Oncofusion

start / graph / gene-eed

Genestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q41-prc2-arm

EED - the PRC2 subunit this disease actually needs, with drugs in the clinic and zero leukaemia studies

also called EED · embryonic ectoderm development · hEED · WAIT-1 · WD protein associating with integrin cytoplasmic tails 1

EED is the part of the Polycomb machine that reads the 'off' tag the machine itself made, and then makes the machine work harder - a self-reinforcing loop. Unlike the enzyme subunits, EED has no back-up copy, so removing it switches the whole machine off; in mouse KMT2A leukaemia that kills the disease, and drugs that block EED already exist in patients - just never tested in leukaemia.

The same thing, for a scientist

EED (UniProt O75530, 441 aa) is the WD40 subunit of PRC2 that binds H3K27me3 through an aromatic cage and allosterically stimulates EZH2 catalysis, creating the read-write loop that spreads the mark [E1][E5]. **It is the PRC2 subunit with no paralogue, and it behaves accordingly.** In MLL-AF9 leukaemia, inactivating *Eed* completely ablates PRC2 function and is incompatible with leukaemia growth, whereas *Ezh2* inactivation is survivable [E2]; shRNA suppression of *Eed* arrests proliferation and forces differentiation with minimal impact on non-transformed haematopoietic lines [E3]; and the anti-leukaemic effect runs through both *Cdkn2a*-dependent and *Cdkn2a*-independent routes, with *Gata2*, *Egr1* and *Cdkn2b* also de-repressed and MLL-fusion direct targets negatively enriched [E4]. **EED is also the best-scoring PRC2 subunit in this project's own DepMap panel** - index mean -0.458 against -0.340 in comparator leukaemia lines, difference -0.117, **rank 824 of 17,728** - and the difference is carried by the *lymphoid* half of the index arm (-0.224) rather than the myeloid half (-0.058) [E6]. **And there is a clean, checkable gap.** Allosteric EED inhibitors exist (EED226 [E5], EEDi-5285, EEDi-5273), a bivalent EED-targeted PRC2 degrader exists, and the clinical candidate MAK683 has completed a 139-patient first-in-human phase 1/2 [E7]. `(EED226 OR MAK683 OR "EED inhibitor") AND (leukemia OR MLL OR KMT2A)` returns **0 PubMed records** [E8]. The mechanistically best-supported PRC2 handle in this disease has clinical chemical matter and has never been pointed at it.

Go deeper
Why EED and not EZH2

Three independent arguments converge, and they are the reason this entry exists separately from gene-ezh2:

  1. No paralogue. EZH1 rescues EZH2 [gene-ezh1]; nothing rescues EED. That is why Eed inactivation ablates PRC2 while Ezh2 inactivation does not 2.
  2. The genetics in this disease already ran the experiment. Eed-null MLL-AF9 leukaemia does not grow 2, and the effect survives removal of the obvious single explanation, Cdkn2a 4.
  3. An allosteric EED inhibitor is not equivalent to a SAM-competitive EZH2 inhibitor. EED226 also inhibits PRC2 carrying an EZH2 mutant that is resistant to SAM-competitive inhibitors 5 - i.e. it covers a resistance mechanism the EZH2-site drugs do not, and by acting on the shared subunit it is agnostic to whether the complex contains EZH1 or EZH2.

Point 3 is the one that changes a plan. The EZH1 problem that makes tazemetostat the wrong drug in this disease (gene-ezh1) is an argument for dual EZH1/EZH2 inhibition or for EED-directed inhibition, and the second route is not discussed anywhere in the KMT2A literature.

