EAF2 (U19) — the EAF paralogue whose ELL-binding domain is a mapped tumour-suppressor module, described entirely in prostate cancer
also called EAF2 · U19 · TRAITS · BM040 · ELL-associated factor 2 · U19/Eaf2
EAF2 is the near-twin of EAF1 and binds the same partner, ELL, which is itself an MLL fusion partner. Its whole functional story was worked out in prostate cancer, where putting it back into cells kills them — and the part of the protein that does the killing is exactly the part that grips ELL.
The same thing, for a scientist
EAF2 (Q96CJ1, 260 aa) is the paralogue of EAF1 and the second ELL-family activation-module partner. Unusually for this beat, it has a **mapped structure-function result**: the region of U19/Eaf2 essential for apoptosis induction and growth suppression maps to **aa 68-113**, and that region is **necessary and sufficient for binding ELL**; co-expressing U19/Eaf2 with ELL in 293 cells increases cell death and growth suppression, so the authors conclude the ELL interaction is essential for the tumour-suppressive activity [E1]. UniProt independently annotates aa **17-104** as necessary for the ELL interaction, aa 177-260 as necessary for transactivation, and aa 246-260 as necessary for interaction with TCEA1 (TFIIS) and transactivation [E2]. The whole functional literature is **prostate cancer**: U19/Eaf2 is an androgen-response gene frequently down-regulated in advanced disease [E1]. In KMT2A-rearranged leukaemia there is essentially nothing — `(MLL OR KMT2A) AND EAF2` returns 5 records [E3] — and DepMap places EAF2 at the anti-dependency extreme (+0.1395, rank 17,372, pan-essentiality -0.0639), i.e. dispensable [E4].
This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
Why a prostate tumour suppressor is in a leukaemia interactome
Because of what it binds. ELL is the t(11;19)(q23;p13.1) MLL fusion partner, and the MLL-ELL fusion retains ELL's EAF-binding region. If EAF2's growth-suppressive activity is delivered through the ELL interaction 1, then an MLL-ELL fusion that titrates EAF proteins away from wild-type ELL is plausibly disabling a tumour suppressor as well as gaining an elongation function. That is the same sequestration logic the EAF1 entry records for Cajal bodies 5, arriving from a different direction.
It is a coherent idea and nobody has tested it in leukaemia. The evidence for it is: a prostate-cancer mapping study 1, an MLL-ELL mislocalisation study for the paralogue 5, and no experiment joining them.
The two ELL-binding boundaries do not agree, and this entry keeps both
| Source | ELL-interaction region | Basis |
|---|---|---|
| Hahn et al. 2007 | aa 68-113 | deletion mapping; necessary and sufficient for ELL binding and for apoptosis/growth suppression 1 |
| UniProt Q96CJ1 | aa 17-104 | curator annotation, 'necessary for interaction with ELL' 2 |
They overlap over aa 68-104 and disagree at both ends by roughly fifty residues at the N-terminus and nine at the C-terminus. This entry records both, attributed, rather than picking one, because a domain lane drawn from the wrong boundary is a silent error of the kind this project has been catching. The defensible common range is aa 68-104, and that is what data.domains carries.
Cell type, cell state, localisation, epigenetics
| Axis | Documented | Where |
|---|---|---|
| Cell type | Human prostate cancer cells and HEK293; no leukaemia system | 1 |
| Cell state | Androgen-responsive; expression frequently down-regulated in advanced prostate cancer; over-expression induces apoptosis and suppresses xenograft growth | 1 |
| Subcellular | Nucleus | 2 |
| Subnuclear | Not established. The Cajal-body result is for EAF1, and no paper places EAF2 there. | — |
| Epigenetic readout | None. No histone-modifying activity; EAF2 acts on ELL-dependent elongation and interacts with TFIIS (TCEA1). | 2 |
The TCEA1 (TFIIS) interaction at aa 246-260 is worth flagging: TFIIS is the anti-pausing factor whose binding NELF blocks in the paused complex. An ELL-partner protein that also contacts TFIIS sits at the junction of the two mechanisms this beat has been mapping, and no one has followed it up in leukaemia.
