Oncofusion

start / graph / gene-smarca2

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q47-swi-snf

SMARCA2 (BRM) - the paralogue nobody measured, and the reason a DepMap rank cannot predict what a BRG1/BRM inhibitor does

also called BRM · hBRM · BAF190B · SNF2A · SNF2L2 · SMCA2_HUMAN

SMARCA2 is the alternative motor that can slot into the same chromatin machine as SMARCA4. Knock out one and the other often covers for it - which is why the drugs that reached the clinic hit both, and why a genetic screen that tests them one at a time understates the target.

The same thing, for a scientist

SMARCA2 (P51531, 1,590 aa) is the mutually exclusive alternative to SMARCA4 as the mSWI/SNF ATPase [E1]. In leukaemia the two are not equivalent: quantitative proteomics shows BAF complexes expressed in leukaemia are **specifically assembled around the BRG ATPase**, and BRG is essential for leukaemia maintenance while being dispensable for immunophenotypic long-term repopulating HSCs [E2]. That asymmetry is the stated basis for a therapeutic window. But redundancy re-appears the moment you perturb: dual inactivation of Brg1 and Brm has only minimal effects on hematopoietic stem and progenitor cells [E3], and every clinical-stage molecule in this class - FHD-286, BRM014 - inhibits both ATPases [E4][E5]. DepMap gives SMARCA2 no KMT2A selectivity (rank 11,581/17,728, difference +0.019) and near-zero pan-essentiality (+0.0285), i.e. the average cell line does not need it [E6]. **This entry exists because SMARCA2 is the single largest reason the SMARCA4 genetic and pharmacological pictures disagree, and because `SMARCA2 AND (MLL OR KMT2A)` returns 6 PubMed records of which none is a primary study of SMARCA2 in this disease [E7].**

What it looks like
loading AF-P51531-F1…
AlphaFold AF-P51531-F1 ↗
colourshape

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.

very high (90+)confident (70–90)low (50–70)very low (<50)
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.

Go deeper
The identifier trap, stated plainly

SMARCA4 is BRG1. SMARCA2 is BRM. They are different genes and swapping them inverts a claim. The usual defences are weak here:

accessionprimarysynonymsprotein name containslength
P51532SMARCA4BAF190A BRG1 SNF2B SNF2L4BRG1-associated factor 190A1,647
P51531SMARCA2BAF190B BRM SNF2A SNF2L2BRG1-associated factor 190B, hBRM1,590

Reading the primary protein name back does not separate them: both contain the string "BRG1-associated factor 190", differing only in the trailing letter. Both full names begin identically ("SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily A member"). Length differs by only 57 residues out of ~1,600 (3.5%). The only clean discriminator is gene_primary 1.

This is a rule-65-class collision and it points in opposite therapeutic directions in the solid-tumour literature, where BRM inhibition in BRG1-deficient tumours is a synthetic-lethal strategy 8 - so a swap there turns a synthetic-lethal proposal into its target.

Why the paralogue is the crux of the beat

Three facts have to be held together:

  1. Leukaemic BAF is assembled around BRG, not BRM 2.
  2. Genetic SMARCA4 knockout shows no KMT2A selectivity and SMARCA2 knockout shows none either 6.
  3. Dual pharmacological inhibition kills MLL1-rearranged AML and works in xenografts 4.

The reconciling model is that the ATPase slot is the dependency and either paralogue can fill it, so no single-gene knockout scores. If that is right, the correct DepMap readout is a double knockout, which the public dataset does not contain, and every single-gene rank in this beat's tables understates the ATPase module. If it is wrong - if FHD-286's leukaemia activity is off-target or driven by something other than the ATPases - then the clinical rationale in this disease is weaker than it looks. Neither has been tested.

