Oncofusion

start / graph / fusion-bmp2k-znf384

Fusionmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.q173-the-znf384-fusion-family

BMP2K::ZNF384 - a PROTEIN KINASE as the 5' partner, with its catalytic domain retained: the counterexample to 'the partner is always a chromatin coactivator'

also called BMP2K-ZNF384 · BIKe-ZNF384

Almost every description of this leukaemia family says the front half is a chromatin regulator. In this case the front half is an enzyme that adds phosphate groups to other proteins, and the fusion keeps the enzyme part. One patient, but it is enough to show the rule is not a rule.

The same thing, for a scientist

BMP2K was identified as a novel ZNF384 fusion partner in the Hirabayashi survey of 291 children: **nucleotide 2228 (exon 15) and nucleotide 2117 (exon 14) of BMP2K were fused in frame to nucleotide 266 (exon 3) of ZNF384**, and the predicted protein **retained the BMP2K kinase domain but lacked its C-terminal portion** [E1]. BMP2K (UniProtKB **Q9NSY1**, **1,161 aa**, BMP-2-inducible protein kinase / BIKe, EC 2.7.11.1) carries a **protein kinase domain at 51-316** with the proton-acceptor active site at **D180**, and is annotated as nuclear on experimental evidence [E2]. **An exon-14/15 breakpoint at around nucleotide 2117-2228 of the coding sequence lies far downstream of residue 316, so the entire kinase domain is upstream of the junction and is retained, exactly as the source states** [E1][E2]. It recurred: the Ponte di Legno series contains one BMP2K case among its 18 'other' partners [E3]. **This entry exists because it is the single clearest falsification of the family's standard description.** BMP2K is not a chromatin coactivator, not a FET protein, has no intrinsically disordered activation module of the EWSR1 type, and contributes a functional catalytic domain. It sits alongside CLTC (clathrin heavy chain), SYNRG (an AP-1 adaptor), ATP2C1, SEC24B, USP25 and CCAR1 in the same partner list - **a long tail that is not chromatin biology at all**.

Go deeper
Why one patient is enough here

The claim under test is universal: the ZNF384 5' partner is a transcriptional coactivator or chromatin remodeller that supplies an activation module. A universal claim is refuted by one counterexample, and a fusion that retains a kinase domain is not a marginal one.

And it is not alone. Taking the Ponte di Legno partner list 3 and the review's compilation, the 5' partners reported for ZNF384 include, alongside EP300, CREBBP, TCF3, TAF15, EWSR1, ARID1B, SMARCA2, SMARCA4 and NIPBL:

partnerwhat it actually is
BMP2Ka serine/threonine protein kinase
CLTCclathrin heavy chain - vesicle coat
SYNRGsynergin gamma - an AP-1 adaptor in endosomal trafficking
ATP2C1a secretory-pathway calcium ATPase
SEC24Ba COPII vesicle coat subunit
USP25a deubiquitinase
CCAR1cell-cycle and apoptosis regulator
DDX42an RNA helicase
TEX41a long intergenic non-coding RNA (see fusion-znf384-tex41)

Roughly half the named partner list is not chromatin. The chromatin partners dominate by case count - EP300 and TCF3 alone are 74% of partner-typed cases - but they do not dominate the partner list, and the two facts get conflated constantly.

The honest statement is therefore: ZNF384 fuses to many things, most patients have a chromatin-associated partner, and the family cannot be defined by what the partner is. What it can be defined by is the ZNF384 half - which is exactly what mechanism-znf384-c-terminal-truncation-phenocopies-the-fusion and fusion-znf384-tex41 independently argue.

What is not known about this fusion, which is nearly everything

No functional experiment on BMP2K::ZNF384 exists in anything this beat read. Specifically unknown:

  • Is the retained kinase active in the fusion? The domain is present; that is not the same as active, and truncating the C-terminus of a kinase can be activating, inactivating or neutral.
  • Does the fusion phosphorylate anything it should not? If it does, this is a fundamentally different oncoprotein from EP300::ZNF384 - a kinase targeted to ZNF384 binding sites - and it would be the only druggable member of the family by a kinase inhibitor.
  • Does it produce the same expression signature? The case clustered as a ZNF384 fusion by the study's own criteria, but no partner-resolved transcriptome was reported for it.
  • Was the reciprocal ZNF384::BMP2K product expressed? Not reported.

