One stain, in the right order: H3K27me3 across a low-grade endometrial stromal sarcoma series first, and the three MBTD1::EZHIP blocks second
also called the archival H3K27me3 stain · does the sarcoma fusion actually remove the mark
There is a routine hospital stain that shows whether a tumour has lost a particular chemical tag on its DNA, and nobody has ever run it on the rare uterine cancer whose fusion gene should destroy that tag. The trick is to stain the ordinary cases first, because if they have lost the tag too then the rare cases would prove nothing.
The same thing, for a scientist
**The claim to test.** MBTD1::EZHIP delivers an H3K27M-mimicking peptide that blocks EZH2 catalysis, and the fusion protein strongly decreases H3K27me2/3 - measured only as a bulk Western blot on an ectopic construct in HEK293T and HCT-116 [E1]. **No chromatin measurement of any kind exists in fusion-positive tumour tissue**, and `MBTD1[tiab] AND H3K27me3[tiab]` returns exactly one record, a 2026 review [E2]. **The assay is already a validated diagnostic.** H3K27me3 immunohistochemistry on formalin-fixed paraffin-embedded material is in routine use for posterior fossa group A ependymoma, and in 2026 was reported across 126 CNS germ cell tumour cases with 98.0% sensitivity and 87.5% specificity for germinoma, alongside EZHIP at 95.1% and 100% [E3]. **So the experiment is a stain, not a project.** **And the order is the whole design.** **Step 1 is the control series and it must come first:** H3K27me3 on ten or more fusion-negative low-grade ESS, ideally including JAZF1-SUZ12 cases, from one archive with one protocol. *If the fusion-negative tumours are also H3K27me3-low, the experiment is over and the answer is that the mark is a property of the histology.* That result costs the same as the rest and it can be obtained without locating a single rare case. **Step 2**, only if the controls retain the mark, is the three reported MBTD1::EZHIP blocks [E4], which requires contacting the reporting institution. **Two reagent problems, both of which the design has to state before it runs.** First, **the expected signature is not loss but redistribution** - EZHIP and H3K27M both produce loss of broad H3K27me3 domains with retention at CpG islands [E5] - and a nuclear stain integrates over the genome, so a partial or preserved stain is **uninterpretable rather than negative**. Second, **an EZHIP antibody raised against the N-terminus will not see this fusion protein**, which retains only EZHIP's C-terminal half [E1], while performing perfectly in every tumour it was validated on. **The epitope has to be established from the vendor datasheet before the EZHIP arm means anything**, and under this project's rule 74 the clone, not the label, is the identifier. The H3K27me3 arm has no such problem and is the one that carries the experiment. **Pre-registered scoring.** Complete loss, mosaic loss, or retained, by the convention used for PFA ependymoma, scored blind to fusion status by two readers, with an internal positive control (stromal or endothelial nuclei in the same section) required on every slide.
Why step 1 before step 2, in numbers
The population being tested is three reported tumours 4, and the comparison that makes them mean anything is fusion-negative low-grade ESS from the same archive and the same protocol.
If the controls are H3K27me3-low too, the fusion cases carry no information. That is not a hypothetical concern. Low-grade ESS is dominated by JAZF1-SUZ12 - a fusion whose C-terminal half is a PRC2 core subunit 1 - and nobody has published what that does to H3K27me3 either. A family in which most members fuse something to a Polycomb component is a family in which an H3K27me3 abnormality could be general.
So the control series is the more valuable experiment, and it is the one that does not require finding rare material. Ten cases from any gynaecological pathology archive, one antibody, one run. It answers:
| control result | consequence |
|---|---|
| controls RETAIN H3K27me3 | proceed to step 2; a loss in the fusion cases would be a real, lesion-attributable finding |
| controls are also LOW | stop; report that H3K27me3 loss is a property of low-grade ESS, which is itself a publishable observation and is currently unknown |
| controls are mixed by fusion subtype | the family becomes a natural experiment in PRC2 gain versus PRC2 loss within one histology - the most informative outcome available here, and entirely unexamined |
The third row is the reason to run this even with no MBTD1::EZHIP case in hand. JAZF1-SUZ12 adds a PRC2 subunit and MBTD1::EZHIP poisons PRC2; if those two produce opposite H3K27me3 phenotypes in the same tumour type, that is a within-histology controlled comparison of a kind this project almost never gets.
