Oncofusion

start / graph / gene-epc1

Genestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q123-the-architecture-from-other-parts-analogy

EPC1 - the NuA4/TIP60 subunit whose C-terminal fragment grips MBTD1, making it the contact a leukaemic fusion cannot do without and the mildest dependency in that whole complex

also called Epl1 · enhancer of polycomb 1 · enhancer of polycomb homolog 1

EPC1 is the part of a gene-activating machine that holds on to MBTD1, and a rare leukaemia fusion protein uses exactly that handhold - break the grip and the fusion stops causing leukaemia. Unlike the machine's other parts, cells can tolerate losing EPC1 reasonably well, which is the only reason it is worth mentioning as a possible drug target.

The same thing, for a scientist

**EPC1 is UniProt Q9H2F5 (EPC1_HUMAN, 836 aa), HGNC:19876 at 10p11.22, primary Ensembl id ENSG00000120616, alias `Epl1`, no previous symbol** [E1]. It is the non-catalytic NuA4/TIP60 subunit that enlists MBTD1 into the complex: a hydrophobic C-terminal fragment of EPC1 engages MBTD1's MBT repeats at a site **distinct from** the H4K20me-binding cage [E2]. In PDB 6NFX at 1.95 A the single 470-residue engineered polymer entity aligns as **MBTD1 Q05BQ5 140-562 joined through a designed linker to EPC1 Q9H2F5 644-671** - spans derived here by aligning the fetched entity sequence against UniProt FASTA downloaded the same session rather than read off the deposit's own numbering [E2]. **Why it earns an entry in this graph.** The ZMYND11::MBTD1 leukaemic fusion **requires** this contact. Deleting MBTD1's four MBT repeats, or the triple substitution A661D/L684D/L688D in fusion numbering, abolishes both Tip60 binding and transformation of murine haematopoietic progenitors - while the H4K20me1/2 cage mutants transform normally [E3]. By the published construct offset those three residues are MBTD1 **A236, L259 and L263**, inside MBT repeats 1 and 2, and the residue identities were checked against the live Q05BQ5 sequence in this beat and match [E4]. **So the fusion's load-bearing protein-protein contact is MBTD1-EPC1, and EPC1 is the half of it that is not in the fusion.** **And it is the only node in this pathway that looks like a target rather than a liability.** Open Targets dataVersion 26.06: EPC1 mean gene effect **-0.336 across 1,258 screens, 22.7% of lines below -0.5, 1.35% below -1.0** - against KAT5 at -0.607 with 70.2% below -0.5 and EP400 at -0.932 with 94.3% below -0.5 [E5]. 'Inhibit the hijacked acetyltransferase' is an essential-function proposal; 'break the MBTD1-EPC1 grip' is, on these numbers, not obviously one. That is a statement about a tolerance margin and not a drug-discovery plan, and the bite-back field says why. **A second reason it is interesting, and it is the reason this beat found it.** EPC1 is itself a recurrent **5' fusion partner** in low-grade endometrial stromal sarcoma - EPC1-PHF1 - in the same fusion family as MBTD1::EZHIP, which Piunti et al. characterise as N-terminal NuA4 component fused to C-terminal Polycomb subunit [E6]. So EPC1 appears twice in this story: as the NuA4 anchor the leukaemic MBTD1 fusion grips, and as a NuA4 component fused to Polycomb in the sarcoma family MBTD1's other fusion belongs to. **Those are two different facts about one protein and their coincidence is not evidence of anything**; `analogy-mbtd1-occupies-both-fusion-roles-polarity-not-parts` says so explicitly rather than building on it. **Coverage verdict, pasted.** `python scripts/is-symbol-covered.py EPC1`, run alone, 2026-09-23: ``` EPC1 no match under 5 candidate names ```

What it looks like
EPC1 - the NuA4/TIP60 subunit whose C-terminal fragment grips MBTD1, making it the contact a leukaemic fusion cannot do without and the mildest dependency in that whole complex836 aa
MBTDMBTD1-binding C-terminal fragment (the span present in PDB 6NFX) · residues 644–6710200400600800
loading 6NFX…
PDB 6NFX ↗
shape

This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.

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.

Go deeper
The contact, in residues, and what the structure can and cannot say

The published statement is that a hydrophobic C-terminal fragment of EPC1 engages the MBT repeats of MBTD1 at a site distinct from the H4K20me-binding site, with cellular assays validating the key residues, and that EPC1 is the subunit that enlists MBTD1 into NuA4/TIP60 2.

