–25) allele and ask whether it phenocopies a menin inhibitor and renders the cells insensitive to one. Arm 2 is the more important of the two, because it tests the limb of the hypothesis on which the entire 'NUP98 is the second menin-dependent family' claim rests, and because **no published experiment anywhere separates 'menin dependency' from 'MLL1 dependency' in NUP98-rearranged leukaemia** — the founding study knocked out *Men1* and *Kmt2a* as whole genes [E1]. A third, cheap control arm is included: genetic *Men1* deletion in a NUP98::KMT2A model, because the resistance claim for that fusion currently rests only on a micromolar tool compound [E2].
Design
Arm 1 — cis addback. Constructs: (a) NUP98::KMT2A as published (exon 1 deleted); (b) NUP98::KMT2A + KMT2A exon 1 restored in frame; (c) NUP98::KMT2A with exon 1 restored but carrying the MBM point mutations known to abolish menin binding, as the specificity control; (d) KMT2A::MLLT3 as the sensitive reference. Readouts: HOXA5/A9/A10, MEIS1, PBX3 expression; menin and MLL1 ChIP at those loci; revumenib and VTP50469 dose-response; CD11b differentiation.
Predicted under the hypothesis: (b) regains HOXA output and drug sensitivity; (c) does not. Predicted under the rival (menin inhibition works on any HOXA-high myeloid leukaemia): (b) and (c) behave the same, and only HOXA level predicts response.
Arm 2 — trans deletion. In NUP98-NSD1 and NUP98-KDM5A transformed cells, CRISPR-replace endogenous Kmt2a with a Δ6–25 allele (and, as controls, with wild-type reinserted, and with a full knockout). Readouts as above plus competitive growth.
Predicted under the hypothesis: Δ6–25 phenocopies the drug — HOXA/MEIS1 down, differentiation up — and the cells become insensitive to further menin inhibition, because the drug's target interaction is already gone. That epistasis is the signature, and it is what distinguishes a menin-interface dependency from a general MLL1 dependency: a full Kmt2a knockout would also lose the programme but would be uninformative about the tether.
Arm 3 — the missing genetic control. Delete Men1 genetically in the NUP98::KMT2A model. Published resistance is to MI-2-2 at 3–12 µM only 2, concentrations at which compound properties can dominate.
Why this is worth doing before another trial is designed
The clinical programme in NUP98-rearranged leukaemia is currently being justified by 'NUP98 fusions are menin-dependent'. That statement is supported by whole-gene knockouts and by inhibitors, neither of which localises the dependency to the interface the drugs actually bind. If the dependency is on the MLL1 complex rather than on the menin tether, then a menin inhibitor is an indirect and partial intervention in this disease, which is what the 5.9-month median remission in the only reported subgroup looks like (trial-save-revumenib-nup98-subgroup).
The experiment also produces a stratification variable for free. If Arm 2 works as predicted, then 'MBM availability' becomes a measurable property of a patient sample, and NUP98-rearranged patients could be selected rather than pooled.
| design | Three-arm construct panel. Arm 1: NUP98::KMT2A with KMT2A exon 1 restored, versus published exon-1-deleted fusion, versus exon-1-restored-but-MBM-point-mutant, versus KMT2A::MLLT3 reference. Arm 2: CRISPR replacement of endogenous Kmt2a with a menin-binding-motif-deleted (delta 6-25) allele in NUP98-NSD1 and NUP98-KDM5A models, with wild-type reinsertion and full knockout as controls. Arm 3: genetic Men1 deletion in the NUP98::KMT2A model. |
|---|---|
| readout | HOXA5/A9/A10, MEIS1 and PBX3 expression; menin and MLL1 ChIP at those loci; revumenib and VTP50469 dose-response; CD11b differentiation; competitive growth. The signature result is EPISTASIS in Arm 2: the delta 6-25 allele phenocopies the drug AND renders further menin inhibition inactive. |
| controls | MBM point-mutant addback (Arm 1c) distinguishes 'exon 1 restored' from 'menin binding restored'; full Kmt2a knockout (Arm 2) distinguishes tether loss from complex loss; KMT2A::MLLT3 is the sensitive reference; the published exon-1-deleted fusion is the resistant reference. |
| cost tier | med |
| timeline | 9-15 months; Arm 1 is a construct panel and could report in 4-6 months on its own |
| why now | The clinical programme in NUP98-rearranged leukaemia is justified by 'NUP98 fusions are menin-dependent', a statement supported only by whole-gene knockouts and inhibitors, neither of which localises the dependency to the interface the drugs bind. |
tests hypotheses
[ "hyp-menin-dependency-tracks-available-mbm" ]
- A negative Arm 2 - the delta 6-25 allele failing to phenocopy the drug - would undercut the rationale for treating NUP98-rearranged patients with menin inhibitors at all, so the experiment is capable of removing a therapeutic option that patients are currently being given off-label.
