Oncofusion

start / graph / fusion-mbtd1-ezhip

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

MBTD1::EZHIP (MBTD1-CXorf67), t(X;17)(p11.2;q21.33) - MBTD1's other fusion, in which a NuA4 subunit delivers an H3K27M-mimicking PRC2 poison to low-grade endometrial stromal sarcoma

also called MBTD1-CXorf67 · MBTD1::CXorf67 · MBTD1-CATACOMB · t(X;17)(p11.2;q21.33)

The same gene that forms a leukaemia fusion also forms a fusion in a rare uterine cancer, but the other way round - here it is at the front, and the piece it drags along is a decoy that jams the cell's main gene-silencing enzyme. This is the entry that lets this project compare the two, because the only thing that changes between them is which side of the join the shared gene sits on.

The same thing, for a scientist

**A reciprocal t(X;17)(p11.2;q21.33) fuses MBTD1 (17q21.33) as the 5' partner to EZHIP (Xp11.22) as the 3' partner**, producing a chimera reported in low-grade endometrial stromal sarcoma of classical histology. It was found in two tumours by whole-transcriptome paired-end RNA sequencing with FISH and banding cytogenetics, validated by RT-PCR and Sanger sequencing, and a third case was recovered with a purpose-built FISH assay from 25 further uterine stromal tumours that were negative for *JAZF1* and *YWHAE* rearrangements; expression profiling clustered the fusion-positive tumours with low-grade JAZF1-associated ESS [E1]. **The 3' partner is the interesting half. EZHIP - previous official symbol *CXorf67*, also called CATACOMB - is a catalytic antagonist of PRC2.** A short, highly conserved peptide in its C-terminal region mimics the sequence around the H3K27M oncohistone, binds the SET domain of EZH2 and blocks methyltransferase activity, de-repressing PRC2 targets; its overexpression is the hallmark of posterior fossa group A ependymoma, a tumour defined by loss of H3K27me3 [E3][E4][E5]. **The fusion protein carries this activity.** Piunti et al. showed that both CATACOMB and the MBTD1-CATACOMB fusion protein interact with PRC2 and **strongly decrease H3K27me2/3** in chromatin, and localised the effect to a single methionine whose substitution abolishes it [E2]. **And the 5' partner places this fusion in a family with a published architectural rule.** Piunti et al. identified JAZF1 as a new NuA4 subunit while characterising these chimeras, call MBTD1 'the NuA4 subunit gene MBTD1', and generalise that **ESS-associated fusion proteins are characterised by an N-terminal NuA4 component fused to a C-terminal Polycomb-group subunit** [E2]. JAZF1-SUZ12, EPC1-PHF1, MEAF6-PHF1 and JAZF1-PHF1 all fit; MBTD1::EZHIP is the member whose C-terminal payload inhibits Polycomb rather than joining it [E1][E2][E6]. **Why this entry exists in a leukaemia graph.** MBTD1 is the 3' partner of ZMYND11::MBTD1, where its job is to dock the chimera into NuA4/TIP60 through EPC1. Here it is 5'. **Piunti's measurement settles what the two fusions share: the sarcoma chimera 'contains the majority of the MBTD1 protein sequence (lacking only the final 39 residues)' - aa 1-589 of 628, retaining the FCS zinc finger and all four MBT repeats [E2][E7], and the leukaemic construct contains the whole ORF [E8].** Nearly the same protein, opposite roles. `analogy-mbtd1-occupies-both-fusion-roles-polarity-not-parts` develops that; this entry holds the sarcoma fusion's own facts. **What is not known is most of it.** The transcript-level exon junction is not in this graph, because the founding report fails all five routes of the unreadable test [E1]. No chromatin measurement exists in fusion-positive tumour tissue - the H3K27me2/3 result is an ectopic construct in HEK293T and HCT-116, and the authors call the construct a 'putative' fusion protein [E2]. No mouse model, no cell line, no occupancy data, no test of whether the chimera still enters NuA4/TIP60, and no test of whether MBTD1's methyl-lysine cage addresses it.

Go deeper
What the translocation makes, as far as the readable literature establishes it

At the DNA level the rearrangement is a reciprocal t(X;17)(p11.2;q21.33), with MBTD1 at 17q21.33 and EZHIP at Xp11.22 - the two cytobands agreeing with the karyotype 179.

At the protein level the chimera retains MBTD1 residues 1-589 (628 minus a final 39) joined to the C-terminal half of EZHIP 2. On UniProt Q05BQ5's feature table that retains the FCS-type zinc finger (45-80) and MBT repeats 1-4 (141-245, 253-350, 351-456, 464-560), including the H4K20me1/2-binding cage in the fourth, which ends 29 residues inside the retained segment 7. On UniProt Q86X51 the PRC2-inhibitory region is at 403-423 of 503 with residues 401-409 sufficient for EZH2 interaction 9, so 'the C-terminal half' is consistent with the inhibitory peptide being present - and the fusion protein's measured activity says it is 2.

At the transcript level, nothing. No exon junction is recorded here. Dewaele et al. validated the fusion transcript by RT-PCR and Sanger sequencing and presumably state the junction, and the paper is unreachable: Europe PMC returns pmcid: null, inEPMC: N, isOpenAccess: N; efetch's 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 1. The exon boundary is therefore a known-but-unreadable fact, not an absence, and is recorded that way rather than guessed.

