Oncofusion

start / graph / gene-brd9

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q5-interactome4

BRD9 — dispensable in the average cell, rank 39 of 17,728 in KMT2A-rearranged lines, and with an on-target proof most epigenetic targets never get

also called BRD9 · Bromodomain-containing protein 9 · LAVS3040 · PRO9856

BRD9 is a reader protein inside one flavour of the SWI/SNF chromatin-remodelling machine. Most cells do not need it at all, but KMT2A-rearranged leukaemia cells do, and there are inhibitors plus an unusually rigorous proof that those inhibitors work by hitting BRD9 and nothing else.

The same thing, for a scientist

BRD9 (Q9H8M2, 597 aa) is the bromodomain subunit (aa 136-240) of the non-canonical BAF (ncBAF) SWI/SNF complex, with a histone H4K5ac/H4K8ac-binding region at aa 214-216 [E1]. AML cells require BRD9 to sustain **MYC transcription**, rapid proliferation and a block in differentiation; small-molecule BRD9 bromodomain inhibitors selectively suppress mouse and human AML line proliferation; and the on-target claim was established by engineering a **bromodomain-swap allele of BRD9** that retains function despite a radically altered pocket and confers resistance to the compound series, which the authors describe as the first evidence for a role of BRD9 in cancer [E2]. The dependency profile is the best-shaped in this beat: BRD9 ranks **39 of 17,728 (0.220th percentile)** with a difference of **-0.3896**, KMT2A-rearranged mean **-0.4793** against other-leukaemia **-0.0897**, and a pan-essentiality of **+0.0348** — positive, meaning the average cell line does not need it at all [E3]. **Classification: dependency, not a demonstrated interactor**; no published experiment places BRD9 on a KMT2A fusion protein, and the literature is thin (`(MLL OR KMT2A) AND BRD9` returns 3 records) [E4].

What it looks like
Bromodomain-containing protein 9597 aa
BRD (bromodomain)BRD (bromodomain) · residues 136–2400200400
loading AF-Q9H8M2-F1…
AlphaFold AF-Q9H8M2-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
Why the numbers here are unusual

Almost everything in this beat fails one of two tests: it is pan-essential, so there is no window; or it has no KMT2A selectivity, so there is no rationale. BRD9 passes both 3:

BRD9MEN1EP300SUPT5H
Pan-essentiality (1,208 lines)+0.035-0.363-0.359-1.507
KMT2A-rearranged mean-0.479-1.263-0.997-1.575
Other-leukaemia mean-0.090-0.397-0.713-1.454
Selectivity difference-0.390-0.866-0.284-0.122
Rank391108767

A gene with a positive mean effect across 1,208 lines, and a -0.48 mean in the rearranged group, is describing a requirement that appears only in this context. That is the shape a therapeutic index is made of, and the shape MEN1 has. BRD9's absolute requirement is much shallower than MEN1's, which is the honest limitation.

Added to that, the chemistry has an on-target proof that most epigenetic probes lack: a bromodomain-swap allele that keeps BRD9 functional while abolishing compound binding, conferring resistance and thereby proving the antiproliferative effect is BRD9-mediated 2. That design is stronger evidence of target engagement than a selectivity panel.

What is missing

Three things, and they are large.

No KMT2A-specific experiment. Hohmann et al. worked in AML broadly and attributed the phenotype to MYC transcription; the KMT2A-rearranged selectivity in this entry comes entirely from the DepMap scan 23. (MLL OR KMT2A) AND BRD9 returns 3 records 4, and none was read here as a functional KMT2A study.

No B-lineage or paediatric data. As with every other entry in this beat.

No interaction evidence. BRD9 is filed as a dependency. The mechanistic link that would make it an interactome member — ncBAF being recruited by, or competing with, the fusion at its target genes — is unexamined in anything cited here. The graph carries no gene-smarca4 or gene-arid1a entry either, so ncBAF is represented in this beat by a single subunit, and complex-level reasoning about it would be premature for exactly the reasons the PRC1.1 entries set out.

Cell type, cell state, localisation, epigenetic readout
AxisDocumentedWhere
Cell typeMouse and human AML cell lines2
Cell stateBlock in differentiation sustained by BRD9; loss releases it and stops rapid proliferation2
SubcellularNucleus1
SubnuclearNot established-
Epigenetic readoutReads H4K5ac and H4K8ac (and the corresponding butyryl marks) through the bromodomain at aa 136-240, with the binding region mapped to aa 214-216; downstream readout is MYC transcription12
LineageMyeloid. No B-lineage, no KMT2A::AFF1.2

The MYC connection is worth putting beside gene-pcgf1: BRD9 loss lowers MYC and kills KMT2A-rearranged cells, while PRC1.1 loss raises MYC and makes them menin-inhibitor-resistant 2 and [PRC1.1 entries]. Two independent nodes in this beat converge on MYC from opposite directions, which makes MYC the single most-implicated output of the whole cycle.