The counted absence, with its queries

Run 2026-09-20. Every count is from PubMed esearch, and each was title-checked before being recorded 8.

queryrecordsafter title-check
(EED226 OR MAK683 OR "EED inhibitor")2828 are about EED-directed PRC2 inhibition (chemistry, prostate, DLBCL, neuroblastoma, NPC, HIV latency, kidney injury, a degrader, a phase 1/2)
(EED226 OR MAK683 OR "EED inhibitor") AND (leukemia OR MLL OR KMT2A)0-
(RBBP4 OR RBBP7) AND (KMT2A OR MLL) AND leukemia20 on topic
SUZ12 AND (KMT2A OR MLL) AND leukemia5see gene-suz12

Per Rule 64 the complex form was run as well and it is larger than the subunit form, which is why it matters: "Polycomb repressive complex 2" AND (KMT2A OR MLL) AND leukemia returns 47 against "PRC2" AND (KMT2A OR MLL-rearranged) AND leukemia's 24. Title-checking the complex-form list recovered no leukaemia EED-inhibitor study - it recovered the PRC2/trithorax response-element literature and the MLL-AF9 genetics already cited here. Per the subtract-before-you-report rule, the repository was also grepped: of the nineteen PMIDs this beat uses, only two (22396593, 22469984) were already cited anywhere in this graph, both in mechanism-polycomb-prc1-prc2-cbx8.

The MAK683 phase 1/2 enrolled 139 patients across clear-cell ovarian carcinoma, CRPC, DLBCL, epithelioid sarcoma, gastric cancer, nasopharyngeal carcinoma and SWI/SNF-mutant sarcoma - no leukaemia cohort at all 7. That is an absence of testing, not a negative result.

Accession resolution

gene_exact:EED AND organism_id:9606 AND reviewed:true returns exactly one reviewed human record: O75530, primary gene name EED, no synonyms, 441 aa, protein name Polycomb protein EED 9. Resolved 2026-09-20 with gene_primary and length in fields.

EED is the safest of the six PRC2 accessions in this beat precisely because it has no listed synonyms and no paralogue - the two conditions that produced Rules 62, 63 and 65.

Subnuclear: PRC2 licenses the compartment PRC1 occupies (Q87, 2026-09-21)

Subcellular. Nucleus and chromosome (UniProt O75530, both UNCODED - no evidence code, so not a measurement); HPA v25.1 IF, reliability Supported, nucleoplasm main with cytosol additional - an independent antibody corroboration, and a cytosolic pool UniProt does not record.

Subnuclear: Polycomb bodies, as a REQUIREMENT. EED and EZH2 are required for BMI1 recruitment to PcG bodies, which overlie pericentric heterochromatin (PMID 16314526). This entry records the requirement and deliberately does NOT assert EED's own steady-state residence there, which nobody has measured. UniProt separately records transient colocalisation with XIST at the inactive X, uncoded.

Cell type and state. MLL-AF9 leukaemia cells in mouse, functionally: Eed inactivation is incompatible with leukaemia growth where Ezh2 inactivation is survivable, and Eed suppression arrests proliferation and forces differentiation. No imaging in any blood cell.

Shown for fusion. MLL-AF9 for function; nobody for localisation. Both fusion-scoped localisation hits were retrieved and rejected.