One naming note that has practical consequences. CXorf67 is the PREVIOUS OFFICIAL SYMBOL of EZHIP, not an alias 9. Every paper that names this fusion names it MBTD1-CXorf67, and MBTD1[tiab] AND EZHIP[tiab] returns 0 while MBTD1[tiab] AND CXorf67[tiab] returns 4 10. Meanwhile EZHIP[tiab] alone returns 62 and CXorf67[tiab] alone returns 21 - the PRC2 literature has migrated to the new symbol and the fusion literature has not. A search on either symbol alone misses most of the relevant work, which is rule 69 operating on a gene this graph had never written.

The mechanism, and the three things that stand between it and being established

What is measured. CATACOMB/EZHIP interacts with PRC2; so does the MBTD1-CATACOMB fusion protein; both strongly decrease the catalytic products H3K27me2 and H3K27me3 in chromatin extracts, by Western blot with histone H3 as loading control, in HEK293T; the effect maps to a short eutherian-conserved sequence containing a methionine resembling H3K27M, and a single substitution of that methionine abolishes it 2. Independently, in ependymoma, the same peptide is shown to bind the EZH2 SET domain and block methyltransferase activity 3, with the mechanistic refinement that H3K27me3-mediated allosteric activation of PRC2 is what makes the inhibition potent, and that the consequence is loss of broad H3K27me3 domains with retention at CpG islands rather than global abolition 4.

Three gaps, in order of how much they matter.

1. Nothing has been measured in the disease. Every fusion experiment is ectopic expression in an immortalised line. No H3K27me3 immunohistochemistry has been reported on a fusion-positive endometrial stromal sarcoma, and MBTD1[tiab] AND H3K27me3[tiab] returns exactly one record, a 2026 review 10. This is the cheapest open question in the entry, because H3K27me3 and EZHIP immunohistochemistry are already routine diagnostics - used for PFA ependymoma, and in 2026 shown to have 98.0% sensitivity for loss of H3K27me3 and 95.1% for EZHIP positivity in CNS germinoma across 126 cases 5. The antibodies, the scoring conventions and the archival blocks all exist. exp-h3k27me3-and-ezhip-ihc-in-archival-mbtd1-ezhip-sarcoma states it.

2. Nobody has asked whether the chimera enters NuA4/TIP60. It retains the MBT repeats and therefore the EPC1-binding surface that the leukaemic fusion requires 811. If it does enter the complex, the lesion is not 'a mislocalised PRC2 inhibitor' but 'an acetyltransferase complex carrying a PRC2 inhibitor', which predicts a different target set and a coupling between two chromatin systems that no paper on either fusion has considered. One co-immunoprecipitation answers it.

3. Nobody has asked what addresses it. MBTD1's only reader module is the MBT4 H4K20me1/2 cage - and in the leukaemic fusion, where MBTD1 is merely fetching the enzyme, that cage is experimentally dispensable 8. Here MBTD1 is the N-terminal module, so the cage is the chimera's only candidate address, and it has never been mutated in this context. The mutant reagents exist: the leukaemia paper's W954A and F951A/W954A/Y958A correspond by the published construct offset to MBTD1 W529 and F526/W529/Y533, and those three residue identities were checked against the live Q05BQ5 sequence in this beat and match 78.

Clinical picture, such as it is

Low-grade endometrial stromal sarcoma of classical histology, indistinguishable on the reported expression profile from the JAZF1-associated low-grade tumours it clustered with 1. Three cases in the founding report - two discovery, one from a 25-tumour FISH screen - and no case series since; two subsequent reviews list the fusion among the low-grade ESS fusions without adding patients 612. No outcome data specific to this fusion exists, and it is not distinguished from other low-grade ESS in any staging or treatment recommendation this beat found.

Detection has the same problem as the leukaemic fusion and worse. The founding detection used paired-end RNA sequencing plus a purpose-built FISH assay for formalin-fixed material 1. Neither gene is a routine fusion-panel target, and the 25-tumour screen's design - tumours already negative for JAZF1 and YWHAE - means the reported frequency is conditioned on two prior negatives. Every frequency statement about this fusion is a floor, and a floor computed inside a doubly-selected denominator.

The therapeutic inference that is floating in the secondary literature, and why it needs care. A 2026 review observes that more than 60% of ESS cases carry translocations involving Polycomb-group genes and asks whether Polycomb inhibitors could be therapeutic candidates 13. For this fusion that question points the wrong way: the lesion is loss of PRC2 catalysis, so an EZH2 inhibitor pushes in the direction the tumour has already gone. The same review also states that MBTD1 'is associated with PRC1', which is wrong against three primary sources that place it in NuA4/TIP60 2811, and that misassignment is exactly what makes the family look like a Polycomb story rather than a NuA4-to-Polycomb story. Recorded here so no entry in this graph inherits it.

