Oncofusion

start / graph / gene-kdm6a

Proteinstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q5c-interactome2

KDM6A / UTX — a commonly asserted KMT2A partner that belongs to a different complex and points the opposite way

also called KDM6A · UTX · ubiquitously transcribed tetratricopeptide repeat X-linked · KABUK2

UTX is often listed as a partner of the KMT2A machinery. It is not: it belongs to the MLL3 and MLL4 complexes, which are different enzymes, and in leukemia it behaves as a brake that gets broken rather than an accelerator that gets hijacked.

The same thing, for a scientist

KDM6A (O15550, 1,401 aa) is an X-linked H3K27me3 demethylase and the defining subunit that distinguishes MLL3/MLL4 (KMT2C/KMT2D) COMPASS-like complexes from MLL1/MLL2 COMPASS. No primary study places UTX in an MLL1 or KMT2A-fusion complex. Its DepMap profile is the inverse of a dependency: mean Chronos +0.096, depleted in 0.8% of lines, and a negative leukemia-restricted correlation with MEN1. In leukemia it is a verified tumour suppressor, lost by somatic mutation in male T-ALL and enriched at relapse in paediatric ALL. This entry exists as a correction, not as a partner entry.

What it looks like
KDM6A1,401 aa
JmjC (H3K27me3 demethylase catalytic domain)JmjC (H3K27me3 demethylase catalytic domain) · residues 1095–125805001,000
loading AF-O15550-F1…
AlphaFold AF-O15550-F1 ↗
colourshape

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.

very high (90+)confident (70–90)low (50–70)very low (<50)
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.

Go deeper
The correction, stated precisely

This graph already records that KDM6A/UTX has no verified direct KMT2A-fusion association. That position holds up under a targeted re-examination, and the evidence for it is positive rather than merely absent.

UTX belongs to MLL3/MLL4. When the MLL3/MLL4-containing H3K4 methyltransferase complex was purified, PTIP, PA1 and UTX were found to associate specifically with that complex, while ASH2L, RBBP5, WDR5 and DPY-30 were the subcomplex shared across all human SET1-like complexes 1. The Drosophila work makes the point even more sharply: Trr and mammalian MLL3/MLL4 complexes are distinguished by bearing a unique subunit, the H3K27 demethylase UTX 2. UTX is the discriminator between the two COMPASS branches, which is the opposite of being a shared partner.

No evidence places UTX in an MLL1 or KMT2A-fusion complex. None of the major MLL-fusion proteomics studies reports UTX 3. Where the claim appears, it is most plausibly a nomenclature artefact from the legacy naming in which 'MLL2' referred to what is now KMT2D/MLL4.

Note also the stoichiometry. Even in the MLL3/MLL4 complex, the original purification describes a substoichiometric amount of UTX 1. It is a specific but loosely associated subunit, not a core one.

The polarity is wrong too

Beyond complex membership, UTX points in the opposite direction from every other protein in this beat.

Somatic loss-of-function UTX mutations occur in human T-ALL, exclusively in male patients; UTX escapes X-inactivation in female T-ALL lymphoblasts and normal T cells, which explains the gender skew; and UTX functions as a bona fide tumour suppressor in vitro and in vivo, with UTX-inactivated T-ALL showing collateral sensitivity to pharmacologic H3K27me3 inhibition 4. KDM6A is also among the epigenetic regulators recurrently mutated and enriched at relapse in paediatric ALL, alongside SETD2, CREBBP and KMT2D 5.

The dependency data agree without being told. KDM6A has a mean Chronos of +0.096 across 1,258 lines — knocking it out makes cells grow marginally better — and is depleted in only 0.8% of lines. Across 54 leukemia lines it correlates with MEN1 at r = -0.19 6. That is a tumour-suppressor profile in numbers.

