Oncofusion

start / graph / gene-smarcd2

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

SMARCD2 (BAF60B) - named in one KMT2A paper in the whole of PubMed, and the dependency it carries is myeloid identity, not the rearrangement

also called BAF60B · SMRD2_HUMAN

SMARCD2 is one of three interchangeable parts that sit in the same slot of the SWI/SNF machine. Leukaemia cells with an MLL rearrangement need it - but when the test is split by cell lineage, the need turns out to belong to myeloid cells generally, not to the rearrangement.

The same thing, for a scientist

SMARCD2 (Q92925, 531 aa) is one of three mutually exclusive BAF60 paralogues (SMARCD1/2/3) present in all three mSWI/SNF assemblies [E1]. It is one of the three subunits in the anchor paper: shRNA depletion in mouse MLL-AF9 cells and human THP-1 reduces expansion, self-renewal and the undifferentiated state, human cDNA rescues the mouse knockdown, and in vivo all 21 transplanted mice survived to termination [E2]. It was independently a hit in an shRNA screen of 93 SWI/SNF-targeting hairpins in MLL-AF9/NrasG12D RN2 cells, where BAF60b is named as a requirement [E3]. **The dependency does not survive a lineage split.** In DepMap, SMARCD2 is depleted in the myeloid half of the KMT2A-rearranged arm (z -1.61) and not at all in the B-lineage half (z -0.04), a gap of -1.57 - the signature this project labels a myeloid-identity artefact [E4]. Genome-wide it ranks 510/17,728 with a small difference (-0.145) [E5]. Separately, primary human AML blasts are reported to show near-complete **loss** of SMARCD2 protein alongside SMARCB1 [E6], which is hard to reconcile with a maintenance requirement and is the sharpest counter-evidence in this beat. **`SMARCD2 AND (MLL OR KMT2A)` returns exactly one PubMed record - the anchor paper [E7].**

What it looks like
loading AF-Q92925-F1…
AlphaFold AF-Q92925-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
One paper, and what it does and does not show

Models in the anchor paper: mouse MLL-AF9-transformed spleen cells, human THP-1 (KMT2A::MLLT3) and MV4-11 (KMT2A::AFF1, myeloid). All myeloid. Controls: knockdown efficiency correlated with phenotype; human cDNA rescued mouse-specific shRNAs; knockdown of one subunit did not change the levels of the other two 2.

What it does not show: any B-lineage model, any infant model, any physical association with a KMT2A fusion, and any evidence that SMARCD2 rather than the BAF60 slot is what matters - SMARCD1/BAF60a was a hit in the parallel RN2 screen 3, and SMARCD1 is the better-ranked of the two in DepMap (rank 59 against 510) 5.

The lineage split, and why it is the headline here

Run for this beat over eight KMT2A-rearranged myeloid and five KMT2A-rearranged B-lineage DepMap lines, scored in SDs of 43 other-leukaemia lines:

halfz
KMT2A-rearranged myeloid-1.61
KMT2A-rearranged B-lineage-0.04
OCI-AML3 (NPM1c, myeloid, no KMT2A rearrangement)-3.24
U-937 (PICALM::MLLT10, myeloid)-4.64

The two myeloid non-KMT2A references are the most SMARCD2-dependent lines in the panel. That is the definition of a lineage effect: the dependency tracks myeloid identity and is stronger in cells without the rearrangement than in cells with it 4. For the infant KMT2A::AFF1 B-ALL this lab studies, the B-lineage column is the relevant one and it is flat.