The residue spans were derived rather than read. PDB 6NFX has one polymer entity, 470 residues, described as 'MBT domain-containing protein 1,Enhancer of polycomb homolog 1' and cross-referenced to both Q05BQ5 and Q9H2F5. Aligned against UniProt FASTA fetched the same session: entity 2-424 matches MBTD1 140-562 and entity 443-470 matches EPC1 644-671, with a designed linker between 2. So the EPC1 fragment in the structure is a 28-residue peptide out of an 836-residue protein.

What that cannot establish. 6NFX is a single-chain engineered chimera, not a two-chain complex. It models the bound state; it does not measure association, affinity, stoichiometry or whether the contact forms at endogenous concentrations. This graph has already had to correct an entry that described a single-chain deposit as a co-structure, so the caveat is house policy and not scepticism. The evidence that the contact matters is the mutagenesis in the leukaemia paper, not the structure 3.

One thing the structure does establish cleanly, and it is load-bearing for two other entries. The EPC1 site and the H4K20me1/2 cage are different sites on the same MBT block 2. That is why the leukaemic fusion can require the MBT repeats while not requiring the cage, and it is why 'MBTD1 recruits NuA4 through its methyl-lysine reader' - a sentence that would read naturally and appears in shortened form in secondary sources - is wrong.

The dependency numbers, and the three ways they could mislead
genemean gene effectn screensbelow -0.5below -1.0
EPC1-0.3361,25822.7%1.35%
MBTD1-0.3531,25817.9%0.24%
KAT5/TIP60-0.6071,25870.2%-
EP400-0.9321,25894.3%-

Open Targets dataVersion 26.06, fetched 2026-09-23, genes resolved through the HGNC primary Ensembl id. EPC1 and MBTD1 were re-fetched in this beat; the KAT5 and EP400 figures are the ones gene-mbtd1 records from the same endpoint and reproduce 5.

Three ways to misread this table.

One: a DepMap number is not lineage-split. None of these figures says anything about haematopoietic lines specifically, and this project has already had two selectivity claims turn out to be lineage artefacts. A mean over 1,258 lines is a statement about cells in general.

Two: 22.7% below -0.5 is not 'safe'. It means roughly one cell line in four suffers measurably when EPC1 is knocked out. The comparison to KAT5's 70.2% is a comparison of two bad options, and the honest reading is 'less pan-essential', not 'dispensable'.

Three: the target is a protein-protein interface, not a protein. Every number here is from a complete knockout. A hypothetical agent breaking the MBTD1-EPC1 contact would leave both proteins intact and leave NuA4/TIP60 otherwise assembled, so the knockout figures are an upper bound on the damage and tell you nothing about the actual perturbation. They are quoted because they bound the worst case, and for no other reason.

