NUP98::HOXD13, t(2;11)(q31;p15) — the myelodysplasia fusion, and a warning against targeting p53
also called NUP98-HOXD13 · NHD13 · t(2;11)(q31;p15) · NUP98/HOXD13
A rare NUP98 fusion found after chemotherapy for another cancer and also in infants. Mice carrying it develop a severe pre-leukaemic bone-marrow failure that always turns into leukaemia — and the obvious idea of blocking p53 to relieve the failure makes it worse.
The same thing, for a scientist
t(2;11)(q31;p15) fuses the **GLFG repeats of NUP98** to the **HOXD13 homeodomain** (P35453, 343 aa, homeodomain **276–335**) [E1][E6]. First described in **therapy-related AML** in a paediatric patient [E1], then in **de novo infant acute myelomonocytic leukaemia** — the report that first linked NUP98 rearrangement to **infant** leukaemia as well as to therapy-related disease, and only the third t(2;11) case at the time [E2]. The human spectrum spans MDS, AML and **CML blast crisis** [E4]. Its lasting contribution is the **NHD13 transgenic mouse**, the canonical MDS model: *"a highly penetrant, severe myelodysplastic syndrome that progresses to acute leukemia"*, **uniformly fatal within 14 months** [E3]. Two independent results show the fusion is **not sufficient** and that the obvious therapeutic inference is wrong. Retroviral insertional mutagenesis identified collaborating loci — ***Meis1, Mn1, Gata2, Erg, Epor*** and the **miR-29a/miR-29b1** locus [E5]. And attempting to rescue the cytopenias by inhibiting p53 **backfires**: pifithrin-α gave only partial, transient rescue with **no improvement in anaemia**, while genetic loss of one or both *p53* alleles **exacerbated the MDS phenotype and accelerated AML** [E4]. Within the menin story, NUP98-HOXD13 is the fusion in which menin-MLL1 inhibition caused **no significant loss of Hoxa occupancy or expression** — the dependency ran through *Meis1* instead [E7].
Nomenclature: two strings are in the verified literature
The founding paper and the infant-leukaemia report both use t(2;11)(q31;p15) 12. The 1999 PRRX1 paper refers to the same lesion as t(2;11)(q35;p15). Both strings exist in the verified literature; prefer q31 and record q35 as a variant so that a search on either retrieves the entry.
Separately, note that PubMed's query parser breaks on translocation parentheses — t(2;11)(q31;p15) is tokenised into unusable fragments and returns nothing. Write such queries as t 2 11 q31.
The p53 result, which is the most actionable thing here
NHD13 mice die of cytopenias as well as of leukaemia, and the cytopenias are p53-dependent apoptosis of dysplastic marrow. The therapeutic idea writes itself: block p53, save the marrow.
It does not work and it is dangerous 4. Pharmacological inhibition with pifithrin-α gave partial, transient rescue with no improvement in anaemia — the cytopenia that actually kills — and genetic loss of one or both p53 alleles exacerbated the MDS phenotype and accelerated progression to AML.
The general lesson for this graph: in a pre-leukaemic state driven by a fusion, the apoptosis that produces the cytopenia is also the tumour suppression. Relieving the symptom removes the brake. This is a directly transferable caution for any supportive-care strategy in fusion-driven MDS, and it is the kind of result that does not appear in a review's summary table.
Where it sits in the menin story
NUP98-HOXD13 binds Hoxb clusters that MLL-AF9 does not, and Mll1 deletion downregulated Hoxa, Hoxb and Meis1 in this fusion's cells, with a single Hoxb gene sufficient to rescue the clonogenic defect 8.
