Oncofusion

start / graph / analogy-mbtd1-occupies-both-fusion-roles-polarity-not-parts

Analogymoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q123-the-architecture-from-other-parts-analogy

MBTD1 is the effector-recruiting half of a leukaemia fusion and the addressing half of a sarcoma fusion while donating nearly all of itself to both - so the junction, not the parts, assigns the role

also called MBTD1 in both orientations · the polarity disanalogy · ZMYND11::MBTD1 versus MBTD1::EZHIP

The same protein appears in two different cancers' fusion genes - at the back of one and at the front of the other - and in both cases almost the whole protein is present. That means the two fusions do not differ because different pieces of it were kept; they differ because of which side of the join it sits on, and this is the second time this graph has found that the join, not the ingredients, decides what a fusion does.

The same thing, for a scientist

**Two fusions contain MBTD1 and its polarity is reversed between them.** In AML, the recurrent t(10;17)(p15;q21) puts MBTD1 **3'** of ZMYND11, and MBTD1's job is to dock the chimera into NuA4/TIP60 through EPC1 - the effector-recruiting role that AFF1 or MLLT3 plays in a KMT2A fusion [E1]. In low-grade endometrial stromal sarcoma, the reciprocal t(X;17)(p11.2;q21.33) puts MBTD1 **5'** of EZHIP (previously CXorf67, also CATACOMB), and MBTD1 occupies the N-terminal position that KMT2A's own CXXC-and-menin-binding block occupies [E2][E3]. **The dispatch brief for this beat, following Q121's own counter-evidence note, predicted that 'what MBTD1 supplies must differ when it donates its N-terminus versus its MBT repeats'. The literature says otherwise, and the measurement is unambiguous.** Piunti et al. state that the sarcoma chimera 'contains the majority of the MBTD1 protein sequence (lacking only the final 39 residues)' [E3] - that is MBTD1 aa **1-589** of 628, which retains the FCS zinc finger (45-80) and **all four MBT repeats** (141-245, 253-350, 351-456, 464-560) [E4]. The leukaemic construct contains the **entire** 628-residue open reading frame [E1]. **Both fusions therefore carry essentially all of MBTD1, including the H4K20me1/2 reader cage in MBT4, and both carry the EPC1-binding surface.** Piunti et al. further place MBTD1 in the same structural class in both settings, calling it 'the NuA4 subunit gene MBTD1' and generalising that **ESS fusion proteins consist of an N-terminal NuA4 component fused to a C-terminal Polycomb-group subunit** [E3]. **So the correct disanalogy is about role, not about parts.** The same nearly-complete protein is the *effector* at one junction and the *address* at the other, and nothing about the protein decides which. That is the same rule `concept-kmt2a-fusion-polarity-is-inverted-in-sarcoma` states for KMT2A itself - leukaemic KMT2A gives the front, sarcomatous KMT2A gives the back, and every consequence follows - **arrived at independently in a gene with no KMT2A involvement, in the same organ, in the same disease family.** Two independent instances make it a property of fusion architecture rather than a peculiarity of KMT2A. **And the one asymmetry that is real.** In the leukaemia the MBT4 methyl-lysine cage is **experimentally dispensable**: W954A and F951A/W954A/Y958A in fusion numbering both bind Tip60 and both transform [E1]. In the sarcoma, where MBTD1 is the addressing half and its cage is the only candidate address the chimera has, **that cage has never been tested at all** [E5]. The single experiment that would decide whether MBTD1 addresses the sarcoma fusion is a cage mutant in the ESS chimera, and the reagent already exists in the leukaemia paper.

