Oncofusion

start / graph / fusion-tcf3-znf384

Fusionstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.q173-the-znf384-fusion-family

TCF3::ZNF384 - the partner whose contribution is now actually demonstrated: steroid resistance, and it is not shared with EP300::ZNF384

also called TCF3-ZNF384 · E2A-ZNF384 · T-Z · t(12;19)(p13;p13)

This version of the ZNF384 fusion uses TCF3 as the front half - the same gene that partners PBX1 in a well-known childhood leukaemia. Children with it are younger, respond worse to steroids and relapse more often, and one experiment has now shown that the steroid resistance really does come from the TCF3 half and not from the shared ZNF384 half.

The same thing, for a scientist

TCF3::ZNF384 joins **TCF3 exon 11 or exon 13 to ZNF384 exon 3**; the predicted protein *"does not have the principal functional domains of wild-type TCF3, whereas it retains almost all of the part of the ZNF384 protein"* [E1]. Against UniProt P15923 (654 aa), that means the **bHLH at 549-602 is lost**, and by the coordinates the leucine zipper at 389-425 sits in the retained N-terminal region containing the activation domains [E5]. It is the **second commonest partner** - 60 of 193 partner-typed Ponte di Legno cases (31%) - and **clinically the worst-behaved**: younger (median 5 years vs 11 for EP300, P < 0.001), 46% allocated high risk vs 28%, 32% MRD-positive at end of induction vs 16%, and a significantly poorer steroid response with higher relapse frequency [E2][E3]. **The decisive experiment is recent and is the reason this entry exists as its own node.** Kusano et al. transduced BCP-ALL lines with TCF3::ZNF384 and EP300::ZNF384 side by side: **both** down-regulated CD10, but **only TCF3::ZNF384** produced resistance to dexamethasone-induced apoptosis and up-regulated CCND2 [E4]. **That is the cleanest available dissection of this family: the shared ZNF384 half carries the lineage phenotype, and the partner adds a separate, partner-specific function.**

Go deeper
The one experiment that separates the two halves

Verbatim from the Kusano abstract 4:

"Flow cytometric analysis and RT-qPCR revealed down-regulation of CD10 in BCP-ALL cells after transduction with both TCF3::ZNF384 and EP300::ZNF384. The annexin-V binding apoptosis assay indicated that TCF3::ZNF384-, but not EP300::ZNF384-, expressing cells exhibited increased resistance to dexamethasone-induced apoptosis. By means of an oligonucleotide microarray and RT-qPCR, we observed that the transduction of TCF3::ZNF384, but not EP300::ZNF384, leads to significant enhancement of cyclin D2 (CCND2) gene expression in BCP-ALL cells, but no growth advantage was observed."

"Other than the common functions of ZNF384-r that contribute to the development of leukemia with a lineage-ambiguous phenotype, TCF3::ZNF384 may exhibit a fusion partner-dependent function distinct from EP300::ZNF384"

Read carefully, this is a two-factor result and both factors matter.

  1. CD10 down-regulation is SHARED. Two fusions with completely different 5' partners - a bHLH transcription factor and a fragment of an acetyltransferase - produce the same immunophenotypic change. That is the ZNF384 half doing it.
  2. Dexamethasone resistance is NOT shared. It tracks the partner, and it corresponds to the clinical difference that motivated the experiment.

This is the model the whole family needed and did not have: a common ZNF384-driven lineage programme, plus partner-specific modifiers. It also means the phrase 'ZNF384-rearranged leukaemia' names a shared biology and a heterogeneous clinical entity at the same time, and that both statements are literally true.

The limits are real. Retroviral transduction into an established BCP-ALL cell line is not the cell of origin, 'no growth advantage was observed' is in the same sentence as the CCND2 result, and this beat read the abstract only.