But in the menin-inhibitor experiment, NUP98-HOXD13 cells showed no significant loss of occupancy at the Hoxa cluster and no change in Hoxa expression — the drug worked through Meis1 7. So for this fusion the HOXA programme is not the node the drug acts on, which is one of the clearest arguments in the family that 'NUP98 fusions are HOXA-driven and therefore menin-dependent' is too loose a statement (mechanism-nup98-fusion-menin-mll1-dependency).
| gene5 | NUP98 |
|---|---|
| gene3 | HOXD13 |
| cytoband | t(2;11)(q31;p15) |
| cytoband variant in literature | t(2;11)(q35;p15) |
| partner class | homeodomain |
| uniprot5 | P52948 |
| uniprot3 | P35453 |
| length AA 5 | 1817 |
| length AA 3 | 343 |
| mouse model | NHD13 transgenic: highly penetrant severe MDS progressing to acute leukaemia, uniformly fatal within 14 months |
| sufficiency | NOT sufficient; collaborating loci Meis1, Mn1, Gata2, Erg, Epor, miR-29a/miR-29b1 |
| p53 warning | pifithrin-alpha gave only partial transient rescue with no improvement in anaemia; genetic loss of one or both p53 alleles exacerbated MDS and accelerated AML |
| menin inhibition hoxa response | no significant loss of Hoxa occupancy or expression; dependency runs through Meis1 |
| prognosis | no cohort outcome data; too few cases |
diseases
[ "therapy-related AML", "de novo infant acute myelomonocytic leukaemia", "MDS", "CML blast crisis" ]
breakpoints
[
{
"id": "bp-nup98-hoxd13-t2-11",
"domains_retained_5": [
"FG repeats 1 (1-156)",
"GLEBS (157-213)",
"FG repeats 2 (214-480)"
],
"domains_lost_5": [
"peptidase S59 (738-880)",
"Nup96 chain (881-1817)"
],
"domains_retained_3": [
"Homeodomain"
],
"domains_lost_3": [
"N-terminal HOXD13 sequence upstream of the breakpoint"
],
"phenotype_notes": "Drives a myelodysplastic rather than an acute phenotype in mice; requires a collaborating event to transform."
}
]- Blocking p53 to relieve the cytopenias of fusion-driven MDS accelerates progression to acute leukaemia in this model; the supportive-care intervention and the oncogenic brake are the same thing.
- The fusion needs a second hit, and Meis1 is on the collaborating list and is also what abrogates menin-inhibitor sensitivity - so the natural evolution of this disease may select for drug resistance before any drug is given.
- An MDS-phenotype fusion is a slow disease, which means any therapy is given for a long time; that puts this fusion at the sharp end of the menin-inhibitor exposure-duration risk this graph records elsewhere.
- Is a NUP98-HOXD13 tumour whose collaborating event is Meis1 insertion menin-inhibitor resistant from the outset? The collaborating-gene list and the drug-resistance node are the same gene and nobody has connected them.
- Does the p53 warning generalise to other fusion-driven MDS, or is it specific to this model? No fusion-free MDS comparator arm was run.
- How rare is NUP98-HOXD13 in the RNA-sequencing era? The 'third reported case' framing dates to 2000 and has not been updated.
- Why does menin-MLL1 inhibition spare the Hoxa cluster in this fusion's cells but not in others?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
NUP98-HOXD13 was first described in therapy-related acute myelogenous leukaemia, with the predicted fusion protein containing both the GLFG repeats from NUP98 and the homeodomain from HOXD13.
NUP98-HOXD13 gene fusion in therapy-related acute myelogenous leukemia · Raza-Egilmez SZ, et al. · Cancer Research 58(19):4269-4273 1998What this evidence cannot show
A case report from 1998; PubMed holds no DOI for this record, so none is given. Domain description is at the level of 'GLFG repeats' and 'homeodomain', not residues; the residue numbers in this entry come from UniProt.