Facts and figures
uniprotP51531
gene symbolSMARCA2
protein nameSWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily A member 2
length AA1590
alphafold IDAF-P51531-F1
accession resolutionUniProt REST gene_exact:SMARCA2 AND organism_id:9606 AND reviewed:true, run 2026-09-20; one reviewed row; gene_primary read back as SMARCA2, length 1590 from the same record.
identifier trapboth protein names contain 'BRG1-associated factor 190'; both full names share a 92-character prefix; length differs by 3.5%. Only gene_primary discriminates.
classificationparalogue of the dependency; not itself a measured KMT2A-rearranged dependency; pharmacologically inseparable from SMARCA4 with current agents
subnuclear localizationNUCLEAR MATRIX, AND THE ASSOCIATION IS PROTEOLYTICALLY REVERSIBLE IN A MYELOID LEUKAEMIA CELL. This is a positive subnuclear record and it is unusual in this tranche for being measured in a haematopoietic cell: full-length hBRM is tightly nuclear-matrix-associated in NB4 cells and the cathepsin-G-generated 160-kDa fragment, lacking the C-terminal bromodomain, is less tightly associated [E_q91_smarca2_loc2]. Add the damage-induced redistribution above and SMARCA2 has two separable subnuclear states, one constitutive and one induced. Note what is NOT claimed: no nucleolar, speckle, PML-body or lamina assignment exists, and the nuclear matrix here is an operational fractionation compartment rather than an imaged body.
localization shown for fusionNO for any KMT2A fusion, in any lineage. But NOT a bare NO under rule 86: the nuclear-matrix result was obtained in NB4, a PML-RARA acute promyelocytic leukaemia line, so the compartment record for SMARCA2 exists inside a DIFFERENT fusion family - the correct statement is 'PML-RARA (NB4), as the cell system rather than as a tested variable; no KMT2A fusion at all'. BAF/SWI-SNF in this project's own complex-enrichment analysis of the Skucha 2018 MLL-fusion interactome (analyses/mll-fusion-interactome-complexes/results.json) is 6 of 17 members recovered, p = 9.4e-05 and q_BH = 1.9e-04 AGAINST A 20,431-PROTEIN PROTEOME BACKGROUND - and that background is the one the analysis's own caveat says inflates significance for any complex of abundant nuclear proteins. Re-cast on a realistic denominator the row is NOT SIGNIFICANT: q = 0.0755 against the 5,754-protein nuclear background and q = 0.186 against the 4,600 proteins actually recovered before filtering. Restricted to the INDEX bait (MLL-AF4 = KMT2A::AFF1) only three BAF members appear at all - SMARCB1, SMARCC1, SMARCC2 - and only one of the three has p < 0.05. So nothing in this entry should be read as establishing BAF as a module of the index fusion's interactome; gene-smarcc1 carries the same correction at subunit level.
paralogue
{
  "gene": "SMARCA4",
  "uniprot": "P51532",
  "length_aa": 1647,
  "relation": "mutually exclusive ATPase; 57 residues apart"
}
depmap
{
  "rank": 11581,
  "of": 17728,
  "difference": 0.019,
  "pan_essentiality_all_lines": 0.0285,
  "lineage_split_z_myeloid": -0.18,
  "lineage_split_z_b_lineage": 0.38,
  "verdict": "no dependency, no selectivity - but a single-gene knockout cannot see a paralogue-buffered requirement"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": 0.0615,
  "chronos_mean_other_leukaemia": 0.0425,
  "difference": 0.019,
  "pan_essentiality_all_lines": 0.0285,
  "selectivity_rank": 11581,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
localization
{
  "CONSTITUTIVE": "NUCLEUS, AND SPECIFICALLY ASSOCIATED WITH THE NUCLEAR MATRIX. UniProt P51531 (fetched live 2026-09-21) gives 'Nucleus' on two ECO:0000269 attributions, PubMed:11259672 and PubMed:25593309, plus a note 'Localizes to sites of DNA damage' (ECO:0000269 PubMed:25593309) [E_q91_smarca2_loc1]. The nuclear-matrix association is the more specific fact and it comes from the first of those papers, which fractionates: full-length hBRM is tightly associated with the nuclear matrix, and its 160-kDa cleavage fragment is LESS tightly associated [E_q91_smarca2_loc2].",
  "INDUCED_1_DNA_DAMAGE": "SMARCA2 is a bromodomain protein, and more than one-third of all human BRD proteins CHANGE LOCALISATION in response to DNA damage in the screen UniProt cites; SMARCA2 is annotated from it as localising to sites of DNA damage [E_q91_smarca2_loc1][E_q91_smarca2_loc3].",