Breakpoints - and a discrepancy in the ZNF384 exon
sourcejunction
Hirabayashi et al. 2017, patient sequencingTCF3 exon 11 to ZNF384 exon 3; also TCF3 exon 13 to ZNF384 exon 3, in-frame, in 7 patients 1
Dickerson et al. 2022, expression constructsTCF3 exon 13 to ZNF384 exon 5 6

These do not agree on the ZNF384 side. Every other fusion in this family joins at ZNF384 exon 3, the first coding exon - Gocho states that explicitly for EP300 and notes it matches the earlier TAF15 and E2A fusions, and Nyquist found the ZNF384 breakpoint in the same intron, between exons 2 and 3, in both of their TAF15 cases. Dickerson's construct methods say exon 5 for TCF3::ZNF384 while saying exon 3 for EP300::ZNF384, TAF15::ZNF384 and CREBBP::ZNF384.

This beat cannot resolve it and records it rather than picking. Possibilities: a different ZNF384 transcript numbering in the construct methods; a genuinely different junction in the patient sample those constructs came from; or a typographical error. It matters because ZNF384 exon 5 would remove coding sequence that exon 3 retains, and because every 'the entire ZNF384 coding region is retained' statement in this family assumes the exon-3 junction.

Isoform note, from this project's own registry: TCF3 has two isoforms, E12 (P15923-1, 654 aa) and E47 (P15923-2, 651 aa), which are identical through residue 529 and diverge only in the bHLH block. Any TCF3::ZNF384 coordinate below 530 is therefore isoform-independent, which covers the retained activation-domain region; only the lost bHLH is isoform-sensitive.

