Oncofusion

start / graph / gene-kmt2d

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q44-kmt2-family

KMT2D (MLL2/MLL4) - the gene named in this graph's menin-resistance entry, written here for the first time

also called MLL2 · MLL4 · ALR · Mll4 (mouse)

KMT2D is the largest member of the KMT2 family and another brake on cancer. Losing it makes mouse blood cancers grow faster, and it leaves those cells unusually dependent on making ribosomes - which is a drug target.

The same thing, for a scientist

KMT2D (UniProt O14686, **5,537 aa** - the largest KMT2 protein and more than twice KMT2B's 2,715) is the MLL4 catalytic subunit of the COMPASS-like MLL3/MLL4 complex, which carries PAXIP1/PTIP and PAGR1 rather than menin and LEDGF [E5]. It is an H3K4 mono-methyltransferase at enhancers and among the most frequently mutated genes in human cancer [E4]. In AML, KMT2D is downregulated or mutated, its deficiency by shRNA or CRISPR accelerates leukaemogenesis in mice, and Kmt2d-deficient HSPCs and AML cells show enhanced ribosome biogenesis, enlarged nucleoli, increased rRNA and protein synthesis and mTOR activation through direct Kmt2d regulation of *Ddit4* - creating a vulnerability to the RNA polymerase I inhibitor CX-5461 in vivo [E1]. Germline loss of function causes Kabuki syndrome, discovered under the name MLL2 [E2]. In this project's disease, KMT2D downregulation is proposed as the resistance mechanism in one of two multiply relapsed KMT2A::AFF1 patients - **and that paper states plainly that none of its patients had ever received a menin inhibitor, so the resistance was acquired without menin-inhibitor exposure** [E3].

Go deeper
The naming trap is at its worst on this gene

KMT2D has been called MLL2 and MLL4 in the primary literature, and so has KMT2B. UniProt lists both aliases on both genes and gives them swapped protein names 6. Three published examples, each resolved differently:

  • Kabuki syndrome, 2010. Ng et al. titled their paper Exome sequencing identifies MLL2 mutations as a cause of Kabuki syndrome 2. That MLL2 is KMT2D - resolved by phenotype, not by symbol: Kabuki syndrome type 1 is KMT2D, while KMT2B's germline phenotype is early-onset dystonia. This is the largest MLL2-means-KMT2D literature in existence and it sits alongside a mouse leukaemia literature in which MLL2 means KMT2B.
  • A 2019 case report. Cryptic ETV6-ABL1 Fusion and MLL2 Truncation... reports an MLL2-ADCY9 fusion as a candidate resistance driver in relapsed B-ALL. The title and abstract are undecidable; the discussion writes "MLL2 (KMT2D)" and settles it 7. A reader who stopped at the abstract could not have.
  • A 2019 review writes the gene as "MLL2(4)/KMT2D" in its own abstract - a published acknowledgement that the alias is broken 4.

The length is the check that survives all of this. KMT2D is 5,537 aa; KMT2B is 2,715. Any entry that asserts one accession and the wrong length is detectable by verify-accessions.py, which is why both are stored here.

What this entry does to `resistance-mll34-utx-tumour-suppressor-failure`

That entry argues KMT2C/KMT2D loss causes menin-inhibitor resistance by disabling the drug's activating MLL3/4-UTX arm, and its patient evidence is the Mahdavi serial-PDX study. Writing the genes surfaces three things that entry does not contain, all from the same abstract or from directly adjacent literature.

  1. The patients never received a menin inhibitor. The abstract says so explicitly: "Because none of the patients had been treated with a menin inhibitor, resistance in these highly pretreated samples was acquired in the absence of menin-inhibitor exposure" 3. The resistant state was produced by conventional therapy. That is compatible with the switch mechanism but it removes any implication that the drug selected for the lesion.
  2. Chemotherapy is a known selective pressure for KMT2C loss. Haploid Kmt2c deletion confers a selective advantage on HSCs only when they are driven into cycle by a strong proliferative stimulus such as chemotherapy, which is the published explanation for why KMT2C deletions dominate therapy-related MDS/AML 8. In a multiply pretreated patient, a KMT2C-deficient clone is the expected finding independent of anything a menin inhibitor does.
  3. The source says KMT3D, not KMT2D. The Mahdavi abstract as PubMed serves it reads "Downregulation of KMT3D may account for resistance in one patient". There is no KMT3D gene; the surrounding sentence about KMT2C/D makes KMT2D the evident intent. The graph entry restates it as KMT2D without flagging that the source does not say so. That is almost certainly the right reading and it should be recorded as an editorial correction rather than made silently.