Human constraint is severe (pLI 1.0, LOEUF 0.155, 11 observed loss-of-function variants against 117.6 expected) 6, and germline KDM6A loss causes Kabuki syndrome — which we state as background rather than as evidence, having not verified a primary citation for it here.

The one clinical idea worth keeping

UTX is not a target to inhibit. But UTX loss may be a targetable state. An epigenetic drug-library screen identified an LSD1 inhibitor that selectively targets UTX-deficient cells for differentiation therapy: Utx loss blocks HSPC differentiation and reduces H3K4 methylation via the COMPASS-like complex, and the LSD1 inhibitor SP2509 selectively reverses this in Utx-null cells and extends survival of Utx-null AML mice 7. Van der Meulen and colleagues separately report collateral sensitivity of UTX-inactivated T-ALL to H3K27me3 inhibition 4.

This graph carries an LSD1/KDM1A target entry already. The relevant question for this project is narrow and answerable: is KDM6A lost in any KMT2A-rearranged cases, particularly at relapse? If so, those patients constitute a defined synthetic-vulnerability subgroup, and the drugs to exploit it are already in development for other reasons.

Facts and figures
uniprotO15550
gene symbolKDM6A
ensemblENSG00000147050
length AA1401
alphafold IDAF-O15550-F1
cytobandXp11.3
complexMLL3/MLL4 (KMT2C/KMT2D) COMPASS-like, substoichiometric; NOT MLL1 and NOT the KMT2A-fusion complex
graph rolecorrection of a commonly asserted edge, not a partner entry
polaritytumour suppressor, lost in leukemia; the opposite polarity to every oncogenic interactor in this beat
druggabilityNot a target to inhibit. UTX loss is a potentially targetable state: LSD1 inhibition and H3K27me3 inhibition both show collateral sensitivity in UTX-deficient models.
subnuclear localizationliquid-like condensates nucleated by UTX's core IDR, into which KMT2D is recruited and its H3K4 methyltransferase activity enriched (PMID 34526716). Nuclear residence itself is conditional: the TPR domain retains UTX in the nucleus via the MLL3/4 complexes, and cancer TPR mutants are preferentially cytoplasmic (PMID 32071397), where UTX instead occupies stress granules (PMID 40536321). MLL3/MLL4 here means KMT2C/KMT2D, not KMT2A and not KMT2B - see brief rule 66.
shown for fusionNO, and the literature that looks closest is about a different KMT2 gene. Measured 2026-09-21 with NCBI esearch: (KDM6A[tiab] OR UTX[tiab]) AND (MLL[tiab] OR KMT2A[tiab] OR MLL-AF4[tiab] OR MLL-AF9[tiab] OR mixed lineage leukemia[tiab]) AND (localiz*[tiab] OR localis*[tiab] OR subcellular[tiab] OR nuclear body[tiab] OR nuclear bodies[tiab] OR speckle*[tiab] OR condensat*[tiab] OR nucleolar[tiab] OR immunofluorescence[tiab] OR compartment*[tiab]) returns 6 records. Only one is a genuine UTX localisation study - PMID 32071397, cited here - and its MLL is MLL3/MLL4, i.e. KMT2C/KMT2D. The other five are prostate-cancer mutational landscape, urothelial Kmt2c/Kmt2d loss redistributing KMT2A-menin, a deregulated-demethylase expression survey, an MLL3 breast preprint and a urothelial review. Operands: gene clause 1,472, gene AND fusion 47. Positive control through the identical shape with (MLLT1[tiab] OR ENL[tiab]) returns 23. Zero in titles and abstracts; supporting figures not searched (brief rule 80). The MLL-term collision is exactly the hazard this entry exists to flag: UTX is repeatedly listed as a KMT2A partner and its real complex is KMT2C/KMT2D.
PDB IDS
[
  "3AVR",
  "3AVS",
  "6FUK",
  "6FUL",
  "6G8F"
]
PDB verified
[
  "3AVR",
  "6FUK"
]
domains
[
  {
    "name": "JmjC (H3K27me3 demethylase catalytic domain)",
    "start": 1095,
    "end": 1258,
    "pfam": null
  }
]
essentiality
{
  "depmap_common_essential": false,
  "chronos_mean": 0.097,
  "chronos_mean_leukemia_lines": 0.1259,
  "n_lines": 1258,
  "n_leukemia_lines": 57,
  "fraction_lines_depleted": 0.0079,
  "gnomad_pLI": 1,
  "gnomad_loeuf": 0.1549,
  "gnomad_obs_lof": 11,
  "gnomad_exp_lof": 117.6,
  "mouse_ko": "Utx loss blocks HSPC differentiation and reduces H3K4 methylation; germline knockout phenotype not verified in this entry",
  "human_germline": "Germline KDM6A loss causes Kabuki syndrome (stated as background; no primary citation verified in this entry)",
  "hpa_ntpm": {
    "bone_marrow": 17.9,
    "thymus": 14.1,
    "liver": 9,
    "spleen": 8.9,
    "lymph_node": 8,
    "testis": 7.4,
    "heart_muscle": 5.3,
