Oncofusion

start / graph / gene-tdp2

Genestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q43-break-repair-machinery

TDP2 - the protein-adduct remover named in this graph's own mechanism entry and until now having no entry of its own

also called TDP2 · TTRAP · EAP2 · tyrosyl-DNA phosphodiesterase 2 · 5'-tyrosyl-DNA phosphodiesterase

When a DNA-untangling enzyme gets stuck welded to a broken chromosome, TDP2 is the protein that cuts the weld so the break can be glued back cleanly. This closes a gap in the graph: the mechanism entry that names TDP2 has existed for cycles and TDP2 itself did not.

The same thing, for a scientist

TDP2 (O95551, 362 aa, also TTRAP/EAP2) hydrolyses the 5'-phosphotyrosyl bond that tethers topoisomerase II to a DNA end in a trapped cleavage complex, liberating a 4-nt 5'-cohesive terminus that classical NHEJ can religate without further processing [E2, E3]. It is the major and possibly the only 5'-tyrosyl-DNA phosphodiesterase in vertebrate cells [E3]. At the locus this graph cares about, TDP2 rejoins DSBs induced during transcription-dependent TOP2 activity at the MLL 'hotspot' and **suppresses** TOP2-induced chromosome rearrangements and translocations - so its loss makes matters worse, and it is not a prevention target [E2]. The nuance that makes it interesting is the junction-class dissociation already recorded in `mechanism-tdp2-repair-of-trapped-top2-complexes`: TDP2-dependent NHEJ is implicated in the formation of a rare translocation subclass characterised by 4 bp of perfect junction homology, the class reported previously for a therapy-related leukaemia translocation [E2]. **This entry is deliberately thin on mechanism** because that mechanism entry holds it; what this node adds is the identity (accession and length, asserted so a paralogue swap with TDP1 is catchable), the human loss-of-function phenotype, and the bounded statement of how little KMT2A-specific TDP2 literature exists.

Go deeper
What this entry adds that the mechanism entry does not

mechanism-tdp2-repair-of-trapped-top2-complexes is the substantive account and should be read first. This node exists for three reasons the mechanism entry cannot serve:

  1. A resolvable identity. TDP2 has a same-named, same-function, different-substrate paralogue in TDP1, and this graph has three standing rules about exactly that failure mode. O95551 is 362 aa; TDP1 is Q9NUW8 at 608 aa 1. The 246-residue difference is what a checker can see.
  2. A human loss-of-function phenotype. Biallelic TDP2 loss causes SCAR23, a recessive paediatric-onset spinocerebellar ataxia with drug-resistant epilepsy and intellectual disability 4. That is the toxicity floor under any TDP2-directed idea, and it is a developmental neurological floor rather than a haematological one, which is unusual for this beat.
  3. A counted absence. (MLL OR KMT2A) AND TDP2 returns 1 record on PubMed as of 2026-09-20, and that record is the Gomez-Herreros paper itself 5. The complex-form check required before writing such an absence was run: TDP2 AND topoisomerase AND translocation returns a small literature, none of it adding a second KMT2A-specific study.
Direction of effect, stated so nobody reads it backwards

TDP2 protects the KMT2A locus. Losing it increases breakage at KMT2A and increases etoposide-induced chromosome rearrangements and translocations 2. Any proposal of the form 'inhibit the repair enzyme to prevent the fusion' has the sign wrong here, and this graph's hyp-block-kmt2a-break-machinery-upstream already lists TDP2 among the targets whose blockade runs the wrong way.

The one place TDP2 activity is implicated in making a junction is narrow and specific: the 4-bp perfect-homology subclass, present in wild-type and absent in TDP2-null cells 2. That is a diagnostic signature, not a therapeutic handle.

No canonical compartment: isoform, stress and infection each move TDP2 (Q87, 2026-09-21)

Subcellular. Nucleus and cytoplasm, partitioned by ISOFORM: TDP2S has a mitochondrial targeting sequence and is enriched in mitochondria and cytosol; full-length TDP2 has the N-terminal NLS (PMID 29438979). UniProt O95551 also carries an infection-scoped comment - picornavirus infection relocalises TDP2 to cytoplasmic sites.