None of this overturns the switch mechanism, which rests on CRISPR screens. It constrains the patient arm of the argument, which is where the entry's confidence currently comes from.

The one actionable thing in the KMT2D leukaemia literature

1 is the only study found in this beat that turns KMT2D loss into a therapeutic handle in myeloid leukaemia, and the handle is not epigenetic. Kmt2d loss activates mTOR by derepressing it - Kmt2d directly regulates Ddit4, a negative regulator of mTOR - which drives ribosome biogenesis: enlarged nucleoli, increased rRNA, increased protein synthesis. The RNA polymerase I inhibitor CX-5461 significantly restrained growth of Kmt2d-loss AML in vivo and extended survival of leukaemic mice.

This is a different countermeasure from the CDK4/6 inhibitors that resistance-mll34-utx-tumour-suppressor-failure proposes for the same lesion class, arrived at from a different mechanism, and neither entry knows about the other. Both are preclinical. A reader deciding what to combine with a menin inhibitor in a KMT2D-low blast now has two candidates and no comparison.

It also rhymes uncomfortably with something this graph already holds: concept-wdr5-what-is-established-for-wild-type-kmt2a-and-what-for-the-fusion records that WDR5 WIN-site inhibition in MLL-rearranged cells acts through ribosome subunit attrition and p53, not through KMT2A. Two independent routes into this family now converge on ribosome biogenesis as the thing that actually moves, which is either a real shared node or a sign that ribosome biogenesis is what any broad perturbation of a proliferating blast reports.