Facts and figures
uniprotQ9H2F5
length AA836
gene primaryEPC1
hgncHGNC:19876
hgnc nameenhancer of polycomb 1
ensemblENSG00000120616
cytoband10p11.22
accession resolutiongene_exact:EPC1 AND organism_id:9606 AND reviewed:true returned exactly one row, Q9H2F5, with EPC1 as the primary gene name; HGNC:19876 independently carries Q9H2F5 and ENSG00000120616. Both fetched live 2026-09-23. The rule-62 several-rows trap does not fire.
structure note6NFX is a SINGLE-CHAIN ENGINEERED CHIMERA at 1.95 A with one polymer entity of 470 residues, cross-referenced to Q05BQ5 and Q9H2F5. Aligned against UniProt FASTA fetched the same session: entity 2-424 = MBTD1 140-562; entity 443-470 = EPC1 644-671. The span was derived by alignment, not read off the deposit's numbering.
complex membershipNuA4/TIP60 histone acetyltransferase complex, as the non-catalytic subunit that enlists MBTD1. NOT a Polycomb-complex subunit - the protein name 'Enhancer of polycomb homolog 1' is a Drosophila-homology name and is a naming hazard of the same family as BICRA's 'BRD4-interacting...' name.
relationship to the index fusionNONE DIRECT. EPC1 has no reported interaction with KMT2A, menin, LEDGF, DOT1L or any SEC component, and no KMT2A fusion involves it. It is here because it is the half, not present in the fusion, of the protein-protein contact that the ZMYND11::MBTD1 leukaemic fusion requires - and because that makes it the analogue in this architecture of the AHD surface a KMT2A fusion's 3' partner uses.
epigenetic readoutNo catalytic activity of its own. As a NuA4/TIP60 subunit it contributes to deposition of histone H4 and H2A acetylation, and its specific documented job is recruiting MBTD1, which brings an H4K20me1/2 reader to the complex.
cell typeBroadly expressed, by the DepMap screen count of 1,258 lines; no cell-type-specific function is asserted here and none was retrieved. The leukaemia-relevant setting is murine haematopoietic stem/progenitor cells, where the MBTD1-EPC1 contact is required for fusion-driven transformation.
cell stateNOT CHARACTERISED in this entry. No self-renewal, differentiation or quiescence phenotype is attributed to EPC1 here; the state statements in this story belong to MBTD1 (HSC quiescence) and to the fusion (self-renewal with differentiation block).
subcellular localizationNOT MEASURED in this beat and not asserted. Nuclear by inference from NuA4/TIP60 membership; no imaging, fractionation or UniProt localisation comment was retrieved, and under rule 91 an unretrieved annotation is recorded as unretrieved.
subnuclear localizationNOT MEASURED.
shown for fusionZMYND11::MBTD1 - the MBTD1-EPC1 contact is required for transformation, shown by three MBTD1-side point mutations in murine HSPCs. NOT shown for any KMT2A fusion, and no EPC1-side mutation has been tested in any leukaemia model.
absence classnever-measured
absence class noteThe absences here need new material. No EPC1-side mutation has been tested in a leukaemia model - every requirement result is from the MBTD1 side. No affinity measurement exists for the MBTD1-EPC1 contact, because the only structure is a designed single-chain chimera in which association cannot be measured. No lineage-split DepMap analysis of EPC1 was run in this beat, and that one is NOT never-measured - it is measured-not-analysed, since the per-line gene effects are public and the split is pure compute; it is named here rather than performed because this beat did no DepMap analysis beyond a three-gene lookup.
alias symbol
[
  "Epl1"
]
prev symbol
[]
PDB IDS
[
  "6NFX"
]
domains
[
  {
    "name": "MBTD1-binding C-terminal fragment (the span present in PDB 6NFX)",
    "start": 644,
    "end": 671,
    "pfam": null
  }
]
essentiality
{
  "depmap_mean_gene_effect": -0.336,
  "depmap_n_screens": 1258,
  "depmap_frac_below_minus_0_5": 0.2273,
  "depmap_frac_below_minus_1_0": 0.0135,
  "open_targets_dataversion": "26.06",
  "fetched": "2026-09-23",
  "comparators_same_call_or_same_endpoint": {
    "MBTD1": "-0.3531 over 1,258 screens, 17.89% below -0.5",
    "KAT5": "-0.607 over 1,258 screens, 70.2% below -0.5 (from gene-mbtd1, same endpoint)",
    "EP400": "-0.932 over 1,258 screens, 94.3% below -0.5 (from gene-mbtd1, same endpoint)"
  },
  "interpretation": "The mildest dependency in the NuA4/TIP60 nodes this graph holds, and 'mildest' is not 'safe' - 22.7% of lines below -0.5 means roughly one line in four is measurably harmed. NOT LINEAGE-SPLIT. A knockout mean is an upper bound on what an interface-directed agent would do, not an estimate."
}
How this could bite us back
  • The whole reason to look at EPC1 is that it is less pan-essential than KAT5 or EP400, and that is an argument about a margin, not about a mechanism. One cell line in four is measurably harmed by losing it, the figure is not lineage-split, and the perturbation anybody would actually attempt - breaking one protein-protein contact - is not the perturbation any of these numbers measures.
  • MBTD1's normal job inside NuA4/TIP60 is promoting homologous-recombination repair and competing with 53BP1 for H4K20me2. An agent that stopped MBTD1 entering the complex would take that with it, which is a DNA-repair liability in every dividing cell, and in a paediatric patient a DNA-repair liability is a second-malignancy liability.
  • MBTD1 also holds adult haematopoietic stem cells quiescent, and conditional Mbtd1 loss drives hyperactive cycling and breaks the stress response. So the toxicity of disrupting this arm lands specifically on the compartment a leukaemia therapy has to preserve.
  • There is no chemical matter. No ligand, no probe and no screen against the MBTD1-EPC1 interface appears anywhere in this beat's reading, and a 28-residue hydrophobic peptide binding a four-repeat block is the shape of interface that has defeated most attempts. Naming EPC1 as 'the druggable node in this pathway' would be naming the least bad option in a pathway with no good ones.
  • And the disease is twelve or thirteen cases. Even a perfect agent against this interface would be an agent for a fusion nobody can detect on a standard panel, which is why the published lead for this fusion is a BET inhibitor and not this.
Still open
  1. What is the affinity of the MBTD1-EPC1 contact? The only structure is a designed single-chain chimera, so no Kd exists. Without one, 'the fusion requires this contact' cannot be translated into whether a competitive inhibitor is plausible, and this project has a standing rule that a PDB entry existing is not a structure suitable for design.
  2. Does an EPC1-side mutation block fusion-driven transformation? All three requirement mutations are on the MBTD1 side, where they could in principle disrupt other interactions of the same hydrophobic surface. An EPC1 mutant rescue or knock-in would establish that EPC1 specifically is the required partner rather than that the surface is required.
  3. Is EPC1 dependency lineage-split, and specifically what is it in haematopoietic lines? The per-line gene effects are public and the split is compute. Every tolerance claim in this entry currently rests on a mean over 1,258 lines of all lineages, which this project's own rules say is not a result.
  4. Does EPC1 have a paralogue that buffers it? EPC2 exists in the human genome and was NOT looked up in this beat. If EPC2 can substitute in NuA4/TIP60, a 22.7%-below-minus-0.5 dependency may be a paralogue-buffered one, which would change both the tolerance argument and the interpretation of every knockout number in this entry. This project closed an analogous question for ARID1B two cycles ago and the answer was paralogue buffering.
  5. Is the ZMYND11::MBTD1 fusion's EPC1 contact the same contact 6NFX models? The structure uses EPC1 644-671 and MBTD1 140-562; the fusion's required residues map to MBTD1 236, 259 and 263, which are inside that span. Consistent, but nobody has shown the fusion protein binding EPC1 through the same surface as wild-type MBTD1 does, and a chimera's MBT block sits in a different molecular context.
Every source behind this entry (7)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    EPC1 is HGNC:19876, 'enhancer of polycomb 1', at 10p11.22, with alias symbol Epl1, no previous symbol, primary Ensembl id ENSG00000120616 and UniProt Q9H2F5 (EPC1_HUMAN, 836 aa, protein name 'Enhancer of polycomb homolog 1'); a gene_exact query for EPC1 restricted to human reviewed records returns exactly one row whose primary gene name is EPC1.

