Oncofusion

start / graph / import-class-ahd-dependent-mechanism

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

Import class 2: AHD-dependent mechanism, which cannot transfer because KMT2A::AFF1 has no AHD

also called AHD import · AF9/ENL-only mechanism · ANC1 homology domain claims

A small floppy piece of the AF9 and ENL proteins, called the AHD, is the hook that grabs the enzyme writing the H3K79 tag and several other partners. The infant fusion MLL-AF4 does not contain that piece at all, so any mechanism described as working through the AHD is a statement about other fusions.

The same thing, for a scientist

The ANC1 homology domain is a ~70-residue intrinsically disordered module at the C-terminus of AF9 (MLLT3) and ENL (MLLT1), cross-referenced as Pfam PF17793 / InterPro IPR040930 on P42568 and Q03111 [E1]. It is a competitive hub: it binds AF4, AFF4, DOT1L, BCOR and CBX8 through coupled folding and binding at an overlapping surface [E2], DOT1L engaging it at three sites with graded functional consequence [E3], and a chemical probe blocking the AF9/ENL-DOT1L/AF4/AFF4 interactions reduces both SEC assembly and H3K79 methylation [E4]. Reconstitution shows the states are mutually exclusive in practice: DOT1 binds AF9 but not AF9-AF4-P-TEFb [E5]. **KMT2A::AFF1 contains no AHD** [E1]. It contains the motif on the other side of that interface (PF18875, AF4_int). **Twenty entries in this graph cite at least one AHD-class paper** [E6]. Every claim in those entries of the form *'the fusion recruits X through the AHD'* is a statement about KMT2A::MLLT3 or KMT2A::MLLT1, and is not merely uncertain for the index fusion - it is structurally inapplicable.

Go deeper
Why this class is the sharpest of the five

Species and lineage imports are matters of degree: a mouse result may hold in humans with a different effect size, and a myeloid result may hold in B cells with a different target list. An interface import is not a matter of degree. Either the polypeptide is present in the chimera or it is not, and here it is not.

That makes this class the easiest to audit and the most consequential to get wrong. It also makes it the class where the graph's counter-argument sweeps have the least excuse for missing an instance, because the check is a domain-table lookup rather than a judgement.

The inversion that makes it interesting rather than merely negative

KMT2A::AFF1 sits on the ligand side of the AHD interface. So AHD biology is not irrelevant to the index fusion - it is relevant with the roles reversed, and the reversal changes the prediction. In MLL-AF9 the fusion supplies the AHD and competes for occupancy among DOT1L, AF4, BCOR and CBX8. In KMT2A::AFF1 the fusion supplies an AF4_int motif that competes with DOT1L for somebody else's AHD. On the reconstitution evidence, the index fusion's engagement of AF9/ENL should displace DOT1L rather than deliver it 5. That is a testable prediction with the opposite sign to the one this graph has been carrying by import.

Facts and figures
classinterface import - structurally inapplicable to the index fusion
interfaceAHD / PF17793 / IPR040930
entries citing class20
index fusion positionligand side - AFF1 carries PF18875 (AF4_int), which binds the AHD
inverted predictionengagement of AF9/ENL by KMT2A::AFF1 should displace DOT1L rather than deliver it
carried by
[
  "MLLT3",
  "MLLT1"
]
absent from
[
  "AFF1"
]
How this could bite us back
  • A drug programme aimed at the AHD-DOT1L interaction is aimed at a surface the index fusion occupies for a different purpose; in t(4;11) cells such a compound could detach the fusion's elongation machinery, kill by a mechanism unrelated to H3K79, and be optimised against the wrong readout.
  • If the inverted prediction is right, the index fusion is a competitive inhibitor of endogenous DOT1L recruitment, and removing the fusion would raise H3K79 methylation somewhere - a pharmacodynamic signal with the opposite sign to the one anyone would look for.
Still open
  1. Does KMT2A::AFF1 expression reduce DOT1L occupancy on endogenous AF9/ENL? The inverted prediction has, to this beat's reading, never been tested.
  2. Would an AHD-interaction inhibitor be worse than useless in t(4;11) disease, given that the fusion itself depends on that surface for P-TEFb?
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Pfam PF17793 (AHD) is cross-referenced on P42568 (MLLT3, 568 aa) and Q03111 (MLLT1, 559 aa) and not on P51825 (AFF1, 1,210 aa), whose only Pfam families are PF05110, PF18875 and PF18876.

    What this evidence cannot show

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

    Evidence that argues against this (1)
    • AFF1 does carry PF18875, a short motif in a disordered region that Pfam detects, so the family's blindness to AFF1's interaction repertoire is not total and a hidden AHD-like element is correspondingly less likely. UniProt Consortium 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The AF9 AHD is intrinsically disordered and recruits multiple partners through coupled folding and binding, with conserved aliphatic residues of each partner forming an integral part of the complex's hydrophobic core, and retains dynamics that may facilitate exchange between disordered partners.

