NPM1 (nucleophosmin) — the nucleolar shuttle whose mutant form binds chromatin with KMT2A and buys the second menin-inhibitor indication
also called NPM1 · NPM · B23 · nucleophosmin · numatrin · NO38 · NPM1c · NPMc+
NPM1 normally lives in the nucleolus, the cell's ribosome factory, and shuttles between compartments carrying other proteins. In about a third of adult AML a tiny four-letter insertion moves it into the cytoplasm instead, and that displaced protein switches on the same HOXA/MEIS1 gene programme that KMT2A fusions switch on — which is why the same menin-blocking drug works in both diseases.
The same thing, for a scientist
NPM1 (P06748, 294 aa) is a pentameric nucleolar phosphoprotein and molecular chaperone that shuttles nucleolus-nucleoplasm-cytoplasm. A 4-bp insertion in exon 12 simultaneously destroys the C-terminal nucleolar-localisation tryptophans and creates a new leucine-rich XPO1/CRM1 nuclear export signal, producing cytoplasmically mislocalised NPM1c [E1][E2]. NPM1c is not merely absent from the nucleus: a fraction binds chromatin directly at HOXA/MEIS1 loci co-occupied by KMT2A (MLL1), and targeted NPM1c degradation collapses expression, Pol II occupancy and activating marks at those targets within hours [E3]. That is the mechanistic basis of the menin dependency: NPM1-mutant AML is exceptionally dependent on the menin-binding site of MLL1 by CRISPR domain scan [E4], menin-MLL1 inhibition reverses preleukaemic self-renewal in Npm1c/Dnmt3a knock-in mice [E5], and revumenib produced CR+CRh in 23.4% of R/R NPM1-mutant AML in the phase 2 AUGMENT-101 arm [E6]. This beat's own lineage-split DepMap run adds a number the graph did not have: **NPM1 itself is not a KMT2A-selective dependency at all** (z +0.05 in KMT2A-rearranged myeloid lines, +0.19 in the single B-lineage line), while **OCI-AML3, the NPM1c line, is the most NPM1-dependent leukaemia line in the set at z -3.16** [E7]. NPM1c is therefore a self-dependency of its own genotype, not a shared KMT2A-axis node.
This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
Why a protein leaving the nucleolus turns on HOX genes
The obvious model — NPM1c is a loss of nuclear NPM1 function — was the field's working assumption for fifteen years and it is now known to be incomplete. Uckelmann and colleagues degraded NPM1c acutely and found that its chromatin targets lose transcription, RNA polymerase II and activating histone modifications, which is a direct gain of function at chromatin, not a passive consequence of the wild-type protein being absent 3. Those targets are co-occupied by KMT2A.
That places NPM1c and the KMT2A fusion at the same loci by two different routes: the fusion arrives through menin-LEDGF tethering of the KMT2A N-terminus, NPM1c arrives by whatever chromatin contact its mutant C-terminus makes. Both then require the menin-MLL1 interaction to sustain output 34. It is the cleanest published reason why one drug has two genotype indications.
The wild-type biology that is being lost, and why it matters for toxicity
UniProt places NPM1 in the nucleolus, the nucleoplasm and the centrosome, with two nuclear localisation signals at aa 152-157 and 191-197 8. Its day job is ribosome biogenesis and histone chaperoning in the nucleolar granular component. The graph already holds the domain architecture as protein-npm1-domains; this entry does not repeat those coordinates.
The therapeutic relevance is that NPM1c disease is also a ribosome-biogenesis lesion. Proteomics of preleukaemic Npm1c knock-in HSPCs and of human NPM1-mutant AML shows ribosome-biogenesis factors depleted, and the cells are correspondingly more sensitive to RNA polymerase I inhibition; TSR3 knockout preferentially kills NPM1c AML via p53 9. This is a second, menin-independent vulnerability of the same genotype and it is an argument against treating NPM1c AML as simply "KMT2A-like".
Cell type, cell state, localisation, epigenetics
- Cell type. NPM1 mutation is essentially restricted to myeloid disease. Falini's founding survey found cytoplasmic NPM in 208/591 primary AML and in none of 135 secondary AML or 980 other haematopoietic and non-haematopoietic neoplasms 1. NPM1-mutant AML is rare in infants and children relative to adults, which matters for a paediatric-focused lab.
