Oncofusion

start / graph / gene-kmt2e

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

KMT2E (MLL5) - the family member with a dead SET domain, and the exclusion from COMPASS is justified

also called MLL5 · Mll5 (mouse)

KMT2E carries the same catalytic module as the rest of its family but it does not work: the crystal structure shows it is missing the parts that grip the target and the methyl donor. It is still a real blood-cancer gene, but for reasons that have nothing to do with the enzyme activity the family is named after.

The same thing, for a scientist

KMT2E (UniProt Q8IZD2, 1,858 aa) was assigned to the KMT2 family on SET-domain homology alone. The 2.1 Angstrom crystal structure of its unliganded SET domain shows that missing key residues, plus an unusually large loop in the SET-I subdomain, preclude interaction with both cofactor and substrate, and the same study demonstrated no in vitro methyltransferase activity on full-length histones or H3 peptides [E1]. **So the routine exclusion of KMT2E from COMPASS discussions is justified on the catalysis, and UniProt's own protein name now reads 'Histone reader KMT2E (Inactive histone-lysine N-methyltransferase 2E)' [E4].** It does not follow that KMT2E is irrelevant to this disease: MLL5 transcript level is an independently validated favourable prognostic marker in 509 AML patients with replication in 167 more [E2], KMT2E-ASNS is a relapse-specific fusion gene in early T-cell precursor ALL [E3], and MLL5 lacks WRAD but binds chromatin through a PHD finger and partners HCF-1 [E5]. **This entry exists partly as a warning about the family-attribution problem: a paper that says 'MLL5 regulates H3K4 methylation' is describing an indirect effect of a catalytically dead protein, and this graph should never let such a claim stand in for KMT2A catalysis.**

Go deeper
Is the exclusion from COMPASS justified in this disease? Yes, on catalysis; no, on relevance

The dispatch asked exactly this. The two halves separate cleanly.

On catalysis: justified, and structurally. 1 is not an assay failure that could be a technical artefact - it is a structure that explains the assay. The SET fold is present and canonical in most respects, but key residues are absent and an oversized SET-I loop blocks the cofactor and substrate channel. The authors tested full-length histones and H3 peptides and found no activity. UniProt has since renamed the protein 'Inactive histone-lysine N-methyltransferase 2E' and reclassified it as a histone reader 4. Nothing about H3K4 writing by the COMPASS family transfers to KMT2E, and it has no WRAD and no menin.

On relevance to leukaemia: not justified. Three independent lines say so:

  • Prognosis. In 509 AML patients from two multicentre trials, high MLL5 expression independently predicted better overall survival (HR 0.66, 95% CI 0.49-0.89, P = 0.007) and higher complete response, with independent validation in 167 patients from AMLSG 07/04 (OS HR 0.5, 95% CI 0.27-0.92, P = 0.023). The effect held in cytogenetically normal AML and in core-binding-factor leukaemias 2.
  • Rearrangement. KMT2E itself is a fusion partner: KMT2E-ASNS is reported as a relapse-specific fusion gene in early T-cell precursor ALL 3. A KMT2-family gene other than KMT2A that rearranges in acute leukaemia is directly on this project's beat and is written nowhere in this graph.
  • Chromatin. MLL5 binds H3 through a PHD finger and associates with HCF-1, so it occupies chromatin without writing on it 5.

The honest summary is that KMT2E belongs in this graph as a reader and prognostic marker, and belongs out of every sentence about H3K4 methyltransferase activity.

Why a dead enzyme in the family is a family-attribution hazard

This graph's standing problem - stated for drugs in the inhibitor-class generalisation and restated for proteins in concept-kmt2-family-attribution-problem - has a sharp instance here. Searches for 'MLL' or 'KMT2' plus a chromatin phenotype return MLL5 papers, several of which describe effects on H3K4 methylation. Those effects are real and they are indirect, because the protein cannot methylate anything 1. A 2009 study titled around MLL5 indirectly regulating H3K4 methylation says so in its own title.

So the rule for this graph: a claim about H3K4 methyltransferase activity that originates in an MLL5/KMT2E paper cannot be used anywhere in the KMT2A arm, however similar the vocabulary. And the converse: a KMT2E prognostic or fusion result must not be quietly upgraded into a statement about the KMT2 catalytic family.

