Oncofusion

start / graph / interface-kmt2a-tad-crebbp-kix

Interfacestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q31-interfaces-structural

The KMT2A transactivation domain on the CBP KIX domain: a cooperative allosteric site at 2842-2858, solved three times, and thrown away by every breakpoint

also called MLL TAD-KIX · CBP KIX · 2AGH · 2LXS

Normal KMT2A carries a short activation segment that docks on one face of the coactivator CBP, and docking there makes CBP grip a second transcription factor on the opposite face more tightly. The leukaemic fusion is cut long before that segment, so it loses a cooperative switch nobody in this field discusses.

The same thing, for a scientist

The KIX domain of CREBBP/CBP has two ligand faces. KMT2A's transactivation domain occupies one; MLL-KIX and pKID/c-Myb binding on the two faces are positively cooperative, and NMR shows the coupling is transmitted through dynamic repacking of the KIX hydrophobic core rather than through a rigid conformational change. Three deposits define it: 2AGH (NMR, CBP KIX with an MLL peptide and a c-Myb peptide) and 2LXS/2LXT (NMR, KIX with MLL and, in 2LXT, CREB pKID). Mapped against the fetched UniProt sequences, the 2LXS MLL entity aligns exactly to Q03164 2842-2858 and the 2AGH MLL entity to approximately 2838-2869. Both lie about 1,250 residues C-terminal to the breakpoint cluster region, so the TAD-KIX contact is absent from every leukaemogenic KMT2A fusion. This graph had no entry for it at all.

Go deeper
Why a lost allosteric site is worth an entry

Most of the subtraction entries in this graph describe modules whose loss removes a function - a reader, a writer, a cleavage site. This one removes a coupling: the wild-type protein, by occupying one KIX face, increases CBP's affinity for a second factor on the other. That is a gain-of-cooperativity function, and a fusion that lacks it cannot supply it however much it is overexpressed.

It also cuts against a lazy reading of MLL fusion biology as 'the fusion does everything the wild type did plus elongation'. The fusion cannot allosterically prime CBP, and nobody has asked whether that matters.

Residues and provenance

KMT2A (Q03164): 2LXS MLL entity = GNILPSDIMDFVLKNTP, which occurs once in Q03164 at 2842-2858; the 2AGH MLL entity is a 31-mer covering approximately 2838-2869 and is not an exact match to the canonical sequence, consistent with a substituted or single-residue-deleted construct - a caution flag rather than a conclusion 3.

CREBBP (Q92793): the KIX domain, deposited as an 87-residue entity in both 2AGH and 2LXS 3.

Facts and figures
partner agene-kmt2a
partner bgene-crebbp
motif aKMT2A transactivation domain, Q03164 2842-2858 (2LXS) and approximately 2838-2869 (2AGH)
motif bCREBBP/CBP KIX domain, deposited as an 87-residue entity; the MLL face is opposite the c-Myb/pKID face
kd statusno dissociation constant recorded in this entry; the primary quantitative claim in the source literature is the cooperativity between the two KIX faces rather than an absolute affinity
fate in fusionlost - the TAD at ~2842-2858 lies about 1,250 residues C-terminal to the breakpoint cluster region and is absent from every leukaemogenic KMT2A fusion
what is lost is a couplingthe wild-type protein increases CBP's affinity for a second factor on the opposite KIX face; the fusion cannot do this at any expression level
druggabilitythe KIX domain is a long-standing and difficult PPI target with published small-molecule and peptidomimetic efforts; the MLL face specifically has no leukaemia-directed chemistry, and the module is absent from the fusion in any case
resolution notePDB metadata and entity sequences fetched from data.rcsb.org 2026-09-20; KMT2A spans aligned to the UniProt Q03164 sequence fetched the same day
chemical matterA designed MLL/c-Myb chimeric ligand occupying both KIX faces has been crystallised at 2.05 A (5SVH) under the deposition title 'Design of a nanomolar affinity ligand to the KIX domain of CBP'. The affinity claim is in the deposition title only; RCSB records no primary publication for this entry, so the nanomolar figure is NOT treated here as a verified measurement.
interface residues a
[]
kd measurements
[]
structures
[
  "2AGH",
  "2LXS",
  "2LXT",
  "5SVH"
]
structure details
[
  {
    "pdb": "2AGH",
    "title_fetched": "Structural basis for cooperative transcription factor binding to the CBP coactivator",
    "method": "NMR",
    "resolution_A": null,
    "contents": "Myb (P06876) + Crebbp (P45481, MOUSE) + KMT2A 31-mer, approx Q03164 2838-2869",
    "fetched": "2026-09-20"
  },
  {
    "pdb": "2LXS",
    "title_fetched": "Allosteric communication in the KIX domain proceeds through dynamic re-packing of the hydrophobic core",
    "method": "NMR",
    "resolution_A": null,
    "contents": "CREBBP (Q92793) 87-mer + KMT2A 19-mer covering Q03164 2842-2858",
    "fetched": "2026-09-20"
  },
  {
    "pdb": "2LXT",
    "title_fetched": "Allosteric communication in the KIX domain proceeds through dynamic re-packing of the hydrophobic core",
    "method": "NMR",
    "resolution_A": null,
    "contents": "CREBBP + KMT2A + CREB pKID (P16220)",
    "fetched": "2026-09-20"
  },
  {
    "pdb": "5SVH",
    "title_fetched": "Crystal structure of the KIX domain of CBP in complex with a MLL/c-Myb chimera",
    "method": "X-ray",
    "resolution_A": 2.05,
    "contents": "CREBBP (Q92793) 87-mer + a 58-residue MLL/c-Myb chimeric ligand annotated to Q03164 and P01103; the MLL portion begins SDDGNILPSDIMDFVLKNTPSMQALGESPES",
    "fetched": "2026-09-20",
    "citation_title_in_rcsb": "Design of a nanomolar affinity ligand to the KIX domain of CBP",
    "citation_status": "RCSB records the primary citation as To Be Published with no PMID or DOI as of 2026-09-20"
  }
]
How this could bite us back
  • Absent target: the TAD is not in the fusion, so this cannot be a fusion-directed site.
  • Promiscuous coactivator: CBP KIX binds many activators, and blocking one face perturbs an entire transcriptional hub.
  • Dynamic allostery: a coupling carried by hydrophobic core repacking is hard to engage selectively with a small molecule.
  • Cross-species mixing: one of the three deposits uses mouse Crebbp, so structure-guided work must not pool residue numbering across them.
Still open
  1. Does losing the MLL-KIX contact change CBP-dependent transcription in a KMT2A-rearranged cell? The question appears to have been asked nowhere.
  2. Is the wild-type allele's TAD-KIX contact part of why wild-type MLL1 is required for MLL-fusion leukaemogenesis?
  3. Does the 2AGH MLL construct carry a substitution relative to the canonical sequence, and if so does it sit at the interface?
  4. What is the affinity of the 5SVH MLL/c-Myb chimeric KIX ligand, and was it ever published? RCSB lists the entry as To Be Published as of 2026-09-20.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The CBP KIX domain binds an MLL transactivation-domain peptide and a c-Myb peptide on opposite faces, and the two bindings are positively cooperative.