Facts and figures
gene5MBTD1
gene3EZHIP
gene3 symbol noteEZHIP is the current HGNC symbol (HGNC:33738). CXorf67 is its PREVIOUS OFFICIAL SYMBOL and CATACOMB an alias. The entire fusion literature writes MBTD1-CXorf67 and MBTD1[tiab] AND EZHIP[tiab] returns 0 records, so both spellings must be searched; the id uses the current symbol per this graph's convention.
cytobandt(X;17)(p11.2;q21.33), reciprocal
gene5 cytoband17q21.33
gene3 cytobandXp11.22
uniprot5Q05BQ5
uniprot3Q86X51
length aa5628
length aa3503
ensembl5ENSG00000011258
ensembl3ENSG00000187690
hgnc5HGNC:19866
hgnc3HGNC:33738
detection methodWhole-transcriptome paired-end RNA sequencing with FISH and banding cytogenetics for discovery; RT-PCR with Sanger sequencing for validation; a purpose-built FISH assay for formalin-fixed paraffin-embedded material. NEITHER GENE IS A ROUTINE FUSION-PANEL TARGET and the 25-tumour screen that found the third case was pre-selected for JAZF1 and YWHAE negativity, so every frequency statement is a floor computed inside a doubly-selected denominator.
mechanism summaryA NuA4 subunit delivering a PRC2 catalytic antagonist. MBTD1's near-complete sequence occupies the N-terminal, addressing position; the retained C-terminal half of EZHIP carries an H3K27M-mimicking peptide that binds the EZH2 SET domain and blocks methyltransferase activity. The fusion protein interacts with PRC2 and strongly decreases H3K27me2/3 in chromatin when expressed ectopically. WHAT ADDRESSES IT IS UNKNOWN, and WHETHER IT ENTERS NuA4/TIP60 IS UNKNOWN.
therapeutic leadNONE ESTABLISHED, and the one floated in the secondary literature points the wrong way. A 2026 review asks whether Polycomb inhibitors could treat ESS cases with Polycomb-gene fusions; for this fusion the lesion IS loss of PRC2 catalysis, so an EZH2 inhibitor pushes in the direction the tumour already went. The rational direction, by analogy with PFA ependymoma and H3K27M glioma work, is restoring rather than further blocking H3K27 methylation - and that analogy is not a treatment plan either.
shown for fusionNOTHING in this entry has been shown in a KMT2A fusion or in any leukaemia, and nothing in the KMT2A literature has been shown on this fusion. Within its own disease: the chromatin effect is shown for the FUSION PROTEIN in HEK293T and HCT-116, and for no patient sample of any kind.
cell typeLow-grade endometrial stromal sarcoma tumour tissue in the three reported patients, in which no chromatin or protein measurement has been made. HEK293T and HCT-116 for all fusion biochemistry. No fusion-positive cell line and no mouse model exist.
cell stateNOT CHARACTERISED. No proliferation, differentiation or self-renewal phenotype has been attributed to this fusion in any system. The only cell-state statement available is the founding report's expression clustering with low-grade JAZF1-associated ESS, which is a transcriptome similarity and not a state assignment.
subcellular localizationNOT MEASURED for the fusion protein. Inferred nuclear from both parents; MBTD1's UniProt locations are Nucleus (ECO:0000305, curator inference) and Chromosome (ECO:0000269).
subnuclear localizationNOT MEASURED. No speckle, PML-body, nucleolar or chromatin-compartment assignment exists, and unlike the leukaemic fusion there is not even a genomic occupancy dataset to substitute for one - no ChIP of this fusion has ever been reported.
epigenetic readoutREMOVES a mark rather than writing one. The fusion protein decreases H3K27me2 and H3K27me3 in bulk chromatin by inhibiting EZH2 catalysis. It carries MBTD1's H4K20me1/me2 reader cage, untested. The expected spatial signature, from the ependymoma work on the same peptide, is loss of BROAD H3K27me3 domains with retention at CpG islands - a redistribution a bulk Western blot cannot resolve.
absence classmeasured-not-reported
absence class noteThe load-bearing absence in this entry is H3K27me3 in fusion-positive TUMOUR TISSUE, and it is measured-not-reported rather than never-measured for a specific and checkable reason: H3K27me3 and EZHIP immunohistochemistry are validated, routine assays on formalin-fixed paraffin-embedded material, in current diagnostic use for PFA ependymoma and reported in 2026 across 126 CNS germ cell tumours with 98.0% and 95.1% sensitivity. The three fusion-positive tumours were formalin-fixed - a FISH assay was built for that material - so archival blocks plausibly exist and the stain is an afternoon. SEVERAL OTHER ABSENCES IN THIS ENTRY ARE never-measured AND ARE NOT COVERED BY THIS CLASS: no co-immunoprecipitation of the chimera with NuA4 subunits, no cage mutant in this fusion, no occupancy data, no cell line, no mouse. Those need new material. One more is neither - the exon-level junction is KNOWN BUT UNREADABLE, not absent, and is recorded under concept-known-but-unreadable-literature's standard rather than as an absence.
diseases
[
  "low-grade endometrial stromal sarcoma, classical histology"
]
breakpoints
[
  {
    "id": "bp-mbtd1-ezhip-protein-level-only",
    "population": "the three reported cases, via the construct characterised in PMID 31281901",