| uniprot | Q96CJ1 |
|---|---|
| gene symbol | EAF2 |
| protein name | ELL-associated factor 2 |
| length AA | 260 |
| alphafold ID | AF-Q96CJ1-F1 |
| accession resolution | UniProt REST query gene_exact:EAF2 AND organism_id:9606 AND reviewed:true, run 2026-09-19; one reviewed hit; primary gene name read back as EAF2 and length 260 confirmed from the same fetched record. |
| alias hazard | U19 and TRAITS are generic tokens; including them inflates the KMT2A co-occurrence count from 5 to 75. |
| paralogue | EAF1 |
| complex | ELL/EAF activation module |
| classification | ELL-family partner and candidate tumour suppressor; NOT a demonstrated KMT2A::AFF1 interactor and untested in leukaemia |
| classification reason | the only functional literature is prostate cancer; the leukaemia relevance is entirely by way of the ELL interaction |
| interaction evidence class | co-immunoprecipitation and deletion mapping (prostate/HEK293); UniProt curation |
| shown for fusion | none |
| epigenetic readout | none - no histone-modifying activity |
aliases retired or alternate
[ "U19", "TRAITS", "BM040" ]
complex partners
[ "gene-ell", "gene-ell2", "gene-eaf1" ]
domains — the numbers behind the chart
[
{
"name": "ELL-binding domain",
"start": 68,
"end": 104,
"source": "intersection of Hahn 2007 deletion mapping (aa 68-113, PMID 17044034) and UniProt Q96CJ1 (aa 17-104); both ranges recorded in prose, the intersection used here to avoid drawing a boundary neither source supports"
},
{
"name": "Transactivation region",
"start": 177,
"end": 260,
"source": "UniProt Q96CJ1, fetched 2026-09-19"
},
{
"name": "TCEA1 (TFIIS) interaction region",
"start": 246,
"end": 260,
"source": "UniProt Q96CJ1, fetched 2026-09-19"
}
]localization
{
"subcellular": "nucleus",
"subnuclear": "foci described as NUCLEAR SPECKLES by confocal microscopy of endogenous EAF2 and ELL (PMID 12446457), with NO speckle marker co-stained, so the assignment is morphological rather than marker-controlled. The marker-controlled Cajal-body assignment (PMID 12686606) is for EAF1, not EAF2. Corrected 2026-09-20: previously read 'not established'."
}essentiality
{
"depmap_scan": "analyses/depmap-kmt2a-selective-scan-full",
"pan_essentiality_all_lines": -0.0639,
"n_all_lines": 1208,
"mean_kmt2a_rearranged": 0.0711,
"mean_other_leukemia": -0.0684,
"kmt2a_selectivity_difference": 0.1395,
"rank": 17372,
"n_scanned": 17728,
"percentile": 97.992,
"interpretation": "dispensable and slightly growth-restraining, which is the expected direction for a tumour suppressor and means the assay is measuring the wrong thing for this gene"
}enumeration record counts — the numbers behind the chart
[
{
"query": "(MLL OR KMT2A) AND EAF2",
"records": 5
},
{
"query": "(MLL OR KMT2A) AND (\"U19/Eaf2\" OR \"U19/EAF2\")",
"records": 3
},
{
"query": "(MLL OR KMT2A) AND (EAF2 OR U19 OR TRAITS)",
"records": 75,
"note": "alias-inflated, discarded"
},
{
"date": "2026-09-19"
}
]depmap kmt2a selectivity
{
"chronos_mean_kmt2a_rearranged": 0.0711,
"chronos_mean_other_leukaemia": -0.0684,
"difference": 0.1395,
"pan_essentiality_all_lines": -0.0639,
"selectivity_rank": 17372,
"rank_denominator": 17728,
"provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
"how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}- EAF2 is a putative tumour suppressor. The therapeutic direction would be restoration, not inhibition, and restoring a nuclear protein is not a tractable small-molecule strategy.