- Cell state. CD34- and CD133-negative, morphologically broad, normal karyotype, frequently FLT3-ITD co-mutant, chemo-responsive at induction 1. In mice the premalignant state is an expanded self-renewing myeloid progenitor compartment that menin inhibition reverses 5.
- Subcellular localisation. This is the defining feature: wild-type nucleolar/nucleoplasmic/centrosomal 8; mutant cytoplasmic by immunohistochemistry, the diagnostic readout itself 12.
- Subnuclear localisation. Wild-type: nucleolar granular component 8. Mutant: nuclear condensates termed C-bodies, which recruit NUP98 and KMT2A 10 — see the caveat on condensate assays below.
- Epigenetic readout. Loss of activating histone modifications and Pol II at NPM1c targets on degradation of NPM1c 3; H3K79me2 and menin-MLL1-dependent HOXA/MEIS1 expression 4.
The condensate claim, stated at the strength the assays support
Datar and colleagues report that NPM1c forms nuclear condensates in cell lines, mouse models and primary patient samples; that phase separation is necessary and sufficient to recruit NUP98 and KMT2A into them; that destabilising phase separation impairs gene expression, leukaemic expansion and the undifferentiated state; and that nucleoporin and KMT2A fusion condensates are biophysically indistinguishable from NPM1c condensates 10.
This project holds a standing methodological rule that 1,6-hexanediol sensitivity and classical FRAP cannot distinguish phase separation from binding to spatially clustered sites, and that hexanediol condenses rather than melts chromatin 11. The Datar claim is stronger than the usual one — it rests on extensive mutagenesis of the phase-separation determinants plus pharmacological destabilisation, not on hexanediol alone — but the abstract does not state that calibrated half-bleach internal-mixing measurements were made, and this beat read the abstract, not the figures. What is safe to assert is that NPM1c forms a concentrated nuclear body that recruits KMT2A and NUP98 and whose integrity is required for the leukaemic programme. Whether the organising principle is liquid-liquid phase separation or high-avidity binding to clustered sites is not settled by what this beat read, and the two models make different predictions about whether the targeting is sequence-specific.
What the dependency data say, split by lineage
Reported as two numbers because the KMT2A-rearranged panel is myeloid-enriched and a myeloid-identity gene scores as KMT2A-selective for free.
| arm | n lines | mean gene effect | z vs other leukaemia |
|---|---|---|---|
| KMT2A-rearranged myeloid | 3 | -1.071 | +0.05 |
| KMT2A-rearranged B-lineage | 1 | -1.005 | +0.19 |
| other leukaemia | 7 | -1.093 | 0 (reference) |
| OCI-AML3 (NPM1c) | 1 | -2.590 | -3.16 |
NPM1 is broadly required (pan-essential mean -0.919 across 391 lines) and shows no KMT2A selectivity whatever. The one striking number is OCI-AML3 7. OCI-AML3 carries NPM1 mutation type A and shows the cytoplasmic NPM immunophenotype 2, so the simplest reading is that NPM1c cells are dependent on the NPM1 locus itself — consistent with NPM1c being a direct chromatin-acting oncoprotein 3 rather than a loss of function.
Caveat that limits this hard: Open Targets returned only 12 leukaemia screens for NPM1 against 57 for MEN1, so the B-lineage arm is a single line and every arm is small. This is a lead, not a measurement.
Bearing on this project's menin-interface hypothesis
It cuts for the hypothesis on mechanism and against any assumption that the KMT2A-rearranged panel is where menin dependence is strongest.
In this beat's own lineage-split run, MEN1 is the single most lineage-independent dependency in the panel (z -2.29 myeloid, -2.28 B-lineage, gap -0.01) — and OCI-AML3, an NPM1c line with no KMT2A rearrangement, scores z -2.80, deeper than either KMT2A-rearranged half 7. That reproduces what ana-menin-programme-vs-rearrangement found by a different route. The menin interface is real and it is not a lineage artefact; but the genotype that needs it most in this cell panel is NPM1c, not KMT2A-rearrangement.
