Oncofusion

start / graph / gene-kmt2c

Genestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q44-kmt2-family

KMT2C (MLL3) - the fetal brake on MLL-rearranged leukaemia, and a 7q tumour suppressor chemotherapy selects against

also called MLL3 · HALR · KIAA1506 · Mll3 (mouse)

KMT2C is a relative of the leukaemia gene KMT2A that acts as a brake rather than an accelerator. In mice, it is the brake that stops a fusion acquired before birth from becoming leukaemia - and deleting just one of its two copies releases it.

The same thing, for a scientist

KMT2C (UniProt Q8NEZ4, 4,911 aa) is the MLL3 subunit of the COMPASS-like MLL3/MLL4 complex, distinguished from the KMT2A/KMT2B branch by carrying PAXIP1/PTIP and NCOA6 instead of menin. It sits at 7q36.1 and is a **haploinsufficient** tumour suppressor in -7/del(7q) AML: an approximately 50% reduction in gene dosage cooperates with other -7/del(7q) events to promote leukaemogenesis, and Mll3-suppressed leukaemias are refractory to conventional chemotherapy but sensitive to JQ1 [E1]. Two results make it directly load-bearing for this project's central disease. First, in a temporally controlled MLL::ENL mouse model, fetal fusion expression creates a heritable leukaemia-resistant state, and **MLL3 is the enforcer of that barrier**: Mll3-sufficient recipients almost all survived beyond 200 days while Mll3-deficient recipients all died with a median survival of 90 days, and deleting a single Mll3 allele greatly accelerated AML onset - an effect that was only modest when the fusion was induced after birth [E2]. Second, in MLL-rearrangement-**negative** infant leukaemia, MLL3 was a compound heterozygote in every infant who developed AML and in 50% of those who developed ALL [E3]. Separately, Kmt2c deletion confers a selective advantage on HSCs specifically when they are driven into cycle by chemotherapy [E4] - which is a serious confound for every report that finds KMT2C loss in a multiply pretreated patient and attributes it to the most recent drug.

Go deeper
Naming: this one is clean, and that is worth saying

MLL3 means KMT2C and has never meant anything else. Unlike MLL2 and MLL4, which each denote two genes across the literature (see concept-kmt2-family-attribution-problem), MLL3 is unambiguous. UniProt Q8NEZ4 lists synonyms HALR, KIAA1506 and MLL3 only, and the protein name is "Myeloid/lymphoid or mixed-lineage leukemia protein 3" 7. The Blood 2026 fetal-barrier paper names its mouse lines as "Mll3/Kmt2c flox" in its own methods, which is the cleanest possible confirmation 2, and the 2013 quantitative complex survey lists MLL3 against accession Q8NEZ4-3 8.

So every MLL3 in this entry resolves to KMT2C, and none of it required inference.

The fetal barrier, and why it matters more here than anywhere else

This project's whole premise is that KMT2A rearrangements arise in utero and that most of them never produce leukaemia. epi-twin-concordance-infant-kmt2a-leukemia, epi-latency-in-utero-to-diagnosis-kmt2a and concept-prevention-window-kmt2a-in-utero all circle the same unexplained fact. 2 proposes a mechanism and names a protein.

The design is a Tet-Off MLL::ENL mouse in which the fusion can be switched on at E10.5 (fetal) or shortly after birth (postnatal). Fetal induction imposed a negative selective pressure on progenitors, followed after birth by loss of self-renewal gene expression and enhanced myeloid differentiation; almost no recipients developed AML despite robust engraftment. Postnatal induction gave highly penetrant AML in every recipient.

Crossing to Mll3 floxed and Mll3 heterozygous lines dissects it:

genotype, fetal inductionoutcome
Mll3 +/+almost all recipients survived >200 days
Mll3 Δ/Δall died, median survival 90 days
Mll3 +/-AML onset greatly accelerated relative to +/+
Mll3 Δ/Δ, postnatal inductiononly a modest effect

The last row is the control that makes it a fetal-specific mechanism rather than a generic tumour-suppressor effect. The authors note Mll3 itself is not differentially expressed through ontogeny or in response to MLL::ENL, so the gene is not being regulated by the fusion - it is a standing requirement 2.

