Oncofusion

start / graph / localization-chromatin-bound-versus-soluble-fraction-kmt2a

Mechanismstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q28-subnuclear

The soluble-nucleoplasm versus chromatin-bound axis - the one localisation readout that keeps a size dimension, and nobody has run it on a KMT2A fusion

also called chromatin fractionation of MLL fusions · salt-extraction profile of MLL-AF4 · nucleoplasmic versus chromatin-bound KMT2A::AFF1

You can ask where a protein sits in the nucleus by taking a picture, or by breaking the cell into layers and weighing what is in each layer. Only the second method can tell the leukemia fusion protein apart from the normal protein it shares an antibody with, and for KMT2A fusions nobody has ever done it.

The same thing, for a scientist

Every published statement about where a KMT2A fusion sits inside the nucleus comes from immunofluorescence, and immunofluorescence has no size dimension, so a shared-epitope antibody images wild-type and fusion together [E1][E2]. The readout that escapes this is biochemical fractionation followed by SDS-PAGE and immunoblot, because the gel supplies the discrimination the antibody lacks: KMT2A::AFF1 runs near 240 kDa and wild-type AFF1 at 125 and 145 kDa on the same blot [E1]. Three papers from 1996-1997 did use cell fractionation alongside imaging [E3][E4][E5], and all three resolved only **nucleus versus cytoplasm** - a question a nuclear localisation signal already answers. None separated soluble nucleoplasm from the chromatin pellet, none ran a salt-extraction series, and the one that states its system used transfected COS7 cells. As of 2026-09-20 a PubMed query for KMT2A-fusion chromatin fractionation returns three records and none fractionates a fusion protein [E6]. So this graph's most-repeated localisation statements - punctate, speckled, nucleolar - all sit on the readout class the fusion-attribution audit found cannot attribute, while the readout class that can attribute has never been pointed at the question.

Go deeper
Why fractionation is the attributable readout

No junction-specific antibody exists for any KMT2A fusion, so every reagent in use is raised against retained sequence and binds the fusion and at least one wild-type parent equally. In a microscope that is fatal. On a gel it is irrelevant: the canonical t(4;11) chimera is a fusion product whose mass differs from both parents 7, and the same rabbit polyclonals that cannot separate the species in an image separate them cleanly on a 3-8% gel 1.

A non-discriminating antibody applied to a separated sample is a discriminating assay.

So the correct form of 'where in the nucleus is the fusion?' is not an image. It is: in a three- or five-step fractionation, which band appears in which fraction, and in what ratio to the wild-type band in the same lane. Every lane carries its own internal control, because the wild-type parent is the comparator.

What the three historical fractionations actually did

Joh 1996 raised antibodies against MLL, LTG9/AF9 and LTG19/ENL and reported that MLL, the chimeric MLL-LTG9 and MLL-LTG19 products and the truncated N-terminal MLL product are synthesised and localised in the nuclei, by immunofluorescence and cell fractionation 3. This is the closest the literature comes, and the resolution is nuclear versus cytoplasmic.

Joh 1997 extended the same approach to MLL-AF6, whose partner is cytoplasmic, and found the chimera nuclear by both immunofluorescence and cell fractionation - and mapped the determinant to a region of the MLL N-terminus containing the AT-hook motifs 4.

Kanda 1997 expressed MEN (ELL), MLL/MEN and a truncated MLL from cDNA in COS7 cells and reported nuclear localisation by immunostaining and subcellular fractionation 5. The abstract's closing sentence - that these proteins 'were localized in the nucleus of leukemic cells' - is not what the experiment measured. The experiment was transfected monkey kidney fibroblasts.

None of the three separated chromatin from nucleoplasm.

What a fractionation would settle that imaging cannot

Four open questions in this graph turn on the soluble-versus-bound ratio, and none needs a new reagent:

  1. Menin inhibition. Occupancy loss is measured by ChIP; the fusion's fate afterwards is not. If the evicted fusion appears in the soluble fraction it is parked; if total fusion signal falls across all fractions it is degraded. That is a two-lane distinction and it decides whether the drug is partly a degrader.
  2. Fusion versus wild-type residence. Whether the fusion is more or less chromatin-retained than wild-type KMT2A is unmeasured, and the wild-type band sits in the same lane.
  3. Salt-extraction class. Whether the fusion releases like a nucleoplasmic factor (150-300 mM NaCl) or like a matrix-associated protein (nuclease or 2 M NaCl) grades this graph's nuclear-matrix entry quantitatively.
  4. Which retained module holds it there. CXXC, AT-hook and menin-binding mutants can each be read out as a shift in the bound fraction rather than as a change in a picture - and the CXXC entry already records that CpG-binding mutants lose transformation without moving in the nucleus, which is exactly the result a picture cannot refine and a fraction can.
Bounded absence, with the exact query

On 2026-09-20 the query (MLL-AF4 OR MLL-AF9 OR KMT2A-AFF1 OR MLL fusion protein) AND (chromatin fractionation OR chromatin-bound fraction OR nucleoplasmic fraction) returns 3 records, with a query translation containing no injected [UID] term. The three - PMIDs 40451434, 38822412 and 23974107 - are a HOXA9/SMYD3 paper, a sequential-drug ALL paper and a review of MLL1 in self-renewal. None fractionates a fusion protein 6.