    "classical_monocyte": 1.4,
    "naive_B_cell": 1.1
  },
  "hpa_specificity": "Low tissue specificity"
}
codependency men1
{
  "r_leukemia_lines": -0.1916,
  "ci_leukemia": [
    -0.408,
    0.0523
  ],
  "crosses_zero": true,
  "p_permutation": 0.15797
}
localization
{
  "subcellular": "NUCLEUS BY RETENTION, with a real and functional cytoplasmic pool. UniProt O15550 annotates Nucleus only and at ECO:0000305 - a curator inference rather than a direct experimental code, which is worth stating because the experimental literature is richer and less tidy than the annotation. UTX is retained in the nucleus by the MLL3/MLL4 (KMT2C/KMT2D) complexes through its TPR domain, and cancer TPR mutants that cannot bind those complexes are preferentially cytoplasmic. Cytoplasmic UTX forms puncta in stress granules and regulates their disassembly through G3BP1.",
  "subnuclear": "PHASE-SEPARATED CONDENSATES. A core intrinsically disordered region of UTX forms liquid condensates, condensation rather than demethylase activity is what carries the tumour-suppressive function, and UTX recruits KMT2D into the same condensates and enriches its H3K4 methylation activity. UTY, the Y-chromosome homologue, condenses with reduced molecular dynamics - a material-state difference that tracks with weaker tumour suppression. This is one of the better-evidenced condensate claims reaching this graph because it uses IDR deletion, mutagenesis AND replacement rather than 1,6-hexanediol.",
  "how_shown": "in vitro reconstitution, IDR deletion/mutagenesis/replacement, engineered cellular systems, fluorescence dynamics and genome-wide histone-modification and chromatin-interaction profiling (PMID 34526716); interaction mapping with ASH2L, PTIP and PA1 plus compartment readout of TPR point mutants including CRISPR-edited endogenous G137V (PMID 32071397); stress-granule co-localisation, G3BP1 interaction and D336G analysis (PMID 40536321).",
  "systems": "in vitro reconstitution, embryonic stem cells, HCT116 and other solid-tumour lines with nude-mouse xenografts; no leukaemia and no KMT2A-rearranged system has ever been used to localise UTX",
  "node": "localization-condensate-evidence-standards"
}
cell type
{
  "expressed_in": "Not resolved to a haematopoietic cell type here. The relevant expression fact is dosage rather than cell type: UTX escapes X-inactivation, so females carry two expressed copies [E4].",
  "required_in": "NOWHERE, in the sense this tranche means - and that is the finding. KDM6A has a POSITIVE mean DepMap gene effect of +0.096 across 1,258 lines with only 0.8 percent of lines depleted [E6]. It is a tumour SUPPRESSOR in this disease family: somatic loss-of-function UTX mutations occur in human T-ALL exclusively in male patients [E4], and KDM6A is among the epigenetic regulators recurrently mutated at RELAPSE in paediatric ALL [E5].",
  "loss_of_function_in": "HAEMATOPOIETIC STEM AND PROGENITOR CELLS: Utx loss blocks HSPC differentiation and reduces H3K4 methylation via the COMPASS-like complex, and an epigenetic drug-library screen identified an LSD1 inhibitor targeting UTX-deficient cells [E7]. That is the cell type where UTX loss has a phenotype, and the polarity is loss-of-function, not dependency.",
  "how_shown": "genome-scale CRISPR fitness (DepMap Chronos via Open Targets, this entry's E6); targeted and whole-exome sequencing of human T-ALL and of paired diagnosis-relapse paediatric ALL (this entry's E4, E5); Utx-deficient murine HSPC differentiation assays with an epigenetic inhibitor screen (this entry's E7)",
  "nodes": [
    "cell-type-hsc-bone-marrow",
    "cell-type-early-t-precursor"
  ]
}
cell state
{
  "maintains_state": "DIFFERENTIATION COMPETENCE, by its presence rather than by its absence: Utx loss BLOCKS HSPC differentiation [E7]. So UTX and the KMT2A fusion push the same state variable in the SAME direction when UTX is lost - which is why this entry records UTX as pointing the opposite way from a KMT2A-fusion dependency, and why an inhibitor of UTX would be an inhibitor in the direction of disease.",
  "state_dependent_localisation": "UTX's compartment is a STRESS-state variable. Cytoplasmic UTX enters stress granules under several stresses and disassembles them through G3BP1, and the clinical D336G mutation increases the cytoplasmic pool and stabilises granules [E_q66_loc2]. A leukaemic blast under chemotherapy or under nutrient stress is a cell in which UTX's distribution should differ, and nobody has measured it.",