The caveats are real. The fusion is MLL::ENL (KMT2A::MLLT1), myeloid AML, mouse. This project's index fusion is KMT2A::AFF1 in a fetal-type pro-B cell, and nothing here is B-lymphoid. NrasG12D also overcame the barrier, so MLL3 is one route out of at least two.

The chemotherapy confound, and the correction it implies for an entry this beat does not own

resistance-mll34-utx-tumour-suppressor-failure argues that KMT2C/KMT2D loss causes menin-inhibitor resistance by disabling the drug's activating MLL3/4-UTX arm. Its strongest patient observation is that KMT2C-edited cells from a multiply relapsed KMT2A::AFF1 patient were selected for under menin-inhibitor-containing conditions.

4 supplies an alternative explanation that entry does not consider and should. Kmt2c deletion markedly enhances murine HSC self-renewal without altering proliferation rates, and haploid deletion confers a selective advantage only when HSCs are driven into cycle by a strong proliferative stimulus, such as chemotherapy. The authors' own framing is that this explains why Kmt2c deletions are commoner in therapy-related MDS/AML than in de novo AML or clonal haematopoiesis. 1 adds that Mll3-suppressed leukaemias are refractory to conventional chemotherapy.

So in a patient who has had multiple lines of cytotoxic therapy, a KMT2C-deficient clone is the expected finding regardless of what happened afterwards, and "selected for under menin-inhibitor-containing growth conditions" cannot be separated from "selected for by everything that came before" without a pre-menin-inhibitor timepoint. This does not overturn the switch mechanism, which rests on CRISPR screens rather than on patients; it constrains the patient evidence. The correction is reported to the coordinator, not applied here.

What KMT2C is not: it is not in a menin complex

The quantitative survey of all six human SET1/MLL complexes found menin and PSIP1/LEDGF associated with MLL1/2 and not with MLL3/4 or SET1A/B; the MLL3/4 branch carries PTIP (PAXIP1) and PA1 (PAGR1) as its specific subunits instead 8. KMT2C also has no N-terminal menin-binding motif of the KMT2A type 7.

That matters for how the resistance literature should be read. When menin inhibition is said to "let MLL3/4-UTX in", the mechanism is competition for chromatin, not competition for menin. A menin inhibitor cannot displace menin from KMT2C because menin was never on it.

