Oncofusion

start / graph / localization-mllt10-af10-wild-type-nuclear-or-both

Mechanismweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q57-localization2

Wild-type AF10 is called nuclear on a construct and found in both fractions in tissue - and that disagreement decides whether CALM-AF10's cytoplasmic life is a gain or an inheritance

also called MLLT10 localization · AF10 nuclear localization signal · Af10 nuclear and cytoplasmic · AF10 bipartite NLS

AF10 is annotated as a nuclear protein on the strength of a paper that found a nuclear-targeting signal in it. But the one study that separated real mouse tissue into nuclear and cytoplasmic parts found AF10 protein in both. That matters, because the leukemic CALM-AF10 fusion is mostly cytoplasmic, and nobody can say whether that is something the fusion does or something AF10 already did.

The same thing, for a scientist

UniProt P55197 annotates MLLT10/AF10 as `Nucleus` with a single experimental code, ECO:0000269|PubMed:10860745 [E1]. That source identifies a **defined bipartite nuclear localisation signal in the N-terminal region** and reports AF10 localised to the nucleus, alongside extended-LAP/PHD-mediated homo-oligomerisation and AT-hook-dependent cruciform-DNA binding [E2]. The earlier characterisation of the murine orthologue reached a different answer by a different method: **subcellular separation of mouse testicular extracts found Af10 as a 140 kDa doublet in BOTH the nuclear and cytoplasmic fractions** [E3]. The graph's existing entry `localization-calm-af10-predominantly-cytoplasmic` records that the PICALM::MLLT10 fusion is predominantly cytoplasmic across five systems with CRM1-dependent export. **Whether that cytoplasmic pool is created by the CALM-derived NES or inherited from a pre-existing AF10 cytoplasmic pool has never been tested**, and the two published readings of wild-type AF10 point in opposite directions [E4].

Go deeper
The disagreement, and why it is not a trivial one
sourcemethodmaterialanswer
Linder 2000 2NLS mapping, construct-basedhuman AF10nucleus, via a defined bipartite NLS
Linder 1998 3subcellular fractionation of native tissuemouse testisnuclear AND cytoplasmic, 140 kDa doublet in both

Same first author, two years apart, two methods, two answers - and UniProt carries only the first. Neither paper is wrong: an NLS experiment asks can this sequence direct import, and a fractionation asks where is the protein actually found. They are different questions and the literature has quietly collapsed them into one annotation.

There is a technical caveat that weakens the fractionation side and should be said: cytoplasmic contamination of nuclear preparations and leakage in the other direction are the classic failure modes of 1998-era fractionation, and no marker-protein controls are reported in the abstract. But the doublet is seen in both fractions, in native tissue, with an antibody raised against AF10's cysteine-rich region - which is a more direct observation of endogenous protein than anything else in this dossier.

Why the answer changes the reading of CALM-AF10

This graph already holds a strong, well-sourced entry: CALM-AF10 is predominantly cytoplasmic, in HEK293, a murine PICALM::MLLT10 leukaemic line, U2OS, COS-7 and MEFs, and leptomycin B makes it entirely nuclear, so the cytoplasmic pool is actively exported by CRM1 rather than excluded. The standard reading is that CALM contributes a nuclear export signal and drags the AF10 moiety out.

If wild-type AF10 already has a substantial cytoplasmic pool 3, a second reading becomes available: the fusion may be enriching a pool that already exists rather than creating one, and the CALM NES may be tuning a balance rather than establishing a direction. The two readings differ in what they predict:

  • NES creates it: endogenous AF10 in a haematopoietic cell should be essentially all nuclear, and leptomycin B should not change it.
  • NES enriches it: endogenous AF10 should already show a leptomycin-B-sensitive cytoplasmic fraction.

That is one experiment - fractionate or image endogenous AF10 in a haematopoietic cell, +/- leptomycin B - and nobody has run it. The entire published AF10 export literature concerns the fusion.

What is annotated versus what is imaged, for the fusions
  • PICALM::MLLT10 (CALM-AF10): extensively imaged. See the existing entry; subnuclear detail is essentially absent, with one unidentified punctate pattern for the CALM-2091 breakpoint variant under leptomycin B.
  • KMT2A::MLLT10 (MLL-AF10): never imaged. A PubMed search for MLL-AF10 with localisation terms returns 7 records, none of which images the fusion 4. This is a canonical-id fusion in this graph (fusion-kmt2a-mllt10) with no localisation data at all.
  • Wild-type MLLT10 in a blood cell: never imaged. The two wild-type observations are in human construct work and mouse testis.

