The super elongation complex, sampled from the RNA side: ELL, ENL, CDK9 and cyclin T1 are all in the proximity record - ELL2, EAF1/2, AF9 and AF10 are at zero
also called SEC proximity recovery · SEC BioID census
Using a public database we counted, for every part of the elongation machine, how many tagging experiments have ever found it sitting nearby. Some parts appear repeatedly; several appear not once.
The same thing, for a scientist
Across the K562 50-bait BioID2 deposition, recovery as prey is: **CCNT1 16, CDK9 11, ELL 10, MLLT1/ENL 7, AFF1 4, AFF4 4**, and **zero for ELL2, ELL3, EAF1, EAF2, MLLT3/AF9, MLLT10/AF10, AFF2 and AFF3** [E1]. Widening to the entire IntAct proximity record (PSI-MI MI:1314, all publications) adds only single rows: AFF4 from a MAC-tag histone H3.1 bait, ELL twice from a CDK7 bait, MLLT1 from an H3.1 bait and from an SP7 bait, MLLT10 three times from transcription-factor baits and once from a centrosome bait, CDK9 from a PPP5C bait and twice from a CFTR bait [E2]. **MLLT3/AF9 and ELL2 have never appeared in any proximity-labelling record at all** [E2]. And across all five contributing publications, **no SEC component has ever been used as a bait** [E2].
The table
| protein | baits in the 50-RBP K562 map | all IntAct proximity rows | ever a bait? |
|---|---|---|---|
| CCNT1 | 16 | 16 | no |
| NELFE | 13 | 13 | no |
| CDK9 | 11 | 14 | no |
| ELL | 10 | 12 | no |
| MLLT1 (ENL) | 7 | 9 | no |
| PAF1 | 6 | 6 | no |
| AFF1 | 4 | 4 | no |
| AFF4 | 4 | 5 | no |
| DOT1L | 4 | 6 | no |
| MEN1 | 3 | 4 | no |
| MLLT10 (AF10) | 0 | 4 | no |
| MLLT6 | 1 | 1 | no |
| ELL2 | 0 | 0 | no |
| EAF1, EAF2 | 0 | 0 | no |
| MLLT3 (AF9) | 0 | 0 | no |
| AFF2, AFF3 | 0 | 0 | no |
All counts retrieved from IntAct on 2026-09-19 12. The method filter passes its control: 64,945 MI:1314 rows exist in total and 259 for LMNA, so the zeros are measurements.
What the zeros do and do not mean
ELL2's zero is the informative one. ELL is recovered by ten baits and its close paralogue ELL2 by none, in the same cells with the same fifty baits. ELL2 is the ELL-family member most associated with the SEC in the KMT2A-fusion literature. Either K562 expresses ELL far above ELL2, or the two occupy different neighbourhoods. The dataset cannot separate those, and expression was not checked here.
AF9's zero is the one that bites this graph. MLLT3/AF9 is a major KMT2A fusion partner and the protein through which a large part of this project's imported mechanism runs. It appears in no proximity-labelling experiment anywhere in IntAct 2. Anything this graph says about AF9's transient partners still rests entirely on affinity purification.
EAF1/EAF2's zero is uninformative. Both are small and lowly expressed; this project has previously measured their literature as thin and their aliases as unusable, and a zero at the detection floor is not a negative result.
Reciprocity: there is none, anywhere
Every SEC row in every proximity publication has the SEC protein in the prey column (PSI-MI MI:0498). No experiment has put a biotin ligase on AFF1, AFF4, ELL, ELL2, MLLT1, MLLT3, MLLT10, CDK9, CCNT1, KMT2A, MEN1 or DOT1L 2.
So every proximity fact in this beat is one-directional, and the single cheapest experiment that would change the picture is any SEC bait in any cell - which is what exp-turboid-map-of-kmt2a-aff1 proposes and what nobody has done.
| script | none (database query) |
|---|---|
| status | run |
| n sec baits ever | 0 |
| result summary | The SEC is present in the curated proximity record only as prey, with CCNT1, CDK9, ELL and ENL recovered repeatedly and ELL2, EAF1/2, AF9, AF10, AFF2 and AFF3 at or near zero. No SEC component has ever been a proximity bait. |
inputs
[ "IntAct PSICQUIC detmethod:\"MI:1314\" intersected with accessions", "IntAct PSICQUIC pubid:39251607" ]
recovery counts k562 map — the numbers behind the chart
{
"CCNT1": 16,
"NELFE": 13,
"CDK9": 11,
"ELL": 10,
"MLLT1": 7,
"PAF1": 6,
"AFF1": 4,
"AFF4": 4,
"DOT1L": 4,
"MEN1": 3,
"MLLT6": 1,
"ELL2": 0,
"EAF1": 0,
"EAF2": 0,
"MLLT3": 0,
"MLLT10": 0,
"AFF2": 0,
"AFF3": 0
}all intact proximity rows — the numbers behind the chart
{
"KMT2A": 46,
"CDK9": 14,
"ELL": 12,
"MLLT1": 9,
"DOT1L": 6,
"AFF4": 5,
"AFF1": 4,
"MEN1": 4,
"MLLT10": 4,
"MLLT3": 0,
"ELL2": 0
}contributing publications
[ "39251607", "35140242", "36931259", "29568061", "26638075", "35384245", "28330616", "36012204" ]
control
{
"total_MI_1314_rows": 64945,
"LMNA_P02545_rows": 259
}- Four of these zeros are for proteins the graph treats as mechanistically central. A reader could take them as evidence those proteins have no transient partners; they are evidence that nobody has looked from this direction.
