KMT2A::GAS7 (MLL-GAS7), t(11;17)(q23;p13) — three patients, and the best mouse model in the partner class
also called MLL-GAS7 · MLL/GAS7 · MLL::GAS7 · KMT2A-GAS7
Only three patients with this fusion appear in the largest survey ever done, yet it is one of the most studied KMT2A fusions in mice — because it was the one used to show that a cytoplasmic partner causes leukaemia simply by making copies of KMT2A stick together. In mice it produces three different leukaemias, including one that is simultaneously myeloid and lymphoid.
The same thing, for a scientist
KMT2A::GAS7 arises from t(11;17)(q23;p13) and is the 28th-ranked direct KMT2A partner at 3 patients in the 3401-case recombinome (1 infant AML, 1 paediatric ALL, 1 unclassified) [E1]. Its index case was a child treated for neuroblastoma whose karyotypic del(11)(q23) was a **cryptic** t(11;17) resolved only by panhandle PCR; two different MLL-GAS7 fusion transcripts were expressed and the translocation was detectable 1.5 months into topoisomerase-II-inhibitor therapy, absent at neuroblastoma diagnosis [E2]. A de novo paediatric precursor-B ALL later carried an **identical** chimera [E3], so the same junction produces t-AML in one child and B-ALL in another. The fusion's importance is experimental, not epidemiological. It is one of the two fusions in which the partner's coiled-coil oligomerisation domain was shown **necessary and sufficient** for transformation, with synthetic dimerisation modules substituting [E4]. It impairs differentiation and expands murine progenitors with multipotent features that induce three pathologically distinct leukaemias in mice, including an acute biphenotypic leukaemia recapitulating the hallmark of human MLL disease [E5]. And it still transforms Hoxa7- or Hoxa9-null progenitors, but less efficiently and with longer latency, reduced penetrance and less mature phenotypes [E6] — making Hoxa7/Hoxa9 quantitative modifiers rather than switches for a dimerisation-driven MLL leukaemia. What does not exist: any menin-inhibitor test, any histone-modification measurement, any localisation image of the fusion protein, any structure, and any human cell line.
This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
The same fusion, two diseases
Three patients in the recombinome 1, and two published cases that between them make the interesting point.
The index case was a child receiving treatment for neuroblastoma who developed AML. Karyotype showed del(11)(q23); panhandle PCR with an unknown partner revealed the deletion was a cryptic t(11;17) with GAS7 at 17p13. Two different MLL-GAS7 fusion transcripts were expressed. Critically, the translocation was already detectable 1.5 months after the start of neuroblastoma treatment — at low cumulative doses of topoisomerase-II inhibitors — and was not detectable in marrow at neuroblastoma diagnosis or in normal-subject lymphocyte DNA 2.
The second was a de novo paediatric precursor-B ALL with t(11;17)(q23;p13), in which FISH and RT-PCR revealed an MLL/GAS7 chimera identical to the one described in the therapy-related AML 3.
So one junction, two lineages, and one aetiology (topoisomerase-II exposure) that covers only one of the two. That is direct human evidence against any account in which the retained partner domains fix the disease lineage, and it matches the mouse model, in which the same construct produces three pathologically distinct leukaemias 5.
Why a three-patient fusion has the class's best mouse model
MLL-GAS7 was chosen as the representative MLL oncoprotein for a cell-of-origin study, not because it is common but because it is mechanistically clean: a cytoplasmic partner with no transcriptional effector domain, activating MLL through dimerisation 46.
What that model showed 5: MLL-GAS7 impairs differentiation and enhances in vitro growth of murine haematopoietic cells with multipotent features; immunophenotype and transcriptional programme indicated multilineage potential; and transplantation produced three pathologically distinct leukaemias, including an acute biphenotypic leukaemia recapitulating the distinctive hallmark features of many human MLL-associated leukaemias. The authors' reading was that the disease originates from multipotential progenitors arrested at a bipotential stage "specifically targeted or induced by MLL oncogenes" — a hedge between cell-of-origin and cell-state-imposition that they did not resolve and that this entry does not resolve either.