Note also, for the reader who reaches for the PZP domain: the extended LAP/PHD finger that mediates AF10 oligomerisation 2 is lost in MLL-AF10, a point the 1998 paper already makes. This graph holds subtraction-mllt10-pzp and mechanism-calm-af10-pzp-loss-is-a-removed-brake on that theme. If oligomerisation contributes to nuclear retention, losing the PZP could itself be a localisation lesion - untested.

Facts and figures
genegene-mllt10
uniprotP55197
length AA1068
compartment subcellularCONTESTED for the wild-type protein: nucleus via a mapped bipartite NLS (human, construct-based) versus both nuclear and cytoplasmic fractions (mouse testis, native tissue fractionation)
compartment subnuclearunpublished - no subnuclear observation of wild-type AF10 exists in any system
contradiction recordedPMID 10860745 (NLS mapping, human) says nucleus; PMID 9878787 (fractionation, mouse testis) finds a 140 kDa doublet in BOTH fractions. UniProt carries only the first. Resolving experiment: image or fractionate ENDOGENOUS AF10 in a haematopoietic cell with and without leptomycin B, with marker-protein controls.
shown for fusionNOT for KMT2A::MLLT10 - MLL-AF10 has never been imaged. For PICALM::MLLT10 the answer is YES and is held separately in localization-calm-af10-predominantly-cytoplasmic: predominantly cytoplasmic in five systems, entirely nuclear under leptomycin B.
open question this createsIs the CALM-AF10 cytoplasmic pool created by the CALM NES or inherited from a pre-existing wild-type AF10 cytoplasmic pool? The two published wild-type readings disagree and the discriminating experiment has not been done.
patient material datafalse
bounded absence date2026-09-20
uniprot evidence codes
{
  "Nucleus": "ECO:0000269|PubMed:10860745"
}
methods
[
  "NLS mapping in constructs",
  "subcellular fractionation of native tissue with an in-house polyclonal",
  "in-vitro oligomerisation and cruciform-DNA binding"
]
antibody reagents named in sources
[
  {
    "target": "AF10",
    "reagent": "in-house polyclonal raised to the AF10 cysteine-rich region; detects a 140 kDa doublet - NOT a catalogue clone",
    "use": "western on fractionated extracts",
    "source_pmid": "9878787"
  }
]
species and systems
[
  "human (construct-based NLS mapping, system unreadable)",
  "mouse testis (native tissue fractionation)"
]
How this could bite us back
  • If endogenous AF10 has a real cytoplasmic pool, then an inhibitor of the DOT1L-AF10 interaction engages a target distributed across two compartments, and the nuclear fraction that matters may be a minority of total target - a dose-setting problem invisible to whole-cell target engagement assays.
  • XPO1/CRM1 inhibition (selinexor) is clinically available and would force AF10 and CALM-AF10 nuclear. In CALM-AF10 disease that is proposed as therapeutic; in an AF10-nuclear-already setting it would do nothing, and the two cases cannot currently be distinguished.
  • Because MLL-AF10 has never been imaged, a resistance mechanism that relocalises the fusion would not be detectable by any assay currently used in this subtype.
  • The only endogenous-protein reagent in this literature is a 1998 in-house polyclonal; any pharmacodynamic assay for AF10 localisation would have to start by making and validating a new antibody.
Still open
  1. Where is endogenous AF10 in a haematopoietic cell, and is any cytoplasmic pool leptomycin-B sensitive? This single experiment decides whether CALM-AF10's cytoplasmic localisation is a gain of function or an amplification of normal behaviour.
  2. Where is KMT2A::MLLT10? A canonical fusion in this graph with no localisation data of any kind, while its sister PICALM::MLLT10 turned out to be mostly cytoplasmic - a result nobody predicted from chromatin data.
  3. Does loss of the extended LAP/PHD (PZP) module, which mediates AF10 oligomerisation and is lost in MLL-AF10, change AF10's nuclear retention? Oligomerisation is a classic retention mechanism and this has never been tested.
  4. What is the 140 kDa doublet? Isoform, phosphorylation or proteolysis - unexplained since 1998, and it is the only endogenous AF10 species anyone has blotted.
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 P55197 (human MLLT10/AF10, 1068 aa) annotates subcellular location as Nucleus with a single experimental code, ECO:0000269|PubMed:10860745, and records no cytoplasmic pool and no subnuclear detail.

    What this evidence cannot show

    A database record. gene_exact:MLLT10 AND organism_id:9606 AND reviewed:true returned one reviewed record, gene_primary MLLT10, accession P55197, length 1068. The annotation rests on a single paper whose localisation result is a construct-based NLS mapping, and the record does not reflect the mouse-tissue fractionation result that disagrees with it.