Facts and figures
gene5TCF3
gene3ZNF384
cytobandt(12;19)(p13;p13) - reported as a cryptic t(12;19)(p13.3;p13.3) in the earliest description of E2A-ZNF384
uniprot5P15923
uniprot3Q8TF68
measurements
proportion0.024proportion: 0.024proportion0.310proportion: 0.31proportion0.320proportion: 0.32event_free_survival_5y0.500event_free_survival_5y: 0.5age_median5.000age_median: 5latency_median74.0latency_median: 74
diseases
[
  "childhood B-cell precursor ALL",
  "B/myeloid MPAL"
]
breakpoints
[
  {
    "id": "bp-tcf3-exon13-znf384-exon3",
    "gene5_exon_last": 13,
    "gene3_exon_first": 3,
    "frequency_pct": null,
    "population": "7 patients with in-frame exon 13 to exon 3 joining in the Hirabayashi 291-patient cohort; an exon 11 to exon 3 junction is also reported",
    "domains_retained_5": [
      "AD1 (9aaTAD, p300/CBP-binding)",
      "NLS (nuclear localisation signal)",
      "LZ (leucine-zipper region)"
    ],
    "domains_lost_5": [
      "bHLH (basic helix-loop-helix)"
    ],
    "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": "AD2 is deliberately NOT listed although the literature places it in the retained region: this graph's protein-tcf3-domains records its residue numbers as null, and naming a null-coordinate domain produces a lane the viewer cannot draw. The retained AD1, NLS and LZ all lie below residue 530, so their coordinates are isoform-independent between E12 and E47 per this project's TCF3 registry row. Hirabayashi's own wording is that the predicted protein 'does not have the principal functional domains of wild-type TCF3', which is broader than the bHLH alone - so the retained list above is the coordinate-resolvable subset, not a functional claim.",
    "conflicting_report": "Dickerson et al.'s construct Methods state TCF3 exon 13 to ZNF384 exon 5, against exon 3 from every other source and for every other partner. Unresolved.",
    "phenotype_notes": "younger patients (median 5 years), higher WBC, poorer steroid response, higher relapse frequency, 46% allocated high risk"
  },
  {
    "id": "bp-tcf3-exon11-znf384-exon3",
    "gene5_exon_last": 11,
    "gene3_exon_first": 3,
    "frequency_pct": null,
    "population": "reported in the same cohort alongside the exon 13 junction; the number of patients with each was not recoverable by this beat",
    "domains_retained_5": [
      "AD1 (9aaTAD, p300/CBP-binding)",
      "NLS (nuclear localisation signal)"
    ],
    "domains_lost_5": [
      "bHLH (basic helix-loop-helix)"
    ],
    "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 exon-11 junction retains a shorter TCF3 fragment than the exon-13 junction, but NO source this beat read maps either junction to a residue, so whether the leucine zipper at 389-425 survives the exon-11 break is UNKNOWN and LZ is deliberately omitted from this variant's retained list rather than guessed."
  }
]
measurements — the numbers behind the chart
[
  {
    "quantity": "proportion",
    "subject": "TCF3 among partner-typed ZNF384 fusions in childhood B-ALL",
    "value": 0.31,
    "units": "fraction",
    "n": 193,
    "universe": "193 of 218 Ponte di Legno cases with partner-gene information; TCF3 60",
    "null": 0.43,
    "note": "the null is EP300's share in the same series - EP300 is commoner, and the TCF3 share is the one that carries the adverse features"
  },
  {
    "quantity": "proportion",
    "subject": "TCF3-ZNF384 among unselected childhood BCP-ALL",
    "value": 0.024,
    "units": "fraction",
    "n": 291,
    "universe": "291 children with BCP-ALL enrolled in a single Japanese clinical trial; 7 patients. Within the B-other subgroup of that cohort the figure is 5.4%.",
    "null": null,
    "note": "the 2.4% is against a whole-BCP-ALL denominator and is the usable one; the 5.4% B-other figure is inflated by construction because B-other is defined by standard genetics having failed"
  },
  {
    "quantity": "age_median",
    "subject": "TCF3::ZNF384 patients at diagnosis",
    "value": 5,
    "units": "years",
    "n": null,
    "universe": "Ponte di Legno series of 218 children; P < 0.001 for the difference by partner gene",
    "null": 11,
    "note": "null is EP300::ZNF384 at median 11 years"
  },
  {
    "quantity": "proportion",
    "subject": "TCF3::ZNF384 patients MRD-positive at end of induction",
    "value": 0.32,
    "units": "fraction",
    "n": 19,
    "universe": "19 TCF3::ZNF384 patients with MRD data among 77 with MRD data in the Ponte di Legno series; MRD negative defined as below 0.01%",
    "null": 0.16,
    "note": "null is the EP300::ZNF384 arm (5 of 32, 16%). The difference across all partner groups was NOT significant, p = 0.491, and only 77 of 218 patients had MRD data."
  },
  {
    "quantity": "event_free_survival_5y",
    "subject": "TCF3-ZNF384 childhood BCP-ALL",
    "value": 0.5,
    "units": "fraction at 5 years",
    "n": 15,
    "universe": "the assembled TCF3-ZNF384 group in Hirabayashi 2017",
    "null": 0.733,
    "note": "null is the ZNF384-wild-type comparator at 73.3%. P = 0.35 - NOT SIGNIFICANT. This figure is widely quoted; the significance is not."
  },
  {
    "quantity": "latency_median",
    "subject": "lethal disease in NSG-SGM3 from human CD34+ cells expressing TCF3::ZNF384",
    "value": 74,
    "units": "days",
    "n": null,
    "universe": "human CD34+ HSPCs; wild-type ZNF384 and the other ZNF384 fusions produced no lethal disease in this system",
    "null": null,
    "note": "the negative arms include EP300::ZNF384, which is the positive arm in the same paper's mouse HSPC assay"
  }
]
How this could bite us back
  • STEROID RESISTANCE IS A BITE-BACK AGAINST THE BACKBONE, NOT A NOVEL AGENT. Glucocorticoids are the spine of every paediatric ALL protocol in this graph's drug list, and this fusion confers resistance to them in a controlled experiment. The implication is not that a new drug is needed but that the existing one is weaker in this group - and steroid response is itself used to allocate risk, which means the fusion partly determines the treatment it then resists.
  • CD19-directed immunotherapy carries the lineage-switch risk specifically for this partner. The Ponte di Legno authors report that lineage switch from ALL to AML after CAR-T has been observed in a patient with TCF3-ZNF384, and advise that anti-CD19 CAR-T 'should be used carefully, if at all, in patients with ZNF384 rearrangements'. This graph already records the same mechanism for KMT2A.
  • Late relapse defeats standard follow-up. The same group reports two patients with TCF3-ZNF384-positive ALL relapsing more than 10 years after diagnosis. A subtype with a decade-plus relapse tail cannot be evaluated on 5-year endpoints, and the 5-year EFS figures quoted for it are therefore optimistic by construction.
  • A CCND2-directed idea inherits the same problem as every cyclin-D strategy: CDK4/6 inhibition hits a cell-cycle module the normal marrow needs, and the fusion did not confer a growth advantage in the one experiment that looked, so the therapeutic rationale would rest on the apoptosis phenotype rather than on proliferation.
Still open
  1. Does the TCF3 fragment alone, with no ZNF384, confer dexamethasone resistance? Kusano et al. show the resistance is partner-specific, but the control that distinguishes 'the TCF3 activation domain does this' from 'the chimera does this' is the TCF3 fragment on its own. Martini et al. ran exactly that control for the TET partners in 2002 and found the partner fragment alone gave a normal phenotype, so the experiment is neither novel nor hard.
  2. Which ZNF384 exon is really used - 3 or 5? Dickerson's construct methods say exon 5 for this fusion alone, against exon 3 from Hirabayashi, Gocho, Nyquist and McClure. If exon 5 is real in some patients, the 'entire ZNF384 coding region is retained' description of the family has an exception in its second commonest member.
  3. Is CCND2 the mechanism of the steroid resistance, or a correlate? The same abstract reports no growth advantage, which is not what a functional cyclin D2 increase predicts. A CCND2 knockdown in TCF3::ZNF384-transduced cells, scored on dexamethasone apoptosis, would settle it.
  4. Why do the clinical ranking and the mouse potency ranking invert? TCF3::ZNF384 has the worse patient outcome and is negative in the mouse HSPC assay where EP300::ZNF384 is positive; TCF3::ZNF384 is the only positive in the human CD34+ assay. If the human assay is the better model, the clinical and experimental rankings agree and the mouse result is the outlier - but nobody has said so in print and this beat is not entitled to.
Every source behind this entry (6)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 291 children with BCP-ALL, TCF3-ZNF384 was the most recurrent ZNF384 fusion at 2.4% of the cohort (7 patients, 5.4% within the B-other subgroup); exon 11 of TCF3 was fused to exon 3 of ZNF384, and exon 13 of TCF3 to exon 3 of ZNF384 with in-frame joining in 7 patients; the predicted protein from all of the fusions does not have the principal functional domains of wild-type TCF3, whereas it retains almost all of the part of the ZNF384 protein. TCF3-ZNF384-positive patients showed a significantly poorer steroid response than ZNF384-wild-type patients (P < 0.01), 5 of 15 (33.3%) relapsed in the bone marrow, and 5-year event-free survival was 50.0% versus 73.3% (P = 0.35).