Evidence that argues against this (1)
- The therapy-related framing from the founding case did not hold: the same fusion was subsequently found in de novo infant leukaemia, so 'therapy-related' is a context in which it was first seen rather than a property of the lesion. Shimada H et al. 2000
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
NUP98-HOXD13 was found in a de novo infant acute myelomonocytic leukaemia, the first report associating NUP98 rearrangement with infant leukaemia as well as with therapy-related AML or MDS, and only the third reported t(2;11) case at the time.
Generation of the NUP98-HOXD13 fusion transcript by a rare translocation, t(2;11)(q31;p15), in a case of infant leukaemia · Shimada H, et al. · British Journal of Haematology 110(1):210-213 2000What this evidence cannot show
A single infant case. 'Third reported case' dates the rarity to 2000; whether NUP98-HOXD13 remains this rare in the RNA-sequencing era was not established in this session.
Evidence that argues against this (1)
- The large paediatric NUP98 landscape cohort finds 20 'NUP98-X' cases spread across 13 partners out of 160 NUP98-rearranged patients, so any individual rare partner including HOXD13 is represented by at most a handful of cases even in the biggest dataset, and infant association rests on very few patients. Bertrums EJM et al. 2023
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
NUP98-HOXD13 transgenic mice develop a highly penetrant, severe myelodysplastic syndrome that progresses to acute leukaemia, and are uniformly fatal within 14 months from leukaemic transformation or severe anaemia and leucopenia.
NUP98-HOXD13 transgenic mice develop a highly penetrant, severe myelodysplastic syndrome that progresses to acute leukemia · Lin YW, et al. · Blood 106(1):287-295 2005What this evidence cannot show
A transgenic overexpression model driven by a haematopoietic promoter, not a knock-in at the endogenous locus, so expression level and cell-of-origin are imposed rather than physiological. Mouse MDS and human MDS differ in several respects and the model's value is as a tractable system rather than as a faithful replica.
Evidence that argues against this (1)
- The transgene is not sufficient on its own: retroviral insertional mutagenesis in this model identified Meis1, Mn1, Gata2, Erg, Epor and the miR-29a/miR-29b1 locus as collaborating events, so what the model reports is the consequence of the fusion plus a second hit. Slape C et al. 2007
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In the NUP98-HOXD13 mouse model, pifithrin-alpha gave only partial and transient rescue of cytopenias with no improvement in anaemia, and genetic loss of one or both p53 alleles exacerbated the MDS phenotype and accelerated progression to AML; the human disease spectrum includes MDS, AML and CML blast crisis.
Loss of p53 accelerates the complications of myelodysplastic syndrome in a NUP98-HOXD13-driven mouse model · Xu H, et al. · Blood 120(15):3089-3097 2012What this evidence cannot show
Pifithrin-alpha is a pharmacological tool with known off-target activity, so the drug arm and the genetic arm are not equivalent tests. The genetic arm is the stronger of the two and it points the same way. Findings are in a transgenic mouse and have not been tested in human NUP98-HOXD13 disease.
Evidence that argues against this (1)
- p53 loss accelerating disease is expected for any tumour-suppressor deletion in a pre-leukaemic state, so this may be a general property of MDS rather than anything specific to the NUP98-HOXD13 fusion; the experiment lacks a fusion-free MDS comparator arm. Lin YW et al. 2005
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Retroviral insertional mutagenesis in NUP98-HOXD13 mice identified Meis1, Mn1, Gata2, Erg, Epor and the miR-29a/miR-29b1 microRNA locus as genes that collaborate with the fusion during leukaemic transformation.
Retroviral insertional mutagenesis identifies genes that collaborate with NUP98-HOXD13 during leukemic transformation · Slape C, et al. · Cancer Research 67(11):5148-5155 2007What this evidence cannot show
Retroviral insertional mutagenesis is biased toward loci that a retrovirus can reach and activate, so the recovered gene list is a sample of the collaborating space and not an enumeration of it. Insertion near a gene does not prove the gene is the effector.
