Last-exon indels that truncate the ZNF384 C-terminus phenocopy the fusion - so the fusion partner may not be the necessary element
also called ZNF384 last-exon indel · noncanonical ZNF384 aberration · ZNF384r-like without a fusion
Two children had leukaemia that looked and behaved exactly like the ZNF384 fusion subtype, but they had no fusion at all - just a small insertion-deletion near the end of the ZNF384 gene. Standard software missed both; a human had to go and look at the raw sequencing reads.
The same thing, for a scientist
Zaliova et al. studied three children whose leukaemias carried the ZNF384-rearranged gene-expression signature with no canonical fusion. One had `fusion-znf384-tex41`; the other **two carried in-frame indels in the last exon of ZNF384**, each replacing a handful of residues near the C-terminus: **patient 2, c.1179_1192del14insGGGCAGGG, p.393_397 KHNPP>RAG, allelic frequency 47% in RNA-seq, located just at the end of the last zinc finger**; **patient 3, c.1448_1451del4insATTTTGAGTGGGACCCGAACGA, p.483_484 CS>YFEWDPND, allelic frequency 52% in RNA-seq and 43% in whole-exome sequencing** [E1]. Both were **missed by the automated variant and fusion callers** and were found only by manual, targeted reanalysis of reads [E1]. Two of the three noncanonical cases met MPAL criteria and all three expressed myeloid markers [E2]. Combining the noncanonical lesions with the canonical fusions, ZNF384 aberrations account for **8% (12/158) of B-other and 1.9% (12/643) of consecutively diagnosed childhood BCP-ALL/B-myeloid MPAL**, with the noncanonical class making up **17% (2/12) of consecutively diagnosed ZNF384-aberrant cases** [E3]. **The mechanistic reading that follows is the important one: what canonical fusions and these indels share is an altered ZNF384 C-terminus with the zinc fingers intact - not a gained coactivator.**
The two indels, verbatim
"We 'manually' reanalyzed the reads mapping to the respective genes in these 2 patients and identified in-frame indels in the last exon of ZNF384 in both. Therefore, unlike canonical ZNF384 fusion, all the novel aberrations resulted in alteration of the C-terminal portion of the ZNF384 protein."
"In patients 2 and 3, the indels led to replacement of 5 and 2 AAs by 3 and 8 new AAs, respectively, with the remaining 119 and 32 C-terminal AAs maintained from the wild-type ZNF384. In patient 2, the indel (c.1179_1192del14insGGGCAGGG; ZNF384 p.393_397KHNPP>RAG) had an allelic frequency 47% in RNAseq and was located just at the end of the last zinc finger. In patient 3, the indel (c.1448_1451del4insATTTTGAGTGGGACCCGAACGA; ZNF384 p.483_484CS>YFEWDPND) had an allelic frequency 52% and 43% in RNAseq and whole-exome sequencing, respectively"
The allelic frequencies near 50% with no other primary lesion, plus mutual exclusivity with known primary aberrations, are what the authors use to argue these are primary genetic hits rather than passengers 1.
A zinc-finger numbering collision hides inside patient 2's indel
Patient 2's indel is at p.393_397, and the paper places it "just at the end of the last zinc finger".
UniProt Q8TF68 annotates eight C2H2 zinc fingers in ZNF384, the last of which spans 433-455 4. Residue 397 is the end of UniProt's sixth finger (373-397), not its eighth. The paper's own Figure 1B schematic, adapted from Liu et al., labels six zinc fingers.
So two respectable sources give ZNF384 a different number of zinc fingers, and the disagreement is not cosmetic: it changes whether patient 2's indel sits inside the DNA-binding array or 36 residues past it. Dickerson et al. give a third answer - ZNF384 fusion oncoproteins retain "five to eight Cys2His2 zinc fingers", and note that exclusion of ZNF384 exon 8 produces in-frame splicing with loss of zinc fingers 4 and 5 5. This is recorded as its own analysis entry, ana-q173-the-znf384-zinc-finger-count-is-not-agreed, because every domain figure and every 'all eight fingers retained' statement in this corpus depends on which count is right.