  "INDUCED_2_APOPTOSIS_AND_IT_IS_MYELOID_SPECIFIC": "In NB4 acute promyelocytic leukaemia cells, apoptosis induced by UV, DNA-damaging agents or staurosporine cleaves hBRM - NOT by caspase-3, -7 or -6, but by CATHEPSIN G, a granule enzyme present in NB4 cells which is itself confined to granules and the Golgi in untreated cells and becomes DIFFUSELY DISTRIBUTED during apoptosis. The cleavage removes a 20-kDa C-terminal fragment containing a bromodomain and DISRUPTS hBRM's association with the nuclear matrix [E_q91_smarca2_loc2]. Two proteins change compartment in this experiment and the relocation of the protease is what reaches the substrate."
}
cell type
{
  "MEASURED_IN_A_MYELOID_LEUKAEMIA_LINE": "NB4, the t(15;17) PML-RARA acute promyelocytic leukaemia line, for the nuclear-matrix association, the apoptotic cleavage and the cathepsin-G dependence [E_q91_smarca2_loc2]. That is a haematopoietic system and a leukaemic one, though the driving fusion is PML-RARA rather than a KMT2A rearrangement - an analogy to another fusion family rather than a result in this one.",
  "WHY_THE_CELL_TYPE_IS_THE_MECHANISM_HERE": "Cathepsin G is a NEUTROPHIL/PROMYELOCYTE AZUROPHIL-GRANULE serine protease. A degradation pathway that depends on it exists only in cells that make granules - so this SMARCA2 proteolysis is expected in myeloid blasts and NOT in a B-lineage KMT2A::AFF1 blast, which makes a lineage prediction this graph can state and nobody has tested.",
  "in_this_disease": "Nothing. SMARCA2 protein has never been detected, quantified or localised in a KMT2A-rearranged cell, and this entry's existing E7 records that SMARCA2[tiab] AND (MLL OR KMT2A) returns 6 records with no primary study."
}
cell state
{
  "APOPTOSIS": "The defining state change: cleavage, loss of the C-terminal bromodomain, and release from the nuclear matrix, all downstream of a protease that itself relocalises [E_q91_smarca2_loc2].",
  "DNA_DAMAGE": "A separate induced state - recruitment to damage sites, annotated from a screen in which over a third of bromodomain proteins move [E_q91_smarca2_loc3].",
  "PARALOGUE_STATE": "The state that governs whether SMARCA2 matters at all is which ATPase the cell's BAF complexes are built around: leukaemic BAF is specifically assembled around BRG/SMARCA4 (this entry's E2), so SMARCA2's compartment may be occupied by very little protein in exactly the cells of interest. That is this entry's second open question and it is a prerequisite for any localisation experiment here."
}
localization bounded absence
{
  "date": "2026-09-22",
  "tool": "NCBI esearch db=pubmed",
  "run_by": "Q91",
  "queries": [
    {
      "query": "SMARCA2[tiab] AND MLL[tiab] (recorded in this entry's existing E7 as SMARCA2 AND (MLL OR KMT2A) = 6)",
      "count": 6,
      "note": "not re-run by Q91; carried from the entry's own E7, and flagged because the grouped form of that query is the rule-94 shape and would have run broader than labelled, which makes the six an upper bound and the absence therefore stronger rather than weaker"
    }
  ],
  "reading": "The absence here is of any measurement of SMARCA2 in a KMT2A-rearranged cell - abundance, compartment or otherwise. It is NOT an absence of a localisation record for the protein, which exists and is good.",
  "absence_class": "never-measured"
}
How this could bite us back
  • A SMARCA4-selective degrader - the obvious way to widen the window - would be the agent most exposed to SMARCA2 compensation, and no experiment in this disease has tested for it.
  • The dual agents carry SMARCA2 inhibition as pure added exposure if leukaemic BAF really is BRG-assembled; that is a toxicity argument with no efficacy offset.
  • Any published result attributed to 'BRG1 inhibition' in this disease is attributable only to the ATPase module, so a resistance mutation in either paralogue could restore the module.
Still open
  1. What is the dependency of a KMT2A-rearranged line on SMARCA4 and SMARCA2 knocked out TOGETHER? No public dataset answers this and it is the measurement the whole ATPase argument turns on.
  2. Do KMT2A-rearranged blasts express SMARCA2 protein at all? If leukaemic BAF is BRG-assembled (E2), a dual inhibitor may be doing nothing useful with its SMARCA2 activity except adding toxicity.
  3. Is SMARCA2 upregulated in cells that survive FHD-286, the way SMARCA2 loss selects for SMARCA4 dependence in solid tumours?
Every source behind this entry (11)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    SMARCA2 is UniProt P51531, 1,590 aa, primary gene name SMARCA2, synonyms BAF190B BRM SNF2A SNF2L2; SMARCA4 is P51532 at 1,647 aa.