    What this evidence cannot show

    Fetched live 2026-09-23. The gene_exact control was run explicitly because of this project's rule 62 (a gene_exact query can return several reviewed records and the first may be the wrong one) and returned a single row, so that trap does not fire here. The FASTA sequence was also downloaded and used to align PDB 6NFX. NOTE THE NAMING HAZARD: the protein name is 'Enhancer of polycomb homolog 1' and EPC1 is a NuA4/TIP60 subunit, NOT a Polycomb-complex subunit - the name is a Drosophila-homology name and this graph has already been burned by protein names that name the wrong partner (rule 69's BICRA case). scripts/is-symbol-covered.py EPC1, run alone, returns 'EPC1 no match under 5 candidate names'.

    Evidence that argues against this (1)
    • A single gene_exact row is reassurance about a query, not about a claim, and the more relevant hazard here is not covered by any accession check: 'Enhancer of polycomb' is a Polycomb-group name on a NuA4 protein, and the secondary literature does conflate them - a 2026 review states that MBTD1, EPC1's binding partner, 'is associated with PRC1'. Length is also no defence for this gene in the way the project's rules assume, because no near-length human paralogue was tested for in this beat. Yu X et al. 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    A crystal structure of the MBTD1-EPC1 complex shows a hydrophobic C-terminal fragment of EPC1 engaging the MBT repeats of MBTD1 at a site distinct from the H4K20me-binding site, with cellular assays validating the key residues, EPC1 being the non-catalytic NuA4/TIP60 subunit that enlists MBTD1 into the complex; PDB 6NFX, at 1.95 A with primary citation PMID 32209463, contains a single 470-residue polymer entity described as 'MBT domain-containing protein 1,Enhancer of polycomb homolog 1' and cross-referenced to Q05BQ5 and Q9H2F5, which aligns as MBTD1 residues 140-562 joined through a designed linker to EPC1 residues 644-671.