| lesion class | in-frame indel in the last exon of ZNF384, altering the C-terminal portion of the protein |
|---|---|
| gene | gene-znf384 |
| uniprot | Q8TF68 |
variants
[
{
"patient": "Zaliova patient 2",
"cdna": "c.1179_1192del14insGGGCAGGG",
"protein": "p.393_397 KHNPP>RAG",
"vaf_rnaseq_pct": 47,
"vaf_wes_pct": null,
"position_note": "the paper places this just at the end of the last zinc finger; UniProt Q8TF68 places residue 397 at the end of its SIXTH finger (373-397) and annotates two further fingers at 403-425 and 433-455",
"c_terminal_aa_retained": 119
},
{
"patient": "Zaliova patient 3",
"cdna": "c.1448_1451del4insATTTTGAGTGGGACCCGAACGA",
"protein": "p.483_484 CS>YFEWDPND",
"vaf_rnaseq_pct": 52,
"vaf_wes_pct": 43,
"position_note": "distal to every annotated zinc finger under either numbering",
"c_terminal_aa_retained": 32
}
]measurements — the numbers behind the chart
[
{
"quantity": "proportion",
"subject": "noncanonical ZNF384 aberrations among consecutively diagnosed ZNF384-aberrant childhood leukaemias",
"value": 0.167,
"units": "fraction",
"n": 12,
"universe": "consecutively diagnosed ZNF384-aberrant cases in 643 Czech children with BCP-ALL or B/myeloid MPAL, Dec 2010 to Dec 2020",
"null": null,
"note": "2 of 12, reported by the authors as 17%. Wilson 95% interval on 2/12 is roughly 4.7-45%, so this is an order-of-magnitude estimate and not a rate."
},
{
"quantity": "proportion",
"subject": "all ZNF384 aberrations (canonical fusion plus noncanonical) among consecutive childhood BCP-ALL/B-myeloid MPAL",
"value": 0.019,
"units": "fraction",
"n": 643,
"universe": "643 consecutive children diagnosed in the Czech Republic Dec 2010 to Dec 2020; 12 events",
"null": null,
"note": "THIS is the population figure. The same cohort read against its 158 B-other cases gives 8%, which is a different denominator and not a subtype frequency."
},
{
"quantity": "allelic_fraction",
"subject": "ZNF384 p.393_397 KHNPP>RAG in RNA-seq",
"value": 0.47,
"units": "fraction of reads",
"n": 1,
"universe": "RNA-seq of Zaliova patient 2 diagnostic sample",
"null": 0.5,
"note": "the null of 0.5 is what a clonal heterozygous variant gives; 47% is consistent with a clonal primary hit, which is the authors' argument"
},
{
"quantity": "allelic_fraction",
"subject": "ZNF384 p.483_484 CS>YFEWDPND in whole-exome sequencing",
"value": 0.43,
"units": "fraction of reads",
"n": 1,
"universe": "WES of Zaliova patient 3 diagnostic sample; the same variant was 52% in RNA-seq of the same patient",
"null": 0.5,
"note": "the RNA-seq and WES values differ by 9 points in the same patient, which is a useful reminder that a single VAF is not a clonality measurement"
}
]- Diagnostic delivery is the bite-back here and it is concrete: a patient with a last-exon ZNF384 indel is negative on every fusion assay and on most automated variant pipelines. If ZNF384 status is ever used to allocate therapy, the pipeline defines the cohort, and this class of patient is systematically excluded while having the same disease. Zaliova's own recommendation - targeted manual reanalysis of ZNF384 in any case with the expression signature and no detected fusion - is a workflow requirement, not a research suggestion.
- If the true necessary lesion is loss of the ZNF384 C-terminus rather than gain of a coactivator, then any therapeutic strategy aimed at the fusion partner - degrading EP300, inhibiting a bromodomain, disrupting an IDR - is aimed at a feature that a substantial minority of patients in this subtype do not have.
- Does a ZNF384 C-terminal truncation, on its own and with no fusion partner, transform a haematopoietic stem and progenitor cell? This is the single cheapest experiment that would discriminate between 'the partner supplies the oncogenic module' and 'the partner is incidental and the lesion is loss of the ZNF384 C-terminus'. Dickerson's HSPC and knockin systems exist and would take the construct directly.