Facts and figures
symbolEED
uniprotO75530
length AA441
uniprot primary gene nameEED
uniprot resolution notegene_exact:EED AND organism_id:9606 AND reviewed:true returned exactly one row; primary gene name EED matched, no synonyms listed; length 441 asserted. Resolved 2026-09-20.
paraloguenone - EED is the only PRC2 core subunit with no paralogue, which is why it is the non-redundant node
complexprotein-prc2-complex
functionWD40 beta-propeller; binds H3K27me3 via an aromatic cage and allosterically stimulates EZH2 catalysis (the PRC2 read-write loop)
expression hpa nTPM median kmt2a r panel29.9
subnuclear localizationPOLYCOMB BODIES, recorded as a REQUIREMENT: EED and EZH2 are required for BMI1 recruitment to PcG bodies, which sit on regions of pericentric heterochromatin (PMID 16314526). Plus transient colocalisation with XIST at the inactive X (UniProt O75530, UNCODED - no evidence code, so not a measurement this graph can stand behind). EED's own steady-state occupancy of PcG bodies has not been measured by a dedicated experiment, and this entry deliberately does not assert it.
cell statePRC2-DEPENDENT LEUKAEMIC PROLIFERATION. shRNA suppression of Eed arrests proliferation and forces differentiation with minimal impact on non-transformed haematopoietic lines - a differentiation-block state that EED maintains - and Eed inactivation is incompatible with MLL-AF9 leukaemia growth. Both are already recorded in this entry's evidence; they are restated here as the cell-state answer because the PI dimension asks for it and the facts were present without being labelled.
shown for fusionMLL-AF9 for FUNCTION (mouse, Eed inactivation); NOBODY for LOCALISATION. Measured 2026-09-21: EED[tiab] AND the eight-term fusion clause AND the ten-term localisation clause returns 2 records, both retrieved and read - PMID 29363553 (PcG-family transcript profiles in aged mouse HSPCs; an expression study, the word 'nuclear' belongs to 'melanoma nuclear protein 18') and PMID 17259173 (a developmentally regulated Eed-Mll interface in mouse brain; a complex-composition study, and MLL here is wild-type in neural tissue, not a fusion). Screened count on topic: 0.
essentiality
{
  "depmap_kmt2a_r_mean": -0.4577,
  "depmap_other_leukaemia_mean": -0.3404,
  "depmap_difference": -0.1173,
  "depmap_rank_of_17728": 824,
  "depmap_pan_essentiality_all_lines": -0.2766,
  "depmap_index_myeloid_mean": -0.3883,
  "depmap_index_lymphoid_mean": -0.5791,
  "depmap_myeloid_difference": -0.0583,
  "depmap_lymphoid_difference": -0.2242,
  "n_index_lines": 11,
  "n_comparator_lines": 43,
  "source": "analyses/prc2-dependency-lineage-split/results.json",
  "interpretation": "best-scoring PRC2 subunit in the index arm; modest lean, below the project's -0.5 selectivity convention; carried by the B-lineage half"
}
chemical matter
{
  "allosteric_inhibitors": [
    "EED226",
    "EEDi-5285",
    "EEDi-5273",
    "MAK683"
  ],
  "clinical_candidate": "MAK683 (NCT02900651, phase 1/2 complete, 139 patients, no leukaemia cohort)",
  "degrader": "a bivalent EED-targeted chemical degrader of PRC2 has been reported (PMID 31831267; not independently read in this beat)",
  "leukaemia_studies": 0,
  "absence_query": "(EED226 OR MAK683 OR \"EED inhibitor\") AND (leukemia OR MLL OR KMT2A) -> 0 records, 2026-09-20"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.4577,
  "chronos_mean_other_leukaemia": -0.3404,
  "difference": -0.1173,
  "pan_essentiality_all_lines": -0.2766,
  "selectivity_rank": 824,
  "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."
}
localization
{
  "subcellular": "NUCLEUS and CHROMOSOME. UniProt O75530 lists both, and both are UNCODED - no ECO:0000269, no ECO:0000305, no citation at all - which is the rule-91 case this graph measures at roughly a quarter of entries: a database statement that is not a measurement. The record does carry one substantive uncoded note: EED 'transiently colocalizes with XIST at inactive X chromosomes'. HPA v25.1 IF, reliability Supported, gives Nucleoplasm main with Cytosol additional, which is an independent antibody-based corroboration of the nuclear call and also reports a cytosolic pool that UniProt does not.",
  "subnuclear": "POLYCOMB BODIES, as a REQUIREMENT rather than as a residence, plus the inactive X. EED and EZH2 are required for BMI1 recruitment to PcG bodies - nuclear structures located on regions of pericentric heterochromatin - so PRC2 builds or licenses the compartment that PRC1 occupies (PMID 16314526). That is a stronger and more useful statement than 'EED is in Polycomb bodies', because it is causal and it predicts what an EED inhibitor does to PRC1's address. Separately, UniProt records transient colocalisation with XIST at the inactive X, uncoded.",