Facts and figures
uniprotQ8NEZ4
length AA4911
gene primary verifiedKMT2C
accession resolutiongene_exact:KMT2C AND organism_id:9606 AND reviewed:true returned exactly one row; gene_primary equals KMT2C; length 4911 asserted
cytoband7q36.1
complex branchCOMPASS-like MLL3/MLL4 (KMT2C/KMT2D) - carries PAXIP1/PTIP and PAGR1/PA1, NOT menin and NOT LEDGF
menin binding motifnone in the canonical N-terminus
tumour suppressor classhaploinsufficient; ~50% dosage reduction is functionally significant
subnuclear localizationUNPUBLISHED. No subnuclear observation of KMT2C exists in any cell type. The family's condensate result belongs to KMT2D and must not be transferred here.
shown for fusionNO for localisation in a KMT2A-rearranged cell. The bounded query returned 3 records and title-checking finds none of them a KMT2C localisation experiment in a KMT2A-fusion context: PMID 32071397 and 30987376 are UTX/KDM6A papers whose MLL is MLL3/MLL4 (the collision registry's UTX row records that UTX is repeatedly mislisted as a KMT2A partner and is not one), and PMID 11718452 is the original HALR/MLL3 cloning and mapping paper. This entry's own fetal-brake result is a genetic and phenotypic result in MLL-rearranged models, not a localisation one.
fetal barrier result
{
  "model": "Tet-Off MLL::ENL (KMT2A::MLLT1), mouse, fetal E10.5 versus postnatal induction",
  "mll3_wt_fetal": "almost all recipients survived >200 days",
  "mll3_null_fetal": "all died, median survival 90 days",
  "mll3_het_fetal": "AML onset greatly accelerated",
  "mll3_null_postnatal": "only a modest effect",
  "pmid": "41032751",
  "title": "Fetal context conveys heritable protection against MLL-rearranged AML that depends on MLL3.",
  "authors": "Mendoza-Castrejon J, Yang W, Denby E, et al.",
  "venue": "Blood",
  "year": 2026,
  "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."
}
aliases searched
[
  "KMT2C",
  "MLL3",
  "HALR",
  "KIAA1506"
]
essentiality
{
  "depmap_common_essential": null,
  "note": "not scored in this beat"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": 0.2231,
  "chronos_mean_other_leukaemia": 0.0542,
  "difference": 0.1689,
  "pan_essentiality_all_lines": 0.0123,
  "selectivity_rank": 17526,
  "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 Q8NEZ4 carries ONE location line on ONE coded citation, ECO:0000269 PubMed:23508102 - the same nuclear-extract quantitative AP-MS of the six human SET1/MLL complexes that supports KMT2D and KMT2B in this graph. Not microscopy. HPA immunofluorescence independently gives main location Nucleoplasm with Cytosol additional at reliability ENHANCED, HPA's highest tier - so for KMT2C, unlike KMT2D, the antibody and the proteomics agree.",
  "subnuclear": "UNPUBLISHED. No subnuclear compartment has been reported for KMT2C in any cell type. The bounded subnuclear query returned exactly ONE record and it is the KMT2D myeloid-leukaemia/ribosome-biogenesis paper, not a KMT2C study.",
  "DO_NOT_TRANSFER_THE_CONDENSATE_RESULT": "The LLPS/transcriptional-condensate result in this family is KMT2D's (PMID 33169020) and the UTX-condensate recruitment result names MLL4 (PMID 34526716). KMT2C and KMT2D are the paralogous enhancer H3K4 monomethyltransferases and are routinely written as 'MLL3/4' as though one object, which is precisely how a KMT2D result would arrive on this entry unearned. Nothing about KMT2C condensates has been shown. Recorded as an absence.",
  "how_shown": "quantitative AP-MS of nuclear extracts (PMID 23508102, no imaging); antibody immunofluorescence at Enhanced reliability (HPA)",
  "tissue_of_evidence": "human cell nuclear extracts and HPA's non-haematopoietic immunofluorescence panel. Nothing haematopoietic has been imaged.",
  "identifier_resolution": "The collision registry's UTX / KDM6A row applies: every primary source placing UTX in a COMPASS-family complex places it with MLL3 and MLL4, meaning KMT2C and KMT2D, NOT KMT2A and NOT KMT2B. KMT2C (Q8NEZ4, 4,911 aa) does not share the MLL2/MLL4 alias tangle that afflicts KMT2B and KMT2D, so the MLL3 name is comparatively safe - but HALR is a further old synonym and carries some of the early literature.",
  "node": "localization-chromatin-bound-versus-soluble-fraction-kmt2a"
}
cell type
{
  "haematopoietic_answer": "PERIPHERAL BLOOD AND FETAL THYMUS by expression, and MYELOID LEUKAEMIA by function; and one absence that is directly relevant to this entry's own argument. UniProt's tissue comment (UNCODED) reads: highly expressed in testis and ovary, then brain and liver; also expressed in placenta, PERIPHERAL BLOOD, FETAL THYMUS, heart, lung and kidney; NOT DETECTED in skeletal muscle and FETAL LIVER. HPA Blood Atlas: low immune cell specificity, Detected in ALL immune cell types.",