Facts and figures
uniprotQ8IZD2
length AA1858
gene primary verifiedKMT2E
accession resolutiongene_exact:KMT2E AND organism_id:9606 AND reviewed:true returned exactly one row; gene_primary equals KMT2E; length 1858 asserted
catalytically activefalse
catalysis evidence2.1 A crystal structure of the unliganded SET domain plus in vitro assays on full-length histones and H3 peptides (PMID 27812132)
wrad bindingnone reported; KMT2E has no WIN motif and is not a WRAD-assembling enzyme
menin bindingnone reported; no MBM-like N-terminal motif
chromatin bindingPHD finger, direct interaction with the H3 N-terminus; partners HCF-1
leukaemia relevancehigh MLL5 transcript level is an independently validated FAVOURABLE prognostic marker in adult AML (n=509 plus 167 validation); KMT2E-ASNS is a relapse-specific fusion in early T-cell precursor ALL
exclusion from compassjustified on catalysis, not on leukaemia relevance
aliases searched
[
  "KMT2E",
  "MLL5"
]
pubmed counts 2026 09 20
{
  "(KMT2E OR MLL5) AND (leukemia OR leukaemia)": 52
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.0832,
  "chronos_mean_other_leukaemia": -0.102,
  "difference": 0.0188,
  "pan_essentiality_all_lines": -0.1086,
  "selectivity_rank": 11562,
  "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."
}
How this could bite us back
  • Any KMT2-family-directed agent that works through the conserved SET fold will bind KMT2E and do nothing there, wasting occupancy on a dead site - a selectivity problem that produces no phenotype and therefore no warning signal in a cell assay.
  • High KMT2E is FAVOURABLE in AML. If a therapy incidentally lowers KMT2E expression, the direction of the prognostic association predicts harm, and nobody measures it.
  • The naming carries the hazard. A protein called 'lysine methyltransferase 2E' in a family defined by methyltransferase activity will keep being cited for methylation claims it cannot support, and this graph has 1,397 entries in which such a claim could hide.
Still open
  1. Is MLL5 expression prognostic in KMT2A-rearranged AML specifically? The 509-patient study reports cytogenetically normal AML and core-binding-factor leukaemias separately but does not report the KMT2A-rearranged subgroup, and that is a re-analysis of an existing dataset rather than a new study.
  2. Does KMT2E expression track with response to menin inhibition? It has no menin contact, so a correlation would have to be indirect - which makes it a clean test of whether MLL5's prognostic effect is a methylation-state proxy rather than a property of the protein.
  3. Can a catalytically dead SET domain still template a complex? The structure shows the cofactor and substrate channel blocked; it does not show the RBBP5/ASH2L docking surface is unusable. Nobody has asked whether KMT2E sequesters any COMPASS component.
  4. What is the KMT2E-ASNS fusion protein? The report is a letter with no abstract in PubMed. Whether it is in frame, whether a protein is made, and how many cases carry it are all unread.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The crystal structure of the unliganded SET domain of human MLL5 at 2.1 Angstrom resolution shows that a lack of key residues, together with an unusually large loop in the SET-I subdomain, precludes interaction with both cofactor and substrate; accordingly MLL5 was shown to be devoid of any in vitro methyltransferase activity on full-length histones and on histone H3 peptides.

    PMID 27812132DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20; full text (PMC5094779) not obtained in this beat. The structure is of the isolated, unliganded SET domain of a 1,858-aa protein - a negative catalytic result on an isolated domain cannot exclude activity that requires a partner subunit, a post-translational modification or a non-histone substrate. The assay substrates were histones and H3 peptides only.

    Evidence that argues against this (1)
    • A review of the same protein notes that the mechanism of action and physiological function of MLL5 remain poorly understood despite the catalytic result, which means 'no activity on these substrates in this construct' has been generalised to 'inactive' faster than the evidence strictly allows. Zhang X et al. 2017
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 509 AML patients treated in trials AML SHG 0199 and AML SHG 0295, high MLL5 transcript level independently predicted better overall survival (HR 0.66, 95% CI 0.49-0.89, P = 0.007), better relapse-free survival (HR 0.72, 95% CI 0.52-1.01, P = 0.057) and a higher complete response rate (OR 1.87, 95% CI 1.08-3.24, P = 0.026), holding in cytogenetically normal AML (n = 268) and core-binding-factor leukaemias (n = 81), and was independently validated in 167 patients from AMLSG 07/04 (OS HR 0.5, 95% CI 0.27-0.92, P = 0.023).

