Oncofusion

start / graph / interface-kmt2a-klhl22-declined-a-cryoem-paper-about-glutamate-dehydrogenase

Interfaceweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.q160-the-prey-only-interactome

KMT2A-KLHL22 - DECLINED. The single source is a cryo-EM structure of KLHL22-CUL3 bound to glutamate dehydrogenase, and KMT2A is a HEK293F pull-down passenger with no nuclear rationale

also called KMT2A-KLHL22 · MLL1-KLHL22 · CUL3-KLHL22

A database lists the ubiquitin-ligase adaptor KLHL22 as a KMT2A partner. The only paper is a structure of that ligase gripping a mitochondrial metabolic enzyme, and KMT2A just came along in the pull-down.

The same thing, for a scientist

**Verdict: DECLINED.** IntAct holds one KMT2A-KLHL22 record: anti-tag co-IP, *association*, HEK293F, KMT2A as prey, intact-miscore 0.35, from **PMID 37788672** [E1]. That paper is Teng, Wang, Liu, Tian, Stjepanovic and Su, *Structure* 2023: a **3.06 Angstrom cryo-EM structure of CUL3 with KLHL22 and the mitochondrial glutamate dehydrogenase complex GDH1**, showing a W-shaped E3 dimer architecture with three CUL3 dimers around a GDH1 hexamer, and in vitro polyubiquitination of GDH1 [E2]. **KMT2A is not its subject and is not in its abstract.** The row is a by-product of the HEK293F over-expression used to make material for cryo-EM. **The compartment argument is unusually strong here and runs against the row**: UniProt records KLHL22 in cytosol, centrosome, spindle, lysosome **and** nucleus, all experimentally coded [E3] - a five-compartment protein whose characterised substrates are metabolic and mitotic.

Go deeper
Why a structural-biology pull-down is a bad interaction source

The purification in a cryo-EM paper is optimised for yield and homogeneity of one complex, not for specificity. Material is over-expressed in HEK293F suspension culture at high copy number and taken through affinity capture; everything else recovered is, by design, what the authors are trying to remove. A prey list from such a preparation is the opposite of an interaction screen: it is the discard fraction.

This is a distinct decline reason from the ones this corpus already records (spoke expansion, non-expressing line, single BRET). It is worth naming because structural papers are increasingly curated into interaction databases, and their pull-down by-products will keep arriving with high-prestige citations attached.

The one thing that would change the verdict is a nuclear rationale, and there is a weak one: KLHL22 does carry an experimental nuclear localization, and CUL3-BTB-Kelch ligases do degrade chromatin proteins. But the substrate that was structurally characterised is a mitochondrial matrix enzyme, and nothing in the paper concerns chromatin. A plausible mechanism that the source did not investigate is not evidence.