  "THE_FETAL_LIVER_LINE_IS_WORTH_STOPPING_ON": "This entry's framing is that KMT2C is a FETAL BRAKE on MLL-rearranged leukaemia, and infant KMT2A::AFF1 is believed to initiate in fetal liver haematopoietic stem cells (cell-type-fetal-liver-hsc). UniProt reports KMT2C as NOT DETECTED IN FETAL LIVER. If that is right, the brake is absent from the tissue where the disease starts - which would be a material qualification of the fetal-brake model. TREAT THIS AS A LEAD, NOT A RESULT: the comment is UNCODED, so it cannot be traced to an experiment, it almost certainly derives from an era of whole-tissue Northern or dot blots rather than from sorted fetal-liver HSCs, and whole fetal liver is dominated by erythroid cells rather than by HSCs. It is recorded because it is checkable and nobody has checked it.",
  "functional_haematopoietic_results": "A KLF4/MLL3 complex axis drives NRBP2 transcription to eliminate AML cells (PMID 42686832), and this entry already records the 7q/del(7q) tumour-suppressor context. Both are myeloid."
}
cell state
{
  "state": "Already recorded in this entry as a fetal/developmental barrier to MLL-rearranged transformation. The localisation-relevant addition is negative: no KMT2C cell-state experiment has ever measured where the protein is, in any state.",
  "bounded": "No KMT2C subnuclear or compartment measurement in any cell state."
}
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": 3,
  "controls": {
    "fusion_clause_alone": 8743,
    "gene_clause_alone": 970
  },
  "reading": "Counted in TITLES AND ABSTRACTS only; supporting figures were not searched (brief rule 80).",
  "title_check": "3 records, 0 of them a KMT2C localisation experiment in a KMT2A-rearranged cell; two are UTX/MLL3-MLL4 papers and one is the HALR/MLL3 cloning paper."
}
How this could bite us back
  • Chemotherapy selects for KMT2C loss. Every cytotoxic backbone this disease is treated with is, on this evidence, enriching for the clone that a later menin inhibitor is reported to fail against - so the resistance lesion may be manufactured by the first-line regimen rather than by the targeted agent.
  • A single allele is enough. Haploinsufficiency means there is no protective heterozygous state and no threshold to stay above; standard variant filtering that demands biallelic loss will call these cases negative.
  • The tumour-suppressor direction is undruggable. Nothing in this entry suggests a way to put KMT2C back. The only actionable consequences are synthetic-lethal - JQ1 in the -7/del(7q) context, CDK4/6 in the menin-resistance context - and both are indirect and neither has worked clinically in AML so far.
  • It complicates the prevention framing this project is drawn to. If a fetal MLL3-dependent barrier is what keeps most in-utero KMT2A rearrangements silent, then anything that perturbs KMT2C function in the fetus - and KMT2C is a huge gene with common variation - is a candidate risk factor, which is a much harder thing to act on than a drug target.
Still open
  1. Does the MLL3 fetal barrier exist in the B-lymphoid lineage? The entire result is MLL::ENL myeloid AML. This project's disease is KMT2A::AFF1 pro-B, and infant ALL is roughly ten times commoner than congenital leukaemia, so the barrier - if it exists there - is leakier or different.
  2. Is KMT2C status measured at diagnosis in any infant KMT2A-rearranged cohort? The one infant-leukaemia germline study that implicates MLL3 is explicitly in the MLL-rearrangement-NEGATIVE subset. Nobody has reported KMT2C compound heterozygosity rates in rearrangement-positive infants, and that is a re-analysis of existing exomes rather than a new experiment.
  3. Are the -7/del(7q) haploinsufficiency result and the chemotherapy-conditional advantage the same mechanism? One says 50% dosage reduction cooperates with other lesions; the other says haploid loss does nothing until the cell is forced to cycle. They may be the same thing seen twice, or two mechanisms, and no experiment distinguishes them.
  4. Would a menin inhibitor be expected to work better or worse in a KMT2C-haploinsufficient blast? The resistance literature says worse, the fetal-barrier literature says such a blast should have been more transformable in the first place, and the two have never been brought together in one model.
  5. UniProt reports KMT2C as not detected in FETAL LIVER (uncoded), and this entry argues KMT2C is a fetal brake on MLL-rearranged leukaemia while the disease is thought to initiate in fetal-liver HSCs. Is KMT2C expressed in sorted human fetal-liver haematopoietic stem cells? A single published atlas lookup would either strengthen the fetal-brake model considerably or force it to be rewritten.
  6. The LLPS/condensate result in this family is KMT2D's, and KMT2C has never been imaged at all. Do KMT2C and KMT2D behave differently as condensate components, and is the routine 'MLL3/4' shorthand hiding a real difference between two proteins the field treats as one?
Every source behind this entry (11)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Using RNAi and CRISPR/Cas9, an approximately 50% reduction in gene dosage of MLL3 (7q36.1) cooperates with other events occurring in -7/del(7q) AMLs to promote leukaemogenesis; Mll3 suppression impairs differentiation of haematopoietic stem and progenitor cells, and Mll3-suppressed leukaemias, like human -7/del(7q) AMLs, are refractory to conventional chemotherapy but sensitive to the BET inhibitor JQ1.

