KMT2A-FGFBP1 is spoke-expanded BioPlex 3.0 AP-MS recovered in BOTH HEK293T and HCT116 - it passes BioPlex's own replication criterion and the partner is a secreted extracellular protein - declined
also called KMT2A-FGFBP1 · MLL-FGFBP1 · HBP17 · EBI-54852101 · EBI-54555390
A database lists FGFBP1, a protein cells release into the space outside them to hand growth factors around, as a KMT2A partner. Two different cell lines gave the same answer - which is exactly the check that is supposed to confirm an interaction, and here it confirms one that cannot be real.
The same thing, for a scientist
**Verdict: DECLINED, and it is the most instructive decline in the beat. Detection method is anti-tag co-immunoprecipitation (AP-MS) in BioPlex 3.0, spoke-expanded, with FGFBP1 as bait and KMT2A as prey** [E1]. Two IntAct records from PMID 33961781: EBI-54555390 in **HEK293T** and EBI-54852101 in **HCT116**, both intact-miscore 0.35 [E1]. **This pair therefore passes the first of the three validation criteria BioPlex names for itself - replication in the second cell line** [E2] - and the partner is UniProt Q14512, 234 aa, subcellular location **Secreted, extracellular space [ECO:0000269] and Cell membrane [ECO:0000269]**, function 'carrier protein that releases fibroblast growth factors from extracellular-matrix storage', tissue specificity suprabasal epidermis, hair follicles and wounded skin [E3]. **Replication across two epithelial cell lines is not independent evidence when the failure mode is shared**: an over-expressed, tagged secretory protein in whole-cell lysate will carry the same abundant nuclear background in both lines. Two runs of one method are one method. Only two of the 13 partners in this beat replicate anywhere, and this is one of them; it still fails.
Passing a validation criterion is not the same as being validated
BioPlex names three confirmations: replication in the second cell line, reciprocal confirmation, and co-elution or co-fractionation. About 60% of its interactions in both networks are confirmed by one of these 2.
FGFBP1-KMT2A passes the first and cannot be tested on the second, because KMT2A is a prey-only node in BioPlex - all 31 rows touching Q03164 have KMT2A as prey, and the 3,969-residue protein is not among the 10,128 baits 1. The third was not checked here.
So the row has the best formal support of any BioPlex partner in this beat, and it is still a secreted extracellular protein paired with a nuclear methyltransferase.
The generalisation worth carrying out of this entry: in AP-MS, cell-line replication controls for cell-specific biology and for sampling; it does not control for a failure mode intrinsic to the reagent and the workflow. Tagged over-expression of a secretory protein, lysis of the whole cell, and affinity capture behave the same way in HEK293T and HCT116. Q128 reached the identical conclusion on menin-CD226 from two different laboratories and two different platforms, which is the stronger version of the same point.
What the partner is, and why there is no compartment in which this could happen
FGFBP1 (Q14512, 234 aa) is a small chaperone for FGFs in the extracellular matrix: it releases matrix-bound FGFs and enhances their mitogenic activity, with roles in tissue repair, angiogenesis and tumour growth. Both of its subcellular-location comments are extracellular - Secreted, extracellular space and Cell membrane, both ECO:0000269 - and it has no nuclear or cytoplasmic comment at any evidence level 3.
Its recorded tissue specificity is suprabasal epidermis, hair follicles, the dermo-epidermal basement membrane and wounded skin 3. Neither HEK293T (embryonic kidney) nor HCT116 (colon carcinoma) is that tissue. Both records are tagged over-expression.
What would overturn this. A documented intracellular or nuclear FGFBP1 pool in a haematopoietic cell, plus a KMT2A-bait or endogenous co-immunoprecipitation recovering FGFBP1. Neither exists, and no archive was queried for the first.
| partner a | gene-kmt2a |
|---|---|
| partner a uniprot | Q03164 |
| partner a length AA | 3969 |
| partner b | FGFBP1 |
| partner b uniprot | Q14512 |
| partner b length AA | 234 |
| verdict | DECLINED |
| adjudicated by | Q129, cycle 92, 2026-09-23 |
| source publication pmid | 33961781 |
| source dataset | BioPlex 3.0 |
| detection method | anti-tag co-immunoprecipitation (AP-MS) |
| experimental role partner | bait |
| experimental role kmt2a | prey |
| kmt2a was never the bait | true |
| expansion method | spoke expansion |
| intact miscore | 0.35 |
| replicated in second bioplex line | true |
| reciprocally confirmed | false |
| reciprocal confirmation possible | false |
| n publications | 1 |
| n methods | 1 |
| n cell lines | 2 |
| orthogonal method exists | false |
| role in this beat | the only BioPlex partner in the beat that satisfies BioPlex's own replication criterion, and it still fails on compartment - the calibration case |
| druggability | Not assessable; no residue or region is mapped on either side and the pair is declined. |
| absence class | never-measured |
| absence class note | No experiment has tested KMT2A against FGFBP1 with KMT2A as the bait, in any cell, and none has tested the pair in a haematopoietic cell. Given the compartment mismatch the useful statement is that the existing rows are uninterpretable rather than that an experiment is owed. |
| shown for fusion | NO - nobody has looked in any KMT2A fusion. Both records are full-length KMT2A as prey in non-haematopoietic epithelial lines. |
| epigenetic readout | none. |
host cell lines
[ "HEK293T embryonic kidney", "HCT116 colon carcinoma" ]
records
[ "EBI-54555390", "EBI-54852101" ]
compartment mismatch
{
"partner_compartments": [
"Secreted, extracellular space [ECO:0000269]",
"Cell membrane [ECO:0000269]"
],
"partner_nuclear_annotation": "none at any evidence level",
"partner_tissue": "suprabasal epidermis, hair follicles, wounded skin",
"host_lines": [
"HEK293T",
"HCT116"
],
"host_expresses_partner_natively": false
}interface residues a
[]
structures
[]
ligands
[ "drug-revumenib", "drug-ziftomenib" ]
- The transferable risk is methodological rather than therapeutic: if this project accepts within-platform replication as confirmation, it will admit exactly the class of edge shown here to be uninterpretable, and several intervention proposals in this graph rest on single-platform evidence.
