AFF4-MED10: AFF4 is prey in a 2004 Mediator MudPIT purification - a lead with an independent mechanistic reason to be true, and no experiment that tests it
also called AFF4-MED10 · AFF4-NUT2 · SEC-Mediator · AFF4-Mediator
AFF4, the scaffold of the elongation complex that KMT2A fusions hijack, turned up as a passenger in a 2004 purification of the Mediator coactivator. There is a known route by which that could be real, and nobody has ever tested it directly.
The same thing, for a scientist
**Verdict: ADMITTED AS A LEAD, on corroboration rather than on the record.** IntAct holds one AFF4-MED10 record: anti-tag co-IP, *association*, HeLa, AFF4 as prey, intact-miscore 0.35, from **PMID 15175163** [E1]. That paper is Sato et al., *Mol Cell* 2004, a MudPIT comparison of six immunoaffinity Mediator preparations purified through **MED10 (Nut2), MED25, MED29, MED19, MED28 and MED26** to define a consensus mammalian Mediator [E2]. AFF4 is not its subject and does not appear in its abstract. **On the record alone this is a single spoke in a 2004 complex purification and would be declined.** It is admitted because of what sits beside it: `gene-med26` is already in this graph, and **MED26 is the Mediator subunit that docks elongation factors** - the Takahashi 2011 result this corpus already carries. One of the six baits in this very purification is MED26. So a super-elongation-complex scaffold appearing in a Mediator preparation is not a surprising contaminant; it is the predicted consequence of a documented Mediator-elongation interface, recovered incidentally seven years before that interface was described.
Why the corroboration argument is allowed here and not elsewhere
This beat declines most single-record prey rows. The difference is the direction of the reasoning.
A typical decline runs: one row, one method, in a cell that does not express the bait, with a compartment mismatch - the row is explained better by an artefact than by biology. Here there is a specific, independently established mechanism that predicts the row: Mediator contacts the super elongation complex through MED26, gene-med26 exists in this graph for that reason, and MED26 is among the six baits Sato et al. purified through.
That is corroboration, not confirmation, and the distinction matters. The MED26-SEC interface being real does not make this row real; the row could still be a MudPIT carry-over. What it does is remove the usual alternative explanation, which is that a nuclear scaffold in an AP-MS pellet is background. A background hypothesis that is also the mechanistic hypothesis cannot discriminate, so the honest verdict is a lead with a named experiment, not an admission.
And the part that is genuinely weak: MED10 is a 135 aa Mediator head-module subunit whose only UniProt subcellular location is Nucleus, and that by curator inference (ECO:0000305), not experiment 3. A 135 aa protein is at the small end where AP-MS recovery is poor and where this project has already had to control for length. There is no reason to think the AFF4 contact is with MED10 specifically rather than with the complex MED10 was used to pull.
Cell type, cell state and what is missing
Cell type: HeLa, from nuclear extract. Cell state: unstimulated, asynchronous - nothing is manipulated. No haematopoietic cell, no KMT2A fusion, no elongation readout.
The experiment that would settle it is cheap and specific: immunoprecipitate AFF4 from a KMT2A::AFF1 cell line and blot for MED26 and MED1, with the reciprocal. If the super elongation complex and Mediator co-occupy the fusion's target loci, that is a second coactivator surface at HOXA that no menin inhibitor touches.
Cross-family note. Mediator-elongation coupling is not specific to KMT2A fusions. MED12 and MED23 mutations recur in other leukaemias, and the NUP98 fusions that phenocopy KMT2A-r HOXA activation do so through condensate formation rather than through a documented Mediator contact. If the SEC-Mediator link is load-bearing for KMT2A::AFF1 and absent in NUP98::NSD1, that is a discriminator between two fusions this corpus repeatedly treats as analogous.
| partner a | gene-aff4 |
|---|---|
| partner b | MED10 |
| partner b uniprot | Q9BTT4 |
| partner b length AA | 135 |
| partner a uniprot | Q9UHB7 |
| partner a length AA | 1163 |
| verdict | ADMITTED AS A LEAD |
| adjudicated by | q160, cycle 146, 2026-09-24 |
| druggability | Not assessable. No region is mapped on either side. The adjacent druggable object is the SEC itself via CDK9, already covered by protein-ptefb-cdk9-ccnt1. |
| n methods | 1 |
| n platforms | 1 |
| n publications | 1 |
| endogenous side | no - immunoaffinity purification through a tagged Mediator subunit |
| intact miscore | 0.35 |
| cell state | unstimulated, asynchronous; nothing manipulated |
| epigenetics | None directly. The relevance is transcriptional rather than epigenetic: Mediator-elongation coupling at the loci the fusion activates. |
| corroborating mechanism | MED26 docks elongation factors (PMID 21729782, already carried by this graph as gene-med26), and MED26 is one of the six baits in this same purification. |
| absence class | never-measured |
| absence class note | No AFF4 immunoprecipitation in any KMT2A-rearranged cell has been blotted for a Mediator subunit. The experiment is a co-IP and two antibodies. |
interface residues a
[]
structures
[]
ligands
[]
records
[ "AFF4-MED10, anti tag coip, association, HeLa, AFF4 prey, miscore 0.35, PMID 15175163" ]
cell systems
[ "HeLa nuclear extract" ]
localization
{
"compartment": "nucleus",
"subnuclear": [],
"evidence_code": "ECO:0000305 (curator inference)",
"shown_for_fusion": "NO - nobody has looked in any fusion. The record is wild-type AFF4 in HeLa."