- Arm 1 uses overexpression constructs, and overexpressing a fusion with a restored menin tether could produce sensitivity that the endogenous protein would not show; a positive result would need an endogenous-level confirmation before it changed anyone's practice.
- Does an MBM-deleted Kmt2a mouse allele already exist in any laboratory? If so Arm 2 becomes a cross rather than a construction.
- Would a degron-tagged menin, allowing acute rather than chronic loss, be a better Arm 3 than deletion, given that chronic MEN1 loss has its own phenotype?
- Is NUP98::KMT2A obtainable as a patient-derived model at all? Only two patients have ever been confirmed, so Arm 1 and Arm 3 may be limited to the published transgenic mouse.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The founding study established menin dependency in NUP98-fusion leukaemia using CRISPR knockout of whole Men1 and Kmt2a genes plus the small-molecule VTP50469, with no menin-binding-motif-deleted allele, so 'menin dependency' and 'MLL1 dependency' are not separated by any published experiment in this disease.
The menin-MLL1 interaction is a molecular dependency in NUP98-rearranged AML · Heikamp EB, et al. · Blood 139(6):894-906 2022What this evidence cannot show
The absence of an MBM-deletion allele was established from a delegated full-text pass in this session; a supplementary experiment could have been missed. Whole-gene knockout plus a pocket-binding inhibitor is strong evidence that the interface matters, so this is a gap in rigour rather than a reason to doubt the conclusion.
Evidence that argues against this (1)
- A small molecule that binds the menin pocket and phenocopies MEN1 knockout is already reasonable evidence that the interface is the relevant function, since menin has other roles the knockout would also remove but the inhibitor would not. The proposed experiment tightens the argument rather than overturning it. Xu H et al. 2016
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The resistance of NUP98::KMT2A to menin inhibition is documented only pharmacologically, with MI-2-2 at 3-12 uM, and no genetic MEN1 deletion arm exists for that model.
Transforming activities of the NUP98-KMT2A fusion gene associated with myelodysplasia and acute myeloid leukemia · Fisher JN, et al. · Haematologica 105(7):1857-1867 2020What this evidence cannot show
Micromolar concentrations of a tool compound can fail for reasons unrelated to target biology - uptake, efflux, serum binding - and a negative result at high concentration is the weakest form of a negative. The comparison to KMT2A-AF9 in the same experiment is the reassuring part, because it shows the compound worked in the assay.
Evidence that argues against this (1)
- MI-2-2 did arrest KMT2A-AF9 cells at 6 and 12 uM in the same experiment (p < 0.0001), so the compound was active under those conditions and the NUP98-KMT2A negative is a genuine differential rather than an assay failure. Fisher JN et al. 2020
cis addback and trans MBM deletion, with epistasis as the signature
MBM-deleted endogenous Kmt2a allele as the decisive reagent
same cross-family interface question, different reagent
IDENTIFIES WHY THIS EXPERIMENT IS NECESSARY RATHER THAN MERELY USEFUL: polarity and MBM-in-cis are perfectly confounded across every naturally occurring lesion (5 cases where both hypotheses speak, 0 disagreements), so no patient tumour can separate the two hypotheses. An engineered MBM addback or deletion is the ONLY configuration that can.