NUP98::NSD1, t(5;11)(q35;p15.5)

This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.

• Targeting the condensate is not a formulable strategy. LLPS is driven by hundreds of weak multivalent contacts with no defined pocket and no stoichiometry, and the FUS-IDR swap experiment shows the sequence is interchangeable [E3]. Anything that dissolves this condensate will dissolve the cell's normal condensates - nucleoli, splicing speckles, stress granules - which is a general-toxicity mechanism, not a therapy.
• The NSD1 SET domain is the tempting enzymatic target, but NSD1 is required for normal development: germline NSD1 loss of function causes Sotos syndrome (overgrowth with intellectual disability). Inhibiting H3K36 methylation in a growing child carries a developmental on-target risk with no adult analogue.
• FLT3 inhibition is the available pharmacology but not the driver. FLT3-ITD is a co-mutation in 91% of cases [E4], and in FLT3-ITD AML generally, single-agent FLT3 inhibition produces short-lived responses with on-target resistance (activation-loop mutations such as D835, and RAS-pathway bypass). Expect the same here, plus the possibility of the fusion-driven HOXA programme surviving FLT3 blockade entirely.
• The lesion is cryptic: 78% of cases have a normal karyotype [E4]. Any centre that does not run RNA-seq or a targeted NUP98 assay will classify these children as intermediate-risk cytogenetically normal AML and under-treat them. Diagnostic under-ascertainment is itself a bite-back on outcomes.
• The prognostic data come from 23 patients [E4]. A high-unmet-need designation resting on that many cases is vulnerable to the same regression-to-the-mean that revised early AALL0031 estimates for Ph+ ALL; the outcome may be less catastrophic in the FLT3-inhibitor era, and the graph should expect that number to move.

1. Does NUP98::NSD1 specifically form liquid-like condensates at ENDOGENOUS expression levels in patient blasts? The family-level evidence is strong [E2][E3] and the original clinical paper reports nuclear aggregates [E4], but FRAP and live-cell imaging at endogenous concentration in a patient-derived model has not, to our knowledge, been done for this fusion. This is the single most important missing experiment in the node.
2. If LLPS competence rather than sequence is what matters [E3], is the condensate itself therapeutically addressable at all? Candidate approaches - aliphatic alcohols, aromatic-interaction disruptors, altering cellular RNA or ATP levels - are all grossly non-selective. Is there any route to a condensate-selective agent, or is the honest answer that this mechanism class requires targeting recruited machinery instead?
3. NSD1 SET catalytic activity is REQUIRED for transformation [E1] and the SET domain has a real SAM pocket [E5]. Why has no NSD1 inhibitor programme reached the clinic, and would a SET-domain inhibitor be tolerable given that germline NSD1 loss causes Sotos syndrome?
4. The FLT3-ITD co-occurrence rate of 91% [E4] is far too high to be coincidence. Does the fusion create a selective requirement for FLT3 signalling (for instance by enforcing a progenitor state dependent on it), or does FLT3-ITD merely provide the proliferative signal the HOXA programme cannot supply? If the former, FLT3 inhibition should be more effective here than in FLT3-ITD AML generally - which is a testable clinical prediction.
5. NUP98::NSD1 and KMT2A fusions converge on the same HOXA9/MEIS1 programme by different mechanisms. Are NUP98-rearranged blasts menin-dependent or DOT1L-dependent, and if so, is the dependency partner-agnostic across the NUP98 family?
6. Does menin inhibition change the H3K36me2 landscape written by the retained NSD1 SET domain, or does the fusion keep writing it at whatever loci it can still reach? That is the experiment that would separate the address explanation from the enzyme explanation directly - H3K36me2 CUT&RUN plus fusion CUT&RUN in the same NUP98::NSD1 cells before and after revumenib. Nobody has run it, and the address model predicts dissociation: fusion occupancy falls at menin-MLL1 target promoters while its methyltransferase activity is unchanged elsewhere.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   NUP98-NSD1 induces AML in vivo, sustains myeloid stem-cell self-renewal, binds elements adjacent to HoxA7/HoxA9, maintains H3K36 methylation and acetylation there and prevents EZH2-mediated repression of the Hox-A locus; deletion of the NUP98 FG-repeat domain, inactivation of the NSD1 H3K36 methyltransferase activity, or loss of Hox-A binding each abolishes Hox-A activation and progenitor immortalisation.