  "the_distinction_that_matters": "A residence claim and a requirement claim are different facts and this entry records the requirement. PMID 16314526 measures BMI1's kinetics at PcG bodies and finds them EED/EZH2- and DNMT1-dependent; it does not establish EED's own steady-state occupancy of the same foci by a dedicated measurement. Recording the stronger-sounding residence claim would be reading a dependency as an image.",
  "how_shown": "FRAP of tagged BMI1 at PcG bodies in mammalian cells with genetic loss of EED, EZH2 and DNMT1 (PMID 16314526); HPA antibody immunofluorescence; UniProt's nuclear and chromosome terms are uncoded",
  "tissue_of_evidence": "non-haematopoietic mammalian lines for the PcG-body work. The EED work IN this disease - Eed inactivation ablating PRC2 function and blocking MLL-AF9 leukaemia - is genetic and transcriptional, and contains no localisation measurement.",
  "node": "localization-polycomb-bodies-kmt2a-r"
}
cell type
{
  "haematopoietic_answer": "MLL-AF9 LEUKAEMIA CELLS in mouse, functionally. Eed inactivation impedes MLL-AF9-mediated leukaemogenesis through both Cdkn2a-dependent and Cdkn2a-independent mechanisms - already recorded in this entry - and Eed is the PRC2 subunit whose loss is not survivable where Ezh2 loss is. Separately, Eed is one of the PcG-family transcripts profiled across young and aged mouse haematopoietic stem/progenitor cells in a senescence study (PMID 29363553), which is an expression measurement, not a localisation one.",
  "cells_actually_imaged": "none for EED in a haematopoietic context; the PcG-body imaging is in non-blood mammalian lines.",
  "expression_atlas": "HPA v25.1: RNA tissue specificity Low; IF reliability Supported."
}
How this could bite us back
  • The dose-limiting toxicities of the only EED inhibitor taken into patients are thrombocytopenia and febrile neutropenia, and its grade 3/4 events are neutropenia, thrombocytopenia and anaemia (PMID 39793445). PRC2 is essential for HSC maintenance, so in a leukaemia the compartment being treated and the compartment being poisoned are the same, and the therapeutic-window claim from a mouse shRNA screen was measured against immortalised lines rather than primary HSPCs.
  • EED has no paralogue, which is exactly why it is attractive and exactly why it is dangerous: there is no buffer in normal cells either. The redundancy that protects normal tissue from an EZH2-selective inhibitor does not exist for an EED-directed one.
  • Resistance to PRC2-directed therapy is reported to arise through PRC2 gatekeeper mutations and through compensatory DNA methylation (PMID 41443386). The second route is upstream of the drug's binding site entirely, so switching subunit does not address it, and a DNA-hypermethylated disease - which t(4;11) infant ALL is - may be pre-adapted to it.
  • The anti-leukaemic mechanism of PRC2 loss is de-repression of CDKN2A and other restraints; CDKN2A deletion is common in KMT2A-rearranged leukaemia, and where it is deleted the mechanism is partly removed (PMID 26118502 shows only PARTIAL rescue, so it is not fully removed - but the residual route is undefined and untargeted).
Still open
  1. Does an allosteric EED inhibitor (EED226 or MAK683) reproduce the Eed-knockout phenotype in a KMT2A-rearranged system, and specifically in a KMT2A::AFF1 B-ALL model? Nobody has run it; see exp-eed-allosteric-inhibition-in-kmt2a-aff1.
  2. Is EED-directed inhibition equivalent to dual EZH1/EZH2 inhibition in this disease, or better? It should be paralogue-agnostic by construction, and EED226 covers at least one SAM-competitive resistance mutant, but nothing compares the two classes head-to-head in any leukaemia.
  3. The MAK683 phase 1/2 used H3K27me3 change from baseline as a pharmacodynamic endpoint in patients. Is that assay transferable to marrow blasts, and would it have the dynamic range to distinguish EZH2-only from EZH1+EZH2 coverage in a KMT2A-rearranged patient?
  4. Why does the EED dependency lean lymphoid in DepMap when every mechanistic paper is myeloid? Four lines is not an answer; a matched panel of KMT2A::AFF1 B-ALL versus KMT2A::MLLT3 AML lines with EED knockout would settle it.
Every source behind this entry (11)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    PRC2 is stimulated upon recognition of its own catalytic product; cryo-EM of human PRC2 bound to bifunctional dinucleosomes shows a single PRC2, via interactions with nucleosomal DNA, positioning the H3 tails of the activating and the substrate nucleosome to interact with the EED subunit and the SET domain of EZH2 respectively, and accommodating varying linker-DNA lengths.