Facts and figures
partner agene-kmt2a
partner bKLHL22
partner b uniprotQ53GT1
partner b length AA634
partner a uniprotQ03164
partner a length AA3969
verdictDECLINED
adjudicated byq160, cycle 146, 2026-09-24
druggabilityNot assessable. CUL3 adaptors are of interest as degrader handles, but nothing here maps a KMT2A-KLHL22 surface and the structurally characterised substrate is unrelated.
n methods1
n platforms1
n publications1
endogenous sideno
intact miscore0.35
cell statenone - preparative over-expression
epigeneticsNone. The paper's substrate is a mitochondrial matrix enzyme.
decline reason classstructural-biology preparative pull-down - the prey list is the fraction the authors were removing
absence classnever-measured
absence class noteNo experiment has asked whether KMT2A or a KMT2A fusion is a CUL3-KLHL22 substrate. A cycloheximide-chase with KLHL22 knockdown would answer it, but nothing in the current evidence motivates running it.
interface residues a
[]
structures
[]
ligands
[]
records
[
  "KMT2A-KLHL22, anti tag coip, association, HEK293F, KMT2A prey, miscore 0.35, PMID 37788672"
]
cell systems
[
  "HEK293F suspension, recombinant over-expression for cryo-EM"
]
localization
{
  "compartment": "cytoplasm, nucleus, lysosome",
  "subnuclear": [],
  "shown_for_fusion": "NO - nobody has looked in any fusion."
}
measurements
[
  {
    "quantity": "intact_miscore",
    "subject": "KMT2A|KLHL22",
    "value": 0.35,
    "units": null,
    "n": 1,
    "universe": "IntAct REST, PMID 37788672",
    "null": null,
    "note": "0.35 is the floor value for a single-record, single-method association in this record set."
  }
]
How this could bite us back
  • Declining a CUL3-adaptor row removes the only pointer this corpus has towards an E3 ligase for KMT2A. If a degrader strategy is ever pursued, this row will be rediscovered and will look like a lead - the decline should be read as 'unsupported', not as 'checked and false'.
Still open
  1. How many other IntAct rows in this corpus originate from structural-biology preparative pull-downs? This is the first one identified as such, and the class is growing as cryo-EM papers get curated.
  2. Is any KMT2A fusion a CUL3 substrate? Fusion protein stability is a live therapeutic question and no E3 ligase has been assigned to any KMT2A fusion in this corpus.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct REST returns one KMT2A-KLHL22 (Q03164-Q53GT1) record: anti-tag co-immunoprecipitation, interaction type 'association', host Homo sapiens HEK293F embryonic kidney cell, KMT2A as PREY, intact-miscore 0.35, from PMID 37788672.

    What this evidence cannot show

    One record, one method, one publication, lowest-tier miscore. HEK293F is the suspension-adapted derivative used for large-scale recombinant expression, which is the tell that this is preparative material rather than a physiological pull-down.

    Evidence that argues against this (1)
    • IntAct curated this row deliberately from a 2023 paper, so a curator judged it worth depositing. Dismissing curated rows on the grounds that the paper was structural would discard some genuine complex members identified exactly that way. IntAct (EMBL-EBI) 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    PMID 37788672 (Teng et al., Structure 2023) presents a 3.06 Angstrom cryo-EM structure of CUL3 in complex with Kelch-like protein 22 (KLHL22) and the mitochondrial glutamate dehydrogenase complex GDH1, describing a W-shaped architecture formed by E3 ligase dimers with three CUL3 dimers dynamically associated with a single GDH1 hexamer, and showing CUL3-mediated polyubiquitination of GDH1 in vitro. Neither KMT2A, MLL nor chromatin appears in the abstract, keywords or MeSH terms.

    Cryo-EM structure of the KLHL22 E3 ligase bound to an oligomeric metabolic enzyme. · Teng F, Wang Y, Liu M, Tian S, Stjepanovic G, Su MY · Structure 2023
    PMID 37788672DOI
    What this evidence cannot show

    Abstract, keywords and MeSH read; the body and supplementary mass-spectrometry tables were not opened. The absence of KMT2A from the abstract is the support for the decline and is declared as such rather than left for an auditor to infer.

    Evidence that argues against this (1)
    • CUL3-BTB-Kelch ligases do have nuclear substrates, and KLHL22 carries an experimentally coded nuclear localization. A KMT2A-KLHL22 contact is therefore not physically absurd - it is unsupported, which is a different and weaker objection than impossible. Teng F et al. 2023
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q53GT1 (KLH22_HUMAN) is 634 aa and carries five experimentally coded (ECO:0000269) subcellular locations: Cytoplasm cytosol, Cytoplasm cytoskeleton microtubule organizing center centrosome, Cytoplasm cytoskeleton spindle, Nucleus, and Lysosome.

    What this evidence cannot show

    Rule 91 applies to all five codes. A protein with five annotated compartments is either genuinely mobile or has accumulated annotations from disparate over-expression studies; UniProt does not distinguish, and the cited papers were not opened.

    Evidence that argues against this (1)
    • The nuclear annotation is experimentally coded and is not weaker than the cytosolic one in UniProt's own scheme. Using 'mostly cytoplasmic' as a decline argument reads a prior into a record that does not state one. UniProt Consortium 2026
Connected to