    "domains_retained_5": [
      "FCS-type zinc finger (45-80)",
      "MBT repeat 1 (141-245)",
      "MBT repeat 2 (253-350)",
      "MBT repeat 3 (351-456)",
      "MBT repeat 4 (464-560), the H4K20me1/2 cage"
    ],
    "domains_lost_5": [
      "the final 39 residues of MBTD1 (590-628), which lie outside every annotated domain"
    ],
    "domains_retained_3": [
      "the PRC2-inhibitory region of EZHIP (403-423 of 503) - INFERRED from 'the C-terminal half of CATACOMB' plus the fusion protein's measured H3K27me2/3 activity, not stated as a residue span by any source"
    ],
    "domains_lost_3": [
      "the N-terminal half of EZHIP, which is annotated as disordered (1-72 and part of 97-462) and carries no assigned function"
    ],
    "phenotype_notes": "NO EXON-LEVEL JUNCTION IS RECORDED. The founding report validated the transcript by RT-PCR and Sanger sequencing and is unreadable through all five routes of this project's unreadable test, so the exon boundary is a known-but-unreadable fact. The protein-level boundary - MBTD1 lacking only its final 39 residues, joined to the C-terminal half of CATACOMB - is from PMID 31281901's full text, and the arithmetic 628-39=589 is this beat's. The reciprocal EZHIP::MBTD1 product has never been looked for."
  }
]
frequency
{
  "reported_cases_total": 3,
  "reported_cases_note": "Two discovery cases plus one from a 25-tumour FISH screen, all in the 2014 founding report. No case series since; two reviews (2018, 2026) list the fusion without adding patients. No population-based incidence figure exists."
}
required modules
{
  "MBTD1 MBT4 H4K20me1/2 cage": "NEVER TESTED in this fusion. It is the chimera's only candidate address. In the leukaemic fusion the same cage is dispensable, but there MBTD1 is the effector-recruiting half, so the two settings are not comparable and neither result predicts the other.",
  "MBTD1 MBT repeats / EPC1 surface": "NEVER TESTED in this fusion. Retained in primary sequence.",
  "EZHIP PRC2-inhibitory peptide": "REQUIRED for the H3K27me2/3 effect - a single methionine substitution abolishes it in the full-length protein and in the fusion construct."
}
analogous fusion families
[
  "ZMYND11::MBTD1 - the SAME gene with the polarity reversed, and the comparison this entry exists to enable; see analogy-mbtd1-occupies-both-fusion-roles-polarity-not-parts",
  "JAZF1-SUZ12, EPC1-PHF1, MEAF6-PHF1, JAZF1-PHF1 - the rest of the low-grade ESS family, all N-terminal NuA4 component fused to C-terminal Polycomb subunit",
  "KMT2A fusions - a 5' module supplying the address and a 3' module supplying the effector, with the polarity invariant that fails in KMT2A-rearranged sarcoma for the same structural reason it differs here",
  "H3K27M diffuse midline glioma - not a fusion, but the lesion this fusion's payload mimics, and the closest thing to a therapeutic precedent"
]
How this could bite us back
  • The obvious therapeutic thought is the wrong one. A tumour whose driver removes H3K27me3 will not be helped by an EZH2 inhibitor, and the 2026 review that floats Polycomb inhibitors for Polycomb-fusion ESS does not distinguish the fusions that ADD a Polycomb component from the one that POISONS it. Any pan-family proposal has to name the fusion.
  • The inverse proposal is no safer. 'Restore H3K27 methylation' has no agent - there is no EZH2 activator - and the nearest precedent, H3K27M glioma, has produced no approved epigenetic therapy despite a much larger effort. Proposing it here would be proposing a drug class that does not exist for a disease with three reported cases.
  • If MBTD1's contribution turns out to be its promoter rather than its protein, then every mechanistic inference in this entry is about the wrong half of the fusion, and the tractable target is EZHIP's inhibitory peptide - a short, disordered, protein-protein interface of exactly the kind that has resisted small molecules everywhere else it has been tried.
  • This entry's main value to the laboratory is comparative, and that is also its main risk. Using a three-case sarcoma fusion to reason about infant leukaemia inverts the usual direction of inference, and rule 73 applies to every claim in it: the systems are HEK293T, HCT-116 and formalin-fixed tumour blocks, and not one is haematopoietic.
  • EZHIP is screened in only 75 DepMap lines with a POSITIVE mean gene effect of +0.178 and no line below -0.5 (Open Targets dataVersion 26.06). That is not a statement that EZHIP is dispensable; it is a statement that DepMap cannot see it, because the gene is expressed in almost nothing DepMap screens. Nobody should quote it either way.
Still open
  1. Does a fusion-positive endometrial stromal sarcoma lose H3K27me3? The antibody is routine, the tumours were formalin-fixed, and the answer is one stain. If it does not, the ectopic HEK293T result does not transfer and the whole mechanism of this fusion is open; if it does, this becomes the first H3K27me3-deficient sarcoma driven by a fusion rather than an oncohistone. See exp-h3k27me3-and-ezhip-ihc-in-archival-mbtd1-ezhip-sarcoma.