   NUP98-NSD1 links H3K36 methylation to Hox-A gene activation and leukaemogenesis · Wang GG, Cai L, Pasillas MP, Kamps MP · Nat Cell Biol 2007
   PMID 17589499DOI
   ▸What this evidence cannot show

   Murine retroviral expression in haematopoietic progenitors with non-physiological fusion levels; the EZH2-antagonism model is inferred from marks and from differentiation-associated repression, not from direct measurement of PRC2 recruitment kinetics. The requirement experiments are loss-of-function in the mouse system and were not repeated in human cells carrying the endogenous translocation.

   ▸Evidence that argues against this (1)

   • NUP98-NSD1 did not induce AML on its own in the follow-up in-vivo test. Thanasopoulou et al. found that retroviral NUP98-NSD1 conferred mouse-strain-dependent aberrant self-renewal on bone-marrow progenitors, but transplantation of cells expressing NUP98-NSD1 ALONE produced no leukaemia; only co-expression with FLT3-ITD did, after short latency, and FLT3-ITD alone did not either. Time to disease tracked the Flt3-ITD-to-wild-type-Flt3 mRNA ratio and FLT3 signalling and NOT the expression level of the fusion or of HoxA5, HoxA7, HoxA9 or HoxA10, and the resulting cells were highly sensitive to a small-molecule FLT3 inhibitor. The Hox-A programme that the domain-dissection experiments show to be NECESSARY is therefore not, on this evidence, the RATE-LIMITING step for disease - the kinase co-mutation present in about 91% of patients [E4] is. Thanasopoulou A et al. 2014
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Across five different NUP98 fusion proteins expressed in human leukaemia cells, shared interactors are enriched for biomolecular-condensation proteins that colocalise with the fusions in nuclear puncta; NUP98 fusions alter the global composition of cellular condensates, and an artificial FG-repeat-containing fusion protein phenocopies both their nuclear localisation and their capacity to drive oncogenic gene expression.

   Biomolecular condensation of NUP98 fusion proteins drives leukemogenic gene expression · Terlecki-Zaniewicz S, Humer T, Eder T, Schmoellerl J, Heyes E, Manhart G, Kuchynka N, Parapatics K, Liberante FG, Muller AC, Tomazou EM, Grebien F · Nat Struct Mol Biol 2021
   PMID 33479542DOI
   ▸What this evidence cannot show

   Ectopic expression in cell lines (including HEK293, HL-60, NIH-3T3) rather than patient blasts; nuclear puncta by confocal imaging are consistent with condensates but the study does not prove liquid-like behaviour for each fusion by FRAP or in vitro reconstitution. biCon-MS is a new method whose specificity for genuine condensate components is not independently validated. Whether NUP98::NSD1 specifically was among the five fusions tested should be checked before citing this for NSD1 alone.

   ▸Evidence that argues against this (1)

   • Held to the graph's own condensate evidence standard (localization-condensate-evidence-standards), this is an association rather than a demonstration, and the failure mode that bites hardest here is fixation. Irgen-Gioro et al. imaged the same cells before and after paraformaldehyde fixation and found fixation can both ENHANCE and DIMINISH apparent liquid-liquid phase separation; for specific proteins it produced droplet-like puncta in cells that had NONE when live, and fixing in the presence of glycine reversed the direction of the artefact. Their kinetic model and live-cell single-molecule data show the outcome depends on the balance between fixation rate and protein-protein interaction dynamics. 'Shared interactors enriched for biomolecular-condensation proteins that colocalise with the fusions in nuclear puncta' is a proteomic enrichment plus imaged puncta; neither observable discriminates phase separation from a stoichiometric multiprotein complex, and such a complex - MLL1 plus NSL - has been shown to be present and functionally required for exactly these fusions. Irgen-Gioro S et al. 2022
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The NUP98 FG-repeat IDR is essential for liquid-liquid phase separation and leukaemic transformation by NUP98-HOXA9; LLPS both promotes chromatin occupancy and is required for the broad super-enhancer-like binding pattern at leukaemogenic genes, and it induces CTCF-independent chromatin loops enriched at proto-oncogenes. Substituting the unrelated LLPS-forming IDR of FUS reproduces the effect.