    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

    **THE NUMBERS IN THIS ITEM COME FROM A PAGE-FETCH SUMMARISATION OF THE PMC RENDERING OF PMC5210242, NOT FROM READING THE PDF** - Europe PMC and NCBI efetch both refuse the full XML for this publisher. Two denominators are mixed in the source and must not be conflated: '7 patients' is the count within the 291-patient trial cohort, while '5 out of 15' and the survival figures are drawn from a larger assembled TCF3-ZNF384 group. The 5-year EFS difference is NOT significant (P = 0.35); only the steroid-response difference reaches P < 0.01. The two breakpoint statements (exon 11 and exon 13, both to ZNF384 exon 3) are reported as extracted and this beat could not confirm how many patients had each.

    Evidence that argues against this (1)
    • The EFS comparison in this study does not reach significance (50.0% vs 73.3%, P = 0.35) even though it is often cited alongside the relapse and steroid findings as if the three were one result. The larger Ponte di Legno series later found no separately reported EFS figure for TCF3::ZNF384 at all - its published contrast is EP300 versus all remaining patients - so the most-cited clinical claim about this fusion rests on a non-significant difference in a 15-patient group plus a significant steroid-response difference, and not on a survival result. Hirabayashi S et al. 2021
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the Ponte di Legno series of 218 children with ZNF384-rearranged B-ALL, TCF3 was the second commonest partner at 60 of 193 partner-typed cases (31%); TCF3-ZNF384 patients were younger (median 5 years versus 11 for EP300-ZNF384, P < 0.001), 46% were allocated to the high-risk group because of high leukocyte counts and poor initial responses versus 28% of EP300-ZNF384 patients, and 6 of 19 (32%) were MRD-positive at the end of induction against 5 of 32 (16%) for EP300-ZNF384 (overall MRD-positive 18 of 77, 23%, median 2.54%, range 0.14-25.6%; p = 0.491 across partner groups).

