Oncofusion

start / graph / mechanism-xpo1-stabilises-menin-chromatin-binding

Mechanismweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q61-export-axis

A 2026 preprint puts XPO1 upstream of the menin interface itself: XPO1 reported to stabilise menin on chromatin and menin's binding to KMT2A and KMT2A::MLLT3

also called XPO1 stabilises menin chromatin binding · ziftomenib plus selinexor · menin-XPO1 synergy

A 2026 preprint reports that blocking the export protein XPO1 loosens menin's grip on DNA and on the KMT2A fusion - the very grip that menin inhibitors are designed to break. Combining the two drugs killed KMT2A-rearranged leukaemia cells better than either alone, in cells, in patient cells and in mice.

The same thing, for a scientist

**This is the entry that answers the beat's headline question and it is a preprint.** Uddin et al. report that the menin inhibitor **ziftomenib** combined with the XPO1 inhibitor **selinexor** synergistically inhibits growth (CI < 1) of multiple **KMT2A-rearranged and NPM1-mutant AML cell lines**, suppresses colony formation in **primary CD34+ KMT2A-rearranged progenitors without affecting normal stem cells**, down-regulates **HOXA9 and MEIS1** while up-regulating CD11b, and improves survival in **MV4;11 (KMT2A::AFF1) and OCI-AML3 (NPM1c) cell-line xenografts and in primary patient-derived KMT2A-rearranged and NPM1-mutant xenografts in NSG mice**, at reduced doses of each drug [E1]. **The mechanistic claim is the load-bearing one and it is new to this graph:** *'a novel role of XPO1 in stabilizing menin's binding to chromatin and its interactions with KMT2A and KMT2A/MLLT3'*, with XPO1 inhibition disrupting those interactions [E1]. **Status: preprint, abstract-level, single group, not peer reviewed.** Nothing downstream should treat the mechanism as established. But two independent 2025-2026 sources point the same way - the C-body work reports **menin and XPO1/CRM1 co-partitioning in the same NPM1c condensate** [E2] - and that coincidence is what raises this from a curiosity to a claim worth testing.

Go deeper
Why this is uncomfortable for the menin-interface hypothesis, in the useful way

This graph has treated the XPO1 address and the menin interface as two different ways of arriving at HOXA: the KMT2A-fusion way (CXXC + menin) and the KMT2A-free way (chromatin-bound XPO1). concept-three-addresses-one-output-what-set-nup214-does-to-the-menin-interface-hypothesis is built on that separation.

If 1 holds, the separation is wrong in a specific way: XPO1 is not an alternative route, it is partly what holds the menin route together. That would mean:

  • A menin inhibitor and an XPO1 inhibitor are not orthogonal in the way drug-selinexor currently says they are. They act on the same assembly from two sides, which is a better reason to combine them than 'different mechanism', and a worse reason to expect non-overlapping resistance.
  • The observed synergy is then mechanistically expected rather than serendipitous, and the prediction is that XPO1 inhibition should lower the menin-inhibitor concentration needed to displace the fusion - testable directly by menin ChIP under selinexor.
  • MEN1 resistance mutations (M327I and relatives) should NOT confer cross-resistance to the XPO1 arm, because the XPO1 contribution is upstream of the pocket. That is a clean, cheap falsifier.

And it is not yet peer reviewed. The correct posture is to write the claim, mark it a preprint, and put the falsifier next to it - not to rewrite the hypothesis around it.

What the combination result does and does not establish

Does: that XPO1 inhibition adds to menin inhibition in KMT2A-rearranged AML models, including primary cells and PDX, and that the addition holds at reduced doses of each - which matters because selinexor's paediatric dose ceiling is set by reversible cerebellar toxicity at 70 mg/m2 (see drug-selinexor).

Does not: show that the effect is XPO1-chromatin-mediated rather than the sum of two unrelated cytotoxicities. XPO1 is common-essential (DepMap mean -2.14 across 1,208 lines, see gene-xpo1), so 'selinexor plus X kills better than X' is close to the null expectation for any X. The abstract's own answer to that is the mechanism claim - the reported loss of menin-chromatin and menin-KMT2A interaction - and a preprint abstract is not enough to adjudicate it.

Nor does it touch infant B-lineage disease. Every model named is AML. MV4;11 carries KMT2A::AFF1, which is this graph's index fusion, but it is a myeloid line from a 10-year-old with biphenotypic leukaemia, not infant B-ALL.

