The CRM1 address is not a menin-free address: NPM1c reaches HOXA the same way and NPM1c AML is menin-inhibitor-sensitive in the clinic
also called CRM1 address menin · NPM1c HOXA CRM1 · separability hypothesis weakened · SET-NUP214 menin
One candidate fusion reaches the HOXA genes by hitching a ride on an export protein, which looked like a way to switch those genes on without needing menin. But the same export protein carries a very common leukaemia mutation to the same genes and those patients do respond to menin drugs, so the route was the wrong thing to look at.
The same thing, for a scientist
Q58 established that SET::NUP214 is addressed to HOXA by **chromatin-bound CRM1**, not by its own DNA binding — the fusion has no DNA-binding domain — and that it **recruits and condenses wild-type MLL1**, which carries MBM1/MBM2. Both findings argue it is a weaker test of menin-independence than CALM-AF10, not a second instance. **There is a sharper argument in the same literature that neither beat drew, and it comes from the clinic.** The paper establishing the CRM1 route reports that chromatin-bound CRM1 recruits **SET-Nup214 *and* NPM1c** onto HOXA clusters, by nucleoporin-CRM1 and NES-CRM1 contacts respectively. **NPM1c AML is menin-inhibitor-responsive.** It is an enrolling genotype in the menin-inhibitor programme and menin inhibition reduces MLL1-fusion targets and synergises with venetoclax and abemaciclib in NPM1c cells. **So a KMT2A-free, CRM1-addressed, HOXA-driving lesion is already known to be menin-dependent.** Whatever separates menin-dependent from menin-independent leukaemia, **it is not the CRM1 address** — and SET::NUP214's prior should move toward menin-*dependent*, not away.
What this does to the separability hypothesis
hyp-menin-axis-separable-from-hoxa-programme rests on CALM-AF10: HOXA-driving, DOT1L-dependent, architecturally menin-free. Q58 was dispatched to find a second instance and found a counter-instance instead, then argued the hypothesis's confidence should fall. This strengthens that argument from a second direction.
The three lesions now line up like this:
| lesion | address | recruits wild-type MLL1? | menin-dependent? |
|---|---|---|---|
| KMT2A::AFF1 | CXXC + menin | is the fusion | yes, established |
| NPM1c | NES-CRM1 | not established | yes, clinically |
| SET::NUP214 | nucleoporin-CRM1 | yes | untested |
| CALM-AF10 | OM-LZ/DOT1L, CRM1, MOZ/ENL, CTCF | not established | untested, inferred no |
The honest reading: the menin-independent column contains one entry and it is an inference. CALM-AF10's menin-independence was never measured — it was argued from composition, a DepMap n=1 that this graph has already bound with a power calculation, and a murine partition. Composition arguments are exactly what failed for SET::NUP214, which looked menin-free until someone checked what it recruits.
That is a general warning about how this hypothesis has been supported. 'The fusion contains no menin-binding motif' does not establish that the leukaemia is menin-independent, because the fusion can bring wild-type MLL1 — and every cell has a wild-type KMT2A allele.
What would actually settle it, and it is one experiment not three
Menin inhibitor dose-response across a panel: an index KMT2A-r line, U-937 (CALM-AF10), LOUCY (SET::NUP214), an NPM1c line as a positive control, and a menin-independent negative control. Q55 and Q58 each independently designed a version of this; it is the same experiment and should be run once.
The NPM1c arm is the part that makes it interpretable. Without it, a positive result in LOUCY could be dismissed as off-target and a negative result in U-937 as insensitivity of that line. With it, the panel has an internal standard for what menin dependence looks like in a KMT2A-free lesion.