    Cryo-EM structures of PRC2 simultaneously engaged with two functionally distinct nucleosomes. · Poepsel S, Kasinath V, Nogales E · Nat Struct Mol Biol 2018
    PMID 29379173DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. Reconstituted human PRC2 on designed dinucleosome substrates in vitro; the geometry is established on a defined template, not in chromatin. No residue numbers are asserted here, in line with the QUEEN_BRIEF rule that spans must be derived by alignment rather than read off a deposit.

    Evidence that argues against this (1)
    • The read-write loop is not the only input, and in transcriptionally active chromatin it is overridden: di- and tri-methylated H3K36 allosterically inhibit PRC2, with unmodified H3K36 occupying a critical position in the EZH2-DNA interface. A model built only on EED-mediated stimulation predicts spreading that active chromatin does not permit. Finogenova K et al. 2020
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Inactivation of the essential PRC2 gene Eed led to complete ablation of PRC2 function, which was incompatible with MLL-AF9 leukaemia growth, with more profound gene-expression changes than in Ezh2-null leukaemic cells including down-regulation of Myc target genes and up-regulation of PRC2 targets; by contrast Ezh2-null leukaemias maintained self-renewal to the third round of transplantation.

    Polycomb repressive complex 2 is required for MLL-AF9 leukemia. · Neff T, Sinha AU, Kluk MJ, Zhu N, Khattab MH, Stein L, et al. · Proc Natl Acad Sci U S A 2012
    PMID 22396593DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. Conditional mouse alleles in an MLL-AF9 model - myeloid, not KMT2A::AFF1, and no infant or B-lineage arm. Genetic ablation removes the protein and therefore PRC2's scaffolding and non-catalytic roles together with the mark, so it is an upper bound on what an inhibitor could achieve rather than a prediction of it.

    Evidence that argues against this (1)
    • EED is essential for normal haematopoietic stem cell maintenance as well: PRC2 components, particularly EZH1/2, SUZ12 and EED, are described as essential for HSC maintenance. A requirement shared with the normal stem-cell compartment is the definition of a narrow therapeutic window, not of a selective target. Yamagishi M et al. 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    shRNA-mediated suppression of the PRC2 component Eed caused proliferation arrest and differentiation of MLL-AF9;Nras-G12D leukaemia cells with minimal impact on the growth of several non-transformed haematopoietic cell lines.

    The Polycomb complex PRC2 supports aberrant self-renewal in a mouse model of MLL-AF9;Nras(G12D) acute myeloid leukemia. · Shi J, Wang E, Zuber J, Rappaport A, Taylor M, Johns C, et al. · Oncogene 2013
    PMID 22469984DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. shRNA knockdown leaves residual protein, and the therapeutic-window claim is measured against immortalised non-transformed lines rather than primary HSPCs or an in vivo compartment. The model carries a co-operating Nras-G12D mutation. Mouse, myeloid.

    Evidence that argues against this (1)
    • The same paper's own framing is the counter to the textbook reading rather than to the result: it concludes that Trithorax and Polycomb group proteins can COOPERATE to maintain aberrant lineage programs in cancer. Any downstream entry that treats PRC2 as the fusion's antagonist is contradicted by the paper this claim comes from. Shi J et al. 2013
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Cdkn2a inactivation only partially rescues in vitro and in vivo growth of Eed-inactivated MLL-AF9 cells; Eed-null Cdkn2a-null cells remain severely compromised, Gata2, Egr1 and Cdkn2b are de-repressed on Eed inactivation, and direct binding targets of MLL fusion proteins are negatively enriched in Eed-inactivated Cdkn2a-null MLL-AF9 cells.