    Evidence that argues against this (1)
    • An annotation derived from one NLS paper answers 'can this protein be imported' rather than 'where is it'. The record is internally consistent and still an incomplete description, and the entry that cites it should say which question the underlying experiment asked.
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Biochemical analysis of human AF10 shows the extended LAP/PHD finger mediates homo-oligomerisation of recombinant AF10, AF10 binds cruciform DNA through a specific interaction with an AT-hook motif, and AF10 is localised to the nucleus by a defined bipartite nuclear localisation signal in the N-terminal region.

    Biochemical analyses of the AF10 protein: the extended LAP/PHD-finger mediates oligomerisation. · Linder B, Newman R, Jones LK, Debernardi S, Young BD, Freemont P, Verrijzer CP, Saha V · J Mol Biol 2000
    PMID 10860745DOI
    What this evidence cannot show

    KNOWN BUT NOT FULLY READ: no PMCID (Europe PMC and efetch both), and route 5 on the DOI returned an 11-character redirect stub - the documented Elsevier failure. So the cell type, the construct and whether endogenous protein was examined are NOT established here; the claim is taken from the abstract, which states the NLS result without naming a system. Oligomerisation and DNA binding are in-vitro results on recombinant protein. Read: abstract only.

    Evidence that argues against this (1)
    • A mapped NLS demonstrates import competence, not steady-state distribution; a protein can carry a functional NLS and still hold a large cytoplasmic pool if export or retention competes, which is exactly what the mouse-tissue fractionation suggests. Because the full text was unreadable, this entry cannot say whether the paper also measured endogenous distribution, and does not assume it did.
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    After subcellular separation of mouse testicular extracts, Af10 was identified as a 140 kDa doublet in BOTH the nuclear and the cytoplasmic extracts, using a polyclonal antibody raised to the cysteine-rich region of AF10.

    Expression pattern and cellular distribution of the murine homologue of AF10. · Linder B, Jones LK, Chaplin T, Mohd-Sarip A, Heinlein UA, Young BD, Saha V · Biochim Biophys Acta 1998
    PMID 9878787DOI
    What this evidence cannot show

    RULE 73: mouse testis, the tissue with the highest Af10 expression - not a haematopoietic cell and not human. Cell-fractionation purity controls (a nuclear and a cytoplasmic marker blot) are not described in the abstract and the full text was not read, so nuclear-preparation contamination cannot be excluded; this is the classic weakness of the method and of the era. The 140 kDa doublet is unexplained in the abstract (isoforms or modification). RULE 74: the antibody is an in-house polyclonal against the cysteine-rich region, not a catalogue clone. Read: abstract only (PubMed efetch, 2026-09-20).

    Evidence that argues against this (1)
    • The strongest counter to this item is methodological rather than bibliographic: a cytoplasmic signal in a 1998 fractionation without reported marker controls is the single commonest false positive in the localisation literature. It is recorded here because it is the only measurement of ENDOGENOUS AF10 distribution that exists, not because it is decisive, and the entry deliberately states the disagreement rather than choosing between the two papers.
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Bounded absence, measured 2026-09-20: the KMT2A::MLLT10 (MLL-AF10) fusion protein has never been imaged, and wild-type AF10 has never been localised in a haematopoietic cell.

    What this evidence cannot show

    QUERIES: (1) `(MLLT10[tiab] OR AF10[tiab] OR "CALM-AF10"[tiab]) AND (localization[tiab] OR localisation[tiab] OR "nuclear export"[tiab] OR shuttling[tiab] OR immunofluorescence[tiab])` = 21 records, all title-checked; every imaging record concerns the CALM-AF10 fusion, CATS, or Ikaros, and none images MLL-AF10 or endogenous AF10 in blood. (2) `("MLL-AF10"[tiab] OR "KMT2A-MLLT10"[tiab] OR "MLL AF10"[tiab]) AND (localization[tiab] OR immunofluorescence[tiab] OR nuclear[tiab])` = 7 records, title-checked, none imaging the fusion. RULE 67: both shapes return large non-zero counts, so the instrument returns ones. NAME LIST: MLLT10, AF10, CALM-AF10, MLL-AF10, KMT2A-MLLT10; the C. elegans orthologue name ZFP-1 was seen in results but not searched deliberately, and AF10's older designation in some literature as 'AF10q23' was not searched. 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)
    • The absence for MLL-AF10 is about IMAGING and should not be read as an absence of information: MLL-AF10 chromatin occupancy and its DOT1L-recruiting OM-LZ module are well studied and recorded elsewhere in this graph. What is missing is a picture, and the picture matters here specifically because the sister fusion PICALM::MLLT10 turns out to be mostly cytoplasmic, which nobody predicted from the chromatin data.
Connected to