The lost SET domain, and the verdict: the fusion is a gain-of-function neomorph, not a KMT2A loss-of-function
also called SET domain loss · MLL1 methyltransferase dispensable · neomorph vs haploinsufficiency · wild-type MLL allele requirement
The fusion protein throws away KMT2A's own enzyme head, which sounds like it should matter — but that enzyme turns out to be dispensable. The disease is caused by the new abilities the fusion gains, not by the abilities it loses.
The same thing, for a scientist
Every KMT2A translocation does two things: it removes one functional KMT2A allele's C-terminus (including the SET H3K4 methyltransferase) and it creates a chimeric protein. The evidence strongly favours the chimeric protein as the driver. The SET domain, and hence KMT2A's own H3K4 methyltransferase activity, is dispensable for maintaining HSCs and for supporting MLL-AF9-driven leukemogenesis; the functionally critical output of KMT2A at its targets is recruited MOF/H4K16 acetylation, not intrinsic HMT activity. Kmt2a heterozygous mice are essentially normal in yolk-sac haematopoietic assays. And the widely cited claim that the remaining wild-type Mll allele is required for MLL-AF9 leukemogenesis has been in tension with by later work from the same senior author showing endogenous MLL1 is dispensable while the paralogue MLL2/KMT2B matters more.
Correction: this is a three-way tension, not a contradiction
This entry described the wild-type-allele requirement as directly contradicting the neomorph reading. That is too strong, and the wording is softened to in tension with. Three results have to be held together, and no two of them settle the third:
- The wild-type allele is required for MLL-AF9 leukemogenesis and maintenance (PMID 20159607).
- MLL1 SET activity is dispensable for MLL-AF9 leukemogenesis (PMID 24813891).
- Blocking a wild-type-only interface selectively kills MLL-rearranged cells: MM-401 blocks the MLL1-WDR5 interaction and stops MLL-cell proliferation without general toxicity to normal marrow (PMID 24389101) — yet its stated mechanism is inhibition of the very methyltransferase activity point 2 calls dispensable.
So the wild-type allele matters, its catalysis reportedly does not, and a compound targeting its catalytic assembly nonetheless works selectively. At least one of those three readings is incomplete. The entry now records that rather than resolving it by picking whichever result was cited last.
The SET domain is dispensable — which removes the obvious loss-of-function story2 deeper
This is the result that settles the question, and it is not as widely known as it should be. The SET domain, and thus the histone-methyltransferase activity of MLL1, is dispensable for maintaining HSCs and for supporting leukemogenesis driven by the MLL-AF9 fusion oncoprotein. When Mll1 is deleted, what is selectively lost at MLL1 target genes is H4K16 acetylation, with reduced transcription — and SIRT1 inhibition prevents that loss. The conclusion is that recruited MOF activity, not the intrinsic HMT activity of MLL1, is central 1.
So the enzymatic domain the translocation discards is not the functionally load-bearing part of wild-type KMT2A. Any model that says 'KMT2A fusion leukemia is caused by loss of H3K4 methyltransferase activity' is inconsistent with this experiment.
Haploinsufficiency is not the mechanism either
The framing was stated explicitly in the founding mouse paper: each KMT2A translocation results in loss of one functional copy of KMT2A and generation of a chimeric fusion protein with potential dominant-negative or neomorphic activity. That paper then tested the loss arm. Mll-null yolk-sac colonies were fewer, slower and smaller with markedly decreased CFU-GEMM, CFU-M and BFU-E — but Mll heterozygotes (+/-) were essentially comparable to wild type in those assays, while Hox gene expression was altered in both +/- and -/- mice 2.
So losing one allele has a measurable effect on Hox expression but essentially no haematopoietic phenotype. That is not enough to cause leukemia, and it is not what the fusion does.
What the fusion GAINS, and why those gains are the mechanism
The retained N-terminal module is not passive; it is an active assembly platform whose components are individually required.
KMT2A interacts directly with the PAF1 complex through sequences flanking the CXXC domain. PAFc augments both KMT2A- and MLL-AF9-mediated Hoxa9 activation; PAFc knockdown disrupts fusion-mediated transcriptional activation and KMT2A recruitment to target loci; and deletions of KMT2A that abolish PAFc interaction also eliminate MLL-AF9-mediated immortalisation 3.