    Inactivation of Eed impedes MLL-AF9-mediated leukemogenesis through Cdkn2a-dependent and Cdkn2a-independent mechanisms in a murine model. · Danis E, Yamauchi T, Echanique K, Haladyna J, Kalkur R, Riedel S, et al. · Exp Hematol 2015
    PMID 26118502DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. Mouse MLL-AF9, myeloid. 'Direct binding targets of MLL fusion proteins are negatively enriched' is an RNA-seq gene-set statement in Eed-inactivated cells and is not accompanied in the abstract by a ChIP demonstration that fusion occupancy itself changes - which is the interpretation a reader is most likely to import.

    Evidence that argues against this (1)
    • That MLL-fusion targets fall when Eed is inactivated is compatible with PRC2 acting UPSTREAM of the fusion programme, but it is equally compatible with the cells having differentiated for Cdkn2a-independent reasons and the fusion programme collapsing as a consequence. Distinguishing the two needs a timecourse or a differentiation-blocked control, and the abstract reports neither. Neff T et al. 2012
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    EED226 is a potent selective PRC2 inhibitor that binds directly to the H3K27me3-binding pocket of EED, induces a conformational change on binding that causes loss of PRC2 activity, blocks H3K27 methylation of PRC2 target genes and induces regression of human lymphoma xenograft tumours comparably to SAM-competitive inhibitors, and also effectively inhibits PRC2 containing an EZH2 protein mutated to resist SAM-competitive inhibitors.

    An allosteric PRC2 inhibitor targeting the H3K27me3 binding pocket of EED. · Qi W, Zhao K, Gu J, Huang Y, Wang Y, Zhang H, et al. · Nat Chem Biol 2017
    PMID 28135235DOI
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. Efficacy is in human LYMPHOMA xenografts; no leukaemia model, no KMT2A-rearranged system and no B-ALL. 'Also inhibits PRC2 with a SAM-competitive-resistant EZH2 mutant' is a cell or enzyme observation about a specific engineered or reported mutant, not a demonstration of clinical cross-resistance coverage.

    Evidence that argues against this (1)
    • Allosteric EED targeting does not escape all PRC2 resistance. Resistance to EZH2 inhibitors in DLBCL has been mapped, and the 2026 Polycomb review states that resistance to dual EZH1/2 inhibition can emerge through PRC2 gatekeeper mutations and compensatory DNA methylation - the second of which is upstream of any PRC2 binding site and would not be addressed by changing which subunit the drug binds. Yamagishi M et al. 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    EED is the best-scoring PRC2 subunit in a scan of 17,728 protein-coding genes across 11 KMT2A-rearranged DepMap leukaemia lines against 43 other leukaemia lines - index mean -0.458, comparator -0.340, difference -0.117, rank 824, pan-essentiality -0.277 - and within the index arm the difference is larger in the B-lineage half (-0.224) than the myeloid half (-0.058).

    What this evidence cannot show

    A difference of -0.117 at rank 824 of 17,728 is a modest lean, not a strong selective dependency, and it sits below this project's -0.5 convention for a selectivity claim. The lymphoid half is 4 lines and the myeloid half 7, so the split is underpowered and no infant KMT2A::AFF1 line exists in DepMap at all. Raw effect sizes only, no z, no permutation null - so this is not a hit, it is a consistency check against the mouse genetics. Cellosaurus fusion annotation may be incomplete. DepMap release version is not stated by the Open Targets API.

    Evidence that argues against this (1)
    • The lymphoid lean is the opposite of what this project has found in every previous epigenetics lineage split, where the index signal turned out to be myeloid identity rather than the fusion. With four lymphoid lines, the honest alternative is that this is noise; the reason it is recorded rather than suppressed is that it agrees in direction with the two index-disease results (PMID 40722046, PMID 36280659) and a suppressed observation cannot later be checked. Ito J et al. 2023
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    MAK683, a first-in-class selective allosteric EED inhibitor, completed a first-in-human phase 1/2 dose-escalation study in 139 patients with advanced malignancies (clear-cell ovarian carcinoma, castration-resistant prostate cancer, DLBCL, epithelioid sarcoma, gastric cancer, nasopharyngeal carcinoma and SWI/SNF-mutated sarcoma); 15 patients had 22 dose-limiting toxicities including thrombocytopenia and febrile neutropenia, grade 3/4 drug-related adverse events included neutropenia, thrombocytopenia and anaemia, median progression-free survival was 1.9 months and the overall response rate was 5.8%, with activity seen in DLBCL and epithelioid sarcoma. Pharmacodynamics were measured as H3K27me3 change from baseline. NCT02900651.