What this experiment cannot do, stated so nobody builds on it
It cannot resolve a redistribution. The published EZHIP signature is loss of broad domains with CpG-island retention 5. Immunohistochemistry gives one intensity per nucleus. A tumour with half its H3K27me3 relocated could stain anywhere from normal to negative depending on what fraction of that cell type's genome sits in broad domains. A retained stain therefore does NOT refute the fusion's mechanism, and the scoring convention has to record 'retained' as uninformative for mechanism while still informative for diagnosis.
It cannot establish that the fusion protein is expressed. An EZHIP stain would, if the epitope is in the retained C-terminal half - and that is unknown 3. Without it, an H3K27me3-low tumour could be H3K27me3-low for an unrelated reason, and this project has a standing rule that an absence of testing is not a negative result.
It cannot say anything about leukaemia. The leukaemia arm of hyp-both-mbtd1-fusions-converge-on-an-h3k27me3-poor-state needs a different instrument entirely - a base-rate-controlled comparison against H3K36me3-matched unbound regions in deposited occupancy data - and that is compute, not a stain.
And the honest scale. Three tumours, from one 2014 report whose full text is unreadable through all five routes 4. Even a clean result is n=3 in a disease with no outcome data, and the claim it would license is 'the fusion protein's ectopic effect on H3K27me3 also occurs in tissue', which is a validation and not a discovery.
| design | TWO STEPS, IN THIS ORDER. STEP 1, the control series, which requires no rare material: H3K27me3 immunohistochemistry on ten or more fusion-negative low-grade endometrial stromal sarcomas from one archive with one protocol and one antibody lot, ideally including JAZF1-SUZ12-positive cases, scored blind. STEP 2, only if step 1 shows the controls retain the mark: H3K27me3 and EZHIP immunohistochemistry on the three reported MBTD1-CXorf67-positive blocks, which requires contacting the reporting institution - and that request should also ask whether an H3K27me3 stain was already performed as part of the original diagnostic workup, since the absence is classed measured-not-reported. |
|---|---|
| readout | H3K27me3 nuclear staining scored by the convention used for posterior fossa group A ependymoma as complete loss, mosaic loss, or retained, by two readers blind to fusion status, with an internal positive control (stromal or endothelial nuclei within the same section) required on every slide. EZHIP scored positive or negative, and ONLY after the antibody's epitope has been established from the vendor datasheet to lie within EZHIP's C-terminal half. |
| controls | Internal positive control nuclei on every slide, because a global negative stain is indistinguishable from a failed run. Fusion-negative low-grade ESS as the between-case control, which is step 1 and is the design's whole point. A PFA ependymoma or H3K27M glioma section as an external assay-performance control. Blind scoring by two readers. One antibody lot across all cases. |
| cost tier | low |
| timeline | step 1, two to four weeks including block retrieval; step 2, months, and gated on institutional access rather than on technique |
| why the order matters | The fusion-positive population is three reported tumours and the control series is available from any gynaecological pathology archive. If the controls are also H3K27me3-low, the rare cases prove nothing, so the cheap experiment that can invalidate the expensive one has to run first. This is the same discipline as running a positive control through the identical query shape before recording a zero. |
| what this cannot show | It cannot resolve a redistribution, which is the published expected signature - immunohistochemistry gives one intensity per nucleus. It cannot establish that the fusion protein is expressed unless the EZHIP antibody's epitope is in the retained C-terminal half. It cannot distinguish the fusion protein's effect from inappropriate EZHIP expression driven off the MBTD1 promoter, because full-length EZHIP produces the same bulk effect. And it says nothing about the leukaemia arm of the hypothesis, which needs a base-rate-controlled analysis of deposited occupancy data instead. |
| status | planned |
| absence class | measured-not-reported |
