Oncofusion

start / graph / ana-q173-the-znf384-zinc-finger-count-is-not-agreed

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

Four sources give ZNF384 three different zinc-finger counts - 8, 6 and 'five to eight' - and the paralogue's feature table explains where the 6 comes from

also called ZNF384 zinc finger count · six versus eight zinc fingers

Nobody agrees how many zinc fingers ZNF384 has. The database and the paper that discovered the fusion family both say eight; the more recent leukaemia papers draw six; one says between five and eight. Looking at ZNF384's closest relative explains it: the relative has exactly six, and the newer papers are drawing the part the two share.

The same thing, for a scientist

**The count matters because every 'the fusion retains all the zinc fingers' statement in this family depends on how many there are, and because one reported leukaemogenic indel sits at residue 393-397 - inside the array under one count and 36 residues past its end under another.** The published answers are: **eight**, from the UniProtKB Q8TF68 feature table (228-250, 256-278, 284-306, 317-339, 345-367, 373-397, 403-425, 433-455) [E1] **and, independently, from the founding paper of the whole fusion family, whose Figure 1A legend describes CIZ/NMP4 as carrying "eight Kruppel-type C2H2 zinc finger domains"** [E6]; **six**, asserted in the text of Li et al. PNAS 2018 ("ZNF384 and ZNF362 are homologous C2H2-type zinc-finger transcription factors containing six zinc fingers") and used in Zaliova et al.'s Figure 1B schematic and its text ("the 6 zinc fingers are mostly intact") [E2][E3]; and **five to eight**, in Dickerson et al. [E4]. **So the split is chronological, not database-versus-literature: the 2002 primary description and the database agree on eight, and the 2018-2021 papers say six.** This analysis adds the observation that resolves where 'six' comes from. **ZNF362, the paralogue, has exactly six C2H2 fingers in UniProt (Q5T0B9, 420 aa: 227-249, 255-277, 283-305, 311-335, 341-363, 371-393) - and their coordinates sit within 6 residues of ZNF384's first six. ZNF384 carries two further annotated fingers, at 403-425 and 433-455, that ZNF362 does not have** [E5]. The newer schematic is a paralogue-shared core, and ZNF384's last two fingers are extra. **Dickerson's 'five to eight' is separately explained: exclusion of ZNF384 exon 8 gives in-frame splicing with loss of zinc fingers 4 and 5, so the number in the expressed protein is also a splicing variable** [E4].

Go deeper
The three answers, side by side
sourcecountwhere stated
UniProtKB Q8TF68 feature table8 (228-250, 256-278, 284-306, 317-339, 345-367, 373-397, 403-425, 433-455)database deposit 1
Martini A et al., Cancer Res 2002 - the founding paper8 - "eight Kruppel-type C2H2 zinc finger domains"Figure 1A legend 6
Li JF et al., PNAS 20186 - "ZNF384 and ZNF362 are homologous C2H2-type zinc-finger transcription factors containing six zinc fingers"body text 2
Zaliova M et al., Blood Adv 20216 - "the 6 zinc fingers are mostly intact"; Figure 1B schematic adapted from Liu et al.body text and figure 3
Dickerson KM et al., Blood Cancer Discov 2022five to eight Cys2His2 zinc fingers, with loss of fingers 4 and 5 on exon-8 exclusionbody text and Figure 6B 4

These are not three measurements of the same thing. UniProt's is a profile match over the sequence. PNAS's is a statement about a gene family. Dickerson's is a statement about the range across expressed fusion proteins. All three can be simultaneously correct while being incompatible as a citation for 'ZNF384 has N zinc fingers'.

Where the six comes from, and the check that produced it

Both feature tables were fetched live in this session and read out in full 15:

ZNF384  Q8TF68  577 aa   228-250  256-278  284-306  317-339  345-367  373-397  403-425  433-455
ZNF362  Q5T0B9  420 aa   227-249  255-277  283-305  311-335  341-363  371-393    --       --
offset                      -1       -1       -1       -6       -4       -2

The first six fingers of the two paralogues are annotated at coordinates within six residues of each other. ZNF384's seventh and eighth have no counterpart in ZNF362.