  "regulates_state": "Embryonic stem cell differentiation, condensation-dependently [E_q66_loc1]. Recorded separately from the haematopoietic result because it is a different system.",
  "how_shown": "Utx-deficient HSPC differentiation assays (this entry's E7); IDR deletion/replacement with differentiation readout in embryonic stem cells (PMID 34526716); stress-granule imaging with G3BP1 perturbation (PMID 40536321)",
  "nodes": [
    "cell-state-differentiation-block"
  ]
}
How this could bite us back
  • Direction-of-effect error: treating UTX as a fusion partner would suggest inhibiting it, and inhibiting a tumour suppressor whose loss drives leukemia is the worst possible intervention.
  • Nomenclature hazard: the legacy 'MLL2' ambiguity between KMT2B and KMT2D has propagated real errors in the literature and is the most likely source of this one.
  • Even the synthetic-lethal idea is conditional: it requires UTX to actually be lost, which has not been shown in KMT2A-rearranged disease.
  • UTX is a tumour suppressor with a positive DepMap gene effect, so there is no inhibitor to bite back at - the bite-back is on the reasoning. Any strategy that lists UTX as a KMT2A partner and proposes perturbing it would be perturbing a COMPASS-like complex that is KMT2C/KMT2D, in the direction of disease, in a gene whose loss already marks relapse in paediatric ALL.
  • The condensate result is a warning about a whole class of strategy rather than a route to one: condensation, not catalysis, carries UTX's tumour suppression (PMID 34526716). Demethylase-inhibitor programmes aimed at the KDM6 family would therefore miss the function that matters, and a condensate-dissolving agent would abolish it.
  • There is one actionable direction and it is synthetic-lethal rather than direct: an epigenetic drug-library screen identified an LSD1 inhibitor active against UTX-deficient cells (this entry's E7), and UTX-deficiency is a relapse-associated state in paediatric ALL. That makes KDM6A status a candidate biomarker for an LSD1 inhibitor, which this graph already carries chemical matter for.
  • Sex is a variable here, not a covariate: UTX escapes X-inactivation and somatic loss-of-function mutations occur in human T-ALL exclusively in male patients (this entry's E4), while UTY condenses with reduced molecular dynamics (PMID 34526716). A UTX-status biomarker therefore behaves differently in male and female patients for structural reasons, not statistical ones.
Still open
  1. Is KDM6A lost in any KMT2A-rearranged cases, and specifically at relapse? If it is, those patients form a defined synthetic-vulnerability subgroup for LSD1 or EZH2 inhibition, and the relapse-cohort data to answer it already exist.
  2. How did UTX enter KMT2A-fusion interactome lists in the first place? If it is the legacy 'MLL2 = KMT2D' naming collision, the same collision may have contaminated other entries in this and other graphs.
  3. Does the X-linked, X-inactivation-escaping behaviour that makes UTX a male-specific tumour suppressor in T-ALL apply in infant B-ALL, where sex ratios and X-inactivation patterns differ?
  4. Does UTX condense in a KMT2A-rearranged cell, and does the KMT2A fusion enter or exclude those condensates? UTX recruits KMT2D into its condensates; whether a KMT2A fusion - which carries the same CxxC address - is admitted is unknown and would distinguish a shared compartment from two separate ones.
  5. Is the cytoplasmic stress-granule pool of UTX present in leukaemic blasts under chemotherapy? If UTX exits the nucleus under stress, then a therapy-stressed cell transiently loses a tumour suppressor from chromatin, which would be a drug-induced and reversible version of the mutation.
  6. Do relapse-acquired KDM6A mutations in paediatric ALL fall in the TPR or in the cIDR? The two disrupt different things - nuclear retention versus condensation - and the clinical mutation spectrum has never been mapped onto the two mechanisms.
Every source behind this entry (9)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    PTIP, PA1 and UTX associate specifically with the MLL3/MLL4-containing H3K4 methyltransferase complex, with UTX present substoichiometrically, whereas ASH2L, RBBP5, WDR5 and hDPY-30 form the subcomplex shared across all human SET1-like complexes.