    PMID 32209463DOI
    What this evidence cannot show

    The paper was read at abstract level; Europe PMC returns no pmcid for it. The residue spans were NOT read off the deposit's numbering but derived by aligning 6NFX's fetched canonical entity sequence (data.rcsb.org polymer_entity endpoint) against UniProt Q05BQ5 and Q9H2F5 FASTA downloaded the same session - the procedure this project requires after three documented cases of deposit-numbering offsets. 6NFX IS A SINGLE-CHAIN ENGINEERED CHIMERA with ONE polymer entity, so it models the bound state rather than measuring association: no affinity, stoichiometry or cellular-concentration statement follows from it. The EPC1 fragment present is 28 residues of an 836-residue protein.

    Evidence that argues against this (1)
    • A single-chain construct is both a legitimate way to solve a weak interface and a way to make an interface look obligate when it is not. The functional evidence that this contact matters comes entirely from mutagenesis on the MBTD1 side in a different paper and a different system; nothing has tested an EPC1-side mutation in a leukaemia model, so 'the fusion requires the MBTD1-EPC1 contact' is supported from one direction only. Li J et al. 2021
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In murine haematopoietic stem/progenitor cells, deletion of the four MBT repeats of the MBTD1 portion of ZMYND11-MBTD1, or the triple point mutation A661D/L684D/L688D that disrupts the MBTD1-EPC1 contact, abolishes both Tip60 binding and transformation, whereas the H4K20me1/2-binding-defective mutants W954A and F951A/W954A/Y958A bind Tip60 normally and transform normally.

    PMID 33594072DOI
    What this evidence cannot show

    Murine primary HSPCs with a retrovirally overexpressed human cDNA; every result is a transformation assay in that system. The three mutated residues are on the MBTD1 side of the interface, so this establishes that the CONTACT is required and not that EPC1 specifically is the required partner - a mutation that disrupts a hydrophobic surface could disrupt other interactions of the same surface. Full text read via PMC7886901; identifiers fetched from PubMed 2026-09-23. gene-mbtd1 E11 records the same result and this item is not independent of it.

    Evidence that argues against this (1)
    • A required protein-protein contact in an overexpression transformation assay is a strong result about the assay and a weak one about a therapeutic window. The same paper's PHD finding shows the system producing uninterpretable requirement calls in the other direction - PHD deletion abolishes leukaemogenesis while only mildly impairing the binding it is supposed to mediate, which the authors flag as unexplained - so a clean requirement in this assay should not be read as a clean mechanism. Devoucoux M et al. 2022
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Applying the construct offset stated in fusion-zmynd11-mbtd1 (MBTD1 residue n equals fusion residue n plus 425) to the published EPC1-contact mutations gives MBTD1 A236, L259 and L263, and to the cage mutations gives MBTD1 F526, W529 and Y533; all six residue identities were read off the live UniProt Q05BQ5 sequence and match the mutation names, so the offset is validated against sequence rather than asserted.

    What this evidence cannot show

    THIS IS A DERIVATION, NOT A PUBLISHED STATEMENT. Neither Li et al. nor Zhang et al. give MBTD1-frame numbers for these mutations; the offset comes from Li et al.'s stated construct boundaries and was first written down in this graph by Q121, labelled there as its own arithmetic. What this item adds is the SEQUENCE CHECK: six positions, six matching residue identities, which is a strong consistency test but not a proof - a uniform offset error that happened to land on matching residues at all six positions is improbable rather than impossible. The offset also assumes the 1,053-residue construct rather than the 1,054-residue patient transcript, a one-residue discrepancy neither source explains.

    Evidence that argues against this (1)
    • Six matching residues is a consistency check whose power depends on the amino-acid composition of the protein, and alanine and leucine are among the commonest residues in any sequence - so three of the six matches carry less information than they appear to. The check is worth more for W529 and Y533, which are rare enough for a coincidental match to be unlikely, and the honest statement is that the offset is strongly supported at the cage and moderately supported at the EPC1 contact. Eryilmaz J et al. 2009
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    DepMap via Open Targets, Open Targets dataVersion 26.06, fetched 2026-09-23: EPC1 mean gene effect -0.336 across 1,258 screens with 22.73% of screens below -0.5 and 1.35% below -1.0; MBTD1 -0.3531 across 1,258 screens with 17.89% below -0.5 and 0.24% below -1.0; both figures reproduce values recorded independently in gene-mbtd1 to three and four decimal places.

