Oncofusion

start / graph / localization-hoxa9-nuclear-bodies-and-the-eif4e-body-switch

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q57-localization2

HOXA9 has its own nuclear bodies, and in monocytic AML they move into the eIF4E body - a subnuclear switch measured in patient marrow and never genotyped

also called HOXA9 nuclear bodies · HOXA9 eIF4E nuclear body · HOXA9 subcellular localization · HOXA9 nuclear import MEIS1 · HoxA9 nuclear dots

HOXA9, the gene KMT2A fusions switch on hardest, does not just float in the nucleus - it collects into small dots. In normal bone marrow those dots sit apart from another set of dots built around a protein called eIF4E, but in one subtype of adult AML the two merge, and blocking NF-kB pulls them apart again.

The same thing, for a scientist

HOXA9 is annotated Nucleus AND Cytoplasm in UniProt (P31269, both ECO:0000269 from a single source, PubMed 15657436) [E1]. That source is a translation-initiation paper, and reading it gives the best subnuclear dataset that exists for HOXA9: in primary human bone-marrow cells, endogenous HOXA9 forms discrete **nuclear dot structures ('HOXA9 nuclear bodies') that are spatially distinct from PML bodies and from eIF4E nuclear bodies**, on a background of diffuse nucleoplasmic and cytoplasmic staining. In **FAB M4/M5 AML and blast-crisis CML specimens** - where eIF4E is up and PRH/HHEX is displaced - HOXA9 almost completely colocalises with abnormally enlarged eIF4E nuclear bodies. Expressing an IkB super-repressor reverses it: HOXA9 reorganises into smaller bodies distinct from eIF4E bodies, as in normal marrow [E2]. Separately, HOXA9's nuclear entry is regulated rather than constitutive: thrombopoietin induces HOXA9 nuclear import in UT-7/TPO and in primary murine Sca-1+/c-kit+/Gr-1- cells in a MAPK-dependent fashion, likely via MEIS1 and two NLSs [E3]. **No specimen in any of this work is reported as KMT2A-rearranged** [E4].

Go deeper
HOXA9 makes bodies of its own, and they are not PML bodies

In primary human bone-marrow cells from apparently healthy individuals and from FAB M1/M2 AML, endogenous HOXA9 gives "diffuse cytoplasmic and nucleoplasmic staining" plus discrete nuclear dots. The paper is explicit that "these bodies... did not have any apparent spatial relationship to those containing PML, PRH, and eIF4E", and that occasional overlap with an eIF4E body "appeared to be random" 2.

That is a rare thing in this graph: a named, patient-derived, endogenous-protein subnuclear structure for a core gene. It is also completely uncharacterised - nobody has asked what else is in a HOXA9 body, whether it contains MEIS1 or PBX, whether it is a transcription site, or whether it is present in lymphoid cells at all.

The switch: HOXA9 moves into the eIF4E body in monocytic AML

In FAB M4/M5 AML and blast-crisis CML CD34+ cells, eIF4E is overexpressed and its nuclear bodies are abnormally enlarged; PRH/HHEX is downregulated and largely cytoplasmic. In that state HOXA9 and eIF4E almost completely colocalise in the enlarged bodies 2. Overexpressing HOXA9 in U937 is sufficient to reproduce the move - and the Y11A mutant, which cannot bind eIF4E, does not enter the bodies, which is the specificity control the claim needs.

The reversal is the strongest part. Adenoviral IkB-super-repressor, which shuts down NF-kB, "leads to downregulation of HOXA9 protein levels and to its reorganization into smaller bodies that are distinct from eIF4E nuclear bodies", restoring the normal-marrow pattern. So HOXA9's subnuclear address is a regulated, drug-reversible variable, not a fixed property.

Rule 73, stated plainly: none of this is KMT2A-rearranged disease. The specimens are FAB-classified AML and bcCML with no fusion genotype reported. It is worth saying, without over-reading it, that FAB M4/M5 monocytic morphology is the subtype in which KMT2A rearrangements are enriched - so the group in which HOXA9 relocates is plausibly fusion-enriched. That is a hypothesis about an unreported variable, not a finding, and it is written here as an open question rather than a claim.