Subnuclear: PML NUCLEAR BODIES, and NUCLEOLAR CAVITIES under stress. Proteasome inhibition drives TDP2 into nucleolar cavities in a PML-dependent manner, where it controls rRNA precursor levels independently of its phosphodiesterase activity (PMID 21921940). HPA v25.1 IF gives 'Nuclear bodies' as a co-equal main location.

Three conditionals, not one address. Isoform, stress, infection. A single 'nucleus' record would lose all three, which is why none is written here.

Shown for fusion. No, in the strong sense - and this is the zero in the tranche that matters most, because this graph's own mechanism entry invokes TDP2 at the KMT2A breakpoint cluster region while none of TDP2's measured addresses is chromatin.

Facts and figures
uniprotO95551
length AA362
gene symbolTDP2
accession resolutiongene_exact:TDP2 AND organism_id:9606 AND reviewed:true returned exactly 1 row; gene_primary read back as TDP2, length 362, entry name TYDP2_HUMAN. Paralogue control Q9NUW8 / TDP1 / 608 aa resolved in the same call.
paralogueTDP1 (Q9NUW8, 608 aa) - acts on 3'-phosphotyrosyl TOP1 adducts, not 5'/TOP2
enzyme activity5'-tyrosyl-DNA phosphodiesterase; removes TOP2 from a trapped cleavage complex leaving a 4-nt 5'-cohesive end
direction of effect at kmt2aPROTECTIVE - loss increases breakage and translocation. Not a prevention target.
human loss of functionSCAR23: recessive paediatric-onset spinocerebellar ataxia, drug-resistant epilepsy, intellectual disability
forward reference closedmechanism-tdp2-repair-of-trapped-top2-complexes named TDP2 with no gene entry behind it
subnuclear localizationPML NUCLEAR BODIES, plus PML-DEPENDENT relocation to NUCLEOLAR CAVITIES under proteasome inhibition, where TDP2 controls rRNA precursor and processing-intermediate levels independently of its phosphodiesterase activity (PMID 21921940). HPA v25.1 IF gives Nuclear bodies as a co-equal main location with nucleoplasm, at reliability Supported. The subnuclear answer is conditional on stress and the nuclear/cytoplasmic answer is conditional on isoform, so no single compartment call is correct for this protein.
cell statePROTEOTOXIC / PROTEASOME-INHIBITED STRESS is the state that moves TDP2, and it is a real state with a named perturbation rather than a label: under proteasome inhibition the protein relocates to nucleolar cavities and takes on a ribosome-biogenesis role it does not have at baseline (PMID 21921940). That is directly relevant here because proteasome inhibition is a clinical intervention, and because this graph's TDP2 interest is topoisomerase-II-adduct removal - a different stress entirely, with no published compartment consequence.
shown for fusionNO - nobody has looked in any fusion. Measured 2026-09-21: (TDP2[tiab] OR TTRAP[tiab] OR EAPII[tiab]) AND the eight-term fusion clause AND the ten-term localisation clause returns 0 records. This matters more than most of the zeros in this tranche, because this graph's own mechanism entry invokes TDP2 for removing topoisomerase II adducts at the KMT2A breakpoint cluster region - so the compartment question is mechanistically live here and has never been asked.
kmt2a literature count
{
  "query": "(MLL OR KMT2A) AND TDP2",
  "count": 1,
  "date": "2026-09-20",
  "aliases_searched": [
    "TDP2",
    "TTRAP",
    "EAP2"
  ]
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.2786,
  "chronos_mean_other_leukaemia": -0.1991,
  "difference": -0.0795,
  "pan_essentiality_all_lines": -0.2362,
  "selectivity_rank": 1771,
  "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 CYTOPLASM, and the partition is set by WHICH PROTEIN IS MADE, not by a trafficking equilibrium - a single 'nucleus' answer for TDP2 would be wrong. An internal ribosome entry site in the TDP2 coding region drives an alternative translation start (PMID 30593505), and a short isoform TDP2S expressed from an alternative transcription start site carries a MITOCHONDRIAL TARGETING SEQUENCE and is enriched in mitochondria and cytosol, while FULL-LENGTH TDP2 carries the N-terminal nuclear localization signal and the ubiquitin-associated domain; both isoforms are present and active in mitochondria (PMID 29438979). UniProt O95551 records Nucleus, Nucleus/PML body, Nucleus/nucleolus and Cytoplasm, plus a second, infection-scoped comment in which picornavirus infection relocalises TDP2 to cytoplasmic sites. HPA v25.1 IF, reliability Supported, gives TWO co-equal main locations, Nucleoplasm and Nuclear bodies.",