    What this evidence cannot show

    Rule 83 observed: the release is recorded because the endpoint returns it, and the screen count is stated per gene rather than implying a corpus-wide denominator. ONE LABELLING CORRECTION, because it matters for the next reader: meta.dataVersion is the OPEN TARGETS PLATFORM release, not a DepMap release name such as 24Q2, so the correct field is 'Open Targets dataVersion 26.06'. gene-mbtd1 E12 renders the same value as 'DepMap release 26/06', which is a mislabel rather than a wrong number - reported, not edited. NOT LINEAGE-SPLIT: nothing here says anything about haematopoietic lines specifically, and this project has had two selectivity claims turn out to be lineage artefacts. A knockout mean is an upper bound on the damage an interface-directed agent would do, not an estimate of it.

    Evidence that argues against this (1)
    • A pooled CRISPR mean is the wrong statistic for a therapeutic-window argument in two ways at once. It is not lineage-split, so the haematopoietic subset - the only one relevant here - is unreported; and 22.7% of lines below -0.5 is a distribution with a long tail, so the mean understates what happens in the quarter of lines that care. This project's own rules require both a lineage split and a distinction between the z-scaled and raw statistics before any selectivity or tolerance claim, and neither was done here. Meyer C et al. 2023
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    EPC1 is a recurrent 5' fusion partner in low-grade endometrial stromal sarcoma as EPC1-PHF1, one of a family of fusions whose members also include JAZF1-SUZ12, JAZF1-PHF1, MEAF6-PHF1 and MBTD1-CXorf67, and which Piunti et al. characterise as consisting of an N-terminal NuA4 acetyltransferase-complex component fused to a C-terminal Polycomb-group subunit - a generalisation made while identifying JAZF1 as a new NuA4 subunit and while describing MBTD1 as 'the NuA4 subunit gene'.

    PMID 31281901DOI
    What this evidence cannot show

    Full text read this session via PMC6609211. The paper's own experiments concern JAZF1 and CXORF67, not EPC1; EPC1-PHF1 appears in its enumeration of the family and no EPC1 experiment is reported. So this item establishes that EPC1 is a member of a fusion family whose architectural rule the paper states, and nothing about EPC1's own fusion mechanism. The family rule is built from four or five fusions in one rare tumour type, and one of its members (JAZF1) was assigned to NuA4 by this same paper, so rule and instance were established together.

    Evidence that argues against this (1)
    • EPC1 appearing as a sarcoma 5' partner AND as MBTD1's NuA4 anchor are two independent facts about one protein, and treating their coincidence as meaningful would be exactly the error the polarity entry warns against. Nothing shows EPC1 appearing on both sides of a junction, which is what a polarity-reversal claim requires, and nothing connects EPC1-PHF1's mechanism to the MBTD1 contact at all. Micci F et al. 2014
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    MBTD1 is a stable subunit of the NuA4/TIP60 acetyltransferase complex and, through a histone reader domain for H4K20me1/2, allows TIP60 to associate with specific gene promoters and to promote homologous-recombination repair of DNA double-strand breaks, with the complex regulating 53BP1 association partly by competing for H4K20me2 and partly through TIP60-mediated acetylation of H2AK15.

    The TIP60 Complex Regulates Bivalent Chromatin Recognition by 53BP1 through Direct H4K20me Binding and H2AK15 Acetylation. · Jacquet K, Fradet-Turcotte A, Avvakumov N, Lambert JP, et al. · Mol Cell 2016
    PMID 27153538DOI
    What this evidence cannot show

    Abstract-level in this beat; gene-mbtd1 E3 carries it in full and this item is not independent of it. Cited here to establish the complex and the normal function that any agent disrupting the MBTD1-EPC1 grip would also disrupt, which is the bite-back argument. The paper is about MBTD1's entry into the complex and the DNA-repair consequence, not about EPC1 specifically. Identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • This paper's mechanism for MBTD1's promoter association runs through the H4K20me1/2 reader - and that is precisely the module the leukaemic fusion does NOT need. So the normal function and the oncogenic function of the MBTD1 arm of NuA4/TIP60 run through different surfaces of the same protein block, which cuts both ways: it weakens 'disrupting the grip will disrupt DNA repair' as an automatic toxicity argument, and it weakens any inference from the normal complex's biology to the fusion's. Li J et al. 2021
Connected to