ZMYND11::MBTD1 is the KMT2A-fusion architecture assembled from different parts - and the part it throws away is a direct physical brake on KMT2A itself

also called the architecture from other parts · ZMYND11-MBTD1 versus KMT2A fusions · reader-plus-effector without menin

A rare leukaemia protein is built on the same plan as the fusion this project studies - a piece that finds a place on packaged DNA welded to a piece that summons a gene-activating machine - but every single component is a different molecule, and it switches on the same handful of genes. The twist nobody has written down is that the half it keeps comes from a protein whose normal job is to hold back the very enzyme this project is about, and the translocation cuts off exactly the bit that does the holding back.

The same thing, for a scientist

**The mapping.** A KMT2A fusion is an N-terminal chromatin-addressing module (AT-hooks, CXXC reading unmethylated CpG, held there by menin and LEDGF) welded to a C-terminal effector-recruiting module (AFF1/MLLT3/MLLT1 pulling in the super-elongation complex and DOT1L), and the output is a HOXA/MEIS1/MYB/MYC stem programme with a differentiation block. **ZMYND11::MBTD1 has the same two-module plan and shares none of the molecules.** The address is the ZMYND11 PHD-bromo-PWWP block reading H3K36me3 specifically on the H3.3 variant [E5]; the effector is NuA4/TIP60 docked through MBTD1's MBT repeats onto EPC1 [E6][E7]; and the output is *Hoxa*, *Meis1*, *Myb*, *Myc* and *Sox4* with a differentiation block released by BET inhibition [E1]. Both modules are genetically required and the disease is AML, not infant B-ALL [E1][E3]. **What this entry adds that neither literature contains.** Q119's `gene-zmynd11` already records, at abstract level, a 2025 preprint reporting that ZMYND11 physically restrains KMT2A. Reading that preprint in full localises the restraint: **the ZMYND11-KMT2A contact is a MYND-domain function.** Deleting the MYND domain collapses most of ZMYND11's interactome, and the commonest ZMYND11-related-syndromic-intellectual-disability missense, **R600W**, selectively strips the top interactors including KMT2A and NSD1; *Zmynd11*-knockout cortex gains H3K4me3 precisely at highly-ZMYND11-bound transcription start sites; a rTetR-MLL-C luciferase reporter is inhibited by full-length ZMYND11; and the transcriptional consequences of degrading ZMYND11 are **partially reversed by revumenib** [E2]. **UniProt places the MYND-type zinc finger at 563-598 of a 602-residue protein, and the fusion retains only ZMYND11 aa 1-409** [E4][E1]. **So the translocation keeps the reading head and amputates the KMT2A brake**, then bolts an acetyltransferase on in its place. The sign of the module at a ZMYND11 address is inverted: wild-type ZMYND11 restrains KMT2A-dependent activation there, and the fusion delivers NuA4/TIP60 instead. **The join is mine.** The preprint never mentions MBTD1 or any fusion - zero occurrences in its full text - and `fusion-zmynd11-mbtd1` does not know that ZMYND11 binds KMT2A. Both halves are published; the inference that the fusion boundary deletes the brake is Q123's, and `hyp-zmynd11-mbtd1-aml-is-revumenib-sensitive-without-a-kmt2a-rearrangement` states it as something that can be killed. **The disanalogies are the valuable half and they are listed in a table below.** No menin, no CXXC, no MBM, no DOT1L, no H3K79me2, no SEC; the reader reads a *deposited* mark rather than the *absence* of one; and the therapeutic logic inverts, because the menin programme targets an interface only the diseased cell needs while BET inhibition targets a reader every cell needs. One disanalogy cuts against this project rather than for it: an artificial ZMYND11-TIP60 chimera transforms equivalently [E1], so in **this** family the 3' partner is interchangeable - and across this graph's twenty-one KMT2A partner entries it is not.

Go deeper

The mapping, stated so each row can be scored

function	KMT2A fusions	ZMYND11::MBTD1	maps?
chromatin address	CXXC reading unmethylated CpG, plus AT-hooks	PWWP reading H3.3K36me3, plus PHD and bromodomain	same role, opposite logic - one reads an absence of methylation, one reads a mark transcription deposited
what holds the address on chromatin	menin (MBM1/MBM2) and LEDGF/PSIP1	nothing analogous; the cage binds the nucleosome directly	NO - and this is where the project's whole druggability argument lives
effector recruited	SEC (AFF1/AFF4, ENL/AF9, P-TEFb) and DOT1L	NuA4/TIP60, catalytic subunit KAT5	same role, different chemistry - elongation + methylation versus acetylation
the junction-side contact that is genetically required	AHD of the 3' partner	MBTD1 MBT repeats gripping EPC1 - dMBTx4, and A661D/L684D/L688D in fusion numbering, each abolish transformation 1	structurally analogous, molecularly unrelated
catalytic mark written	H3K79me2 (DOT1L), H3K4me3 retained in cis on wild-type MLL1	histone H4 and H2A acetylation	NO
output programme	HOXA9, MEIS1, MYB, MYC	Hoxa, Meis1, Myb, Myc, Sox4 1	YES - this is the strongest row
cell state	self-renewal with differentiation block	self-renewal with differentiation block, released by I-BET151 into terminal differentiation 1	YES
repressive chromatin at targets	bivalent/H3K27me3-contested at a subset	peaks devoid of H3K27me3; authors propose the state is refractory to PRC2 1	partly - see `hyp-both-mbtd1-fusions-converge-on-an-h3k27me3-poor-state`
5' partner's normal job relative to KMT2A	KMT2A is the enzyme	ZMYND11 inhibits the enzyme, through its MYND domain 2	INVERTED, and the fusion deletes the inhibiting module 24
disease and lineage	infant B-lineage ALL, and AML	adult and paediatric AML with minimal differentiation, CD7+/CD56-bright/MPO-	NO (rule 73)
is the 3' partner interchangeable?	no - partner identity tracks lineage, latency, outcome	yes - an artificial ZMYND11-TIP60 chimera transforms equivalently 1	NO, and it cuts against a general modularity story
druggable interface	menin-KMT2A, a contact the normal cell also has but the fusion depends on	none identified; the lead is BET inhibition downstream	NO