The Hox follow-up 6 is the most-quoted and most-misquoted result attached to this fusion. Hoxa7-/- and Hoxa9-/- progenitors remained susceptible to MLL-GAS7 transformation, so neither gene is individually necessary — but they generated transformed blast CFUs less efficiently and produced leukaemias with longer latency, reduced penetrance and less mature phenotypes. The correct summary is that Hoxa7 and Hoxa9 tune the efficiency and maturity of a dimerisation-driven MLL leukaemia without gating it, and that single-gene deletions in a redundant cluster cannot bound the HOXA programme's role.
Cell type, cell state, localization, epigenetics
Cell type. Multipotent murine haematopoietic progenitors, by immunophenotype and transcriptional programme, transplantable into three distinct leukaemias 5. In humans, one infant AML, one paediatric precursor-B ALL, one unclassified 123. This is the only fusion in this beat with an experimentally identified target compartment.
Cell state. Impaired differentiation with enhanced in vitro growth; arrest at a bipotential stage 5. The mouse disease spectrum includes a biphenotypic leukaemia, so the state driven by this fusion is lineage-ambiguous rather than lineage-committed.
Subcellular localization. Not reported for the fusion protein. Wild-type GAS7 is cytoplasmic and membrane/F-actin-associated; the MLL-SEPTIN6 experiment showing nuclear-speckle relocalisation of a cytoplasmic-partner fusion has never been repeated for MLL-GAS7.
Subnuclear localization. Not reported.
Epigenetics. No H3K79 methylation, H3K4 methylation or HOXA-locus ChIP was found for MLL-GAS7. The fusion has a genetic Hox readout 6 and a transcriptional-programme readout 5, neither of which is a chromatin measurement. Stated so that the entry is not read as covering epigenetics when it does not.
What is missing, and the one experiment worth naming
No menin-inhibitor test has ever been published for MLL-GAS7, and the reagent gap is smaller here than anywhere else in this beat: the retroviral construct, the immortalised progenitor cultures and the transplant model are all published and the fusion activates MLL by a mechanism with no elongation component whatsoever 45.
That makes MLL-GAS7 the most informative available test of the menin-interface hypothesis outside the elongation partners. If a fusion whose partner supplies only an oligomerisation surface is killed by revumenib, menin dependence belongs to the KMT2A N-terminus and generalises across partner classes. If it is not, the hypothesis is confined to the fusions that recruit elongation machinery — which would be the single most consequential negative result this project could obtain.
Also missing: any structure of the GAS7 F-BAR or coiled coil (only the SH3 and WW accessory domains are solved), any measurement of the fusion's oligomeric stoichiometry, and any human KMT2A::GAS7 cell line.
| gene5 | KMT2A |
|---|---|
| gene3 | GAS7 |
| cytoband | t(11;17)(q23;p13); recorded in the recombinome as t(11;17)(q23.3;p13.1). Frequently cryptic: the index case presented as del(11)(q23) on karyotype. |
| uniprot5 | Q03164 |
| uniprot3 | O60861 |
| uniprot3 verified | O60861 = GAS7_HUMAN, PRIMARY gene name GAS7, canonical displayed isoform O60861-3, 476 aa, Homo sapiens (UniProt REST, fetched 2026-09-19; primary gene name and length read back from the record). |
| prognosis | Not estimable. Three patients in the largest series; no cohort, no survival data. |
| PDB note | 2LX7 = 'Solution NMR structure of SH3 domain of growth arrest-specific protein 7 (GAS7) (fragment 1-60) from Homo sapiens' (NESG target HR8574A); 2YSH = 'Solution structure of the WW domain from the human growth-arrest-specific protein 7, GAS-7'. Titles and methods verified against the RCSB entry API on 2026-09-19. Neither covers the F-BAR or the coiled coil, and there is no structure of the fusion. |
