USP2 (ubiquitin carboxyl-terminal hydrolase 2) - the ninth commonest KMT2A partner and the only one that is a deubiquitinase
also called UBP41 · USP2a · USP2-69 · ubiquitin-specific-processing protease 2 · deubiquitinating enzyme 2
USP2 is an enzyme that removes ubiquitin, the small tag cells use to mark proteins for destruction, so USP2 makes its targets last longer. It is the ninth most common gene fused to KMT2A in leukaemia, and it is not the same gene as USP22, which this graph covers separately.
The same thing, for a scientist
USP2 (UniProt O75604, 605 aa, chromosome 11q23.3) is a cysteine-protease-family deubiquitinase whose catalytic USP/UCH domain occupies residues 267-599 with the nucleophilic cysteine at 276 and the proton-acceptor histidine at 557 [E1]. Its best-characterised substrate is MDM2: USP2a deubiquitinates MDM2 without acting on p53, so raising USP2a stabilises MDM2 and thereby *promotes* p53 degradation, the opposite sign to HAUSP/USP7 [E2]. In leukaemia USP2 matters because it sits ~1 Mbp telomeric to KMT2A on the same chromosome arm and is transcribed toward the centromere, so a KMT2A::USP2 fusion arises by intrachromosomal inversion; it is the ninth most frequent KMT2A partner at 30 patients in the 3401-patient recombinome and is found exclusively in the minor breakpoint cluster region [E3][E4]. The fusion retains USP2's UCH domain, which is the basis of the standing hypothesis that a partner-supplied deubiquitinase activity counteracts the PHD-finger-dependent degradation of MLL [E4]. Direct experimental characterisation of USP2 in a KMT2A-rearranged context does not exist: wild-type USP2 is not a dependency in KMT2A-rearranged leukaemia lines in either the myeloid or the B-lineage half of the DepMap panel [E5].
This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.
I have never read one of these pictures before
A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.
Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.
Anything in yellow sticks is a bound small molecule, which is usually the drug.
USP2 is not USP22. Read this before citing either.1 deeper
This graph already holds gene-usp22 and exp-usp22-macrocycle-direction-test-kmt2a. USP2 and USP22 are different genes with different chromosomes, different lengths, different complexes and opposite standing in the KMT2A field. The similarity of the symbols, plus both being ubiquitin-specific proteases, makes this a high-probability transcription error, so the discriminating facts are set out here in full:
| USP2 | USP22 | |
|---|---|---|
| UniProt | O75604 (UBP2_HUMAN) | Q9UPT9 (UBP22_HUMAN) |
| Length | 605 aa | 525 aa |
| Locus | 11q23.3, ~1 Mbp telomeric to KMT2A | 17p11.2 |
| Verified aliases | UBP41 | KIAA1063, USP3L |
| Relationship to KMT2A | fusion partner (30 patients, rank 9) | not a fusion partner; a SAGA/DUB-module component studied as a co-dependency |
| Substrate work cited here | MDM2 2 | H2B monoubiquitination, covered in its own entry |
| DepMap in KMT2A-r lines | no dependency in either lineage half (z +0.58 myeloid, +0.31 B-lineage) 5 | dependency in both halves (z -1.79 myeloid, -0.76 B-lineage) 5 |
Note the last row especially: the two genes give opposite answers in the same analysis, run in the same session on the same panel, so a substitution of one for the other would invert a conclusion rather than blur it.
The symbols do not actually collide in PubMed retrieval
Checked rather than assumed. (MLL OR KMT2A) AND USP2 returns 11 records and (MLL OR KMT2A) AND USP22 returns 1; the sets do not overlap and no title-checking was needed to separate them 6. The alias UBP41 returns 0 records in intersection with MLL/KMT2A, so it adds nothing. The risk of confusing these two genes is therefore not a literature-retrieval risk - it is a writing-and-reading risk inside documents like this one, which is why the discriminating table is here rather than in a methods note.
Domain architecture and what the fusion keeps
UniProt O75604 annotates a single catalytic domain, the USP (ubiquitin carboxyl-terminal hydrolase, Pfam PF00443) domain, at residues 267-599, with catalytic residues Cys276 (nucleophile) and His557 (proton acceptor) 1. InterPro places the Pfam PF00443 match at 267-596 1. Two separate regions, 1-200 and 403-503, are annotated as necessary for interaction with MDM4 1.