Go deeper
The two fusions side by side
ZMYND11::MBTD1MBTD1::EZHIP
translocationt(10;17)(p15;q21) 1t(X;17)(p11.2;q21.33), reciprocal 2
diseaseAML with minimal differentiationlow-grade endometrial stromal sarcoma
MBTD1 polarity3'5'
MBTD1 retainedthe entire 628-residue ORF in the functional construct 1aa 1-589 - 'lacking only the final 39 residues' 3
FCS zinc finger 45-80retained; deletion still transforms with a reduced phenotype 1retained; untested
MBT repeats 1-4retained; deletion abolishes Tip60 binding and transformation 1retained; untested
MBT4 H4K20me1/2 cage 464-560retained and DISPENSABLE - two cage mutants transform 1retained and NEVER TESTED 5
EPC1-binding surfaceretained and REQUIRED - A661D/L684D/L688D abolishes transformation 1retained; untested
MBTD1's structural roleeffector recruiter - it brings NuA4/TIP60address - the N-terminal position, by the ESS family rule 3
partner's payloada chromatin reader: PWWP reading H3.3K36me3 1a catalytic antagonist of PRC2: an H3K27M-mimicking peptide that blocks EZH2 367
direction of the chromatin effectADDS an activating mark (H4/H2A acetylation)REMOVES a repressive mark (H3K27me2/3 falls)
net chromatin state at targetsacetylation-rich, H3K36me3-positive, H3K27me3-depleted 1globally H3K27me2/3-depleted 3

Read the last three rows together. Both fusions end at chromatin poor in H3K27me3, by opposite mechanisms - one by adding an activating mark to a state the authors propose is refractory to PRC2 invasion, the other by directly poisoning PRC2's catalysis. hyp-both-mbtd1-fusions-converge-on-an-h3k27me3-poor-state states that as something that can be killed, and it is the only claim in this entry that is not already in print somewhere.

Why the brief's version of the disanalogy was wrong, and why the wrong version was more attractive

The dispatch brief said the opposite orientation 'argues against a single mechanistic story about MBTD1 - what it supplies must differ when it donates its N-terminus versus its MBT repeats.' That sentence contains a hidden factual premise: that a 5' partner donates its N-terminus and a 3' partner donates its C-terminal repeats. For MBTD1 that premise is false in both directions. As a 3' partner it contributes its whole ORF including its N-terminus; as a 5' partner it contributes everything except 39 C-terminal residues, including all four MBT repeats. Neither fusion is a fragment of MBTD1 in any useful sense.

The wrong version was attractive because it is usually true. In a KMT2A fusion the 5' partner really does donate its N-terminus and lose its SET domain, and the 3' partner really does donate a C-terminal AHD; the whole breakpoint-cluster-region literature is organised around that. Generalising it to MBTD1 costs nothing to say and is testable in one sentence of one paper. This is the pattern this project keeps paying for: a premise that is true of the index fusion, applied to another family, where it is not.

What survives of the brief's point, and it survives in a stronger form. The brief's underlying instinct - that one mechanistic story about MBTD1 cannot cover both fusions - is correct. It is just not correct for the stated reason. The story that cannot cover both is role: in the leukaemia MBTD1 is the thing that fetches the enzyme, and in the sarcoma it is the thing that decides where. An entry reasoning from 'MBTD1 recruits NuA4/TIP60' to a prediction about the sarcoma fusion would be making a category error, and so would an entry reasoning from 'MBTD1 reads H4K20me1/2' to a prediction about the leukaemic one - because that reading function is exactly the one the leukaemia paper showed to be unnecessary.

The third instance nobody has looked for

Two instances of a rule are a pattern; three would make it worth a systematic scan. This graph now holds two independent cases in which a single gene supplies opposite structural roles in two fusions of two different diseases - KMT2A (concept-kmt2a-fusion-polarity-is-inverted-in-sarcoma) and MBTD1 (this entry). Both were found by accident, KMT2A's while writing about sarcoma and MBTD1's while writing about leukaemia.

The scan that would find the rest is cheap and has not been run: for every gene with two or more recurrent fusion partners in a curated fusion catalogue, ask whether it appears as both the 5' and the 3' partner, and if so whether the retained domain sets differ. A gene that appears on both sides of a junction is a natural experiment in whether the parts or the position carry the mechanism, and it needs no new material. EPC1 is an immediate candidate from the same sarcoma family - it is a 5' partner in EPC1-PHF1 and, as the subunit that grips MBTD1's MBT repeats, is functionally on the effector side of ZMYND11::MBTD1 without being in it 38. That is not the same thing as a polarity reversal and is not claimed as one; it is where a scan would start.

This is recorded as a task rather than performed, because no fusion catalogue was queried in this beat and asserting a count from a catalogue nobody opened is the failure mode this project has a rule about.