| mechanism class note | Assigned on direct experimental evidence — the GAS7 coiled coil was shown necessary AND sufficient, and synthetic dimerisation modules substituted. The open structural question is whether that module dimerises or, as the F-BAR work implies, assembles into a two-dimensional sheet; the class assignment does not depend on which. |
| druggability class | unexplored |
| druggability note | No menin-inhibitor test has ever been published for MLL-GAS7 despite published constructs, immortalised cultures and a transplant model. This is the most informative available menin experiment outside the elongation partners. |
| stage of hematopoiesis | multipotent haematopoietic progenitor, established experimentally by immunophenotype, transcriptional programme and the induction of three distinct leukaemias including a biphenotypic one |
| age peak | paediatric in all reported cases |
| subcellular localization | NOT REPORTED for the fusion protein. Wild-type GAS7 is cytoplasmic, F-actin-associated and membrane-shaping. |
| subnuclear localization | NOT REPORTED. |
| epigenetic readout | NONE PUBLISHED. The fusion has a genetic Hox readout (Hoxa7-/-, Hoxa9-/-) and a transcriptional-programme readout, but no histone-modification or chromatin-occupancy measurement. |
| cell lines note | No human KMT2A::GAS7 cell line exists. The available models are murine retroviral transduction cultures and transplant recipients. |
| key interactors note | menin binding is inferred from the retained KMT2A N-terminus and untested for this fusion. HOXA9 and MEIS1 are listed because they were tested genetically (HOXA9) or are upregulated by the equivalent synthetic-dimerisation construct (MEIS1), not because ChIP or expression data exist for MLL-GAS7 itself. |
diseases
[ "therapy-related AML (post topoisomerase-II inhibitor)", "de novo paediatric precursor-B ALL", "infant AML" ]
frequency — the numbers behind the chart
{
"dcal_recombinome_n": 3,
"dcal_recombinome_rank": 28,
"dcal_recombinome_total": 3401,
"infant_aml": 1,
"paediatric_all": 1,
"unclassified_other": 1
}breakpoints
[
{
"id": "bp-kmt2a-gas7-cryptic-t11-17",
"population": "the index therapy-related AML case, in which a karyotypic del(11)(q23) proved to be a cryptic t(11;17) and TWO different MLL-GAS7 fusion transcripts were expressed; an identical chimera was later reported in a de novo paediatric precursor-B ALL",
"domains_retained_5": [
"MBM1",
"MBM2",
"AT-hook 1",
"AT-hook 2",
"AT-hook 3",
"SNL1/2",
"CXXC"
],
"domains_lost_5": [
"PHD1",
"PHD2",
"PHD3",
"Bromodomain",
"PHD4",
"FYRN",
"TAD",
"FYRC",
"WIN",
"SET"
],
"domains_retained_3": [
"F-BAR domain (aa 196-456)",
"coiled coil (aa 309-419, lying within the F-BAR)"
],
"domains_lost_3": [],
"phenotype_notes": "Exon numbers are deliberately omitted for BOTH genes: the sources report a cryptic rearrangement identified by panhandle PCR and two distinct fusion transcripts, without a single canonical exon junction, and none is inferred here. The retained-3 array lists only the modules the functional literature identifies as present and load-bearing; whether the SH3 and WW domains are retained depends on the GAS7 isoform and breakpoint and is not established, so they appear in neither array."
}
]PDB IDS
[ "2LX7", "2YSH" ]
mechanism class
[ "forced-oligomerisation" ]
cell lines
[]
drugs
[]
trials
[]
key interactors
[ "gene-men1", "gene-hoxa9", "gene-meis1" ]
- The index case shows a KMT2A fusion arising 1.5 months into topoisomerase-II-inhibitor therapy at low cumulative dose, in a child being treated for a different cancer, and presenting cytogenetically as a deletion rather than a translocation. Any intensification of such regimens carries this risk, and the resulting leukaemia is invisible to karyotype.
- Because the same junction produced myeloid disease in one child and B-lymphoid disease in another, a therapy chosen on the fusion alone is being chosen without knowing which lineage programme it will have to act in — and lineage switch under lineage-directed immunotherapy is documented for the neighbouring cytoplasmic-partner fusion KMT2A::EPS15.