    PTIP associates with MLL3- and MLL4-containing histone H3 lysine 4 methyltransferase complex. · Cho YW, Hong T, Hong S, Guo H, Yu H, Kim D, Guszczynski T, Dressler GR, Copeland TD, Kalkum M, Ge K · Journal of Biological Chemistry 2007
    PMID 17500065DOI
    What this evidence cannot show

    Complex purification from cell lines. Substoichiometric association is a weaker claim than core subunit membership, and this should be reflected wherever UTX is drawn into a COMPASS-like complex diagram.

    Evidence that argues against this (1)
    • UTX associating SUBSTOICHIOMETRICALLY with the MLL3/MLL4 complex is the qualifier that matters: a substoichiometric subunit is present on some complexes and not others, so a bulk composition result does not license treating it as a constituent. Cell Reports 2014
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Trr and mammalian MLL3/MLL4 complexes are distinguished from other COMPASS family members by bearing a unique subunit, the H3K27 demethylase UTX.

    Enhancer-associated H3K4 monomethylation by Trithorax-related, the Drosophila homolog of mammalian Mll3/Mll4. · Herz HM, Mohan M, Garruss AS, Liang K, Takahashi YH, Mickey K, Voets O, Verrijzer CP, Shilatifard A · Genes & Development 2012
    PMID 23166019DOI
    What this evidence cannot show

    Primarily Drosophila genetics with mammalian comparison; the claim about UTX as a discriminating subunit is a statement about complex composition, not about function in leukemia.

    Evidence that argues against this (1)
    • This is MLL3/MLL4 and Drosophila Trr biology. KMT2A is a different COMPASS family member, and the entry's own E1 says PTIP, PA1 and UTX associate specifically with the MLL3/MLL4 complex — that is, not with KMT2A. Cell Reports 2014
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    A targeted search found no primary study placing UTX/KDM6A in an MLL1 or KMT2A-fusion complex, and none of the major MLL-fusion proteomics studies reports UTX as a component.