Facts and figures
domainengineering
domain noteA within-biology architectural comparison, filed as an analogy for the PI's fifth dimension. The comparison here is between two fusions of the SAME gene rather than between two fusion families, which is what makes the polarity variable isolable.
the one asymmetry that is realMBT4's H4K20me1/2 cage is retained in both and tested in only one. In the leukaemia it is dispensable (two independent cage mutants transform). In the sarcoma, where it is the chimera's only candidate address, it has never been tested - and the mutant reagent already exists in the leukaemia paper.
shown for fusionNothing in this comparison has been measured in a KMT2A fusion. The sarcoma arm's chromatin result is in HEK293T and HCT-116 with an ectopic construct the authors themselves call a 'putative' fusion protein; no ESS cell or tissue appears in it. The leukaemia arm is murine primary HSPCs.
epigenetic readoutMBTD1 reads H4K20me1/me2 through MBT repeat 4; the leukaemic fusion writes H4 and H2A acetylation through recruited KAT5 and does NOT need the reader; the sarcoma fusion removes H3K27me2/3 by blocking EZH2 catalysis. Both fusions end at H3K27me3-poor chromatin.
cell typeMurine bone-marrow HSPCs and human AML blasts on the leukaemia side; low-grade endometrial stromal sarcoma tumour tissue on the sarcoma side, in which no chromatin measurement has been made. HEK293T and HCT-116 for the sarcoma biochemistry.
absence classnever-measured
absence class noteThe polarity comparison has never been instrumented. No experiment has tested the H4K20me1/2 cage in the sarcoma chimera, no chromatin measurement exists in MBTD1::EZHIP tumour tissue, and no fusion catalogue has been scanned for other genes appearing on both sides of a junction. The first two need new material - a mutant construct, a tissue series - so never-measured. THE THIRD IS DIFFERENT AND IS DELIBERATELY LEFT UNCLASSIFIED HERE: a both-polarities scan over a curated fusion catalogue is pure compute, but no catalogue was queried in this beat, so classifying it would be classifying an analysis I have not attempted.
mapping
[
  {
    "bio": "MBTD1 as the 3' partner of ZMYND11::MBTD1: entire 628-residue ORF retained, role = effector recruiter, brings NuA4/TIP60 via EPC1, EPC1 contact REQUIRED, H4K20me1/2 cage DISPENSABLE",
    "analog": "the role AFF1, MLLT3, MLLT1 or ELL plays in a KMT2A fusion"
  },
  {
    "bio": "MBTD1 as the 5' partner of MBTD1::EZHIP: aa 1-589 retained (all but the final 39), role = address by the ESS N-NuA4/C-PcG rule, cage retained and NEVER TESTED",
    "analog": "the role KMT2A's own AT-hook/CXXC/MBM block plays in a leukaemic KMT2A fusion"
  },
  {
    "bio": "the partner's payload in the leukaemia is a chromatin READER (ZMYND11 PWWP, H3.3K36me3)",
    "analog": "the partner's payload in the sarcoma is a catalytic ANTAGONIST (EZHIP, an H3K27M-mimicking EZH2 blocker)"
  },
  {
    "bio": "leukaemic outcome: acetylation added at H3K36me3-marked, H3K27me3-devoid chromatin",
    "analog": "sarcoma outcome: H3K27me2/3 removed globally - opposite mechanism, same H3K27me3-poor endpoint"
  },
  {
    "bio": "KMT2A: front in leukaemia, back in sarcoma, every consequence follows (concept-kmt2a-fusion-polarity-is-inverted-in-sarcoma)",
    "analog": "MBTD1: back in leukaemia, front in sarcoma - the SECOND independent instance, in a gene with no KMT2A involvement"
  }
]
generated hypotheses
[
  "hyp-both-mbtd1-fusions-converge-on-an-h3k27me3-poor-state"
]
the measurement that settles the brief
{
  "question": "Does MBTD1 donate different parts of itself to its two fusions?",
  "answer": "NO. Leukaemic construct: MBTD1 aa 1-628, the entire ORF. Sarcoma fusion protein: MBTD1 aa 1-589, 'lacking only the final 39 residues'. Both retain the FCS zinc finger and all four MBT repeats.",
  "source_for_the_39": "PMID 31281901, full text",
  "arithmetic_is_mine": "628 - 39 = 589",
  "what_differs_instead": "the structural ROLE the junction assigns - effector recruiter at one, address at the other"
}
How this could bite us back
