GAS7 — growth-arrest-specific F-BAR protein, the second pillar of the forced-oligomerisation model

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• GAS7 is required for phagocytosis in macrophages. An agent that inhibited GAS7 itself, rather than the fusion's oligomerisation surface, would be expected to impair innate immune clearance in patients who are already neutropenic and infection-prone from chemotherapy — a toxicity that would be easy to attribute to the underlying disease.
• If the oligomerisation surface is an F-BAR sheet rather than a coiled coil, then inhibitors designed against a coiled-coil interface would fail for reasons nobody would diagnose, because the field's stated model of the target would be wrong rather than the chemistry.
• The index MLL-GAS7 case appeared 1.5 months into topoisomerase-II-inhibitor therapy at low cumulative dose. Any treatment intensification using such agents in another cancer carries a measurable risk of creating exactly this class of fusion, and the resulting leukaemia is cytogenetically cryptic.

1. Is GAS7's transforming module a coiled coil or an F-BAR sheet? The UniProt coiled coil (aa 309-419) sits inside the F-BAR (aa 196-456), and the F-BAR is the part shown to self-assemble — into two-dimensional sheets, not dimers. Re-running the 2003 deletion series against the 2019 loop mutants would settle whether the model's founding module was correctly identified.
2. What is the stoichiometry of MLL-GAS7? 'Dimerisation' is asserted throughout this literature and has never been measured for the fusion. If GAS7 templates a sheet, the fusion may present many MLL N-termini rather than two, which is a different mechanism with different drug implications.
3. Which GAS7 isoform is in the fusion, and which is expressed in haematopoietic progenitors? The three characterised human products differ by an SH3 domain and a WW domain, and no source read here states which was used in the transforming constructs.
4. Does MLL-GAS7 relocalise to the nucleus, and does the F-BAR still bind membranes in the fusion? A fusion protein carrying an intact membrane-binding module is an odd object and nobody has imaged it.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   Human GAS7 is UniProt O60861 (GAS7_HUMAN), primary gene name GAS7, synonym KIAA0394, canonical displayed isoform O60861-3 of 476 aa with four annotated splice isoforms, domains SH3 at aa 1-62, WW at aa 77-110 and F-BAR at aa 196-456, an annotated coiled coil at aa 309-419 lying within the F-BAR, disordered region at aa 100-171, subcellular location cytoplasm, and cross-referenced structures 2LX7 and 2YSH.

   UniProtKB O60861 (GAS7_HUMAN) Growth arrest-specific protein 7, Homo sapiens, 476 aa · UniProt Consortium · UniProtKB 2026
   ▸What this evidence cannot show

   Canonical isoform O60861-3 only; three further isoforms are described and carry different domain content, so every residue number here is isoform-specific. The cytoplasmic localisation is annotated by similarity (ECO:0000250), not from a direct human experiment. Resolved by fetching the accession and reading back the PRIMARY gene name (GAS7) and length (476 aa).

   ▸Evidence that argues against this (1)

   • The two characterised human GAS7 transcripts are 330 aa (FCH domain only) and 412 aa (WW plus FCH), and the shorter one is roughly fourteen times more abundant across the tissues tested — so the 476-residue canonical entry, which additionally contains an SH3 domain, may not be the dominant species in any tissue, and residue numbers taken from it can mislead. Chao CC et al. 2005
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   gas7, identified in serum-starved murine fibroblasts, is expressed in vivo selectively in neuronal cells of the mature cerebral cortex, hippocampus and cerebellum; its transcripts encode a 48 kDa protein containing a structural domain resembling sequences of the POU transcription factor OCT2 and of synapsins; GAS7 expression is prominent in cerebellar Purkinje cells; inhibiting its production in terminally differentiating cultures of embryonic murine cerebellum impedes neurite outgrowth from maturing Purkinje cells; and GAS7 overexpression in undifferentiated neuroblastoma cultures dramatically promotes neurite-like outgrowth.

   gas7: A gene expressed preferentially in growth-arrested fibroblasts and terminally differentiated Purkinje neurons affects neurite formation. · Ju YT, Chang AC, She BR, Tsaur ML, Hwang HM, Chao CC, et al. · Proc Natl Acad Sci U S A 1998
   PMID 9736752DOI
   ▸What this evidence cannot show

   Mouse, 1998, with antisense-style inhibition in primary culture as the loss-of-function tool. The stated resemblance to OCT2 and synapsins is a 1998 sequence inference that later structural work did not sustain — the region is now understood as F-BAR — and is reported here as what the founding paper said, not as current annotation.