    MLL3 is a haploinsufficient 7q tumor suppressor in acute myeloid leukemia. · Chen C, Liu Y, Rappaport AR, et al. · Cancer Cell 2014
    PMID 24794707DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20; full text (PMC4206212) not obtained in this beat. A mouse model of -7/del(7q) AML driven by combinations of lesions chosen by the authors; MLL3 dosage reduction cooperates with those events and is not shown to be sufficient alone. Myeloid throughout, adult-type disease, no KMT2A rearrangement and no infant material. The JQ1 sensitivity is a cell-line and mouse result from 2014 and BET inhibitors have since failed to deliver in AML clinically.

    Evidence that argues against this (1)
    • Haploinsufficiency inferred from a 50% knockdown in a mouse is not the same as the human lesion: -7/del(7q) removes hundreds of genes, and this project's own standing rule is that an engineered dosage reduction models one candidate out of a deleted segment. The same laboratory group's later work shows Kmt2c loss acts only under a proliferative stimulus, which is an additional requirement the 2014 model does not isolate. Chen R et al. 2021
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a Tet-Off MLL::ENL mouse, fetal (E10.5) fusion induction established a heritable leukaemia-resistant state in which almost all recipients of Mll3+/+ LK cells survived beyond 200 days, whereas all recipients of Mll3-deficient (Mll3 delta/delta) cells died with a median survival of 90 days, and deleting a single Mll3 allele greatly accelerated AML onset; Mll3 deletion had only a modest effect when the fusion was induced shortly after birth, and Mll3 itself is not differentially expressed through ontogeny or in response to MLL::ENL.

    Fetal context conveys heritable protection against MLL-rearranged AML that depends on MLL3. · Mendoza-Castrejon J, Yang W, Denby E, et al. · Blood 2026
    PMID 41032751DOI
    What this evidence cannot show

    Full text read via Europe PMC fullTextXML (PMC12758594) on 2026-09-20, 54,881 stripped characters; the mouse lines are named in the methods as 'Mll3/Kmt2c flox', which is how the gene identity was resolved. **MLL::ENL is KMT2A::MLLT1 and the disease is AML** - this project's index fusion is KMT2A::AFF1 in a fetal-type pro-B cell and no B-lymphoid arm exists here. Transplantation of P0 liver progenitors is a surrogate for the human fetal-to-neonatal transition, not a model of it. NrasG12D also overcame the barrier, so MLL3 is one of at least two routes out.

    Evidence that argues against this (1)
    • The same group's earlier work shows Kmt2c loss confers an HSC self-renewal advantage in general, and specifically under proliferative stress, so 'Mll3 enforces a fetal-specific barrier' has to be read against a gene that also promotes leukaemia-relevant HSC behaviour outside any fetal context. The postnatal-induction control is what separates them, and it is one experiment. Chen R et al. 2021
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In the largest exome-sequencing study of matched mother-infant non-cancer DNA pairs in MLL-rearrangement-negative infant leukaemia, rare potentially functional congenital variation was significantly enriched in acute-leukaemia-associated genes, and MLL3 specifically was a compound heterozygote in every infant who developed AML and in 50% of infants who developed ALL.

    PMID 24301523DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20; full text (PMC4045651) not obtained, so n per arm is not recorded here and 'every infant who developed AML' is a small denominator. **The cohort is explicitly MLL-rearrangement-NEGATIVE infant leukaemia** - the complement of this project's population, not its population. KMT2C is an enormous gene (4,911 aa) with a correspondingly large neutral variant burden, and 'compound heterozygote' from exome data without functional validation is a statistical statement about two non-synonymous alleles, not a demonstration of loss of function.

    Evidence that argues against this (1)
    • Population-scale data show putatively functional germline variation in KMT2A-D is far commoner in cancer-free adults than previously appreciated - 1:112 by manual interpretation - so an enrichment of rare KMT2C variants in a small infant series has a high prior of being background in a very large gene. Larson JK et al. 2023
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Kmt2c deletions markedly enhance murine HSC self-renewal capacity without altering proliferation rates; haploid Kmt2c deletions convey a selective advantage ONLY when HSCs are driven into cycle by a strong proliferative stimulus such as chemotherapy; cycling Kmt2c-deficient HSCs fail to differentiate appropriately, particularly in response to interleukin-1, and impair enhancer recruitment during differentiation - explaining 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. Mouse HSCs, adult, no fusion oncoprotein and no leukaemia endpoint in the core experiments. The chemotherapy used and its schedule are not recorded here. The inference to human therapy-related disease is the authors' and rests on the epidemiological pattern rather than on a human cohort.