| uniprot | P06748 |
|---|---|
| gene symbol | NPM1 |
| protein name | Nucleophosmin |
| length AA | 294 |
| alphafold ID | AF-P06748-F1 |
| ensembl | ENSG00000181163 |
| accession resolution | UniProt REST gene_exact:NPM1 AND organism_id:9606 AND reviewed:true, run 2026-09-19; exactly one reviewed hit; primary gene name read back as NPM1, UniProtKB id NPM_HUMAN, length 294 confirmed from the same fetched record. Ensembl id taken from the cached HGNC complete set and checked against the Open Targets approvedSymbol, which read back NPM1. |
| domains note | Domain coordinates are held by protein-npm1-domains and are deliberately not duplicated here. UniProt annotates two nuclear localisation signals (aa 152-157, 191-197) and no folded domain. |
| classification | NOT a documented physical interactor of KMT2A::AFF1. The mutant form NPM1c co-occupies chromatin with KMT2A (MLL1) and shares the menin dependency; the relationship is target-level convergence, not a protein-protein interface with the fusion. |
| interaction evidence class | ChIP/degron co-occupancy with KMT2A; CRISPR domain screen; genetic mouse model; phase-2 clinical trial; unbiased DepMap scan |
| shown for fusion | none — NPM1c is a separate driver genotype, not a KMT2A fusion partner. NPM1 is separately the 5' partner of fusion-npm1-alk. |
| epigenetic readout | Degradation of NPM1c causes loss of activating histone modifications and of RNA polymerase II at its chromatin targets; HOXA/MEIS1 output is menin-MLL1- and DOT1L-dependent |
| cell type | myeloid — adult AML with normal karyotype; absent from secondary AML and from 980 other neoplasms in the founding survey; uncommon in infant disease |
| cell state | CD34-negative, CD133-negative, frequently FLT3-ITD co-mutant, chemo-responsive at induction; preleukaemic state is an expanded self-renewing myeloid progenitor pool |
aliases retired or alternate
[ "NPM", "B23", "numatrin", "NO38" ]
domains
[]
localization
{
"subcellular_wildtype": "nucleus (nucleolus and nucleoplasm) and centrosome",
"subcellular_mutant": "cytoplasm — the diagnostic hallmark of NPMc+ AML",
"subnuclear_wildtype": "nucleolar granular component",
"subnuclear_mutant": "nuclear condensates ('C-bodies') that recruit NUP98 and KMT2A; the organising principle (phase separation vs clustered binding) is not settled by what this beat read"
}essentiality
{
"depmap_scan": "analyses/depmap-interactome5-lineage-split",
"pan_essentiality_all_lines": -0.919,
"n_all_lines": 391,
"n_leukaemia_screens_returned": 12,
"mean_kmt2a_rearranged_myeloid": -1.071,
"n_kmt2a_myeloid": 3,
"z_kmt2a_myeloid": 0.05,
"mean_kmt2a_rearranged_b_lineage": -1.005,
"n_kmt2a_b_lineage": 1,
"z_kmt2a_b_lineage": 0.19,
"mean_other_leukaemia": -1.093,
"sd_other_leukaemia": 0.473,
"n_other_leukaemia": 7,
"gene_effect_oci_aml3_npm1c": -2.59,
"z_oci_aml3_npm1c": -3.16,
"lineage_split_reported": true,
"interpretation": "no KMT2A selectivity in either lineage half; a single extreme value in the NPM1c line, most plausibly oncogene addiction to the mutant allele rather than a wild-type requirement"
}enumeration record counts — the numbers behind the chart
[
{
"query": "NPM1 AND (menin OR MEN1) AND (leukemia OR leukaemia)",
"records": 144,
"date": "2026-09-19"
},
{
"query": "(NPM1 mutant OR NPM1c OR NPM1-mutated) AND (menin OR MEN1) AND (HOXA OR HOX)",
"records": 28,
"date": "2026-09-19"
},
{
"query": "NPM1 AND (nuclear export OR XPO1 OR CRM1) AND (mutant OR cytoplasmic) AND leukemia",
"records": 70,
"date": "2026-09-19"
}
]trials
[ "NCT04065399" ]
- Resistance: TP53 inactivation is a documented revumenib-resistance mechanism, and the leading menin-independent vulnerability in this genotype (TSR3 loss, Pol I inhibition) is itself p53-dependent — so the backup axis fails in the same patients as the primary one.
- Essential-function toxicity: NPM1 is broadly required (pan-essentiality -0.92 across 391 lines) and is central to ribosome biogenesis. Direct NPM1-directed therapy has no obvious window; the tractable handle is the menin interface, not the protein itself.