How to read this table. Rows 1, 3, 4, 6 and 7 are the analogy: two modules, an address and an effector, producing one programme and one cell state. Rows 2, 5, 10, 11 and 12 are the disanalogy, and row 2 is the load-bearing one for this laboratory - the reason the menin interface is attractive is that the fusion cannot let go of it, and this fusion has no equivalent hostage. Rows 9 and 11 are the two that were not in the dispatch brief and are the reason this entry exists.

The brake the breakpoint deletes, in residues

Live UniProt Q15326 (ZMY11_HUMAN, 602 aa) carries, in order: a SAMD1-like winged-helix domain at 6-82, a PHD-type zinc finger at 100-148, a bromodomain at 149-255, a PWWP domain at 280-331 and a MYND-type zinc finger at 563-598 4. The functional ZMYND11::MBTD1 construct in the founding study is ZMYND11 aa 1-409 joined through a 16-residue UTR-derived segment to the whole MBTD1 open reading frame 1.

409 < 563. The fusion therefore retains WH, PHD, bromodomain and PWWP and loses the MYND-type zinc finger outright, together with residue R600 - the site of the commonest disease-causing ZMYND11 missense and the residue whose substitution selectively abolishes the KMT2A interaction 24.

The arithmetic is trivial and the consequence is not. The preprint's model is a negative-feedback loop: SETD2 writes H3K36me3, H3K36me3 recruits ZMYND11, ZMYND11 inhibits KMT2A, transcription falls, H3K36me3 falls, ZMYND11 leaves 2. The fusion breaks that loop in the worst possible place. It keeps the recruiting arm - the PWWP still reads H3.3K36me3, and is required for transformation - and replaces the inhibiting arm with an acetyltransferase. What used to be a brake applied at transcribed chromatin becomes an accelerator applied at the same coordinates.

Three things this does NOT establish, and they are the reason this section is an argument and not a result. The preprint is not peer reviewed. It is mouse cortical neurons and HEK293T, not haematopoietic cells, so the ZMYND11-KMT2A contact has never been shown in a blood cell. And nobody has tested whether the fusion protein itself can still bind KMT2A - the inference is from domain boundaries, not from a co-immunoprecipitation on the chimera, and a 409-residue N-terminal fragment could in principle retain a contact the MYND domain merely strengthens.

Where this sits against the rest of the graph, so nothing is rediscovered

fusion-zmynd11-mbtd1 holds the fusion's own biology, its twelve-case clinical series, the module-requirement table and the two mechanism papers' disagreement about promoter-proximal versus gene-body mistargeting. gene-zmynd11 holds ZMYND11's identity, its H3.3K36me3 reader chemistry, the preprint at abstract level, and the measurement that put ZMYND11 in this graph in the first place - the chr10:180,000 bin at 183.0x the MLL1-N genome median in RS4;11, 405 bp from the ZMYND11 transcription start. gene-mbtd1 holds MBTD1's NuA4 membership, its HSC-quiescence function and its second fusion. This entry adds only the comparison and the MYND inference; it does not restate any of that.

On the KMT2A side, the natural comparators already argued in this graph are the two extremes of address availability: fusion-kmt2a-aff1, where the CXXC and both menin-binding motifs are retained in cis, and fusion-nup98-kmt2a, where the polarity is reversed and exon 1 - and with it the menin-binding motif - is gone. concept-kmt2a-fusion-polarity-is-inverted-in-sarcoma states the rule those two instantiate. ZMYND11::MBTD1 is the third case: an address module with no menin dependence available at all, because the gene has no MBM to lose. That makes it the natural negative control for any claim of the form this programme requires the menin interface.

And the honest limit on the whole comparison. Not one experiment has ever been performed jointly on ZMYND11::MBTD1 and a KMT2A fusion. MBTD1[tiab] AND KMT2A[tiab] returns 0 records and ZMYND11[tiab] AND MLL[tiab] returns 0 8. Everything above is a comparison of two separate literatures made here for the first time, which is what an analogy node is for and also exactly why it is filed at moderate.

The cheapest way to find out whether this analogy is about coordinates or only about gene names

"The same programme" is doing a lot of work in this entry and in fusion-zmynd11-mbtd1. It currently means the same gene list - Hoxa, Meis1, Myb, Myc on one side and HOXA9, MEIS1, MYB, MYC on the other. It does not yet mean the same sub-elements.

