Oncofusion

start / graph / localization-calm-af10-predominantly-cytoplasmic

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q55-calm-af10

CALM-AF10 is predominantly cytoplasmic in transfected non-haematopoietic cells - and NUCLEAR in leukaemia cells (title corrected 2026-09-21; the five systems are five instances of one design)

also called CALM-AF10 subcellular localization

A protein that switches genes on should live in the nucleus. Imaging with tagged constructs in five non-blood cell types puts the CALM-AF10 fusion mostly in the cytoplasm - but in mouse leukaemia cells the full-length fusion was found in the nucleus, so which compartment the fusion occupies appears to depend on the cell it is in.

The same thing, for a scientist

**CORRECTED 2026-09-21 (Q67). The original title of this entry - 'five independent systems agree' - overstated what the evidence shows, and the correction matters because this entry is load-bearing for the argument that CALM-AF10 cannot be doing chromatin work the way a KMT2A fusion does.** Every imaging study in the original reading used an epitope-tagged, overexpressed construct in a non-haematopoietic line, and they agree: YFP-CALM-AF10 overwhelmingly cytoplasmic with some nuclear signal [E1]; predominantly cytoplasmic in HEK293 and in a murine PICALM::MLLT10-positive leukaemic line [E2]; primarily cytoplasmic in U2OS [E3]; cytoplasmic in COS-7 [E4]; and CALM-AF10 redistributing IKAROS into the cytoplasm [E5]. **Leptomycin B makes it entirely nuclear** in HEK293 and in a murine PICALM::MLLT10 line [E2], so export is active and CRM1-dependent rather than the nucleus being closed to it. **Three published results the original reading did not carry, and they point the other way.** (i) In a mouse model of CALM-AF10 leukaemia, full-length CALM-AF10 **in leukaemia cells localised to the NUCLEUS**, with no consistent effect on growth-factor endocytosis and suppression of H3K79 methylation regardless of clathrin [see E3 counter-evidence]. (ii) The first molecular characterisation of a patient CALM-AF10 reported the chimaeric product localising to **both the nucleus and the cytoplasm** [see E4 counter-evidence]. (iii) The distribution is **conditional on a partner**: CATS is nuclear with a nucleolar preference, its binding site on CALM (aa 221-335) is retained in the fusion, and expressing CATS markedly increases the nuclear localisation of CALM and of CALM/AF10 - and CATS is predominantly expressed in thymus and spleen, the tissues this leukaemia arises in [see E1 and E5 counter-evidence]. **What survives.** The fusion shuttles, the export is CRM1-dependent, and in transfected non-haematopoietic cells at steady state most of it is cytoplasmic. **What does not survive is the inference in the original title** - that a predominantly cytoplasmic steady state in those systems is evidence against a chromatin mechanism. Two breakpoint variants differ in texture as well as amount: CALM-1926-AF10-GFP is diffusely cytoplasmic and becomes diffusely nuclear under leptomycin B, while CALM-2091-AF10-GFP clusters at discrete cytoplasmic sites and becomes **punctate** in the nucleus [E6]; **nobody has identified the puncta**. In MEFs about **44% of nuclei retain both DOT1L and CALM-AF10** [E2].

Go deeper
CORRECTION, 2026-09-21: the title overstated this, and writing the counter-evidence is what changed it

This entry was written on 2026-09-20 asserting that five independent systems agree the fusion is predominantly cytoplasmic, and that this is 'the opposite of what a chromatin oncoprotein should look like'. Writing the counter-evidence for its evidence items changed the reading in three ways, and the title has been corrected rather than quietly rewritten, because the original title is what gets quoted into the menin-independence argument.