- How often are these lesions present and unreported? Zaliova state that no similar ZNF384 mutation has been reported in any published genomic study or public database, and found two of their own only by manual read inspection after five callers missed them. That is a testable claim about the field: reanalyse deposited RNA-seq from ZNF384r-like cases in other cohorts and count.
- What does the ZNF384 C-terminus actually do? Both indels and the TEX41 truncation converge on it, and the paper notes it 'lacks well-defined functional domains'. A region that three independent leukaemogenic lesions target and that has no annotated function is exactly the kind of gap this graph should be filling - is it a degron, an autoinhibitory element, a partner-binding surface?
- Is the 119-residue versus 32-residue difference in retained C-terminus between patients 2 and 3 functionally meaningful, or does any disruption of this region suffice? Two patients with very different amounts of remaining protein produced the same signature, which weakly favours 'any disruption suffices' - but n = 2.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Two of three children with the ZNF384-rearranged expression signature and no canonical fusion carried in-frame indels in the last exon of ZNF384: p.393_397 KHNPP>RAG (c.1179_1192del14insGGGCAGGG, 47% allelic frequency in RNA-seq, located just at the end of the last zinc finger) and p.483_484 CS>YFEWDPND (c.1448_1451del4insATTTTGAGTGGGACCCGAACGA, 52% in RNA-seq and 43% in whole-exome sequencing). Both were missed by the automated variant and fusion callers used and were found only by manual, targeted reanalysis of the mapped reads.
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 2021What this evidence cannot show
FULL TEXT READ via Europe PMC fullTextXML for PMC8579251, retrieved 2026-09-24; these sentences are from the Results and discussion section. n = 2 for the indel class. The callers used were VarScan, samtools, TopHat, deFuse and Cicero, and the authors state explicitly 'We cannot exclude that some other variant callers, which we did not test, may have found these mutations' - so the instrument-failure claim is bounded to the tools they ran. The functional consequence of either indel was NOT tested in any assay.
Evidence that argues against this (1)
- The strongest argument against these being causal is that nothing was done to show they are. There is no reporter assay, no expression of the mutant protein, no occupancy experiment. The case for causality is circumstantial and the authors say so: allelic frequency near 50%, mutual exclusivity with other primary lesions, and a matching expression signature. All three are consistent with a passenger in a clone that acquired its phenotype some other way, and the authors' own sentence concedes 'functional studies would be necessary to directly demonstrate the biological consequences of these aberrations'. Zaliova M et al. 2021
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
All three patients with noncanonical ZNF384 aberrations expressed myeloid markers, CD10 was weak in two, two of three were classified as MPAL, and all three expressed CD66c, which is rarely expressed in ZNF384-rearranged ALL.
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 2021What this evidence cannot show
n = 3 and the marker statements are pooled across all three noncanonical cases, one of which is the ZNF384::TEX41 fusion rather than an indel. The per-antigen percentages are in the paper's Table 1, which the retrieved XML renders as a caption only, so no numeric marker value is recorded here.
Evidence that argues against this (1)
- CD66c is the marker that does not fit. The authors flag it as rarely expressed in canonical ZNF384-rearranged ALL and report it in all three noncanonical cases. Either the noncanonical class is immunophenotypically distinct after all - which would weaken the phenocopy claim this entry is built on - or 3 of 3 on a rare marker is noise at n = 3. The paper does not have the numbers to separate these and neither does this entry. Zaliova M et al. 2021
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In 643 children diagnosed with BCP-ALL or B/myeloid MPAL in the Czech Republic between December 2010 and December 2020, of whom 158 were B-other, canonical ZNF384 rearrangements accounted for 6% (10/158) of B-other and 1.6% (10/643) of all cases; adding the noncanonical aberrations gives 8% (12/158) and 1.9% (12/643); the noncanonical class comprises 17% (2/12) of consecutively diagnosed ZNF384-aberrant cases.