Facts and figures
uniprotO14686
length AA5537
gene primary verifiedKMT2D
accession resolutiongene_exact:KMT2D AND organism_id:9606 AND reviewed:true returned exactly one row; gene_primary equals KMT2D; length 5537 asserted, which is the only robust discriminator from KMT2B (Q9UMN6, 2715 aa) because both genes carry the aliases MLL2 and MLL4
cytoband12q13.12
complex branchCOMPASS-like MLL3/MLL4 (KMT2C/KMT2D) - carries PAXIP1/PTIP and PAGR1/PA1, NOT menin and NOT LEDGF
menin binding motifnone; menin is not a component of the MLL3/4 complexes
principal catalytic roleH3K4 mono-methylation at enhancers
germline phenotypeKabuki syndrome type 1 (KDM6A causes type 2)
leukaemia mechanismloss activates mTOR via derepression of Ddit4, driving ribosome biogenesis; vulnerability to RNA Pol I inhibition (CX-5461)
subnuclear localizationTranscriptional condensates formed by liquid-liquid phase separation; KMT2D loss impairs Polycomb-dependent chromatin compartmentalisation and alters nuclear architecture (PMID 33169020, Kabuki syndrome, mesenchymal stem cells). No haematopoietic imaging.
shown for fusionNO for localisation in a KMT2A-rearranged cell. The bounded query returned 7 records and TITLE-CHECKING finds none of them a KMT2D localisation experiment in a KMT2A-fusion context: two are UTX/KDM6A papers whose MLL is MLL3/MLL4 and not KMT2A (PMIDs 32071397, 30987376), one is a menin/IRF1 pre-mRNA-processing paper (22240255), one is the HALR/MLL3 cloning paper (11718452), and the remaining three are MLL2/MLL4 cancer-expression papers whose gene assignment is itself ambiguous under rule 66 (22713656, 20433758, 19060922). Note also that the gene clause for this entry necessarily includes MLL2 and MLL4, which are contaminated tokens, so this count is an upper bound on relevance and not a literature.
naming instances resolved
207111752,01020711175: 2010300288192,01930028819: 2019311282162,01931128216: 2019
aliases searched
[
  "KMT2D",
  "MLL2",
  "MLL4",
  "ALR"
]
naming instances resolved — the numbers behind the chart
[
  {
    "pmid": "20711175",
    "written_as": "MLL2",
    "resolves_to": "KMT2D",
    "how": "phenotype - Kabuki syndrome type 1",
    "title": "Exome sequencing identifies MLL2 mutations as a cause of Kabuki syndrome.",
    "authors": "Ng SB, Bigham AW, Buckingham KJ, et al.",
    "venue": "Nat Genet",
    "year": 2010,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "30028819",
    "written_as": "MLL2",
    "resolves_to": "KMT2D",
    "how": "stated in the full-text discussion as 'MLL2 (KMT2D)'",
    "title": "Cryptic ETV6-ABL1 Fusion and MLL2 Truncation Revealed by Integrative Clinical Sequencing in Multiply Relapsed Acute Lymphoblastic Leukemia.",
    "authors": "Mankuzhy NP, Walling E, Anderson B, et al.",
    "venue": "J Pediatr Hematol Oncol",
    "year": 2019,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "31128216",
    "written_as": "MLL2(4)",
    "resolves_to": "KMT2D",
    "how": "the review writes both aliases together with the KMT2D symbol",
    "title": "COMPASS Ascending: Emerging clues regarding the roles of MLL3/KMT2C and MLL2/KMT2D proteins in cancer.",
    "authors": "Fagan RJ, Dingwall AK",
    "venue": "Cancer Lett",
    "year": 2019,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  }
]
essentiality
{
  "depmap_common_essential": null,
  "note": "not scored in this beat"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.5398,
  "chronos_mean_other_leukaemia": -0.606,
  "difference": 0.0663,
  "pan_essentiality_all_lines": -0.3917,
  "selectivity_rank": 15817,
  "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. UniProt O14686 carries ONE location line on ONE coded citation, ECO:0000269 PubMed:23508102 - and that experiment is LABEL-FREE QUANTITATIVE MASS SPECTROMETRY OF NUCLEAR EXTRACTS, not microscopy. The protein is recovered from a nuclear extract, which is where the experiment looked. This is exactly the situation this graph already records for gene-kmt2b, whose nuclear call rests on the same publication, and the two entries should be read together.",