Facts and figures
uniprotQ92925
gene symbolSMARCD2
protein nameSWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily D member 2
length AA531
alphafold IDAF-Q92925-F1
accession resolutionUniProt REST gene_exact:SMARCD2 AND organism_id:9606 AND reviewed:true, run 2026-09-20; one reviewed row; gene_primary and length read back from the fetched record.
complexBAF60 slot, shared by cBAF, PBAF and ncBAF
classificationmaintenance dependency in myeloid KMT2A-rearranged models; lineage artefact by DepMap split; NOT a demonstrated interactor of a KMT2A fusion
subnuclear localizationNOT MEASURED. No subnuclear compartment has been published for SMARCD2 in any system. What is published is CHROMATIN OCCUPANCY BY FUNCTION: SMARCD2-containing SWI/SNF complexes are required for CEBP-epsilon recruitment to neutrophilic secondary granule gene promoters [E_q71_ct1], which is a locus-level statement, not a compartment. [EXPLICIT NEGATIVE] UNPUBLISHED, not not-applicable. SMARCD2 is a 531-residue nuclear BAF subunit and a subnuclear pool is entirely plausible; nobody has imaged it. Nothing at all has been measured in a KMT2A-rearranged cell, which matches this entry's existing finding that SMARCD2 is named in exactly one KMT2A paper in all of PubMed.
paralogues
[
  "SMARCD1 (Q96GM5, 515 aa)",
  "SMARCD3"
]
depmap
{
  "rank": 510,
  "of": 17728,
  "difference": -0.145,
  "mean_kmt2a_r": -0.4163,
  "mean_other_leukaemia": -0.2713,
  "pan_essentiality_all_lines": -0.1633,
  "lineage_split_z_myeloid": -1.61,
  "lineage_split_z_b_lineage": -0.04,
  "z_split_gap": -1.57,
  "z_OCI_AML3_NPM1c": -3.24,
  "z_U937_PICALM_MLLT10": -4.64,
  "verdict": "MYELOID-IDENTITY ARTEFACT"
}
literature size
{
  "query": "SMARCD2 AND (MLL OR KMT2A)",
  "records": 1,
  "date": "2026-09-20",
  "tool": "PubMed E-utilities esearch"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.4163,
  "chronos_mean_other_leukaemia": -0.2713,
  "difference": -0.145,
  "pan_essentiality_all_lines": -0.1633,
  "selectivity_rank": 510,
  "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
{
  "subcellular": "NUCLEUS - AND NOTE THE EVIDENCE CODE. UniProt Q92925 annotates Nucleus at ECO:0000305, which is 'inferred by curator', NOT an experimental annotation. Every other gene in this tranche with a nuclear call has ECO:0000269. For a BAF subunit the inference is safe, but it is an inference and this entry says so rather than presenting it as a measurement.",
  "subnuclear": "NOT MEASURED. No subnuclear compartment has been published for SMARCD2 in any system. What is published is CHROMATIN OCCUPANCY BY FUNCTION: SMARCD2-containing SWI/SNF complexes are required for CEBP-epsilon recruitment to neutrophilic secondary granule gene promoters [E_q71_ct1], which is a locus-level statement, not a compartment.",
  "explicit_no": "UNPUBLISHED, not not-applicable. SMARCD2 is a 531-residue nuclear BAF subunit and a subnuclear pool is entirely plausible; nobody has imaged it. Nothing at all has been measured in a KMT2A-rearranged cell, which matches this entry's existing finding that SMARCD2 is named in exactly one KMT2A paper in all of PubMed.",
  "how_shown": "UniProt Q92925 SUBCELLULAR LOCATION comment at ECO:0000305, fetched 2026-09-21; locus-level occupancy inferred from CEBP-epsilon recruitment assays (PMID 28369034)",
  "systems": "murine granulopoiesis and human patient cells; no KMT2A-rearranged system",
  "shown_for_fusion_localisation": "no",
  "retired_symbol_note": "Brief rule 69 records that PMID 29483235 is reachable via BAF60B and invisible to SMARCD2. BAF60B is this gene's alias and both of the papers cited here are indexed under SMARCD2, but any future absence claim on this gene must be run over BAF60B as well.",
  "node": "localization-chromatin-bound-versus-soluble-fraction-kmt2a"
}
cell type
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "identifier_note": "SMARCD2 (Q92925, 531 aa) versus SMARCD1 (531 aa region, 63% identical) is a live paralogue trap, and brief rule 69 documents the adjacent SMARCA4/SMARCA2 case where the protein names are '...190A' and '...190B' and a swap inverts a therapeutic proposal. Here the papers explicitly test non-substitution, so the distinction is load-bearing: a result attributed to the wrong paralogue would be exactly backwards.",
  "expressed_in": {
    "class": 1,
    "summary": "broadly expressed; the SPECIFICITY is functional rather than expressional - SMARCD1 is present in the same cells and cannot substitute."
  },
  "required_in": {
    "class": 2,
    "claim": "In MICE, HUMANS AND ZEBRAFISH, SMARCD2 is required for GRANULOCYTE and broader MYELOID differentiation: Smarcd2-deficient mice fail to make mature neutrophils and eosinophils [E_q71_ct1], and human patients with biallelic loss have neutropenia, specific granule deficiency and myelodysplasia, with the defect at EARLY steps of myeloid-erythroid progenitor differentiation [E_q71_ct2]. This is a MYELOID IDENTITY requirement and it exists in three species with no fusion oncoprotein.",
    "pmids": [
      "28369034",
      "28369036"
    ],
    "system_named_in_the_claim": "murine, human and zebrafish normal myelopoiesis - no fusion present",
    "nodes": [
      "cell-type-gmp",
      "cell-type-cmp"
    ]
  },
  "the_selectivity_and_what_it_does_not_show": {
    "claim": "Rank 510 of 17,728: Chronos -0.4163 in KMT2A-rearranged against -0.2713 in other leukaemias, difference -0.1450, pan-essentiality -0.1633. Selective half, no floor compression, but a small absolute effect.",