So the literature's 'six' is the shared core of the ZNF384/ZNF362 family, and the PNAS sentence that states it says exactly that - it is a sentence about the family, applied to both members. UniProt gives ZNF384 two more.

What this analysis does NOT establish. Coordinate proximity is not a sequence alignment; no alignment was run here and none is claimed. It is possible that ZNF384's fingers 7 and 8 are degenerate profile matches that a curator drawing a functional schematic would rightly omit, rather than real fingers ZNF362 lost. Distinguishing those two readings needs the alignment and, ideally, a DNA-binding experiment on truncation constructs - neither of which exists in what this beat read.

Why it is load-bearing, not pedantry

Three concrete consequences:

  1. Zaliova patient 2's indel is at p.393_397, which the paper places "just at the end of the last zinc finger." Under UniProt's numbering, 373-397 is the sixth of eight and there are two fingers beyond it. The paper's own interpretation - that DNA binding is likely preserved because the fingers are mostly intact - is unaffected either way, but the statement "located just at the end of the last zinc finger" is only true under the six-finger convention, and a reader who looks it up in UniProt will think the paper made an error. It did not; it used a different count.

  2. gene-znf384 in this graph currently records eight fingers spanning 228-455 and describes the array as "retained intact in ZNF384 fusions". With Dickerson's exon-8 splicing result, that is not safe as written: the expressed protein can lack fingers 4 and 5. The domain payload should stay as UniProt (it is a database record and correct as such), but the claim about retention should be qualified. This is reported as a correction to an entry this beat does not own, per the standing rule, with the citation attached: PMID 35247902, DOI 10.1158/2643-3230.BCD-21-0163.