| absence class note | H3K27me3 immunohistochemistry is a routine diagnostic stain on formalin-fixed material, in current use for PFA ependymoma and validated in 2026 across 102 CNS germ cell tumour cases; the fusion-positive tumours were formalin-fixed, because a FISH assay was built for that material. So the stain may well have been performed as part of a diagnostic workup and recorded only in a hospital report - which is why step 2's institutional request asks for existing stains as well as for blocks, and why this is measured-not-reported rather than never-measured. STEP 1 IS A DIFFERENT CASE AND IS ALSO measured-not-reported for the same reason, on a much larger population: H3K27me3 has surely been stained on some low-grade ESS somewhere, and nobody has published the series. |
prearranged interpretation
{
"controls_retain_and_fusion_cases_lose": "SUPPORTS hyp-both-mbtd1-fusions-converge-on-an-h3k27me3-poor-state on its sarcoma arm; the first tissue-level evidence that this fusion does what the ectopic construct does",
"controls_retain_and_fusion_cases_retain": "the sarcoma arm is REFUTED at tissue level, subject to the redistribution caveat - a retained stain cannot exclude a redistribution, so the correct report is 'no detectable global loss' and not 'the mechanism is wrong'",
"controls_also_low": "STOP. H3K27me3 loss is a property of low-grade ESS and the fusion cases carry no information. This is a publishable observation in its own right and is currently unknown.",
"controls_differ_by_fusion_subtype": "THE BEST AVAILABLE OUTCOME: the family becomes a within-histology controlled comparison of PRC2 gain (JAZF1-SUZ12, a PRC2 core subunit at the C-terminus) against PRC2 loss (MBTD1::EZHIP, a PRC2 antagonist), and that comparison has never been made.",
"any_result_with_a_failed_internal_control": "uninterpretable; a global negative stain is indistinguishable from a failed run and must not be reported as loss"
}- The experiment's most likely outcome is the one that stops it: fusion-negative low-grade ESS turning out to be H3K27me3-low as well, making the rare cases uninformative. That is a real result and should be reported as one rather than dressed as a partial success, and the design puts it first so that the cost of learning it is minimal.
- A retained H3K27me3 stain will be tempting to read as refuting the fusion's mechanism, and it does not - the published signature is a redistribution that a nuclear stain cannot see. The scoring convention has to fix that interpretation before the slides are read, because afterwards the ambiguity always resolves in the direction the reader expected.
- The EZHIP arm can produce a clean, confident false negative with no signature at all, if the antibody's epitope sits in the half of the protein this fusion does not retain. That is the same failure mode as this project's antibody-clone rule, and it is the reason the design gates that arm on a datasheet.
- Even a perfect positive result licenses only 'the ectopic effect also occurs in tissue', in three tumours, in a disease with no outcome data and no therapeutic option pointing the right way. It is a validation, not a discovery, and its value to this laboratory is entirely comparative.
- Which EZHIP antibody clone has a C-terminal epitope? This gates half the experiment, it is answerable from vendor datasheets in an afternoon, and under rule 74 the clone rather than the label is the identifier. No clone was retrievable at abstract level in this beat.
- Do the 2014 blocks still exist, and is there governance to stain them? The experiment's step 2 may fail for administrative rather than scientific reasons, and that is the second reason step 1 comes first.
- Was an H3K27me3 stain already done on any of these cases? The absence is classed measured-not-reported precisely because this is plausible. One request to the reporting institution could close the question without any new laboratory work, and it should be made before the stain is planned.
- Can CUT and RUN be run on decades-old formalin-fixed material? If a nuclear stain cannot resolve the redistribution the published mechanism predicts, this is the only fallback, and it was not costed here. If the answer is no, then the honest conclusion is that the cheapest experiment cannot answer the question and step 1's control series is the only defensible output.