  • If the junction rather than the parts assigns the role, then a therapeutic argument built on a partner's known biology is unsafe in any fusion whose polarity differs from the index case. This project's own menin argument is polarity-locked: it works because leukaemic KMT2A is 5' and keeps exon 1. Every entry that reasons from 'this gene does X' to 'therefore the fusion does X' inherits that exposure, and concept-kmt2a-fusion-polarity-is-inverted-in-sarcoma exists because the inference was made and was wrong.
  • A PRC2-inhibiting fusion is a warning about EZH2 inhibitors reaching this family at all. If a tumour's lesion IS loss of PRC2 catalysis, then an EZH2 inhibitor is pushing in the direction the tumour already went, and the rational move is the opposite - which is why PFA ependymoma work has looked at restoring rather than further blocking H3K27 methylation. Any pan-Polycomb therapeutic proposal for low-grade ESS, of the kind the 2026 review floats, has to say which fusion it means.
  • The dispensable-cage result is the most quotable line in the MBTD1 literature and it is a negative from one overexpression transformation assay in mouse cells. If it is wrong - if the cage contributes at endogenous expression - then H4K20me1/me2 re-enters the mechanism of a fusion this graph has already written down as not needing it, and the MBTD1 arm of the H4K20 story would have to be rebuilt. A negative that convenient deserves a replication before anything is built on it.
  • EZHIP is screened in only 75 DepMap cell lines, with a POSITIVE mean gene effect of +0.178 and no line below -0.5 (Open Targets dataVersion 26.06). So there is no dependency signal and, more importantly, no denominator worth interpreting. Nobody should read 'not a dependency' out of 75 screens of an X-linked gene expressed in almost no cell line.
Still open
  1. Does the MBTD1::EZHIP chimera need MBTD1's H4K20me1/2 cage? In the leukaemia the cage is dispensable because MBTD1 is only fetching the enzyme; in the sarcoma it is the only address the chimera has. The cage mutants already exist as reagents (W954A and the FWY>AAA triple, in leukaemia-fusion numbering; MBTD1 W529 and F526/W529/Y533 by the published offset). This is the single experiment that decides whether MBTD1 addresses the sarcoma fusion, and it is one construct.
  2. Where exactly is the MBTD1::EZHIP transcript junction? The founding report is unreadable through all five routes, so the exon boundary is not in this graph. The protein-level boundary - MBTD1 minus its final 39 residues - comes from a second paper and is not the same fact. One library request, or one email to the authors, closes it.
  3. Is there a third gene that appears as both the 5' and the 3' partner of two recurrent fusions in different diseases? Two instances exist (KMT2A, MBTD1) and both were found by accident. The scan over a curated fusion catalogue is compute, needs no new material, and would say whether polarity-assigns-role is a general property of fusion architecture or a coincidence of two genes.
  4. Does MBTD1::EZHIP still enter NuA4/TIP60? It retains the EPC1-binding MBT repeats, so the prediction is yes - which would make it a fusion that delivers a PRC2 poison INTO an acetyltransferase complex. Nobody has run the co-immunoprecipitation, and the answer changes the mechanism from 'a mislocalised PRC2 inhibitor' to 'an acetyltransferase complex carrying a PRC2 inhibitor', which are different lesions with different predicted target sets.
  5. Is the H4K20me1/2 cage doing anything in ANY MBTD1 context that has been tested? It is MBTD1's defining biochemical feature, it is how MBTD1 is described in every review, it is required for the TIP60 complex's 53BP1 competition in DNA repair - and it is dispensable in the one fusion where it has been mutated. The mark entry mark-h4k20me1 records what that costs the H4K20 story; this entry records that the cost has not been paid twice, because the second fusion was never asked.
Every source behind this entry (8)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the leukaemic fusion the functional construct contains the entire 628-residue MBTD1 open reading frame; deletion of the four MBT repeats or the triple substitution A661D/L684D/L688D in fusion numbering abolishes both Tip60 binding and transformation of murine haematopoietic stem/progenitor cells; deletion of the FCS zinc finger still transforms with a reduced phenotype; and the H4K20me1/2-cage-defective mutants W954A and F951A/W954A/Y958A both bind Tip60 normally and transform normally.