    UniProtKB P51531 (SMCA2_HUMAN) · UniProt Consortium · UniProtKB 2026
    What this evidence cannot show

    gene_exact:SMARCA2 AND organism_id:9606 AND reviewed:true, run 2026-09-20; one reviewed row each; primary names and lengths read back from the fetched records. Note the record's own protein-name field contains the typographical variant '(SAMRCA2)'.

    Evidence that argues against this (1)
    • Both records' protein names contain 'BRG1-associated factor 190', so reading the protein name back - the standard defence under rule 65 - cannot separate BRG1 from BRM. UniProt Consortium 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Quantitative proteomics shows BAF complexes in leukaemia are specifically assembled around the BRG ATPase; BRG is essential for leukaemia maintenance in a mouse AML model and dispensable for immunophenotypic long-term repopulating HSCs.

    PMID 24478402DOI
    What this evidence cannot show

    Mouse AML model; abstract read, full text not retrieved by this beat, so the specific driver oncogene of that model is not asserted here. 'Dispensable for LT-HSC' is an immunophenotypic readout, not a serial-transplant functional one.

    Evidence that argues against this (1)
    • The anchor MLL-rearranged paper found SMARCA4 knockdown reduced LSK colony formation and LSK proliferation, which is a normal-progenitor effect the immunophenotypic LT-HSC readout would not capture. Cruickshank VA et al. 2015
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Dual inactivation of Brg1 and Brm leads to only minimal effects on hematopoietic stem and progenitor cells, which is the stated basis for a therapeutic window for targeting Brg1 in leukaemia.

    Role of SWI/SNF in acute leukemia maintenance and enhancer-mediated Myc regulation. · Shi J, Whyte WA, Zepeda-Mendoza CJ, Milazzo JP, et al. · Genes Dev 2013
    PMID 24285714DOI
    What this evidence cannot show

    Reported by Shi et al. in discussion, citing Willis et al. 2012, which this beat did not retrieve and does not cite. Graded moderate for that reason: it is a secondary statement in a primary paper, not a result of the paper cited.

    Evidence that argues against this (1)
    • A mouse dual-knockout tolerance result does not transfer to sustained pharmacological ATPase inhibition in a child, where the relevant question is cumulative marrow reserve rather than steady-state stem-cell number. Fiskus W et al. 2024
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    FHD-286 is an orally bioavailable selective inhibitor of BRG1/BRM in clinical development in AML and is active against MLL1-rearranged and mutant-NPM1 AML preclinically.

    PMID 38437498DOI
    What this evidence cannot show

    Preclinical efficacy; the dual mechanism means no result in this paper attributes activity to SMARCA2 or SMARCA4 individually.

    Evidence that argues against this (1)
    • Because the compound is dual, its activity cannot be used as evidence that either single ATPase is the dependency; it is evidence about the module. Abhinav Ayyadevara VSS et al. 2025
  5. E5weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.preprint

    The tool inhibitor BRM014 and FHD-286 were both used as 'BRG1 inhibitors' in a B-ALL study, illustrating that the class is routinely described by one paralogue while inhibiting both.