    "but": "The DepMap KMT2A-rearranged arm is myeloid-enriched (roughly 7 myeloid to 4 lymphoid, per this project's cycle-41 finding), and SMARCD2's documented requirement IS myeloid identity. A myeloid-identity gene scoring selective in a myeloid-enriched index arm is the CEBPA/SPI1 artefact this graph already named, and the difference should not be read as fusion selectivity until the arm is split. This entry's own title already says the dependency is myeloid identity, not the rearrangement; the localisation and cell-type record added here supports that reading rather than qualifying it."
  },
  "context": "murine, human and zebrafish normal myelopoiesis; this project's DepMap scan for the selectivity"
}
cell state
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "maintains_state": {
    "class": 3,
    "states": [
      "cell-state-differentiation-block"
    ],
    "claim": "In MICE AND HUMANS, losing SMARCD2 IMPOSES a myeloid differentiation block - failure to generate mature neutrophils and eosinophils, and myelodysplasia with excess blasts in patients [E_q71_ct1, E_q71_ct2]. The direction is the opposite of the usual leukaemia framing: here the differentiation block is the consequence of LOSING the gene, so SMARCD2 is a gene whose presence RELIEVES the block rather than one that sustains it.",
    "pmids": [
      "28369034",
      "28369036"
    ],
    "system_named_in_the_claim": "murine and human normal myelopoiesis, plus an AML promyelocytic line for the chromatin changes"
  },
  "confidence": "strong",
  "caveats": "Nothing here is KMT2A-rearranged, B-lineage or infant. The differentiation-block direction is inferred from loss-of-function phenotypes in normal myelopoiesis; no KMT2A-rearranged cell has been tested. And the DepMap tilt is confounded by the index arm's myeloid enrichment.",
  "counter_evidence": {
    "note": "The authors of the human study conclude SMARCD2 is a potential TUMOUR SUPPRESSOR in leukaemia, on the strength of patients whose biallelic loss produces myelodysplasia with excess blast cells. If that is right, the selectivity number in this entry is describing a gene whose loss is oncogenic, and reading rank 510 as a therapeutic opportunity would invert the direction.",
    "citation": {
      "title": "Chromatin-remodeling factor SMARCD2 regulates transcriptional networks controlling differentiation of neutrophil granulocytes.",
      "authors": "Witzel M, Petersheim D, Fan Y, Bahrami E, Racek T, Rohlfs M, Puchałka J, Mertes C, Gagneur J, Ziegenhain C, Enard W, Stray-Pedersen A, Arkwright PD, Abboud MR, Pazhakh V, Lieschke GJ, Krawitz PM, Dahlhoff M, Schneider MR, Wolf E, Horny HP, Schmidt H, Schäffer AA, Klein C",
      "year": 2017,
      "venue": "Nat Genet",
      "pmid": "28369036",
      "doi": "10.1038/ng.3833",
      "type": "paper"
    }
  }
}
depmap release backfill
{
  "open_targets_dataversion": "26.06",
  "note": "Added by Q71 per brief rule 83; Open Targets dataVersion fetched live 2026-09-21 as year 26 month 06. Scan denominator for this gene: n_all_lines 1208."
}
How this could bite us back
  • No SMARCD2-directed agent exists, so any intervention would have to go through the shared ATPase and would carry that module's whole liability profile.
  • If the requirement is really for an occupied BAF60 slot, removing SMARCD2 invites SMARCD1 or SMARCD3 substitution - a resistance route with no assay currently watching for it.
  • The reported loss of SMARCD2 in primary AML blasts means a patient population could be selected for the drug target being absent.
  • SMARCD2 loss in humans causes neutropenia and specific granule deficiency (PMID 28369036) - a clinically serious, infection-prone phenotype in exactly the compartment a leukaemia patient cannot spare. Inhibiting it would reproduce a known human immunodeficiency.
  • The paralogue will not rescue. SMARCD1 is 63% identical, present in the same cells, and explicitly cannot substitute for SMARCD2 in granulocyte development (PMID 28369034). The usual 'a paralogue buffers the toxicity' argument is unavailable here and the evidence points the other way.
  • The human genetics say tumour suppressor. Patients with biallelic SMARCD2 loss get myelodysplasia with excess blasts, and the authors call SMARCD2 a potential tumour suppressor in leukaemia. A drug that removes it in a child with leukaemia risks selecting for the disease it treats.
Still open
  1. Do the three BAF60 paralogues substitute for each other in leukaemia, so that the anchor paper's SMARCD2 phenotype is really a BAF60-slot phenotype? SMARCD1 and SMARCD2 were both screen hits in RN2 and neither has been tested against the other.
  2. Is SMARCD2 protein present in primary KMT2A-rearranged blasts at all? One report finds near-complete loss in primary AML while the maintenance data come from cell lines.
  3. Does SMARCD2 matter in infant KMT2A::AFF1 B-ALL? No B-lineage or infant experiment exists, and the DepMap B-lineage column is flat.
  4. Does SMARCD2 have any subnuclear distribution at all? UniProt's nuclear call for this gene is ECO:0000305 - a curator inference - and no imaging exists, which is unusual for a BAF subunit in a graph with 47 localization entries.
  5. Split the DepMap KMT2A-rearranged arm by lineage for SMARCD2 specifically. If the -0.145 difference lives entirely in the myeloid lines, the number is a lineage readout and this entry's title is confirmed by the scan rather than merely consistent with it.
Every source behind this entry (9)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    SMARCD2 is UniProt Q92925, 531 aa, primary gene name SMARCD2, synonym BAF60B; it is one of three BAF60 paralogues.