    PMID 33594072DOI
    What this evidence cannot show

    Murine primary HSPCs with a retrovirally overexpressed human cDNA; every mutant result is a transformation assay in that system and none is in a human cell. The cage mutants' NEGATIVE result - the load-bearing row of this entry's table - is a negative in one assay and one readout, so it establishes that the cage is not required for THAT phenotype, not that the cage does nothing. Full text read via PMC7886901; identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • A dispensability result from an overexpression assay is the weakest kind of negative available. If the construct is expressed far above endogenous levels, a mutant with a partial defect in target selection can still reach enough sites to transform, and the cage would then be a genuine contributor that the assay cannot see. The paper's own PHD result shows the system producing exactly this kind of ambiguity in the other direction - PHD deletion abolishes leukaemogenesis while only mildly impairing H3K36me3 binding in vitro, which the authors flag as unexplained. Jacquet K et al. 2016
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In low-grade endometrial stromal sarcoma, MBTD1 is the 5' partner of a novel reciprocal t(X;17)(p11.2;q21.33) producing an MBTD1-CXorf67 fusion transcript, identified in two independent tumours by whole-transcriptome paired-end RNA sequencing with FISH and banding cytogenetics and validated by RT-PCR with Sanger sequencing, with a third case found by a purpose-built FISH assay among 25 further uterine stromal tumours negative for JAZF1 and YWHAE rearrangements; expression profiling clustered the fusion-positive tumours with low-grade JAZF1-associated ESS.

    Identification of a novel, recurrent MBTD1-CXorf67 fusion in low-grade endometrial stromal sarcoma. · Dewaele B, Przybyl J, Quattrone A, Finalet Ferreiro J, et al. · Int J Cancer 2014
    PMID 23959973DOI
    What this evidence cannot show

    ABSTRACT LEVEL ONLY, and this is a case where all five routes of the unreadable test fail: Europe PMC's core record returns pmcid null, inEPMC N and isOpenAccess N; the efetch ArticleIdList carries no pmc id; and route 5, curl -sL on the DOI with a browser User-Agent, returns HTTP 403 with a 5,562-byte Cloudflare interstitial that strips to 60 characters of text. THE TRANSCRIPT-LEVEL EXON JUNCTION OF THIS FUSION IS THEREFORE NOT ESTABLISHED IN THIS BEAT; the protein-level boundary used throughout this entry comes from E3 instead. The expression-clustering result is 11 tumours total and is an association. Identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • Three cases, one institution, one discovery cohort. 'Recurrent' at n=3 in a referral series is a statement about ascertainment as much as about biology, and the 22 fusion-negative tumours in the same screen are the denominator that matters: the fusion is rare even among JAZF1- and YWHAE-negative uterine stromal tumours. Two later reviews list it as one of five or six low-grade ESS fusions without adding cases, so the reported total has not grown much in a decade. Hoang L et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Piunti et al. call MBTD1 'the NuA4 subunit gene MBTD1', report that the MBTD1-CATACOMB fusion protein contains the majority of the MBTD1 protein sequence lacking only the final 39 residues fused to the C-terminal half of CATACOMB, show that both CATACOMB and the MBTD1-CATACOMB fusion protein interact with PRC2 and strongly decrease H3K27me2/3 in chromatin extracts, localise the inhibitory function to a short eutherian-conserved sequence containing a methionine resembling the H3K27M oncohistone whose single substitution abolishes the effect, identify JAZF1 as a new subunit of the NuA4 acetyltransferase complex, and generalise that ESS-associated fusion proteins are characterised by an N-terminal NuA4 component fused to a C-terminal Polycomb-group subunit.