  3. Any 3D domain-viewer lane drawn for a ZNF384 fusion will show eight bars, because the viewer reads the UniProt payload, while every schematic in the source literature shows six. A reader comparing the two will conclude one of them is wrong.
Facts and figures
statusrun
result summaryThree published counts (8, 6, five-to-eight) are all defensible against different definitions. The literature's 6 corresponds to the six C2H2 fingers UniProt annotates in the paralogue ZNF362, whose coordinates lie within 6 residues of ZNF384's first six; ZNF384 carries two additional annotated fingers at 403-425 and 433-455 with no ZNF362 counterpart. Separately, exon-8 exclusion removes two fingers from the expressed protein, so the count varies by splicing as well as by convention.
measurements
count2.000count: 2count6.000count: 6offset_max6.000offset_max: 6count8.000count: 8
inputs
[
  "https://rest.uniprot.org/uniprotkb/Q8TF68.json",
  "https://rest.uniprot.org/uniprotkb/Q5T0B9.json",
  "PMID 30487223 full text (PMC6294900)",
  "PMID 34529760 full text (PMC8579251)",
  "PMID 35247902 (PMC9414741)"
]
measurements — the numbers behind the chart
[
  {
    "quantity": "count",
    "subject": "C2H2 zinc fingers annotated in ZNF384 by UniProt",
    "value": 8,
    "units": "domains",
    "n": 1,
    "universe": "UniProtKB Q8TF68 feature table, 577 aa, retrieved 2026-09-24",
    "null": 6,
    "note": "the null here is the count the leukaemia literature uses, which is what makes this a disagreement rather than a fact"
  },
  {
    "quantity": "count",
    "subject": "C2H2 zinc fingers annotated in the paralogue ZNF362 by UniProt",
    "value": 6,
    "units": "domains",
    "n": 1,
    "universe": "UniProtKB Q5T0B9 feature table, 420 aa, retrieved 2026-09-24, same profile detector as Q8TF68",
    "null": 8,
    "note": "the null is ZNF384's count. The paralogue has exactly the number the leukaemia literature attributes to BOTH proteins."
  },
  {
    "quantity": "offset_max",
    "subject": "largest coordinate difference between ZNF384 finger n and ZNF362 finger n, for n = 1 to 6",
    "value": 6,
    "units": "residues",
    "n": 6,
    "universe": "start coordinates 228/227, 256/255, 284/283, 317/311, 345/341, 373/371 from the two UniProt feature tables",
    "null": null,
    "note": "a coordinate comparison, NOT an alignment. Recorded to make the observation checkable, not to assert orthology of individual fingers."
  },
  {
    "quantity": "count",
    "subject": "ZNF384 zinc fingers lost when exon 8 is excluded by alternative splicing",
    "value": 2,
    "units": "domains",
    "n": null,
    "universe": "Dickerson et al. Figure 6B: exclusion of exon 8 gives in-frame splicing with loss of zinc fingers 4 and 5",
    "null": 0,
    "note": "this is why the expressed count is not fixed even once the convention is fixed"
  }
]
How this could bite us back
  • This analysis supports no intervention. Its interventional relevance is indirect: if a therapeutic strategy ever targets the ZNF384 DNA-binding surface - a molecular glue, a decoy oligonucleotide, a degron aimed at the fusion - then which fingers exist and which are expressed is the whole design brief, and the field currently holds three answers.
  • A domain figure drawn from UniProt and a domain figure drawn from the literature will not match for this protein. Anyone comparing a ZNF384 lane in this graph's 3D viewer against a published schematic will see eight bars against six and should not conclude either is wrong.
Still open
  1. Are ZNF384 fingers 7 and 8 (403-425, 433-455) real DNA-contacting fingers or degenerate profile matches? A pairwise alignment of Q8TF68 and Q5T0B9, plus an EMSA or SELEX on ZNF384 truncated at 397 versus 455, would settle both this and the paralogue question. Neither has been done and both are cheap.
  2. How often is ZNF384 exon 8 excluded, and does it differ between fusion partners or between B-ALL and MPAL presentations? If the expressed protein routinely lacks two fingers, then the 'complete ZNF384 coding sequence is retained' description of the family is true at the DNA level and false at the protein level, and the discrepancy is systematic rather than occasional.
  3. Does the ZNF362 fusion subgroup, which shares the partners TAF15 and SMARCA2 with ZNF384 and clusters with it transcriptionally, differ in any measurable way? Four cases exist. A protein with six fingers and one with eight, fused to the same partners at similar breakpoints, producing the same expression signature, would be strong evidence that fingers 7 and 8 do not matter.
  4. Does any other entry in this corpus quote a zinc-finger count for ZNF384 that would now need qualifying? gene-znf384 records eight and describes the array as retained intact in fusions; the exon-8 splicing result makes the retention clause unsafe as written. That is reported, not edited.
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q8TF68 (ZN384_HUMAN, 577 aa) annotates eight C2H2-type zinc fingers at 228-250, 256-278, 284-306, 317-339, 345-367, 373-397, 403-425 and 433-455.

    What this evidence cannot show

    A UniProt C2H2 feature is a PROSITE-class profile match, not an experiment. Profile-based finger calls routinely exceed the number a functional study or a review schematic recognises, because a degenerate finger can match the profile without contributing to DNA binding. This item establishes what the database says and nothing about which fingers bind DNA.

    Evidence that argues against this (1)
    • Two primary sources in the leukaemia literature state six, not eight, and one states a range of five to eight. The database count is the outlier in the field even though it is the most mechanically reproducible number. The direction of the disagreement - a profile detector finding MORE fingers than curators draw - is the expected one for degenerate C2H2 arrays, which argues that fingers 7 and 8 at 403-425 and 433-455 are the ones in doubt. Li JF et al. 2018
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Li et al. state that ZNF384 and ZNF362 are homologous C2H2-type zinc-finger transcription factors containing six zinc fingers belonging to the ZFAM4 gene family, that the zinc-finger domains are retained in both ZNF384 and ZNF362 fusion proteins, and that the ZNF362 partners TAF15 and SMARCA2 also fuse to ZNF384 with similar breakpoints.