   ▸Evidence that argues against this (1)

   • Neuron-selective expression does not hold for the protein's best-characterised function, which is in macrophages: GAS7 F-BAR assemblies drive phagocytic cup formation, and loop mutations abolish phagocytosis. A myeloid role matters here because MLL-GAS7 disease is myeloid or biphenotypic. Hanawa-Suetsugu K et al. 2019
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Two human Gas7 cDNAs were identified: hGas7-a (2427 nt, encoding 330 aa, containing only the FCH domain) and hGas7-b (2610 nt, encoding 412 aa, containing a WW domain and an FCH domain), arising by canonical alternative splicing of one genomic locus; hGas7-a mRNA was about fourteen times more abundant than hGas7-b across brain, breast cancer, placenta and head-and-neck cancer; transfection yielded 38 kDa and 50 kDa proteins respectively and both caused neurite-like processes in mouse Neuro-2a and human SH-SY5Y cells, with hGas7-a preferentially eliciting lamellipodia and hGas7-b filopodia; Gas7 interacts with F-actin and colocalises with the terminal part of actin microfilaments where membrane outgrowth occurs.

   Human Gas7 isoforms homologous to mouse transcripts differentially induce neurite outgrowth. · Chao CC, Chang P, Lu HH · J Neurosci Res 2005
   PMID 15948147DOI
   ▸What this evidence cannot show

   Quantitative PCR abundance across a small and heterogeneous tissue panel; overexpression morphology in two neuroblastoma lines. Neither isoform corresponds to the 476-residue UniProt canonical sequence, and the paper does not address which isoform is present in haematopoietic cells or which was used in the leukaemia constructs.

   ▸Evidence that argues against this (1)

   • UniProt records four GAS7 isoforms and displays a 476-residue canonical form that contains an SH3 domain absent from both transcripts characterised here, so the two-isoform picture is incomplete and the field's residue numbering is not consistent across sources. UniProt Consortium 2026
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Two-dimensional sheet-like assembly of GAS7 plays a critical role in phagocytic cup formation in macrophages: the GAS7 F-BAR domain possesses unique hydrophilic loops that allow two-dimensional sheet formation on flat membranes, super-resolution microscopy reveals similar assemblies on phagocytic cups and on liposomes, and mutating those loops abolishes both membrane localisation of GAS7 and phagocytosis — in contrast to most BAR-domain proteins, which assemble into helical submicron-diameter filaments to make invaginations or protrusions.

   Phagocytosis is mediated by two-dimensional assemblies of the F-BAR protein GAS7. · Hanawa-Suetsugu K, Itoh Y, Ab Fatah M, Nishimura T, Takemura K, Takeshita K, et al. · Nat Commun 2019
   PMID 31628328DOI
   ▸What this evidence cannot show

   Purified protein on liposomes plus macrophage imaging; the assemblies are membrane-templated, and nothing here shows that GAS7 or an MLL-GAS7 fusion oligomerises in the nucleus or in the absence of a membrane. It is cited to challenge the assumption that GAS7's oligomerisation is simple coiled-coil dimerisation, not to describe the fusion.

   ▸Evidence that argues against this (1)

   • The leukaemia result that defines GAS7's role in this graph attributes transformation specifically to a COILED-COIL oligomerisation domain and shows that synthetic dimerisation modules substitute for it — which a membrane-templated two-dimensional sheet cannot easily explain, since a synthetic dimeriser makes a dimer, not a lattice. Either the transforming module is a coiled coil functioning independently of the F-BAR sheet, or dimerisation and sheet formation are both sufficient by different routes. So CW et al. 2003
5. E5established 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 and 1 in the unclassified-other category; 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 2023
   PMID 37019990DOI
   ▸What this evidence cannot show

   Read directly from the Europe PMC full-text XML of PMC10169636 with the Table 1 column structure verified from the header. Three patients is a number where a single ascertainment decision changes the distribution entirely. Referral cohort, ascertainment-weighted.

