In a cell with no fusion, a menin inhibitor is substantially a KMT2B drug
also called the KMT2B half of menin-inhibitor toxicity · normal-tissue menin pharmacology
In a leukaemia cell, a menin inhibitor mostly attacks the fusion protein. In every other cell in the child's body there is no fusion, so the drug can only be acting on menin's two normal partners - and one of those, KMT2B, has never been measured under the drug at all.
The same thing, for a scientist
**Statement.** In non-malignant tissue, where no KMT2A fusion exists, the pharmacologically relevant consequences of menin-MBM-pocket occupancy are distributed between menin-KMT2A and menin-KMT2B roughly in proportion to their occupancy, and the KMT2B share is non-trivial - plausibly near half, given that menin is recovered with MLL1 and MLL2 at equal stoichiometry in the one quantitative dataset that measured it [E1]. **Origin: literature gap.** The efficacy version of this idea was tested in mouse MLL-AF9 and largely refuted [E2]; the normal-tissue version has never been tested by anyone, and this graph's endocrine, latency and metabolic entries are all written as though menin had a single KMT2 partner. **Prior probability 0.5** that the KMT2B share of normal-tissue menin-KMT2 occupancy exceeds 25%; **0.15** that any measurable normal-tissue toxicity of a clinical menin inhibitor is attributable predominantly to the KMT2B arm. The first number is high because the biochemistry is already suggestive; the second is low because nothing establishes that KMT2B loss in an adult tissue produces a phenotype on the timescale of drug exposure.
Why this is a real hypothesis and not a restatement of the mechanism
mechanism-menin-binds-two-kmt2-proteins-not-one records what is established. This entry makes a claim that goes beyond it, in a direction nobody has tested, and is falsifiable.
The move is to separate two questions the literature conflates:
- Does the KMT2B arm contribute to killing leukaemia cells? Asked and answered, mostly no, in mouse MLL-AF9 2. This entry accepts that.
- Does the KMT2B arm account for a material share of what the drug does to a child's normal tissues? Never asked. The only study that compared Men1 deletion to Mll1;Mll2 co-deletion did it in leukaemia cells 2; the study that mapped the Menin-KMT2A/B unit did it in K-562 and stem cells 3; the quantitative complex measurement was in HeLa 1. There is no normal haematopoietic, islet, hepatic or neural arm anywhere.
That second question is where this graph's exposure lies. risk-menin-inhibitor-exposure-duration-vs-men1-latency, the endocrine-* entries and mechanism-acute-men1-deletion-reverses-established-hyperglycaemia all reason about what happens when you interfere with menin in a normal cell - and they reason about menin-KMT2A.
Predictions
- Biochemical. A KMT2B MBM1 peptide (spanning approximately aa 12-30, containing R17-G18-R19-F20-P21-G22-R23-P24) will bind menin with a Kd within one order of magnitude of the KMT2A MBM1's approximately 10 nM, and revumenib, ziftomenib, bleximenib and enzomenib will each inhibit menin-KMT2B with an IC50 within 10-fold of their menin-KMT2A IC50. If the KMT2B Kd is more than 100-fold weaker, the hypothesis is dead.
- Cellular. In a fusion-negative human cell line, menin-inhibitor treatment will displace KMT2B from chromatin with kinetics comparable to KMT2A displacement, measurable only with an endogenously tagged KMT2B allele because antibody reagents are inadequate 3.
- Genetic. In a fusion-negative context, KMT2B knockout will phenocopy a larger share of menin-inhibitor-induced transcriptional change than KMT2A knockout does - the reverse of what is observed in MLL-AF9 leukaemia cells.
- Clinical, weak. Adverse events in menin-inhibitor trials that do not resemble differentiation syndrome or QTc prolongation - particularly neurological events in young children - would be the place a KMT2B signal could appear. This is the weakest prediction and should not be over-read: germline KMT2B haploinsufficiency is a developmental phenotype and a drug is neither germline nor developmental.
What would kill it
Any one of these:
- A measured menin-KMT2B Kd more than 100-fold weaker than menin-KMT2A's, or clinical menin inhibitors with more than 100-fold selectivity for the KMT2A interface.
- Endogenously tagged KMT2B showing no chromatin displacement at drug concentrations that fully displace KMT2A.
- Demonstration that KMT2B is not expressed, or is expressed at negligible levels relative to KMT2A, in the normal tissues where menin-inhibitor toxicity is observed.
- A conditional Kmt2b knockout in adult mouse tissues producing none of the phenotypes menin inhibition produces there.