- Compensation: menin inhibition in NPM1c achieved CR+CRh in 23.4% with a 4.7-month median duration. Whatever compensates is fast and common, and the published mechanism is that menin-MLL1 inhibition alone is inferior to menin plus DOT1L inhibition in this genotype.
- Indication drift: a paediatric-focused programme that leans on the NPM1c indication is leaning on adult data — median age 63, one patient under 18 in the AUGMENT-101 NPM1 arm.
- If the C-body model is right, dissolving condensates is the intervention, and a reagent that dissolves them non-specifically (hexanediol and its relatives) also condenses bulk chromatin. The obvious pharmacology for this mechanism is the pharmacology this project's own methods rule says is uninterpretable.
- Is OCI-AML3's extreme NPM1 dependency a requirement for wild-type NPM1 or addiction to the mutant allele? An allele-specific CRISPR or a degron restricted to NPM1c would separate them, and the answer changes whether NPM1 is a target or only a marker.
- NPM1c recruits KMT2A into condensates and KMT2A fusion proteins form condensates said to be biophysically indistinguishable. Does a KMT2A::AFF1 condensate recruit wild-type NPM1, and does menin inhibition dissolve it or merely evict the fusion from chromatin?
- Why is an NPM1c line more menin-dependent than the KMT2A-rearranged lines in the same panel? Either the KMT2A panel is not as menin-dependent as the field assumes, or the NPM1c route to the menin-MLL1 complex is more obligate. This graph has now seen it twice by two methods and has no mechanistic explanation.
- NPM1-mutant AML is overwhelmingly an adult disease. What is the paediatric NPM1c fraction, and is the menin dependency the same in a fetal-origin cell of origin?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Cytoplasmic NPM was found in 208 of 591 primary AML specimens (35.2%) and in none of 135 secondary AML or 980 other haematopoietic and non-haematopoietic neoplasms; it associated with normal karyotype, CD34/CD133 negativity, high FLT3-ITD frequency and responsiveness to induction chemotherapy, and the associated exon-12 NPM mutations caused cytoplasmic localisation when transfected.
Cytoplasmic nucleophosmin in acute myelogenous leukemia with a normal karyotype · Falini B, Mecucci C, Tiacci E, Alcalay M, et al. · N Engl J Med 352(3):254-66 2005What this evidence cannot show
Abstract fetched from PubMed and read 2026-09-19; full text not read. Adult cohort, immunohistochemistry as the primary assay. The 35.2% figure is adult AML and should not be carried into paediatric or infant disease, where NPM1 mutation is far less common.
Evidence that argues against this (1)
- The same paper reports NPM1c AML as chemo-RESPONSIVE at induction. A genotype with good chemotherapy response is a weaker argument for a new targeted agent than a KMT2A rearrangement is, and the menin-inhibitor indication in this genotype is relapsed/refractory precisely for that reason. Arellano ML et al. 2025
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The OCI-AML3 cell line carries a TCTG duplication at NPM exon 12 corresponding to mutation type A, the commonest mutation in primary NPMc+ AML, and displays aberrant cytoplasmic NPM expression, CD34 negativity and macrophage markers; it was identified by screening 79 myeloid cell lines and engrafts NOD/SCID mice retaining cytoplasmic NPM.
Cell line OCI/AML3 bears exon-12 NPM gene mutation-A and cytoplasmic expression of nucleophosmin · Quentmeier H, Martelli MP, Dirks WG, Bolli N, et al. · Leukemia 19(10):1760-7 2005What this evidence cannot show
Abstract fetched and read 2026-09-19. This citation establishes the GENOTYPE of OCI-AML3, which is the load-bearing fact for every DepMap statement in this entry and in three of the coordinator's analyses. It says nothing about dependency.
Evidence that argues against this (1)
- OCI-AML3 is a single cell line and it is also DNMT3A R882-mutant and NRAS-mutant. Any OCI-AML3-specific dependency attributed to NPM1c is confounded by the rest of its genotype, and this entry's z -3.16 rests on that one line. Uckelmann HJ et al. 2020
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
NPM1c binds directly to specific chromatin targets that are co-occupied by KMT2A (MLL1); targeted degradation of NPM1c causes a rapid decrease in gene expression, loss of RNA polymerase II and loss of activating histone modifications at those targets, defining the mechanism by which menin-MLL1 inhibitors produce clinical responses in NPM1-mutated AML.