  2. Does the chimera enter NuA4/TIP60? It retains the MBT repeats that constitute the EPC1-binding surface. If yes, the lesion is an acetyltransferase complex carrying a PRC2 poison - a coupling between two chromatin systems no paper on either MBTD1 fusion has considered - and the predicted target set is MBTD1's, not EZHIP's. One co-immunoprecipitation.
  3. What addresses this fusion? MBTD1's only reader is the MBT4 H4K20me1/2 cage. The mutant reagents exist by transposition from the leukaemia paper (MBTD1 W529; F526/W529/Y533). If the cage is dispensable here too, then MBTD1 contributes only a promoter and a scaffold and the fusion is functionally an EZHIP overexpression event - which is the null model this entry's own counter-evidence raises and which nobody has excluded.
  4. Does the MBTD1 promoter matter more than the MBTD1 protein? EZHIP is normally expressed in almost no somatic tissue - it is screened in only 75 DepMap lines - so placing its coding half under a broadly expressed MBTD1 promoter is itself a plausible complete mechanism. Comparing fusion-positive tumours against EZHIP-overexpressing tumours of other types would separate 'a chimeric protein with a new address' from 'an inappropriate promoter', and needs only expression data.
  5. Has the reciprocal EZHIP::MBTD1 product ever been looked for? The translocation is described as reciprocal and the founding report is unreadable. In the leukaemic fusion the reciprocal is predicted to lack a productive start codon; here nobody has said either way, and this graph maintains a reciprocal-product arm precisely because the question is usually not asked.
  6. Why is the same chromosome pair, 10 and 17, involved in both the leukaemic ZMYND11::MBTD1 t(10;17)(p15;q21) and high-grade ESS t(10;17)(q22;p13)? Almost certainly coincidence - different arms, unrelated genes - but it is a collision that will trip any query or reader matching on the cytogenetic string alone, and it is recorded here for that reason rather than as a biological question.
Every source behind this entry (13)
  1. E1strong 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 of classical histology by whole-transcriptome paired-end RNA sequencing with FISH and banding cytogenetics and validated by RT-PCR with Sanger sequencing; a purpose-built FISH assay recovered a third positive case among 25 further uterine stromal tumours negative for JAZF1 and YWHAE rearrangements, and 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 unreadable through all five routes: Europe PMC core returns pmcid null, inEPMC N, isOpenAccess N; the efetch ArticleIdList carries no pmc id, so routes 2-4 cannot be attempted; and route 5 (curl -sL https://doi.org/10.1002/ijc.28440 with a browser User-Agent) returns HTTP 403 with a 5,562-byte Cloudflare interstitial whose stripped length is 60 characters. The exon-level junction, the FISH probe identities and the per-case clinical detail are therefore KNOWN-BUT-UNREADABLE and nothing about them is asserted. Three cases, one discovery cohort, and the 25-tumour screen was pre-selected for JAZF1 and YWHAE negativity. Identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • 'Recurrent' at n=3 in a referral series is a statement about ascertainment as much as about biology, and the reported total has barely grown in a decade: two later reviews list the fusion among low-grade ESS fusions without adding cases. The 22 fusion-negative tumours in the same screen are the denominator that matters, and they say the fusion is rare even among the JAZF1- and YWHAE-negative subset it was sought in. Hoang L et al. 2018
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Both CATACOMB/CXORF67 and the MBTD1-CXORF67 fusion protein interact with PRC2 and strongly decrease the catalytic products H3K27me2 and H3K27me3 in chromatin extracts of cells expressing them; the inhibitory function maps to a short highly conserved region containing a methionine whose single substitution abolishes the effect and whose flanking sequence resembles the H3K27M oncohistone; the fusion protein contains the majority of the MBTD1 protein sequence, lacking only the final 39 residues, joined to the C-terminal half of CATACOMB; MBTD1 is described as a NuA4 subunit gene and JAZF1 is newly identified as a NuA4 acetyltransferase complex subunit; and ESS-associated fusion proteins are generalised as 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. EVERY FUSION EXPERIMENT IS ECTOPIC: HA-FLAG constructs in HEK293T and HCT-116. The authors' own figure legend calls it 'the MBTD1-CATACOMB PUTATIVE fusion protein', because the junction was taken from the transcript report rather than re-derived, so even the construct's identity to the patient protein is the authors' assumption. NO ENDOMETRIAL STROMAL SARCOMA CELL OR TISSUE APPEARS IN THE PAPER. The readout is Western blot of bulk chromatin, so it is a global level and not an occupancy measurement; no ChIP of the fusion is reported. The arithmetic 628 minus 39 equals 589 is mine; the 39 is the paper's. check-citation-reuse reports this PMID as not previously cited anywhere in this graph.