- Wild-type GAS7 is required for macrophage phagocytosis, so partner-directed inhibition would be expected to impair innate immune clearance in neutropenic patients, a toxicity easily misattributed to the underlying disease.
- Is MLL-GAS7 killed by a menin inhibitor? The constructs and the mouse model are published and the partner supplies no elongation machinery, making this the cleanest available test of whether menin dependence belongs to the KMT2A N-terminus rather than to partner-recruited elongation complexes.
- Does MLL-GAS7 dimerise or form a sheet? The synthetic-dimeriser substitution says a dimer suffices; the F-BAR structural work says GAS7 itself assembles into a two-dimensional lattice. Nobody has measured the fusion's stoichiometry, and the answer changes what an interface inhibitor would have to block.
- Why did the identical chimera produce therapy-related AML in one child and de novo precursor-B ALL in another? This is the sharpest lineage-plasticity observation in the whole non-elongation partner class and has never been followed up.
- Does MLL-GAS7 deposit H3K79me2 at HOXA? For a fusion with no DOT1L-recruiting partner, this is the experiment that would separate 'dimerisation activates HOXA through the MLL moiety' from 'dimerisation recruits elongation machinery indirectly'.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In Table 1 of the 3401-patient 2023 KMT2A recombinome, GAS7 is the 28th-ranked direct translocation partner with 3 patients (1 infant AML, 1 paediatric ALL, 1 unclassified other), and GAS7 is listed among the historically recognised partners at t(11;17)(q23.3;p13.1), 17p13.1, attributed to Megonigal et al. (2000) with disease annotation t-AML.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, Burmeister T, Gröger D, Sutton R, et al. · Leukemia 2023What this evidence cannot show
Read directly from the Europe PMC full-text XML of PMC10169636, Table 1 column structure verified from the header. Three patients is a count where any single ascertainment decision changes the whole distribution. Referral cohort, ascertainment-weighted, a floor rather than an incidence.
Evidence that argues against this (1)
- This partner is specifically prone to being missed: the index case's karyotype read as del(11)(q23) and the t(11;17) was cryptic, requiring panhandle PCR with an unknown partner to identify. A count derived largely from karyotype- or panel-directed testing will under-represent it by an unknown factor. Megonigal MD et al. 2000
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a child who developed AML during neuroblastoma therapy, the leukaemic clone's karyotypic del(11)(q23) was shown by panhandle PCR to be a cryptic t(11;17) with GAS7 at 17p13 as the MLL partner; two different MLL-GAS7 fusion transcripts were expressed; the translocation was already detectable 1.5 months after the start of neuroblastoma treatment and was not detectable in marrow DNA at neuroblastoma diagnosis nor in peripheral-blood lymphocyte DNA from six normal subjects, establishing GAS7 as a new MLL partner gene in treatment-related AML and showing MLL translocations can be present early at low cumulative doses of topoisomerase-II inhibitors.
Detection of leukemia-associated MLL-GAS7 translocation early during chemotherapy with DNA topoisomerase II inhibitors. · Megonigal MD, Cheung NK, Rappaport EF, Nowell PC, Wilson RB, Jones DH, et al. · Proc Natl Acad Sci U S A 2000What this evidence cannot show
A single patient tracked retrospectively. 'Not detectable at diagnosis' is bounded by PCR sensitivity and is not evidence of absence; six normal controls is a small specificity panel. That two different fusion transcripts were expressed from one rearrangement means 'the MLL-GAS7 protein' is already ambiguous in the index case.
Evidence that argues against this (1)
- An aetiology tied to topoisomerase-II-inhibitor exposure cannot be general for this fusion, because an identical MLL/GAS7 chimera was subsequently found in a DE NOVO paediatric precursor-B ALL with no such exposure. Panagopoulos I et al. 2006
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
MLL/GAS7 resulting from t(11;17)(q23;p13), previously reported in one case of treatment-related AML, was found in a de novo case of paediatric precursor-B acute lymphoblastic leukaemia, where FISH and RT-PCR revealed an MLL/GAS7 chimera identical to the one previously described in AML.
