Oncofusion

start / graph / gene-zranb2

Geneweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q20-proximity-interactome

ZRANB2 - the only bait in any published proximity experiment that reaches AFF1, AFF4, ELL, MLLT1, CCNT1, KMT2A and DOT1L at once

also called ZNF265 · ZIS · zinc finger RANBP2-type containing 2 · Zinc finger, splicing

ZRANB2 is a protein that helps decide how RNA is spliced. When researchers tagged it and looked at everything it sits near, the list included both halves of the KMT2A::AFF1 fusion's parent proteins and most of the elongation machinery they work with.

The same thing, for a scientist

ZRANB2 (O95218) is a zinc-finger splicing regulator. As a BioID2 bait in K562 it recovers **566 preys**, and among them are **AFF1, AFF4, ELL, MLLT1, CCNT1, KMT2A, DOT1L, BRD4, WDR5, ASH2L, NELFE, POLR2A, ICE2, CDC73, RTF1, CTR9, SETD1A, KMT2D, HCFC1, CBX8, ARID1A and CHD8** [E1]. It is the **only bait in the entire IntAct proximity record - across all five publications that contain any of these proteins - that reaches both SEC scaffolds, the P-TEFb cyclin, ELL, ENL, KMT2A and DOT1L in one experiment** [E2]. That breadth is simultaneously the reason it is interesting and the reason to distrust it: a bait with 566 preys is sampling a large part of the nucleoplasm.

Go deeper
Why this bait, of all baits

The dispatch that created this beat flagged ZRANB2 from IntAct as an uncited AFF1 partner. It turns out to be more than that. Of the 50 baits in the K562 map, ZRANB2 is the one whose neighbourhood contains the largest coherent slice of this graph's subject matter 1:

  • both SEC scaffolds: AFF1 and AFF4
  • the elongation module: ELL, MLLT1/ENL, CCNT1, NELFE, POLR2A, ELOA, TCEA1
  • the KMT2A side: KMT2A itself, plus WDR5, ASH2L, HCFC1, SETD1A, KMT2D
  • DOT1L
  • PAF-complex members CDC73, CTR9, RTF1
  • Polycomb CBX8 and chromatin remodellers ARID1A, CHD8

What it does not recover is MEN1 1. The one protein this project's therapeutic hypothesis rests on is absent from the neighbourhood that contains everything else.

And why to distrust it

ZRANB2's 566 preys are roughly a fifth of every distinct prey in the dataset. A list that long recovers a coherent slice of nuclear biology whatever the bait is, and the correct comparison is not 'does ZRANB2 reach the SEC' but 'does ZRANB2 reach the SEC more than a comparable bait does' 1.

On that comparison it is genuinely unusual for AFF1: only four of fifty baits reach AFF1 at all, and ZRANB2 is one of them. It is much less unusual for KMT2A, which 25 of 50 baits reach.

And nothing in this record is a binding claim. ZRANB2 has no published co-immunoprecipitation with AFF1, no structure with it and no functional experiment in KMT2A-rearranged cells 3.

What is known about ZRANB2 independently

ZRANB2 is a zinc-finger protein that regulates alternative splicing; its two RanBP2-type zinc fingers bind single-stranded RNA, and displacement of their zinc by arsenite alters splicing genome-wide, which is the mechanism proposed for arsenic-induced carcinogenesis 3.

That is a thin base. There is no ZRANB2 leukaemia literature, and this entry does not claim one.

Retained region

Can this be mapped to a region the fusion retains? No, and the reason is structural, not a gap in the reading. AFF1 was a PREY in this experiment, not a bait. What was labelled is endogenous, full-length, wild-type AFF1 in a cell with no KMT2A rearrangement. Proximity labelling biotinylates accessible lysines anywhere on the prey and the mass spectrometry was not run for site-level biotinylation, so the record carries no positional information at all. Nothing here distinguishes a contact with the AFF1 N-terminus, which KMT2A::AFF1 discards, from one with the C-terminal half it keeps. Any entry that infers otherwise is over-reading the data.

The splicing-factor compartment, and a pointed cell-type absence (Q87, 2026-09-21)

Subcellular. Nucleus, and nuclear residence REQUIRES the RS domain (ZNF265-EGFP deletion constructs, HT-1080; UniProt O95218 ECO:0000269 PubMed:11448987, which really is the paper for the fact). HPA v25.1 IF, reliability Enhanced: nucleoplasm, single main location.

Subnuclear: the splicing-factor compartment, nuclear speckles being the best-supported single name. Confocal colocalisation with SC35 (speckles) and U1-70K, and also with SMN (gems) and with p300 and YY1. Speckles and gems are distinct bodies and this work overlaps markers of both without resolving between them, so the compartment is recorded broadly.