   ▸Evidence that argues against this (1)

   • The index MLL-GAS7 case demonstrates the exact failure mode that makes such counts floors: its karyotype read as del(11)(q23) and the t(11;17) was cryptic, requiring panhandle PCR with an unknown partner gene to identify. A compendium relying on karyotype-directed or panel-directed testing will systematically miss this rearrangement. Megonigal MD et al. 2000
6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The coiled-coil oligomerisation domains of AF1p/Eps15 and GAS7, proteins that normally reside in the cytoplasm, are necessary and sufficient for leukaemogenic transformation induced by the respective MLL fusion proteins; homodimerisation of MLL by synthetic dimerisation modules mimics bona fide MLL fusion proteins, resulting in Hox gene activation and enhanced self-renewal of haematopoietic progenitors, supporting an oligomerisation-dependent mechanism for oncogenic conversion of MLL.

   Dimerization contributes to oncogenic activation of MLL chimeras in acute leukemias. · So CW, Lin M, Ayton PM, Chen EH, Cleary ML · Cancer Cell 2003
   PMID 12957285DOI
   ▸What this evidence cannot show

   Read from the abstract; the full text is subscription-only and cell.com returned HTTP 403, so the residue boundaries of the coiled-coil constructs, the identity of the synthetic dimerisation module, the self-association assay and the quantitative transformation data are NOT verified and are asserted nowhere in this entry. Mouse progenitor transduction assays. The authors' own conclusion is hedged: transformation proceeds 'presumably in part by recruitment of accessory factors through the dimerized MLL moiety'.

   ▸Evidence that argues against this (1)

   • Structural work on GAS7 published since locates its self-assembly determinant in the F-BAR domain's hydrophilic loops, which form two-dimensional sheets rather than a simple coiled-coil dimer, and the UniProt-annotated coiled coil (aa 309-419) lies inside the F-BAR (aa 196-456). The identity of GAS7's transforming module as 'a coiled coil' is therefore not independent of a domain with quite different assembly behaviour. Hanawa-Suetsugu K et al. 2019
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   MLL-GAS7 impairs 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 recapitulating the distinctive hallmark features of many MLL-associated leukaemias in humans — modelling that disease as arising 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 2003
   PMID 12620410DOI
   ▸What this evidence cannot show

   Retroviral transduction of mouse progenitors with a fusion found in three human patients worldwide, so the model's clinical representativeness rests on the mechanism generalising, not on the fusion's frequency. Read from the abstract; the constructs, the GAS7 isoform used and the marker panels are not verified here. 'Specifically targeted or induced by' is the authors' own hedge between a cell-of-origin and a cell-state-imposition interpretation, and this entry does not resolve it.

   ▸Evidence that argues against this (1)

   • The multipotent-target reading is constrained by the same group's later genetic work, in which transformation by MLL-GAS7 still occurred in Hoxa7- or Hoxa9-null progenitors but produced leukaemias with LESS MATURE phenotypes, longer latency and reduced penetrance — indicating the resulting disease phenotype is shaped by the progenitor's downstream programme rather than fixed by the fusion. So CW et al. 2004
8. E8strong 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; separately, Hoxa7-/- and Hoxa9-/- mice showed selective reductions in committed progenitors but not haematopoietic stem cells, with MEP reductions and reticulocytosis/thrombocytopenia in Hoxa7-/- and marked lymphopenia with reduced CLPs, CMPs and GMPs in Hoxa9-/-.

   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 2004
   PMID 15070702DOI
   ▸What this evidence cannot show

   Single-gene germline knockouts in a family with extensive paralogue redundancy: Hoxa7 and Hoxa9 were deleted separately, never together, and other HOXA cluster genes plus MEIS1 remain intact, so this bounds the requirement for each gene alone and not for the HOXA programme. Read from the abstract. Results are for MLL-GAS7 and may not transfer to elongation-partner fusions.