}measurements
[
{
"quantity": "count",
"subject": "IntAct records for AFF4|MED10",
"value": 1,
"units": "records",
"n": 171,
"universe": "all IntAct REST records for Q9UHB7, 2026-09-24",
"null": null,
"note": "One of 171 AFF4 records; AFF4 has 47 prey-only partners in the PSICQUIC grading and more under REST."
}
]- Mediator is required for essentially all Pol II transcription. Any intervention aimed at a Mediator-SEC interface has no therapeutic window argument available to it, and the MED12/MED13 kinase-module inhibitors that exist are broadly cytotoxic.
- Admitting this row on corroboration rather than on the record sets a precedent this beat should be uncomfortable with: the corroborating mechanism also predicts the artefact, so the same reasoning would admit a false row.
- Does AFF4 immunoprecipitated from a KMT2A::AFF1 line recover MED26 or MED1? One co-IP answers it, and no such blot exists anywhere in this corpus.
- If Mediator and the SEC co-occupy HOXA in a KMT2A-rearranged blast, is that a coactivator surface a menin inhibitor leaves intact? Every resistance mechanism in this graph so far is menin-proximal; this would be a menin-independent one.
- Is the SEC-Mediator link present in NUP98::NSD1, which phenocopies the HOXA program by a different route? A difference there would be the first mechanistic discriminator between two fusions this corpus treats as analogous.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
IntAct REST returns one AFF4-MED10 (Q9UHB7-Q9BTT4) record: detection method anti-tag co-immunoprecipitation, interaction type 'association', host Homo sapiens HeLa epitheloid cervical carcinoma cell, AFF4 recorded as PREY, intact-miscore 0.35, from PMID 15175163. It is the only record for this pair.
IntAct molecular interaction database, REST findInteractions on Q9UHB7, retrieved 2026-09-24 by q160 (partner MED10) · IntAct (EMBL-EBI) 2026What this evidence cannot show
One record, one publication, one method. IntAct does not flag this row with an expansion method, but a six-bait MudPIT complex purification reports all recovered proteins against the tagged subunit, so it is spoke-like in practice whether or not it is labelled as such.
Evidence that argues against this (1)
- A 2004 MudPIT preparation is the kind of experiment that recovers abundant nuclear proteins non-specifically, and the paper's own framing is that it is comparing six preparations to find which subunits are CONSENSUS - i.e. it is explicitly aware that individual preparations carry non-consensus material. AFF4 is by definition in the non-consensus fraction. Sato S et al. 2004
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
PMID 15175163 (Sato et al., Mol Cell 2004) uses multidimensional protein identification technology to compare six Mediator preparations immunoaffinity-purified through their Nut2 (MED10), MED25 (Med25), Intersex (MED29), LCMR1 (MED19), AK007855 (MED28) or CRSP70 (MED26) subunits, in order to define a set of consensus mammalian Mediator subunits, and additionally identifies CDK11 and MED13L as Mediator-associated. Neither AFF4 nor the super elongation complex appears in its abstract.
A set of consensus mammalian mediator subunits identified by multidimensional protein identification technology. · Sato S, Tomomori-Sato C, Parmely TJ, Florens L, Zybailov B, Swanson SK, Banks CAS, Jin J, Cai Y, Washburn MP, Conaway JW, Conaway RC · Mol Cell 2004What this evidence cannot show
Abstract read in full; the MudPIT protein lists were not opened, so the AFF4 identification rests entirely on IntAct's curation of the supplementary data. `expected_absent_terms` records that AFF4 is absent from the abstract, which is the point: the paper is not about this pair.
Evidence that argues against this (1)
- That MED26 is one of the six baits cuts both ways. It supports the mechanistic reading, and it also means the AFF4 recovery may come from the MED26 preparation rather than the MED10 one, in which case attaching the row to MED10 is an artefact of how IntAct assigned the pair. Sato S et al. 2004
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB Q9BTT4 (MED10_HUMAN) is 135 aa. Its only subcellular location comment is Nucleus, and it carries evidence code ECO:0000305 - curator inference - not ECO:0000269 experimental evidence. No tissue-specificity comment is recorded.
UniProtKB entry Q9BTT4, subcellular location, length and tissue specificity, retrieved 2026-09-24 · UniProt Consortium 2026What this evidence cannot show
Rule 91 clause 4: ECO:0000305 is curator inference, not a measurement. For a Mediator head-module subunit a nuclear call is not controversial, but it is worth recording that this graph's localization field for MED10 would be carrying an inference rather than an experiment.
Evidence that argues against this (1)
- Absence of an experimental localization code for a core Mediator subunit is far more likely to reflect that nobody bothered to publish the immunofluorescence than that the localization is uncertain. Reading ECO:0000305 as doubt would be over-reading. UniProt Consortium 2026
AFF4 prey in a MED10 Mediator MudPIT preparation, HeLa, miscore 0.35
MED26 docks elongation factors and is one of the six baits in the same purification - the corroborating mechanism
- related to →Plain-English: what transcription elongation is, and why a traffic light explains this leukemia
a second coactivator surface at the fusion's targets that no menin inhibitor touches
one of the 135 uncovered human prey-only partners