    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. expected_absent_terms names 'steroid' because this paper contains NO steroid-response data - the words steroid, prednis* and dexamethasone return zero hits in the complete fetched text - so the steroid claim about TCF3::ZNF384 must be attributed to Hirabayashi 2017 and Kusano 2025, never to this cohort. MRD data were available for only 77 of 218 patients and the partner-group difference did not reach significance (p = 0.491), so the 32%-versus-16% contrast is descriptive. This is a retrospective consortium collection spanning 1992-2018 with no total B-ALL denominator, MPAL cases excluded by design.

    Evidence that argues against this (1)
    • The MRD difference this item quotes is NOT statistically significant across partner groups (p = 0.491) and rests on 19 TCF3 patients with MRD data out of 60. The high-risk allocation difference is also not an outcome - it is a treatment decision made partly on the basis of the same initial response that is being used to characterise the fusion, so 46% versus 28% high-risk is partly definitional rather than prognostic. The one difference in this paper that is significant and independent is age. Hirabayashi S et al. 2021
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In human CD34+ haematopoietic stem and progenitor cells transplanted into NSG-SGM3 mice, TCF3::ZNF384 - but not wild-type ZNF384 and not the other ZNF384 fusion oncoproteins tested - produced lethal disease with a median latency of 74 days, with expansion of cells expressing CD33, CD19 and MPO consistent with the immunophenotype of human B/myeloid MPAL. TCF3 was also the commonest partner among ZNF384-rearranged B/myeloid MPAL cases in the St Jude ALAL cohort (n = 8 of 15 partner-named cases).

    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; mouse numbers and statistics were not extracted. NSG-SGM3 mice express human SCF, GM-CSF and IL-3 and are known to favour myeloid output, which is a direct confounder for a B/myeloid readout. The MPAL partner counts come from a different paper (Alexander et al. 2018, PMID 30209392) and are combined into this claim, which is flagged here rather than left implicit.

    Evidence that argues against this (1)
    • In the SAME paper, the mouse HSPC assay gives the opposite partner ranking: 'only EP300::ZNF384 was able to promote formation of serially transplantable leukemia without concomitant genomic alterations', and EP300::ZNF384 is the fusion with the BEST clinical outcome. So across the two systems in one paper, each of the two commonest partners is the unique positive in one assay and negative in the other. No statement of the form 'TCF3::ZNF384 is the more potent fusion' is supported without naming the assay and the species. Dickerson KM et al. 2022
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Retroviral transduction of BCP-ALL cell lines with TCF3::ZNF384 and EP300::ZNF384 side by side showed that BOTH fusions down-regulate CD10, but only TCF3::ZNF384-expressing cells were resistant to dexamethasone-induced apoptosis and only TCF3::ZNF384 significantly enhanced CCND2 expression; no growth advantage was observed. The authors conclude that TCF3::ZNF384 exhibits a fusion-partner-dependent function distinct from EP300::ZNF384, over and above the common ZNF384-rearrangement functions that produce the lineage-ambiguous phenotype.

    TCF3::ZNF384 induces steroid resistance in B-cell precursor acute lymphoblastic leukemia cells. · Kusano S, Ueno-Yokohata H, Hori M, Ishibashi T, Fujimura J, Shimizu T, Ohki K, Kiyokawa N · Pediatr Int 2025
    PMID 40391410DOI
    What this evidence cannot show

    ABSTRACT ONLY. The cell lines used are not named in the abstract, nor the number of independent transductions, nor the magnitude of the dexamethasone-resistance or CCND2 effects - so this item supports a DIRECTION, not a size. The system is retroviral over-expression in established BCP-ALL lines, which is not the cell of origin identified for this subtype (an HSC) and carries expression levels that need not match a patient's. Only two of the family's twenty-plus partners were compared, so 'partner-dependent' here means 'these two differ', not that every partner differs.