    UniProtKB Q92925 (SMRD2_HUMAN) · UniProt Consortium · UniProtKB 2026
    What this evidence cannot show

    gene_exact:SMARCD2 AND organism_id:9606 AND reviewed:true, run 2026-09-20; one reviewed row; gene_primary read back as SMARCD2 and length 531 from the same record.

    Evidence that argues against this (1)
    • SMARCD1 (Q96GM5, 515 aa) and SMARCD2 differ by 16 residues, the same order as the SMC1A/SMC3 collision rule 66 calls the limit of the length discriminator. Assert the accession, not the size. UniProt Consortium 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    shRNA depletion of Smarcd2 reduces expansion, clonal expansion and self-renewal of mouse MLL-AF9 cells and of human THP-1, causes apoptosis, S-phase loss, c-Kit loss and Mac1 gain, and extends survival in transplanted mice; human SMARCD2 cDNA rescues the mouse knockdown.

    SWI/SNF Subunits SMARCA4, SMARCD2 and DPF2 Collaborate in MLL-Rearranged Leukaemia Maintenance. · Cruickshank VA, Sroczynska P, Sankar A, Miyagi S, et al. · PLoS One 2015
    PMID 26571505DOI
    What this evidence cannot show

    shRNA only; all models myeloid; the in vivo readout is survival of 21 mice to experiment termination rather than a quantified leukaemia burden. Read in full via PMC4646637.