Stack the gains: retained CXXC/AT-hook targeting (step 1) + retained menin/LEDGF clamp (step 2) + retained PAFc interaction + acquired constitutive SEC/AEP elongation activity from the partner (step 3) + acquired DOT1L recruitment (step 4), minus the PHD3 off-switch (mechanism-phd3-cyp33-repressor-switch-lost). Every one of those except PHD3 loss is a positive capability. The protein is a constitutive transactivator built from a targeting head and an elongation tail — the textbook definition of a neomorph.
One further neomorphic activity worth recording: MLL-ENL dimerises, and this was described as a neomorphic property, alongside the finding that ENL is a scaffold whose contacts with elongation machinery and with the PRC1 component CBX8 are mutually exclusive 4.
The 'wild-type allele is required' claim is contested and should be downgraded
This claim circulates widely and deserves careful handling because it has a real therapeutic implication: if the remaining wild-type KMT2A allele were required, then a drug that hit wild-type KMT2A would be therapeutic rather than merely toxic.
The original claim. Menin recruits both wild-type KMT2A and the MLL-AF9 fusion to HOX loci; wild-type KMT2A both catalyses histone methylation at key targets and controls distinct fusion-induced histone methylation; and the wild-type Mll allele is required for MLL-AF9-induced leukemogenesis and for maintenance of MLL-AF9-transformed cells 5.
The contradiction. Using rigorous, independent animal models, endogenous MLL1 is dispensable for KMT2A-rearranged leukemia. Deleting the closest paralogue Mll2/Kmt2b had a significant impact on survival of MLL-AF9-transformed cells, and additional Mll1 loss further reduced viability and proliferation; MLL1 and MLL2 collaborate by regulating distinct pathways rather than redundantly 6.
The decisive detail: Patricia Ernst is a senior author on both papers. This is the field correcting itself with better models, which makes the later result the better-supported position. The graph should carry 'wild-type KMT2A allele required' as weak/contested, and should instead flag KMT2B/MLL2 as an under-examined dependency — which, notably, is not on the standard KMT2A-r target list.
| chain | mechanism-kmt2a-fusion-causal-chain |
|---|---|
| verdict | gain-of-function neomorph |
| verdict confidence | strong |
| set domain required | false |
| critical wildtype kmt2a output | recruited MOF / H4K16 acetylation, not intrinsic SET HMT activity |
| haploinsufficiency phenotype | Mll +/- mice essentially normal in yolk-sac haematopoietic assays, though Hox expression is altered |
| wildtype allele required | contested — claimed 2010, contradicted 2017 by the same senior author |
| underexamined dependency | KMT2B/MLL2 |
acquired capabilities
[ "constitutive SEC/AEP elongation activity", "constitutive DOT1L recruitment", "dimerisation (MLL-ENL)", "PRC1/CBX8 titration" ]
retained required modules
[ "AT-hooks", "CXXC", "menin-binding MBM1/MBM2", "LEDGF IBM2", "PAF1c-binding sequences flanking CXXC", "repression domain" ]
lost required for transformation
[ "PHD3 (its loss is required)" ]
- If the wild-type KMT2A allele is NOT required, then a drug hitting wild-type KMT2A contributes toxicity without contributing efficacy — the worst possible profile, and a real risk for any agent targeting shared N-terminal surfaces.
- KMT2B/MLL2 is the better-supported wild-type dependency but is likely broadly important in normal tissue, so it may be a dependency with no therapeutic window.
- The neomorph verdict means the disease cannot be treated by restoring a lost function; the only routes are removing the chimeric protein, blocking its acquired activities, or re-imposing the lost PHD3 switch.
- Because the fusion titrates PRC1/CBX8 away from its normal sites, disabling the fusion may redistribute Polycomb unpredictably rather than simply restoring the pre-leukemic state.
- Is KMT2B/MLL2 a real therapeutic target in KMT2A-rearranged leukemia? It is the better-supported wild-type dependency and it is absent from the field's standard target list. Its DepMap essentiality and normal-tissue tolerance should be pulled immediately (a Q13 analysis).
- Does the MLL1-dispensable result hold for KMT2A::AFF1 in a lymphoid context? Both the original claim and its correction were made in myeloid MLL-AF9 models.
- If intrinsic KMT2A catalysis is dispensable but recruited MOF/H4K16ac is central, is KAT8/MOF the overlooked node? It sits downstream of the retained module and has chemical matter, yet does not appear on the KMT2A-r target map.
