Oncofusion

start / graph / interface-kmt2a-prkra-declined-a-cytoplasmic-pkr-activator

Interfaceweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q129-the-never-discussed-partners-of-kmt2a-and-psip1

KMT2A-PRKRA is one spoke-expanded BioPlex 3.0 AP-MS row with PRKRA (PACT) as bait and KMT2A as prey; PRKRA's only UniProt compartments are cytoplasm and perinuclear region, both uncoded - declined

also called KMT2A-PRKRA · MLL-PRKRA · PACT · RAX · EBI-54614065

A database lists PACT, a cytoplasmic protein that switches on the antiviral kinase PKR and helps make small RNAs, as a KMT2A partner. The one supporting experiment tagged PACT in kidney cells and KMT2A came along in the catch; PACT is not recorded as a nuclear protein and the pair has never been tested the other way round.

The same thing, for a scientist

**Verdict: DECLINED. Detection method is anti-tag co-immunoprecipitation (AP-MS) in BioPlex 3.0, spoke-expanded, with PRKRA as bait and KMT2A as prey** [E1]. One IntAct record, EBI-54614065, PMID 33961781, HEK293T only, intact-miscore 0.35 - the floor band shared with the experimentally secreted ANOS1 and FGFBP1 [E1]. UniProt O75569 gives PRKRA as 313 aa with subcellular location **Cytoplasm, perinuclear region and Cytoplasm, both UNCODED**, and no nuclear comment at any evidence level [E3]. Function: activates EIF2AK2/PKR in the absence of dsRNA, driving eIF2-alpha phosphorylation, translational shutdown and apoptosis; also required for siRNA production by DICER1. **Two separate problems compound.** The compartment argument is real but weaker than for ANOS1 and FGFBP1 because PRKRA's compartments are uncoded, so the record is an assertion without a cited experiment on both sides of the comparison (rule 91). And **PRKRA is a dsRNA-binding protein over-expressed with a tag in a whole-cell lysate**, the configuration in which nucleic-acid-mediated co-capture of an abundant chromatin protein is least distinguishable from a protein-protein contact - and no nuclease control is recorded in the IntAct row.

Go deeper
The nucleic-acid bridge, which is the specific objection for this partner

PRKRA is a double-stranded-RNA-binding protein: its documented functions are PKR activation and participation in DICER1-dependent siRNA production 3. KMT2A is a chromatin-bound protein that is recovered with DNA and RNA from any lysate that is not nuclease-treated.

An AP-MS row between an RNA-binding protein and a chromatin protein can be produced entirely by a shared nucleic acid, with no direct or indirect protein contact. IntAct's record carries no feature or annotation indicating benzonase or RNase treatment 1, and BioPlex's published workflow lyses whole cells and captures by tandem FLAG/HA affinity 2; the paper does not describe a nuclease step in the text read here.

This objection is testable and nobody has tested it, for this pair or for the three splicing factors (SRSF4, SRSF5, SRSF6) that sit in the same 0.35 band on the same prey 1. The experiment is one comparison: the same pull-down with and without nuclease.

And note the limit of the objection. A nucleic-acid bridge is still a real co-occupancy on a shared transcript or locus, which is a legitimate biological statement - it is just not the statement 'these two proteins bind'. Recording which one the record supports is the point.

Everything else this row lacks
  • Spoke expansion: the binary KMT2A-PRKRA pair was never observed. It was inferred by joining every prey in a PRKRA pull-down to the bait 1.
  • No second cell line: HEK293T only; it did not appear in BioPlex's HCT116 arm 1.
  • No reciprocal arm, and none possible: all 31 BioPlex rows touching Q03164 have KMT2A as prey, because the 3,969-residue protein is not among the 10,128 baits 1.
  • No orthogonal method: one publication, one detection method in the whole of IntAct 1.
  • No haematopoietic context: HEK293T is embryonic kidney; nothing has been done in any blood cell.
Facts and figures
partner agene-kmt2a
partner a uniprotQ03164
partner a length AA3969
partner bPRKRA
partner b uniprotO75569
partner b length AA313
verdictDECLINED
adjudicated byQ129, cycle 92, 2026-09-23
source publication pmid33961781
source datasetBioPlex 3.0
detection methodanti-tag co-immunoprecipitation (AP-MS)
experimental role partnerbait
experimental role kmt2aprey
kmt2a was never the baittrue
expansion methodspoke expansion
intact miscore0.35
replicated in second bioplex linefalse
reciprocally confirmedfalse
reciprocal confirmation possiblefalse
n publications1
n methods1
orthogonal method existsfalse
partner nuclear annotationnone at any evidence level, and the cytoplasmic comments are uncoded
nucleic acid bridge riskhigh - the bait is a dsRNA-binding protein and the prey is chromatin-associated; no nuclease control is recorded in the IntAct row and none was found in the keyword scan of the paper's methods
nuclease control recordedfalse
druggabilityNot assessable. No residue, domain or region is mapped on either side and no structure of the pair exists.
absence classnever-measured
absence class noteNo experiment has tested KMT2A against PRKRA with KMT2A as the bait, in any cell, and no nuclease-controlled version of the existing pull-down exists. The nuclease comparison is the cheap discriminating experiment and it is wet work, not re-analysis. A secondary measured-not-searched gap: the BioPlex portal's CompPASS-Plus probability for this pair was not retrieved.
shown for fusionNO - nobody has looked in any KMT2A fusion.
epigenetic readoutnone - no chromatin, histone-mark or occupancy measurement exists for this pair.
host cell lines
[
  "HEK293T embryonic kidney"
]
records
[
  "EBI-54614065"
]
partner compartments
[
  "Cytoplasm, perinuclear region [UNCODED]",
  "Cytoplasm [UNCODED]"
]
interferon cluster members in this beat
[
  "OASL",
  "PRKRA",
  "IRF2"
]
interface residues a
[]
structures
[]
ligands
[
  "drug-revumenib",
  "drug-ziftomenib"
]
How this could bite us back
  • PRKRA activates PKR, which shuts down translation and triggers apoptosis. If a KMT2A fusion did constrain that axis, a menin inhibitor releasing it would be therapeutically useful - and if it did the opposite, the drug would be blunting an apoptotic pathway. Neither direction has been measured, so this is a symmetric unknown rather than an opportunity.
  • An RNA-bridged association is not nothing: it would place KMT2A and PRKRA on shared transcripts, which would matter for any RNA-directed therapeutic. But it would be a statement about co-occupancy, not a druggable protein-protein interface, and the two get conflated whenever an AP-MS row is read as 'binds'.
Still open
  1. How many KMT2A interaction records in this graph come from experiments with no nuclease control, and how many of those partners are RNA-binding proteins? Chromatin proteins and RNA-binding proteins co-purify through nucleic acid, and the graph has never audited for it.
  2. Would a benzonase-treated pull-down remove PRKRA, the three SRSF proteins and the ribosomal block together? If one control collapses several of these partners at once, that is a single measurement adjudicating many rows.
  3. Does PRKRA have a nuclear pool in any human cell? UniProt's compartments for it are uncoded, so the question is open in the record rather than settled.
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 curates one record of a KMT2A-PRKRA association, EBI-54614065, from PMID 33961781 (BioPlex 3.0), annotated to Supplementary table 1. Detection method anti-tag co-immunoprecipitation, interaction type 'association', expansion method SPOKE EXPANSION, host Homo sapiens HEK293T embryonic kidney cell. PRKRA (O75569) is the BAIT and KMT2A (Q03164) is the PREY. intact-miscore 0.35. No mutation and no nuclease-treatment feature is recorded on either side. Three splicing factors - SRSF4, SRSF5 and SRSF6 - sit at the same 0.35 score on the same prey in the same dataset.