  1. The five systems are one design. Four of the five are a tagged construct overexpressed in a non-haematopoietic line (U2OS, COS-7, HEK293/293T, MEF). Replicating a design does not make the result independent of the design.
  2. A leukaemia-cell result exists and disagrees. In a mouse model of CALM-AF10 leukaemia, full-length CALM-AF10 localised to the nucleus in leukaemia cells. That is the only haematopoietic, disease-relevant measurement in the set, and it is the one the original entry did not carry.
  3. The distribution is partner-conditional. CATS, whose binding site on CALM is retained in the fusion, drives CALM/AF10 into the nucleus, and CATS is a thymus- and spleen-enriched, proliferation-linked protein. So a cell line chosen for imaging convenience is also a choice of CATS background.

confidence_overall is lowered from strong to moderate. The shuttling and the CRM1-dependence of the export are unaffected; the inference drawn from the steady state is.

Why a cytoplasmic oncoprotein is not a contradiction here

Two readings coexist and this entry keeps both.

Reading 1 - the cytoplasm is where the damage is done to DOT1L. Exporting the fusion drags DOT1L out with it, producing the genome-wide H3K79 depletion, while the nuclear minority produces the focal HOXA gain 2.

Reading 2 - the export step is a chromatin-docking step. CRM1 is prebound at HOXA chromatin, so the NES is an address rather than a removal signal, and the cytoplasmic pool is a by-product of a shuttling cycle whose functional half is nuclear.

The imaging cannot distinguish them, and the field has not run the experiment that would. Note the trap: a steady-state image of a shuttling protein reports the ratio of import to export rates, not where the protein does its work.

Third reading, added 2026-09-21: the steady state is set by the host cell, and neither of the two readings above is being tested where the disease is. If CATS level moves the fusion between compartments and CATS is thymus- and spleen-enriched, an image taken in U2OS or COS-7 is not a measurement of where the fusion sits in a thymocyte. And the one leukaemia-cell measurement - nuclear fusion that still suppresses H3K79 - is compatible with neither reading as stated, because reading 1 requires the fusion to be out of the nucleus and reading 2 requires the cytoplasmic pool to be the by-product rather than the minority. The experiment that would settle it is an endogenously tagged allele in a fusion-positive human line, and it has not been done for this fusion in any laboratory located in this beat's reading.

The subnuclear gap

This graph maintains a detailed subnuclear layer for KMT2A fusions - speckles, lamina/LADs, chromocentres, nucleolus, nuclear matrix. For CALM-AF10 the only subnuclear observation anywhere in this beat's reading is 'punctate' under leptomycin B for one breakpoint variant 6, with no marker co-staining and no identification. That is a one-experiment, one-construct, drug-forced observation and it should not be quoted as a subnuclear localisation. Bounded absence: no study has co-stained CALM-AF10 with a nuclear-speckle, PML-body, lamina or nucleolar marker.