Facts and figures
uniprotQ9H8M2
gene symbolBRD9
protein nameBromodomain-containing protein 9
length AA597
alphafold IDAF-Q9H8M2-F1
accession resolutionUniProt REST query gene_exact:BRD9 AND organism_id:9606 AND reviewed:true, run 2026-09-19; one reviewed hit; primary gene name read back as BRD9 and length 597 confirmed from the same fetched record.
complexncBAF (non-canonical BAF) SWI/SNF
domains noteUniProt also annotates a three-residue H4K5ac/H4K8ac binding region at aa 214-216; it is inside the bromodomain and is not given its own lane.
classificationdependency and tractable drug target; NOT a demonstrated interactor of KMT2A::AFF1
classification reasonno published experiment places BRD9 on a KMT2A fusion protein; the KMT2A selectivity comes from an unbiased dependency scan and the mechanism reported is MYC transcription
interaction evidence classsmall-molecule inhibition with an engineered bromodomain-swap resistance allele; unbiased DepMap scan
shown for fusionnone; AML broadly
epigenetic readoutreads H4K5ac/H4K8ac (and butyryl equivalents); downstream readout is MYC transcription and a differentiation block
on target validationbromodomain-swap allele of BRD9 retaining function with an altered pocket confers resistance to the compound series - stronger evidence of target engagement than a selectivity panel (PMID 27376689)
converges onMYC - from the opposite direction to PRC1.1, where subunit loss RAISES MYC and confers menin-inhibitor resistance
domains
[
  {
    "name": "BRD (bromodomain)",
    "start": 136,
    "end": 240,
    "source": "UniProt Q9H8M2, fetched 2026-09-19; the target of the Hohmann compound series and the site of the bromodomain-swap resistance allele"
  }
]
localization
{
  "subcellular": "nucleus",
  "subnuclear": "not established"
}
essentiality
{
  "depmap_scan": "analyses/depmap-kmt2a-selective-scan-full",
  "pan_essentiality_all_lines": 0.0348,
  "n_all_lines": 1208,
  "mean_kmt2a_rearranged": -0.4793,
  "mean_other_leukemia": -0.0897,
  "kmt2a_selectivity_difference": -0.3896,
  "rank": 39,
  "n_scanned": 17728,
  "percentile": 0.22,
  "comparators": {
    "SMARCA4": {
      "difference": 0.0791,
      "rank": 16333
    },
    "ARID1A": {
      "difference": 0.0276,
      "rank": 12593
    }
  },
  "interpretation": "positive pan-essentiality with a -0.48 requirement in rearranged lines - the cleanest window shape in this beat, though the absolute requirement is shallow"
}
enumeration record counts
[
  {
    "query": "(MLL OR KMT2A) AND BRD9",
    "records": 3
  },
  {
    "query": "BRD9 bromodomain acute myeloid leukemia",
    "records": 18
  },
  {
    "date": "2026-09-19"
  }
]
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.4793,
  "chronos_mean_other_leukaemia": -0.0897,
  "difference": -0.3896,
  "pan_essentiality_all_lines": 0.0348,
  "selectivity_rank": 39,
  "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."
}
How this could bite us back
  • The whole KMT2A rationale for BRD9 rests on one computational scan over eleven AML-heavy, non-infant cell lines. No wet experiment has tested it in this genotype, and the founding paper did not look at KMT2A rearrangement at all.
  • A positive pan-essentiality is encouraging for tolerability and discouraging for potency: the absolute requirement in rearranged lines is -0.48, a fraction of MEN1's -1.26, so single-agent activity is unlikely to be deep.
  • MYC is the reported output, and lowering MYC is a mechanism shared with several agents; resistance by MYC restoration through any other route would be expected.
  • BRD9 sits in ncBAF, and SWI/SNF perturbation has context-dependent and sometimes tumour-promoting effects. In a B-lineage disease with no data, that risk is unquantified.
  • The compounds are chemical probes from 2016 with no in vivo efficacy reported in the abstract read here. Treating BRD9 as clinically tractable today would over-read the chemistry.
Still open
  1. Does BRD9 inhibition or degradation work in a KMT2A-rearranged model specifically, and in KMT2A::AFF1 B-ALL? The dependency signal is strong and entirely computational; no wet experiment in this genotype has been published.
  2. Is the requirement BRD9-specific or ncBAF-specific? SMARCA4 and ARID1A score on the opposite side, and the ncBAF-defining subunits BICRA and BICRAL were not examined in this beat.
  3. BRD9 loss lowers MYC and PRC1.1 loss raises it. Would BRD9 inhibition resensitise a PRC1.1-deficient, menin-inhibitor-resistant cell? That is a directly testable combination and it follows from two independent entries in this cycle.
  4. Does BRD9 bromodomain inhibition and BRD9 degradation give the same phenotype? For several epigenetic targets they do not.
  5. Is ncBAF recruited by, or in competition with, the KMT2A fusion at its target genes?
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q9H8M2 (BRD9_HUMAN, Bromodomain-containing protein 9) is 597 aa and annotates a bromodomain at aa 136-240 and a region at aa 214-216 that binds histone H4K5ac and H4K8ac and the corresponding H4K5bu and H4K8bu butyryl marks.