  "subnuclear": "TRANSCRIPTIONAL CONDENSATES formed by liquid-liquid phase separation. KMT2D (explicitly 'also known as MLL4' in the source) contributes to the assembly of transcriptional condensates through LLPS; its loss of function impairs Polycomb-dependent chromatin compartmentalisation and alters nuclear architecture and the mechanical properties of the nucleus (PMID 33169020). Separately, UTX/KDM6A condensation recruits MLL4 = KMT2D (PMID 34526716), and cancer-derived UTX TPR mutations that impair interaction with MLL3/4 complexes change UTX's subcellular localisation (PMID 32071397) - the MLL3/4 complexes are a nuclear RETENTION factor for UTX.",
  "how_shown": "quantitative AP-MS of nuclear extracts for the compartment (PMID 23508102, no imaging); imaging of condensates and chromatin compartmentalisation in Kabuki-syndrome mesenchymal stem cells (PMID 33169020)",
  "tissue_of_evidence": "human cell nuclear extracts; MESENCHYMAL STEM CELLS and chondrocyte commitment for the condensate work. Nothing haematopoietic has been imaged and nothing leukaemic.",
  "HPA_DISAGREES_AND_IS_PROBABLY_WRONG": "HPA v25.1 assigns KMT2D main subcellular locations PLASMA MEMBRANE and CYTOSOL, with Nucleoplasm only as an ADDITIONAL location, at reliability Approved (the weaker tier). For a 5,537-residue H3K4 methyltransferase that co-occupies more than a thousand enhancers with GATA1, a plasma-membrane main call is almost certainly an antibody artefact. It is recorded because suppressing a disagreeing measurement is worse than explaining it - NOT because it is believed. Compare KMT2C in the sibling entry, where HPA gives Nucleoplasm main at the higher Enhanced tier and agrees with everything else.",
  "identifier_resolution": "Brief rule 66 and the collision registry's MLL2 / MLL4 rows govern every citation in this field. KMT2B (Q9UMN6, 2,715 aa) and KMT2D (O14686, 5,537 aa) BOTH carry MLL2 and MLL4 as synonyms and each one's UniProt protein name carries the OTHER one's number, so reading the protein name back returns a confident wrong answer either way. Length is the only reliable discriminator. Each paper below was resolved individually and the reasoning is written into its caveat.",
  "node": "localization-condensate-evidence-standards"
}
cell type
{
  "haematopoietic_answer": "ERYTHROID PRECURSORS, with a named mechanism. KMT2D is an essential cofactor for GATA1: it co-localises with GATA1 at more than a thousand erythroid enhancers, co-occupancy predicts stronger transcription than GATA1 alone, acute KMT2D depletion drops H3K4me1 and H3K27ac at a subset of those enhancers, and loss of KMT2D in human erythroid precursors causes developmental arrest (PMID 42409830).",
  "why_this_matters_here": "This is a human, haematopoietic, enhancer-level result for a KMT2 family member other than KMT2A, and it is the cleanest cell-type answer in this tranche. It also sharpens the family's division of labour: KMT2A works at CpG-island promoters through its CXXC domain, KMT2D at lineage enhancers with a master transcription factor.",
  "expression_atlas": "UniProt tissue comment (UNCODED): expressed in most adult tissues including a variety of haematopoietic cells, with the exception of the liver. HPA Blood Atlas: low immune cell specificity, detected in some.",
  "identifier_resolution": "Brief rule 66 and the collision registry's MLL2 / MLL4 rows govern every citation in this field. KMT2B (Q9UMN6, 2,715 aa) and KMT2D (O14686, 5,537 aa) BOTH carry MLL2 and MLL4 as synonyms and each one's UniProt protein name carries the OTHER one's number, so reading the protein name back returns a confident wrong answer either way. Length is the only reliable discriminator. Each paper below was resolved individually and the reasoning is written into its caveat."
}
cell state
{
  "state": "DIFFERENTIATION TRANSIT. The two dedicated results are both about commitment rather than steady state - erythroid precursor maturation, which arrests without KMT2D (PMID 42409830), and mesenchymal commitment to chondrocytes, which is rescued by relieving nuclear mechanical stress (PMID 33169020).",
  "the_mechanical_reading": "PMID 33169020 makes an unusual claim this graph should not lose: KMT2D haploinsufficiency changes the MECHANICAL properties of the nucleus, and inhibiting the mechanosensor ATR re-established commitment. If a chromatin cofactor's loss is felt as nuclear mechanics, that is a compartment-level phenotype rather than a gene-by-gene one."
}
localization bounded absence
{
  "date": "2026-09-21",