Two published genomic datasets exist for the fusion 19 and this project holds binned KMT2A::AFF1 occupancy from GSE127507. Overlapping them at the HOXA cluster is pure compute and would separate three outcomes that the current evidence cannot: the two fusions occupy the same sub-elements (the address chemistry is interchangeable at this locus); the same genes by different elements (the programme is a property of the locus, not the reader); or different genes that happen to share four famous names (the analogy is largely nominal). The third outcome is the one nobody expects and the one that would matter most.

This is deliberately not written as an evidence item, because no accession was looked up in this beat. Naming a GEO series from memory is the failure this project has a rule about. ana-mbtd1-fusion-comparison-is-uncomputed is not written either; the task is recorded here and in the open questions instead.

Facts and figures

domain	engineering
domain note	Not a cross-domain metaphor. This is a within-biology architectural comparison between two fusion families, filed under the analogy type because that is what this graph uses for the PI's fifth dimension, 'the analogous biology in other fusion families'. The nearest sibling in form is analogy-brg1-to-ring1b-swap-at-myc-enhancers-inv16.
shown for fusion	NOTHING in this comparison has been measured in any KMT2A fusion, and nothing in the KMT2A literature has been measured on ZMYND11::MBTD1. MBTD1[tiab] AND KMT2A[tiab] returns 0. The ZMYND11-KMT2A contact itself is shown only in mouse neurons and HEK293T, in a preprint, on wild-type ZMYND11 and never on the fusion protein.
epigenetic readout	Reads H3.3K36me3 (ZMYND11 PWWP, required). Writes H4 and H2A acetylation (recruited KAT5). Carries but does not require an H4K20me1/2 reader (MBTD1 MBT4). Target chromatin is H3K36me3-positive, acetylation-rich and H3K27me3-depleted. On the wild-type side, ZMYND11 loss raises H3K4me3 at its own bound promoters - the mark KMT2A writes.
cell type	Murine primary bone-marrow HSPCs for transformation; human AML blasts with minimal differentiation in patients; mouse cortical neurons and HEK293T for the ZMYND11-KMT2A contact.
cell state	Self-renewal with a differentiation block on both sides of the comparison, which is the second-strongest row after the shared output programme.
absence class	never-measured
absence class note	The comparison itself has never been instrumented. No experiment has been performed jointly on ZMYND11::MBTD1 and any KMT2A fusion, no co-immunoprecipitation has asked whether the fusion protein retains KMT2A binding, and no menin inhibitor has been given to a ZMYND11::MBTD1 model. All three need new material - constructs, cells, a mouse - so the class is never-measured and not measured-not-analysed. ONE SUB-ARM IS WEAKER THAN THAT AND IS DELIBERATELY NOT UPGRADED HERE: the two published ZMYND11::MBTD1 genomic datasets (PMID 33594072, PMID 35705031) could be intersected with this project's binned KMT2A::AFF1 occupancy as pure compute, which would be measured-not-analysed. No GEO accession is named in this entry because none was looked up in this beat, and naming one from memory is the exact failure this field exists to prevent.
what would make this a finding rather than an analogy	A co-immunoprecipitation showing whether ZMYND11 aa 1-409 and the ZMYND11::MBTD1 fusion protein retain KMT2A binding. One blot separates 'the fusion deleted the brake' from 'the fusion kept the brake and added an accelerator', and those two models predict opposite responses to a menin inhibitor.

mapping

[
  {
    "bio": "KMT2A N-terminus: AT-hooks plus CXXC reading unmethylated CpG, held on chromatin by menin and LEDGF",
    "analog": "ZMYND11 N-terminus: winged-helix, PHD, bromodomain and PWWP reading H3K36me3 on the H3.3 variant, binding the nucleosome directly with no adaptor"
  },
  {
    "bio": "3' partner AHD recruiting SEC and DOT1L",
    "analog": "MBTD1 MBT repeats gripping EPC1 and recruiting NuA4/TIP60"
  },
  {
    "bio": "H3K79me2 deposited across fusion target bodies",
    "analog": "histone H4 and H2A acetylation deposited at fusion targets"
  },
  {
    "bio": "HOXA9, MEIS1, MYB, MYC maintained; differentiation block",
    "analog": "Hoxa, Meis1, Myb, Myc, Sox4 maintained; differentiation block released by BET inhibition - THE STRONGEST ROW"
  },
  {
    "bio": "the menin-KMT2A interface is the drug target because the fusion cannot let go of it",
    "analog": "NO EQUIVALENT. No adaptor is hostage; the published lead is BET inhibition acting downstream on a reader every cell needs"
  },
  {
    "bio": "KMT2A is the methyltransferase the fusion misdirects",
    "analog": "ZMYND11 is a physical INHIBITOR of that same methyltransferase, acting through its MYND domain - and the fusion boundary at aa 409 deletes the MYND domain (563-598). INVERTED SIGN AT THE SAME ADDRESS."
  },
  {
    "bio": "3' partner identity tracks lineage, latency and outcome across the recombinome",
    "analog": "3' partner identity is DISPENSABLE - an artificial ZMYND11-TIP60 chimera transforms equivalently. This row cuts against a general modularity story."
  },
  {
    "bio": "infant B-lineage ALL, and AML",
    "analog": "adult and paediatric AML with minimal differentiation, CD7+/CD56-bright/HLA-DR-/MPO-. NOT the index disease."
  }
]