Nuclear entry is regulated, and MEIS1 may be the carrier

HOXA9 is not simply a nuclear protein. Thrombopoietin induces nuclear import of HOXA9 in the human UT-7/TPO line and in primary murine Sca-1+/c-kit+/Gr-1- cells, in a MAPK-dependent fashion, with total HOXA9 levels unchanged 3. The authors attribute it to effects on its partner MEIS1 (whose mRNA TPO raises via PI3K) and to two newly identified NLSs.

This is the mechanistic bridge to gene-meis1: in the TALE/HOX literature the MEINOX partner is what retains the homeodomain protein in the nucleus, and the same paper is one of the few that measures it in blood cells rather than in limb bud or genital tract. It also means a cytoplasmic HOXA9 pool is a real, regulated thing in haematopoietic cells, not an artefact - which is why the UniProt Cytoplasm line here, unlike DOT1L's, carries an experimental code.

What is missing, precisely

No published image shows HOXA9 in a KMT2A-rearranged cell 4. Given that HOXA9 induction is the canonical readout of KMT2A-fusion activity and the endpoint of the menin-inhibitor pharmacodynamic argument, the questions this opens are cheap:

  1. Do KMT2A::AFF1 blasts have HOXA9 nuclear bodies, and are they the normal-marrow kind or the eIF4E kind?
  2. Does a menin inhibitor dissolve them, and does that happen before or after HOXA9 mRNA falls?
  3. Is HOXA9 protein partly cytoplasmic in these blasts, i.e. is some of the measured HOXA9 protein not doing transcription at all?

Question 3 matters for interpretation of every HOXA9 western blot in the field, because a whole-cell HOXA9 level is being read as a nuclear transcription-factor level.

Facts and figures
genegene-hoxa9
uniprotP31269
length AA272
compartment subcellularnucleus and cytoplasm, both experimentally annotated; nuclear entry is cytokine-regulated (TPO/MAPK) rather than constitutive
compartment subnucleardiscrete HOXA9 nuclear bodies, distinct from PML bodies and from eIF4E nuclear bodies in normal human marrow and FAB M1/M2 AML; relocated into enlarged eIF4E nuclear bodies in FAB M4/M5 AML and blast-crisis CML, reversibly under NF-kB inhibition
patient material datatrue
endogenous protein imagedtrue
shown for fusionNO - no specimen in the localisation work is reported as KMT2A-rearranged; the closest is FAB M4/M5 AML, a morphological class enriched for KMT2A rearrangements but not genotyped in the source
bounded absence date2026-09-20
uniprot evidence codes
{
  "Nucleus": "ECO:0000269|PubMed:15657436",
  "Cytoplasm": "ECO:0000269|PubMed:15657436"
}
methods
[
  "indirect immunofluorescence and laser-scanning confocal microscopy on primary patient marrow",
  "nuclear/cytoplasmic fractionation",
  "co-immunoprecipitation",
  "point-mutant (Y11A) specificity control",
  "adenoviral IkB super-repressor rescue"
]
antibody reagents named in sources
[
  {
    "target": "HOXA9",
    "reagent": "rabbit polyclonal (Upstate), IF 1:100, WB 1:500; specificity by recombinant-antigen pre-absorption",
    "use": "IF and western",
    "source_pmid": "15657436"
  },
  {
    "target": "eIF4E",
    "reagent": "mouse monoclonal (BD Transduction Laboratories), IF 1:100",
    "use": "IF and western",
    "source_pmid": "15657436"
  },
  {
    "target": "PML",
    "reagent": "mouse monoclonal CLONE 5E10, IF 1:10",
    "use": "IF compartment marker",
    "source_pmid": "15657436"
  }
]
species and systems
[
  "primary human bone marrow (healthy)",
  "primary human FAB M1/M2 AML",
  "primary human FAB M4/M5 AML",
  "human blast-crisis CML CD34+",
  "human U937",
  "human UT-7/TPO",
  "primary murine Sca-1+/c-kit+/Gr-1- progenitors"
]
regulated by
[
  "NF-kB (IkB super-repressor reverses the eIF4E-body localisation)",
  "thrombopoietin/MAPK (nuclear import)",
  "eIF4E binding via HOXA9 Y11"
]
How this could bite us back
  • HOXA9 nuclear import is cytokine-driven (TPO/MAPK). A marrow niche supplying that cytokine could keep HOXA9 nuclear while a drug lowers its transcription - a niche-based resistance route invisible to mRNA pharmacodynamics.
  • If HOXA9's oncogenic contribution runs partly through eIF4E-dependent mRNA export rather than through its own transcription, then targeting the KMT2A-fusion transcriptional axis leaves that arm intact, and an eIF4E-directed agent (e.g. ribavirin, already trialled in AML) becomes a rational combination rather than an unrelated idea.
  • NF-kB inhibition moves HOXA9 out of eIF4E bodies but also lowers HOXA9 levels, so any combination reasoning has to separate relocation from depletion; a trial reading only protein level would credit the wrong mechanism.
  • HOXA9 bodies are present in NORMAL bone marrow. An intervention aimed at the structure rather than at the fusion has no obvious therapeutic window in haematopoiesis.
Still open
  1. Do KMT2A::AFF1 blasts contain HOXA9 nuclear bodies, and are they the normal-marrow type or the eIF4E type? The imaging exists as a published protocol with a validated antibody; nobody has pointed it at this disease.
  2. The FAB M4/M5 specimens in which HOXA9 enters eIF4E bodies were never genotyped. What fraction were KMT2A-rearranged? Re-analysis of banked material, or a small new cohort, would answer whether this is a fusion phenotype hiding inside a morphological class.
  3. What else is in a HOXA9 nuclear body? MEIS1, PBX3, the fusion itself, nascent RNA - none has been co-stained, and the structure has essentially not been studied since 2005.
  4. Is HOXA9 protein partly cytoplasmic in KMT2A-rearranged blasts? Every HOXA9 western in this field measures whole-cell protein and is interpreted as transcription-factor activity.
  5. Does a menin inhibitor change HOXA9's nuclear/cytoplasmic ratio or its bodies before it changes HOXA9 mRNA? That ordering would distinguish a transcriptional from a post-translational arm of the response.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt P31269 (human HOXA9, 272 aa) annotates both Nucleus and Cytoplasm with ECO:0000269|PubMed:15657436, i.e. both compartments derive experimentally from one source, which is a study of eIF4E-dependent mRNA export rather than a localisation study.