- What is the H3K27me3 status of JAZF1-SUZ12 low-grade ESS? This is the single most informative unasked question in the whole design - a fusion that adds a PRC2 core subunit, in the same histology as one that poisons PRC2 - and it is obtainable entirely within step 1, from material any archive holds.
- E1strong 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 H3K27me2 and H3K27me3 in chromatin extracts of cells expressing them; the fusion protein contains the majority of MBTD1, lacking only its final 39 residues, fused to the C-terminal half of CATACOMB; and the low-grade endometrial stromal sarcoma fusion family, including JAZF1-SUZ12, is characterised as an N-terminal NuA4 acetyltransferase-complex component fused to a C-terminal Polycomb-group subunit.
CATACOMB: An endogenous inducible gene that antagonizes H3K27 methylation activity of Polycomb repressive complex 2 via an H3K27M-like mechanism. · Piunti A, Smith ER, Morgan MAJ, Ugarenko M, et al. · Sci Adv 2019What this evidence cannot show
Full text read this session via PMC6609211. ECTOPIC HA-FLAG EXPRESSION IN HEK293T AND HCT-116 and the authors' own figure legend calls it the 'putative' fusion protein; no endometrial stromal sarcoma cell or tissue appears in the paper, which is precisely the gap this experiment addresses. The readout is a bulk-chromatin Western blot. THE RETAINED-HALF DETAIL IS THE ONE THAT DECIDES THE EZHIP ANTIBODY ARM: only the C-terminal half of EZHIP is present in the fusion.
Evidence that argues against this (1)
- Full-length CATACOMB produces the same H3K27me2/3 decrease as the fusion in this paper, so a positive stain in a fusion-positive tumour would not establish that the fusion protein rather than inappropriate EZHIP expression is responsible. Distinguishing them needs the EZHIP stain to work on the chimera, which is the arm with the epitope problem, or a cage mutant, which is a different experiment. Hubner JM et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Bounded literature absence, 2026-09-23: MBTD1[tiab] AND H3K27me3[tiab] returns exactly 1 record, PMID 42238311, which is a 2026 review and not a measurement, so no primary report has described H3K27me3 in MBTD1::EZHIP tumour tissue; MBTD1[tiab] AND EZHIP[tiab] returns 0 and MBTD1[tiab] AND CXorf67[tiab] returns 4, so the whole fusion literature is four records.
What this evidence cannot show
One [tiab] tag per term, never the parenthesised group form; querytranslations read back; pubmed_safe.py raises rather than returning a zero on an empty translation. The name list searched was MBTD1, CXorf67, EZHIP, CATACOMB on the gene side and H3K27me3 only on the mark side - NOT 'H3K27 trimethylation', 'trimethylated H3K27', 'H3 lysine 27' or 'Polycomb' - so the counts are a floor. TITLES AND ABSTRACTS ONLY: a diagnostic stain reported in a table or a supplementary figure is invisible, which is exactly why this absence is classed measured-not-reported rather than never-measured.
Evidence that argues against this (1)
- H3K27me3 immunohistochemistry is routine enough in soft-tissue and gynaecological pathology that it could easily have been performed on one of these cases as part of a diagnostic workup and recorded only in a hospital report. The right next step is therefore not a better query but a request to the reporting institution, and this experiment's step 2 is that request as much as it is a stain. Dewaele B et al. 2014
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
H3K27me3 and EZHIP immunohistochemistry are validated, routine assays on formalin-fixed paraffin-embedded tissue: across 126 central nervous system germ cell tumour cases, loss of H3K27me3 was seen in 98.0% (100/102) of germinomas and germinoma components against 12.5% of non-germinomatous components (98.0% sensitivity, 87.5% specificity), and EZHIP was positive in 95.1% (97/102) against 0/24 (95.1% sensitivity, 100% specificity).