generated hypotheses

[
  "hyp-zmynd11-mbtd1-aml-is-revumenib-sensitive-without-a-kmt2a-rearrangement",
  "hyp-both-mbtd1-fusions-converge-on-an-h3k27me3-poor-state"
]

the modules

{
  "address_module_kmt2a_fusion": "KMT2A aa 1-~1400: AT-hooks, CXXC, MBM1/MBM2, IBM1/IBM2",
  "address_module_zmynd11_fusion": "ZMYND11 aa 1-409: WH 6-82, PHD 100-148, Bromo 149-255, PWWP 280-331",
  "effector_module_kmt2a_fusion": "AFF1 / MLLT3 / MLLT1 / MLLT10 / ELL - SEC and DOT1L",
  "effector_module_zmynd11_fusion": "MBTD1 aa 1-628: FCS ZnF 45-80, MBT 141-245, 253-350, 351-456, 464-560 - NuA4/TIP60 via EPC1",
  "module_lost_that_matters_most": "ZMYND11 MYND-type zinc finger 563-598, the KMT2A-interaction module, plus residue R600"
}

How this could bite us back

THE SHARPEST WAY THIS ENTRY COULD DAMAGE THE MENIN PROGRAMME is the same mechanism that makes it interesting. If revumenib works in ZMYND11::MBTD1 AML - a leukaemia with no KMT2A rearrangement - then the drug's effect is not a readout of fusion disruption, and the corpus's cycle-82 K562 result already points the same way (a menin inhibitor moved H3K79me2 in a fusion-free line about as much as in fusion-bearing ones). Two independent lines converging on 'menin inhibitors act on wild-type KMT2A too' would widen the drug's indication and simultaneously dissolve the argument that its selectivity comes from the fusion.
The published lead here is BET inhibition, acting DOWNSTREAM on BRD4 reading the acetylation the fusion causes. There is no interface that exists only in the diseased cell, so the selectivity argument that makes the menin-KMT2A contact attractive has no counterpart. A laboratory that reads this analogy as encouragement should notice it is an encouragement to a drug class that has repeatedly under-delivered.
If a partner-independent 'address plus effector' architecture is really this permissive - and the artificial ZMYND11-TIP60 chimera says it is, in this family - then resistance by effector substitution becomes thinkable. A cell under menin inhibition does not need to restore the menin contact if any module that delivers a coactivator to the same address will do. That is a compensation route this graph has not costed.
KAT5/TIP60 is near-pan-essential and EP400 worse, so 'inhibit the hijacked acetyltransferase' is an essential-function toxicity proposal. EPC1, the subunit the fusion actually grips, is milder - mean DepMap gene effect -0.336 across 1,258 screens, 22.7% of lines below -0.5, Open Targets dataVersion 26.06 - which makes the MBTD1-EPC1 contact the only node in this pathway that looks like a target rather than a liability. See gene-epc1 for why that is still not encouraging.
The whole comparison rests on twelve or thirteen reported cases and two mechanism papers that disagree with each other. Treating it as a template for reasoning about a disease with thousands of patients inverts the usual direction of inference, and rule 73 applies to every row of the mapping table: the systems are murine, neuronal or heterologous, and not one of them is infant B-lineage KMT2A::AFF1.

Still open

Does the ZMYND11::MBTD1 fusion protein still bind KMT2A? The fusion retains ZMYND11 aa 1-409 and the MYND domain is 563-598, so the prediction is no - but a feature-table interval is not a folding unit, and one co-immunoprecipitation on the chimera and on the 1-409 fragment settles a claim this entry currently rests on arithmetic.
Do ZMYND11::MBTD1 and KMT2A::AFF1 occupy the same HOXA sub-elements, the same genes by different elements, or different genes that share four famous names? Both fusion datasets are public and this project holds binned KMT2A::AFF1 occupancy; the intersection is compute. The third outcome is the one nobody expects and the one that would matter most, because 'the same programme' currently means only 'the same gene list'.
Is ZMYND11::MBTD1 AML sensitive to a menin inhibitor? fusion-zmynd11-mbtd1 predicts independence on the reasonable ground that there is no CXXC and no MBM. The preprint inverts that expectation, because revumenib partially reverses the consequences of losing ZMYND11 and this fusion deletes ZMYND11's KMT2A brake. hyp-zmynd11-mbtd1-aml-is-revumenib-sensitive-without-a-kmt2a-rearrangement states it; the reagents are commercial and there is no cell line.
Is ZMYND11::MBTD1 partly a LOSS-of-ZMYND11 lesion? ZMYND11 is a tumour suppressor and a KMT2A inhibitor; a balanced translocation disrupts one allele. Nobody has asked whether the untranslocated allele is expressed, whether the fusion is dominant-negative over wild-type ZMYND11 at shared addresses, or whether ZMYND11 haploinsufficiency contributes to the phenotype. Every existing experiment is an overexpression gain-of-function assay, which cannot see this.
Does the H3K4me3 gain seen at ZMYND11-bound promoters in Zmynd11-knockout cortex also occur at ZMYND11 addresses in a fusion-bearing haematopoietic cell? If it does, the fusion is simultaneously adding acetylation and de-restraining KMT2A's own mark at the same coordinates, which is a two-hit chromatin lesion nobody has described.
Is the winged-helix domain at ZMYND11 6-82 doing anything in the fusion? It is retained, it is absent from fusion-zmynd11-mbtd1's domain list, and no mutant in the founding mutagenesis touches it. It is the only retained module of the fusion with no functional test at all.