    What this evidence cannot show

    A database record. gene_exact:HOXA9 AND organism_id:9606 AND reviewed:true returned exactly one reviewed record with gene_primary HOXA9, accession P31269, length 272. The point of this item is the provenance of the annotation, not the annotation itself: a single paper supplies both compartments, so the record is only as good as that paper and should be read with it.

    Evidence that argues against this (1)
    • UniProt gives no subnuclear detail for HOXA9 at all, although the cited paper contains a well-characterised subnuclear structure. This is the general pattern the beat was sent to test: UniProt records compartments, not bodies, so a graph that imports UniProt localisation wholesale gets zero subnuclear coverage by construction.
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In primary human bone-marrow cells, endogenous HOXA9 forms discrete nuclear dot structures spatially distinct from PML bodies and eIF4E nuclear bodies on a background of diffuse nucleoplasmic and cytoplasmic staining; in FAB M4/M5 AML and blast-crisis CML specimens HOXA9 colocalises almost completely with abnormally enlarged eIF4E nuclear bodies, an eIF4E-binding-dependent move (abolished by the HOXA9 Y11A mutant) that is reversed by expression of an IkB super-repressor, which reorganises HOXA9 into smaller bodies distinct from eIF4E bodies.

    Eukaryotic translation initiation factor 4E activity is modulated by HOXA9 at multiple levels. · Topisirovic I, Kentsis A, Perez JM, Guzman ML, Jordan CT, Borden KL · Mol Cell Biol 2005
    PMID 15657436DOI
    What this evidence cannot show

    RULE 73: the systems are primary human bone marrow, FAB M4/M5 AML, blast-crisis CML CD34+ cells and the U937 monocytic line - NOT KMT2A-rearranged disease and not B-lineage; no fusion genotype is reported for any specimen. RULE 74: immunofluorescence used rabbit polyclonal anti-HOXA9 (Upstate) at 1:100, mouse monoclonal anti-eIF4E (BD Transduction Laboratories) at 1:100, and anti-PML monoclonal CLONE 5E10 at 1:10; HOXA9 antibody specificity was checked by pre-incubation with recombinant HOXA9 antigen, which abolished the signal in both IF and western. Confocal resolution only - 'distinct bodies' means resolvably separate foci, not molecular exclusion. Several supporting panels are referenced to a laboratory website rather than to the article, and were not retrievable. Read: full text via PMC HTML (PMC544005) after fullTextXML returned HTTP 500 and efetch db=pmc returned no <body>.