   Phase separation drives aberrant chromatin looping and cancer development · Ahn JH, Davis ES, Daugird TA, Zhao S, Quiroga IY, Uryu H, Li J, Storey AJ, Tsai YH, Keeley DP, Mackintosh SG, Edmondson RD, Byrum SD, Cai L, Tackett AJ, Zheng D, Legant WR, Phanstiel DH, Wang GG · Nature 2021
   PMID 34163069DOI
   ▸What this evidence cannot show

   The mechanistic work is on NUP98::HOXA9, NOT NUP98::NSD1, so applying it here is a family-level inference. Much of the imaging and Hi-C is in HEK293/HeLa and murine systems with ectopic expression; LLPS at endogenous fusion concentrations in patient blasts is not demonstrated. The FUS-IDR substitution shows sufficiency of an IDR in this assay but does not prove that the native FG repeats work purely by LLPS rather than also by specific interactions.

   ▸Evidence that argues against this (1)

   • A general caution rather than a specific refutation: the field has repeatedly found that puncta observed under ectopic overexpression can be artefacts of concentration, and several high-profile transcriptional-condensate claims have been challenged on exactly those grounds. The NUP98 case is stronger than most because of the IDR-swap sufficiency experiment, but the concentration caveat applies. We did not find a paper specifically refuting NUP98-fusion LLPS. The concentration caveat is now cited rather than asserted: Chong et al. (Science 2018, PMID 29930090) show by live-cell single-molecule imaging that under physiological conditions LCD-LCD interactions occur without detectable phase separation, and Muzzopappa et al. (Nat Commun 2022, PMID 36526633) show that 1,6-hexanediol and classical FRAP fail to separate phase separation from binding to spatially clustered sites. The residual 'we found no paper specifically refuting NUP98-fusion LLPS' is a bounded absence with nothing to cite. Chong S et al. 2018
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   NUP98/NSD1 is a cryptic fusion present in 16.1% of paediatric and 2.3% of adult cytogenetically normal AML, with 78% normal karyotype, FLT3-ITD in 91%, WT1 mutation in 45%, median WBC 147 x 10^9/L, FAB M4/M5 in 63%, mutual exclusivity with all recurrent type-II aberrations, aberrant localisation of the fusion protein in nuclear aggregates, and 4-year event-free survival below 10% as an independent predictor of poor prognosis.

   NUP98/NSD1 characterizes a novel poor prognostic group in acute myeloid leukemia with a distinct HOX gene expression pattern · Hollink IHIM, van den Heuvel-Eibrink MM, Arentsen-Peters STCJM, Pratcorona M, Abbas S, Kuipers JE, van Galen JF, Beverloo HB, Sonneveld E, Kaspers GJJL, Trka J, Baruchel A, Zimmermann M, Creutzig U, Reinhardt D, Pieters R, Valk PJM, Zwaan CM · Blood 2011
   PMID 21813447DOI
   ▸What this evidence cannot show

   Only 23 fusion-positive cases were found among >1000 screened, so the survival estimate rests on small numbers and the confidence interval around 'EFS <10%' is wide. The percentage frequencies are conditional on the cytogenetically normal subgroup, which is a selected denominator. Therapy predates routine FLT3 inhibition, so the dismal outcome reflects an era in which the 91% FLT3-ITD co-mutation was untreated.