    A first-in-human phase 1/2 dose-escalation study of MAK683 (EED inhibitor) in patients with advanced malignancies. · Ribrag V, Iglesias L, De Braud F, Ma B, Yokota T, Zander T, et al. · Eur J Cancer 2025
    PMID 39793445DOINCT02900651
    What this evidence cannot show

    Abstract read from the eutils record; full text not read. **No leukaemia cohort was enrolled**, so this trial says nothing about efficacy in leukaemia - it is cited for tolerability, for the existence of clinical chemical matter against EED, and for the fact that H3K27me3 has already been used as a clinical pharmacodynamic endpoint. Adults with advanced solid tumours and lymphoma; the dose-limiting toxicities are haematological, which is the most relevant fact for any leukaemia extension and the least favourable.

    Evidence that argues against this (1)
    • The dose-limiting toxicities of an EED inhibitor are thrombocytopenia and febrile neutropenia, and grade 3/4 events are neutropenia, thrombocytopenia and anaemia - i.e. the toxicity of this class is haematopoietic suppression. That is exactly consistent with PRC2 being essential for HSC maintenance, and it is the specific reason a leukaemia extension is harder than the solid-tumour experience suggests: the target organ and the toxic organ are the same one. Yamagishi M et al. 2026
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    PubMed returns 28 records for (EED226 OR MAK683 OR "EED inhibitor") and 0 records for (EED226 OR MAK683 OR "EED inhibitor") AND (leukemia OR MLL OR KMT2A); title-checking the 28 confirms all are about EED-directed PRC2 inhibition and none concerns leukaemia.

    What this evidence cannot show

    A counted absence is only as good as its symbol set. This one covers two named compounds and one phrase; it does not cover EEDi-5285, EEDi-5273, the bivalent EED degrader by name, unnamed chemical-series papers, or any study framed at 'PRC2 inhibitor' without naming EED. The complex form was run separately (Rule 64) and recovered no leukaemia EED-inhibitor study. The absence means nobody has PUBLISHED such a study indexed under these terms - it is an absence of testing, not a negative result.

    Evidence that argues against this (1)
    • The functional equivalent HAS been done and is positive, just not with an EED-directed compound: combined pharmacologic inhibition of EZH2 and EZH1 has a strong anti-leukaemic effect in MLL-AF9, and genetic Eed inactivation does the same. So the absence is of a specific chemical modality, not of the biological test, and anyone reading this absence should not conclude that PRC2 inhibition is untested in this disease. Danis E et al. 2015
  9. E9established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    EED resolves to a single reviewed human UniProt record, O75530 (EED_HUMAN), primary gene name EED, no listed gene synonyms, 441 amino acids, protein name Polycomb protein EED.

    What this evidence cannot show

    Resolved 2026-09-20. UniProt length is for the canonical isoform; EED is translated from multiple in-frame start codons giving isoforms that differ at the N-terminus, so 441 aa is the reference value and not a statement that only one species exists in cells.

    Evidence that argues against this (1)
    • The symbol EED is a three-letter string that PubMed's parser treats as a free-text token, so a literature search on the symbol alone is unreliable in the opposite direction from the paralogue traps: it over-retrieves. The searches in this entry were therefore run as compound queries with compound names rather than the bare symbol. NCBI PubMed 2026
  10. E10moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    EED and EZH2 are required for BMI1 recruitment to Polycomb bodies - nuclear structures located on regions of pericentric heterochromatin - and the maintenance DNA methyltransferase DNMT1 is required too. PRC2 therefore licenses the subnuclear compartment that PRC1 occupies, which makes EED's subnuclear relevance causal rather than merely positional.