    Evidence that argues against this (1)
    • The same knockdowns reduced LSK colony number and LSK proliferation, so the leukaemia-versus-normal window claimed from c-Kit-enriched bone marrow is not clean. Xiu Y et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In an shRNA screen of 93 hairpins covering SWI/SNF subunits in MLL-AF9/NrasG12D RN2 cells, BAF60b (SMARCD2) was identified as a requirement for proliferation, alongside BAF53a, BAF250a, BAF60a, BAF155 and BAF45d.

    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

    Mouse line, competition assay, criterion was multiple independent hairpins reducing GFP+ more than threefold over 10 days with RT-qPCR-validated knockdown. BAF60c, BAF45b, BAF45c and BAF53b were excluded as not expressed, so the screen is not complete over the complex.

    Evidence that argues against this (1)
    • A proliferation screen in one mouse line cannot distinguish a subunit-specific requirement from a requirement for an intact complex; every hit is consistent with destabilising the assembly. Mashtalir N et al. 2018
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Split by lineage, the SMARCD2 dependency is present only in the myeloid half of the KMT2A-rearranged arm (z -1.61 myeloid, -0.04 B-lineage), and is stronger still in two myeloid lines without a KMT2A rearrangement (OCI-AML3 -3.24, U-937 -4.64).

    SWI/SNF panel lineage-split DepMap dependency scan, 28 genes · Oncofusion knowledge graph, analyses/swi-snf-lineage-split/run.py · DepMap Chronos gene effect via Open Targets GraphQL 2026
    What this evidence cannot show

    Eight myeloid and five B-lineage index lines; one line moves a mean. Split is by lineage label. z is in SDs of the comparator arm and is not an effect size - SMARCD2's genome-wide raw difference is only -0.145. U-937 carries PICALM::MLLT10, a fusion that converges on the same elongation biology, so calling it a non-KMT2A reference is a simplification.

    Evidence that argues against this (1)
    • DepMap contains no infant-derived KMT2A-rearranged B-ALL line, so a flat B-lineage column is a statement about five adolescent/adult lines, not about infant disease. Oncofusion knowledge graph et al. 2026
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Genome-wide, SMARCD2 ranks 510 of 17,728 with difference -0.145 (KMT2A-r -0.4163, other leukaemia -0.2713, pan-essentiality -0.1633).

    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

    11 index against 43 comparator lines; Cellosaurus annotation may be incomplete; a rank in the top 3% with a raw difference of -0.145 is a narrowness artefact of the kind this project has documented - report the difference, not only the rank.

    Evidence that argues against this (1)
    • SMARCD1 outranks it at 59 with a larger difference (-0.3574) and survives the lineage split, so if a BAF60 slot matters in this disease the evidence points at SMARCD1 rather than SMARCD2. Oncofusion knowledge graph et al. 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Primary human AML cells are reported to exhibit near-complete loss of SMARCB1 (BAF47) and SMARCD2 (BAF60B), with the residual complex nucleating around SMARCC1 and colocalising with H3K27ac at oncogenic loci.

    PMID 29483235DOI
    What this evidence cannot show

    Primary AML blasts, not KMT2A-rearranged-selected; 'near complete loss' is a protein-level observation in that cohort and is not a general property of AML. The paper's mechanistic focus is SMARCB1; SMARCD2 is reported alongside it without its own loss-of-function follow-up.

    Evidence that argues against this (1)
    • The anchor paper requires SMARCD2 for maintenance in THP-1 and MV4-11, which are AML lines. Both cannot be generally true; the most likely reconciliation is cohort or antibody heterogeneity, and it is unresolved. Cruickshank VA et al. 2015
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    SMARCD2 has exactly one PubMed record in this disease: `SMARCD2 AND (MLL OR KMT2A)` returns 1 (PMID 26571505, the anchor).

    What this evidence cannot show

    Bounded absence. Bounded to PubMed on 2026-09-20 over SMARCD2 and BAF60B. The alias run per the retired-symbol rule recovered four extra records, one of which (29483235) is the loss-in-primary-AML paper cited as E6 and is unreachable from the current symbol - so the current-symbol count alone would have hidden the strongest counter-evidence in this entry.