Cell type: a genuine absence, and a pointed one. ZRANB2 is the only bait in any published proximity experiment reaching AFF1, AFF4, ELL and MEN1, and it has never been examined in a haematopoietic cell of any kind. Cell state: nothing published at all.

Shown for fusion. No, in the strong sense. The retired symbol ZNF265 was searched deliberately - the only localisation paper publishes under it.

Facts and figures
uniprotO95218
gene symbolZRANB2
length AA330
uniprot resolution noteAccession taken from the IntAct MITAB record's interactor A column; primary gene name and length read back from the UniProt REST API on 2026-09-19 as ZRANB2, 330 aa.
n preys total566
methodBioID2 in K562 (PSI-MI MI:1314)
recovers men1false
subnuclear localizationTHE SPLICING-FACTOR COMPARTMENT - NUCLEAR SPECKLES as the best-supported single name. Confocal microscopy shows ZRANB2/ZNF265 colocalising with the speckle marker SC35 and with U1-70K, and also with SMN (a gems marker) and with p300 and YY1; nuclear localisation requires the RS domain (PMID 11448987). Speckles and gems are distinct bodies and this work overlaps markers of both without resolving between them, so the compartment is recorded as the splicing-factor compartment rather than as speckles alone.
cell stateNone recorded and none published. ZRANB2 has no state-scoped result of any kind - no cell-cycle dependence, no differentiation dependence, no stress relocalisation. Recorded as an absence rather than left blank.
shown for fusionNO - nobody has looked in any fusion. Measured 2026-09-21: (ZRANB2[tiab] OR ZNF265[tiab]) AND the eight-term fusion clause AND the ten-term localisation clause returns 0 records. The retired symbol ZNF265 was included deliberately: the founding localisation paper publishes under it and would be invisible to a current-symbol-only query.
proximity baits recovered
[
  "AFF1",
  "AFF4",
  "ELL",
  "MLLT1",
  "CCNT1",
  "KMT2A",
  "DOT1L",
  "BRD4",
  "WDR5",
  "ASH2L",
  "NELFE",
  "POLR2A",
  "ICE2",
  "CDC73",
  "CTR9",
  "RTF1",
  "SETD1A",
  "KMT2D",
  "HCFC1",
  "CBX8",
  "ARID1A",
  "CHD8"
]
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.2778,
  "chronos_mean_other_leukaemia": -0.4377,
  "difference": 0.1599,
  "pan_essentiality_all_lines": -0.4111,
  "selectivity_rank": 17497,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
localization
{
  "subcellular": "NUCLEUS, and nuclear residence REQUIRES the RS domain - a mapped determinant, not a default. UniProt O95218 lists one location term, Nucleus, on ECO:0000269 PubMed:11448987, and that citation is genuinely the paper for the fact (rule 91 satisfied): ZNF265-EGFP fusion constructs transfected into HT-1080 cells show that nuclear localisation of ZNF265 requires the RS domain. HPA v25.1 IF, reliability Enhanced, gives a single main location, Nucleoplasm. No cytoplasmic pool is recorded by either source.",
  "subnuclear": "THE SPLICING-FACTOR COMPARTMENT, established by four-way colocalisation rather than by a single marker. Confocal microscopy shows ZRANB2/ZNF265 colocalising with SC35 - the canonical NUCLEAR SPECKLE marker - with the snRNP protein U1-70K, with the motor neuron gene product SMN, whose own nuclear bodies are GEMS, and with the transcriptosomal components p300 and YY1 (PMID 11448987). Colocalisation with both SC35 and SMN is worth stating precisely: speckles and gems are adjacent but distinct bodies, and this paper reports overlap with markers of both without resolving which body ZRANB2 belongs to. The honest answer is the splicing-factor compartment broadly, with speckles the best-supported single name.",
  "the_zinc_finger_reads_single_stranded_nucleic_acid": "ZRANB2's two RanBP2-type zinc fingers and its RS domain together make it an RS-domain splicing regulator, which is why the speckle address is the expected one - but this entry's reason for existing is different: ZRANB2 is the only bait in any published proximity experiment that reaches AFF1, AFF4 and ELL. A speckle address and a proximity relationship with the elongation axis are two separate facts and neither explains the other.",
  "how_shown": "confocal immunofluorescence colocalisation with SC35, U1-70K, SMN, p300 and YY1; RS-domain deletion in ZNF265-EGFP constructs in HT-1080 cells (PMID 11448987)",
  "tissue_of_evidence": "HEK 293 and HT-1080 human lines. Nothing haematopoietic and nothing about a fusion, despite this entry's proximity result being the reason the gene is here.",
  "node": "localization-nuclear-speckles-kmt2a-r"
}
cell type
{
  "haematopoietic_answer": "NONE. No ZRANB2 localisation or expression experiment has been done in a haematopoietic cell. The compartment work is in HEK 293 and HT-1080. HPA v25.1 records low RNA tissue specificity, i.e. broad expression, which is a transcript measurement and not a cell-type finding.",
  "bounded": "This is a genuine cell-type absence, and it is notable because ZRANB2 is, by this entry's own account, the only bait in any published proximity experiment that reaches AFF1, AFF4, ELL and MEN1 - so the protein that comes closest to the index complex in a proximity map has never been looked at in a blood cell."
}
How this could bite us back
  • This is a splicing factor found near an elongation complex by a method that labels neighbourhoods. Promoting it to a target would be the exact error the method's own literature warns against.
  • ZRANB2 is broadly expressed and its zinc fingers are a common fold. Any ligand aimed at them is a poor bet for selectivity.
  • If ZRANB2 does matter, its arsenite sensitivity means an existing leukaemia drug already perturbs it non-selectively, which complicates attributing any effect to it.
Still open
  1. Does ZRANB2 co-immunoprecipitate with AFF1 or with KMT2A::AFF1 in a t(4;11) line? No such experiment exists.
  2. Is ZRANB2 required in KMT2A-rearranged cells? It was not in this project's 79-gene RNA-layer DepMap panel and should be added.
  3. Does ZRANB2's absence of MEN1 recovery reproduce, and if so does it mean menin sits in a different sub-neighbourhood from the rest of the KMT2A complex?
  4. Do the arsenite-sensitive zinc fingers matter here, i.e. does arsenic trioxide - a drug already used in leukaemia - perturb the AFF1 neighbourhood through ZRANB2?
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    As a BioID2 bait in K562, ZRANB2 (O95218) recovers 566 distinct preys, including AFF1, AFF4, ELL, MLLT1, CCNT1, KMT2A, DOT1L, BRD4, WDR5, ASH2L, NELFE, POLR2A, ICE2, CDC73, CTR9, RTF1, SETD1A, KMT2D, HCFC1, CBX8, ARID1A and CHD8, but NOT MEN1.