Mutant NPM1 Directly Regulates Oncogenic Transcription in Acute Myeloid Leukemia · Uckelmann HJ, Haarer EL, Takeda R, Wong EM, et al. · Cancer Discov 13(3):746-765 2023What this evidence cannot show
Abstract fetched and read 2026-09-19; full text and figures not read. PubMed's publication-type field for this record lists 'Editorial' and 'Comment' alongside the research-support types, which is a metadata error — the pagination (746-765) and content are those of a primary research article, and the record itself points to a companion article by Wang et al. at p. 724. Cited here as a paper. Cell-line and degron based; the degron system forces acute loss, which is not how the protein behaves in a patient.
Evidence that argues against this (1)
- A companion paper published back to back in the same issue reached the same conclusion from a different group, which is corroboration rather than refutation, but it means the finding rests on two papers from a closely connected field rather than on independent replication years apart. This beat read neither full text, and the specific claim that NPM1c binds chromatin DIRECTLY (rather than being recruited by a partner) is the part most in need of independent structural support. Datar GK et al. 2025
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A CRISPR/Cas9 genome-editing domain screen showed NPM1-mutant AML to be exceptionally dependent on the menin-binding site in MLL1; MLL1 and DOT1L control HOX, MEIS1 and FLT3 expression and differentiation in NPM1-mutant AML, and small-molecule menin-MLL1 inhibition had profound antileukaemic activity in human and murine NPM1-mutant models.
Targeting Chromatin Regulators Inhibits Leukemogenic Gene Expression in NPM1 Mutant Leukemia · Kühn MW, Song E, Feng Z, Sinha A, et al. · Cancer Discov 6(10):1166-1181 2016What this evidence cannot show
Abstract fetched and read 2026-09-19. The domain screen is a functional mapping in cell lines, not a structural demonstration; 'exceptionally dependent' is the authors' wording and the comparator set is not stated in the abstract.
Evidence that argues against this (1)
- The same paper reports that menin-MLL1 inhibition alone was inferior to the combination with DOT1L inhibition for suppressing HOX/FLT3 and inducing differentiation. That is an argument that the menin interface is not sufficient as a single target in this genotype - and pinometostat, the DOT1L arm of that combination, has not succeeded clinically as monotherapy. Arellano ML et al. 2025
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In Npm1c/Dnmt3a mutant knock-in mice, leukaemia is preceded by extended myeloid progenitor proliferation and self-renewal, and that self-renewal is reversed by oral administration of the menin-MLL1 inhibitor VTP-50469, supporting targeted epigenetic therapy in a preventative setting.
Therapeutic targeting of preleukemia cells in a mouse model of NPM1 mutant acute myeloid leukemia · Uckelmann HJ, Kim SM, Wong EM, Hatton C, et al. · Science 367(6477):586-590 2020What this evidence cannot show
Abstract fetched and read 2026-09-19. Mouse knock-in requiring a second cooperating Dnmt3a lesion; 'preventative' is a mouse-model claim and no human preventative trial exists. VTP-50469 is the tool compound whose clinical descendant is revumenib.
Evidence that argues against this (1)
- Preventative treatment of clonal haematopoiesis with a menin inhibitor would expose people who will mostly never develop AML to a drug whose class causes differentiation syndrome and QTc prolongation. The mouse result does not address that trade-off, and the indication that actually exists is relapsed/refractory disease. Arellano ML et al. 2025
- E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In the phase 2 NPM1-mutant arm of AUGMENT-101 (NCT04065399), 84 R/R NPM1-mutant AML patients received revumenib; in the 64-patient efficacy-evaluable population the CR+CRh rate was 23.4% (1-sided P = .0014), ORR 46.9%, median duration of CR+CRh 4.7 months, with 5 of 30 responders proceeding to transplant and treatment-related adverse events causing discontinuation in 4 patients (4.8%).
Menin inhibition with revumenib for NPM1-mutated relapsed or refractory acute myeloid leukemia: the AUGMENT-101 study · Arellano ML, Thirman MJ, DiPersio JF, Heiblig M, et al. · Blood 146(9):1065-1077 2025What this evidence cannot show
Abstract fetched and read 2026-09-19. Median age 63 and only one patient under 18 — this is an adult data set being used in a paediatric-focused graph. Single-arm, no comparator. 75% had previous venetoclax.
