BMP2K (BIKe) - a nuclear serine/threonine kinase, and the ZNF384 fusion partner that is not a chromatin protein

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.

• The retained kinase domain invites a kinase inhibitor and nothing supports one. BMP2K has no selective clinical inhibitor; the closest agents would reach it as an off-target of broader kinase inhibitors, which in a patient population of one or two would be an uncontrolled polypharmacology experiment.
• BMP2K is a nuclear NAK-family kinase and its family members AAK1 and GAK are essential to clathrin-mediated endocytosis, so a non-selective agent aimed at this family carries a general-cell-biology toxicity risk rather than a leukaemia-specific one. This beat did NOT research that literature and flags it as an unexamined hazard rather than a measured one.
• This entry is a minimal partner stub written to close a figure join. Anyone building an argument about BMP2K on it is building on one UniProt record and two sentences from two cohort papers, and the scope section says so.

1. Is the BMP2K kinase catalytically active within the BMP2K::ZNF384 fusion, and does the fusion phosphorylate anything at ZNF384 binding sites? If yes, it is a categorically different oncoprotein from the rest of the family and the only one with an obvious small-molecule handle.
2. Does truncating BMP2K after residue ~700 activate or inactivate the kinase? Three quarters of the protein is deleted in the fusion and none of it has an annotated function, so autoinhibition, dimerisation and substrate-recruitment elements could all be in there.
3. What does BMP2K do in haematopoietic cells? Its only UniProt function annotation is an orthologue inference about osteoblast differentiation, and this beat did not search the NAK-kinase-family literature at all. A partner nobody can characterise is a partner whose contribution cannot be assessed.
4. Is BMP2K a dependency in any leukaemia line? A DepMap lookup was not run by this beat and would take minutes; it would say whether the wild-type gene matters in this lineage before anyone asks whether the fusion does.

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

   UniProtKB Q9NSY1 (BMP2K_HUMAN, gene BMP2K, aliases BIKe and HRIHFB2017) is a reviewed 1,161 aa human entry, EC 2.7.11.1, with a protein kinase domain at 51-316, an ATP-binding region at 57-65 and a binding residue at 79, and a proton-acceptor active site at 180; its subcellular location is nucleus with an ECO:0000269 code citing PubMed 9853615, and its function comment 'may be involved in osteoblast differentiation' carries an ECO:0000250 code inferred from the mouse orthologue Q91Z96.

   UniProtKB entry Q9NSY1 (BMP2K_HUMAN) feature table and comments, retrieved live 2026-09-24 by q173 via an organism-restricted reviewed gene_exact search that returned exactly one record · UniProt Consortium 2026
   ▸What this evidence cannot show

   Fetched live in this session, not recalled; the gene_exact search returned exactly one reviewed human record, so there is no symbol collision here. Rule 91 has NOT been discharged: neither the paper behind the ECO:0000269 localisation comment nor any source for the function comment was opened, so both are carried as database pointers. The function annotation in particular is an orthologue inference and says nothing about human haematopoiesis. No PDB structure is recorded in this entry.

   ▸Evidence that argues against this (1)

   • The C-terminal ~845 residues of BMP2K - three quarters of the protein, and the part deleted in the leukaemia fusion - have NO annotated domain in UniProt at all, so 'the fusion loses the C-terminal portion' names a region whose function is unknown. That makes it impossible to say from the database whether the truncation is activating, inactivating or neutral for the retained kinase, and this beat found no paper that addresses it. The kinase-domain retention is therefore a structural statement with no functional consequence that anyone has measured. Hirabayashi S et al. 2016
2. E2moderate 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, with nucleotide 2228 (exon 15) and nucleotide 2117 (exon 14) of BMP2K fused in frame to nucleotide 266 (exon 3) of ZNF384, retaining the BMP2K kinase domain and lacking 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. The nucleotide coordinates should be re-checked against the PDF before anything is built on them. n = 1 patient, no functional experiment, and two nucleotide/exon pairs are reported which most likely means two transcript variants in one patient but the extraction does not say.

   ▸Evidence that argues against this (1)

   • No independent confirmation of this breakpoint exists. The Ponte di Legno series records a BMP2K case but no breakpoint, and no case report or functional study of BMP2K::ZNF384 appears among the 112 PubMed records this beat enumerated for ZNF384 with fusion or rearrangement terms in title and abstract. So the kinase-domain-retention claim - the load-bearing one for the counterexample argument - rests on a single unconfirmed extraction from a single paper about a single patient. Hirabayashi S et al. 2021
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   BMP2K recurs as a ZNF384 partner: the Ponte di Legno childhood series records one BMP2K case among the 18 'other' partners in 193 partner-typed patients.

   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. No breakpoint, domain content or outcome is reported for the BMP2K case. This consortium series and the 2016 discovery cohort share authors and a Japanese contributing group, so 'recurs' means a second study and a second denominator, not fully independent ascertainment - the same patient could in principle appear in both.

   ▸Evidence that argues against this (1)

   • One case in each of two overlapping series is the thinnest possible basis for calling a partner recurrent, and the honest reading is that BMP2K may be a single reported event counted twice. That does not weaken the argument this entry exists to support - a kinase appearing once as a ZNF384 partner is still a counterexample to a universal claim about partners - but it does mean BMP2K should not be described as an established recurrent partner alongside TAF15 or CREBBP. Hirabayashi S et al. 2016

• fuses with →ZNF384 - a C2H2 transcription factor that is the invariant 3' half of a whole B-ALL fusion subtype, including AKAP8::ZNF384

  BMP2K::ZNF384, exon 14/15 to ZNF384 exon 3; the kinase domain is retained

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

  BMP2K is the 5' partner, contributing its protein kinase domain at residues 51-316

• refutes →The ZNF384 family is not the inverse of KMT2A::AFF1 - it is the SAME grammar written in the opposite gene order, and three of the four claims in the inverse framing are false

  a nuclear serine/threonine kinase, not a chromatin coactivator, as a ZNF384 5' partner