The 2019 founding report states that every USP2 and USP8 rearrangement it found fused the conserved UCH domain to an extended 5'-MLL portion 4. That is the functionally load-bearing observation: the catalytic module is what survives into the fusion. It also means the N-terminal 1-200 MDM4-interaction region is at risk of being lost, which the papers do not address and which matters for whether the fusion retains substrate selectivity (see open questions).
What USP2 does in a normal cell1 deeper
The clean primary result is the MDM2 one 2. A bacterial two-hybrid screen using MDM2 as bait recovered a peptide resembling USP2a; full-length USP2a associates with MDM2 in cells and deubiquitinates it while showing no deubiquitinating activity toward p53. Ectopic USP2a accumulates MDM2 dose-dependently and promotes MDM2-mediated p53 degradation, and suppressing endogenous USP2a destabilises MDM2, accumulates p53 and activates it 2. This is the opposite sign to HAUSP/USP7, which deubiquitinates p53 as well as MDM2 and so protects p53.
The 2019 KMT2A letter summarises this correctly and takes one further inferential step - that stabilised MDM2 degrades p53, which activates MYC because active p53 induces microRNAs targeting MYC mRNA 4. That second step is the letter's own reasoning, not a measurement in a KMT2A::USP2 cell, and this entry does not carry it as established.
Cell type, cell state, localisation and epigenetic readout: what is and is not known
- Subcellular localisation: UniProt records cytoplasm, cytoplasmic perinuclear region, nucleus and membrane for USP2 1. So it is not a constitutively nuclear protein, unlike most KMT2A partners.
- Subnuclear localisation: not known. No speckle, PML-body, nucleolar or chromatin-compartment assignment has been published for USP2, and none for the KMT2A::USP2 fusion protein.
- Cell type and cell state of the leukaemia-initiating cell: not known for this fusion. The patients are overwhelmingly infants and young children with B-ALL or B/myeloid MPAL 3, which is consistent with the pro-B/LMPP origin proposed for other KMT2A fusions, but no sorted-population or xenograft work exists for KMT2A::USP2.
- Epigenetic readout: none published. No ChIP-seq, CUT&RUN, H3K79me2 or H3K4me3 profile of a KMT2A::USP2 cell has been reported. The closest thing is a transcriptional observation, that LAMP5 expression estimated the occurrence of all KMT2A-USP2 fusions in a machine-learning classifier 7.
Essentiality, lineage-split before it is stated
USP2 is not a dependency of KMT2A-rearranged leukaemia lines, and the statement is made only after splitting the index arm by lineage, per the standing rule in this repo.
| gene | KMT2A-r myeloid (z) | KMT2A-r B-lineage (z) | gap | verdict |
|---|---|---|---|---|
| USP2 | +0.58 | +0.31 | 0.28 | no dependency in either half |
| USP22 | -1.79 | -0.76 | -1.03 | dependency in both halves |
| MEN1 | -2.29 | -2.28 | -0.01 | dependency in both halves |
| KMT2A | -1.48 | -1.22 | -0.26 | dependency in both halves |
z-scores are in standard deviations of the other-leukaemia arm; negative means more depleted 5. USP2 scores positive in both halves, i.e. if anything slightly less essential in KMT2A-rearranged lines than in other leukaemia lines. The hard limit on this: the DepMap panel contains no KMT2A::USP2 line at all. This measures wild-type USP2 in lines carrying other KMT2A fusions, which is a different question from whether the fusion protein's own UCH activity is required. See ana-depmap-usp2-maml2-lineage-split for the full method and its limits.