    Selective Sensitivity of Ph-like B-ALL to BRG1 Inhibition Reveals a Novel Targeted Therapy Strategy. · Abhinav Ayyadevara VSS, Gaur S, Paik A, Perencsik R, et al. · bioRxiv 2025
    PMID 40672274DOI
    What this evidence cannot show

    bioRxiv preprint. Cited here only for the naming practice and the reported KMT2A-R resistance, not for any mechanistic claim.

    Evidence that argues against this (1)
    • Naming a dual inhibitor after one target is a literature convention, not a measurement; any selectivity claim needs the compound's own profiling data, which is not in the abstract. Perner F et al. 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    SMARCA2 ranks 11,581 of 17,728 for KMT2A-rearranged selectivity (difference +0.019) with pan-essentiality +0.0285, i.e. the average cell line does not require it.

    Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Oncofusion knowledge graph, analyses/depmap-kmt2a-selective-scan-full/run.py · DepMap Chronos gene effect via Open Targets GraphQL 2026
    What this evidence cannot show

    Single-gene CRISPR knockout cannot detect a dependency buffered by a paralogue; this number is uninformative about the ATPase module and is reported to make that point.

    Evidence that argues against this (1)
    • In SMARCA4-deficient solid tumours SMARCA2 is a strong dependency, so a near-zero pan-essentiality across a mixed panel conceals a large context-dependent effect. Perner F et al. 2026
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    SMARCA2 has essentially no primary literature in KMT2A-rearranged leukaemia: `SMARCA2 AND (MLL OR KMT2A)` returns 6 PubMed records (2026-09-20) and none is a primary study of SMARCA2 in this disease.

    What this evidence cannot show

    Bounded absence. Bounded to PubMed on 2026-09-20 over the symbols SMARCA2 and BRM. `BRM` is a short word-like alias and its 5 hits were not individually title-checked, so the real on-beat number is at most 5 and probably lower. The complex-level form `SWI/SNF AND (MLL OR KMT2A)` (27 records) was also run per rule 64 and adds no SMARCA2-specific study.

    Evidence that argues against this (1)
    • Complex-framed papers can contain a subunit experiment invisible to a symbol query; the FHD-286 papers are exactly that for SMARCA2, since the compound inhibits it and neither paper's abstract makes a SMARCA2-specific claim. Fiskus W et al. 2024
  8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    BRM inhibition in BRG1-deficient tumours is an established synthetic-lethal strategy, so a BRG1/BRM identifier swap inverts a therapeutic proposal rather than merely naming the wrong protein.

    Therapeutic targeting of chromatin alterations in leukemia and solid tumors. · Perner F, Berg T, Sasca D, Mersiowsky SL, et al. · Int J Cancer 2026
    PMID 40181550DOI
    What this evidence cannot show

    Review, cited as a map of the field rather than as primary evidence for the synthetic lethality itself.

    Evidence that argues against this (1)
    • This graph's own resistance entry uses FHD-286, a dual agent, so in leukaemia the synthetic-lethal framing does not apply and the two paralogues are removed together. Fiskus W et al. 2026
  9. E_q91_smarca2_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB P51531 (SMCA2_HUMAN, primary gene SMARCA2, 1,590 aa, read back live 2026-09-21) carries ONE subcellular location value, 'Nucleus', on two ECO:0000269 attributions - PubMed:11259672 and PubMed:25593309 - plus one location note, 'Localizes to sites of DNA damage' (ECO:0000269 PubMed:25593309). There is no annotated NLS and no isoform-scoped comment. The length, 1,590 aa against SMARCA4's 1,647, is the only field that separates this record from its paralogue's, as this entry's existing identifier-trap section records.

    What this evidence cannot show

    Both cited papers study something other than SMARCA2's resting compartment - one is an apoptotic-proteolysis study and one is a genome-scale bromodomain screen - so this is formally a rule-91 configuration. Unusually, opening them improved rather than degraded the annotation: see E_q91_smarca2_loc2.