Two limits on that number, stated so the next reader does not over-trust it. PubMed collapses MLL::AF4 to MLL-AF4, so a query listing four spellings searches fewer. And fractionation is a routine step that often appears only in a figure, never in an abstract, so an abstract-level zero is weaker evidence of absence here than for a headline finding.

Facts and figures
readout classbiochemical fractionation + SDS-PAGE + immunoblot
size dimensiontrue
attributabletrue
published for kmt2a fusionfalse
historical fractionations
89345411,9968934541: 199693495011,9979349501: 199792770491,9979277049: 1997
expected band positions kda
{
  "KMT2A::AFF1": 240,
  "AFF1_wild_type": [
    125,
    145
  ]
}
historical fractionations — the numbers behind the chart
[
  {
    "pmid": "8934541",
    "resolution": "nucleus vs cytoplasm",
    "system": "not stated in abstract",
    "title": "Identification of MLL and chimeric MLL gene products involved in 11q23 translocation and possible mechanisms of leukemogenesis by MLL truncation.",
    "authors": "Joh T, Kagami Y, Yamamoto K, et al.",
    "venue": "Oncogene",
    "year": 1996,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "9349501",
    "resolution": "nucleus vs cytoplasm",
    "system": "not stated in abstract",
    "title": "Chimeric MLL products with a Ras binding cytoplasmic protein AF6 involved in t(6;11) (q27;q23) leukemia localize in the nucleus.",
    "authors": "Joh T, Yamamoto K, Kagami Y, et al.",
    "venue": "Oncogene",
    "year": 1997,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "9277049",
    "resolution": "nucleus vs cytoplasm",
    "system": "transfected COS7",
    "title": "Subcellular localization of the MEN, MLL/MEN and truncated MLL proteins expressed in leukemic cells carrying the t(11;19)(q23;p13.1) translocation.",
    "authors": "Kanda Y, Mitani K, Tanaka T, et al.",
    "venue": "Int J Hematol",
    "year": 1997,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  }
]
How this could bite us back
  • If menin inhibition moves the fusion into the soluble nucleoplasm rather than destroying it, intermittent dosing allows immediate re-loading and the drug-holiday strategy fails.
  • If the fusion is predominantly soluble at steady state, a chromatin-displacement drug is acting on a minority species and the apparent potency of occupancy loss overstates the effect on the protein pool.
  • A fractionation readout adopted as a pharmacodynamic biomarker would be confounded by any treatment that changes global chromatin compaction, which differentiation-inducing agents do.
Still open
  1. In a t(4;11) line, what fraction of KMT2A::AFF1 is in the chromatin pellet versus the soluble nucleoplasm, and is that ratio different from wild-type AFF1 in the same lane?
  2. Does the fusion release at the same salt concentration as wild-type KMT2A, or does it behave as a matrix-associated species?
  3. After menin inhibition, does the fusion appear in the soluble fraction or disappear from all fractions?
  4. Does any published MLL-AF4 occupancy paper contain an unabstracted fractionation blot in its supplementary figures?
Every source behind this entry (7)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In t(4;11) lines the same rabbit polyclonal reagents resolve MLL-AF4 at about 240 kDa from wild-type AF4 at 125 and 145 kDa on a blot, while in the microscope both t(4;11) and non-t(4;11) leukaemic cells show a similar punctate nuclear pattern, and the fusion is the less abundant species.

    PMID 9808577
    What this evidence cannot show

    Abstract-level plus this graph's fusion-attribution audit of the same paper; the full text is not in PMC by any of the four routes and the record carries no DOI. The blot-versus-image contrast is the paper's own design, not a reinterpretation.

    Evidence that argues against this (1)
    • Chimeric ALL-1 proteins in transfected cells do show a punctate distribution mappable to two trithorax-conserved elements in the N-terminal ~1,100 residues, so the punctate pattern is not purely a wild-type artefact - it is simply not attributable in the 1998 experiment. Yano T et al. 1997
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Western blots of extracts from cells carrying t(11;19) and t(4;11) resolved chimeric HRX proteins at the sizes predicted from the fusion mRNAs, while immunocytochemistry showed the same punctate nuclear distribution for wild-type and chimeric HRX in cells with and without 11q23 translocations - a finding the authors state precludes an immunocytochemical approach to diagnosis.