| uniprot | O75604 |
|---|---|
| uniprot verified | O75604 = UBP2_HUMAN, PRIMARY gene name USP2, synonym UBP41, 605 aa, Homo sapiens. Fetched from the UniProt REST API on 2026-09-19 and the primary gene name read back rather than trusting a gene_exact synonym match; length agrees with the value asserted here. Explicitly distinct from Q9UPT9 = UBP22_HUMAN, primary gene name USP22, 525 aa, which is a different protein covered by gene-usp22. |
| length AA | 605 |
| ensembl | ENSG00000036672 |
| ensembl verified | ENSG00000036672 confirmed to resolve to approvedSymbol USP2 by the Open Targets Platform target query used in analyses/depmap-usp2-maml2-lineage-split; that script aborts any gene whose Ensembl id resolves to a different symbol, and no gene in the panel errored. |
| locus | 11q23.3, approximately 1 Mbp telomeric to KMT2A, transcribed toward the centromere |
| PDB note | No PDB entry is asserted here. No structure of KMT2A::USP2 exists, and no structure of a USP2 UCH domain was verified against the RCSB API in this session, so none is listed rather than listing one from recall. |
| alphafold ID | AF-O75604-F1 |
| subnuclear localization | Not known. No subnuclear compartment has been assigned to USP2 or to KMT2A::USP2 in any published work. |
| epigenetic readout | None published. No chromatin-profiling experiment of any kind has been performed on a KMT2A::USP2 cell. |
PDB IDS
[]
domains — the numbers behind the chart
[
{
"name": "UCH (ubiquitin carboxyl-terminal hydrolase catalytic domain)",
"start": 267,
"end": 599,
"pfam": "PF00443"
},
{
"name": "MDM4-interaction region (N-terminal)",
"start": 1,
"end": 200
},
{
"name": "MDM4-interaction region (C-terminal)",
"start": 403,
"end": 503
}
]catalytic residues
{
"nucleophile_cys": 276,
"proton_acceptor_his": 557
}subcellular localization
[ "cytoplasm", "cytoplasm, perinuclear region", "nucleus", "membrane" ]
essentiality
{
"depmap_common_essential": false,
"kmt2a_r_myeloid_z": 0.58,
"kmt2a_r_b_lineage_z": 0.31,
"lineage_split_done": true,
"verdict": "no dependency in either lineage half of the KMT2A-rearranged panel",
"hard_limit": "the DepMap panel contains no KMT2A::USP2 line"
}kmt2a partnership
{
"rank_2023_recombinome": 9,
"patients_2023": 30,
"rank_2019_letter": 14,
"patients_2019": 17,
"bcr": "minor BCR (BCR2) only",
"mechanism_class": "intrachromosomal inversion at 11q23.3"
}disambiguation
{
"not_to_be_confused_with": "gene-usp22 (USP22, Q9UPT9, 525 aa, 17p11.2, SAGA DUB module) - a different gene that is NOT a KMT2A fusion partner",
"why_it_matters": "the two give opposite DepMap answers in the same lineage-split analysis, so substituting one for the other inverts a conclusion"
}depmap kmt2a selectivity
{
"chronos_mean_kmt2a_rearranged": 0.0279,
"chronos_mean_other_leukaemia": -0.0255,
"difference": 0.0533,
"pan_essentiality_all_lines": 0.0076,
"selectivity_rank": 14997,
"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."
}cell type
{
"evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
"expressed_in": {
"class": 1,
"summary": "broadly expressed deubiquitinase with a well-studied circadian oscillation and prostate biology; no haematopoietic cell-type-restricted expression claim is made here, and none was found."
},
"required_in": {
"class": 2,
"claim": "CONDITIONAL, AND THE CONDITION IS A CELL STATE RATHER THAN A CELL TYPE. USP2 is the specific deubiquitinase for cyclin D1: USP2 knock-down destabilises cyclin D1 and induces GROWTH ARREST in human cancer lines whose growth depends on cyclin D1, has only very mild effects in normal human fibroblasts, and has NO EFFECT in cancer cells that do not express cyclin D1 [E_q69_cs1]. The requirement tracks cyclin D1 addiction, not lineage.",
"pmids": [
"19917254"
],
"system_named_in_the_claim": "human cancer cell lines and normal human fibroblasts - not haematopoietic and not leukaemia"
},
"the_haematopoietic_cell_type_is_a_PATIENT_level_answer": {
"claim": "The only cell types in which USP2 is documented in this disease are the patient ones: KMT2A-USP2 has been reported in INFANT ACUTE MYELOID LEUKAEMIA by whole-transcriptome sequencing, and as a cryptic/atypical rearrangement in infant and childhood leukaemia by mate-pair sequencing, and a recurrent second breakpoint cluster region in KMT2A gives rise to recurrent MLL-USP2 fusions. Those PMIDs are already cited in this entry and are not re-cited here.",
"why_it_is_recorded_as_a_cell_type_answer_at_all": "for a subtraction-layer partner gene, the disease-and-lineage distribution of the fusion IS the cell-type evidence available, and recording it as such is more honest than leaving the dimension blank. It is class 1 evidence (where the gene is found), not class 2 (where it is required)."