    Structural basis for cooperative transcription factor binding to the CBP coactivator. · De Guzman RN, Goto NK, Dyson HJ, Wright PE · J Mol Biol 2006
    PMID 16253272DOI
    What this evidence cannot show

    NMR of isolated peptides with an isolated 87-residue KIX domain; cooperativity is measured on peptides, and the biological relevance to full-length MLL1 in cells is not established by this work. No PMCID; the abstract-level record was fetched from PubMed and the structural content read from the deposited entries.

    Evidence that argues against this (1)
    • Bruschweiler et al. show the coupling is carried by dynamic repacking of the hydrophobic core rather than by a defined conformational relay, so a description in terms of two static sites understates how much the effect depends on the domain's internal dynamics. Bruschweiler S et al. 2013
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Allosteric communication between the two KIX ligand faces proceeds through dynamic repacking of the hydrophobic core, characterised by NMR with MLL and pKID ligands.

    PMID 23651431DOI
    What this evidence cannot show

    A dynamics argument from relaxation and chemical-shift data on a small isolated domain; the ligands are short peptides. Full text available in PMC (PMC3719477) and the structural content confirmed from the deposits.

    Evidence that argues against this (1)
    • A dynamics-based allosteric mechanism is harder to exploit pharmacologically than a conformational one, and this work does not show the effect operates at endogenous protein concentrations. De Guzman RN et al. 2006
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PDB 2LXS contains a CREBBP KIX entity and an MLL entity whose sequence aligns exactly to Q03164 2842-2858; 2AGH contains KIX, a c-Myb peptide and a 31-residue MLL entity covering approximately 2838-2869.

    What this evidence cannot show

    Titles, methods and entity sequences fetched from data.rcsb.org 2026-09-20 and aligned to UniProt Q03164 and Q92793 fetched the same day. The 2AGH MLL entity does not match the canonical sequence exactly over its full length, so its span is given as approximate and the construct is flagged as possibly substituted. 2AGH's CBP entity is annotated to P45481 (mouse Crebbp) rather than to human Q92793.

    Evidence that argues against this (1)
    • 2AGH's coactivator chain is mouse Crebbp while 2LXS/2LXT use human CREBBP, so the two descriptions of this interface are not from the same protein and residue indices should not be mixed. De Guzman RN et al. 2006
  4. E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

    A designed MLL/c-Myb chimeric ligand for the CBP KIX domain has been crystallised at 2.05 A (PDB 5SVH), exploiting the cooperativity between the two KIX faces.

    PDB 5SVH: Crystal structure of the KIX domain of CBP in complex with a MLL/c-Myb chimera · RCSB entry records no primary citation authors; deposition listed as To Be Published · RCSB Protein Data Bank 2016
    What this evidence cannot show

    Entry metadata and both entity sequences fetched from data.rcsb.org 2026-09-20. RCSB lists the primary citation as 'To Be Published' with NO PMID and NO DOI, so the deposition title's claim of nanomolar affinity is unverifiable and is not recorded here as a measurement. The ligand is a chimera of the MLL TAD and c-Myb, i.e. a designed molecule rather than either natural partner.

    Evidence that argues against this (1)
    • Whatever the ligand's affinity, the KMT2A TAD it is built from is absent from every leukaemogenic fusion, so a KIX ligand of this design is not a fusion-directed agent. Bruschweiler S et al. 2013
Connected to