Every source behind this entry (9)

E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ZMYND11-MBTD1 transforms primary murine haematopoietic stem/progenitor cells and causes AML; it binds 16,114 sites overlapping H3K36me3 and devoid of H3K27me3, recruits NuA4/Tip60, and maintains high expression of Hoxa, Meis1, Myb, Myc and Sox4; both an intact H3.3K36me3-binding PWWP and the Tip60-interaction surface are required, deletion of the four MBT repeats or the triple mutation A661D/L684D/L688D in fusion numbering each abolishing transformation; an artificial ZMYND11-TIP60 chimera transforms equivalently; the fusion construct is ZMYND11 aa 1-409 plus a 16-residue UTR-derived segment plus the entire MBTD1 open reading frame; and the resulting AML is sensitive to the BET inhibitor I-BET151 in vitro and in vivo.
ZMYND11-MBTD1 induces leukemogenesis through hijacking NuA4/TIP60 acetyltransferase complex and a PWWP-mediated chromatin association mechanism. · Li J, Galbo PM, Gong W, Storey AJ, et al. · Nat Commun 2021
PMID 33594072 DOI
What this evidence cannot show
MURINE throughout the functional arm: primary mouse HSPCs, retroviral overexpression of a human cDNA, syngeneic transplant, no human ZMYND11::MBTD1 cell line in existence. The occupancy data are on a tagged construct, not an endogenous allele. Transformation by ZM alone had long latency (mean 154 days) and incomplete penetrance (8/9). The citation identifiers here were taken from a PubMed efetch record fetched 2026-09-23, not copied from the sibling entries that also cite this paper; note PubMed renders the second author as 'Galbo PM' where the sibling entries carry 'Galbo PM Jr'.
Evidence that argues against this (1)
- The equivalent transformation by an artificial ZMYND11-TIP60 chimera is reported by the same authors as evidence for modularity, and it argues AGAINST the analogy being informative in the direction this entry wants. If the 3' partner's identity is dispensable provided an acetyltransferase arrives, then 'a reader fused to a complex recruiter' may be a very weak architectural constraint - satisfiable by many molecules - and the resemblance to KMT2A fusions would carry little mechanistic content. The KMT2A family behaves differently: Meyer et al.'s recombinome shows partner identity tracking lineage and outcome across hundreds of cases, which is not what an interchangeable-effector model predicts. Meyer C et al. 2023
E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.preprint
In a 2025 preprint, ZMYND11 co-immunoprecipitates KMT2A and NSD1 from mouse brain and from HEK293T; deletion of the MYND domain drastically reduces the majority of ZMYND11 interactions while the ZRSID-associated R600W substitution selectively removes the top interactors including KMT2A and NSD1; Zmynd11 conditional-knockout cortex shows a significant increase in H3K4me3 at highly-ZMYND11-bound transcription start sites but not at lowly-bound ones and no change in H3K4me1, with strongly-bound genes up-regulated; full-length ZMYND11 inhibits activation of a 9xTetO-minCMV-luciferase reporter driven by rTetR-MLL-C while the MYND domain alone represses without activating; and 24-hour dTAG degradation of endogenous ZMYND11 in primary cortical neurons produces transcriptional defects that revumenib significantly attenuates, with revumenib's own small effects anti-correlated with those of ZMYND11 loss and the rescue absent for dTAG off-target genes in wild-type neurons.
ZMYND11 Restrains KMT2A to Enable a Neuronal Developmental Program. · Greben AW, Wu XS, Robb JE, Vogel IR, et al. · bioRxiv 2025
PMID 41279818 DOI
What this evidence cannot show
A PREPRINT, not peer reviewed, and filed at moderate for that reason alone; the experiments themselves are unusually complete. Full text read this session via PMC12636335. THE SYSTEM IS NEURONAL: mouse cortical neurons, mouse cortex and HEK293T. No haematopoietic cell appears anywhere in it, so the ZMYND11-KMT2A contact has never been demonstrated in a blood cell. The revumenib rescue is explicitly PARTIAL and at one timepoint. The interactome work is FLAG-IP of overexpressed ZMYND11 in 293T for the mutant comparison, and the authors themselves note that overexpression recovers lower-affinity interactions. The paper says nothing about MBTD1 or about any fusion - a grep of its full text returns zero occurrences of MBTD1 - so every statement in this entry connecting it to ZMYND11::MBTD1 is Q123's inference and is labelled as such. Q119's gene-zmynd11 E5 already cites this record at abstract level; what is new here is the full text and the MYND localisation.
Evidence that argues against this (1)
- The same authors' model makes the opposite prediction awkward for a leukaemia reading: they position ZMYND11 as a tumour SUPPRESSOR acting through KMT2A inhibition, and note that SETD2 - the writer of the mark ZMYND11 reads - is mutated more often in KMT2A-rearranged than in non-KMT2A-rearranged leukaemia. On that model, losing ZMYND11 function should phenocopy part of what a KMT2A rearrangement does, which makes ZMYND11::MBTD1 potentially a LOSS-of-ZMYND11 lesion as much as a gain-of-NuA4 one. The fusion literature treats it purely as a gain of function and has never tested haploinsufficiency of the untranslocated allele, so the two readings are unresolved and this entry's framing chooses one. Wen H et al. 2014
E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Across twelve assembled cases, AML with t(10;17)(p15;q21)/ZMYND11::MBTD1 shows minimal differentiation with a CD7-positive, CD33-positive, CD56-bright, CD117-positive, HLA-DR-negative and MPO-negative immunophenotype, simple karyotypes, rare co-occurring mutations, and generally poor outcomes in adults with long-term remission in two paediatric patients.
Acute myeloid leukemia with t(10;17)(p15;q21)/ ZMYND11::MBTD1: a subtype with minimal differentiation, CD7/CD56 expression, and generally poor outcomes in adults. · Wei Q, Pemmaraju N, Wang SA, Hu S, et al. · Haematologica 2026
PMID 41262054 DOI
What this evidence cannot show