},
"nodes": [],
"context": "human cancer lines for the conditional requirement; patient case series for the lineage distribution"
}cell state
{
"evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
"maintains_state": {
"class": 3,
"states": [
"cell-state-cell-cycle-g1-cdk6"
],
"claim": "USP2 maintains the CYCLIN D1-DEPENDENT G1 STATE. Screening ubiquitinated cyclin D1 against a deubiquitinase library identified USP2 as its specific deubiquitinase; USP2 interacts with cyclin D1 directly and stabilises it by antagonising ubiquitin-dependent degradation, and USP2 knock-down destabilises cyclin D1 and induces growth arrest [E_q69_cs1].",
"pmids": [
"19917254"
],
"system_named_in_the_claim": "human cancer cell lines"
},
"the_conditionality_is_the_point": "USP2 inactivation has very mild effects in normal human fibroblasts and NO effect in cancer cells that do not express cyclin D1. So USP2 is a state-conditional dependency: the same genotype is required or dispensable depending on whether the cell is running its G1 on cyclin D1. That is the shape of dependency this beat's dimension exists to capture, and it cannot be read off a lineage.",
"bounded_query": {
"date": "2026-09-21",
"tool": "NCBI esearch db=pubmed",
"query": "(USP2 OR USP2a)[tiab] AND (hematopoie* OR haematopoie* OR leukemia OR leukaemia OR myeloid OR lymphoid)[tiab]",
"count": 25,
"on_topic_after_title_check": 4,
"what_the_four_are": "all four are the KMT2A-USP2 fusion reports and a KMT2A-rearranged review - i.e. USP2 appears in the leukaemia literature only as a FUSION PARTNER and never as a studied gene in a blood cell. No paper measures USP2 function, requirement or state-dependence in any haematopoietic cell.",
"second_query": "USP2[ti] AND (macrophage OR monocyte OR inflammat* OR \"T cell\" OR myeloid OR neutrophil)[tiab] -> 27 records, of which the closest (PMID 37670038, USP2 stabilising PD-L1) is tumour immune evasion in a solid-tumour setting rather than a haematopoietic cell-type result.",
"positive_control": "both shapes return 25 and 27 rather than 0, so the instrument demonstrably returns a one (brief rule 67).",
"bound": "titles and abstracts only; supporting figures not searched (brief rule 80)"
},
"depmap_reading": "Chronos +0.0279 in the KMT2A-rearranged arm against -0.0255 in other leukaemias, difference +0.0533, rank 14,997 of 17,728 - the ANTI-SELECTIVE half, and pan-essentiality is +0.0076, i.e. USP2 knockout has essentially NO effect on proliferation in any leukaemia line. Consistent with PMID 19917254: a cell that is not cyclin D1-addicted does not need USP2, and nothing about a KMT2A rearrangement makes a cell cyclin D1-addicted.",
"confidence": "moderate",
"caveats": "The cyclin D1 result is human cancer cell lines, none of them leukaemic, and it is one 2009 study from one laboratory. Nothing in the literature searched measures USP2 in a haematopoietic cell of any kind. The cell-type entry for this gene is therefore a patient-distribution answer, which is the weakest of the three evidence classes, and that is stated rather than dressed up.",
"counter_evidence": {
"note": "The authors' own controls bound the dependency hard, and this is the rare case where the paper's limitation IS the counter-evidence: 'cyclin D1 is not universally required for cell-cycle progression', USP2 inactivation has 'either very mild effects on cell growth in normal human fibroblasts or no effect in the cancer cells that do not express cyclin D1'. Any proposal to drug USP2 in this disease has to first show that KMT2A-rearranged blasts are cyclin D1-addicted, and this project's own scan (pan-essentiality +0.0076) says they are not.",
"citation": {
"title": "Suppression of cancer cell growth by promoting cyclin D1 degradation.",
"authors": "Shan J, Zhao W, Gu W",
"year": 2009,
"venue": "Mol Cell",
"pmid": "19917254",
"doi": "10.1016/j.molcel.2009.10.018",
"type": "paper"
}
}
}- If a deubiquitinase inhibitor were ever proposed for KMT2A::USP2, note that USP2's documented effect is to stabilise MDM2 and thereby lower p53. Inhibiting it would be expected to RAISE p53 - a plausible therapeutic direction, but it also means the fusion is not obviously a gain-of-degradation target, and an inhibitor would hit wild-type USP2 in every tissue that expresses it, including heart and muscle.