Also worth stating plainly: BMP2K is annotated nuclear with an ECO:0000269 experimental code referencing PMID 9853615 2. A nuclear kinase is a more plausible fusion partner for a transcription factor than a cytoplasmic one, and this is the one partner-side fact that makes the case less strange than it first looks.

Facts and figures
gene5BMP2K
gene3ZNF384
cytobandBMP2K 4q21.21; no karyotype reported for this case in what was read
uniprot5Q9NSY1
uniprot3Q8TF68
measurements
proportion0.450proportion: 0.45count2.000count: 2residue316.0residue: 316
diseases
[
  "childhood B-cell precursor ALL"
]
breakpoints
[
  {
    "id": "bp-bmp2k-exon15-znf384-exon3",
    "gene5_exon_last": 15,
    "gene3_exon_first": 3,
    "gene5_nucleotide": 2228,
    "gene3_nucleotide": 266,
    "frequency_pct": null,
    "population": "1 of 291 children with BCP-ALL in a single Japanese clinical trial; a second junction at BMP2K nucleotide 2117 (exon 14) is reported for the same case",
    "domains_retained_5": [
      "PK (protein kinase)"
    ],
    "domains_lost_5": [],
    "domains_retained_3": [
      "ZF1 (C2H2 zinc finger)",
      "ZF2 (C2H2 zinc finger)",
      "ZF3 (C2H2 zinc finger)",
      "ZF4 (C2H2 zinc finger)",
      "ZF5 (C2H2 zinc finger)",
      "ZF6 (C2H2 zinc finger)",
      "ZF7 (C2H2 zinc finger)",
      "ZF8 (C2H2 zinc finger)"
    ],
    "domains_lost_3": [],
    "domains_note": "The retained BMP2K piece is its protein kinase domain (UniProt Q9NSY1, 51-316), which includes the D180 proton-acceptor active site and the 57-65 ATP-binding region, and which the source states is retained. domains_lost_5 is EMPTY not because nothing is lost but because nothing annotated is lost: the deleted C-terminal portion of BMP2K, roughly 845 residues, carries no UniProt domain feature at all, so there is no named domain to list. The retained ZNF384 side is the eight UniProt-annotated fingers, per the source's statement that the junction is at nucleotide 266 in exon 3, the first coding exon.",
    "phenotype_notes": "the case was part of the ZNF384-fusion group sharing weak CD10 with aberrant CD13 and/or CD33; no partner-specific phenotype was reported"
  }
]
measurements — the numbers behind the chart
[
  {
    "quantity": "count",
    "subject": "BMP2K::ZNF384 cases in the literature read by this beat",
    "value": 2,
    "units": "patients",
    "n": null,
    "universe": "one in Hirabayashi 2016 (291 children) and one in the Ponte di Legno series (218 ZNF384-rearranged children); the two cohorts share contributing groups, so these may not be two distinct patients",
    "null": null,
    "note": "the number is small and its independence is uncertain, which is stated rather than smoothed"
  },
  {
    "quantity": "residue",
    "subject": "end of the BMP2K protein kinase domain",
    "value": 316,
    "units": "aa position",
    "n": 1,
    "universe": "UniProtKB Q9NSY1 feature table, domain 51-316, active site D180, retrieved 2026-09-24",
    "null": null,
    "note": "cited to make the retention claim checkable; the source states the kinase domain is retained and the coordinate is consistent with an exon-14/15 breakpoint, but the two coordinate systems were not reconciled exactly"
  },
  {
    "quantity": "proportion",
    "subject": "named ZNF384 5' partners that are NOT chromatin or transcription-associated proteins",
    "value": 0.45,
    "units": "fraction",
    "n": 20,