    Evidence that argues against this (1)
    • A global H3K27me2/3 decrease by Western blot cannot distinguish the mechanism this fusion is interesting for - targeted PRC2 inhibition at MBTD1 addresses - from the mechanism the free EZHIP protein already has, which is untargeted inhibition wherever it is expressed. The paper shows full-length CATACOMB doing the same thing as the fusion, so nothing in it establishes that the MBTD1 half contributes anything at all. On this evidence the fusion could be, functionally, an EZHIP overexpression event driven off the MBTD1 promoter. Hubner JM et al. 2019
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    A small, highly conserved peptide in the C-terminal region of CXorf67/EZHIP mimics the sequence of K27M-mutated histones and binds the SET domain of EZH2, blocking EZH2 methyltransferase activity and inhibiting PRC2 function with de-repression of PRC2 target genes including neurodevelopmental genes; overexpression of the gene is a hallmark of posterior fossa group A ependymoma, a tumour characterised by lack of H3K27me3.

    PMID 30923826DOI
    What this evidence cannot show

    Abstract-level in this beat. The system is ependymoma; nothing in the paper concerns MBTD1, any fusion, or any sarcoma, so it establishes what the 3' payload does and not what the fusion does. Mass spectrometry, peptide modelling, immunocytochemistry, Western blot and methyltransferase assays are the stated methods; the peptide-modelling arm is computational. The first author's surname carries an umlaut in the source and is transliterated here. Genuinely new to this graph. Identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • Ragazzini et al. describe EZHIP in its physiological setting as CONSTRAINING PRC2 activity in germ cells rather than switching it off, which makes the protein a tuner whose effect is dose- and context-dependent. A fusion that relocates or overexpresses a tuner is a different lesion from one that delivers an inhibitor, and the two predict different residual H3K27me3 landscapes. Nothing in the sarcoma literature distinguishes them. Ragazzini R et al. 2019
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    A conserved sequence in EZHIP is necessary and sufficient to inhibit PRC2 catalytic activity in vitro and in vivo, contacting the active site of EZH2 in a manner similar to the H3K27M oncohistone; expression of H3K27M or of EZHIP produces similar chromatin profiles, namely loss of broad H3K27me3 domains with retention of H3K27me3 at CpG islands; and H3K27me3-mediated allosteric activation of PRC2 substantially increases the inhibitory potential of both, which explains the observed loss of H3K27me3 spreading in tumours.

    PFA ependymoma-associated protein EZHIP inhibits PRC2 activity through a H3 K27M-like mechanism. · Jain SU, Do TJ, Lund PJ, Rashoff AQ, et al. · Nat Commun 2019
    PMID 31086175DOI
    What this evidence cannot show

    Abstract-level in this beat, and cited specifically for the refinement that the lesion is loss of BROAD H3K27me3 DOMAINS WITH CpG-ISLAND RETENTION rather than global loss of the mark - which is the detail any immunohistochemical or ChIP experiment on the sarcoma fusion has to be designed around, since a bulk Western blot cannot see a redistribution. No MBTD1, no fusion and no sarcoma appears in this paper. Genuinely new to this graph. Identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • The CpG-island retention result is the reason the ectopic Western-blot evidence for the fusion (E2) is weaker than it looks: a global decrease in bulk H3K27me2/3 is consistent with the redistribution this paper describes, with wholesale loss, and with an overexpression artefact, and the three have different consequences for which genes de-repress. Any entry that reads 'the fusion removes H3K27me3' off E2 is over-reading it by one level of spatial resolution. Piunti A et al. 2019
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    EZHIP and H3K27me3 immunohistochemistry are established, validated diagnostic assays on formalin-fixed paraffin-embedded tissue: in a 126-case series of central nervous system germ cell tumours, EZHIP was positive in 95.1% (97/102) of germinomas and germinoma components and loss of H3K27me3 was observed in 98.0% (100/102), against 0/24 EZHIP positivity and 12.5% H3K27me3 loss in non-germinomatous components, giving EZHIP a sensitivity of 95.1% with 100% specificity and H3K27me3 loss a sensitivity of 98.0% with 87.5% specificity.

    PMID 42453183DOI
    What this evidence cannot show

    Abstract-level in this beat, and cited ONLY to establish that the two antibodies are routine, scoreable and validated on archival material - which is what makes the proposed sarcoma experiment an afternoon rather than a project. The disease is a CNS germ cell tumour and has no relationship to endometrial stromal sarcoma; the denominators differ between markers because not every case was stained for every marker (102, 82, 97 across the panel), and the clone identities are not in the abstract. Under this project's rule 74 an immunohistochemical readout is identified by its CLONE and not its label, and no clone is recorded here because none was retrievable at abstract level - so this item supports 'the assay exists and performs' and not 'this specific reagent will work'. Genuinely new to this graph.

    Evidence that argues against this (1)
    • High sensitivity and specificity in a tumour type where the lesion is EZHIP OVEREXPRESSION says little about a tumour where the lesion is an EZHIP FUSION. A chimeric protein may not present the epitope the antibody was raised against, and if the antibody targets the N-terminus of EZHIP it will not detect a fusion that retains only the C-terminal half - which is precisely the fusion described here. The experiment proposed on the back of this item must therefore validate the antibody against the fusion protein, or read H3K27me3 alone. Jain SU et al. 2019
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MBTD1/CXorf67 is one of a set of recurrent low-grade endometrial stromal sarcoma fusions alongside JAZF1/SUZ12, YWHAE/FAM22, ZC3H7/BCOR and recombinations of PHF1 with JAZF1, EPC1 and MEAF6, all of whose partner genes are involved in epigenetic modification.

    MEAF6/PHF1 is a recurrent gene fusion in endometrial stromal sarcoma. · Micci F, Gorunova L, Gatius S, Matias-Guiu X, et al. · Cancer Lett 2014
    PMID 24530230DOI
    What this evidence cannot show

    Abstract-level; a two-case report on a DIFFERENT fusion of the same family, cited only for its enumeration of the family membership, which is the list this entry's family claim rests on. It is not independent of E1 for MBTD1/CXorf67 - it cites it. The paper's framing is 'epigenetic modification' generally and does NOT make the N-terminal-NuA4/C-terminal-Polycomb claim, which comes from E2 alone. Identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • An enumeration in a case report is a snapshot of what was known in 2014 and the family has been revised since - ZC3H7B-BCOR and YWHAE-NUTM2A/B are now the preferred namings, and the high-grade/low-grade boundary has moved. A family membership list is a taxonomy, not a mechanism, and this graph should not read the architectural rule in E2 as endorsed by every paper that lists the same fusions. Hoang L et al. 2018
  7. E7established 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, HGNC:19866 at 17q21.33, primary Ensembl id ENSG00000011258, 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 keeps the FCS zinc finger and all four MBT repeats, and the leukaemia-fusion cage mutants W954A and F951A/W954A/Y958A correspond by the published construct offset to MBTD1 W529 and F526/W529/Y533, whose residue identities match the live sequence.