    Evidence that argues against this (1)
    • This record would be read as thin and it is the opposite, which is worth stating because it cuts against the reflex this tranche was dispatched with. Rule 91 predicts that a compartment cited to an off-topic paper is over-claimed. Here the cited apoptosis paper turns out to contain an explicit nuclear-matrix FRACTIONATION of hBRM, in a leukaemia cell line, with a proteolytic perturbation - i.e. the annotation is UNDER-claimed, because UniProt recorded 'Nucleus' from a paper that supports the finer 'nuclear matrix'. Rule 91's instruction is to open the paper, and the reason to open it is not only to catch over-claims. Biggs JR et al. 2001
  10. E_q91_smarca2_loc2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In NB4 leukaemic cells, hBRM/SMARCA2 is cleaved after apoptosis induced by UV irradiation, DNA-damaging agents or staurosporine. The protease is not an effector caspase - hBRM could not be cleaved by caspase-3, -7 or -6 in vitro - but CATHEPSIN G, a granule enzyme present in NB4 cells, which cleaves hBRM in a pattern matching the in-vivo apoptotic pattern; a peptide inhibitor of cathepsin G blocks hBRM cleavage during apoptosis without blocking caspase activation or PARP cleavage. Cathepsin G is localised in GRANULES AND THE GOLGI in untreated cells and becomes DIFFUSELY DISTRIBUTED during apoptosis. Cleavage removes a 20-kDa C-terminal fragment containing a bromodomain, and this DISRUPTS THE ASSOCIATION BETWEEN hBRM AND THE NUCLEAR MATRIX: the 160-kDa fragment is less tightly nuclear-matrix-associated than full-length hBRM.

    The human brm protein is cleaved during apoptosis: the role of cathepsin G. · Biggs JR, Yang J, Gullberg U, Muchardt C et al. · Proc Natl Acad Sci U S A 2001
    PMID 11259672DOI
    What this evidence cannot show

    One cell line, NB4, whose driver is PML-RARA; nuclear matrix is an operational fractionation compartment, not an imaged structure, and this project's own vocabulary entry for it says as much. 2001, read from the abstract. Whether the same proteolysis occurs in a B-lineage blast is untested and the granule dependence predicts it does not.

    Evidence that argues against this (1)
    • The tempting synthesis - that cathepsin G supplies a myeloid-specific route to destroying a BAF ATPase and might therefore be exploited - collapses on the paralogue asymmetry this entry is built around. Quantitative proteomics shows leukaemic BAF complexes are specifically assembled around the BRG/SMARCA4 ATPase, and BRG rather than BRM is what leukaemia maintenance requires. Cleaving SMARCA2 in a myeloid blast may remove an ATPase that is barely incorporated, which would explain why nobody has pursued the observation in twenty-five years. The localisation result is solid and its therapeutic reading is blocked by the same fact that makes this entry exist. Buscarlet M et al. 2014
  11. E_q91_smarca2_loc3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    SMARCA2's damage-induced relocalisation annotation derives from a genome-scale survey rather than a dedicated experiment: the screen reports that MORE THAN ONE-THIRD OF ALL HUMAN BROMODOMAIN-CONTAINING PROTEINS CHANGE LOCALISATION IN RESPONSE TO DNA DAMAGE, and names ZMYND8 as the factor it then pursues, recruiting the NuRD complex to damaged chromatin in a transcription-associated DDR pathway. SMARCA2 is a bromodomain protein and is one of the unnamed movers in that survey.

    PMID 25593309DOI
    What this evidence cannot show

    The abstract names ZMYND8 and no other mover, so SMARCA2's individual behaviour in this screen is not readable from the abstract and the per-protein data were not retrieved. Confidence is capped at moderate for that reason. The screen is in human cells and is not haematopoietic.

    Evidence that argues against this (1)
    • This is the SAME paper that brief rule 91 names as one of the mis-citations on KMT2A's own UniProt record - and the two cases diverge, which is the useful part. KMT2A is not a bromodomain protein, so a bromodomain screen is a category error as evidence for its compartment; SMARCA2 IS one, so the same screen is exactly the right instrument for it. An automated audit that flagged 'UniProt cites PMID 25593309 for a localisation' would score these two identically and be wrong on one of them. The check rule 91 demands is whether the paper is about YOUR fact, not whether other entries have been burned by it. UniProt Consortium 2026
Connected to