Retrospective case series, five new single-institution cases plus seven assembled from the literature, no control group and no survival statistic with an interval. Cited here only to fix the lineage and age of the comparator disease, which is the load-bearing disanalogy under rule 73; the clinical detail belongs to fusion-zmynd11-mbtd1 and is not restated. Identifiers fetched from PubMed 2026-09-23.
Evidence that argues against this (1)
- A 2026 case report reaches the same immunophenotypic description and argues the entity is myeloid/NK cell precursor leukaemia rather than AML with minimal differentiation, concluding that earlier cases were misclassified. The two 2026 papers agree on the markers and disagree on the label, so the lineage disanalogy in this entry is secure at the level of 'not B-lineage' and unsettled below that. Liang H et al. 2026
E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProt Q15326 (ZMY11_HUMAN) is 602 amino acids and its feature table places a SAMD1-like winged-helix domain at 6-82, a PHD-type zinc finger at 100-148, a bromodomain at 149-255, a PWWP domain at 280-331 and a MYND-type zinc finger at 563-598, so a fusion retaining only ZMYND11 residues 1-409 keeps the winged-helix, PHD, bromodomain and PWWP modules and loses the MYND-type zinc finger and residue 600 entirely.
UniProtKB Q15326 (ZMY11_HUMAN), Zinc finger MYND domain-containing protein 11; feature table fields ft_domain and ft_zn_fing · UniProt 2026
What this evidence cannot show
Fetched live 2026-09-23 with an explicit field list; the record is the single reviewed human entry with ZMYND11 as primary gene name. The domain boundaries are curation, revised between releases, and the 563-598 MYND boundary does not itself contain residue 600 - the preprint describes R600 as sitting at the C-terminal END of the MYND domain, which is a statement about the structural element rather than about the feature-table interval, and the two need not coincide. The arithmetic 409 < 563 is mine. Note that fusion-zmynd11-mbtd1 E11 omits the winged-helix domain at 6-82, which is inside the retained segment; gene-zmynd11 E1 lists it. Reported, not edited.
Evidence that argues against this (1)
- A feature-table interval is not a folding unit, and the whole inference in this entry rests on treating one as the other. If the KMT2A-binding surface extends N-terminal of residue 409 - or if, as the preprint's own dMYND result allows, the MYND domain stabilises a contact made elsewhere rather than making it - then the fusion could retain KMT2A binding and this entry's central claim collapses. The experiment that settles it is a co-immunoprecipitation of KMT2A with the fusion protein and with ZMYND11 aa 1-409, and it has never been done. Greben AW et al. 2025
E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
ZMYND11 recognises H3K36me3 only on the histone variant H3.3 and not on H3.1 or H3.2, caging the trimethyl-lysine in an aromatic pocket of the PWWP domain with H3.3 specificity conferred by encapsulation of the H3.3-specific Ser31 residue in a composite pocket formed by the tandem bromodomain-PWWP module, and it acts as an unconventional transcription co-repressor by modulating RNA polymerase II at the elongation stage while functioning as a tumour suppressor.
ZMYND11 links histone H3.3K36me3 to transcription elongation and tumour suppression. · Wen H, Li Y, Xi Y, Jiang S, et al. · Nature 2014
PMID 24590075 DOI
What this evidence cannot show
Abstract-level in this beat; gene-zmynd11 E3 carries the fuller treatment and this item exists only to supply the address chemistry for the mapping table. The tumour-suppressor and prognostic arms are breast-cancer systems, not haematopoietic. Identifiers fetched from PubMed 2026-09-23.
Evidence that argues against this (1)
- The variant specificity makes the analogy to KMT2A's CXXC weaker than the table's 'same role' suggests, and in a direction that matters for this project: an H3.3-restricted reader can only address chromatin where H3.3 has been deposited, which is largely actively transcribed and regulatory chromatin, whereas CXXC reads unmethylated CpG at promoters irrespective of variant. The two modules therefore sample partly non-overlapping genomes by construction, so shared output genes are a stronger observation than shared addressing would be. Devoucoux M et al. 2022
E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
MBTD1 is a stable subunit of the NuA4/TIP60 acetyltransferase complex and, through a histone reader domain for H4K20me1/2, allows TIP60 to associate with specific gene promoters and to promote homologous-recombination repair of double-strand breaks, regulating 53BP1 association partly by competing for H4K20me2 and partly through TIP60-mediated acetylation of H2AK15.
The TIP60 Complex Regulates Bivalent Chromatin Recognition by 53BP1 through Direct H4K20me Binding and H2AK15 Acetylation. · Jacquet K, Fradet-Turcotte A, Avvakumov N, Lambert JP, et al. · Mol Cell 2016
PMID 27153538 DOI
What this evidence cannot show
Abstract-level in this beat; gene-mbtd1 E3 carries it in full. Cited here to establish that the effector module of the fusion is a NuA4/TIP60 component in its normal life too, which is what makes the 3'-partner comparison to AFF1/MLLT3 a comparison of roles rather than of accidents. Identifiers fetched from PubMed 2026-09-23.
Evidence that argues against this (1)
- The H4K20me1/2 reader function this paper establishes as MBTD1's route into promoter association is precisely the function the leukaemic fusion does NOT need: the cage mutants W954A and F951A/W954A/Y958A both bind Tip60 and both transform. So MBTD1's documented normal mechanism of target selection is dispensable in the fusion, and it is the SCAFFOLDING half rather than the reading half of MBTD1 that the oncoprotein uses. An entry that cited this paper as the fusion's targeting mechanism would be wrong. Li J et al. 2021
