Why a nucleoporin fusion needs menin-MLL1: the second menin-dependent fusion family, and what it borrows rather than carries
also called NUP98 fusion menin dependency · menin-MLL1 in NUP98-rearranged AML · NUP98 MLL1 convergence
NUP98 fusions contain no piece of the MLL1 protein at all, yet blocking the menin-MLL1 handshake shuts them down. The reason is that they borrow the cell's own normal MLL1 and use it to hold open the genes that keep the leukaemia cell immature.
The same thing, for a scientist
KMT2A fusions carry the menin-binding motif **in the fusion protein itself**, so menin inhibition removes the oncoprotein from chromatin directly. NUP98 fusions carry **none of KMT2A** — they are the N-terminal FG-repeat region of NUP98 joined to a partner's C-terminus — and are nonetheless sensitive to menin-MLL1 inhibition with an IC50 *"similar to what we have previously reported for MLL-rearranged and NPM1c leukemia cells"* [E1]. The mechanism is borrowed rather than carried: **wild-type MLL1 is a molecular dependency of NUP98-fusion leukaemia**, menin tethers wild-type MLL1 to chromatin through the same MBM interface that the KMT2A fusions exploit, and disrupting it evicts **both MLL1 and the NUP98 fusion protein itself** from a critical set of target genes [E1]. Downstream this looks like the KMT2A-r programme: HOXA cluster down, *Meis1* down, CD11b and tissue differentiation markers up [E1] — although the load-bearing node is **MEIS1, not HOXA**, because MEIS1 re-expression abrogates sensitivity while NUP98-HOXD13 cells lose no Hoxa occupancy at all [E5]. The dependency is also **partner-graded**, not uniform: a roughly six-fold IC50 spread across partners, one of three PDX refractory, and NUP98::KMT2A resistant outright [E5][E6]. This is the strongest structural argument in this graph that the menin-KMT2A interface is a **family-crossing** target rather than a KMT2A-specific quirk — and the eviction-of-the-fusion result is the part that makes it more than correlation. Three things bound the claim and are set out below: an alternative acetyltransferase dependency in the same disease [E4], a shared myeloid requirement that is not fusion-specific at all (see `mechanism-kat8-mof-requirement-is-myeloid-not-fusion-specific`), and the short clinical remissions actually observed.
The structural puzzle, stated precisely
Menin binds the extreme N-terminus of KMT2A through the MBM1/MBM2 motifs. Every oncogenic KMT2A fusion retains that N-terminus, which is why menin inhibition works there: the drug detaches the oncoprotein from its chromatin anchor.
A NUP98 fusion retains the NUP98 FG-repeat region and the partner's C-terminal module. It contains no KMT2A sequence and therefore no menin-binding motif. On the KMT2A logic, menin inhibition should do nothing to it.
It does something. The resolution is that the relevant menin-MLL1 complex in a NUP98-fusion cell is the endogenous, wild-type one, and the NUP98 fusion depends on it rather than being it 1. That is a genuinely different topology of dependency and it predicts different things — most importantly, that menin inhibition in NUP98 disease removes a cofactor of the oncoprotein rather than the oncoprotein itself, which is a weaker intervention on its face and may explain why the clinical remissions are shorter (trial-save-revumenib-nup98-subgroup).
The eviction result is the load-bearing experiment
The founding study reports that on menin-MLL1 inhibition "MLL1 and the NUP98 fusion protein itself are evicted from chromatin at a critical set of genes that are essential for the maintenance of the malignant phenotype" 1.
This matters more than the viability data. A drug that kills NUP98-fusion cells could be doing so through any downstream node. A drug that physically removes the fusion oncoprotein from its target genes, by an intervention aimed at a protein the fusion does not contain, establishes that the fusion's chromatin residence is dependent on the menin-MLL1 module. That is a mechanistic dependency, not a correlation of sensitivity.
