Oncofusion

start / graph / concept-partner-import

Explainerstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q-partner-provenance

Partner import: when a mechanism measured in one fusion is asserted about another

also called cross-partner import · fusion provenance · whose fusion was it measured in

Almost everything known about how these leukaemia fusion proteins work was worked out in one particular fusion, MLL-AF9, usually in mice. Borrowing that knowledge for the infant fusion MLL-AF4 is often reasonable and sometimes badly wrong, and the difference depends on whether the borrowed step needs a protein part that MLL-AF4 does not have.

The same thing, for a scientist

A **partner import** is an assertion in this graph about the index fusion (KMT2A::AFF1) whose load-bearing evidence was obtained in a different fusion, a different species, or a different lineage. It is a provenance property of a claim, not a verdict on it. Imports are usually unavoidable - the mechanistic literature is dominated by murine MLL-AF9 myeloid transformation, MLL-ENL, MLL-AF10 and human AML lines - and partner-shared biology is real: all common KMT2A chimeras co-purify the super elongation complex [E1], and endogenous AF10 is required for transformation by three fusions that do not contain it [E2]. The measured import fraction in this graph is 54% of fusion-bearing load-bearing evidence (`ana-partner-provenance-audit`). **The rule that separates safe from unsafe imports is whether the imported step names an interface.** A claim of the form *'this complex is present at fusion targets'* usually survives the partner boundary. A claim of the form *'the fusion recruits X through domain D'* survives only if the index fusion has D. KMT2A::AFF1 has no AHD and no AF10 OM-LZ, so every DOT1L-recruitment claim built on either interface fails the test [E3][E4].

Go deeper
Three axes, and they are not equivalent
  1. Fusion identity. The 3' partner determines which interfaces the chimera carries. This is the axis where imports break hardest, because an interface is either present or absent - there is no partial credit.
  2. Species. Murine MLL-AF9 myeloid transformation is the field's workhorse assay. Species differences matter most for expression, dosage and immune context and least for direct protein-protein contacts, which are usually conserved.
  3. Lineage and age. The index disease is infant B-lymphoid. Most mechanism is myeloid. Lineage changes the target gene set (HOXA9/MEIS1 dependence differs), the cell of origin, and the differentiation block being enforced.

An entry can import on one axis and not the others, and the three should be stated separately. 'Mouse MLL-AF9 myeloid' is three imports, not one.

The interface test, written to be applied

Before importing a mechanism into KMT2A::AFF1, ask: does the claim survive if the 3' partner is replaced?

  • Menin binds the KMT2A N-terminus - survives. Every fusion retains it; this is why the menin interface is the partner-independent part of this project's bet.
  • The fusion recruits the super elongation complex - probably survives; shown for multiple chimeras 1.
  • The fusion recruits DOT1L through the AHD - does not survive. The AHD belongs to AF9/ENL 3.
  • DOT1L is directed by the AF10 OM-LZ - does not survive. The OM-LZ belongs to AF10 4.
  • H3K79me2 is elevated at fusion targets - survives, and is independently established in MLL-AF4 5.

The last line matters: an import can be unnecessary. Several claims in this graph are imported when index-fusion evidence exists and was simply not cited.

Facts and figures
definitionan assertion about KMT2A::AFF1 whose load-bearing evidence was obtained in a different fusion, species or lineage
testdoes the claim survive replacement of the 3' partner? interface-naming claims usually do not
measured import fraction PCT54
axes
[
  "fusion identity",
  "species",
  "lineage and age"
]
How this could bite us back
  • Treating every import as suspect would discard most of what is known about this disease and leave the graph unable to say anything mechanistic; the concept is only useful if it is applied by the interface test rather than as a blanket discount.
  • A target nominated from an interface-naming claim that fails the test inherits a mechanism the index disease may not run, and the failure will only appear at the point where a compound is tested in t(4;11) cells.
Still open
  1. Is there a systematic way to flag interface-naming claims automatically, so that an entry asserting 'the fusion recruits X via domain D' is checked against the index fusion's domain list at write time?
  2. How often does a mechanism established in KMT2A::AFF1 fail to transfer to MLL-AF9? The import is almost entirely one-directional in the literature, so the reverse failure rate is unmeasured.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    AF4 and ENL family proteins form a higher-order complex with P-TEFb (AEP) that MLL oncoproteins fused to AEP components constitutively hijack, and even MLL-AF6, whose partner is cytoplasmic, constitutively recruits AEP by an alternative route.