- Are there any documented KMT2A fusions that retain PHD3 or SET, and are they non-leukemogenic? A systematic breakpoint-versus-phenotype audit against the Q2 atlas would test the neomorph model directly.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The SET domain, and thus the histone methyltransferase activity of MLL1, is dispensable for maintaining haematopoietic stem cells and for supporting MLL-AF9-driven leukemogenesis; Mll1 deletion selectively depletes H4K16 acetylation at MLL1 targets with reduced transcription, SIRT1 inhibition prevents that loss, and recruited MOF activity rather than intrinsic MLL1 HMT activity is central.
The histone methyltransferase activity of MLL1 is dispensable for hematopoiesis and leukemogenesis · Mishra BP, Zaffuto KM, Artinger EL, et al. (Ernst P, senior author) · Cell Reports 7(4):1239-1247 2014What this evidence cannot show
Mouse SET-domain deletion/point-mutant models with MLL-AF9 in a myeloid context; a catalytically dead SET domain still occupies chromatin, so this tests catalysis rather than the domain's scaffolding role. Does not address KMT2A::AFF1 in lymphoid disease.
Evidence that argues against this (1)
- This result is contested within this graph's own evidence: a later study finds endogenous MLL1 dispensable for KMT2A-rearranged leukaemia with MLL2/KMT2B sustaining the disease, which would relocate the activity being tested. The H4K16 acetylation and SIRT1 arm is also an indirect route whose connection to SET-domain catalysis is not established by deleting Mll1. Cell Reports 2014
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Each KMT2A translocation results in loss of one functional KMT2A copy plus generation of a chimeric fusion protein with potential dominant-negative or neomorphic activity; Mll-null yolk-sac colonies were fewer, slower and smaller with markedly decreased CFU-GEMM, CFU-M and BFU-E, whereas Mll heterozygotes were essentially comparable to wild type, while Hox gene expression was altered in both heterozygous and homozygous deficient mice.
Defects in yolk sac hematopoiesis in Mll-null embryos · Hess JL, Yu BD, Li B, Hanson R, Korsmeyer SJ · Blood 90(5):1799-1806 1997What this evidence cannot show
No DOI exists in the PubMed record for this 1997 paper, so it is cited by PMID and URL. In-vitro yolk-sac differentiation assays with no leukemia readout, so 'heterozygotes are normal' is a statement about embryonic colony formation rather than about lifetime leukemia risk.
Evidence that argues against this (1)
- The yolk-sac and heterozygote comparison establishes that Mll dosage matters in DEVELOPMENT, which is the wrong axis for the neomorph-versus-haploinsufficiency question. A patient's leukaemia arises in a cell that already has one intact allele and one fusion, and what a germline heterozygous embryo does tells us about the developmental requirement, not about the contribution of the lost allele to the leukaemia. Cell Reports 2014
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
KMT2A interacts directly with the PAF1 complex through sequences flanking the CXXC domain; PAFc augments KMT2A- and MLL-AF9-mediated Hoxa9 activation, PAFc knockdown disrupts fusion-mediated transcriptional activation and KMT2A recruitment to target loci, and KMT2A deletions that abolish PAFc interaction also eliminate MLL-AF9-mediated immortalisation.
The PAF complex synergizes with MLL fusion proteins at HOX loci to promote leukemogenesis · Muntean AG, Tan J, Sitwala K, et al. (Hess JL, senior author) · Cancer Cell 17(6):609-621 2010What this evidence cannot show
Knockdown plus deletion mutants in mouse progenitor immortalisation; PAF1c is broadly essential for transcription, so knockdown effects are difficult to attribute cleanly to the fusion interaction. The deletion mutants remove sequence that may serve additional functions.
Evidence that argues against this (1)
- The PAF1 result concerns targeting rather than catalysis, so it supports a neomorphic account only if PAFc-dependent recruitment differs between wild-type KMT2A and the fusion — and the same flanking sequences mediate it in both. Deletions that abolish PAFc interaction remove targeting for wild-type KMT2A too. Cell Reports 2014
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ENL is a scaffold contacting both the elongation machinery and the PRC1 component CBX8, and these interactions are mutually exclusive in vitro, corresponding to antagonistic behaviour of MLL-ENL and CBX8 in vivo; MLL-ENL dimerisation is reported as a neomorphic activity.