    PMID 9129043
    What this evidence cannot show

    Abstract only. No PMCID by the Europe PMC core route, no DOI in the PubMed record, and the publisher page returned HTTP 403, so routes 2-4 were not reachable. This is the MLL-side twin of the 1998 anti-AF4 result and was not previously cited anywhere in this graph.

    Evidence that argues against this (1)
    • A fusion introduced into a cell that does not otherwise contain it is attributable by imaging, and in that setting HRX/ALL1-eps15 localised exclusively to the nucleus in bodies smaller and more numerous than wild-type - so imaging is not useless, only unattributable in the endogenous setting. Rogaia D et al. 1997
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Antibodies against MLL, LTG9/AF9 and LTG19/ENL showed that MLL, the chimeric MLL-LTG9 and MLL-LTG19 products and the truncated N-terminal MLL product are synthesised and localised in the nuclei, using immunofluorescence and cell fractionation studies.

    PMID 8934541
    What this evidence cannot show

    Abstract only - no PMCID by the Europe PMC core route and no DOI in the PubMed record, so the four-route test stops at route 1. The abstract does not state which cells were fractionated, nor whether the fractions were blotted against a size marker, and the resolution is nucleus versus cytoplasm only.

    Evidence that argues against this (1)
    • The same group's follow-up maps the nuclear-localisation determinant to a region containing the AT-hook motifs, i.e. retained KMT2A sequence present in every fusion - so a nuclear finding carries almost no information about which fusion is present. Joh T et al. 1997
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Immunofluorescence staining and cell fractionation showed that MLL-AF6 chimeric products localise in the nuclei despite AF6 itself being cytoplasmic, and the region of the MLL N-terminus responsible for that nuclear localisation was found to be a region containing the AT-hook motifs.

    PMID 9349501DOI
    What this evidence cannot show

    Abstract only; no PMCID. The abstract does not name the cell system, and the determinant mapping is performed on MLL fragments rather than on an endogenous fusion, so it localises a construct.

    Evidence that argues against this (1)
    • A different so-called cytoplasmic partner, EEN/endophilin A2, is predominantly nuclear in haematopoietic, fibroblast and epithelial lines and shuttles, so the premise that partners are cytoplasmic and must be dragged in does not hold for every partner. Cheung N et al. 2004
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Normal MEN (ELL), MLL/MEN chimeric and truncated MLL proteins expressed from cDNA in COS7 cells were nuclear by both immunostaining and subcellular fractionation.

    PMID 9277049DOI
    What this evidence cannot show

    The system is transfected COS7 monkey kidney cells, not leukaemic cells, although the paper's own concluding sentence says the proteins 'were localized in the nucleus of leukemic cells'. Abstract only; no PMCID. Overexpression of a construct is the weakest system for a localisation claim.

    Evidence that argues against this (1)
    • An N-terminal tag on MLL::AF9 failed to reproduce the untagged construct's leukaemogenesis kinetics, which the authors attribute to interference with chromatin localisation because the DNA-binding and menin-binding modules are N-terminal - so constructs can perturb the very property being measured. Olsen SN et al. 2022
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    As of 2026-09-20 PubMed returns 3 records for the stated chromatin-fractionation query, with a query translation containing no injected [UID] term, and none of the three fractionates a KMT2A fusion protein.

    What this evidence cannot show

    A counted absence is only as good as its symbol set. PubMed collapses MLL::AF4 to MLL-AF4 so four listed spellings search fewer, and the query is fusion-level rather than partner-symbol-level. All three returned records were title-checked individually (PMIDs 40451434, 38822412, 23974107).

    Evidence that argues against this (1)
    • Fractionation is a routine step that frequently appears only in a figure and never in an abstract, so an abstract-level zero understates how often it may have been done; a figure-level search of the MLL-AF4 occupancy literature could still turn one up. Godfrey L et al. 2017
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The t(4;11)(q21;q23) translocation generates an MLL-AF4 fusion product, which is what makes SDS-PAGE a discriminating step for this fusion.

    Acute mixed-lineage leukemia t(4;11)(q21;q23) generates an MLL-AF4 fusion product · Domer PH, et al. · Proceedings of the National Academy of Sciences of the United States of America 1993
    PMID 7689231DOI
    What this evidence cannot show

    The 1993 founding report establishes the fusion product. Breakpoint heterogeneity means the expected band position varies between patients, so a fractionation blot must be read against that patient's predicted size rather than a universal number.

    Evidence that argues against this (1)
    • Because the predicted size varies with breakpoint, a blot read against a single expected position could miss a genuine fusion band; the 1997 HRX study handled this by predicting each line's size from its own fusion mRNA. Butler LH et al. 1997
Connected to