    Evidence that argues against this (1)
    • A conditional requirement - advantage only under proliferative stress - cuts against reading KMT2C as a constitutive tumour suppressor, and therefore also against the haploinsufficient-7q-tumour-suppressor framing that the same gene carries in the -7/del(7q) literature. Both are published, they are not obviously the same mechanism, and nobody has reconciled them. Chen C et al. 2014
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    KMT2C and KMT2D are among the genes most frequently mutated in human cancer, are the paralogous MLL3 and MLL2(4) subunits of COMPASS-like complexes, and their heterozygous somatic mutations were revealed by large-scale tumour sequencing rather than by the leukaemia rearrangement literature that established KMT2A.

    PMID 31128216DOI
    What this evidence cannot show

    A review, cited as a map only. Its own abstract writes the second gene as 'MLL2(4)/KMT2D', which is a published acknowledgement of the alias collision rather than a resolution of it. Mutation-frequency rankings across pan-cancer datasets are strongly length-biased and KMT2C is 4,911 aa, so 'most frequently mutated' needs a length-corrected denominator that this review does not supply.

    Evidence that argues against this (1)
    • Frequency of mutation in pan-cancer sequencing is not evidence of driver status, and for very large genes it is systematically inflated. The functional case for KMT2C in myeloid disease rests on the dosage experiments, not on the mutation counts. Chen C et al. 2014
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Loss of Kmt2c or Kmt2d in mouse urothelium impaired differentiation and sensitised to oncogenic transformation, and mechanistically caused redistribution of KMT2A-menin away from KMT2D-localised enhancers onto CpG-high and bivalent promoters with derepression of signal-induced immediate early genes.

    PMID 39806204DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20. **Urothelium, not haematopoiesis**; no leukaemia, no fusion protein, no menin inhibitor. Cited here strictly as orthogonal evidence that the KMT2C/D arm and the KMT2A-menin arm compete for the same chromatin. This paper is already cited in `resistance-mll34-utx-tumour-suppressor-failure`; the identifiers here were re-fetched from PubMed rather than copied from that entry.

    Evidence that argues against this (1)
    • Enhancer landscapes are lineage-specific, so a redistribution demonstrated in urothelium does not establish that the same redistribution happens in a fetal-type pro-B blast. The fetal-barrier result is the only Kmt2c experiment in a haematopoietic MLL-rearranged context, and it says nothing about menin redistribution. Mendoza-Castrejon J et al. 2026
  7. E7established 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 KMT2C as Q8NEZ4, 4,911 aa, gene_primary KMT2C, synonyms HALR/KIAA1506/MLL3 only, protein name 'Histone-lysine N-methyltransferase 2C ... (Myeloid/lymphoid or mixed-lineage leukemia protein 3)'; the canonical N-terminal sequence begins MSSEEDKSVEQPQPPPPPPEEPGAPAPSPAAADKRPRGRPRKD and contains no motif corresponding to the KMT2A menin-binding motif MBM1.

    What this evidence cannot show

    gene_exact:KMT2C AND organism_id:9606 AND reviewed:true returned exactly one row and its gene_primary equals KMT2C, so rules 62/63/65 are satisfied. The absence of an MBM-like motif is an observation about the canonical N-terminus only; menin could in principle contact an internal surface, and no menin-KMT2C binding experiment was found either way.

    Evidence that argues against this (1)
    • Absence of a sequence motif is not absence of an interaction. The positive statement that menin is not in the MLL3/4 complexes comes from the quantitative pulldown, which is a measurement; the sequence observation on its own would not settle it. van Nuland R et al. 2013
  8. E8strong 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 menin and PSIP1 to MLL1/2 and PA1 and PTIP (PAXIP1) to MLL3/4, listing MLL3/4 against accessions O14686-3 and Q8NEZ4-3, and found that menin and Psip1 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 routes 2 and 3 failed. HeLa, GFP-tagged overexpressed baits. The MLL3 and MLL4 baits are isoform-3 accessions, i.e. fragments rather than the full 4,911-aa and 5,537-aa proteins, which is a material limitation for a negative result about an interaction that could in principle be N-terminal.