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 2021What this evidence cannot show
This is the one place in this beat's reading where a ZNF384 frequency is quoted against a CONSECUTIVE, population-level denominator (643 children in one country over a decade) rather than against a B-other residual group. That is why it is recorded. It is still a single national cohort, it mixes BCP-ALL with B/myeloid MPAL in the denominator, and 12 events is a small numerator - the Wilson 95% interval on 12/643 runs roughly 1.1-3.2%.
Evidence that argues against this (1)
- The 6% and 8% B-other figures should NOT be read as subtype frequencies, and this is the trap this graph already recorded on gene-znf384. B-other is defined by standard-of-care genetics having FAILED to assign a subtype, so any lesion that standard-of-care genetics cannot see is enriched in that denominator by construction. The 1.6% and 1.9% figures against 643 consecutive cases are the ones that mean something. Note also that the same paper reports ZNF384 rearrangement at '5% to 10% of B-other leukemias' in its introduction, citing other cohorts, so 6% is at the low end of the published spread and cohort-to-cohort variation exceeds the difference the novel class adds. Zaliova M et al. 2021
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB Q8TF68 annotates EIGHT C2H2 zinc fingers in ZNF384, at 228-250, 256-278, 284-306, 317-339, 345-367, 373-397, 403-425 and 433-455, in a 577 aa protein.
UniProtKB entry Q8TF68 (ZN384_HUMAN) feature table, as recorded in this graph's gene-znf384 entry and confirmed against the live record · UniProt Consortium 2026What this evidence cannot show
Cited here only to establish the numbering conflict, not to settle it. A UniProt C2H2 feature is a profile match, and profile-based finger calls routinely differ from the count a paper's schematic draws, because a degenerate finger can match a profile without being a functional DNA contact. This entry does not claim UniProt is right and the papers wrong; it claims they disagree.
Evidence that argues against this (1)
- Zaliova et al. write 'the 6 zinc fingers are mostly intact' and their Figure 1B legend labels 'ZFs, Kruppel-type C2H2 zinc-finger domains' on a schematic adapted from Liu et al. Dickerson et al. write that ZNF384 fusion oncoproteins retain 'five to eight Cys2His2 zinc fingers'. Three independent sources give three answers - 8, 6, and 'five to eight'. The disagreement is real and unresolved, and it is directly load-bearing here because it decides whether patient 2's p.393_397 indel is inside or outside the array. Dickerson KM et al. 2022
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ZNF384 fusion oncoproteins retain five to eight Cys2His2 zinc fingers, and exclusion of ZNF384 exon 8 results in in-frame splicing with loss of zinc fingers 4 and 5 - so the number of fingers in the expressed protein varies by splicing as well as by breakpoint.
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 2022What this evidence cannot show
Retrieved from the PMC rendering of PMC9414741 in this session. The exon-8 splicing statement is tied to Figure 6B and the accompanying text, so support_location is 'figure'; the surrounding quantitative detail (how often exon 8 is skipped, in which cases) was not extracted and is not claimed here. The 'five to eight' range is the paper's own summary phrasing for the family, not a per-fusion measurement.
Evidence that argues against this (1)
- This item is itself the counter-evidence to the tidy version of the family's architecture, so the honest disclosure is the opposite one: this beat found NO source that reports a systematic, case-level census of how many zinc fingers each ZNF384 fusion actually retains. 'Five to eight' is a range asserted in a review sentence within a primary paper, not a distribution anyone measured and published. Nobody appears to have disagreed with it, but nobody appears to have checked it either, and the searches run here were ZNF384 plus fusion or rearrangement terms in title and abstract, which would not reach a supplementary table where such a census would live. Dickerson KM et al. 2022
an out-of-frame truncation at ZNF384 p.481 produces the subtype's expression signature with no partner contribution
in-frame indels in the last exon of ZNF384, p.393_397 KHNPP>RAG and p.483_484 CS>YFEWDPND
two of three noncanonical-lesion patients met MPAL criteria
the empirical basis for the hypothesis: the subtype occurs with no fusion partner at all
the p.393_397 indel is 'just at the end of the last zinc finger' under the six-finger convention and between fingers 6 and 7 under UniProt's eight
the noncanonical lesions reproduce the immunophenotype but add CD66c, which is rare in canonical ZNF384r
adding the noncanonical lesions moves the population figure from 1.6% to 1.9%