Facts and figures
mechanism classsubcellular localization
predominant compartmentcytoplasm in transfected non-haematopoietic lines; NUCLEUS reported for full-length CALM-AF10 in murine leukaemia cells (Oncogene 2012); both compartments in the first patient-derived characterisation (Leukemia 2001)
nuclear fraction reportedminority; 44% of MEF nuclei retain both DOT1L and CALM-AF10
export receptorCRM1/XPO1; leptomycin B produces complete nuclear accumulation
subnuclear localizationessentially unstudied; the only observation is 'punctate' for the CALM-2091-AF10 variant under leptomycin B in 293T cells, with no marker co-staining and no identification
counted absenceNo study located on 2026-09-20 co-stains CALM-AF10 with a nuclear-speckle, PML-body, nuclear-lamina or nucleolar marker.
method warningEvery observation in the original reading is from an overexpressed, epitope-tagged construct in a non-leukaemic cell line except one murine leukaemic line; there is no endogenously tagged CALM-AF10 imaging anywhere. A steady-state image of a shuttling protein reports a rate ratio, not a site of action. AND: five agreeing results obtained with one experimental design are five replicates of that design, not five independent systems - which is the error the original title made.
compartment is conditionalYes. Expression of the nuclear/nucleolar CALM interactor CATS markedly increases nuclear localisation of CALM and CALM/AF10, and CATS is predominantly expressed in thymus and spleen - so a cell line's CATS level is an uncontrolled variable in every one of these images.
systems agreeing
[
  "HEK293",
  "murine PICALM::MLLT10 leukaemic line",
  "U2OS",
  "COS-7",
  "293T",
  "MEF"
]
How this could bite us back
  • If the functional pool is the nuclear minority, then any measurement of total fusion protein - western, flow, mass spectrometry - is dominated by the inactive pool and will mis-report target engagement for any agent.
  • A drug that traps the fusion in the nucleus could increase its chromatin-associated fraction. Two papers say nuclear restriction abolishes transformation and one says nuclear retention is the oncogenic state; until that is settled, an export inhibitor is a bet on which of them is right.
Still open
  1. Where is endogenous CALM-AF10 in a patient blast? Every image in the original reading is an overexpressed tagged construct, mostly in non-haematopoietic cells, and the one leukaemia-cell result says nucleus.
  2. Does CATS level explain the discrepancy between the transfection images and the leukaemia-cell image? CATS is thymus- and spleen-enriched and drives CALM/AF10 nuclear; nobody has measured CATS in the lines used for the imaging.
  3. What are the nuclear puncta? One construct, one drug, one cell line, no markers - and it is the only subnuclear observation this fusion has.
  4. If the fusion is nuclear in a leukaemia cell and still suppresses H3K79 globally, what is the mechanism of the global loss? Cytoplasmic sequestration of DOT1L cannot be it in that system.
Every source behind this entry (6)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    YFP-tagged CALM localises primarily to the cytoplasm and plasma membrane, and YFP-CALM-AF10 is overwhelmingly cytoplasmic with some nuclear protein; both CALM and CALM/AF10 interact with CATS, which is nuclear.

    PMID 19383357DOI
    What this evidence cannot show

    Fluorescent-protein tagging and overexpression. A YFP tag on a 170 kDa chimera can alter distribution, and none of the imaging studies here used an endogenously tagged allele.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT FROM THE SAME GROUP, two years earlier, and it makes this distribution conditional rather than intrinsic. CATS localises to the nucleus with a preference for nucleoli, its interaction domain on CALM maps to aa 221-335 and is retained in the fusion, and expression of CATS MARKEDLY INCREASES the nuclear localisation of CALM and of CALM/AF10. CATS is predominantly expressed in thymus, spleen and colon. So a YFP-CALM-AF10 imaged in a cell line with low CATS is measuring the fusion in the wrong CATS background, and 'overwhelmingly cytoplasmic' is a statement about the host cell's CATS level as much as about the fusion. Archangelo LF et al. 2006
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CALM-AF10 localises predominantly to the cytoplasm in HEK293 cells and in a murine PICALM::MLLT10-positive leukaemic cell line; leptomycin B makes it entirely nuclear in both; and about 44% of MEF nuclei retain both DOT1L and CALM-AF10.

    PMID 23487024DOI
    What this evidence cannot show

    Overexpression in HEK293 plus one murine leukaemic line; the 44% figure is from MEFs, a fibroblast. Leptomycin B is a covalent CRM1 inhibitor and a tool compound.

    Evidence that argues against this (1)
    • The cytoplasmic majority does not settle where the fusion acts, and the literature disagrees on the sign: Okada et al. report nuclear colocalisation of CALM-AF10 with hDOT1L and treat nuclear retention as contributing to transformation, while the NES work treats nuclear restriction as abolishing it. Okada Y et al. 2006
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Immunofluorescence in U2OS cells shows primarily cytoplasmic CALM-AF10 with some nuclear distribution, and nuclear colocalisation of CALM-AF10 with hDOT1L.