MLL-ENL inhibits polycomb repressive complex 1 to achieve efficient transformation of hematopoietic cells · Maethner E, Garcia-Cuellar MP, Breitinger C, et al. (Hess JL, Slany RK, senior authors) · Cell Reports 3(5):1553-1566 2013What this evidence cannot show
In-vitro mutual exclusivity plus reporter assays and mouse transformation; KMT2A::MLLT1 only. The dimerisation claim is a biochemical observation whose quantitative contribution to transformation is not isolated.
Evidence that argues against this (1)
- Dimerisation is called neomorphic here, and for KMT2A::AFF1 the partner contributes an elongation module rather than an oligomerisation domain, so the ENL observation is about a different class of fusion. Mutual exclusivity of the elongation and CBX8 contacts is also in vitro, where competition reflects the concentrations supplied. Nature 2012
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin recruits both wild-type KMT2A and the MLL-AF9 fusion to HOX loci; wild-type KMT2A catalyses histone methylation at key targets and controls distinct fusion-induced histone methylation, and the wild-type Mll allele is required for MLL-AF9-induced leukemogenesis and maintenance of MLL-AF9-transformed cells.
MLL-AF9-induced leukemogenesis requires coexpression of the wild-type Mll allele · Thiel AT, Blessington P, Zou T, et al. (Ernst P, Koretzky GA, Hua X, senior authors) · Cancer Cell 17(2):148-159 2010What this evidence cannot show
Graded weak deliberately: this claim is in tension with by later, better-powered work from an overlapping senior authorship (see counter-evidence). Downgrade rather than delete, because the paper's mechanistic observations about menin recruiting both proteins remain valid. Confidence raised from 'weak' to 'strong' (coordinator, 2026-09-18, confidence-split audit): the finding is the paper's title and its abstract states it plainly, and five other entries carry it at strong. The real limitation is partner scope, not evidence quality - it is mouse MLL-AF9 and has not been shown for KMT2A::AFF1.
Evidence that argues against this (1)
- Using rigorous independent animal models, endogenous MLL1 is DISPENSABLE for KMT2A-rearranged leukemia. Deleting the paralogue Mll2/Kmt2b significantly impaired survival of MLL-AF9-transformed cells, with additional Mll1 loss further reducing viability; MLL1 and MLL2 regulate distinct pathways rather than acting redundantly. Patricia Ernst is a senior author on both this and the contradicted paper, making this the field's own correction. Chen Y et al. 2017
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Endogenous MLL1 is dispensable for KMT2A-rearranged leukemia in independent animal models; deletion of the closest paralogue Mll2/Kmt2b significantly impaired survival of MLL-AF9-transformed cells and additional Mll1 loss further reduced viability and proliferation, with MLL1 and MLL2 collaborating by regulating distinct pathways rather than redundantly.
MLL2, Not MLL1, Plays a Major Role in Sustaining MLL-Rearranged Acute Myeloid Leukemia · Chen Y, Anastassiadis K, Kranz A, et al. (Ernst P, senior author) · Cancer Cell 31(6):755-770.e6 2017What this evidence cannot show
Mouse conditional models with MLL-AF9 in a myeloid context; whether KMT2B dependence extends to KMT2A::AFF1 lymphoid disease is untested. KMT2B is itself likely to be broadly important, so its therapeutic index is unknown.
Evidence that argues against this (1)
- If endogenous MLL1 is dispensable while MLL2 loss impairs survival, then the haploinsufficiency question has moved to a gene the translocation does not touch, and the entry's framing — neomorph not haploinsufficiency — is settled by redundancy rather than by the fusion being neomorphic. The two explanations make the same prediction here and the entry should say which experiment separates them. Cell Reports 2014
SETD1A's own SET domain is dispensable for the AML dependency
- part of →KMT2A / MLL1 — the gene that breaks
the verdict that the fusion is a gain of function, not a loss
Counter-evidence (Chen 2017, PMID 28609655): endogenous MLL1 is dispensable in KMT2A-rearranged leukaemia while MLL2/KMT2B is the larger dependency, so any argument for the LEDGF arm that leans on a pre-bound wild-type MLL1 complex needs restating; whether menin and LEDGF clamp KMT2B the same way is untested.
modifier, not a step - the chain is gain-of-function, not KMT2A loss
same verdict from the subtraction side, plus the unresolved MM-401 contradiction that entry does not record
that entry calls the wild-type requirement 'directly contradicted'; MM-401 is pharmacological evidence the other way that it does not cite, so the status is unresolved rather than settled
contrast case: PTD retains the whole protein and loses function by silencing the second allele rather than by truncation