   ▸Evidence that argues against this (1)

   • Independence does not survive adjustment for FLT3-ITD, and the frequency does not survive the larger denominator. In 1421 patients across five consecutive COG/CCG and SWOG trials, NUP98/NSD1 was found in 7% of cytogenetically normal AML - not 16.1% - and in 15% of FLT3-ITD AML, with 82% of NUP98/NSD1 cases also carrying FLT3-ITD. Patients with both had a complete-remission rate of 27% versus 69% for FLT3-ITD without NUP98/NSD1 (p<.001) and 3-year overall survival of 31% versus 48% (p=.011). In multivariable analysis the DUAL lesion remained an independent predictor of poor outcome, while NUP98/NSD1 WITHOUT FLT3/ITD 'lost its prognostic significance'. The dismal prognosis this entry attributes to the fusion is, on the larger cohort, a property of the interaction with a co-mutation rather than of the fusion alone. Ostronoff F et al. 2014
5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   The retained NSD1 SET domain is a structurally characterised H3K36 methyltransferase with a bound SAM cofactor and an autoregulatory autoinhibitory mechanism (PDB 3OOI, 6KQP, 6KQQ). The fourth entry mapped to the NSD1 accession, 8C5L, is an NR2F6 ligand-binding-domain structure with a 19-residue NSD1 peptide and contains no SET domain.

   PDB entry 3OOI: Crystal Structure of Human Histone-Lysine N-methyltransferase NSD1 SET domain in Complex with S-adenosyl-L-methionine (primary citation: The structure of NSD1 reveals an autoregulatory mechanism underlying histone H3K36 methylation) · RCSB Protein Data Bank · RCSB PDB 2010
   ▸What this evidence cannot show

   Verified by fetching the RCSB entry records and by a UniProt cross-reference query on Q96L73 (4 entries: 3OOI, 6KQP, 6KQQ, 8C5L). Only THREE of those four are SET-domain structures; 8C5L is an NR2F6 ligand-binding domain with a 19-residue NSD1 peptide and was removed from the claim on 2026-09-19 (see the correction section). The three that remain cover the isolated SET domain of wild-type NSD1, not the fusion, and not the PWWP or PHD modules in context. The autoinhibitory loop is a feature of the isolated domain and its behaviour in the fusion is unknown.

   ▸Evidence that argues against this (1)

   • 8C5L is not an NSD1 SET-domain structure. Fetched from RCSB on 2026-09-19, it is 'NR2F6 ligand binding domain in complex with NSD1 peptide', X-ray at 2.6 A, with two polymer entities: a 568-residue maltose-binding-protein / NR2F6 nuclear-receptor ligand-binding-domain fusion (UniProt P0AEX9 and P10588) and a 19-residue NSD1 peptide (UniProt Q96L73). It contains no SET domain, no AWS or post-SET, no SAM and no methyltransferase activity. It entered the list only because a UniProt cross-reference query on Q96L73 returns four entries and this is one of them - the accession-mapping artefact. Worse for the entry's logic, the 19 residues it does contain are the NSD1 nuclear-receptor-interaction motif, which lies in the N-terminal region the FUSION DELETES. The claim has been corrected to 3OOI, 6KQP and 6KQQ. RCSB Protein Data Bank 2023
6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   The NUP98 C-terminal domain lost in the fusion is the autoproteolytic / NUP96-binding module that anchors NUP98 in the nuclear pore complex, so the fusion is released from the pore into the nucleoplasm.

   PDB entries 1KO6 (Crystal Structure of C-terminal Autoproteolytic Domain of Nucleoporin Nup98) and 2Q5X (Crystal Structure of the C-terminal domain of hNup98) · RCSB Protein Data Bank · RCSB PDB 2002
   ▸What this evidence cannot show

   Verified by fetching the RCSB entry records. The structures establish what the deleted domain is; the inference that losing it releases the fusion from the pore is a reasonable structural deduction supported by the observed nucleoplasmic/nuclear-aggregate localisation of the fusion [E4], not a direct localisation experiment in these papers.

   ▸Evidence that argues against this (1)