EZHIP and H3K27me3 are useful diagnostic markers for germinoma and for the germinoma component of germ cell tumors of central nervous system. · Han L, Hu Z, Zhao L, Teng L, Wang L · Neurooncol Pract 2026What this evidence cannot show
Abstract-level. Cited to establish that both stains are routine, scoreable and validated on archival material. The disease is a CNS germ cell tumour, unrelated to endometrial stromal sarcoma. NO ANTIBODY CLONE IS RECORDED because none was retrievable at abstract level, and under this project's rule 74 the clone is the identifier - so this supports 'the assay exists and performs' and NOT 'this reagent will work here'. Denominators differ between markers (102, 82, 97). Genuinely new to this graph.
Evidence that argues against this (1)
- Performance where the lesion is EZHIP OVEREXPRESSION does not transfer to a fusion retaining only EZHIP's C-terminal half. An N-terminal-epitope antibody would miss the fusion protein entirely while performing perfectly in every tumour it was validated on - an epitope-scoped false negative with no visible signature. That is why the EZHIP arm of this experiment is gated on establishing the epitope from the vendor datasheet, and why H3K27me3 is the arm that carries the design. Piunti A et al. 2019
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Three MBTD1-CXorf67-positive low-grade endometrial stromal sarcomas have been reported - two found by whole-transcriptome paired-end RNA sequencing with FISH and banding cytogenetics and validated by RT-PCR with Sanger sequencing, and a third recovered by a purpose-built FISH assay for formalin-fixed paraffin-embedded material from among 25 further uterine stromal tumours negative for JAZF1 and YWHAE rearrangements.
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 2014What this evidence cannot show
ABSTRACT LEVEL ONLY and unreadable through all five routes: Europe PMC core returns pmcid null, inEPMC N and 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. So the institutional provenance of the blocks, the FISH probe identities and the per-case detail are KNOWN-BUT-UNREADABLE. THE ONE FACT THIS EXPERIMENT DEPENDS ON IS READABLE AND IS IN THE ABSTRACT: a FISH assay was developed for formalin-fixed paraffin-embedded material, so formalin-fixed blocks existed for at least one case. Whether they still exist twelve years later is unknown.
Evidence that argues against this (1)
- A FISH assay having been run on formalin-fixed material in 2013 or 2014 does not establish that blocks remain, that they are retrievable, or that consent and governance permit their use for a new stain. Tissue archives are discarded on institutional schedules. The experiment's step 2 may be unrunnable for administrative rather than scientific reasons, which is the second reason step 1 - the control series, obtainable from any archive - has to come first. Hoang L et al. 2018
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Expression of H3K27M or of EZHIP produces similar chromatin profiles, namely loss of broad H3K27me3 domains with retention of H3K27me3 at CpG islands, with H3K27me3-mediated allosteric activation of PRC2 substantially increasing the inhibitory potential of both.
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 2019What this evidence cannot show
Abstract-level, and cited for the single most design-relevant fact in this entry: THE EXPECTED SIGNATURE IS A REDISTRIBUTION, NOT A LOSS. That is why a retained or partial stain in this experiment must be scored as uninformative for mechanism rather than as a refutation, and it is the limitation the whole design is built around. In vitro reconstitution plus cellular expression; no fusion and no sarcoma. Genuinely new to this graph.
Evidence that argues against this (1)
- If the signature is a redistribution, the proposed immunohistochemical experiment may be incapable of detecting the very thing it is designed to detect, and the fallback - ChIP or CUT and RUN on archival formalin-fixed material - is technically hard and was not costed here. In that case the honest conclusion is that the cheapest available experiment cannot answer the question, and the control series in step 1 becomes the only defensible output. Han L et al. 2026
tests the sarcoma arm, and runs the fusion-negative control series FIRST because that is what decides whether the rare cases mean anything
H3K27me3 immunohistochemistry, scored by the PFA ependymoma convention, with the redistribution caveat fixed in advance
the EZHIP arm, gated on establishing that the antibody epitope lies in the C-terminal half the fusion retains
the control arm's most informative comparator: JAZF1-SUZ12 puts a PRC2 CORE subunit where MBTD1::EZHIP puts a PRC2 poison, in the same histology