    What this evidence cannot show

    Fetched live 2026-09-23 with the FASTA sequence. gene_exact:MBTD1 AND organism_id:9606 AND reviewed:true returns exactly one record with MBTD1 as primary gene name, so the several-reviewed-rows trap does not fire. The 1-589 span is arithmetic on a boundary stated in E2 and is not itself published. The offset mapping applies the derivation stated in fusion-zmynd11-mbtd1 (MBTD1 n = fusion n + 425); the three residue identities F, W and Y at MBTD1 526, 529 and 533 were read off the live sequence in this beat and match the mutation names, which validates the offset but does not make it a published statement. Domain boundaries are curation and are revised between releases.

    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 only 29 residues; a truncation that close could destabilise the adjacent repeat, and no folding, binding or thermal-stability assay has been performed on the sarcoma chimera. The claim that the cage is present in this fusion is a claim about primary sequence and nothing more. Eryilmaz J et al. 2009
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the leukaemic fusion ZMYND11::MBTD1 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, while the H4K20me1/2-cage-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. Cited here for the CONTRAST with this fusion - the same protein, nearly the same retained sequence, a different role - and not for anything about sarcoma. The cage-dispensability result is a negative in one assay and one readout in an overexpression system, which is the weakest kind of negative available. Full text read via PMC7886901; identifiers fetched from PubMed 2026-09-23.

    Evidence that argues against this (1)
    • Dispensability in the leukaemic fusion does not predict dispensability here and the entry must not be read as implying it - the whole point of the comparison is that the same module sits in a different structural role. It is equally possible that the cage is dispensable in both, in which case the sarcoma chimera has no reader at all and is functionally an EZHIP overexpression event driven off the MBTD1 promoter, which is the alternative model E2's counter-evidence raises. Piunti A et al. 2019
  9. E9established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    EZHIP is HGNC:33738, 'EZH inhibitory protein', at Xp11.22, with CXorf67 as its previous official symbol and CATACOMB as an alias, primary Ensembl id ENSG00000187690 and UniProt Q86X51 (EZHIP_HUMAN, 503 aa) whose feature table marks residues 401-409 as sufficient for interaction with EZH2 and 403-423 as necessary and sufficient for inhibition of PRC2/EED-EZH1 and PRC2/EED-EZH2 complex activity, with a FUNCTION comment stating that it down-regulates H3K27me3 and probably acts by limiting the stimulatory effect of cofactors such as AEBP2 and JARID2.

    What this evidence cannot show

    Both fetched live 2026-09-23. NOTE A TOOL DEFECT FOUND WHILE ESTABLISHING THIS: scripts/is-symbol-covered.py reported 'CXorf67 NOT IN HGNC under any current, previous or alias symbol - treat any answer as unsafe' in a multi-symbol invocation, which is false; run alone it resolves CXorf67 to EZHIP correctly. The cause is an order-dependence bug recorded in ana-is-symbol-covered-is-order-dependent-and-reports-a-false-unresolvable, and the verdict lines in this beat were all re-taken one symbol at a time. The 401-409 and 403-423 regions are curator summaries of in vitro reconstitution, not measurements this entry made, and 'necessary and sufficient' is the feature description's wording.

    Evidence that argues against this (1)
    • A UniProt region is a curation snapshot and this project has already recorded a case where a feature table omitted a domain a paper had characterised. More specifically here, the FUNCTION comment's mechanism - limiting cofactor stimulation - is attributed to one of four cited papers and is framed as 'probably', which is a weaker statement than the SET-domain-occupancy mechanism E3 and E4 report. The two mechanisms are not the same and the record carries both without adjudicating. Jain SU et al. 2019
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Bounded literature counts, 2026-09-23: MBTD1[tiab] AND CXorf67[tiab] returns 4 (PMIDs 42238311, 29275929, 24530230, 23959973); MBTD1[tiab] AND EZHIP[tiab] returns 0, so no indexed title or abstract has yet named this fusion under the current HGNC symbol; MBTD1[tiab] AND sarcoma[tiab] returns 6; MBTD1[tiab] AND H3K27me3[tiab] returns 1, which is a 2026 review and not a measurement; CXorf67[tiab] alone returns 21 and EZHIP[tiab] alone returns 62.

    What this evidence cannot show

    One [tiab] tag per term, never the parenthesised group form; every querytranslation read back; pubmed_safe.py raises rather than returning a zero on an empty translation, so the MBTD1-plus-EZHIP zero is a genuine zero for that string pair. THE NAME LIST IN FULL: MBTD1, CXorf67, EZHIP, CATACOMB. CATACOMB[tiab] returns 20 and is CONTAMINATED by the ordinary English word - hits date back to 1996 and include non-genomic uses - so it cannot be used as a bare count and was not. The 62/21 split across the current and retired symbols means neither symbol alone reaches the whole literature, which is rule 69's point. Titles and abstracts only; rule 80 applies, and the H3K27me3 count is the clearest illustration - a measurement in a supplementary figure would be invisible.