The caveat is that eviction was shown at a selected gene set, and chromatin eviction assays cannot distinguish 'the fusion needs MLL1 to be recruited' from 'the fusion and MLL1 co-occupy a template that collapses when either leaves'. The second is a real alternative and no published experiment separates them.
The programme that is shared, and the programme that is not
Shared with KMT2A-rearranged leukaemia 1:
- HOXA cluster transcriptional output, suppressed by menin-MLL1 inhibition
- MEIS1 as a downstream proleukemogenic transcription factor, downregulated
- differentiation release as the therapeutic readout (CD11b up)
- response to the same tool compound at a similar IC50
Not shared, or not yet shown to be shared:
- the fusion's own chromatin-targeting module. KMT2A fusions target through CXXC and AT-hooks; NUP98 fusions target through the partner's reader or writer domain — the KDM5A PHD finger reading H3K4me3, the NSD1 SET domain writing H3K36me2 — which are partner-specific and give partner-specific target sets.
- the resistance route. The reversible-agent trial reports 13% emergent menin-binding-site mutations across genotypes; the one published NUP98 patient escaped through RAS-pathway co-mutation with no MEN1 mutation (
pt-story-nup98-nsd1-icovamenib-complete-remission).
Correction, made in this cycle. An earlier draft of this entry predicted that menin inhibition should work across the whole NUP98 family regardless of partner, on the reasoning that the dependency is on a borrowed module. That prediction is withdrawn: the dependency is partner-graded, and in one case abolished.
- VTP50469 IC50 is 84 nM for NUP98-KDM5A against 517 nM for NUP98-HOXA9, a roughly six-fold spread within the family 5.
- NUP98::KMT2A is outright resistant to menin inhibition, because it deletes KMT2A exon 1 and with it the menin-binding motif (
fusion-nup98-kmt2a) 6. - Of three NUP98-rearranged PDX models, two responded and one was refractory 5.
The surviving prediction is narrower and better: menin dependency should track the availability of an intact menin-MLL1 tether, in cis or in trans, not the identity of the 3' partner as such. That is the hypothesis hyp-menin-dependency-tracks-available-mbm. No trial has yet reported partner-resolved response, so the clinical half remains untested.
What would refute this
- A NUP98-fusion model whose growth is unaffected by MEN1 knockout while remaining sensitive to the small-molecule inhibitor. That would show the compound works through something other than the menin interface and would collapse the family analogy to a pharmacology coincidence.
- A menin-binding-deficient MLL1 rescue that restores NUP98-fusion leukaemia. If MLL1 lacking MBM1/MBM2 rescues, the dependency is on MLL1 rather than on the interface, and the drug target moves.
- Partner-resolved clinical data showing response confined to one partner. A graded dependency across partners is already on record and is compatible with the borrowed-module model; response in only one partner, with none in the rest, would point instead at something that partner supplies and would not be a menin story at all.
- Demonstration that KAT6A/KAT7 inhibition and menin inhibition are epistatic in the same direction, making menin dependence a downstream readout of an acetyltransferase dependency rather than a parallel one 4.