- It is paralogue-buffered by EAF1, so even a successful restoration might be redundant.
- The entire functional case rests on over-expression phenotypes in prostate cells, which are among the least transferable results in cancer biology.
- Its androgen-responsiveness means any expression-based reasoning about EAF2 in a paediatric, pre-pubertal population is on very thin ground.
- Does EAF2 suppress growth in a KMT2A-rearranged leukaemia cell as it does in prostate cells, and is that suppression ELL-dependent? One restoration experiment in a KMT2A::AFF1 line would answer it, and no one has run it.
- Does MLL-ELL, or any KMT2A fusion, titrate EAF2 away from wild-type ELL? The paralogue result exists; the EAF2 result does not.
- Is EAF2 expression lower in KMT2A-rearranged leukaemia than in other leukaemias? A transcriptomic look costs nothing and has apparently never been done.
- EAF2 contacts TFIIS, the anti-pausing factor that NELF excludes from paused polymerase. Does the EAF2-TFIIS contact matter at fusion target genes?
- Which ELL-binding boundary is right, aa 68-113 or aa 17-104?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
U19/Eaf2 is an androgen-response gene frequently downregulated in advanced human prostate cancer; it interacts with ELL, a fusion partner of MLL in the t(11;19)(q23;p13.1) translocation in acute myeloid leukaemia; over-expression induces apoptosis and suppresses xenograft tumour growth; the region essential for apoptosis and growth suppression maps to amino acids 68-113 and this region is necessary and sufficient for binding ELL; and co-expression of U19/Eaf2 with ELL in 293 cells leads to a significant increase in cell death and growth suppression, indicating the ELL interaction is essential for these activities.
Apoptosis induction and growth suppression by U19/Eaf2 is mediated through its ELL-binding domain · Hahn J, Xiao W, Jiang F, Simone F, Thirman MJ, Wang Z · The Prostate 67(2):146-53 2007What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. Prostate cancer cells and HEK293; **no leukaemia experiment of any kind**. The apoptosis and growth suppression are over-expression phenotypes, which are prone to artefact, and 'necessary and sufficient' for binding is from deletion mapping rather than from a measured affinity.
Evidence that argues against this (2)
- UniProt's curated ELL-interaction region for EAF2 is aa 17-104, not 68-113. The two assignments overlap only over aa 68-104, so the claim that the apoptotic module and the ELL-binding module are the same region depends on which boundary is right, and this entry cannot resolve that from the abstract alone. UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
- SUBNUCLEAR LOCALIZATION IS NOT 'NOT ESTABLISHED'. Confocal microscopy of endogenous EAF2 shows it colocalising with ELL in a nuclear speckled pattern. The caveat that matters is that no speckle marker (SON, SRRM2, SC35) was co-stained, so 'speckles' there is a description of appearance rather than a marker-controlled assignment - and a sister beat records that the same ambiguity produces a live Cajal-body-versus-speckle contradiction for ELL. Simone F et al. 2002
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB Q96CJ1 (EAF2_HUMAN, ELL-associated factor 2) is 260 aa and annotates aa 17-104 as necessary for interaction with ELL, aa 177-260 as necessary for transactivation activity, aa 246-260 as necessary for interaction with TCEA1 and for transactivation, and disordered regions at aa 116-154 and 170-234.
UniProtKB Q96CJ1 (EAF2_HUMAN), ELL-associated factor 2 · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026What this evidence cannot show
Record fetched 2026-09-19; primary gene name read back as EAF2 and length 260 confirmed from the same fetched record, and all coordinates copied from it. These are curator aggregations across primary papers and can move between releases. Note the asymmetry with the paralogue: UniProt annotates an ELL-interaction region for EAF2 but none for EAF1.