    Evidence that argues against this (1)
    • The abstract states in the same sentence as the CCND2 result that 'no growth advantage was observed', so the mechanism proposed for the steroid resistance does not produce the proliferative phenotype one would expect from a cyclin D2 increase. That is a real internal tension in the result. The wider caution is that this is an over-expression experiment in a cell line: TCF3::ZNF384 contains most of the TCF3 N-terminal activation region, and over-expressing a strong activation domain can confer apoptosis resistance in a way that does not depend on the ZNF384 half at all. The discriminating control - the TCF3 fragment alone, without ZNF384 - is not mentioned in the abstract and this beat could not check whether it was done. Kusano S et al. 2025
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB P15923 (TFE2_HUMAN, TCF3/E2A) is a reviewed 654 aa human entry whose only annotated region features are a bHLH domain at 549-602 and a leucine-zipper region at 389-425; UniProt annotates no AD1 or AD2 activation domain on this record. Isoforms E12 (P15923-1, 654 aa) and E47 (P15923-2, 651 aa) are identical through residue 529 and diverge only in the bHLH-encoding block, so any coordinate below 530 is isoform-independent.

    What this evidence cannot show

    A reviewed gene_exact search for TCF3 returns TWO human records: P15923 (TFE2_HUMAN, 654 aa, the E2A protein) and Q9HCS4 (TF7L1_HUMAN, TCF7L1, 588 aa, which carries TCF3 as a synonym). The 654 aa record is the correct one for this fusion. Note that Hirabayashi's 'the predicted protein does not have the principal functional domains of wild-type TCF3' is a statement about AD1/AD2/bHLH in the literature's vocabulary, and UniProt annotates NO AD1 or AD2 at all on P15923 - so the retained activation domains cannot be given residue boundaries from the database and are not given any here.

    Evidence that argues against this (1)
    • The most important limitation is the one the database creates rather than resolves: because UniProt annotates no activation domain on TCF3, the claim that TCF3::ZNF384 retains TCF3's activation domains - which is the mechanistic heart of the partner-dependent steroid-resistance story - cannot be checked against a coordinate. It rests on the literature's AD1/AD2 convention, which this project's own collisions registry records as 'a literature construct' that 'cannot be resolved from the database'. Any domain figure drawn for this fusion from UniProt will show an empty N-terminal region where the mechanism is supposed to live. Oncofusion Knowledge Graph / UniProt Consortium 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Dickerson et al.'s expression constructs record the TCF3::ZNF384 breakpoint as TCF3 exon 13 fused to ZNF384 exon 5, while recording EP300::ZNF384 as exon 6 to exon 3, TAF15::ZNF384 as exon 6 to exon 3 and CREBBP::ZNF384 as exon 6 to exon 3 - so the ZNF384 exon differs for TCF3 alone in that methods list.

    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

    This is a Methods sentence describing the constructs the authors built and validated, not a survey of patient breakpoints. It is recorded because it conflicts with every other source on the ZNF384 exon for TCF3::ZNF384, and a conflict in a Methods list is exactly the kind of thing that gets copied forward silently. This beat did NOT resolve it and offers no verdict on which number is right.

    Evidence that argues against this (1)
    • Three independent sources put the ZNF384 breakpoint at exon 3 or in the intron immediately before it, for every partner examined. Gocho states 'exon 3 (the first coding exon)' for EP300 and adds that 'the fusion point of ZNF384 is the same as those of previously reported ZNF384-related fusion genes between TAF15 and E2A genes' - E2A being TCF3. Nyquist found the ZNF384 breakpoint 'in the same intron in both cases, between exons 2 and 3'. McClure found ZNF384 exon 3 in 8 of 9 cases and exon 2 in the ninth. The weight of evidence is against exon 5, and the most likely explanations are a different transcript numbering or a typographical error - but this beat did not verify either, and a Methods discrepancy in a major paper is worth recording rather than quietly discarding. Gocho Y et al. 2015
Connected to