  "subnuclear": "PML NUCLEAR BODIES at baseline, and NUCLEOLAR CAVITIES under stress - and the move between them is PML-dependent. TDP2 (as TTRAP) is described as a PML-nuclear-bodies-associated protein; on PROTEASOME INHIBITION it accumulates in NUCLEOLAR CAVITIES in a promyelocytic-leukaemia-protein-dependent manner, and there it controls levels of the ribosomal RNA precursor and processing intermediates - a function INDEPENDENT of its 5'-tyrosyl DNA phosphodiesterase activity (PMID 21921940). UniProt carries both terms with the note 'Localizes to nucleolar cavities following stress; localization to nucleolus is dependent on PML protein'. HPA's 'Nuclear bodies' main location is an independent antibody-based corroboration of the PML-body call.",
  "the_partition_is_conditional_three_ways": "(1) by ISOFORM - TDP2S has a mitochondrial targeting sequence, full-length TDP2 has an NLS; (2) by STRESS - proteasome inhibition drives the nuclear pool into nucleolar cavities, and that move requires PML; (3) by INFECTION - picornavirus infection relocalises TDP2 to cytoplasmic sites distinct from those holding the viral replication proteins. Recording 'nucleus' alone would lose all three.",
  "how_shown": "immunofluorescence with proteasome inhibition and PML dependence, plus rRNA precursor quantification and a phosphodiesterase-dead separation-of-function test (PMID 21921940); alternative transcription start mapping, MTS prediction and mitochondrial fractionation with isogenic TDP2-knockout DT40 cells (PMID 29438979); HPA antibody IF",
  "node": "localization-pml-nuclear-bodies-pmlrara"
}
cell type
{
  "haematopoietic_answer": "NONE MEASURED. No TDP2 localisation experiment has been done in a haematopoietic cell. The compartment work is in human epithelial and avian DT40 systems; DT40 is a chicken B-cell line, which is lymphoid but is not a human haematopoietic cell and was used for genetics rather than for imaging.",
  "expression_atlas": "HPA v25.1: RNA tissue specificity 'Tissue enhanced'; IF reliability Supported."
}
How this could bite us back
  • TDP2 inhibitors are being pursued as chemosensitisers to increase etoposide killing. In a patient population whose disease originated in an etoposide-type break, that is the same exposure with the adduct-removal step disabled, and the published direction of effect is that TDP2 loss increases translocation.
  • The human TDP2-null phenotype is neurological and paediatric. Any systemic TDP2-directed agent in a child inherits a developmental-neurology risk that would be difficult to detect within the timeframe of a leukaemia trial.
Still open
  1. Does TDP2 loss change breakage at KMT2A in a haematopoietic cell? The only locus-level measurement is in a retinal epithelial line and a breast cancer line.
  2. How many of the sequenced KMT2A genomic junctions in the published recombinome carry exactly 4 bp of perfect homology, and does the fraction differ between therapy-related and de novo infant cases? This is the only aetiological readout in the entire beat that could be computed from data already in the public domain.
  3. Is TDP2 expressed at all in human fetal liver HSC? If the enzyme that cleans TOP2 adducts is low in the cell where the de novo break happens, that is a candidate explanation for the infant age peak and nobody has measured it.
Every source behind this entry (7)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    TDP2 is UniProt O95551, tyrosyl-DNA phosphodiesterase 2, 362 residues, synonyms EAP2 and TTRAP; the paralogue TDP1 is Q9NUW8 at 608 residues. Both primary gene names and lengths were read back from the records in this session.

    What this evidence cannot show

    A naming fact. Resolved by gene_exact:TDP2 AND organism_id:9606 AND reviewed:true, one row, gene_primary TDP2. Note the entry name is TYDP2_HUMAN, not TDP2_HUMAN, so a check that matches on entry name rather than gene_primary would fail here even though the record is correct.