- Because there is no reciprocal experiment anywhere, every proximity statement in this beat could be reversed by one bait construct that does not yet exist.
- Is ELL2's zero an expression effect? Measuring ELL and ELL2 protein in K562 would decide between 'different neighbourhood' and 'below detection'.
- AF9 has never been in any proximity experiment. Given how much of this graph's mechanism is imported from AF9 systems, is that the single most consequential gap in the beat?
- Would an AFF1 or AFF4 bait recover the same four or five RBPs reciprocally, or a different set entirely?
- Does the Human Cell Map, which IntAct does not serve, contain any of these proteins as prey and would it change the zeros?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
In the K562 50-bait BioID2 deposition, prey recovery counts are CCNT1 16, NELFE 13, CDK9 11, ELL 10, MLLT1 7, PAF1 6, AFF1 4, AFF4 4, DOT1L 4, MEN1 3, MLLT6 1, and zero for ELL2, EAF1, EAF2, MLLT3, MLLT10, AFF2 and AFF3.
IntAct PSICQUIC query: pubid:39251607, full 19,403-row deposition (IMEx IM-30059), retrieved 2026-09-19 · IntAct molecular interaction database (EMBL-EBI) 2026What this evidence cannot show
Retrieved 2026-09-19 from the authors' curated bona fide pair list, which has an unstated filtering threshold, so all zeros are zeros after filtering. Counts are abundance-weighted: a lowly expressed protein can be absent for reasons unrelated to its neighbourhood, and expression in K562 was not checked. K562 has no KMT2A rearrangement.
Evidence that argues against this (2)
- In the largest human BioID survey (192 baits, HEK293) a prey's bait profile is dominated by subcellular compartment rather than by binding, so co-recovery of nuclear proteins by nuclear baits is the expected background. Go CD et al. 2021
- An absence of testing is not a negative result. The fifty baits are RNA-binding proteins chosen for a post-transcriptional study; no bait was chosen because it would sample the elongation complex, so a zero says the RBP-side sampling did not reach that protein, not that the protein has no neighbours. Khoroshkin M et al. 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Across the whole IntAct proximity record (detection method MI:1314, all publications), rows exist for KMT2A (46), CDK9 (14), ELL (12), MLLT1 (9), DOT1L (6), AFF4 (5), AFF1 (4), MEN1 (4) and MLLT10 (4), and none at all for ELL2 or MLLT3; in every row the protein is in the prey role, and no super elongation complex component, KMT2A-complex component or fusion protein has ever been used as a bait. The filter passes its control with 64,945 MI:1314 rows in total and 259 for LMNA (P02545).
IntAct PSICQUIC query: detmethod:"MI:1314" (proximity-dependent biotin identification) intersected with the UniProt accessions of the super elongation complex, KMT2A, MEN1 and DOT1L, retrieved 2026-09-19 · IntAct molecular interaction database (EMBL-EBI) 2026What this evidence cannot show
IntAct is one database with one curation policy and does not serve every proximity resource - notably the Human Cell Map (192 BioID baits, HEK293) is absent. These numbers therefore bound IntAct's holdings, not the world's proximity data, and the correct claim is 'of the curated proximity record'.
Evidence that argues against this (1)
- The Human Cell Map alone contains 192 BioID baits in HEK293 and is not served through this query, so an IntAct-bounded zero can only be a statement about curation coverage. A complete statement would require querying that resource, which has no REST endpoint found on 2026-09-19. Go CD et al. 2021
ELL recovered by ten baits; ELL2 by none
CDK9 recovered by eleven baits and cyclin T1 by sixteen
- related to →MLLT3 / AF9 — a famous fusion partner that is not a dependency, and a hub domain that is
MLLT3/AF9 has never appeared in any proximity-labelling record
MLLT1/ENL recovered by seven baits
principal data source
no SEC component has ever been a proximity bait, which is what that experiment proposes
same proximity record, KMT2A side
second proximity dataset in the same census
contributes to the SEC recovery census
contributes to the SEC recovery census
contributes to the SEC recovery census
contributes to the SEC recovery census
contributes to the SEC recovery census
the census behind the verdict
ILF3 is one of the baits behind the ELL recovery count in that analysis