    PubMed searches for KDM6A/UTX in MLL1 and KMT2A-fusion complexes, conducted for this graph on 2026-09-18; no primary study retrieved · Oncofusion Knowledge Graph project, using the NCBI PubMed E-utilities · Literature search 2026
    What this evidence cannot show

    A negative search is bounded by query terms and by which proteomics studies report full component lists. Absence from a published component list can also mean the subunit was below detection rather than absent.

    Evidence that argues against this (1)
    • The negative result should not be read as settling that KDM6A is irrelevant to KMT2A-fusion biology, because KDM6A turns out to act on leukemia cells through routes that MLL-fusion proteomics would not detect. In AML, KDM6 demethylases regulate DNA-damage-repair gene transcription and KDM6A level sets sensitivity to venetoclax and to PARP inhibition; in CML, KDM6A is recruited by YY1 to the NTRK1 promoter and sustains imatinib resistance independently of its demethylase activity. Neither mechanism requires KDM6A to be a member of an MLL1 or KMT2A-fusion complex, and neither would appear in a co-immunoprecipitation of the fusion. 'Not in the complex' and 'not relevant to the disease' are different claims and this entry sometimes runs them together. Boila LD et al. 2023
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Somatic loss-of-function UTX mutations occur in human T-ALL exclusively in male patients, UTX escapes X-inactivation in female T-ALL lymphoblasts and normal T cells, UTX functions as a bona fide tumour suppressor in vitro and in vivo, and UTX-inactivated T-ALL shows collateral sensitivity to pharmacologic H3K27me3 inhibition.

    The H3K27me3 demethylase UTX is a gender-specific tumor suppressor in T-cell acute lymphoblastic leukemia. · Van der Meulen J, Sanghvi V, Mavrakis K, Durinck K, Fang F, Matthijssens F, et al., Wendel HG, Van Vlierberghe P · Blood 2015
    PMID 25320243DOI
    What this evidence cannot show

    T-ALL, not KMT2A-rearranged B-ALL. Whether the tumour-suppressor role transfers to KMT2A-rearranged disease is untested, and the gender-specific mechanism depends on X-inactivation escape in T cells specifically.

    Evidence that argues against this (2)
    • KDM6A is not a tumour suppressor everywhere in leukemia, and in at least one leukemia it is the opposite. In CML, KDM6A is upregulated relative to normal marrow, and of five upregulated KDMs only KDM6A depletion sensitised cells to imatinib-induced apoptosis; re-introducing either wild-type OR a demethylase-dead KDM6A restored imatinib resistance. Mechanistically KDM6A is recruited by YY1 to the NTRK1 promoter and upregulates TRKA. The authors state the finding is 'contrary to its reported role as a tumor suppressor and independent of its demethylase function'. Two things follow for this entry: the tumour-suppressor framing is context-dependent rather than general, and the catalytic activity that a KDM6 inhibitor would target is not the activity doing the work. Zhang C et al. 2021
    • In AML the direction of the KDM6A effect flips with disease stage and with the drug being used. A significant percentage of relapsed AML has UPREGULATED KDM6A, higher KDM6A expression correlates with venetoclax tolerance, and loss of KDM6A increases mitochondrial activity and BCL2 expression and sensitises AML cells to venetoclax; KDM6A-mutant patient-derived xenografts are sensitive to olaparib. So KDM6A status is a therapy-selection variable in AML in both directions, not a one-way tumour-suppressor story, and a KDM6A-high relapsed AML is a setting where reducing KDM6A would help rather than harm. This bears directly on the entry's claim that KDM6A 'points the opposite way'. Boila LD et al. 2023
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KDM6A is among the epigenetic-regulator genes with recurrent somatic mutations gained at relapse in paediatric acute lymphoblastic leukaemia, alongside SETD2, CREBBP, MSH6 and KMT2D.