Evidence that argues against this (1)
- A 23.4% CR+CRh with a 4.7-month median duration is a modest result, and this arm is the basis of the entry's framing of NPM1c as 'the second indication'. The honest reading is that menin inhibition in NPM1c is active but far from curative as monotherapy, which is why the current trials are combinations. Damaskou A et al. 2025
- E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
In this beat's lineage-split DepMap run, NPM1 shows no KMT2A selectivity (KMT2A-rearranged myeloid mean -1.071, z +0.05; KMT2A-rearranged B-lineage n=1, mean -1.005, z +0.19; other-leukaemia arm mean -1.093, SD 0.473, n=7), while OCI-AML3 (NPM1c) has gene effect -2.590, z -3.16. Pan-essentiality across the 391 lines Open Targets returns is -0.919. In the same run MEN1 scores z -2.29 (myeloid) and -2.28 (B-lineage) with OCI-AML3 at z -2.80.
Interactome round 5 panel, lineage-split DepMap analysis (Chronos gene effect via Open Targets GraphQL) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-interactome5-lineage-split 2026What this evidence cannot show
Run 2026-09-19. Open Targets returned only 12 leukaemia screens for NPM1 (against 57 for MEN1), so the B-lineage arm is ONE line (RS4;11) and no z from it should be treated as a measurement. The NPM1 z-scores were computed outside the main script, which rejected NPM1 for having too few lines; the arithmetic was done from the twelve gene-effect values read out of the same API response. Lineage assignment is by label. Ensembl ENSG00000181163 was checked against the API's approvedSymbol, which read back NPM1.
Evidence that argues against this (1)
- OCI-AML3's extreme NPM1 score could equally reflect that the cell line is heterozygous NPM1c and that CRISPR cutting the locus removes the mutant allele - i.e. it may be a readout of NPM1c oncogene addiction rather than of wild-type NPM1 requirement. Chronos cannot distinguish those, and a guide RNA set targeting NPM1 hits both alleles. The published NPM1c degron work makes the oncogene-addiction reading the more likely one. Uckelmann HJ et al. 2023
- E8established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB P06748 (NPM_HUMAN, Nucleophosmin) is 294 aa with nuclear localisation signals at aa 152-157 and 191-197 and is annotated to the nucleolus, the nucleoplasm and the centrosome.
UniProtKB P06748 (NPM_HUMAN), Nucleophosmin · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026What this evidence cannot show
Record fetched 2026-09-19 via gene_exact:NPM1 AND organism_id:9606 AND reviewed:true, one reviewed hit; primary gene name read back as NPM1 and length 294 confirmed from the same fetched record. UniProt annotates no folded domain for NPM1 — the pentameric core is a structural, not a curated-domain, feature, and its coordinates live in protein-npm1-domains.
Evidence that argues against this (1)
- UniProt's subcellular annotation is for the WILD-TYPE protein and carries no annotation of the mutant's cytoplasmic localisation, so a reader taking localisation from the database alone would get the diagnostically important fact exactly backwards for a third of AML. Falini B et al. 2005
- E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
NPM1c mutations disrupt a nucleolar localisation signal and create a novel nuclear export signal; proteomics of conditional Npm1cA/+ knock-in murine HSPCs and of human NPM1-mutant AML shows ribosome-biogenesis proteins significantly depleted, preleukaemic Npm1cA/+ HSPCs are more sensitive to RNA polymerase I inhibitors including actinomycin D, and knockout of the 40S maturation factor TSR3 preferentially inhibits NPM1c+ AML proliferation through a p53-dependent apoptotic response.
Posttranscriptional depletion of ribosome biogenesis factors engenders therapeutic vulnerabilities in NPM1-mutant AML · Damaskou A, Wilson R, Gozdecka M, Giotopoulos G, et al. · Blood 146(10):1239-1252 2025What this evidence cannot show
Abstract fetched and read 2026-09-19. Mouse knock-in plus human AML proteomics; actinomycin D is not a selective Pol I inhibitor in the clinic and the TSR3 effect is p53-dependent, which restricts it to TP53-wild-type disease.