MLL/GAS7 fusion in a pediatric case of t(11;17)(q23;p13)-positive precursor B-cell acute lymphoblastic leukemia. · Panagopoulos I, Lilljebjörn H, Strömbeck B, Hjorth L, Olofsson T, Johansson B · Haematologica 2006What this evidence cannot show
A single case. No DOI is registered in the PubMed record for this article, so the citation is anchored on PMID and URL. Read from the abstract. 'Identical' refers to the transcript junction; identity of the expressed protein, including which of the index case's two transcripts it matches, is not established.
Evidence that argues against this (1)
- The index case expressed TWO different MLL-GAS7 fusion transcripts, so describing a second patient's chimera as 'identical to the one previously described in AML' identifies it with one of two products and leaves open whether the same protein is being compared. Megonigal MD et al. 2000
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Following fusion with GAS7 and with AF1p/Eps15, proteins that normally reside in the cytoplasm, the coiled-coil oligomerisation domains of those proteins are necessary and sufficient for leukaemogenic transformation induced by the respective MLL fusion proteins, and homodimerisation of MLL by synthetic dimerisation modules mimics bona fide MLL fusion proteins, producing Hox gene activation and enhanced self-renewal of haematopoietic progenitors.
Dimerization contributes to oncogenic activation of MLL chimeras in acute leukemias. · So CW, Lin M, Ayton PM, Chen EH, Cleary ML · Cancer Cell 2003What this evidence cannot show
Read from the abstract; the full text is subscription-only and cell.com returned HTTP 403, so construct boundaries, the synthetic dimerisation module's identity, the self-association assay and the quantitative data are NOT verified and are asserted nowhere in this entry. Mouse progenitor assays. The authors hedge the mechanism themselves: transformation proceeds 'presumably in part by recruitment of accessory factors through the dimerized MLL moiety'.
Evidence that argues against this (1)
- Structural work published since places GAS7's self-assembly determinant in the F-BAR domain's hydrophilic loops, which form two-dimensional membrane-templated SHEETS rather than a coiled-coil dimer, and the annotated coiled coil (aa 309-419) lies inside the F-BAR (aa 196-456). The identity of the transforming module as a simple dimerising coiled coil is therefore not settled. Hanawa-Suetsugu K et al. 2019
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MLL-GAS7 impairs the differentiation and enhances the in vitro growth of murine haematopoietic cells with multipotent features; their multilineage differentiation potential was suggested by immunophenotype and transcriptional programme and confirmed by their ability to induce three pathologically distinct leukaemias in mice, including an acute biphenotypic leukaemia that recapitulates the distinctive hallmark features of many MLL-associated leukaemias in humans, modelling that disease as originating from multipotential progenitors arrested at a bipotential stage specifically targeted or induced by MLL oncogenes.
MLL-GAS7 transforms multipotent hematopoietic progenitors and induces mixed lineage leukemias in mice. · So CW, Karsunky H, Passegué E, Cozzio A, Weissman IL, Cleary ML · Cancer Cell 2003What this evidence cannot show
Retroviral transduction of mouse progenitors with a fusion found in three patients worldwide; the model's relevance rests on the dimerisation mechanism generalising, not on the fusion's frequency. Read from the abstract, so marker panels, construct boundaries and the GAS7 isoform used are not verified here. 'Specifically targeted or induced by' is the authors' own unresolved hedge between cell-of-origin and cell-state-imposition.
Evidence that argues against this (1)
- The phenotype is not fixed by the fusion: in the same group's genetic follow-up, MLL-GAS7 transformation of Hoxa7- or Hoxa9-null progenitors yielded leukaemias with LESS MATURE phenotypes, longer latency and reduced penetrance, so the disease that emerges is shaped by the progenitor's own downstream programme rather than dictated by the oncoprotein. So CW et al. 2004
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In retroviral transduction/transplantation assays, Hoxa7-deficient and Hoxa9-deficient progenitors remained susceptible to transformation by MLL-GAS7, which activates MLL through a dimerisation-dependent mechanism, but were less efficient at generating transformed blast colony-forming units in vitro and induced leukaemias with longer disease latency, reduced penetrance and less mature phenotypes; Hoxa7 and Hoxa9 therefore contribute to progenitor homeostasis and modify disease latency, penetrance and phenotype without being necessary for MLL-GAS7-mediated leukaemogenesis.