And the readout should be HOXA cluster-wide, per Q55's correction that CALM-AF10 runs through Hoxa5 rather than Hoxa9, and per Q58's finding that SET::NUP214 output spans HOXA, HOXB and HOXC. A single-gene readout would fail for reasons that have nothing to do with menin.
| readout | HOXA cluster-wide, not a single gene |
|---|
lesions
{
"KMT2A::AFF1": {
"address": "CXXC + menin",
"menin_dependent": "established"
},
"NPM1c": {
"address": "NES-CRM1",
"menin_dependent": "clinically established"
},
"SET::NUP214": {
"address": "nucleoporin-CRM1",
"recruits_wt_MLL1": true,
"menin_dependent": "untested"
},
"CALM-AF10": {
"address": "OM-LZ/DOT1L, CRM1, MOZ/ENL, CTCF",
"menin_dependent": "untested, inferred independent"
}
}proposed panel
[ "index KMT2A-r line", "U-937 (CALM-AF10)", "LOUCY (SET::NUP214)", "an NPM1c line (positive control)", "menin-independent negative control" ]
- This argues AGAINST a hypothesis this project has been building for several cycles, and it should be read that way rather than as a refinement. The menin-independent column now contains one lesion whose independence was inferred from composition and never measured - and composition arguments are exactly what just failed for SET::NUP214.
- It is also not a demonstration. Nothing here shows SET::NUP214 responds to a menin inhibitor; it shows that the reason for expecting it not to was bad. Do not let a corrected prior be cited as a result.
- Is NPM1c's menin sensitivity mediated by wild-type KMT2A engagement, or by something about the CRM1-addressed complex? If the former, this argument weakens considerably.
- Does a menin inhibitor reduce HOXA output in LOUCY? One arm of the proposed panel, and the single most informative unrun experiment for the separability hypothesis.
- Do the two abstract-only MLL1 papers survive full-text reading? They carry the weight of the SET::NUP214 half of this argument.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.journal-article
Chromatin-bound CRM1 recruits both SET-Nup214 and the NPM1 mutant NPM1c onto HOXA cluster regions, via nucleoporin-CRM1 and NES-CRM1 interactions respectively, causing aberrant HOXA gene activation in human leukaemia cells, so the CRM1 addressing route is shared between a nucleoporin fusion and a point-mutant nucleophosmin.
Chromatin-bound CRM1 recruits SET-Nup214 and NPM1c onto HOXA clusters causing aberrant HOXA expression in leukemia cells. · Oka M, Mura S, Otani M, et al. · eLife 2019What this evidence cannot show
Human leukaemia cell lines; the SET-Nup214 arm rests on LOUCY, which with MEGAL is essentially the entire model space for this fusion (Q58 counted two lines out of 141). **Not the index disease** — T-ALL and AML rather than infant B-lineage KMT2A::AFF1 (rule 73). Read from the abstract; the abstract's gene names are partially stripped in the PubMed record, so 'HOXA' is read from the title and keywords as well as the body.
Evidence that argues against this (1)
- Sharing an addressing route does not establish a shared drug dependency. NPM1c and SET::NUP214 differ in almost every other respect - lineage, co-mutations, the rest of the complex - and NPM1c's menin sensitivity may run through its own wild-type KMT2A engagement rather than through anything CRM1 does. The argument here raises the prior on SET::NUP214 being menin-dependent; it does not demonstrate it, and only the dose-response would. Oka M et al. 2023
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.journal-article
SET::NUP214 interacts with MLL through the SET acidic region and the two cooperatively enhance the HOXA10 promoter with neither region alone sufficient, and the phase-separated nuclear bodies of SET::NUP214 promote condensation of MLL1 in robust association with CRM1 and co-localised on chromatin, so this fusion brings wild-type MLL1 - which carries the menin-binding motifs - to its target promoters.