Facts and figures
mechanism classcofactor stabilisation / drug combination
claim statuspreprint, abstract-level, single group, unreplicated
mechanistic claimXPO1 stabilises menin binding to chromatin and menin interactions with KMT2A and KMT2A::MLLT3; XPO1 inhibition disrupts them
falsifierMEN1 pocket resistance mutations (e.g. M327I) should not confer cross-resistance to the XPO1 arm if the XPO1 contribution is upstream of the pocket
not testedinfant B-lineage KMT2A::AFF1; any lymphoid model; any clinical combination
drugs
[
  "drug-ziftomenib",
  "drug-selinexor"
]
models
[
  "multiple KMT2A-rearranged AML cell lines",
  "multiple NPM1-mutant AML cell lines",
  "primary CD34+ KMT2A-rearranged progenitors",
  "MV4;11 cell-line xenograft (KMT2A::AFF1, myeloid)",
  "OCI-AML3 cell-line xenograft (NPM1c)",
  "primary patient-derived KMT2A-r and NPM1-m xenografts in NSG mice"
]
readouts
[
  "combination index <1",
  "colony formation",
  "apoptosis",
  "G2/M fraction",
  "HOXA9 and MEIS1 expression",
  "CD11b",
  "RNA-seq",
  "proteomics",
  "survival"
]
How this could bite us back
  • Selinexor's paediatric ceiling is set by reversible cerebellar toxicity at 70 mg/m2 and its class toxicities are anorexia, weight loss and thrombocytopenia; ziftomenib brings differentiation syndrome and QTc. A combination inherits both toxicity profiles and there is no paediatric safety data for it.
  • If XPO1 really does stabilise the menin-KMT2A assembly, then the two drugs are NOT orthogonal, and resistance that loosens the assembly could reduce sensitivity to both at once - the opposite of the usual argument for combining.
  • The synergy was measured at reduced doses in mice. A dose reduction that preserves efficacy in NSG xenografts says nothing about whether it preserves efficacy against a human marrow burden.
  • This is a preprint. Building a clinical argument on an unreviewed abstract is the exact reasoning pattern that produced the DOT1L programme.
Still open
  1. Does menin ChIP occupancy at HOXA fall under selinexor alone, in a KMT2A-rearranged cell? That single experiment converts the preprint's central claim from an assertion into a measurement, and nothing in this graph has it.
  2. Do MEN1 resistance mutations cross-resist selinexor? If they do, the 'upstream of the pocket' reading is wrong and the synergy is two independent cytotoxicities.
  3. Is the synergy present in a lymphoid KMT2A-rearranged model, or is it an AML phenomenon? Every model in the preprint is myeloid.
  4. What is the schedule of the combination? The NPM1c literature says intermittent XPO1 inhibition fails; a combination designed around weekly selinexor may inherit that failure.
Every source behind this entry (2)
  1. E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.preprint

    In KMT2A-rearranged and NPM1-mutant AML models, the menin inhibitor ziftomenib combined with the XPO1 inhibitor selinexor synergistically inhibited growth of multiple cell lines (CI<1), suppressed colony formation in primary CD34+ KMT2A-rearranged progenitors without affecting normal stem cells, induced apoptosis, down-regulated HOXA9 and MEIS1 and up-regulated CD11b, and improved survival in MV4;11 and OCI-AML3 cell-line and primary patient-derived xenografts in NSG mice at reduced doses; the authors report a novel role of XPO1 in stabilizing menin's binding to chromatin and its interactions with KMT2A and KMT2A/MLLT3, disrupted by XPO1 inhibition.

    PMID 41959134DOI
    What this evidence cannot show

    PREPRINT, NOT PEER REVIEWED, posted 2026-03-13, read at abstract level only on 2026-09-20. Single group. All models are AML - KMT2A-rearranged and NPM1-mutant - and none is infant B-lineage ALL; MV4;11 is a myeloid line from a paediatric patient, not an infant B-ALL model. The synergy readout is a combination index, which is sensitive to the concentration grid chosen and is not a mechanism. XPO1 is common-essential, so an additive kill with any partner is close to the null. The mechanistic claim - XPO1 stabilising menin on chromatin - is the most consequential sentence in the abstract and no figure supporting it was read in this beat. The claim must be re-checked against the published version and against the paper's own ChIP data before it is relied on.

    Evidence that argues against this (1)
    • The same combination logic has a documented failure mode in the other direction: selinexor given once or twice weekly produced only a temporary disruption of the XPO1-NPM1c interaction and no clinical benefit in NPM1-mutated patients, which is why prolonged inhibition with eltanexor was needed to get irreversible HOX down-regulation. A synergy shown under continuous in vitro exposure does not transfer automatically to an intermittent clinical schedule. Pianigiani G et al. 2022
  2. E2weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.review

    A commentary on the 2025 C-body work reports that NPM1c nuclear condensates co-partition NUP98, KMT2A/MLL1, menin and XPO1/CRM1, and that inhibiting either the XPO1/CRM1 or the menin interaction significantly alters condensate composition and functionality.

    A Unified Model: Chromatin-Bound Multicomponent Condensates. · Lau BK, Haarer EL, Ahn JH, et al. · Cancer Research 2026
    PMID 41481253DOI
    What this evidence cannot show

    A COMMENTARY describing another group's figures, read at abstract level. Cited here only as an independent pointer in the same direction as [E1], not as a measurement. Co-partitioning in a condensate is not the same claim as XPO1 stabilising menin on chromatin, and conflating them would be exactly the kind of drift this graph exists to prevent.

    Evidence that argues against this (1)
    • NAMED METHODOLOGICAL DEFEATER for reading the commentary's co-partitioning statement as a pointer in the same direction. The commentary describes composition and functionality of condensates, read from another group's images; Muzzopappa et al. showed that the two assays that establish such structures as condensates - 1,6-hexanediol sensitivity and classical FRAP - cannot distinguish liquid-liquid phase separation from molecules binding spatially clustered sites, and that only calibrated half-bleaching does. Worse for the inference: if XPO1 is an abundant chromatin-associated cofactor rather than a specific partner, then 'inhibiting the XPO1/CRM1 interaction significantly alters condensate composition' is what removing any abundant component would do. The caveat is right that co-partitioning is not stabilisation; the assay cannot currently establish either. Muzzopappa F et al. 2022
Connected to