   • 'Released from the pore' is correct about the loss and misleading about what replaces it, and the difference is where the function lives. Xu et al. showed that five NUP98 fusions - NUP98-HOXA9, -HOXD13, -NSD1, -PHF23 and -TOP1 - physically interact with MLL1 and with the non-specific-lethal (NSL) histone-modifying complex; NUP98-HOXA9 and MLL1 co-localise on chromatin at Hox gene promoters; MLL1 is REQUIRED for the proliferation of NUP98-HOXA9 cells in vitro and in vivo; and inactivating MLL1 decreases expression of genes co-bound by the fusion and reverses the gene-expression signature of NUP98-rearranged human leukaemias. The fusion is therefore re-anchored into a defined multiprotein complex, not liberated into bulk nucleoplasm where 'only then can its IDR condense on chromatin'. This is also the single most consequential result in this node for the lab's menin-interface hypothesis: a fusion family containing no KMT2A at all is nonetheless dependent on MLL1, whose menin interface is the lab's target. Xu H et al. 2016
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In NUP98::NSD1 AML - a fusion that supplies its own catalytic H3K36 methyltransferase - the borrowed menin-MLL1 tether is still required: in human primary NUP98-rearranged AML cells and in immunodeficient mice transplanted with NUP98::NSD1 FLT3-ITD-positive patient-derived AML cells, the clinical menin inhibitor revumenib impaired proliferation and clonogenicity, drove myeloid differentiation, downregulated MEIS1 and CDK6, reduced both wild-type FLT3 and FLT3-ITD expression, and blocked leukaemic engraftment and leukaemia-associated death.

   The MLL-Menin Interaction is a Therapeutic Vulnerability in NUP98-rearranged AML. · Rasouli M, Blair H, Troester S, Szoltysek K, Cameron R, Ashtiani M, Krippner-Heidenreich A, Grebien F, McGeehan G, Zwaan CM, Heidenreich O · Hemasphere 2023
   PMID 37520776DOI
   ▸What this evidence cannot show

   Human primary cells and a patient-derived xenograft, which is the right system, but a single NUP98::NSD1 PDX and it is FLT3-ITD-positive - the co-mutation present in 74-91% of this genotype - so the fusion contribution and the kinase contribution are not separated here, and the paper itself reports synergy with a FLT3 inhibitor. The effect is differentiating and antiproliferative rather than apoptotic. No experiment in this paper tests whether the NSD1 SET activity is dispensable once menin is inhibited, which is the question the analogy actually raises.

   ▸Evidence that argues against this (1)

   • The other laboratory's NUP98::NSD1 model was the one that did NOT respond. In the founding menin-dependency study, 2 of 3 NUP98-rearranged patient-derived xenografts responded to VTP50469 and the refractory one was a NUP98-NSD1 model carrying WT1, ASXL1, IDH1, BCORL1 and FLT3-ITD; the same study found sensitivity to be partner-graded (IC50 84 nM for NUP98-KDM5A against 517 nM for NUP98-HOXA9) and abrogated outright by MEIS1 re-expression, and in NUP98-HOXD13 cells found no significant loss of Hoxa occupancy. So the placement of NUP98::NSD1 on the menin-sensitive side of the axis currently rests on two single-model results from two laboratories that disagree about the same co-mutation combination, and the graph records the dispute as hyp-nup98-menin-response-is-comutation-gated. Heikamp EB et al. 2022
8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   The discriminating variable on the menin axis is which MLL1 species carries an intact menin-binding motif, not whether the fusion owns a chromatin-modifying enzyme: NUP98::KMT2A, the one NUP98 fusion that supplies an MLL1 moiety of its own, 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 micromolar and to JQ1 at doses where KMT2A-AF9 cells arrest.

   Transforming activities of the NUP98-KMT2A fusion gene associated with myelodysplasia and acute myeloid leukemia. · Fisher JN, Thanasopoulou A, Juge S, Tzankov A, Bagger FO, Mendez MA, Peters AHFM, Schwaller J · Haematologica 2020
   PMID 31558671DOI
   ▸What this evidence cannot show

   Murine retroviral system with a single menin-inhibitor tool compound at micromolar concentrations, which is a weak pharmacological probe by current standards; resistance at 3-12 uM of MI-2-2 is not the same measurement as the nanomolar VTP50469 IC50 values reported for the sensitive NUP98 partners, so the comparison crosses compounds as well as fusions. This entry infers that the MBM deletion is the cause; the paper does not perform the MBM add-back that would prove it, which is exactly the experiment the graph proposes in exp-mbm-addback-and-trans-deletion-panel.