    Mutations in epigenetic regulators including SETD2 are gained during relapse in paediatric acute lymphoblastic leukaemia. · Mar BG, Bullinger LB, McLean KM, Grauman PV, Harris MH, Stevenson K, et al., Ebert BL, Armstrong SA · Nature Communications 2014
    PMID 24662245DOI
    What this evidence cannot show

    A mixed paediatric ALL relapse cohort; the paper does not report whether the KDM6A mutations occurred in KMT2A-rearranged cases specifically, which is exactly the number this project would want.

    Evidence that argues against this (1)
    • Recurrent somatic mutation gained at relapse in paediatric ALL is a pan-subtype observation; KDM6A mutation in KMT2A-rearranged infant disease specifically is not established by it, and infant KMT2A-r carries among the lowest mutation burdens of any sequenced paediatric cancer. Oncofusion knowledge graph (internal analysis) 2026
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    KDM6A has a positive mean DepMap gene effect of +0.096 across 1,258 lines with only 0.8% of lines depleted and a leukemia mean of +0.126, and correlates with MEN1 at r = -0.19 across 54 leukemia lines; gnomAD reports pLI 1.0 and LOEUF 0.155 with 11 observed against 117.6 expected loss-of-function variants.

    What this evidence cannot show

    DepMap release version not stated by the API. A slightly positive mean gene effect is within noise of zero for most genes and should not be over-read as a growth advantage; the informative statement is that KDM6A is not required. KDM6A is X-linked, which complicates CRISPR scoring across male and female lines in ways this analysis does not adjust for.

    Evidence that argues against this (1)
    • A positive mean gene effect across unselected cell lines does not mean KDM6A is never a dependency; it means the dependency is confined to a genotype that is rare in the panel. SMARCA4-inactivated cells show impaired transactivation, strongly reduced KDM6A/KDM6B levels and a strong dependency on these demethylases, such that KDM6 inhibition compromises viability; GSK-J4 had strong anti-tumour effects in mice orthotopically implanted with SMARCA4-mutant lung cancer cells and with primary small cell carcinoma of the ovary, hypercalcaemic type. The DepMap +0.096 mean should therefore be read as 'no dependency in the average line', not as 'KDM6A is not a target'. Romero OA et al. 2021
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Utx loss blocks HSPC differentiation and reduces H3K4 methylation via the COMPASS-like complex, and an epigenetic drug-library screen identified the LSD1 inhibitor SP2509 as selectively reversing this in Utx-null cells and extending survival of Utx-null AML mice.

    Epigenetic drug library screening identified an LSD1 inhibitor to target UTX-deficient cells for differentiation therapy. · Wu B, Pan X, Chen X, Chen M, Shi K, Xu J, Zheng J, Niu T, Chen C, Shuai X, Liu Y · Signal Transduction and Targeted Therapy 2019
    PMID 31044091DOI
    What this evidence cannot show

    Mouse models of UTX-deficient AML/MDS, not KMT2A-rearranged leukemia. SP2509 is a tool compound; LSD1 inhibitors as a class have had mixed clinical results in AML. The synthetic-lethal relationship depends on UTX being lost, which has not been shown to occur in KMT2A-rearranged disease.

    Evidence that argues against this (1)
    • Utx loss blocking HSPC differentiation and an LSD1 inhibitor reversing it is a drug-library screen result in a normal-haematopoiesis model, so the rescue is of a differentiation block rather than of a leukaemia. Oncofusion knowledge graph (internal analysis) 2026
  8. E_q66_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    UTX/KDM6A forms phase-separated nuclear condensates and that, rather than its demethylase activity, is what makes it a tumour suppressor: a core intrinsically disordered region of UTX forms phase-separated liquid condensates, loss of the cIDR caused by the most frequent cancer mutation of UTX is mainly responsible for abolishing tumour suppression, deletion/mutagenesis/replacement of the IDR show condensation is critical for tumour suppression and embryonic stem cell differentiation, UTX recruits MLL4/KMT2D into the same condensates and enriches its H3K4 methylation activity, and UTY - the Y-chromosome homologue with weaker tumour-suppressive activity - forms condensates with REDUCED molecular dynamics.