    Transcriptional landscape of B cell precursor acute lymphoblastic leukemia based on an international study of 1,223 cases. · Li JF, Dai YT, Lilljebjorn H, Shen SH, Cui BW, Bai L, et al. · Proc Natl Acad Sci U S A 2018
    PMID 30487223DOI
    What this evidence cannot show

    FULL TEXT READ via the PMC HTML rendering of PMC6294900 (pnas.org returns 403). The six-finger statement is a descriptive sentence in the Results, referenced to the paper's reference 14, which this beat did not open - so the count is being carried from a source one step further back. The retention claim points to SI Appendix Figure S8B, a supplementary schematic that was not retrieved.

    Evidence that argues against this (1)
    • This paper's own table contradicts its body text on a neighbouring fact, which is a reason for caution about its descriptive sentences generally: the G5 row lists 'SMARCA2-ZNF362 (4)', collapsing all four ZNF362 cases under one partner, while the body text says SMARCA2-ZNF362 (n = 3) and TAF15-ZNF362 (n = 1). The body text is the more specific statement and is the one used here, but a paper that disagrees with itself in a table is not a strong authority for a domain count it cites to someone else. Li JF et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Zaliova et al. write that 'DNA binding is likely preserved in ZNF384 with novel aberrations, as the 6 zinc fingers are mostly intact', and locate the p.393_397 indel of their patient 2 'just at the end of the last zinc finger'; their Figure 1B schematic of the ZNF384 protein is adapted from Liu et al. and labels the fingers collectively as 'ZFs, Kruppel-type C2H2 zinc-finger domains'.

    A novel class of ZNF384 aberrations in acute leukemia. · Zaliova M, Winkowska L, Stuchly J, Fiser K, Triska P, Zwyrtkova M, Hrusak O, Starkova J, Sramkova L, Stary J, Trka J, Zuna J · Blood Adv 2021
    PMID 34529760DOI
    What this evidence cannot show

    FULL TEXT READ via Europe PMC fullTextXML for PMC8579251. The Figure 1B image itself was not viewed - only its legend and the text that refers to it - so the schematic's finger count is inferred from the legend's plural label plus the body text's 'the 6 zinc fingers', not from counting bars in the figure.

    Evidence that argues against this (1)
    • Under the UniProt numbering, residue 397 ends the SIXTH of eight fingers and two further fingers lie at 403-425 and 433-455, so 'just at the end of the last zinc finger' is not literally true against the database. This is not an error by the authors - they are using the six-finger convention consistently - but it means the sentence cannot be quoted as positional evidence without stating which convention it assumes. Had the eight-finger numbering been used, the same indel would be described as sitting between fingers 6 and 7, which reads very differently. UniProt Consortium 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Dickerson et al. describe ZNF384 fusion oncoproteins as retaining five to eight Cys2His2 zinc fingers, with full-length ZNF384 as the C-terminal partner juxtaposed to a range of N-terminal partners, and report that exclusion of ZNF384 exon 8 results in in-frame splicing with loss of zinc fingers 4 and 5.

    ZNF384 Fusion Oncoproteins Drive Lineage Aberrancy in Acute Leukemia. · Dickerson KM, Qu C, Gao Q, Iacobucci I, Gu Z, Yoshihara H, Backhaus EA, Chang Y, Janke LJ, Xu B, Wu G, Papachristou EK, D'Santos CS, Roberts KG, Mullighan CG · Blood Cancer Discov 2022
    PMID 35247902DOI
    What this evidence cannot show

    Read from the PMC rendering of PMC9414741. The exon-8 skipping observation is tied to Figure 6B, hence support_location 'figure'; the frequency of exon-8 exclusion, in which samples, and whether both isoforms are expressed in the same patient were NOT extracted and are not claimed. Note that 'loss of zinc fingers 4 and 5' uses yet another numbering - it cannot be mapped onto UniProt's eight or the literature's six without seeing the figure.