    PMID 31281901DOI
    What this evidence cannot show

    Full text read this session via PMC6609211. THE FUSION EXPERIMENTS ARE ECTOPIC: HA-FLAG constructs expressed in HEK293T and HCT-116, with the fusion protein described in the figure legend as 'the MBTD1-CATACOMB putative fusion protein' - the authors' own word, because the junction was taken from the transcript report and not re-derived. No endometrial stromal sarcoma cell or tissue appears in the paper, so the H3K27me2/3 decrease is a property of the construct in a heterologous cell and not a measurement in the disease. The '39 residues' figure is the paper's; the arithmetic 628 minus 39 equals 589 is mine. Genuinely new to this graph - check-citation-reuse reports this PMID as not cited anywhere before cycle 83.

    Evidence that argues against this (1)
    • The generalisation 'N-terminal NuA4 component fused to a C-terminal PcG subunit' 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 the rule and one of its instances were established together. A 2026 review of the same gene reaches the opposite subunit assignment for MBTD1, stating that 'while MBTD1 is not a subunit of PRC2, it is associated with PRC1'. That is wrong against three primary papers, but its existence shows the NuA4 assignment is not universally carried in the secondary literature, and a rule whose category membership is contested is weaker than it reads. Yu X et al. 2026
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    MBTD1 is UniProt Q05BQ5 (MBTD1_HUMAN), 628 amino acids, with an FCS-type zinc finger at 45-80 and four MBT repeats at 141-245, 253-350, 351-456 and 464-560; retaining residues 1-589 therefore retains the FCS zinc finger and all four MBT repeats including the H4K20me1/2-binding cage in the fourth.

    What this evidence cannot show

    Fetched live 2026-09-23, including the FASTA sequence, which was used independently in this beat to align PDB 4C5I (an exact match to residues 130-566 after a 3-residue cloning tag) and PDB 6NFX (residues 140-562 in a single-chain chimera with EPC1 644-671). gene_exact:MBTD1 with organism_id:9606 and reviewed:true returns exactly one record with MBTD1 as primary gene name, so the several-rows trap does not fire. Domain boundaries are curation and are revised between releases; the inference that 1-589 spans them is arithmetic on those boundaries, not a statement from any paper.

    Evidence that argues against this (1)
    • Retaining a domain interval is not retaining a function. MBT repeat 4 ends at 560 and the retained segment ends at 589, so the cage is inside the fusion by 29 residues - but a C-terminal truncation 39 residues from the end could still destabilise the adjacent repeat, and nothing in either fusion literature reports a folding or binding assay on the sarcoma chimera. The claim that the cage is present in the sarcoma fusion is therefore a claim about primary sequence only. Eryilmaz J et al. 2009
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Bounded literature absence, 2026-09-23: the whole primary literature naming both partners of the sarcoma fusion is four records. MBTD1[tiab] AND CXorf67[tiab] returns 4 (PMIDs 42238311, 29275929, 24530230, 23959973 - a review, a review, a different-fusion paper and the founding report); MBTD1[tiab] AND EZHIP[tiab] returns 0, so no paper has yet written the fusion under the current HGNC symbol; MBTD1[tiab] AND sarcoma[tiab] returns 6; and MBTD1[tiab] AND H3K27me3[tiab] returns 1, which is the 2026 review and not a measurement.

    What this evidence cannot show

    One [tiab] tag per term, never the parenthesised group form; every querytranslation read back from the response; pubmed_safe.py refuses to return a zero on an empty translation. THE NAME LIST MATTERS MORE THAN USUAL HERE AND IS GIVEN IN FULL: MBTD1, CXorf67, EZHIP, CATACOMB. CXorf67[tiab] alone returns 21 and EZHIP[tiab] alone returns 62, so the literature is split across a retired and a current symbol and the fusion is written only under the retired one - a paper describing it as MBTD1::EZHIP would be invisible to the query that finds the founding report and vice versa. CATACOMB[tiab] returns 20 and is CONTAMINATED: the hits include records from 1996 and 2005 and archaeological or anatomical uses of the ordinary English word, so it must be title-checked and cannot be used as a bare count. Titles and abstracts only; rule 80 applies.