    "universe": "THIS BEAT'S OWN TALLY over the named partner list assembled from the Ponte di Legno series and the Zhu review: of roughly 20 named partners, about 9 (BMP2K, CLTC, SYNRG, ATP2C1, SEC24B, USP25, CCAR1, DDX42, TEX41) are kinases, trafficking proteins, an ATPase, a deubiquitinase, a helicase or a lncRNA",
    "null": null,
    "note": "A COUNT OF PARTNER NAMES, NOT OF PATIENTS, and the two answer different questions. By PATIENT count the chromatin partners dominate overwhelmingly - EP300 and TCF3 alone are 74% of partner-typed cases. The classification of individual partners is this beat's, is debatable at the margins, and is offered to make the point checkable rather than as a measurement."
  }
]
How this could bite us back
  • The single interventional idea this fusion suggests is also its main hazard: a retained kinase domain invites a kinase inhibitor, and nothing is known about whether the kinase is active, what it phosphorylates, or whether the leukaemia depends on it. BMP2K has no selective inhibitor in clinical use, and the closest agents would hit it as an off-target of broader kinase inhibitors - which means any attempt would be an uncontrolled polypharmacology experiment in a patient population of one or two.
  • More importantly, this fusion is a bite-back against the family's therapeutic framing rather than a target itself. Strategies aimed at the partner - degrading EP300, blocking a bromodomain, disrupting a prion-like IDR - have nothing to act on in a BMP2K, CLTC, SYNRG, ATP2C1 or TEX41 case. If the partner-directed approach is the one pursued, a minority of patients in this subtype are excluded from it by their own genotype, and nobody currently tracks which.
Still open
  1. Is the retained BMP2K kinase domain catalytically active in the fusion, and does the fusion phosphorylate substrates at ZNF384 binding sites? If yes, BMP2K::ZNF384 is a categorically different oncoprotein from the rest of the family and the only one with an obvious small-molecule handle. If no, the partner is inert and the case becomes further evidence that the ZNF384 half is what matters.
  2. Do the non-chromatin partners produce the same gene-expression signature as EP300 and TCF3? Every cohort that has clustered these cases has done so on the whole ZNF384 group, and no study this beat read has asked whether the BMP2K, CLTC, SYNRG or ATP2C1 cases sit inside or at the edge of the cluster. If they sit squarely inside it, the partner's identity is close to irrelevant and the family is defined by the ZNF384 lesion - which would be the strongest possible support for the C-terminal-truncation model.
  3. Is the non-chromatin tail under-ascertained? A novel partner is only found by an unbiased method, and 25 of 218 Ponte di Legno patients have no partner information at all. The chromatin-partner dominance could be partly a detection artefact of targeted assays designed around the known partners.
  4. What, if anything, does BMP2K do in a haematopoietic cell? Its only UniProt function annotation is 'may be involved in osteoblast differentiation', inferred from mouse. A partner nobody can characterise is a partner whose contribution cannot be assessed, and that is the honest state of this case.
Every source behind this entry (3)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    BMP2K was identified as a novel ZNF384 fusion partner in one patient among 291 children with BCP-ALL: nucleotide 2228 (exon 15) and nucleotide 2117 (exon 14) of BMP2K were fused in frame to nucleotide 266 (exon 3) of ZNF384, and the predicted protein retained the BMP2K kinase domain but lacked its C-terminal portion.