    PMID 16921363DOI
    What this evidence cannot show

    U2OS osteosarcoma cells with transfected constructs - a non-haematopoietic, non-leukaemic system.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT IN THE SYSTEM THAT MATTERS, and it is the strongest objection in this entry. In a mouse model of CALM-AF10 leukaemia, IN LEUKAEMIA CELLS full-length CALM-AF10 LOCALISED TO THE NUCLEUS - with no consistent effect on growth-factor endocytosis, and suppressing H3K79 methylation regardless of the presence of clathrin. U2OS osteosarcoma cells with transfected constructs and a leukaemic blast are not the same measurement, and where the two have been compared they disagree. The cytoplasmic result is reproducible across transfection systems; it has not been reproduced in a haematopoietic cell carrying the fusion. Stoddart A et al. 2012
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In COS-7 cells both CALM and CALM-AF10 localise to the cytoplasm, and CALM-AF10 mutants lacking the NES localise to the nucleus.

    PMID 24397609DOI
    What this evidence cannot show

    COS-7 (African green monkey kidney) with transfected constructs. Its value is as an independent replication of the cytoplasmic finding in a fourth cell type.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT, and it predates all four transfection replicates. The first molecular characterisation of a patient CALM-AF10, in acute megakaryoblastic leukaemia, reports that although CALM is a cytoplasmic protein the CHIMAERIC FUSION PRODUCT IS ABLE TO LOCALISE TO BOTH THE NUCLEUS AND THE CYTOPLASM, and the authors' analysis of the fusion variants led them to the opposite inference from this entry's - that the critical product is the cytoplasmic one - which is a hypothesis, not an observation. Note also that AF10 carries a functional nuclear localisation signal, so a fusion retaining AF10 sequence has an import determinant. Four independent replications of one experimental design (a tagged construct overexpressed in a non-haematopoietic line) are not five independent systems. Jones LK et al. 2001
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    IKAROS expressed alone concentrates at dense centromeric heterochromatin, and becomes both nuclear and cytoplasmic when CALM-AF10 is present, with IKAROS and CALM-AF10 colocalising in the cytoplasm.

    PMID 18037964DOI
    What this evidence cannot show

    Co-transfection imaging plus GST pulldown and co-immunoprecipitation. Redistribution of an overexpressed partner by an overexpressed fusion is suggestive, not physiological. It is nonetheless the only reported subnuclear consequence of the fusion, and it is a loss of a subnuclear localisation rather than a gain.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT on the direction of the redistribution. Co-transfection imaging reports which construct is in excess, and the reciprocal experiment has been published: a nuclear, nucleolus-preferring CALM interactor whose binding site is retained in the fusion pulls CALM/AF10 INTO the nucleus when it is the one supplied. So the same assay format yields fusion-moves-partner-to-cytoplasm with IKAROS and partner-moves-fusion-to-nucleus with CATS. Neither establishes what the endogenous fusion does to the endogenous partner, and this entry should not carry the IKAROS result as 'the only reported subnuclear consequence of the fusion' without the CATS result beside it. Archangelo LF et al. 2006
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Two CALM-AF10 breakpoint variants differ in distribution: CALM-1926-AF10-GFP is diffusely cytoplasmic and becomes diffusely nuclear under leptomycin B, while CALM-2091-AF10-GFP clusters at discrete cytoplasmic sites and becomes punctate in the nucleus; the variants also alter the phenotype of the myeloid neoplasms they cause in mice.

    PMID 21706055DOI
    What this evidence cannot show

    GFP fusions in 293T cells. The nuclear puncta are unidentified - no marker co-staining - and appear only under a CRM1 inhibitor. Per the brief's condensate rules, a punctate appearance in an overexpression system is weak evidence about the endogenous protein and cannot distinguish phase separation from binding to clustered sites.

    Evidence that argues against this (1)
    • Restoring the AF10 PZP domain also promotes NUCLEAR localisation of CALM-AF10 and increases its chromatin association - while abolishing transformation. So 'more nuclear' tracks with 'less oncogenic' in at least one well-controlled experiment, which no simple localisation model in this entry predicts. Klein BJ et al. 2021
Connected to