    Evidence that argues against this (1)
    • A zero on MBTD1 plus EZHIP is a fact about naming practice, not about biology, and it will become false the moment any group writes the fusion under the current symbol - so it is a dated measurement with a short shelf life and must not be re-quoted in a later cycle without re-running. Worse, it means the two literatures on this gene are structurally separated: a PRC2 biologist searching EZHIP will not find the sarcoma fusion, and a sarcoma pathologist searching CXorf67 will miss most of the mechanism. Yu X et al. 2026
  11. E11strong 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, a hydrophobic C-terminal fragment of EPC1 engaging the MBT repeats at a site distinct from the H4K20me-binding site; in PDB 6NFX at 1.95 A the single 470-residue engineered polymer entity aligns as MBTD1 Q05BQ5 residues 140-562 joined through a designed linker to EPC1 Q9H2F5 residues 644-671.

    PMID 32209463DOI
    What this evidence cannot show

    Paper read at abstract level; the residue spans were derived by aligning 6NFX's fetched canonical entity sequence against UniProt Q05BQ5 and Q9H2F5 FASTA downloaded the same session, not read off the deposit's numbering. 6NFX IS A SINGLE-CHAIN ENGINEERED CHIMERA with one polymer entity, so it models the bound state rather than measuring association. Cited here to establish that the MBT repeats this fusion retains are the NuA4-binding surface, which is what makes 'does the sarcoma chimera enter NuA4/TIP60' a well-posed question.

    Evidence that argues against this (1)
    • Retaining a binding surface in primary sequence is not the same as forming the complex, and there is a specific reason for doubt here: the sarcoma chimera's C-terminal 39-residue truncation and its bulky EZHIP appendage could both interfere sterically with an interaction that a designed single-chain construct was built to guarantee. Nothing about 6NFX speaks to affinity or to competition in a cell, and no co-immunoprecipitation of the sarcoma fusion with any NuA4 subunit has been reported. Piunti A et al. 2019
  12. E12strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.review

    Low-grade endometrial stromal sarcoma bears close histopathological resemblance to proliferative-type endometrial stroma, about half harbour t(7;17)(p15;q21) with JAZF1-SUZ12 fusion, and less common JAZF1-PHF1, EPC1-PHF1, MEAF6-PHF1 and MBTD1-CXorf67 fusions have also been reported, while high-grade endometrial stromal sarcoma is defined by t(10;17)(q22;p13) with YWHAE-NUTM2A/B fusion and distinct morphology.

    PMID 29275929DOI
    What this evidence cannot show

    A REVIEW, cited as a map of where this fusion sits in a classification this beat has no independent competence in, and not as primary evidence for anything. Abstract-level. It adds no cases to E1 and is not independent of it. Note the collision hazard the review itself surfaces: the HIGH-grade ESS translocation is t(10;17)(q22;p13) while the ZMYND11::MBTD1 leukaemia translocation is t(10;17)(p15;q21) - same chromosome pair, different arms, different disease, and a query or a reader matching on 't(10;17)' alone will conflate two unrelated entities.

    Evidence that argues against this (1)
    • A 2018 review's classification predates the current WHO and ICC revisions and the reviews themselves note the taxonomy has moved repeatedly - the term 'high-grade endometrial stromal sarcoma' was re-introduced, retired and re-scoped within the period this literature covers. Any statement in this entry about where MBTD1::EZHIP sits in a classification should be read as 'as of 2018, and unchanged in the 2026 review that repeats it', not as current nomenclature. Yu X et al. 2026
  13. E13moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    A 2026 review of CXorf67 in malignancies states that CXorf67 interacts with EZH2 and SUZ12 and thereby influences H3K27me3, that it is overexpressed in posterior fossa ependymoma, diffuse midline glioma, endometrial stromal sarcoma, non-small cell lung cancer and Merkel cell carcinoma, that in endometrial stromal sarcoma it forms fusion genes with MBTD1 potentially disrupting Polycomb-group functions, that more than 60% of ESS cases carry translocations involving Polycomb-group genes including MBTD1, SUZ12 and PHF1, and asks whether Polycomb inhibitors could be therapeutic candidates in that setting.

    PMID 42238311DOI
    What this evidence cannot show

    A REVIEW, full text read this session via PMC13226146, and cited partly as a map and partly as a DOCUMENTED ERROR. It states that 'while MBTD1 is not a subunit of PRC2, it is associated with PRC1, another major component of the PcG complexes' - WHICH IS WRONG. MBTD1's established complex membership is NuA4/TIP60: PMID 27153538 establishes it as a stable NuA4/TIP60 subunit, PMID 32209463 solves the EPC1 contact that recruits it, and PMID 31281901 calls it 'the NuA4 subunit gene MBTD1'. MBTD1 belongs to the MBT-repeat family, which is Polycomb-group by homology, and that is presumably the origin of the slip - but PRC1 membership is a specific claim and it is not supported. The review's therapeutic suggestion is filed here as a hypothesis of the authors' and is argued against in the sections above. Confidence moderate, not strong, because a review carries no primary observation.

    Evidence that argues against this (1)
    • The primary literature contradicts this review on MBTD1's complex membership, and the correction is load-bearing rather than pedantic: the family rule that makes MBTD1::EZHIP interpretable is 'N-terminal NuA4 component fused to C-terminal Polycomb subunit', and if MBTD1 were a PRC1 protein the fusion would be a Polycomb-to-Polycomb event with a quite different predicted mechanism. Reading the review's version would also make an EZH2 inhibitor look rational for this fusion, when the lesion is loss of PRC2 catalysis. Jacquet K et al. 2016
Connected to