   ▸Evidence that argues against this (1)

   • Induced dimerisation of truncated MLL upregulates Hoxa7, Hoxa9 and Meis1 and dimerised MLL binds with higher affinity to a CpG island within the Hoxa9 locus, so the dimerisation mechanism is tightly coupled to Hoxa9 activation — which makes the persistence of transformation in Hoxa9-null progenitors a genuinely awkward result requiring either paralogue compensation or a Hox-independent arm that has not been identified. Martin ME et al. 2003
9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

   GAS7 is not a dependency in any leukaemia context measured: DepMap Chronos mean gene effect -0.0533 across 1208 lines, KMT2A-rearranged mean -0.048 versus -0.073 in other leukaemia lines (difference +0.025, i.e. marginally LESS required in KMT2A-rearranged lines), rank 12,299 of 17,728 (69.4th percentile) for KMT2A selectivity; splitting the index arm by lineage gives +0.89 SD in the 8-line myeloid half and -0.89 SD in the 5-line B-lineage half, with no dependency in either.

   DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API: genome-wide KMT2A-selectivity scan (analyses/depmap-kmt2a-selective-scan-full/results.json) plus the lineage-split re-analysis of the deferred KMT2A partners run for this beat (analyses/depmap-deferred-partners-lineage-split/) · Broad Institute DepMap; Open Targets Platform; lineage split run by this project · Database and derived analysis 2026
   ▸What this evidence cannot show

   Decisive limitation: no MLL-GAS7 cell line exists in DepMap, so this measures wild-type GAS7 in leukaemias driven by other partners. The ±0.89 SD split across 8 and 5 lines on an essentially zero absolute effect is noise, not a lineage signal, and is reported in both halves precisely so it is not mistaken for one. DepMap release version not stated by the API.

   ▸Evidence that argues against this (1)

   • A wild-type-partner CRISPR null says nothing about whether the partner's module is required within the fusion: Sept6-null mice develop normally and MLL-SEPT6 disease is unchanged on that background, while the SEPT6 coiled coil is individually required for transformation when present in the fusion. The same logic applies to GAS7, whose coiled coil is reported necessary and sufficient in the fusion. Ono R et al. 2005
10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a child who developed acute myeloid leukaemia 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 gene; 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 at neuroblastoma diagnosis nor in peripheral-blood lymphocyte DNA from six normal subjects — establishing GAS7 as a new MLL partner in treatment-related AML and showing MLL translocations can appear 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 2000
    PMID 10706619DOI
    ▸What this evidence cannot show

    A single patient tracked retrospectively, with PCR sensitivity as the detection limit — 'not detectable at diagnosis' is not 'absent'. Six normal controls is a small specificity panel for a PCR assay.

    ▸Evidence that argues against this (1)

    • MLL-GAS7 is not confined to therapy-related disease: a de novo paediatric t(11;17)(q23;p13) precursor-B ALL carried an MLL/GAS7 chimera identical to the one described in the therapy-related AML, so an aetiological account tied to topoisomerase-II-inhibitor exposure does not cover the partner. Panagopoulos I et al. 2006
11. E11moderate 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 also found in a de novo case of paediatric 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 2006
    PMID 16956839
    ▸What 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 chimera' is a statement about the transcript junction, and the same junction producing B-ALL in one child and t-AML in another is itself the finding that limits how much the junction explains.

    ▸Evidence that argues against this (1)

    • That the identical chimera produced treatment-related AML in one patient and de novo precursor-B ALL in another argues the fusion does not determine lineage, which is consistent with the mouse model in which MLL-GAS7 transforms multipotent progenitors and produces three pathologically distinct leukaemias including a biphenotypic one. So CW et al. 2003

• ← part ofKMT2A::GAS7 (MLL-GAS7), t(11;17)(q23;p13) — three patients, and the best mouse model in the partner class

  GAS7 F-BAR and the coiled coil lying within it

• ← fuses withKMT2A / MLL1 — the gene that breaks

  cryptic t(11;17)(q23;p13); 3 patients in 3401

• related to →Non-elongation KMT2A partners considered and not written: what the searches actually returned, partner by partner

  deferred by that enumeration as mechanistically load-bearing but clinically negligible; written here