    Evidence that argues against this (1)
    • The paralogue control matters because TDP1 and TDP2 are routinely discussed together and act on opposite topoisomerase classes - TDP1 on 3'-phosphotyrosyl adducts from TOP1, TDP2 on 5' adducts from TOP2. A swap would silently move a claim from the camptothecin literature to the etoposide literature. The accessions differ and so do the lengths: Q9NUW8 is 608 aa against O95551's 362. UniProt Consortium 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    TDP2 rejoins DSBs induced during transcription-dependent TOP2 activity at the MLL translocation hotspot, suppresses TOP2-induced chromosome rearrangements and reduces TOP2-induced translocations arising during gene transcription; TDP2-dependent NHEJ is nevertheless implicated in the formation of a rare translocation subclass characterised by junctions with 4 bp of perfect homology, previously associated with therapy-related leukaemia.

    TDP2 suppresses chromosomal translocations induced by DNA topoisomerase II during gene transcription. · Gómez-Herreros F, Zagnoli-Vieira G, Ntai I, Martínez-Macías MI, Anderson RM, Herrero-Ruíz A, et al. · Nat Commun 2017
    PMID 28794467DOI
    What this evidence cannot show

    Breast cancer cells and hTERT-RPE-1 with etoposide and FISH probes flanking the KMT2A locus. Not a haematopoietic cell, not a fetal cell, and no KMT2A fusion protein anywhere in the study. The 4-bp perfect-homology subclass is rare and its genetic dissociation rests on its absence in the TDP2-null arm.

    Evidence that argues against this (1)
    • Conversion of a trapped cleavage complex into an irreversible break is itself TDP2- and proteasome-dependent and is accelerated about two-fold at transcribed loci, so TDP2 sits on both sides of the ledger: it is required to make the lesion irreversible and required to rejoin it cleanly. A net-protective phenotype in one system does not establish which of the two roles dominates at the KMT2A BCR in another. Canela A et al. 2019
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    TDP2/TTRAP is the major, if not the only, 5'-tyrosyl-DNA phosphodiesterase activity in vertebrate cells, and TDP2-deleted DT40 cells are highly sensitive to etoposide but not hypersensitive to camptothecin or methyl methanesulfonate.

    PMID 21030584DOI
    What this evidence cannot show

    Chicken DT40 genetics plus biochemistry. The selectivity of the etoposide phenotype (no camptothecin or MMS hypersensitivity) is the strongest single argument that the enzyme's cellular job is TOP2 adduct removal specifically. The enzyme was originally identified as a 5'-tyrosyl DNA phosphodiesterase in Cortes Ledesma et al., Nature 2009 (PMID 19794497).

    Evidence that argues against this (1)
    • A selective etoposide phenotype in a DT40 knockout establishes a dominant activity in that cell, not exclusivity in humans: other nucleases including the MRE11-CtIP route can remove protein-linked ends, and their contribution is not measured by a TDP2 knockout that remains viable. Gómez-Herreros F et al. 2017
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Biallelic loss-of-function mutations in TDP2 cause a recessive paediatric-onset spinocerebellar ataxia (SCAR23) with drug-resistant epilepsy, intellectual disability and progressive cerebellar atrophy.

    PMID 31410782DOI
    What this evidence cannot show

    Very few reported families. A germline complete-loss phenotype is not a model for partial pharmacological inhibition in an adult, and this graph's standing caution on constraint metrics applies. The relevance here is that the phenotype is neurological and not haematological, which is itself informative.

    Evidence that argues against this (1)
    • With only a handful of reported patients, the absence of a reported haematological malignancy excess in TDP2-null humans carries almost no statistical weight - it is an absence of testing, not a negative result. The mouse data point the other way: Tdp2-deleted mice show lymphoid toxicity and increased bone-marrow genome instability after TOP2 poisoning. Gómez-Herreros F et al. 2013
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    As of 2026-09-20 the PubMed query (MLL OR KMT2A) AND TDP2 returns exactly 1 record, which is the Gomez-Herreros 2017 study already cited here; there is no second KMT2A-specific TDP2 study.

    What this evidence cannot show

    A counted absence and therefore only as good as its symbol set. Searched: TDP2 and the retired/alternative symbols TTRAP and EAP2. TTRAP and EAP2 are short acronyms used in other fields, so their raw counts were title-checked rather than taken at face value. The complex form was also run because TDP2 acts within an NHEJ context; no KMT2A-specific record was added by it. Symbols NOT searched exhaustively: the enzyme-activity phrase '5-tyrosyl-DNA phosphodiesterase' as a quoted tiab term in singular and plural.