Evidence that argues against this (1)
- The deletion-mapping paper places the ELL-binding region at aa 68-113, extending nine residues past UniProt's C-terminal boundary and starting fifty-one residues later. A domain lane drawn from either set alone would be wrong at one end, which is why this entry carries the intersection (aa 68-104) in its structured data and both ranges in prose. Hahn J et al. 2007
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
As of 2026-09-19, PubMed returns 5 records for '(MLL OR KMT2A) AND EAF2' and 3 for '(MLL OR KMT2A) AND ("U19/Eaf2" OR "U19/EAF2")'. The broader query '(MLL OR KMT2A) AND (EAF2 OR U19 OR TRAITS)' returns 75, an inflation driven by the generic aliases U19 and TRAITS rather than by relevant literature.
PubMed E-utilities esearch record counts for EAF2 and its aliases against KMT2A and MLL, run for this beat · NCBI PubMed, queried 2026-09-19 2026What this evidence cannot show
The 75-record figure was not title-checked; 'inflation driven by generic aliases' is an inference from the fifteen-fold jump and from the fact that U19 and TRAITS are both ordinary English-looking tokens. This is the second alias-precision failure in this beat, after NELFE's 'RD'.
Evidence that argues against this (1)
- Not reading the 75 titles leaves open the possibility that a genuine EAF2 leukaemia paper is indexed under U19 only. The project's standing instruction is to search retired symbols precisely because older functional characterisation hides there, and U19 is exactly such a symbol - so discarding that query without inspection trades one known error mode for another. NCBI PubMed, queried 2026-09-19; finding originally made by this project's non-elongation-partners beat, cycle 40 2026
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
In the project's genome-wide KMT2A-rearranged selectivity scan of 17,728 genes, EAF2 ranks 17,372 (97.992nd percentile) with a difference of +0.1395 (KMT2A-rearranged mean +0.0711 against other-leukaemia mean -0.0684) and a pan-essentiality of -0.0639 across 1,208 lines - dispensable, and if anything less required in KMT2A-rearranged lines. Its paralogue EAF1 scores +0.0469 (rank 14,482).
Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026What this evidence cannot show
Rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19. EAF1 and EAF2 are paralogues and neither single knockout is informative about EAF function; no double knockout is available. Eleven KMT2A-rearranged lines, AML-heavy, none infant-derived.
Evidence that argues against this (1)
- A tumour suppressor should score positive in a proliferation assay - losing it helps cells grow - so EAF2's positive gene effect is consistent with the prostate literature rather than contradicting it, and reading the DepMap number as 'EAF2 is irrelevant' would be backwards. The assay is measuring the wrong direction for this gene. Hahn J et al. 2007
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Transfection of MLL-ELL delocalises the paralogue EAF1 from a nuclear speckled distribution to a diffuse nucleoplasmic pattern, and EAF1 speckles are undetectable in leukaemic cell lines from mice transplanted with MLL-ELL-transduced bone marrow; the authors conclude that expression of the MLL-ELL fusion protein may have a dominant effect on the normal protein-protein interactions of ELL.
EAF1, a novel ELL-associated factor that is delocalized by expression of the MLL-ELL fusion protein · Simone F, Polak PE, Kaberlein JJ, Luo RT, Levitan DA, Thirman MJ · Blood 98(1):201-9 2001What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. **This result is for EAF1, not EAF2**, and is carried here only to state the sequestration idea that would connect EAF2's tumour-suppressor function to MLL-ELL. No equivalent experiment has been done with EAF2.
Evidence that argues against this (1)
- Transferring an EAF1 localisation result to EAF2 is exactly the paralogue conflation this beat criticised in the BCOR/BCORL1 case. EAF1 and EAF2 differ enough that UniProt annotates an ELL-interaction region for one and not the other, and the sequestration idea remains untested for EAF2. UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
ELL-binding domain; aa 68-113 by deletion mapping (necessary and sufficient), aa 17-104 by UniProt curation
paralogues; UniProt annotates an ELL-interaction region for EAF2 and none for EAF1, and only EAF1 has a Cajal-body phenotype
the second EAF-family partner of the ELL/EAF activation module
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
NOT an interactor and untested in leukaemia: 5 PubMed records with MLL/KMT2A, all functional work in prostate cancer
the U19/TRAITS alias inflation (75 records against 5)