    What this evidence cannot show

    Retrieved from IntAct 2026-09-19. A 566-prey neighbourhood is about a fifth of the dataset's entire prey space, so co-occurrence in it is weak evidence about any individual pair. The bait is an overexpressed BioID2 fusion in a cell line with no KMT2A rearrangement. Absence of MEN1 from a hit list is not evidence that menin is elsewhere - it may simply be below the detection threshold.

    Evidence that argues against this (2)
    • In the largest published human BioID survey (192 baits, HEK293) a prey's bait profile is dominated by subcellular compartment rather than by binding. A nuclear elongation factor recovered by nuclear RNA-processing baits is what shared compartment occupancy alone predicts. Go CD et al. 2021
    • The founding BioID report describes the reagent as labelling proteins that are near the bait without binding it. Promiscuous labelling of the vicinity is the designed behaviour, so a hit list is a neighbourhood census before it is anything else. Roux KJ et al. 2012
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Across all five publications in IntAct that contribute proximity-dependent biotin identification records for any of AFF1, AFF4, ELL, MLLT1, CDK9, CCNT1, KMT2A, MEN1 or DOT1L, ZRANB2 is the only bait whose neighbourhood contains both SEC scaffolds together with ELL, MLLT1, CCNT1, KMT2A and DOT1L.

    What this evidence cannot show

    A superlative from a sample: the sample is the union of five depositions curated into one database, four of which had nothing to do with transcription elongation. It says ZRANB2 is the broadest such bait among the baits anyone happened to run, not that it is the broadest possible. The comparison is also unfair to baits with smaller prey lists.

    Evidence that argues against this (1)
    • The comparison is confounded by prey-list length. Baits with 600-800 preys - GPKOW, PPIL4, HNRNPC, SBDS - reach large overlapping slices of the same biology, and PPIL4 reaches more of the KMT2A-side proteins than ZRANB2 does. Breadth is a property of the pulldown, not of the protein. IntAct molecular interaction database (EMBL-EBI) 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    ZRANB2 is a zinc-finger protein regulating alternative splicing; arsenite displaces zinc from its RanBP2-type zinc fingers and alters splicing, the proposed mechanism for arsenic-induced skin carcinogenesis.

    PMID 32274925DOI
    What this evidence cannot show

    Toxicology paper in keratinocytes, not haematopoietic cells and not leukaemia. Abstract read; full text not read. This is background on the protein, not support for any claim about AFF1.