    Evidence that argues against this (1)
    • A negative pulldown result obtained with isoform-3 fragment constructs cannot exclude a menin contact on a region the construct does not contain. The absence of menin from MLL3/4 is consistent, widely repeated and never independently retested with full-length protein. Klonou A et al. 2021
  9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    UniProt Q8NEZ4's sole nuclear location line for KMT2C rests on ECO:0000269 PubMed:23508102 - quantitative mass spectrometry of nuclear extracts of the human SET1/MLL complexes - and HPA immunofluorescence independently gives main location Nucleoplasm with Cytosol additional at reliability ENHANCED, HPA's highest tier.

    What this evidence cannot show

    One coded citation, and it is a fractionation experiment with no imaging and no subnuclear resolution. The agreement between an antibody at Enhanced reliability and a proteomics experiment is the strongest joint subcellular call in this tranche, and it still says nothing below 'nucleus'. Nothing haematopoietic in either source.

    Evidence that argues against this (1)
    • The agreement is less independent than it looks and neither source resolves anything below 'nucleus'. UniProt's coded citation is a single quantitative AP-MS study of nuclear extracts, and THAT SAME PUBLICATION is the sole support for the nuclear calls on KMT2B and KMT2D in this graph - three entries resting on one experiment, which is a correlated rather than a corroborating structure. A fractionation result plus an antibody that reports 'Nucleoplasm' leaves the subnuclear question entirely open. Mol Cell Biol 2013
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Human Protein Atlas v25.1 assigns KMT2C main subcellular location Nucleoplasm (Cytosol additional) at reliability ENHANCED and records it as detected in all immune cell types at low specificity - and the same pipeline gives the paralogue KMT2D a Plasma membrane main call at the weaker Approved tier, which is the internal control showing the KMT2D call is the outlier.

    What this evidence cannot show

    Antibody immunofluorescence in non-haematopoietic lines. The paralogue comparison is an argument about antibody quality within one pipeline, not an independent measurement of either protein. Blood Atlas values are RNA.

    Evidence that argues against this (1)
    • Comparing two antibodies inside one pipeline is not an independent measurement, and the argument could run the other way: KMT2D's outlier call might reflect a genuine extranuclear pool that nobody has looked for rather than a bad antibody. There is precedent for regulated distribution in this branch of the family - cancer-derived UTX TPR mutations that impair interaction with MLL3/4 complexes change UTX's subcellular localisation, showing the complex governs where its members sit. If retention is partner-dependent, an unexpected compartment for one paralogue is not automatically an artefact. Oncogene 2020
  11. E11weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

    UniProt's UNCODED tissue comment for KMT2C reports expression in peripheral blood and fetal thymus and states that KMT2C is NOT DETECTED IN FETAL LIVER - the tissue in which infant KMT2A::AFF1 is thought to initiate, and therefore a direct and checkable qualification of this entry's fetal-brake framing.

    What this evidence cannot show

    The comment carries NO evidence code, so it cannot be traced to an experiment and is almost certainly from whole-tissue blotting rather than from sorted fetal-liver HSCs; whole fetal liver is dominated by erythroid cells, in which a low signal would not exclude expression in the rare HSC compartment. Recorded as a LEAD for someone to check against a fetal-liver HSC RNA-seq dataset, not as a finding. Brief rule 91 item 2: read the uncoded comments, record them as uncoded, do not discard them and do not promote them.

    Evidence that argues against this (1)
    • A negative from an uncoded whole-tissue comment is outweighed by demonstrated function in blood cells: a KLF4/MLL3 complex axis drives NRBP2 transcription in acute myeloid leukaemia, which requires KMT2C to be present and active in haematopoietic cells. The fetal-liver line should therefore be treated as a lead about one developmental tissue at one detection threshold, not as evidence that KMT2C is absent from the haematopoietic system, and it must not be allowed to weaken the fetal-brake model on its own. Leukemia 2026
Connected to