    What this evidence cannot show

    Read live from IntAct REST on 2026-09-23, not PSICQUIC (rule 93). The BioPlex supplementary table was not opened. The absence of a nuclease feature in the IntAct row is not proof that no nuclease was used - IntAct records experimental preparations inconsistently, and the field 'experimentalPreparations' was null for every record in this pull. So the nucleic-acid-bridge objection rests on the absence of a recorded control, not on a recorded absence of the control.

    Evidence that argues against this (1)
    • Spoke expansion, single cell line and floor score are properties shared by every partner in this beat, so they cannot distinguish PRKRA from the rest; the distinguishing objection is the RNA bridge, and that one is an untested hypothesis about the experiment rather than an observation. IntAct (EMBL-EBI) 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In human HEK293T and HCT116 epithelial cell lines, BioPlex 3.0 affinity-purified 10,128 tagged human proteins plus 5,522 HCT116 immunoprecipitations by tandem FLAG/HA affinity from whole-cell lysate, scored by CompPASS-Plus against a GFP control at 1% protein-level FDR with entropy-based dataset-level filtering, and reports about 60% of interactions in both networks confirmed by replication, reciprocal recovery or co-fractionation.

    Dual proteome-scale networks reveal cell-specific remodeling of the human interactome. · Huttlin EL, Bruckner RJ, Navarrete-Perea J, Cannon JR, Baltier K, Gebreab F, et al. (28 authors), Gygi SP · Cell 2021
    PMID 33961781DOI
    What this evidence cannot show

    Full text read 2026-09-23 through the PMC HTML route (PMC8165030) after fullTextXML returned HTTP 500. Neither 'KMT2A' nor 'PRKRA' appears in the paper's text; the row is a supplementary-table entry not opened here. The detailed affinity-purification protocol in the STAR Methods was skimmed for the words 'contamin', 'background' and 'false positive' rather than read line by line, so the statement that no nuclease step is described is a statement about a keyword scan and not a full reading of the methods.

    Evidence that argues against this (1)
    • BioPlex recovered WDR5, RBBP5 and ASH2L on KMT2A at 0.91-0.96, and those are protein-protein contacts recovered from the same lysate workflow, so the workflow plainly can return genuine chromatin-complex partners rather than only nucleic-acid co-riders. IntAct (EMBL-EBI) 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB O75569 is 313 aa, primary gene name PRKRA, recommended name 'Interferon-inducible double-stranded RNA-dependent protein kinase activator A'. Its only subcellular-location comments are Cytoplasm, perinuclear region and Cytoplasm, and both are UNCODED - no evidence code is attached; there is no nuclear comment at any evidence level. Function: activates EIF2AK2/PKR in the absence of double-stranded RNA, leading to EIF2S1 phosphorylation, inhibition of translation and induction of apoptosis; required for siRNA production by DICER1 and for subsequent siRNA-mediated post-transcriptional gene silencing.

    What this evidence cannot show

    Both compartment comments are uncoded, so this is a curator assertion without a cited experiment in the record (rule 91). That weakens the compartment argument in both directions: it cannot be cited as a measured exclusion of nuclear PRKRA. The cited function papers were not opened and no PubMed search was run on nuclear PRKRA in this beat, so the absence is unbounded in the literature sense.

    Evidence that argues against this (1)
    • PRKRA participates in DICER1-dependent small-RNA biogenesis, and components of that machinery do have documented nuclear pools in several systems. An uncoded UniProt compartment list is exactly the kind of record rule 91 warns is incomplete, so 'PRKRA is cytoplasmic' should be read as 'UniProt says so without citing an experiment'. UniProt Consortium 2026
Connected to