Evidence that argues against this (1)
- Meis1 appearing as a collaborating locus here and as the node through which menin-inhibitor sensitivity runs in the same fusion family raises the possibility that the collaborating event and the drug target are the same axis, in which case a Meis1-collaborating tumour would be predicted to be drug-resistant from the outset. That prediction has not been tested. Heikamp EB et al. 2022
- E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
HOXD13 is P35453, 343 aa, with the homeodomain at residues 276-335.
UniProtKB entry P35453, Homeobox protein Hox-D13 (HOXD13) · UniProt Consortium · UniProtKB 2026What this evidence cannot show
Primary gene name read back from the fetched record, not from a gene_exact search, which matches synonyms; the only listed synonym is HOX4I. The exact HOXD13 residue retained by the patient fusion was not established.
Evidence that argues against this (1)
- Germline HOXD13 mutations cause synpolydactyly, so HOXD13 has a substantial developmental role outside haematopoiesis; a homeodomain annotation says nothing about which of its functions the fusion co-opts. UniProt Consortium 2026
- E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In NUP98-HOXD13 cells, menin-MLL1 inhibition caused no significant loss of occupancy at the Hoxa cluster and no change in Hoxa expression; the dependency ran through Meis1, shown by two independent rescues.
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
Read via a delegated full-text pass in this session. A null result at the Hoxa cluster in one fusion's cells does not establish that Hoxa is irrelevant across the family; the founding abstract reports Hoxa downregulation for the family as a whole.
Evidence that argues against this (1)
- The same paper's abstract states that menin-MLL1 inhibition downregulates the HOXa cluster in NUP98-fusion leukaemia generally, so the Hoxa-null result in NUP98-HOXD13 cells is a within-paper exception that the abstract does not flag - a reader of the abstract alone would get the opposite impression. Heikamp EB et al. 2022
- E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
NUP98-HOXD13 binds Hoxb clusters that MLL-AF9 does not, Mll1 deletion downregulates Hoxa, Hoxb and Meis1 in its cells, and overexpression of a single Hoxb gene rescues the clonogenic defect.
NUP98 Fusion Proteins Interact with the NSL and MLL1 Complexes to Drive Leukemogenesis · Xu H, et al. · Cancer Cell 30(6):863-878 2016What this evidence cannot show
Read via a delegated full-text pass in this session. This paper's central interaction claim has a published failure to replicate by unbiased mass spectrometry (see mechanism-nup98-fusion-mll1-nsl-interaction), which does not bear directly on the ChIP-seq and rescue results but does argue for caution with the paper as a whole.
Evidence that argues against this (1)
- The HOXB result has not been revisited by any independent group in the decade since, and a single-gene rescue of a clonogenic defect is a permissive rather than a specific readout - many genes rescue clonogenicity. Terlecki-Zaniewicz S et al. 2021
- analogous to →NUP98::HOXA9, t(7;11)(p15;p15) — the founding NUP98 fusion, and an incomplete oncogene
THE TWO NUP98-HOMEODOMAIN FUSIONS - same 5' FG/GLFG repeats, same partner class, different HOX cluster. Both keep the NUP98 repeats and a homeodomain that supplies DNA binding directly, so they are the arm in which the fusion reaches HOX targets **through its own homeodomain** rather than through a KMT2A CXXC address. Their phenotypes diverge - HOXA9 gives AML, HOXD13 gives a highly penetrant MDS that transforms - which locates the difference in the 3' half.
- part of →NUP98 — domain architecture and residue coordinates (with the 1817/920 numbering trap resolved)
GLFG repeats fused to the HOXD13 homeodomain
Meis1 among the collaborating loci and the node menin inhibition acts through
first report linking NUP98 rearrangement to infant leukaemia (infant AMML, not ALL)
menin inhibition spares the Hoxa cluster in this fusion; the node is Meis1
3' partner coordinates for the fusion viewer lane
contributes the 5' half: the FG/GLFG region and the off-pore chromatin-facing module