    PMID 20153263DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read on 2026-09-19; full text not read. Biochemical purification plus murine myeloid transformation. 'Most frequently used strategy' is a statement about the partners tested, not a survey of all KMT2A partners.

    Evidence that argues against this (1)
    • Shared AEP recruitment does not imply shared DOT1L recruitment: reconstitution shows DOT1 does not join AF9-AF4-P-TEFb complexes, so partners that share AEP can still differ in whether they deliver the H3K79 writer. Biswas D et al. 2011
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Endogenous AF10 regulates progressive H3K79 methylation and HOX expression, and AF10 inactivation impairs transformation by MLL-AF9, MLL-AF6 and NUP98-NSD1, none of which contain AF10.

    AF10 regulates progressive H3K79 methylation and HOX gene expression in diverse AML subtypes. · Deshpande AJ, Deshpande A, Sinha AU, Chen L, Chang J, Cihan A, Fazio M, Chen CW, Zhu N, Koche R, Dzhekieva L, Ibáñez G, Dias S, Banka D, Krivtsov A, Luo M, Roeder RG, Bradner JE, Bernt KM, Armstrong SA · Cancer Cell 2014
    PMID 25464900DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read on 2026-09-19; full text not read. Mouse models; all three fusions tested are myeloid and none is KMT2A::AFF1, so the best available demonstration that mechanisms cross partner boundaries is itself an import on two of the three axes.

    Evidence that argues against this (1)
    • A shared requirement for an endogenous cofactor is weaker than a shared mechanism: three fusions can all need AF10 for different reasons, and the experiment cannot distinguish a common pathway from a common bottleneck. Okuda H et al. 2017
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The AHD - the DOT1L-binding module - is a domain of AF9 and ENL, cross-referenced as Pfam PF17793 on P42568 and Q03111 and absent from AFF1 (P51825).

    What this evidence cannot show

    Accessions fetched directly on 2026-09-19 with primary gene names and lengths read back. Pfam absence is absence of a detected match, and AFF1 is annotated as overwhelmingly disordered, where profile HMMs under-detect short motifs.

    Evidence that argues against this (1)
    • AFF1 does carry PF18875, the motif that binds the AHD, so the index fusion is not disconnected from AHD biology - it is on the other side of the same interface, which is a subtler position than simple absence. Leach BI et al. 2013
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt P55197 annotates MLLT10 residues 703-784 as 'Transactivation domain; required for DOT1L-binding' with a leucine zipper at 750-778; no equivalent region is annotated on AFF1.

    What this evidence cannot show

    Fetched directly on 2026-09-19. UniProt regions are curatorial calls from primary literature rather than crystallographic footprints.

    Evidence that argues against this (1)
    • The AF10-DOT1L interaction depends on a higher-order coiled-coil configuration rather than on simple heterodimerisation, so describing it as a transferable 'interface' understates how much of its activity is oligomeric state. Song X et al. 2019
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    H3K79 methylation profiles define murine and human MLL-AF4 leukaemias and DOT1L suppression inhibits expression of critical MLL-AF4 target genes, so the H3K79 phenotype in the index fusion is independently established and need not be imported.

    H3K79 methylation profiles define murine and human MLL-AF4 leukemias. · Krivtsov AV, Feng Z, Lemieux ME, Faber J, Vempati S, Sinha AU, Xia X, Jesneck J, Bracken AP, Silverman LB, Kutok JL, Kung AL, Armstrong SA · Cancer Cell 2008
    PMID 18977325DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read on 2026-09-19; full text not read. ChIP-chip, uncalibrated, RNAi-era knockdown; the human arm is partner-mixed MLL-rearranged ALL rather than t(4;11) specifically.

    Evidence that argues against this (1)
    • Establishing the phenotype is not establishing the mechanism, and the recruitment step remains unexplained for this fusion - so 'the import was unnecessary' holds for the mark and not for how the mark gets there. Biswas D et al. 2011
Connected to