E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A crystal structure shows a hydrophobic C-terminal fragment of EPC1 engaging the MBT repeats of MBTD1 at a site distinct from the H4K20me-binding site, with EPC1 the non-catalytic NuA4/TIP60 subunit that enlists MBTD1 into the complex; in PDB 6NFX at 1.95 A the single 470-residue polymer entity aligns as MBTD1 Q05BQ5 residues 140-562 joined through a designed linker to EPC1 Q9H2F5 residues 644-671, and the three residues whose mutation abolishes fusion-driven transformation map by the published construct offset to MBTD1 A236, L259 and L263 within MBT repeats 1 and 2.
Structural Basis for EPC1-Mediated Recruitment of MBTD1 into the NuA4/TIP60 Acetyltransferase Complex. · Zhang H, Devoucoux M, Song X, Li L, et al. · Cell Rep 2020
PMID 32209463 DOI
What this evidence cannot show
The paper was read at abstract level; the structural spans were NOT read off the deposit's own numbering but derived by aligning 6NFX's fetched canonical entity sequence against UniProt Q05BQ5 and Q9H2F5 FASTA downloaded the same session, which is the procedure this project requires after three cases of deposit-numbering offsets. 6NFX IS A SINGLE-CHAIN ENGINEERED CHIMERA, not a two-chain complex, so it models the bound state rather than measuring association - the same caveat this graph carries for 2LM0. The A236/L259/L263 mapping applies Q121's stated offset (MBTD1 n = fusion n + 425) and the residue identities were checked against the live Q05BQ5 sequence and match; the offset itself is a derivation, not a published statement.
Evidence that argues against this (1)
- A single-chain construct is a legitimate way to solve a weak interface and is also a way to make an interface look obligate when it is not. Nothing in a chimeric deposit measures affinity, stoichiometry or whether the contact forms in a cell at endogenous concentrations, and this project has already had to correct an entry that described a single-chain deposit as a co-structure. The functional evidence that the contact matters comes from the mutagenesis in the leukaemia paper, not from the structure. Li J et al. 2021
E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Bounded literature absence, 2026-09-23: no indexed title or abstract joins this fusion to the KMT2A family. MBTD1[tiab] AND KMT2A[tiab] returns 0 records; ZMYND11[tiab] AND MBTD1[tiab] AND menin[tiab] returns 0; ZMYND11[tiab] AND H3K27me3[tiab] returns 0; and ZMYND11[tiab] AND KMT2A[tiab] returns 4, of which three are Cornelia-de-Lange or neurodevelopmental-disorder gene lists in which both symbols appear incidentally and the fourth is the 2025 preprint cited as E2.
NCBI E-utilities esearch db=pubmed via scripts/pubmed_safe.py, 2026-09-23. querytranslations read back from the responses: "MBTD1"[Title/Abstract] AND "KMT2A"[Title/Abstract]; "ZMYND11"[Title/Abstract] AND "MBTD1"[Title/Abstract] AND "menin"[Title/Abstract]; "ZMYND11"[Title/Abstract] AND "H3K27me3"[Title/Abstract]; "ZMYND11"[Title/Abstract] AND "KMT2A"[Title/Abstract]. Counts and id lists persisted to analyses/q123-the-architecture-from-other-parts-analogy/bounded-searches.json · NCBI PubMed 2026
What this evidence cannot show
Every query used one [tiab] tag per term, never the parenthesised group form, and every querytranslation was read back from the response; pubmed_safe.py raises rather than returning a zero on an empty translation, so these zeros are genuine zeros for the strings searched. TITLES AND ABSTRACTS ONLY - rule 80 applies with full force, and the clearest demonstration is inside this very item: ZMYND11[tiab] AND H3K27me3[tiab] returns 0 while the founding paper reports 16,114 fusion peaks devoid of H3K27me3, because its abstract never names the mark. The alias arm was run on ZMYND11 only (BS69, in gene-zmynd11 E9) and not on MLL-family aliases beyond MLL and KMT2A; a paper writing 'MLL1' or 'HRX' with MBTD1 would be missed.
Evidence that argues against this (1)
- A zero on MBTD1 plus KMT2A is entirely compatible with the comparison having been made and rejected in a discussion section, a review, or a grant, none of which an abstract-level query can see. It is also compatible with the comparison having been made under complex names rather than gene symbols - 'NuA4' plus 'SEC', or 'acetyltransferase hijacking' plus 'elongation hijacking' - which is precisely the complex-name unreachability this project has already been burned by twice, once concluding a FACT-subunit literature empty while citing the disproving paper in four other entries. Crump NT et al. 2023
E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In human cells the ZMYND11-MBTD1 fusion protein is stably incorporated into the endogenous NuA4/TIP60 complex and mislocalises it onto the body of genes normally bound by ZMYND11, increasing chromatin acetylation and altering transcription most notably of MYC as well as alternative splicing, favouring Myc-driven pluripotency during embryonic stem cell differentiation and self-renewal of haematopoietic stem/progenitor cells.
Oncogenic ZMYND11-MBTD1 fusion protein anchors the NuA4/TIP60 histone acetyltransferase complex to the coding region of active genes. · Devoucoux M, Fort V, Khelifi G, Xu J, et al. · Cell Rep 2022
PMID 35705031 DOI
What this evidence cannot show
ABSTRACT LEVEL ONLY. Europe PMC returns no pmcid for this record and the efetch ArticleIdList carries no pmc id, so routes 2-4 of the unreadable test cannot be attempted; the DOI route was not tried in this beat. Cited here for the existence of a second genomic dataset and for the disagreement with E1 about where the decisive mistargeting happens, which fusion-zmynd11-mbtd1 records in full. No numbers are quoted because the abstract gives none. Identifiers fetched from PubMed 2026-09-23.
Evidence that argues against this (1)
- This paper's emphasis - gene bodies and MYC - is the emphasis that makes the analogy to KMT2A fusions WORSE, not better, because the strongest row of the mapping table is the shared HOXA/MEIS1 stem programme and that row comes from E1 alone. If Devoucoux et al. are right that the decisive event is body-of-gene acetylation with MYC as the headline target, then ZMYND11::MBTD1 resembles a generic MYC-driving lesion more than it resembles a KMT2A fusion, and this entry's central row is a murine-specific result. Li J et al. 2021