    Evidence that argues against this (1)
    • A splicing regulator's presence in a transcription-elongation neighbourhood is exactly what compartment-driven proximity labelling produces, because splicing is co-transcriptional. The independent biology of ZRANB2 makes the proximity hit MORE explicable without any direct interaction, not less. Go CD et al. 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ZRANB2 (published as ZNF265) is a nuclear RS-domain splicing regulator whose nuclear localisation REQUIRES the RS domain - shown with ZNF265-EGFP constructs in HT-1080 cells - and which colocalises by confocal microscopy with the nuclear speckle marker SC35, the snRNP protein U1-70K, the gems component SMN, and the transcriptosomal components p300 and YY1. It also immunoprecipitates from splicing extracts in association with mRNA and interacts with U1-70K and U2AF35. UniProt O95218's single location term rests on this paper (ECO:0000269 PubMed:11448987) and it is genuinely the paper for the fact.

    ZNF265--a novel spliceosomal protein able to induce alternative splicing. · Adams DJ, van der Weyden L, Mayeda A, Stamm S, Morris BJ, Rasko JE · J Cell Biol 2001
    PMID 11448987DOI
    What this evidence cannot show

    HEK 293 and HT-1080 human lines, 2001, with EGFP-tagged constructs for the RS-domain mapping. The paper reports colocalisation with markers of two DIFFERENT nuclear bodies - SC35 for speckles and SMN for gems - and does not resolve which one ZRANB2 occupies, so 'nuclear speckles' is the best-supported single name rather than a settled assignment. Nothing haematopoietic and nothing about any fusion.

    Evidence that argues against this (1)
    • Colocalisation with SC35 is the weakest common form of subnuclear evidence and should not be read as residence, because nuclear speckles are large, bright and adjacent to a great deal of the nucleoplasm - the same objection this graph raises against reading punctate overexpression signals as compartments. The sharpest version of the problem is visible in a sibling protein: AQR/IBP160 is also called a nuclear speckle protein, but the evidence there is FUNCTIONAL - depleting it releases MALAT-1 from speckles - and no equivalent functional test has ever been done for ZRANB2. Until one is, ZRANB2's speckle assignment rests on an image, and the additional SMN overlap means the image is compatible with more than one body. Miyagawa R et al. 2012
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    No ZRANB2 localisation result exists in any KMT2A-fusion context, and no ZRANB2 experiment of any kind has been done in a haematopoietic cell. Measured 2026-09-21: (ZRANB2[tiab] OR ZNF265[tiab]) AND (KMT2A[tiab] OR MLL[tiab] OR "MLL-AF4"[tiab] OR "MLL-AF9"[tiab] OR "MLL-ENL"[tiab] OR "MLL-AF10"[tiab] OR "t(4;11)"[tiab] OR "t(9;11)"[tiab]) AND (localization[tiab] OR localisation[tiab] OR subnuclear[tiab] OR nuclear[tiab] OR nucleolar[tiab] OR speckle[tiab] OR speckles[tiab] OR "PML bodies"[tiab] OR chromatin-bound[tiab] OR cytoplasmic[tiab]) returns 0 records. The retired symbol ZNF265 was included because the founding localisation paper publishes under it.

    What this evidence cannot show

    Bounded by titles and abstracts only (rule 80). The bound is over the gene symbols; a proximity-labelling paper framed at the level of a complex would be unreachable from either of them (rule 64), and the proximity result this entry rests on is itself of that kind. Rule-67 positive control for this query SHAPE: the identical construction, run over all sixteen genes of this tranche on the same day, returned non-zero for seven of them - H2AZ1 1, EED 2, CHAF1B 1, CEBPA 3, TIFAB 1, SAMHD1 1, PROM1 1 - so the instrument can return a one. The querytranslation was read back on every run and confirms the quoted cytogenetic terms expand to "t 4 11"[Title/Abstract] and "t 9 11"[Title/Abstract] and not to a UID filter (rule 95); every term carries its own [tiab] tag rather than sitting inside a parenthesised group (rule 94).

    Evidence that argues against this (1)
    • The absence is more pointed than most in this tranche and is worth arguing as a gap rather than recording as a fact. ZRANB2 is, on this entry's own account, the ONLY bait in any published proximity experiment that reaches AFF1, AFF4 and ELL - the core of the elongation complex this whole graph is about - and yet the protein has never been imaged in a blood cell, let alone a KMT2A-rearranged one. That combination is a specific, cheap and well-motivated experiment rather than an open-ended gap. Note also that a second protein in this same tranche, AQR, sits in BOTH the speckle compartment and an AFF4-ELL2-containing promoter-proximal chromatin assembly; if that dual address generalises, then ZRANB2's speckle assignment may be hiding a chromatin one that nobody has looked for. Xiang Y et al. 2026
Connected to