    ZNF384-related fusion genes define a subgroup of childhood B-cell precursor acute lymphoblastic leukemia with a characteristic immunotype. · Hirabayashi S, Ohki K, Nakabayashi K, Ichikawa H, Momozawa Y, Okamura K, et al., Kiyokawa N · Haematologica 2016
    PMID 27634205DOI
    What this evidence cannot show

    **READ THROUGH A PAGE-FETCH SUMMARISATION OF THE PMC RENDERING OF PMC5210242, NOT FROM THE PDF** - Europe PMC and NCBI efetch both refuse the full XML for this publisher, and the nucleotide numbers in this claim were extracted by a summariser rather than read by this beat directly. They should be re-checked against the PDF before anything is built on them. n = 1 patient. Two nucleotide/exon pairs are reported (2228/exon 15 and 2117/exon 14), which most likely means two transcript variants from one patient rather than two patients, but the extraction does not say and this beat did not confirm it. No functional experiment was performed on this fusion.

    Evidence that argues against this (1)
    • This beat could find NO independent confirmation of the BMP2K::ZNF384 breakpoint. The Ponte di Legno series records one BMP2K case but reports no breakpoint, and no case report or functional study of this fusion was found in 112 enumerated PubMed records for ZNF384 plus fusion or rearrangement terms in title and abstract. So the architecture claim rests on one extraction from one figure-adjacent passage in one paper, for one patient. The counterexample to the coactivator rule is real at the level of 'BMP2K is a partner' - which two independent cohorts support - but the specific claim that the KINASE DOMAIN is retained rests on a single unconfirmed source. Hirabayashi S et al. 2021
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q9NSY1 (BMP2K_HUMAN, BMP-2-inducible protein kinase, alias BIKe, EC 2.7.11.1) is a reviewed 1,161 aa human entry with a protein kinase domain at residues 51-316, a proton-acceptor active site at residue 180 and an ATP-binding region at 57-65 and 79; it is annotated as nuclear with an ECO:0000269 experimental evidence code citing PubMed 9853615, and its function is given as 'may be involved in osteoblast differentiation' by similarity to the mouse orthologue.

    What this evidence cannot show

    Fetched live in this session, not recalled. The FUNCTION comment is ECO:0000250 - inferred from the mouse orthologue Q91Z96, not experimentally established in human - so 'involved in osteoblast differentiation' is a database inference and should not be quoted as a human finding. Per this project's rule 91, the nuclear localisation comment's ECO:0000269 citation (PMID 9853615) means a curator read AN experiment in that paper; this beat did NOT open it to confirm it is a localisation experiment for BMP2K.

    Evidence that argues against this (1)
    • The kinase-domain retention argument in this entry combines a UniProt residue coordinate (domain ends at 316) with a source's nucleotide coordinate (breakpoint at cDNA nt 2117-2228), and those are different coordinate systems. Converting nt 2228 to a residue requires knowing where the coding sequence starts in the transcript Hirabayashi numbered, which this beat does not know. The conclusion survives because the gap is large - even with the most generous 5'UTR, nt 2117 is several hundred codons past residue 316 - but the arithmetic is not exact and is not presented as such. The source's own statement that the kinase domain is retained is the load-bearing evidence, not this cross-check. Hirabayashi S et al. 2016
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The Ponte di Legno childhood series independently records one BMP2K case among the 18 'other' ZNF384 partners in 193 partner-typed patients, alongside single cases of ARID1B, C11orf74, CCAR1, CLTC, DUX4, NIPBL, SEC24B, SMARCA2 and USP25 and six cases of EWSR1.

    Clinical characteristics and outcomes of B-ALL with ZNF384 rearrangements: a retrospective analysis by the Ponte di Legno Childhood ALL Working Group. · Hirabayashi S, Butler ER, Ohki K, Kiyokawa N, Bergmann AK, Moricke A, Boer JM, Cave H, Cazzaniga G, Yeoh AEJ, et al., Moorman AV, Manabe A · Leukemia 2021
    PMID 33692463DOI
    What this evidence cannot show

    FULL TEXT READ via Europe PMC fullTextXML for PMC8550960. This confirms BMP2K as a recurrent partner across two independent series but reports NO breakpoint, NO domain content and NO outcome for the case. Note that the Hirabayashi 2016 discovery cohort and this consortium series share authors and a Japanese contributing group, so 'independent' means a different study and denominator, not fully independent ascertainment - the same patient could in principle appear in both.

    Evidence that argues against this (1)
    • The partner tail in this table is shaped by detection method as much as by biology: rearrangements were detected 'by FISH, RT-PCR, and/or NGS as per individual study group criteria', and a FISH-only centre contributes a case with no partner at all - 25 of 218 patients have no partner information. A tail of one-case partners assembled across 16 consortia over 26 years cannot be read as a stable spectrum, and the non-chromatin partners in it are the ones most likely to be under-ascertained, because a novel partner is only found by an unbiased method. Zhu L et al. 2023
Connected to