   ▸Evidence that argues against this (1)

   • An alternative reading of the same result that does not need the menin axis at all: this fusion is HOXA-low. If NUP98::KMT2A simply does not run a HOXA/MEIS1 programme, its menin-inhibitor resistance is the trivial consequence of not having the programme menin inhibition removes, and says nothing about the motif. The founding menin study makes the same point from the other side - the load-bearing node is MEIS1 rather than HOXA occupancy, since MEIS1 re-expression abrogates sensitivity while NUP98-HOXD13 cells lose no Hoxa occupancy - so resistance and HOXA-low status are confounded in this model and the motif explanation is not the only one on offer. Heikamp EB et al. 2022

• ← refutesNUP98 — domain architecture and residue coordinates (with the 1817/920 numbering trap resolved)

  Counter-evidence added cycle 41: the FG-repeat requirement is not FG-specific. Replacing the NUP98 FG repeats with an unrelated phase-separating IDR from FUS gave similar enhancement of genome-wide chimera binding and target-gene activation (Ahn 2021, PMID 34163069), and condensation alone is insufficient - the PHD finger targets the condensate to H3K4me3/2 genes while the FG repeats set its composition (Ahn 2025, PMID 39922194).

• ← related toNUP98 — domain architecture and residue coordinates (with the 1817/920 numbering trap resolved)

  Counter-evidence (InterPro P52948): the peptidase domain's N-terminal tolerance is ~17 residues, not the +/-8 stated, and the bar labelled 'NUP96' at 881-1817 is a polyprotein cleavage product rather than a domain -- the only annotated fold inside it is 1332-1623. None of this moves the NUP98::NSD1 breakpoint bracket at 486-531.

• ← part ofNUP98 — domain architecture and residue coordinates (with the 1817/920 numbering trap resolved)

  5-prime partner domain architecture; the FG-repeat region is retained and the peptidase domain lost

• ← supportsRepairing 85 defective citation blocks, and the finding that the queue's own numbers were wrong in three directions at once

  same repair: the overexpression-concentration caveat is cited rather than asserted

• ← related toNUP98 fusions share the menin dependency (confirmed preclinically) — and the partner-supplied reader/writer module is not why

  Status change: menin-MLL1 inhibition (VTP50469) prolongs survival in NUP98-NSD1 patient-derived xenografts and downregulates the HOXA cluster and Meis1 (PMID 34582559). This is confirmed preclinically, not predicted -- and no menin-inhibitor trial has enrolled by NUP98 genotype.

• ← related toMechanism class 1 — chromatin and elongation hijack (KMT2A-r, NUP98-r)

  Condensate-transformation dissociation (Jevtic 2022, PMID 35073946): SMARCA5 inhibition abolishes NUP98-NSD1 transformation while leaving the condensates intact; condensate formation is not sufficient for the transformed phenotype.

• ← related toMechanism class 4 — condensate-driven transcriptional rewiring (NUP98-r, EWSR1::FLI1)

  Necessity/sufficiency split (Jevtic 2022, PMID 35073946): SMARCA5 inhibition abolishes NUP98-NSD1 transformation while the condensates persist, so the material property can be present without the disease state.

• ← related toExperiment (proposed) — cross-family interface point-mutant panel: is the obligate interface real in each fusion?

  Puncta are the wrong readout for the NUP98 arm (Jevtic 2022, PMID 35073946): SMARCA5 inhibition destroys NUP98-NSD1 transformation while the condensates persist, so the panel needs a constituent-composition readout.

• related to →FLT3 in KMT2A-r leukemia — near-universal overexpression, but activating mutation in only ~3% of infants

  Sufficiency bound (Thanasopoulou 2014, PMID 24951466): NUP98-NSD1 alone produced NO leukaemia on transplantation; only co-expression with FLT3-ITD did, and latency tracked the FLT3-ITD ratio rather than fusion or HoxA expression.

• related to →The case that the KMT2A-fusion condensate is an artefact — what 1,6-hexanediol, overexpressed tags and in-vitro droplets actually prove

  Condensate-standard bound (Irgen-Gioro 2022, PMID 36444977): paraformaldehyde fixation can create droplet-like puncta in cells that have none when live, so imaged puncta plus a proteomic enrichment do not discriminate phase separation from a stoichiometric complex.