    Evidence that argues against this (1)
    • A four-record literature cannot support a claim that the cage has never been tested - only that no title or abstract says it was. A mutagenesis panel on the sarcoma chimera could sit in a supplementary figure of the Piunti paper, which was read in full this session for its text but not exhaustively for its supplements, or of any paper about EZHIP in another tumour. The absence is bounded by what an abstract index can see, which for a mutant in a figure panel is close to nothing. Piunti A et al. 2019
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    EZHIP/CXorf67 inhibits PRC2 by a mechanism closely resembling the H3K27M oncohistone: a short, highly conserved peptide in its C-terminal region mimics the sequence of K27M-mutated histones and binds the SET domain of EZH2, blocking methyltransferase activity and de-repressing PRC2 target genes, and its overexpression is the hallmark of posterior fossa group A ependymoma, a tumour characterised by loss of H3K27me3.

    PMID 30923826DOI
    What this evidence cannot show

    Abstract-level in this beat. The system is ependymoma, not sarcoma and not leukaemia, and nothing in it concerns MBTD1 or any fusion; it is cited to establish what the sarcoma fusion's 3' payload does. The first author's surname carries a diacritic in the source (Hubner with an umlaut) which is transliterated here. Genuinely new to this graph. Identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • The H3K27M analogy is a mechanism claim about a peptide, and the two lesions are not equivalent in effect. Jain et al. show that both EZHIP and H3K27M cause loss of BROAD H3K27me3 domains while H3K27me3 is RETAINED at CpG islands, and that H3K27me3-mediated allosteric activation of PRC2 is what makes the peptide inhibitors potent - so the effect is a redistribution of a mark rather than its abolition, and an entry describing either lesion as 'PRC2 off' would overstate it. Jain SU et al. 2019
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    EZHIP is UniProt Q86X51 (EZHIP_HUMAN), 503 amino acids, whose feature table marks residues 401-409 as sufficient for interaction with EZH2 and residues 403-423 as necessary and sufficient for inhibition of PRC2/EED-EZH1 and PRC2/EED-EZH2 complex activity; the gene is HGNC:33738 at Xp11.22, primary Ensembl id ENSG00000187690, with CXorf67 as its PREVIOUS OFFICIAL SYMBOL and CATACOMB as an alias.

    What this evidence cannot show

    Both records fetched live 2026-09-23. The inhibitory region at 403-423 sits in the C-terminal 20% of a 503-residue protein, which is consistent with a fusion retaining 'the C-terminal half of CATACOMB' carrying it - but that consistency is an inference from two independent sources and not a statement that the retained fragment includes 403-423, because the transcript junction is unreadable (see E2). CXorf67 being a PREVIOUS OFFICIAL SYMBOL and not merely an alias matters under this project's rule 69: the literature is split across the two spellings and neither search reaches the other.

    Evidence that argues against this (1)
    • A feature-table region labelled 'necessary and sufficient' is a curator's summary of in vitro reconstitution, and the germ-cell literature gives a different emphasis: Ragazzini et al. describe EZHIP as CONSTRAINING rather than abolishing PRC2 activity in germ cells, a physiological role in which the protein is a tuner. A fusion that delivers a tuner to a new address is a different proposition from one that delivers an off switch, and the sarcoma literature has never distinguished them. Ragazzini R et al. 2019
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    EPC1 is the non-catalytic NuA4/TIP60 subunit that enlists MBTD1 into the complex, its hydrophobic C-terminal fragment engaging the MBT repeats at a site distinct from the H4K20me-binding site; in PDB 6NFX the single 470-residue engineered entity aligns as MBTD1 Q05BQ5 140-562 joined to EPC1 Q9H2F5 644-671.

    PMID 32209463DOI
    What this evidence cannot show

    Paper read at abstract level; the spans were derived by aligning 6NFX's fetched canonical entity sequence against UniProt FASTA downloaded the same session rather than read off the deposit's numbering. 6NFX is a SINGLE-CHAIN ENGINEERED CHIMERA, so it models the bound state rather than measuring association. Cited here only to identify EPC1 as the contact, because EPC1 is the candidate third instance named in the last section of this entry.

    Evidence that argues against this (1)
    • EPC1's appearance as a 5' fusion partner in EPC1-PHF1 and its role as MBTD1's NuA4 anchor are two different facts about one protein, and treating their coincidence as evidence for a polarity rule would be a category error of exactly the kind this entry warns against. EPC1-PHF1 is a NuA4-to-Polycomb fusion in the same family; nothing shows EPC1 appearing on both sides of a junction, which is what the rule requires. Piunti A et al. 2019
Connected to