    Evidence that argues against this (1)
    • The paper itself notes that HOXA9 and eIF4E are both abundant and diffuse in the cytoplasm, so its immunoprecipitation evidence for interaction in M1/M2 and normal specimens 'likely reflect[s] interactions within the populations... found diffusely throughout the cytoplasm and/or nucleus' rather than in bodies. Colocalisation and interaction are therefore measuring different pools here, and the body-level claim rests on the imaging alone. Separately, PRH/HHEX loss is the dominant change in the M4/M5 state, so HOXA9's entry into eIF4E bodies could be a consequence of a vacancy rather than an active relocation.
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Thrombopoietin induces nuclear import of HOXA9, without changing total HOXA9 levels, in the human UT-7/TPO line and in primary murine Sca-1+/c-kit+/Gr-1- haematopoietic cells, in a MAPK-dependent fashion; TPO also raises MEIS1 mRNA through PI3K, and the authors identify two nuclear localisation signals in HOXA9.

    PMID 15254242DOI
    What this evidence cannot show

    Mixed human cell line and primary murine progenitors; a cytokine-stimulation experiment in normal haematopoiesis, not leukaemia, and not KMT2A-rearranged. The MEIS1 route is inferred from correlated mRNA changes plus the known TALE partnership rather than demonstrated by a MEIS1-null control in the same experiment. Read: abstract only (PubMed efetch, 2026-09-20).

    Evidence that argues against this (1)
    • If HOXA9 nuclear entry is cytokine-regulated in normal progenitors, then HOXA9 protein abundance and HOXA9 transcriptional output can dissociate - and every whole-cell HOXA9 western in the KMT2A field is a measurement of abundance. A menin inhibitor that lowered nuclear HOXA9 without lowering total HOXA9, or the reverse, would be scored wrongly by the standard assay.
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Bounded absence, measured 2026-09-20: no published study reports HOXA9 protein subcellular or subnuclear localisation in a KMT2A-rearranged specimen or cell line.

    What this evidence cannot show

    QUERIES: `MEIS1[tiab] AND (nuclear localization[tiab] OR subcellular[tiab] OR immunofluorescence[tiab] OR nuclear import[tiab])` = 23 records (run for the partner and title-checked; it surfaced the HOXA9 transport paper and no KMT2A imaging); and the HOXA9-specific literature was reached through the UniProt evidence chain and through that result set rather than through a dedicated HOXA9 imaging query, which is a gap in this bound and is stated as such. RULE 67: the operand `MEIS1[tiab] AND ...` returns 23, so the shape returns ones. NAME LIST: HOXA9, HoxA9, HOX A9; the retired/alternate spellings HOX1G and HOX1.7 were NOT searched, and older HOXA9 literature does publish under HOX1G - so this bound is weaker than the DOT1L one and is recorded as moderate rather than strong. RULE 80 (added mid-beat): this count is bounded by TITLES AND ABSTRACTS, not by papers. Localisation panels are very often supporting figures whose abstracts never mention them, so a PubMed zero here means 'no paper ANNOUNCES this measurement', not 'no such figure exists'. Supporting figures and supplementary material were NOT searched.

    Evidence that argues against this (1)
    • This bound is explicitly incomplete: HOX1G and HOX1.7 are genuine previous symbols for HOXA9 (the brief's retired-symbol rule) and were not searched, and the query was routed through MEIS1 rather than run natively on HOXA9 imaging terms. A future pass should redo it as `(HOXA9 OR HOX1G OR HOX1.7)[tiab] AND (immunofluorescence OR confocal OR subcellular)[tiab] AND (MLL OR KMT2A)[tiab]` before this absence is relied on.
Connected to