| mechanism class | borrowed-cofactor dependency (the oncoprotein depends on a complex it does not contain) |
|---|---|
| contrast with kmt2a | KMT2A fusions retain MBM1/MBM2 and are themselves the menin client; NUP98 fusions retain no KMT2A sequence and depend on endogenous wild-type MLL1 |
| tool compound founding study | VTP50469 |
| prediction status | partner-graded in models; clinically untested, because no trial has reported partner-resolved response in NUP98-rearranged disease |
| withdrawn prediction | menin inhibition works across the whole NUP98 family independent of the 3' partner - WITHDRAWN this cycle: a ~6-fold IC50 spread (84 nM NUP98-KDM5A vs 517 nM NUP98-HOXA9), 1 of 3 PDX refractory, and NUP98::KMT2A resistant outright |
| surviving prediction | menin dependency tracks the availability of an intact menin-MLL1 tether rather than the identity of the 3' partner; see hyp-menin-dependency-tracks-available-mbm |
| downstream node is meis1 not hoxa | In NUP98-HOXD13 cells, menin-MLL1 inhibition caused no significant loss of occupancy or expression at the Hoxa cluster; retroviral Meis1 or CRISPRa at the endogenous Meis1 promoter abrogated sensitivity. The dependency runs through MEIS1. |
| eviction is not sufficient | Eviction of the fusion-menin-KMT2A complex does not by itself silence target genes; silencing requires PRC1.1 accumulation, and PRC1.1 loss causes menin-inhibitor resistance in vivo. |
| pdx response | 2 of 3 NUP98-rearranged PDX responded; 1 (a NUP98-NSD1 with co-occurring mutations) was refractory |
shared downstream programme
[ "HOXA cluster", "MEIS1", "differentiation block release (CD11b)" ]
partner specific targeting modules
{
"KDM5A": "C-terminal PHD finger reading H3K4me3/me2",
"NSD1": "SET domain writing H3K36me2",
"HOXA9": "homeodomain DNA binding"
}clinical agents with nup98 preclinical data
[ "revumenib", "icovamenib" ]
vtp50469 ic50 NM
{
"NUP98-KDM5A": 84,
"NUP98-HOXA9": 517
}- Removing a borrowed cofactor is a weaker intervention than removing the oncoprotein, so a NUP98-directed menin programme may have a lower efficacy ceiling than the KMT2A experience suggests, and the clinical durability data so far are consistent with that.
- Wild-type MLL1 and menin are required in normal haematopoiesis and in endocrine tissue; targeting a cofactor the leukaemia borrowed from the host means the therapeutic window is set by the host's need for it, not by anything about the fusion.
- If KAT6A/KAT7 turn out to be the dominant dependency, a menin-first strategy in NUP98-rearranged AML wastes the one clinical window these patients get, since they relapse fast and are mostly children.
- The dependency is partner-graded and at least one NUP98 partner (KMT2A itself) is resistant, so a partner-blind 'NUP98-rearranged' eligibility criterion will enrol patients who cannot benefit and will dilute the observed response rate of the patients who can.
- MEIS1 overexpression abrogates sensitivity in models, so any resistance mechanism that raises MEIS1 - and MEIS1 is downstream of several leukaemic lesions - gives escape without a MEN1 mutation, which surveillance sequencing of the menin pocket would not detect.
- Does MEN1 genetic knockout, as opposed to small-molecule inhibition, kill NUP98-fusion leukaemia? The founding study used a compound throughout; a genetic arm would separate target from chemotype and is the single cheapest experiment that could refute the whole analogy.
- Is the dependency on the menin-MLL1 INTERFACE or on MLL1 itself? An MBM-deficient MLL1 rescue in MLL1-null NUP98-fusion cells would settle it, and it has not been done.
- Does the borrowed-cofactor topology predict a lower ceiling on response than the KMT2A topology, and is the 5.9-month SAVE remission duration that ceiling showing itself?
- Are KAT6A/KAT7 and menin-MLL1 dependencies parallel or epistatic in NUP98-rearranged AML?
- Does the partner gradient (84 nM vs 517 nM) reflect the fusion or the cell line? A matched isogenic comparison expressing different NUP98 partners in one background has never been done and would settle whether partner identity really modulates menin dependency.
- Should trials stratify NUP98-rearranged patients by MEIS1 expression rather than by partner, given that MEIS1 overexpression abrogates menin-inhibitor sensitivity in models?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MLL1 is a molecular dependency in NUP98-fusion leukaemia; menin-MLL1 inhibition with VTP50469 kills NUP98-HOXA9 and NUP98-JARID1A leukaemia cells at an IC50 similar to MLL-rearranged and NPM1c cells, evicts both MLL1 and the NUP98 fusion protein from critical target genes, downregulates the HOXA cluster and Meis1, upregulates differentiation markers, and prolongs survival of NUP98-NSD1 and NUP98-JARID1A PDX mice.