    Association of BMI1 with polycomb bodies is dynamic and requires PRC2/EZH2 and the maintenance DNA methyltransferase DNMT1. · Hernández-Muñoz I, Taghavi P, Kuijl C, Neefjes J, van Lohuizen M · Mol Cell Biol 2005
    PMID 16314526DOI
    What this evidence cannot show

    The measurement is FRAP of tagged BMI1, so the quantity measured is PRC1's kinetics and EED enters as a genetic dependency; the paper does not establish EED's own steady-state occupancy of the same foci. Non-haematopoietic mammalian cell lines, tagged constructs, 2005. And FRAP cannot by itself distinguish a phase-separated body from binding to spatially clustered sites, which is the standing condensate caution in this graph.

    Evidence that argues against this (1)
    • This is a requirement, not a residence, and the distinction bites in the one place it would be used. The PcG-body compartment that EED licenses has been characterised on the PRC1 side - tagged RING1, CBX proteins and BMI1 form foci with slow internal FRAP recovery - and the natural next sentence, 'so EED is in Polycomb bodies too', is not what either experiment shows. It is also worth noting that PcG proteins 'exert their activities elsewhere in the nucleoplasm', so even for the PRC1 subunits the foci are not where most of the protein or most of the function is; reading Polycomb bodies as the site of PRC2 action would overstate what the compartment does. Vandenbunder B et al. 2014
  11. E11moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    No EED localisation result exists in a KMT2A-fusion context. Measured 2026-09-21: EED[tiab] AND (KMT2A[tiab] OR MLL[tiab] OR "MLL-AF4"[tiab] OR "MLL-AF9"[tiab] OR "MLL-ENL"[tiab] OR "MLL-AF10"[tiab] OR "t(4;11)"[tiab] OR "t(9;11)"[tiab]) AND (localization[tiab] OR localisation[tiab] OR subnuclear[tiab] OR nuclear[tiab] OR nucleolar[tiab] OR speckle[tiab] OR speckles[tiab] OR "PML bodies"[tiab] OR chromatin-bound[tiab] OR cytoplasmic[tiab]) returns 2 records, both retrieved by efetch and read, and neither reports EED's compartment: PMID 29363553 profiles PcG-family transcripts in young versus aged mouse haematopoietic stem/progenitor cells, and PMID 17259173 characterises a developmentally regulated Eed-Mll complex interface in mouse brain, where MLL is wild-type and no fusion is present. Screened count on topic: 0.

    What this evidence cannot show

    Bounded by titles and abstracts only (rule 80). EED[tiab] is also a short three-letter token and the operand control EED[tiab] AND (KMT2A[tiab] OR MLL[tiab] OR "MLL-AF9"[tiab] OR "MLL-AF4"[tiab]) returns 13, so the intersection being probed is genuinely small rather than empty. Rule-67 positive control for this query SHAPE: the identical construction, run over all sixteen genes of this tranche on the same day, returned non-zero for seven of them - H2AZ1 1, EED 2, CHAF1B 1, CEBPA 3, TIFAB 1, SAMHD1 1, PROM1 1 - so the instrument can return a one. The querytranslation was read back on every run and confirms the quoted cytogenetic terms expand to "t 4 11"[Title/Abstract] and "t 9 11"[Title/Abstract] and not to a UID filter (rule 95); every term carries its own [tiab] tag rather than sitting inside a parenthesised group (rule 94).

    Evidence that argues against this (1)
    • The 2-record hit list is a good worked example of why a fusion-scoped localisation count is a weak instrument, and reading this absence as 'EED is unstudied in MLL-fusion leukaemia' would be flatly wrong: one of the two hits is a mouse-brain paper showing that Eed associates with MLL itself upon brain maturation, having partnered EZH2 during neurodevelopment - a direct, developmentally switched PcG-trxG interface on this graph's central protein, found only because a localisation-term query happened to catch the word 'nuclear'. The query bounds the localisation question and nothing else, and rule 77 applies in reverse here: the compartment term, not the disease term, is what hid the interesting result. Kim SY et al. 2007
Connected to