    Evidence that argues against this (1)
    • This item is itself a counter to the tidy account, so the honest disclosure is about its own limits: 'five to eight' is a summary phrase in a primary paper, not a census anyone published. This beat found NO systematic, case-level tabulation of how many zinc fingers each reported ZNF384 fusion retains, and the searches run - ZNF384 combined with fusion or rearrangement terms in title and abstract, 112 records enumerated - would not reach such a table if it exists in a supplement. The absence is bounded by titles and abstracts only. Dickerson KM et al. 2022
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q5T0B9 (ZN362_HUMAN, ZNF362, 420 aa) annotates exactly SIX C2H2-type zinc fingers, at 227-249, 255-277, 283-305, 311-335, 341-363 and 371-393 - coordinates lying within six residues of ZNF384's first six fingers - and has no counterpart to ZNF384's fingers at 403-425 and 433-455.

    What this evidence cannot show

    THIS IS A COORDINATE COMPARISON, NOT AN ALIGNMENT. No sequence alignment between Q8TF68 and Q5T0B9 was run in this session, and none is claimed. The two proteins differ in length by 157 residues, so near-identical finger coordinates in the 227-393 window are a striking observation but do not by themselves prove one-to-one orthology of the individual fingers. Both records use the same profile detector, which is what makes the counts directly comparable and is also why a shared detector bias would affect both.

    Evidence that argues against this (1)
    • The explanation offered here - that the literature's 'six' is the paralogue-shared core - is an inference from two feature tables, not something any source states. An equally consistent reading is that ZNF384's fingers 7 and 8 are degenerate matches the profile detector over-calls, in which case ZNF362 is simply the cleaner protein and the paralogue comparison is a coincidence of a shared detector rather than a shared architecture. This analysis cannot distinguish the two, and the discriminating work - a pairwise alignment plus a DNA-binding assay on ZNF384 truncated at 397 versus 455 - has not been done by anyone. UniProt Consortium 2026
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The founding paper of the ZNF384 fusion family describes CIZ/NMP4 as containing leucine-rich, serine-rich and proline-rich domains at the N-terminus, followed by a nuclear localization signal, EIGHT Kruppel-type C2H2 zinc finger domains, and a Gln-Ala repeat at the C-terminus; it also reports the longer CIZ cDNA as encoding an open reading frame of 578 amino acids, with a shorter splice variant of 517.

    Recurrent rearrangement of the Ewing's sarcoma gene, EWSR1, or its homologue, TAF15, with the transcription factor CIZ/NMP4 in acute leukemia. · Martini A, La Starza R, Janssen H, Bilhou-Nabera C, Corveleyn A, Somers R, Aventin A, Foa R, Hagemeijer A, Mecucci C, Marynen P · Cancer Res 2002
    PMID 12359745
    What this evidence cannot show

    FULL TEXT READ, but by an unusual route: aacrjournals.org is Cloudflare-blocked to every fetch method tried, and the complete published full text was recovered from an Internet Archive snapshot of the original HighWire page, dated 2009-09-29. The quoted sentence is the Figure 1A legend. This paper predates DOI assignment for AACR content and no DOI exists in PubMed, Europe PMC or Crossref for it, so the archived URL is the only resolvable locator besides the PMID. expected_absent_terms names ZNF384 because the gene is called CIZ/NMP4 throughout this 2002 paper. Note also a length discrepancy worth carrying: Martini reports a 578-residue ORF where UniProt Q8TF68 is 577 aa.

    Evidence that argues against this (1)
    • This item weakens the tidy 'database over-calls, papers get it right' reading that the rest of this analysis might suggest, and it is recorded for that reason. The 2002 primary description, which characterised the CIZ cDNA itself, says eight - agreeing with UniProt and disagreeing with the 2018 and 2021 papers. So the disagreement is chronological rather than methodological, and the more recent 'six' is the departure, not the correction. This beat found no paper that explicitly argues against the eight-finger count or explains the change; the newer papers simply state six without addressing the earlier figure. Li JF et al. 2018
Connected to