• related to →Nobody needs to enrol anyone: the PTH question is a regression on data that already exists

  Covariate correction (Ostronoff 2014, PMID 25145343): in 1421 patients NUP98/NSD1 was 7% of CN-AML, not 16.1%, and NUP98/NSD1 WITHOUT FLT3-ITD lost its prognostic significance in multivariable analysis.

• related to →Dataset: PDB and AlphaFold - the structural census, including the two absences that shape every plan

  Structural correction (RCSB 8C5L, fetched 2026-09-19): 8C5L is an NR2F6 ligand-binding domain with a 19-residue NSD1 peptide, no SET domain and no SAM; it maps to the N-terminal region the fusion DELETES. Entry corrected.

• related to →Hypothesis — the menin lesson generalises: every non-enzymatic fusion has an obligate partner interface that is more druggable than the fusion

  Re-anchoring, not release (Xu 2016, PMID 27889185): five NUP98 fusions physically interact with MLL1 and the NSL complex, co-occupy Hox promoters, and require MLL1 for proliferation in vitro and in vivo - a KMT2A-free fusion family dependent on MLL1.

• ← related toThe DOT1L-AF10 octamer - the second route to H3K79 methylation, required in MLL-AF10 and absent from KMT2A::AFF1

  AF10 inactivation impairs NUP98-NSD1 transformation and those cells are DOT1L-inhibitor sensitive

• ← bindsNSL complex - the MOF assembly that five NUP98 fusions bind alongside MLL1, and that competes with MLL/COMPASS for WDR5

  NUP98-NSD1 is one of five NUP98 fusions shown to interact physically with MLL1 and the NSL complex.

• depends on →Mechanism class 4 — condensate-driven transcriptional rewiring (NUP98-r, EWSR1::FLI1)
• ← testsNUP98 fusions share the menin dependency (confirmed preclinically) — and the partner-supplied reader/writer module is not why
• ← part ofNSD1 (nuclear receptor-binding SET domain protein 1) domain architecture

  reference domain coordinates for the NSD1 side of this fusion

• ← related toThe nuclear pore and NUP98 fusions — a fusion that takes a pore protein's disordered repeats off the pore and builds nuclear droplets with them

  FG-repeat condensates carry the NUP98-NSD1 interactome, but formation is not sufficient

• ← treated bySAVE (NCT05360160): the first prospective menin-inhibitor trial to report a NUP98-rearranged subgroup outcome

  NUP98r subgroup, 21% of 42, median CR/CRh 5.9 months

• ← treated byThe first published complete remission of NUP98-rearranged AML on a menin inhibitor (icovamenib/BMF-219, n=1)

  first published CR in NUP98-rearranged AML on a menin inhibitor

• ← regulatesWhy a nucleoporin fusion needs menin-MLL1: the second menin-dependent fusion family, and what it borrows rather than carries

  PDX survival benefit on menin-MLL1 inhibition

• ← related toNUP98 rearrangements are 7.2% of paediatric AML and 1.6% of adult, with 35% five-year overall survival — and the partner changes everything

  108 of 160 NUP98-r cases; 5-year OS 36%, EFS 17%, CR after course 1 38%

• ← related toNUP98::NSD1 without FLT3-ITD loses its prognostic significance — the confound that undermines the fusion's own reputation

  74-91% FLT3-ITD co-occurrence; without it the fusion loses prognostic significance

• ← part ofImport class 5: evidence borrowed from fusions that are not KMT2A fusions at all

  a menin-dependent non-KMT2A fusion

• ← related toH3K36me2/3 - the mark that physically switches PRC2 off, that LEDGF reads to reach chromatin, and that a comparator fusion uses to keep EZH2 off HOXA

  NUP98-NSD1 maintains H3K36 methylation at HoxA7/A9 and thereby prevents EZH2-mediated repression

• ← related toThe fusion-oncoprotein TurboID experiment this graph proposes has already been published - for PML::RARA, RUNX1::RUNX1T1, CBFB::MYH11, NUP98::HOXA9 and SET::NUP214, and not for any KMT2A fusion