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 in vitro models are MOUSE leukaemia cell lines transformed by NUP98-HOXA9 and NUP98-JARID1A, not human patient cells; the compound is the tool molecule VTP50469, not a clinical agent; survival prolongation in PDX is reported without magnitude in the abstract. 'Molecular dependency' is asserted for MLL1 in the opening sentence as prior knowledge rather than established in this work, so the MLL1-requirement claim rests on work upstream of this paper that was not read in this session.
Evidence that argues against this (2)
- A parallel dependency on KAT6A and KAT7 acetyltransferase complexes has since been reported in the same disease, so menin-MLL1 is one of at least two targetable nodes in NUP98-rearranged AML and the relative importance of the two is unresolved. Michmerhuizen NL et al. 2025
- The same study contains a non-responding NUP98-NSD1 PDX, carrying WT1, ASXL1, IDH1, BCORL1 and FLT3-ITD, so sensitivity is not uniform even within the models used to establish the dependency. Carraway HE et al. 2025
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Revumenib, a clinical menin inhibitor, blocks leukaemic engraftment and prevents leukaemia-associated death in mice transplanted with NUP98::NSD1 FLT3-ITD patient-derived AML cells, though the effect is differentiating and antiproliferative rather than apoptotic.
The MLL-Menin Interaction is a Therapeutic Vulnerability in NUP98-rearranged AML · Rasouli M, et al. · HemaSphere 7(8):e935 2023What this evidence cannot show
Single-partner (NUP98::NSD1), FLT3-ITD co-mutant PDX; the abstract's phrasing was read, not the primary figures. A cytostatic mechanism in a xenograft predicts relapse on withdrawal and the abstract does not report a treatment-cessation arm.
Evidence that argues against this (1)
- The founding study reports eviction of the fusion from chromatin and PDX survival benefit, a more definitive phenotype than 'no apoptotic death' implies. The two reports characterise the same drug class in the same disease differently and are not reconciled in the literature read here. Heikamp EB et al. 2022
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Menin inhibition combined with kinase inhibition is more effective than menin inhibition alone in leukaemia with NUP98 translocations.
Combination of menin and kinase inhibitors as an effective treatment for leukemia with NUP98 translocations · Miao H, et al. · Leukemia 38(8):1674-1687 2024What this evidence cannot show
Cited from the title and abstract-level summary; the primary combination data and the specific kinase targets were not read in detail in this session. A combination benefit in models does not establish that monotherapy is insufficient in patients.
Evidence that argues against this (1)
- A combination requirement is itself an argument that menin dependency in NUP98-rearranged disease is weaker than in KMT2A-rearranged disease, where single-agent activity was sufficient to obtain an approval. The need to add a kinase inhibitor is a quantitative concession, not a confirmation. Syndax Pharmaceuticals 2026
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
KAT6A and KAT7 histone acetyltransferase complexes are molecular dependencies and therapeutic targets in NUP98-rearranged AML.
KAT6A and KAT7 Histone Acetyltransferase Complexes Are Molecular Dependencies and Therapeutic Targets in NUP98-Rearranged Acute Myeloid Leukemia · Michmerhuizen NL, et al. · Cancer Discovery 15(10):2096-2116 2025What this evidence cannot show
Identified and verified by PubMed record in this session; the primary data were not read. Whether these dependencies are parallel to, upstream of, or downstream of the menin-MLL1 dependency is not addressed here and is an open question.
Evidence that argues against this (1)
- KAT7 is already recorded in this graph as a dependency of KMT2A-rearranged leukaemia as well, so an acetyltransferase requirement shared by both families may be a myeloid-leukaemia requirement rather than a NUP98-specific alternative to menin. The distinction is the subject of a separate entry in this beat. Heikamp EB et al. 2022
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin-MLL1 dependency in NUP98-fusion leukaemia is partner-graded and MEIS1-mediated: VTP50469 IC50 is 84 nM for NUP98-KDM5A versus 517 nM for NUP98-HOXA9; retroviral Meis1 or CRISPRa at the endogenous Meis1 promoter abrogates sensitivity; in NUP98-HOXD13 cells there is no significant loss of Hoxa occupancy or expression; and of three NUP98-rearranged PDX models two responded and one was refractory.