SET-NUP214 and MLL cooperatively regulate the promoter activity of the HoxA10 gene. · Cigdem S, Saito S, Nishikata D, et al. · Genes Cells 2021What this evidence cannot show
**Both supporting papers were read at abstract level only by the beat that found them, and I verified them the same way** - neither is in PMC. That is the weakest link in the strongest claim of this argument and it is flagged here rather than buried. The promoter-activity work is a reporter assay, which measures cooperation on a construct rather than occupancy at the endogenous locus. 'MLL' in the 2021 paper is wild-type MLL1 by context. **Moderate for that reason**, despite two independent groups.
Evidence that argues against this (1)
- Recruiting wild-type MLL1 does not by itself make a lesion menin-dependent. The wild-type MLL1-menin interaction exists in every cell including normal ones, and whether displacing menin from it has any consequence at these particular promoters is precisely what has not been measured. The composition argument being used here in reverse is the same kind of argument this entry criticises when it is used the other way. Oka M et al. 2023
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin inhibition is active in NPM1-mutant (NPM1c) acute myeloid leukaemia as well as in MLL1-rearranged disease: SNDX-50469 reduces MLL1 and MLL1-fusion targets and induces differentiation and loss of viability, and combinations with venetoclax or abemaciclib are synergistic in cell lines and patient-derived AML cells harbouring either lesion.
Effective Menin inhibitor-based combinations against AML with MLL rearrangement or NPM1 mutation (NPM1c) · Fiskus W, Boettcher S, Daver N, Mill CP, Sasaki K, Birdwell CE, Davis JA, Takahashi K, Kadia TM, DiNardo CD, et al. · Blood Cancer Journal 12:5 2022What this evidence cannot show
AML rather than infant B-lineage KMT2A::AFF1 (rule 73), and pre-clinical - cell lines and patient-derived cells, with the in vivo arm being the venetoclax combination. The clinical counterpart is that NPM1c is an enrolling genotype in the menin-inhibitor programme, which is a fact about trial design rather than a reported response rate, and this entry does not quote one.
Evidence that argues against this (1)
- NPM1c menin dependence is usually explained through wild-type KMT2A/menin maintaining the HOXA programme, not through CRM1. If that is the mechanism, then NPM1c tells us about wild-type KMT2A engagement and says nothing about whether a CRM1-addressed lesion is menin-dependent - which would remove most of this entry's force. Oka M et al. 2019
LOWERS ITS PRIOR FROM A SECOND DIRECTION: NPM1c reaches HOXA by the same chromatin-bound CRM1 route as SET::NUP214 and NPM1c AML is menin-inhibitor-sensitive, so the CRM1 address is not a menin-free address.
CORRECTS ITS CENTRAL FRAMING: SET::NUP214 recruits wild-type MLL1 and is a weaker test of menin-independence than CALM-AF10, not a second instance.
THE INTERNAL STANDARD: a KMT2A-free, CRM1-addressed, HOXA-driving lesion that IS menin-inhibitor-sensitive.
WHAT SEPARATES MENIN-DEPENDENT FROM MENIN-INDEPENDENT IS NOT THE ADDRESS, on this argument.
THE REMAINING MENIN-INDEPENDENT INSTANCE, and it is an inference from composition plus an n=1 DepMap value already bound by a power calculation.
THE STANDING OBJECTION THIS HYPOTHESIS MUST ANSWER: chromatin-bound CRM1 recruits SET-NUP214 AND NPM1c onto HOXA, and NPM1c AML is clinically menin-inhibitor sensitive. So a KMT2A-free, CRM1-addressed, HOXA-driving lesion is already known to be menin-DEPENDENT. A shared address therefore cannot by itself explain menin-independence, and the hypothesis is strongest read as a claim about druggable addressing rather than about what separates the two classes.
IF XPO1 STABILISES MENIN ON CHROMATIN, THIS PRIOR FINDING BECOMES THE EXPECTED RESULT rather than a puzzle: a CRM1-addressed lesion would be menin-dependent because menin is part of the same assembly, not despite sharing the address. Carried weak - the XPO1-menin claim is a 2026 bioRxiv preprint, not peer reviewed, single group, all models myeloid.