    UTX condensation underlies its tumour-suppressive activity. · Shi B; Li W; Song Y; Wang Z; Ju R; Ulman A; Hu J; Palomba F; Zhao Y; Le JP; Jarrard W; Dimoff D; Digman MA; Gratton E; Zang C; Jiang H · Nature 2021
    PMID 34526716DOI
    What this evidence cannot show

    Reconstitution in vitro plus engineered systems in cells and embryonic stem cells - not leukaemia, not haematopoietic, and not KMT2A-rearranged. IDENTIFIER TRAP: the MLL4 recruited here is KMT2D (O14686, 5,537 aa), the enhancer H3K4 monomethyltransferase, NOT KMT2A and NOT KMT2B - and brief rule 66 records that MLL4 is an alias of both KMT2B and KMT2D with the protein names swapped, so this sentence must never be shortened to 'UTX recruits MLL4'. The IDR-replacement design is the strongest part of the evidence because it separates condensation from sequence. Read: abstract only. Identifiers copied field by field from the efetch record fetched 2026-09-21, ArticleId anchored per brief rule 81.

    Evidence that argues against this (1)
    • UTX's nuclear residence is not intrinsic - it is imposed by the complex, and cancer mutations release it into the cytoplasm. Interaction-compromised TPR mutants G137V and D336G and a TPR-deleted mutant were PREFERENTIALLY LOCALISED TO THE CYTOPLASM, and the authors conclude UTX is RETAINED in the nucleus by MLL3/4 complexes through the TPR; the preferential cytoplasmic localisation was reproduced for endogenous G137V created by CRISPR-Cas9 in HCT116. So a condensate model built on the nuclear pool depends on a complex interaction that the commonest class of clinical mutation disrupts. Kato H; Asamitsu K; Sun W; Kitajima S; Yoshizawa-Sugata N; Okamoto T; Masai H; Poellinger L 2020
  9. E_q66_loc2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    UTX/KDM6A has a genuine cytoplasmic function in a second, non-chromatin compartment: UTX localises both in the cytoplasm and the nucleus, cytoplasmic UTX forms puncta and co-localises in STRESS GRANULES under different stresses, the TPR domain regulates stress-granule disassembly by interacting with G3BP1 to disrupt the granule scaffold network under endoplasmic reticulum stress, and the clinical TPR mutation D336G increases cytoplasmic UTX and STABILISES stress granules - with wild-type UTX inhibiting and D336G promoting cell growth and tumorigenesis in a G3BP1-dependent manner.

    Histone Demethylase UTX Suppresses Tumor Cell Proliferation by Regulating Stress Granules. · Liu X; Liu X; Xue M; Xiao Y; Chen Y; Qiu R; Wu D; Zhou Y; Wang J; Yuan Y; Yu L; Shi T; Li Y; Su H; Chen H; Liu Y; Huang K; Zheng L · Adv Sci (Weinh) 2025
    PMID 40536321DOI
    What this evidence cannot show

    Solid-tumour cell lines and nude-mouse xenografts, not leukaemia and not haematopoietic. The compartment conclusion rests on puncta co-localisation with stress-granule markers under induced stress, so the resting cytoplasmic fraction is not quantified. D336G appears in both this work and PMID 32071397 with consistent direction - more cytoplasmic UTX - which is reassuring, but the two papers attribute different consequences to it. Read: abstract only.

    Evidence that argues against this (1)
Connected to