- Read the other way, an over-strict rule discards real partners. FGF signalling is genuinely active in some leukaemias, and a reflex that deletes anything secreted would also delete a legitimate autocrine-loop hypothesis if one ever arrived with better evidence.
- How many edges in this graph rest on cell-line replication within a single platform, and are any of them load-bearing? Replication within one method is currently treated as confirmation in several entries and this pair shows what that is worth when the failure mode is shared.
- Does BioPlex's own CompPASS-Plus probability separate FGFBP1-KMT2A from WDR5-KMT2A? IntAct's pair-level miscore does (0.35 versus 0.96) but that score largely reflects accumulated literature rather than this experiment.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
IntAct curates two records of a KMT2A-FGFBP1 association from PMID 33961781 (BioPlex 3.0), both annotated to Supplementary table 1: EBI-54555390 in Homo sapiens HEK293T embryonic kidney cell and EBI-54852101 in Homo sapiens HCT116 colon carcinoma cell. Detection method anti-tag co-immunoprecipitation, interaction type 'association', expansion method SPOKE EXPANSION in both. FGFBP1 (Q14512) is the BAIT and KMT2A (Q03164) is the PREY in both. intact-miscore 0.35. All 31 BioPlex rows touching Q03164 have KMT2A as prey, so reciprocal confirmation is structurally unavailable.
IntAct molecular interaction database, REST findInteractions on Q03164 (KMT2A), 376 records retrieved 2026-09-23 by Q129 · IntAct (EMBL-EBI) 2026What this evidence cannot show
Read live from IntAct REST on 2026-09-23, not PSICQUIC (rule 93). The BioPlex supplementary table was not opened. Note that two IntAct records is not two publications: both come from PMID 33961781, and treating them as independent would be the exact error rule 93's sibling cases warn about. Spoke expansion means the binary pair was never observed in either line.
Evidence that argues against this (1)
- Recovery in two cell lines is a real observation and BioPlex treats it as confirmatory for good reasons - most shared interactions do sit in core complexes. Dismissing a replicated edge requires an argument about shared failure modes, which is what this entry supplies; it is an interpretation, not a fact in the record. Huttlin EL et al. 2021
- 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 (118,162 interactions among 14,586 proteins) plus 5,522 HCT116 immunoprecipitations, scored by CompPASS-Plus against a GFP control at 1% protein-level FDR with entropy-based dataset-level filtering. It names replication in the second cell line, reciprocal confirmation and co-elution/co-fractionation as its confirmations and reports that about 60% of interactions in both networks are confirmed by one of them, with median clone-wise reproducibility around 60% within the same cell line. Its stated headline is that shared interactions reside in core complexes while cell-specific interactions rewire between lines.
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 2021What this evidence cannot show
Full text read 2026-09-23 through the PMC HTML route (PMC8165030, 221,127 stripped characters) after fullTextXML returned HTTP 500. Neither 'KMT2A' nor 'FGFBP1' appears in the paper's text; the rows are supplementary-table entries not opened here. The paper's claim that shared interactions reside in core complexes is a population statement and is used here to show what the replication criterion is supposed to mean, not as a claim about this pair.
Evidence that argues against this (1)
- If BioPlex's own finding is correct - that shared interactions preferentially sit in core complexes - then a shared FGFBP1-KMT2A edge is evidence AGAINST this entry's verdict, because it places the pair in the population the authors consider most reliable. That is the strongest available argument for the row and it is the reason this decline is argued at length rather than asserted. Huttlin EL et al. 2021
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB Q14512 is 234 aa, primary gene name FGFBP1, recommended name 'Fibroblast growth factor-binding protein 1'. Its only subcellular-location comments are Secreted, extracellular space [ECO:0000269] and Cell membrane [ECO:0000269]; there is no nuclear or cytoplasmic comment at any evidence level. Function: carrier protein that releases FGFs from extracellular-matrix storage and enhances their mitogenic activity, including FGF2 signalling in tissue repair, angiogenesis and tumour growth. Tissue specificity: suprabasal epidermis, hair follicles, the dermo-epidermal basement membrane and wounded skin.
UniProtKB entry Q14512 (FGFBP1): primary gene name, sequence length, subcellular location, function and tissue specificity, retrieved 2026-09-23 · UniProt Consortium 2026What this evidence cannot show
UniProt location comments are curated pointers, not measurements (rule 91); the cited papers were not opened. 'No nuclear comment' is an absence in UniProt's record, not a published negative - no PubMed query was run on nuclear FGFBP1 in this beat, so the absence is unbounded in the literature sense.
Evidence that argues against this (1)
- FGFBP1 is transcriptionally induced in many carcinomas and is a well-studied tumour-associated gene, so it is not an obscure protein whose annotation is thin by neglect; the absence of a nuclear pool in its record carries more weight than it would for an uncharacterised protein. That argument cuts in favour of the decline, and is recorded here so that the counter-evidence field is not used to manufacture a doubt the record does not support. UniProt Consortium 2026
- related to →KMT2A / MLL1 — the gene that breaks
adjudicates the two IntAct KMT2A-FGFBP1 rows; DECLINED - replicated in HEK293T and HCT116 and the partner is still a secreted protein
the second experimentally secreted protein in the same 0.35 floor band; this one additionally passes BioPlex's own replication criterion