  "tool": "NCBI E-utilities esearch, db=pubmed",
  "shape": "gene clause AND localisation clause AND KMT2A/MLL-fusion clause, one [tiab] tag per term (brief rule 94), every cytogenetic term quoted (brief rule 95)",
  "querytranslation_fusion_clause": "\"KMT2A\"[Title/Abstract] OR \"MLL\"[Title/Abstract] OR \"MLL1\"[Title/Abstract] OR \"MLL-AF4\"[Title/Abstract] OR \"MLL-AF9\"[Title/Abstract] OR \"MLL-ENL\"[Title/Abstract] OR \"KMT2A-rearranged\"[Title/Abstract] OR \"MLL-rearranged\"[Title/Abstract] OR \"t 4 11\"[Title/Abstract] OR \"t 9 11\"[Title/Abstract] OR \"MLL fusion\"[Title/Abstract]",
  "count": 7,
  "controls": {
    "fusion_clause_alone": 8743,
    "gene_clause_alone": 3016
  },
  "reading": "Counted in TITLES AND ABSTRACTS only; supporting figures were not searched (brief rule 80).",
  "title_check": "7 records, 0 of them a KMT2D localisation experiment in a KMT2A-rearranged cell; two are UTX/MLL3-MLL4 papers, and three carry rule-66 ambiguous gene assignments."
}
How this could bite us back
  • The resistance lesion precedes the drug. The only patient evidence linking KMT2C/D loss to menin-inhibitor resistance comes from patients who had never received a menin inhibitor, so this is a pre-existing state that better menin chemistry cannot prevent and that upfront use would have to outrun rather than avoid.
  • Prior chemotherapy is a plausible manufacturer of the lesion. If cytotoxic therapy selects KMT2C/D-deficient clones, then the standard backbone is building the resistant population before the targeted agent is ever given - an argument for moving menin inhibitors earlier that is stronger than the one usually made for it.
  • The countermeasure is a DNA-damaging agent in disguise. CX-5461 is described as an RNA Pol I inhibitor but also stabilises G-quadruplexes and damages DNA; giving it to an infant to exploit a ribosome-biogenesis vulnerability carries second-malignancy risk in a disease whose survivors are measured in decades.
  • There is no way to restore a tumour suppressor. Every route here is synthetic-lethal and indirect, which means every route depends on the lesion being correctly identified - and a KMT2D DOWNREGULATION without mutation is invisible to the sequencing panels these patients actually get.
Still open
  1. Is KMT2D expression measured at diagnosis in any infant KMT2A-rearranged cohort? The only KMT2D observation in this disease is one patient in a two-patient PDX series, at relapse, after multiple lines of therapy. A baseline distribution does not exist.
  2. Does the CX-5461 vulnerability hold in KMT2A-rearranged B-ALL with low KMT2D? The result is myeloid and the lesion in this project's population is a downregulation rather than a knockout. Those are different perturbations and the ribosome-biogenesis phenotype may be dose-dependent.
  3. CX-5461 or CDK4/6? Two preclinical countermeasures now exist for the same lesion class, derived from different mechanisms in different models, and no study compares them. That comparison is cheap and nobody has run it.
  4. How much of the pan-cancer KMT2D mutation burden is length artefact? At 5,537 aa this is the family's largest gene and the field's frequency claims are rarely length-corrected. A corrected ranking would change how seriously a KMT2D variant in a leukaemia panel should be taken.
  5. KMT2D builds transcriptional condensates and its loss changes nuclear MECHANICS (PMID 33169020), while KMT2A works at CpG-island promoters. If a KMT2A fusion redistributes elongation machinery, does it change KMT2D condensate number or size - and would that be visible as a nuclear-architecture phenotype rather than as a gene-expression one? No imaging of KMT2D exists in any haematopoietic cell.
Every source behind this entry (12)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KMT2D is downregulated or mutated in AML, and its deficiency by shRNA knockdown or CRISPR/Cas9 editing accelerates leukaemogenesis in mice; Kmt2d-loss HSPCs and AML cells show enhanced ribosome biogenesis, enlarged nucleoli and increased rRNA and protein synthesis, driven by mTOR pathway activation through direct Kmt2d regulation of Ddit4; the RNA polymerase I inhibitor CX-5461 significantly restrained growth of Kmt2d-loss AML in vivo and extended survival of leukaemic mice.