Leukemic transformation of hematopoietic progenitors by MLL-GAS7 in the absence of Hoxa7 or Hoxa9. · So CW, Karsunky H, Wong P, Weissman IL, Cleary ML · Blood 2004What this evidence cannot show
Single-gene germline knockouts in a highly redundant cluster: Hoxa7 and Hoxa9 were deleted separately and never together, and the remaining HOXA genes and MEIS1 are intact, so this bounds each gene's individual requirement and not the HOXA programme's. Read from the abstract. The result is for a dimerisation-driven fusion and should not be transferred to elongation-partner fusions without testing.
Evidence that argues against this (1)
- The dimerisation mechanism is otherwise tightly coupled to Hoxa9: induced dimerisation of truncated MLL upregulates Hoxa7, Hoxa9 and Meis1, and dimerised MLL binds with higher affinity than undimerised MLL to a CpG island within the Hoxa9 locus. Transformation persisting in Hoxa9-null progenitors therefore requires either paralogue compensation or a Hox-independent arm, neither of which has been identified. Martin ME et al. 2003
- E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
The GAS7 protein retained in this fusion (UniProt O60861, GAS7_HUMAN, primary gene name GAS7, canonical isoform O60861-3, 476 aa) carries an SH3 domain at aa 1-62, a WW domain at aa 77-110, an F-BAR domain at aa 196-456 and a coiled coil at aa 309-419 that lies entirely within the F-BAR; the only solved structures are of the SH3 domain (PDB 2LX7) and the WW domain (PDB 2YSH), both by solution NMR, and there is no structure of the F-BAR, of the coiled coil, or of any MLL-GAS7 fusion.
UniProtKB O60861 (GAS7_HUMAN) Growth arrest-specific protein 7, Homo sapiens, 476 aa · UniProt Consortium · UniProtKB 2026What this evidence cannot show
Canonical isoform only, and GAS7 has four annotated isoforms with materially different domain content, so these coordinates do not transfer between isoforms. Which GAS7 isoform is present in the fusion is not established by any source read for this entry. PDB titles and methods for 2LX7 and 2YSH were verified against the RCSB entry API on 2026-09-19. Resolved by fetching the accession and reading back the primary gene name and length.
Evidence that argues against this (1)
- The two characterised human GAS7 transcripts encode 330 aa (FCH domain only) and 412 aa (WW plus FCH), neither containing the SH3 domain present in the canonical entry, and the shorter is roughly fourteen times more abundant — so the canonical coordinates may not describe the species actually expressed in any relevant tissue. Chao CC et al. 2005
anti-tag co-IP with the MLL-GAS7 construct in HEK293, IntAct interaction type 'association' (the weaker of the two terms in this enumeration), PMID 26766589
- part of →KMT2A / MLL1 — the gene that breaks
retained KMT2A N-terminus (MBM1/2, AT-hooks, CXXC)
- part of →GAS7 — growth-arrest-specific F-BAR protein, the second pillar of the forced-oligomerisation model
GAS7 F-BAR and the coiled coil lying within it
the GAS7 coiled coil is necessary AND sufficient; synthetic dimerisers substitute
transforms Hoxa9-null progenitors but with longer latency, reduced penetrance and less mature phenotypes
multipotent progenitor target; transformed cells induce three distinct leukaemias including a biphenotypic one
acute biphenotypic leukaemia in the mouse model; identical human chimera produced AML in one child and precursor-B ALL in another
the two partners whose coiled-coil necessity AND sufficiency founded the forced-oligomerisation model, in the same experiment
both are the non-chromatin counterexample inside their own family: a kinase 5' partner with its catalytic domain RETAINED, against a family otherwise built from chromatin and elongation partners
same partner class - a cytoplasmic protein supplying a coiled-coil oligomerisation module - and both are weak relative to the elongation-complex fusions