  BioID has been applied to NUP98::HOXA9 and SET::NUP214, not to this NUP98 fusion

• ← analogous toZMYND11::MBTD1, t(10;17)(p15;q21) - a chromatin reader welded to the NuA4/TIP60 acetyltransferase, maintaining Hoxa/Meis1/Myb/Myc/Sox4 in AML: the KMT2A-fusion architecture built from different parts

  a 5' module supplying chromatin targeting fused to a partner supplying the effector, both maintaining HOXA; and NSD1 is the second protein the ZMYND11 MYND domain co-immunoprecipitates

• ← part ofNUP98 - the mobile nucleoporin that leaves the pore to put H3K4me3 on haematopoietic promoters, and the shared 5' donor of seven fusions in this graph

  contributes the 5' half: the FG/GLFG region and the off-pore chromatin-facing module

• ← part ofNSD1 - the H3K36 dimethyltransferase whose catalytic activity a NUP98 fusion cannot do without, and which cells can otherwise lose almost for free

  contributes the 3' half; its H3K36 methyltransferase activity is required - catalytic-dead NSD1 abolishes Hox-A activation and immortalisation

• ← related toNUP98 fusions retain a 469-residue FG repeat region that carries ZERO annotated modification sites - the cleanest retained-portion PTM absence in the four fusion families checked

  the retained NUP98 moiety carries no annotated modification site

• ← refutesWhy this one rare leukaemia was thought to decide an argument about a drug used in a commoner one - and why it decides less than the title says (corrected 2026-09-21)

  a KMT2A-free nucleoporin fusion that IS menin-MLL1-dependent, so 'contains no KMT2A' does not isolate menin

• ← refutesMenin has never been tested against SET::NUP214 - a bounded absence in three independent registers: literature, cell-line dependency data, and the founding paper's own blots

  the analogous experiment was run in the other nucleoporin-fusion family and came out menin-dependent, so this absence is not neutral

• ← related toTwo pools of one protein: the chromatin-bound XPO1 that matters here has never been separated biochemically from the soluble export machine

  loss of NUP98-fusion chromatin binding after leptomycin B is consistent with docking being disrupted AND with the whole nucleocytoplasmic economy being disrupted; the agent's measured in-human effect is generalised toxicity on every schedule tested

• analogous to →NUP98::KMT2A, inv(11)(p15q23) — the NUP98 fusion that contains KMT2A and is menin-inhibitor RESISTANT

  AXIS: WHICH MLL1 SPECIES CARRIES THE MENIN-BINDING MOTIF. Same 5-prime FG-repeat module, opposite standing on the menin axis. NUP98::NSD1 carries no KMT2A at all, borrows endogenous wild-type MLL1 in trans and is menin-inhibitor sensitive; NUP98::KMT2A supplies an MLL1 moiety with exon 1 deleted, removing the MBM from the very MLL1 species it provides, is HOXA-low and is resistant. Holding the FG module constant is what makes the pair informative. DISANALOGY: the resistant member is also HOXA-low, so absence of response and absence of the programme are confounded in it.

• analogous to →NUP98::KDM5A (NUP98-JARID1A), t(11;12)(p15;p13)

  AXIS: SHARED 5-PRIME PARTNER, OPPOSITE 3-PRIME MODULE CLASS. Identical NUP98 FG-repeat IDR; NSD1 contributes a WRITER whose SET catalysis is required for transformation, KDM5A contributes a READER whose catalytic demethylase domain is discarded at the breakpoint and whose H3K4me3-binding PHD finger is required instead. Same condensate physics, same HOXA output, and different diseases - monocytic cytogenetically normal AML against megakaryoblastic AML - so lineage is set by the 3-prime partner and not by the FG repeats.

• supports →Why a nucleoporin fusion needs menin-MLL1: the second menin-dependent fusion family, and what it borrows rather than carries

  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.

• ← analogous toTPR / Megator (P12270) - the nuclear-basket nucleoporin that builds heterochromatin exclusion zones, in the pan-MLL-fusion core

  nucleoporin biology in leukaemia - but TPR appears here as a partner of the fusion, not as a fusion component, so the analogy is structural and not mechanistic

• ← related toClec12a is REQUIRED for NUP98::NSD1 AML pathogenesis - a marker turning out to be a dependency, in a different fusion family, and untested in KMT2A-r

  the fusion family in which the dependency was demonstrated