- USP2's annotated localisation includes cytoplasm and membrane, not only nucleus. A partner that is not constitutively nuclear raises the possibility that the fusion mislocalises or is only partly nuclear, which would confound any assumption that it behaves like a nuclear elongation-complex fusion.
- The absence of a USP2 dependency in KMT2A-rearranged lines means there is no cell-line-based route to validating a USP2-directed agent today. Any programme would have to build the model first.
- A USP2 inhibitor has no demonstrated target state in this disease. USP2 loss arrests growth only in cells addicted to cyclin D1, has very mild effects in normal human fibroblasts and no effect where cyclin D1 is absent (PMID 19917254); and in this project's genome-wide scan USP2 knockout has essentially no proliferation effect in any leukaemia line (pan-essentiality +0.0076, KMT2A-rearranged arm +0.0279). Before any USP2-directed chemistry, the cyclin D1 dependence of KMT2A-rearranged blasts has to be measured - it has not been.
- Does the KMT2A::USP2 fusion protein retain catalytic deubiquitinase activity? Nobody has expressed it and assayed it. This is a one-week biochemistry experiment and it gates every downstream interpretation.
- The N-terminal 1-200 region of USP2 is annotated as necessary for MDM4 interaction and is lost or truncated in the fusion. Does the fusion therefore have a different substrate repertoire from wild-type USP2, and if so what does it deubiquitinate?
- Is the ~1 Mbp KMT2A-USP2 interval a single topological domain? If the two genes share a TAD or a loop anchor, the inversion may reflect proximity rather than any property of either breakpoint, which would make this the cleanest available test of the proximity model of fusion formation.
- Which USP2 isoform does the fusion transcript correspond to? The leukaemia literature does not say, and USP2a versus the shorter isoforms differ in exactly the N-terminal region the fusion replaces.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
USP2 is UniProt O75604, 605 aa, with a USP/UCH catalytic domain at residues 267-599 (Pfam PF00443 match 267-596), catalytic Cys276 and His557, MDM4-interaction regions at 1-200 and 403-503, and annotated localisation to cytoplasm, perinuclear cytoplasm, nucleus and membrane.
UniProtKB entry O75604 (UBP2_HUMAN), Ubiquitin carboxyl-terminal hydrolase 2, with InterPro/Pfam domain boundaries · UniProt Consortium; InterPro (EMBL-EBI) · UniProtKB / InterPro 2026What this evidence cannot show
Database annotation, much of it inferred by similarity rather than measured on the human protein. The MDM4-interaction regions are curated from experimental papers not read in this session. Domain boundaries apply to the canonical 605-aa isoform (USP2a); USP2 has shorter isoforms, and which isoform a fusion transcript corresponds to is not stated in the leukaemia literature. Resolved by fetching the accession and reading back the PRIMARY gene name (USP2, synonym UBP41) and the length (605 aa) rather than trusting a gene_exact synonym match.
Evidence that argues against this (1)
- UniProt's subcellular-location set for USP2 is broad (four compartments) and is not compartment-resolved per isoform or per cell type, so it cannot be used to claim that the fusion protein is nuclear. Nothing in the leukaemia literature localises KMT2A::USP2 at all. Meyer C et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
USP2a associates with and deubiquitinates MDM2 but not p53, so raising USP2a accumulates MDM2 and promotes MDM2-mediated p53 degradation, while suppressing endogenous USP2a destabilises MDM2 and activates p53.
The deubiquitinating enzyme USP2a regulates the p53 pathway by targeting Mdm2. · Stevenson LF, Sparks A, Allende-Vega N, Xirodimas DP, Lane DP, Saville MK · The EMBO Journal 2007What this evidence cannot show
Cell-line overexpression and RNA-interference work in non-haematopoietic tumour lines; no leukaemia cell, no KMT2A-rearranged context, and no fusion protein. The direction of effect is established for full-length USP2a, and the fusion protein is not full-length. Abstract and central claims read in this session; figures not re-derived.
Evidence that argues against this (1)
- A p53-axis mechanism is hard to reconcile with the clinical picture: KMT2A::USP2 disease is infant/young-child B-ALL and MPAL, a setting in which TP53 mutation is rare and p53 pathway disruption is not a characteristic lesion. No study has shown p53 pathway output to be altered in a KMT2A::USP2 patient sample, so the MDM2 link remains an inference from wild-type USP2 biology rather than an observation in this disease. Meyer C et al. 2019
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
USP2 is the ninth most frequent KMT2A translocation partner in the 3401-patient DCAL recombinome, with 30 patients, and USP2 breakpoints occur only in the minor breakpoint cluster region.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, Burmeister T, Gröger D, et al., Marschalek R · Leukemia 2023What this evidence cannot show
Single-centre referral cohort, so rank 9 is a rank within what DCAL receives, not a population rank. USP2 is a fusion that classical PCR methods systematically miss, so its rank rose as sequencing replaced PCR and will probably rise further; 30 is a floor. Counts read from the published article in this session.