    UniProtKB Q9H8M2 (BRD9_HUMAN), Bromodomain-containing protein 9 · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
    What this evidence cannot show

    Record fetched 2026-09-19; primary gene name read back as BRD9 and length 597 confirmed from the same fetched record, and both coordinates copied from it. A three-residue 'binding region' is a curator summary of contact residues and not a domain boundary; it should not be used as a lane in a domain figure.

    Evidence that argues against this (1)
    • The acetyl-lysine reading function is not obviously what matters therapeutically. BRD9 also acts as a structural ncBAF subunit, and degrader work in other systems has shown bromodomain inhibition and protein degradation give different phenotypes - the same genetic-versus-chemical divergence this graph documented for USP22. A bromodomain map does not tell you which activity the AML phenotype needs. Hohmann AF et al. 2016
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Acute myeloid leukaemia cells require the BRD9 subunit of the SWI-SNF chromatin-remodelling complex to sustain MYC transcription, rapid cell proliferation and a block in differentiation; small-molecule inhibitors of the BRD9 bromodomain selectively suppress the proliferation of mouse and human AML cell lines; and a bromodomain-swap allele of BRD9 that retains functionality despite a radically altered bromodomain pocket confers resistance to the antiproliferative effects of the compound series, establishing BRD9 as the relevant cellular target. The authors state this is the first evidence for a role of BRD9 in cancer.

    Sensitivity and engineered resistance of myeloid leukemia cells to BRD9 inhibition · Hohmann AF, Martin LJ, Minder JL, Roe JS, Shi J, Steurer S, Bader G, McConnell D, Pearson M, Gerstberger T, Gottschamel T, Thompson D, Suzuki Y, Koegl M, Vakoc CR · Nature Chemical Biology 12(9):672-9 2016
    PMID 27376689DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. **AML broadly, not KMT2A-rearranged specifically** - no fusion genotype is named in the abstract, so the KMT2A selectivity in this entry is supplied by the DepMap scan and not by this paper. Mouse and human cell lines only; no in vivo efficacy and no primary patient material in the abstract. The compounds are chemical probes, not clinical candidates.

    Evidence that argues against this (1)
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    In the project's genome-wide KMT2A-rearranged selectivity scan of 17,728 genes, BRD9 ranks 39 (0.220th percentile) with a difference of -0.3896 (KMT2A-rearranged mean -0.4793 against other-leukaemia mean -0.0897) and a pan-essentiality of +0.0348 across 1,208 lines, i.e. dispensable on average. Its ncBAF-adjacent SWI/SNF partners score in the opposite direction: SMARCA4 +0.0791 (rank 16,333, pan -0.3636) and ARID1A +0.0276 (rank 12,593, pan -0.2337).

    Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
    What this evidence cannot show

    All rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19. Eleven KMT2A-rearranged against 43 other leukaemia lines, AML-heavy, none infant-derived. Not permutation-tested for BRD9. Note that ARID1A is a canonical-BAF subunit rather than ncBAF, so its score is not a clean complex-level comparison.

    Evidence that argues against this (1)
    • BRD9 moves and the other SWI/SNF subunits in the scan do not - SMARCA4 and ARID1A are both on the anti-dependency side. Either the requirement is BRD9-specific rather than ncBAF-specific, or the scan is reading one gene's guide behaviour. Without an ncBAF-resolved subunit set (BRD9, BICRA, BICRAL, SMARCA4) the complex-level interpretation this entry gestures at is unsupported, and the beat did not write those entries. Hohmann AF et al. 2016
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    PubMed returns 3 records for '(MLL OR KMT2A) AND BRD9' and 18 for 'BRD9 bromodomain acute myeloid leukemia' as of 2026-09-19, so BRD9 has a modest AML literature and essentially no KMT2A-rearranged-specific literature.

    What this evidence cannot show

    BRD9 has no widely used retired symbol, so the alias-recall problem that affects other entries in this beat is smaller here; the UniProt-listed alternatives LAVS3040 and PRO9856 are clone identifiers that do not appear in the literature. The 3 records were not read, so 'essentially no KMT2A-specific literature' rests on the count plus the absence of BRD9 from the KMT2A-rearranged reviews this beat consulted.

    Evidence that argues against this (1)
Connected to