    PMID 21205756DOI
    What this evidence cannot show

    Abstract fetched and read 2026-09-20; the fetched record carries no PMCID, so full text was not obtained through routes 1-4. **Adult AML, 2011 treatment era, no KMT2A-rearranged subgroup analysis and no paediatric or infant arm.** The cut-point between 'high' (n = 127) and 'low' (n = 382) expression is not recorded here and a 1:3 split suggests a chosen rather than a natural threshold. RT-PCR transcript level is not protein. The title calls MLL5 a histone methyltransferase, which the 2016 structure shows it is not.

    Evidence that argues against this (1)
    • A later study frames MLL5 expression as a biomarker for DNA hypermethylation and sensitivity to epigenetic therapy rather than as a prognostic marker in its own right, which would make the survival association a proxy for methylation state rather than a property of MLL5. Zhang X et al. 2017
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    KMT2E-ASNS is reported as a novel relapse-specific fusion gene in early T-cell precursor acute lymphoblastic leukaemia - that is, a KMT2-family gene other than KMT2A participating in a leukaemia fusion.

    PMID 28069604DOI
    What this evidence cannot show

    **The PubMed record carries an empty abstract** - it is a Letter, and the entire claim above is the title. Full text (PMC5374844) was not retrieved in this beat, so the number of cases, the breakpoint, the reading frame and whether a protein product was demonstrated are all unknown here. This is deliberately graded moderate and should not be cited for anything beyond the existence of the fusion until the letter is read. Early T-cell precursor ALL is not this project's population.

    Evidence that argues against this (1)
    • A single relapse-specific fusion reported in a letter, in a disease subtype distant from infant KMT2A-rearranged leukaemia, is the weakest kind of evidence that a family member is oncogenic. The substantive KMT2E leukaemia evidence is expression-prognostic, and that points the opposite way - high MLL5 is FAVOURABLE. Damm F et al. 2011
  4. E4established 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 KMT2E as Q8IZD2, 1,858 aa, gene_primary KMT2E, sole synonym MLL5, and protein name 'Histone reader KMT2E (Inactive histone-lysine N-methyltransferase 2E) (Inactive lysine N-methyltransferase 2E) (Myeloid/lymphoid or mixed-lineage leukemia protein 5)' - the database has reclassified the protein as a reader rather than a writer.

    What this evidence cannot show

    gene_exact:KMT2E AND organism_id:9606 AND reviewed:true returned exactly one row whose gene_primary equals KMT2E. A curated protein name reflects curators' reading of the literature at a point in time and is not itself an experiment; the experiment is the structure in E1. The alias MLL5 is unambiguous - unlike MLL2 and MLL4 it denotes only this gene.

    Evidence that argues against this (1)
    • Database reclassification propagates: once UniProt calls a protein inactive, downstream tools and reviews repeat it, and any later demonstration of conditional activity has to overturn an entrenched annotation rather than an open question. Mas-Y-Mas S et al. 2016
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    MLL5 was assigned to the KMT2 family on SET-domain homology alone, appears to lack intrinsic histone methyltransferase activity towards histone substrates, is distinct from the other family members, and has reported roles in cell cycle progression, genomic stability maintenance, adult haematopoiesis and spermatogenesis, with chromatin binding through a PHD domain.

    MLL5 (KMT2E): structure, function, and clinical relevance. · Zhang X, Novera W, Zhang Y, Deng LW · Cell Mol Life Sci 2017
    PMID 28188343DOI
    What this evidence cannot show

    A review, cited as a map. Its own conclusion is that the mechanism of action and physiological function of MLL5 remain poorly understood, so nothing in it should be treated as settled beyond the negative catalytic result that E1 establishes independently. Abstract read 2026-09-20; full text not obtained.

    Evidence that argues against this (1)
    • A review cannot settle whether the exclusion of KMT2E from COMPASS is appropriate, because the question is about what the protein does in a specific disease and the review is pan-biology. The AML prognostic data are the only large human dataset and they were generated before the catalytic result and describe MLL5 as a methyltransferase. Damm F et al. 2011
Connected to