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
These details come from the full text rather than the abstract and were read via a delegated research pass in this session, not by the author of this entry. The IC50 comparison is between two mouse cell lines and a six-fold shift in a cell-viability IC50 is within the range that expression level or growth rate can produce. The refractory PDX carried co-occurring mutations, so partner identity and co-mutation are confounded in that comparison.
Evidence that argues against this (1)
- That MEIS1 overexpression abrogates sensitivity means the drug's effect can be bypassed downstream without touching menin at all. A leukaemia with high constitutive MEIS1, from any cause, would be predicted to be resistant regardless of its NUP98 partner - which makes MEIS1 level, not partner, the variable to stratify on, and no trial measures it. Heikamp EB et al. 2024
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
NUP98::KMT2A, which deletes KMT2A exon 1 and with it the menin-binding motif, expresses reduced rather than elevated HOXA genes and is resistant to the menin inhibitor MI-2-2 at 3-12 uM and to JQ1, doses at which KMT2A-AF9 cells arrest.
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
Single laboratory, transgenic overexpression model, tool compound at micromolar concentrations with no genetic MEN1 deletion arm. Only two patients with this fusion have ever been molecularly confirmed. A negative at high compound concentration is weaker than a genetic negative would be.
Evidence that argues against this (1)
- This fusion's mouse model has a median latency of 80 weeks and an MDS-like rather than acute phenotype, so its cells may differ from other NUP98-fusion leukaemias in ways unrelated to the menin tether; attributing the resistance specifically to loss of the menin-binding motif is an inference the paper supports but does not prove with a rescue experiment. Kaltenbach S et al. 2010
NUP98 fusions depend on endogenous menin-MLL1 although they contain no KMT2A sequence
- depends on →KMT2A / MLL1 — the gene that breaks
wild-type MLL1 is the borrowed cofactor
- regulates →NUP98::NSD1, t(5;11)(q35;p15.5)
PDX survival benefit on menin-MLL1 inhibition
NUP98-JARID1A models sensitive to VTP50469 in vitro and in PDX
HOXA cluster downregulated on menin-MLL1 inhibition
Meis1 downregulated on menin-MLL1 inhibition
- related to →KAT7 / HBO1 — an enzyme, a leukaemia-stem-cell dependency, and an inhibitor already in a crystal
KAT6A/KAT7 as a parallel, possibly competing dependency in the same disease
a borrowed-cofactor dependency is an interface dependency by another route
the clinical readout of the borrowed-cofactor prediction
- part of →NUP98 — domain architecture and residue coordinates (with the 1817/920 numbering trap resolved)
the retained FG-repeat region is the NUP98 contribution to the fusion
orthogonal DepMap check of the menin dependency, closed for want of lines
bounds the analogy: shared requirement is not convergence without a lineage-matched comparator
the NUP98 fusion that carries KMT2A is the one that cannot use menin
the physical basis for the borrowed-cofactor dependency
replaces the withdrawn partner-independence prediction
- ← refutesNUP98::HOXD13, t(2;11)(q31;p15) — the myelodysplasia fusion, and a warning against targeting p53
menin inhibition spares the Hoxa cluster in this fusion; the node is Meis1
eviction of the fusion-menin-KMT2A complex is not sufficient to silence target genes
the axis that would discriminate the two framings; it is partner-graded, which cuts against the strong form of the shared-address claim
- ← supportsNUP98::NSD1, t(5;11)(q35;p15.5)
A fusion that carries its own required methyltransferase is still menin-MLL1 dependent, which argues that the borrowed thing is the chromatin ADDRESS rather than a missing enzymatic activity.
the same nucleoporin-fusion-recruits-MLL1 shape, already documented for NUP98 fusions and already menin-dependent there