    PMID 37142882DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20; full text (PMC10323629) not obtained in this beat, so the AML models used, the magnitude of the survival extension and the CX-5461 dosing are not recorded here. Myeloid throughout; no KMT2A rearrangement and no B-lymphoid arm. CX-5461 has known clinical liabilities including photosensitivity and is also a G-quadruplex binder and DNA-damaging agent, so 'ribosome biogenesis inhibition' may not be the whole of its action.

    Evidence that argues against this (1)
    • Ribosome biogenesis and p53 keep appearing as the mechanism whenever a KMT2-family protein is perturbed in leukaemia cells - the WDR5 WIN-site inhibitor result in MLL-rearranged cells is the same shape. That raises the possibility that ribosome-biogenesis vulnerability is a generic property of the perturbed blast rather than a specific consequence of losing KMT2D. Howard GC et al. 2024
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Exome sequencing of ten unrelated probands, with genotypic and phenotypic stratification, identified nonsense or frameshift mutations in MLL2 in seven, with follow-up Sanger sequencing detecting MLL2 mutations in two of the three remaining cases and in 26 of 43 additional cases, establishing MLL2 as a major cause of Kabuki syndrome; the gene in question is KMT2D, resolved by phenotype rather than by symbol.

    Exome sequencing identifies MLL2 mutations as a cause of Kabuki syndrome. · Ng SB, Bigham AW, Buckingham KJ, et al. · Nat Genet 2010
    PMID 20711175DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20. Cited here for two purposes only: to name the germline phenotype of KMT2D loss, and as the canonical published instance of MLL2 denoting KMT2D. **The gene identity is resolved by phenotype** - Kabuki syndrome type 1 is KMT2D and KMT2B's germline phenotype is dystonia - not by any statement in the paper, which predates KMT2 nomenclature. As with KMT2B's dystonia, constitutional haploinsufficiency is not a safety model for partial reversible pharmacology.

    Evidence that argues against this (1)
    • Kabuki syndrome is also caused by KDM6A (Kabuki type 2), the H3K27 demethylase in the same MLL3/4-UTX complex, so the phenotype does not uniquely mark KMT2D; it marks the complex. A naming resolution by phenotype is therefore sound for separating KMT2D from KMT2B and would not separate KMT2D from its own complex partners. Fagan RJ et al. 2019
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In serial patient-derived xenografts, highly pretreated KMT2A::AFF1 ALL samples were much less sensitive to menin inhibition than earlier samples from the same patients; none of the patients had been treated with a menin inhibitor, so the resistance was acquired in the absence of menin-inhibitor exposure; transcriptomic analysis documented sustained on-target efficacy toward the canonical targets; the abstract attributes possible resistance in one patient to downregulation of a gene it writes as 'KMT3D' in a sentence otherwise about KMT2C/D, and reports that KMT2C-edited cells from that patient were selected for in menin-inhibitor-containing growth conditions.

    Upfront menin-inhibitor resistance in multiply pretreated leukemias. · Mahdavi L, Alikarami F, Goodrow H, et al. · Exp Hematol 2026
    PMID 41027490DOI
    What this evidence cannot show

    Abstract fetched from PubMed and read 2026-09-20; the fetched record carries no PMCID, so full text was not obtained through routes 1-4 and route 5 was not attempted. **Two patients.** The KMT2D observation is offered as a possible explanation, not a demonstration. **The abstract as PubMed serves it says 'KMT3D', a gene that does not exist**; this entry reads it as KMT2D from context and records the discrepancy rather than correcting it silently. Also note the abstract writes MLLT4 as 'mixed-lineage leukemia-4', which is AFDN - another instance of the family's naming disorder in a 2026 paper.

    Evidence that argues against this (1)
    • Haploid KMT2C loss gives HSCs a selective advantage specifically when chemotherapy drives them into cycle, which is the published explanation for KMT2C deletions dominating therapy-related myeloid disease. In patients who received multiple cytotoxic lines and no menin inhibitor, a KMT2C/D-deficient clone is the expected finding, so 'selected for in menin-inhibitor-containing conditions' cannot be separated from selection by prior therapy without a pre-treatment timepoint. Chen R et al. 2021
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    KMT2C and KMT2D are H3K4 methyltransferases embedded in large COMPASS-like complexes, and heterozygous somatic mutations in the paralogous MLL3/KMT2C and MLL2(4)/KMT2D genes are now among the most frequently associated with human cancer in large-scale tumour sequencing studies.

    PMID 31128216DOI
    What this evidence cannot show

    A review, cited as a map and as a published instance of the 'MLL2(4)' spelling. Mutation frequency across pan-cancer datasets is strongly length-biased and KMT2D is the longest gene in the family at 5,537 aa, so 'among the most frequently mutated' requires a length-corrected denominator this review does not supply. Abstract read; full text (PMC6638576) not obtained.

    Evidence that argues against this (1)
    • The functional case in myeloid disease rests on the dosage and knockout experiments, not on the mutation counts, and those experiments are mouse. A pan-cancer frequency ranking says where sequencing has looked, not what drives this disease. Xu J et al. 2023
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Quantitative label-free AP-MS across the human SET1/MLL complexes assigned PA1 and PTIP (PAXIP1) as the specific subunits of the MLL3/4 complexes (accessions O14686-3 and Q8NEZ4-3) and found that menin and Psip1/LEDGF do not interact with the Set1a/b or Mll3/4 methyltransferases.