    Evidence that argues against this (1)
    • A count of 1 is fragile in exactly the way this project has been burned before: the graph's own bibliography is a better index than PubMed for this field. Grepping the repository for TDP2 finds the mechanism entry and this graph's break-machinery hypothesis, both citing the same single paper - which is consistent with the count, but the check is what makes the absence usable rather than the count. Gómez-Herreros F et al. 2017
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    TDP2's subnuclear address is conditional on stress and requires PML. TDP2/TTRAP is a PML-nuclear-bodies-associated protein, and in response to proteasome inhibition it accumulates in NUCLEOLAR CAVITIES in a promyelocytic-leukaemia-protein-dependent manner; in the nucleolus it contributes to controlling levels of the ribosomal RNA precursor and its processing intermediates, and this function is INDEPENDENT of its 5'-tyrosyl DNA phosphodiesterase activity. UniProt O95551 records both Nucleus/PML body and Nucleus/nucleolus, and HPA v25.1 immunofluorescence independently gives 'Nuclear bodies' as a co-equal main location.

    The PML nuclear bodies-associated protein TTRAP regulates ribosome biogenesis in nucleolar cavities upon proteasome inhibition. · Vilotti S, Biagioli M, Foti R, Dal Ferro M, Lavina ZS, Collavin L, et al. · Cell Death Differ 2012
    PMID 21921940DOI
    What this evidence cannot show

    The nucleolar relocation is induced by proteasome inhibition, a strong and pleiotropic perturbation, so the nucleolar address is a stress-state property rather than a baseline one. The system is a non-haematopoietic human line; the separation-of-function test rules out the phosphodiesterase activity but does not establish what the nucleolar activity IS. Nothing here involves a topoisomerase-II adduct, which is the reason this graph cares about TDP2.

    Evidence that argues against this (1)
    • A single nuclear answer for TDP2 is wrong in a second, independent way that this stress story does not cover, and the two together mean the protein has no canonical compartment. A short isoform, TDP2S, is expressed from an alternative transcription start site, carries a MITOCHONDRIAL TARGETING SEQUENCE, and is enriched in mitochondria and cytosol, while full-length TDP2 carries the N-terminal NLS and UBA domain - and both isoforms are present and ACTIVE in mitochondria, with TDP2-null cells hypersensitive to mitochondrially-targeted doxorubicin. So 'where is TDP2' has an isoform answer before it has a stress answer, and any antibody or tagging strategy that cannot separate the two will report a blend. Huang SN et al. 2018
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    TDP2's nuclear/cytoplasmic partition is set by isoform choice. An internal ribosome entry site in the TDP2 coding region drives an alternative translation start, and a short isoform TDP2S expressed from an alternative transcription start site carries a mitochondrial targeting sequence contributing to its enrichment in mitochondria and cytosol, while full-length TDP2 carries the N-terminal nuclear localization signal and the ubiquitin-associated domain.

    Mitochondrial tyrosyl-DNA phosphodiesterase 2 and its TDP2S short isoform. · Huang SN, Dalla Rosa I, Michaels SA, Tulumello DV, Agama K, Khiati S, et al. · EMBO Rep 2018
    PMID 29438979DOI
    What this evidence cannot show

    The isoform work is largely in isogenic wild-type and TDP2-knockout DT40 chicken cells complemented with human TDP2 - a genetics system, not a human haematopoietic one - and the mitochondrial enrichment rests on fractionation plus a targeting-sequence prediction. The relative abundance of TDP2S versus full-length TDP2 in any human cell type, let alone a leukaemic blast, is not established here.

    Evidence that argues against this (1)
    • If the isoform split is real then the protein this graph reasons about may not be the protein most antibodies see, and that cuts against the mechanism entry that invokes TDP2 for removing topoisomerase II adducts at the KMT2A breakpoint cluster region: that job requires the NLS-bearing full-length species, in the nucleus, at chromatin - and neither this paper nor any other reports which species predominates in a haematopoietic cell. Worse, the one clear subnuclear address the protein does have at baseline is PML bodies, which are not where a topoisomerase-II adduct is, and under stress the nuclear pool leaves for the nucleolus altogether. The compartment evidence does not currently support the mechanism this graph asserts. Vilotti S et al. 2012
Connected to