    Evidence that argues against this (1)
    • Rule 64 applies: a complex-framed paper can contain a SMARCD2 experiment with no symbol in the abstract. The complex form was run and title-checked; it adds no further SMARCD2-specific study, but that is a bound on abstracts, not on figures. Shi J et al. 2013
  8. E_q71_ct1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In MICE, granulocyte development requires SMARCD2: Smarcd2-deficient mice fail to generate functionally mature NEUTROPHILS and EOSINOPHILS, a phenotype reminiscent of human neutrophil-specific granule deficiency; SMARCD2-containing SWI/SNF complexes are necessary for CEBP-epsilon recruitment to the promoters of neutrophilic secondary granule genes, the 63%-identical paralogue SMARCD1 CANNOT substitute, and the functional specificity maps to SMARCD2's divergent coiled-coil 1 and SWIB domains.

    SMARCD2 subunit of SWI/SNF chromatin-remodeling complexes mediates granulopoiesis through a CEBPɛ dependent mechanism. · Priam P, Krasteva V, Rousseau P, D'Angelo G, Gaboury L, Sauvageau G, Lessard JA · Nat Genet 2017
    PMID 28369034DOI
    What this evidence cannot show

    Abstract read in full; full text not read. MURINE normal granulopoiesis - not leukaemia, not human, not KMT2A-rearranged, and myeloid rather than the B lineage this graph exists for (brief rule 73). The paralogue-non-substitution result is the strongest part and it is a specificity claim about a subunit, not about a complex.

    Evidence that argues against this (1)
    • An independent group reaching the same gene from human patients reports the opposite therapeutic direction. Three unrelated pedigrees with neutropenia, specific granule deficiency and myelodysplasia with excess blasts carry homozygous loss-of-function SMARCD2 mutations, and the authors conclude SMARCD2 is a POTENTIAL TUMOUR SUPPRESSOR in leukaemia. A tumour suppressor is not a dependency, and this entry must not be read as proposing SMARCD2 inhibition. Witzel M et al. 2017
  9. E_q71_ct2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In HUMANS, MICE AND ZEBRAFISH, SMARCD2 (BAF60b) is a critical regulator of MYELOID DIFFERENTIATION: patients from three unrelated pedigrees with neutropenia, specific granule deficiency, myelodysplasia with excess of blast cells and various developmental aberrations carry homozygous loss-of-function SMARCD2 mutations; SMARCD2 controls EARLY steps in the differentiation of myeloid-erythroid progenitor cells, interacts with CEBP-epsilon, and defective SMARCD2 produces transcriptional and chromatin changes in AML human promyelocytic cells - the authors' conclusion being that SMARCD2 is a potential TUMOUR SUPPRESSOR in leukaemia.

    Chromatin-remodeling factor SMARCD2 regulates transcriptional networks controlling differentiation of neutrophil granulocytes. · Witzel M, Petersheim D, Fan Y, Bahrami E, Racek T, Rohlfs M, Puchałka J, Mertes C, Gagneur J, Ziegenhain C, Enard W, Stray-Pedersen A, Arkwright PD, Abboud MR, Pazhakh V, Lieschke GJ, Krawitz PM, Dahlhoff M, Schneider MR, Wolf E, Horny HP, Schmidt H, Schäffer AA, Klein C · Nat Genet 2017
    PMID 28369036DOI
    What this evidence cannot show

    Abstract read in full; full text not read. Human patients plus mouse and zebrafish models of NORMAL myelopoiesis, and an AML promyelocytic line - not KMT2A-rearranged, not B-lineage, not infant (brief rule 73). 'Early steps in myeloid-erythroid progenitor differentiation' is a compartment statement from flow cytometry and expression, not a single-cell fate map. The tumour-suppressor reading is the authors' interpretation, not a tested genotype.

    Evidence that argues against this (1)
    • The paralogue question cuts the other way for a drug. The companion murine study shows SMARCD1, 63% identical, cannot replace SMARCD2 in granulocyte development, and that the specificity sits in the divergent coiled-coil 1 and SWIB domains. So the two paralogues are NOT redundant, which means a pan-SMARCD tool would remove a non-redundant differentiation function rather than being buffered by the paralogue - the opposite of the usual paralogue-redundancy assumption. Priam P et al. 2017
Connected to