    Quantitative dissection and stoichiometry determination of the human SET1/MLL histone methyltransferase complexes. · van Nuland R, Smits AH, Pallaki P, Jansen PW, Vermeulen M, Timmers HT · Mol Cell Biol 2013
    PMID 23508102DOI
    What this evidence cannot show

    Full text read via the PMC HTML route (PMC3647974) on 2026-09-20 after Europe PMC fullTextXML returned HTTP 500 and efetch db=pmc returned no body. HeLa, GFP-tagged overexpressed baits. The MLL4 bait is accession O14686-3, an isoform record rather than the full 5,537-aa protein, which limits what a negative menin result can exclude.

    Evidence that argues against this (1)
    • A menin contact on a region absent from an isoform-3 construct would be invisible to this experiment, and the absence of menin from MLL3/4 has been repeated in reviews far more often than it has been independently retested. Klonou A et al. 2021
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt reviewed human record fetched 2026-09-20 gives KMT2D as O14686, 5,537 aa, gene_primary KMT2D, synonyms ALR/MLL2/MLL4, protein name 'Histone-lysine N-methyltransferase 2D ... (Myeloid/lymphoid or mixed-lineage leukemia protein 2)'; KMT2B is Q9UMN6, 2,715 aa, carries the same two aliases MLL2 and MLL4, and its protein name is 'Myeloid/lymphoid or mixed-lineage leukemia protein 4' - so the two genes' protein names are swapped relative to their shared aliases.

    What this evidence cannot show

    gene_exact:KMT2D AND organism_id:9606 AND reviewed:true returned exactly one row whose gene_primary equals KMT2D, satisfying rules 62, 63 and 65. Note the accession's visual similarity to NCOA6's Q14686 - one character apart, and NCOA6 is a specific subunit of the same MLL3/MLL4 complex, so the two can be confused in exactly the context where they co-occur.

    Evidence that argues against this (1)
    • A curated synonym list is a snapshot: UniProt has changed these alias sets before and the literature will not change with it. The durable discriminator is length, 5,537 against 2,715, which is why this entry stores it. Larson JK et al. 2023
  7. E7weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A clinical-sequencing case report of a multiply relapsed paediatric B-ALL identified an MLL2-ADCY9 fusion causing truncation of MLL2 as a candidate driver of therapy resistance after an initial response to imatinib for a cryptic ETV6-ABL1 fusion; the gene identity is undecidable from the title and abstract and is resolved only in the discussion, which writes 'MLL2 (KMT2D)'.

    PMID 30028819DOI
    What this evidence cannot show

    Full text read via efetch db=pmc (PMC6339603) on 2026-09-20, 20,274 stripped characters. **n = 1, a case report**, and the authors themselves call the MLL2 truncation 'a possible molecular driver'. No functional work. It is cited here as a documented instance of the naming ambiguity reaching a clinical report, and as an example of a gene identity that only full text can settle - not as evidence about KMT2D's role in resistance.

    Evidence that argues against this (1)
    • The patient's disease is ETV6-ABL1 Ph-like B-ALL, not KMT2A-rearranged, and the resistance in question is to imatinib rather than to a menin inhibitor. Nothing about this case transfers to this project's population beyond the naming lesson. Mahdavi L et al. 2026
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Haploid Kmt2c deletions convey a selective advantage to HSCs only when they are driven into cycle by a strong proliferative stimulus such as chemotherapy, which the authors offer as the explanation for why Kmt2c deletions are commoner in therapy-related MDS/AML than in de novo AML or clonal haematopoiesis.

    PMID 33596429DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20; full text (PMC7951951) not obtained. This is a KMT2C result cited inside a KMT2D entry, and the two are paralogues in the same complex rather than the same gene - the transfer to KMT2D is this entry's inference and is stated as such. Mouse HSCs, adult, no fusion.

    Evidence that argues against this (1)
    • KMT2C and KMT2D are not interchangeable: the Kmt2d-loss phenotype in AML is an mTOR and ribosome-biogenesis phenotype with no reported chemotherapy-conditional selection, so the chemotherapy-confound argument is established for KMT2C and assumed for KMT2D. Xu J et al. 2023
  9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    UniProt O14686's sole nuclear location line for KMT2D rests on ECO:0000269 PubMed:23508102, which is label-free quantitative mass spectrometry of NUCLEAR EXTRACTS of the six human SET1/MLL complexes - a fractionation result, not microscopy.