Evidence that argues against this (1)
- A p53-dependent vulnerability is exactly the kind that disappears in the relapsed/refractory setting where menin inhibitors are used, and TP53 inactivation is already a published resistance mechanism for revumenib. So this alternative axis may not be available in the patients who most need an alternative. Arellano ML et al. 2025
- E10moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
NPM1c forms nuclear condensates in human cell lines, mouse models and primary patient samples; NPM1c phase separation is reported to be necessary and sufficient to recruit NUP98 and KMT2A to those condensates; destabilising phase separation by mutagenesis or pharmacologically impairs gene regulation, in vivo leukaemic expansion and maintenance of the undifferentiated state; and nucleoporin and KMT2A fusion proteins form condensates the authors describe as biophysically indistinguishable from NPM1c condensates. The authors name these bodies 'coordinating bodies' (C-bodies).
Disparate leukemia mutations converge on nuclear phase-separated condensates · Datar GK, Khabusheva E, Anand A, Beale J, et al. · Cell 188(25):7118-7136.e21 2025What this evidence cannot show
Abstract fetched and read 2026-09-19; full text and figures NOT read. Confidence is held at moderate rather than strong for one reason only: the abstract does not state which discriminating assay was used, and this project holds that the two commonest condensate assays cannot distinguish phase separation from binding to spatially clustered sites. The mutagenesis-plus-destabilisation design is better than the usual evidence, and a reader of the full text may well be able to raise this to strong.
Evidence that argues against this (1)
- 1,6-hexanediol sensitivity and classical FRAP recovery both fail to distinguish liquid-liquid phase separation from binding to spatially clustered sites; only model-free calibrated half-bleaching that measures internal mixing discriminates them. Any condensate claim resting on the two common assays supports 'concentrated body' and not 'phase-separated droplet'. Muzzopappa F et al. 2022
- E11established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
1,6-hexanediol rapidly immobilises and condenses chromatin in living human cells, an action the authors state is distinct from melting a droplet and which they instruct readers to use in reinterpreting chromatin-associated droplet results.
1,6-Hexanediol rapidly immobilizes and condenses chromatin in living human cells · Itoh Y, Iida S, Tamura S, Nagashima R, et al. · Life Sci Alliance 4(4):e202001005 2021What this evidence cannot show
Abstract fetched and read 2026-09-19. Not a leukaemia paper; this is a reagent-behaviour result carried into this entry because the NPM1c condensate literature is chromatin-associated. It does not say the Datar result is wrong — it says one class of evidence for such results is uninterpretable.
Evidence that argues against this (1)
- This is a reason for caution, not a refutation, and it can be over-applied. 'Binding to spatially clustered sites' is a substantive rival model and in this system arguably the more interesting one, because clustered binding predicts sequence-specific targeting of NPM1c to HOXA loci while bulk phase separation does not - and the degron work shows the targeting IS specific. Uckelmann HJ et al. 2023
the exon-12 frameshift mutant of this gene
NPM1c binds chromatin targets co-occupied by KMT2A and confers menin dependence, which is the basis of the second approved menin-inhibitor indication.
NPM1c binds directly to specific chromatin targets co-occupied by KMT2A (MLL1).
The NPM1c line OCI-AML3 is the programme arm of that analysis and is MORE menin-dependent than the KMT2A-rearranged arm.
Domain architecture of the same protein, including the nucleolar localisation signal disrupted by NPM1c.
Revumenib carries a separate approval for relapsed or refractory NPM1-mutant AML.
Ziftomenib is approved for adults with relapsed or refractory NPM1-mutant AML.
- related to →The nucleolus — the one compartment with direct published images of a KMT2A fusion inside it
Wild-type NPM1 is a nucleolar shuttle; NPM1c mutations disrupt the nucleolar localisation signal and create a nuclear export signal.
NPM1c forms nuclear condensates, and the entry's hexanediol/FRAP caveat applies to that claim.
NPM1 itself shows no KMT2A selectivity on a lineage-split DepMap run; the dependency is conferred by the mutant, not the gene.
THE INTERNAL STANDARD: a KMT2A-free, CRM1-addressed, HOXA-driving lesion that IS menin-inhibitor-sensitive.
NPM1 as the bait
NPM1-mutant AML has no fusion and no menin-binding motif and is menin-inhibitor responsive
chromatin-bound CRM1 recruits NPM1c to the same HOX clusters, and NPM1c AML is menin-inhibitor sensitive - so the CRM1 address is not a menin-free address