Connected to

related to →ZMYND11::MBTD1, t(10;17)(p15;q21) - a chromatin reader welded to the NuA4/TIP60 acetyltransferase, maintaining Hoxa/Meis1/Myb/Myc/Sox4 in AML: the KMT2A-fusion architecture built from different parts
the analogy entry Q121's fusion entry points at and which did not exist until cycle 83
related to →ZMYND11 (BS69) - an H3.3K36me3 reader that restrains elongation, is itself the 5' partner of a chromatin-hijacking AML fusion driving HOXA/MEIS1/MYB, and whose promoter is the third-strongest MLL1-N bin in RS4;11
the address module; gene-zmynd11 already carries the KMT2A-restraint preprint at abstract level and this entry adds its full text and the MYND localisation
related to →MBTD1 - the H4K20me1/me2 reader inside NuA4/TIP60, the quiescence brake on adult haematopoietic stem cells, and the 3' partner of ZMYND11::MBTD1 whose methyl-lysine cage the fusion carries and does not use
the effector-recruiting module
related to →Revumenib (SNDX-5613, Revuforj) — first approved menin–KMT2A interface inhibitor
revumenib partially reverses the transcriptional consequences of degrading ZMYND11 in mouse cortical neurons, which is why this analogy generates a menin-inhibitor prediction in a fusion with no KMT2A rearrangement
supports →In leukaemia KMT2A gives the front of the protein; in sarcoma it gives the back. Every consequence follows from that.
the same rule in a gene with no KMT2A involvement: which modules survive is set by the junction, not by the partner list
related to →DOT1L — the target where excellent chemistry met an indirect interaction
the effector this architecture does NOT use - an acetyltransferase stands where the methyltransferase stands in a KMT2A fusion
related to →MEN1 / menin — the adaptor the fusion cannot work without
the absent adaptor - the disanalogy that matters most to this laboratory, since no protein is hostage in this fusion
related to →In inv(16), disrupting the fusion interface swaps BRG1 for RING1B at three MYC enhancers - the cleanest version anywhere of the experiment nobody has run on menin
the nearest sibling in form: another within-biology cross-fusion-family comparison in this graph, written to the same shape
depends on →H3K36me2/3 - the mark that physically switches PRC2 off, that LEDGF reads to reach chromatin, and that a comparator fusion uses to keep EZH2 off HOXA
the mark the address module reads, and the mark whose writer SETD2 the preprint places in the same axis
related to →KAT5 / Tip60 - the acetyltransferase in the only direct KMT2A-fusion histone-variant result, and both fusions tested were myeloid
the catalytic subunit delivered in place of DOT1L
← related toEPC1 - the NuA4/TIP60 subunit whose C-terminal fragment grips MBTD1, making it the contact a leukaemic fusion cannot do without and the mildest dependency in that whole complex
in this architecture the MBTD1-EPC1 grip is the analogue of the AHD surface a KMT2A fusion's 3' partner uses
← proposed byZMYND11::MBTD1 AML is menin-inhibitor sensitive despite having no KMT2A rearrangement, because the translocation deletes ZMYND11's MYND domain and with it a physical brake on wild-type KMT2A
generated by the architectural comparison, from the MYND-domain inference