    What this evidence cannot show

    Recovering a protein from a nuclear extract shows that some of it is nuclear; it cannot exclude other pools and gives no subnuclear resolution. The same publication is the sole support for gene-kmt2b's nuclear call in this graph, so the two entries share one experiment. Rule 66: O14686 is 5,537 aa and is KMT2D; the source paper's own accession table was the discriminator used in the sibling entry.

    Evidence that argues against this (1)
    • HPA v25.1 gives KMT2D main subcellular locations Plasma membrane and Cytosol with Nucleoplasm only additional, at reliability Approved - a call this entry records and does not believe. Human Protein Atlas 2026
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KMT2D (named in the source as 'KMT2D, also known as MLL4') contributes to the assembly of transcriptional condensates through liquid-liquid phase separation, and its haploinsufficiency in Kabuki syndrome impairs Polycomb-dependent chromatin compartmentalisation and alters nuclear architecture and mechanics.

    PMID 33169020DOI
    What this evidence cannot show

    RULE 66 RESOLUTION: the paper's MLL4 is KMT2D, settled by the paper's own abstract, which writes 'KMT2D (also known as MLL4)' and by its subject, Kabuki syndrome type 1, which is caused by KMT2D haploinsufficiency. System is mesenchymal stem cells and chondrocyte commitment - not haematopoietic, not leukaemic. Abstract-level reading. Condensate claims should be graded against localization-condensate-evidence-standards.

    Evidence that argues against this (1)
    • Two objections. The system is Kabuki-syndrome mesenchymal stem cells and chondrocyte commitment, and the rescue is by inhibiting ATR - a checkpoint kinase - which is compatible with an indirect route from KMT2D loss to the phenotype rather than a direct consequence of condensate failure. Second, where this family's phase-separation evidence is most direct it names UTX as the condensing species, with MLL4 recruited into UTX condensates; that ordering makes KMT2D a client rather than a scaffold and would change how the compartment should be described. Nature 2021
  11. E11strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KMT2D is an essential cofactor for GATA1 at erythroid enhancers: it co-localises with GATA1 at more than a thousand erythroid enhancers, co-occupancy is associated with stronger transcription than GATA1 alone, acute depletion rapidly reduces H3K4me1 and H3K27ac at a subset of them, and loss of KMT2D in human erythroid precursors causes developmental arrest.

    PMID 42409830DOI
    What this evidence cannot show

    Human erythroid precursors - haematopoietic but erythroid, not B-lineage, and no KMT2A fusion. This is a chromatin-occupancy result, so it locates KMT2D on the genome and not in the nucleus's three-dimensional space. Abstract-level reading. Rule 66 is not at risk here: the paper uses the current symbol KMT2D throughout.

    Evidence that argues against this (1)
    • An essential differentiation cofactor and a tumour suppressor are different cell-type readings of the same gene, and KMT2D has both: KMT2D DEFICIENCY promotes myeloid leukaemias. A record that says 'KMT2D is required for erythroid maturation' and stops will mislead anyone reasoning about KMT2D loss in myeloid disease, where losing it is transforming rather than arresting. Neither reading generalises to B lineage, and the entry's cell-type field should be read as lineage-plural. Adv Sci (Weinh) 2023
  12. E12weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

    Human Protein Atlas v25.1 assigns KMT2D main subcellular locations Plasma membrane and Cytosol with Nucleoplasm only additional, at reliability Approved - a call that contradicts every other line of evidence for this protein and is recorded rather than suppressed.

    What this evidence cannot show

    Do NOT cite this as evidence that KMT2D is a membrane or cytosolic protein. Approved is HPA's weaker reliability tier, and a plasma-membrane main call for a 5,537-residue H3K4 methyltransferase that occupies a thousand erythroid enhancers is almost certainly an antibody artefact. Logged so the disagreement is visible; the sibling gene KMT2C gets Nucleoplasm main at the higher Enhanced tier from the same pipeline, which is the internal control.

    Evidence that argues against this (1)
    • UniProt's coded experimental citation places KMT2D in nuclear extracts (ECO:0000269 PubMed:23508102), and KMT2D co-occupies more than a thousand erythroid enhancers with GATA1. Nat Commun 2026
Connected to