Evidence that argues against this (1)
- The same partner was ranked fourteenth and counted at 17 patients four years earlier by the same group, which shows how unstable the rank is with respect to assay and accrual rather than biology. Meyer C et al. 2019
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
All KMT2A::USP2 and KMT2A::USP8 rearrangements in the discovery cohort fused the conserved UCH domain to an extended 5'-MLL portion, which the authors propose may counteract the PHD-finger-dependent polyubiquitination and proteasomal degradation of MLL.
Human MLL/KMT2A gene exhibits a second breakpoint cluster region for recurrent MLL-USP2 fusions. · Meyer C, Lopes BA, Caye-Eude A, Cavé H, Arfeuille C, et al., Marschalek R · Leukemia 2019What this evidence cannot show
The UCH retention is an observation from sequence; the degradation-rescue idea is explicitly speculative in the source ('may counteract'), with no biochemistry, no half-life measurement and no catalytic-dead control. Confidence is set at moderate for the sequence observation and the hypothesis is carried separately rather than as fact.
Evidence that argues against this (1)
- The same letter reports that two AFF1 and one MLLT3 case also broke in the minor BCR. Those partners supply no deubiquitinase activity at all, so if PHD-retaining fusions need a UCH domain to survive degradation, those three patients should not exist. Either the rescue is not necessary, or those cases work by a different route. Muntean AG et al. 2008
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In DepMap Chronos data accessed through Open Targets, USP2 shows no dependency in KMT2A-rearranged leukaemia lines in either the myeloid (z +0.58) or the B-lineage (z +0.31) half of the index arm, while USP22, MEN1 and KMT2A are depleted in both halves.
DepMap Chronos gene-effect scores for USP2, MAML2, USP22, USP8, MEN1 and KMT2A in KMT2A-rearranged leukaemia lines, lineage-split (analyses/depmap-usp2-maml2-lineage-split) · Broad Institute DepMap; Open Targets Platform; analysis run for this graph · DepMap / Open Targets Platform 2026What this evidence cannot show
Eight myeloid and five B-lineage KMT2A-rearranged lines, so a single line moves a mean. The split is by lineage LABEL, not expression signature. Decisively: the panel contains NO KMT2A::USP2 line, so this bounds the requirement for wild-type USP2 in KMT2A-rearranged cells and says nothing about whether the fusion protein's UCH activity is needed. Ensembl ENSG00000036672 was asserted to resolve to approvedSymbol USP2 by the query itself, which aborts on any mismatch.
Evidence that argues against this (1)
- A null dependency result in a panel that lacks the genotype in question is close to uninformative about that genotype. The strongest counter to reading anything into this is simply that KMT2A::USP2 has never been modelled in a cell line, a mouse or a xenograft. Blackburn PR et al. 2020
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
PubMed retrieval does not confuse USP2 with USP22 in the KMT2A context: (MLL OR KMT2A) AND USP2 returns 11 records and (MLL OR KMT2A) AND USP22 returns 1, with no overlap, and the alias UBP41 adds no records.
PubMed E-utilities esearch counts for KMT2A partner symbol and alias queries, run 2026-09-19 · National Center for Biotechnology Information (NCBI), PubMed · PubMed (NCBI E-utilities) 2026What this evidence cannot show
Counts are a snapshot of one day's index and will drift. The queries use [All Fields], so they also match reference lists and keyword blocks; every one of the 11 USP2 records was checked against its title and abstract and all 11 genuinely concern KMT2A-USP2, so raw and checked counts coincide here, which is unusual. The absence of overlap does not mean no paper discusses both genes, only that none in this intersection does.
Evidence that argues against this (1)
- A clean retrieval separation is exactly the condition under which a human or an agent stops checking, and the error that has actually occurred in this project is transcription, not retrieval. The counter to this evidence item is that it measures the wrong failure mode, which is why the discriminating table is kept in the body. Meyer C et al. 2023
- E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In a machine-learning classifier trained to predict KMT2A rearrangement from expression, high LAMP5 expression estimated the occurrence of all KMT2A-USP2 fusions, and the authors note KMT2A-USP2 fusions are often missed during FISH screening.