    PMID 23260655DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read on 2026-09-19; full text not read. NMR of isolated domains and peptides, so the exchange model is inferred from relaxation measurements rather than observed on chromatin.

    Evidence that argues against this (1)
    • Conformational plasticity cuts against strict mutual exclusivity: a disordered hub that refolds around each partner could in principle accommodate partners at non-identical, partially overlapping surfaces. Kuntimaddi A et al. 2015
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    DOT1L has three AF9-binding sites, and structure-guided point mutations give graded reduction of DOT1L recruitment to MLL-AF9, graded loss of H3K79me2/me3 at target genes and graded loss of transformation potential.

    Degree of recruitment of DOT1L to MLL-AF9 defines level of H3K79 Di- and tri-methylation on target genes and transformation potential. · Kuntimaddi A, Achille NJ, Thorpe J, Lokken AA, Singh R, Hemenway CS, Adli M, Zeleznik-Le NJ, Bushweller JH · Cell Rep 2015
    PMID 25921540DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read on 2026-09-19; full text not read. Entirely MLL-AF9 - this is the quantitative backbone of the AHD-DOT1L axis and it exists only for the partner the index fusion does not have.

    Evidence that argues against this (1)
    • Three binding sites weaken the simple competition picture: a multivalent DOT1L could in principle retain one site while AF4 occupies another, which is exactly the ternary arrangement the reconstitution data argue against. Biswas D et al. 2011
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The AHD of AF9/ENL binds AF4, AFF4 or DOT1L, and a small-molecule inhibitor of those interactions (IC50 0.9-3.5 uM) reduces SEC assembly and DOT1L-mediated H3K79 methylation, suppresses onco-MLL, DOT1L, HOXA9 and MYC signatures, and has antitumour activity in a mouse MLL-rearranged leukaemia model.

    Small-molecule inhibitor of AF9/ENL-DOT1L/AF4/AFF4 interactions suppresses malignant gene expression and tumor growth. · Wu F, Nie S, Yao Y, Huo T, Li X, Wu X, Zhao J, Lin YL, Zhang Y, Mo Q, Song Y · Theranostics 2021
    PMID 34373735DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read on 2026-09-19; full text not read. A micromolar probe with a broad transcriptional footprint; the abstract does not describe a separation-of-function control distinguishing AHD occupancy from general toxicity, and the in vivo model is not specified by fusion in the abstract. **Half resolved.** The in vitro arm is **MOLM-13 at 5 microM** (GSE151660). **The in vivo arm is not deposited and stays unresolved** - do not rewrite this item as 'the model is MOLM-13'. Resolved 2026-09-19 by the Q18b unstated-systems sweep from the study's deposition, not from its text; bounded as a statement about the deposited data rather than about the whole study.

    Evidence that argues against this (2)
    • If a single compound blocking the AHD removes both SEC and DOT1L, then in KMT2A::AFF1 - where the fusion itself uses that surface - the same compound would detach the fusion's own elongation machinery, so the class's therapeutic reading inverts along with its mechanistic one. Biswas D et al. 2011
    • Resolving one arm of a two-arm study and leaving the other unresolved is the case most likely to be mis-summarised later, because the resolved half reads as the answer. The in vivo claim remains exactly as unsupported as it was. NCBI Gene Expression Omnibus 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In reconstitution, DOT1 forms a stable complex with AF9 and does not interact with AF9-AF4-P-TEFb complexes.

    Function of leukemogenic mixed lineage leukemia 1 (MLL) fusion proteins through distinct partner protein complexes. · Biswas D, Milne TA, Basrur V, Kim J, Elenitoba-Johnson KS, Allis CD, Roeder RG · Proc Natl Acad Sci U S A 2011
    PMID 21896721DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read on 2026-09-19; full text not read. Purified minimal complexes 'in the absence of other factors' - a sufficiency test that can miss interactions stabilised in cells by chromatin or bridging factors. The functional validation in this paper is MLL-AF9 immortalisation.

    Evidence that argues against this (1)
    • A steady-state negative in a tube cannot exclude transient DOT1L access to an AF4-occupied AHD in cells, which is the timescale on which a distributive methyltransferase would deposit H3K79me2 anyway. Leach BI et al. 2013
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Twenty of the 848 entries in this graph cite at least one AHD-class paper (PMIDs 23260655, 25921540, 32954361, 34373735, 34174329, 38747396 or 21896721).

    What this evidence cannot show

    Our own count by literal PMID string match over node JSON on 2026-09-19. Counts an entry once whether the citation is load-bearing or a counter-evidence note; misses entries that describe AHD mechanism without citing these seven papers.

    Evidence that argues against this (1)
    • Several of the twenty are structural or interactome entries about AF9 and ENL themselves, where AHD mechanism is the subject rather than an import into the index fusion - so the count is an upper bound on the number of entries at risk. Oncofusion knowledge graph et al. 2026
Connected to