Novel Diagnostic and Therapeutic Options for KMT2A-Rearranged Acute Leukemias. · Lopes BA, Poubel CP, Teixeira CE, Caye-Eude A, Cavé H, Meyer C, Marschalek R, Boroni M, Emerenciano M · Frontiers in Pharmacology 2022What this evidence cannot show
Retrospective classifier on assembled transcriptome cohorts; 'all KMT2A-USP2 fusions' refers to the handful present in the training data, so the claim rests on very few cases. SKIDA1 (AUC 0.839) and LAMP5 (AUC 0.746) outperformed CSPG4/NG2 (AUC 0.722) for KMT2A-r generally, but no AUC is reported for the USP2 subset specifically. Overlapping authorship with the DCAL group that reported the fusion. Abstract read in full text form; model not re-run.
Evidence that argues against this (1)
- LAMP5 overexpression is described across KMT2A-rearranged leukaemia generally and is not specific to USP2, so its perfect separation of USP2 cases in one training set is more likely a small-n artefact than a USP2-specific marker. The same paper reports LAMP5's overall AUC as 0.746, well short of a diagnostic-grade discriminator. Li EW et al. 2025
- E_q69_cs1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Screening ubiquitinated cyclin D1 as a substrate against a deubiquitinase library identified USP2 as a specific deubiquitinase for cyclin D1; USP2 directly interacts with cyclin D1 and promotes its stabilisation by antagonising ubiquitin-dependent degradation, USP2 knock-down destabilises cyclin D1 and induces growth arrest in human cancer lines whose growth depends on cyclin D1 expression, and inactivation of USP2 has either very mild effects on cell growth in normal human fibroblasts or no effect in cancer cells that do not express cyclin D1.
Suppression of cancer cell growth by promoting cyclin D1 degradation. · Shan J, Zhao W, Gu W · Mol Cell 2009What this evidence cannot show
Abstract read in full; full text not read. Human cancer cell lines and normal human fibroblasts, none of them haematopoietic and none leukaemic (brief rule 73). One 2009 study from one laboratory. The conditionality - mild or no effect without cyclin D1 addiction - is the authors' own finding and is the load-bearing part of this item.
Evidence that argues against this (1)
- This project's genome-wide Chronos scan finds USP2 knockout has essentially no proliferation effect in any leukaemia line (pan-essentiality +0.0076) and is anti-selective in the KMT2A-rearranged arm (rank 14,997 of 17,728, difference +0.0533). Whatever cyclin D1-conditional dependency exists in solid-tumour lines is not detectable in leukaemia, so the state this evidence describes has not been shown to exist in the cells this graph is about. source
- ← fuses withKMT2A / MLL1 — the gene that breaks
inv(11)(q23.3q23.3), 30 patients, rank 9 of all KMT2A partners; minor BCR only
USP2 is the partner that defined the second/minor BCR; 17 of 21 minor-BCR cases in the discovery cohort
DIFFERENT GENE, easily confused: USP2 = O75604, 605 aa, 11q23.3, KMT2A fusion partner; USP22 = Q9UPT9, 525 aa, 17p11.2, SAGA DUB module, not a fusion partner. Opposite DepMap verdicts in the same analysis.
USP2 breakpoints occur exclusively in BCR2
conserved UCH deubiquitinase catalytic domain (USP2 267-599)
asks whether the fusion's UCH domain has catalytic activity at all, which has never been shown
the two UCH-domain KMT2A partners; different chromosomes, lengths and normal functions, same catalytic fold
- ← testsLineage-split DepMap: neither USP2 nor MAML2 is a dependency in KMT2A-rearranged lines, and USP22 is
no dependency in either lineage half (z +0.58 myeloid, +0.31 B-lineage); panel contains no KMT2A::USP2 line
11 raw records, 11 title-checked on-topic; the feared USP2/USP22 retrieval collision does not occur
USP2 is the specific deubiquitinase for cyclin D1 and stabilises it; USP2 knock-down destabilises cyclin D1 and arrests growth ONLY in cyclin D1-addicted cells (PMID 19917254)
a worked case where required_in is conditional on a cell STATE (cyclin D1 addiction) rather than on a cell type, and the haematopoietic cell-type answer exists only at patient level