Note what would not kill it: another demonstration that the KMT2B arm does not contribute to anti-leukaemic efficacy. That question is already settled in the mouse and this hypothesis does not depend on it.
| statement | In non-malignant tissue, where no KMT2A fusion exists, the effects of menin-MBM-pocket occupancy are distributed between menin-KMT2A and menin-KMT2B, and the KMT2B share is non-trivial. |
|---|---|
| origin | literature-gap |
| prior probability | 0.5 |
| prior definition | P(KMT2B share of normal-tissue menin-KMT2 occupancy > 25%); the separate proposition that normal-tissue toxicity is PREDOMINANTLY KMT2B-driven is held at 0.15 |
| does not falsify | further evidence that the KMT2B arm contributes little to anti-leukaemic EFFICACY - that question is already settled in mouse MLL-AF9 and this hypothesis is restricted to normal tissue |
predictions
[ "a KMT2B MBM1 peptide binds menin within one order of magnitude of the KMT2A MBM1 Kd of ~10 nM", "clinical menin inhibitors show <10-fold selectivity between the menin-KMT2A and menin-KMT2B interfaces", "endogenously tagged KMT2B is displaced from chromatin by menin inhibitor with kinetics comparable to KMT2A", "in a fusion-negative cell, KMT2B knockout phenocopies more of the drug's transcriptional effect than KMT2A knockout does" ]
experiments
[ "exp-menin-kmt2b-mbm-affinity-and-inhibitor-ic50", "exp-kmt2b-endogenous-tag-occupancy-under-menin-inhibitor" ]
analyses
[ "ana-kmt2-family-enumeration" ]
falsifiers
[ "a measured menin-KMT2B Kd more than 100-fold weaker than menin-KMT2A's", "clinical menin inhibitors with >100-fold selectivity for the KMT2A interface", "endogenously tagged KMT2B showing no displacement at drug concentrations that fully displace KMT2A", "KMT2B not expressed, or negligible relative to KMT2A, in the normal tissues where menin-inhibitor toxicity appears", "adult conditional Kmt2b knockout producing none of the tissue phenotypes menin inhibition produces" ]
- If true, it is an on-target liability that no amount of chemical optimisation against the menin pocket removes, because the pocket is the same one.
- If true and large, a genuinely KMT2A-selective agent would be safer - but if the KMT2B arm also contributes to efficacy in B-lymphoid disease, which is untested, the same agent would be less effective. The hypothesis does not tell you which way to optimise.
- It is currently unmeasurable on the KMT2B side in cells, so a null clinical safety result will always be confounded with not having looked, and this hypothesis could survive indefinitely on that confound. That is a reason to run the biochemistry, which is not blocked by reagents.
- What are the relative KMT2A and KMT2B protein levels in the human tissues where menin inhibitors cause effects? This is a proteomics-atlas question that could be answered from existing public data in an afternoon and would set the hypothesis's central number on something better than HeLa.
- Do the four clinical menin inhibitors differ from each other on this axis? They are chemically distinct MBM-pocket binders and nothing requires them to share a KMT2A-versus-KMT2B selectivity ratio. If one is markedly more KMT2A-selective, that is a differentiating property nobody has measured.
- Is there an adult-tissue Kmt2b conditional knockout phenotype at all? Without one, prediction 3 has no comparator and the hypothesis cannot be evaluated even if the biochemistry comes out as predicted.
- E1strong 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 recovered menin with MLL1 and MLL2/KMT2B and with neither SET1A/B nor MLL3/4, the authors stating menin binds the two with equal affinity; menin was 4- to 5-fold less abundant than MLL1/2 overall, so the menin-bound population is a minority of both enzymes.
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 2013What this evidence cannot show
Full text read via the PMC HTML route (PMC3647974) on 2026-09-20. HeLa, an epithelial cancer line, with GFP-tagged overexpressed baits - so the 'equal' is a statement about equal recovery of two overexpressed constructs in one cell type and is the entire quantitative basis of this hypothesis's prior. Normal tissue is not represented.
Evidence that argues against this (1)
- Relative complex occupancy in HeLa need not hold in an islet beta cell or a haematopoietic progenitor, where KMT2A and KMT2B expression ratios could differ by an order of magnitude. This hypothesis's central number is extrapolated from the least relevant available cell type. Sparbier CE et al. 2023
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin-MLL interaction compounds inhibit menin-MLL1 and menin-MLL2 as well as the menin-fusion interaction in biochemical and structural assays, but in MLL-AF9 leukaemia cells Men1 deletion produces distinct consequences from Mll1;Mll2 co-deletion, the compounds phenocopy menin loss, Mll1;Mll2-deficient cells are more sensitive to the inhibitors, and the predominant antileukaemic action is concluded to be direct fusion inhibition.
Distinct pathways affected by menin versus MLL1/MLL2 in MLL-rearranged acute myeloid leukemia. · Chen Y, Jones KL, Anastassiadis K, et al. · Exp Hematol 2019What this evidence cannot show
Abstract fetched and read 2026-09-20; full text (PMC6472707) not obtained. **This is the paper that refutes the efficacy version of this hypothesis and it is cited here for that reason** - the hypothesis is deliberately restricted to the normal-tissue question the paper does not address. All arms are leukaemia cells; mouse; myeloid.
Evidence that argues against this (1)
- If the KMT2B arm contributes nothing detectable to the drug's action even in cells where the fusion competes for menin, a reasonable prior is that it contributes little anywhere - the hypothesis relies on normal tissue being different in kind, which is asserted rather than shown. van Nuland R et al. 2013
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Commercially available reagents to monitor KMT2B binding at chromatin are inadequate, and the authors state their KMT2A-only analyses therefore underestimate the consequences of liberating KMT2A/B from active genes after Menin loss or VTP50469 treatment.
Targeting Menin disrupts the KMT2A/B and polycomb balance to paradoxically activate bivalent genes. · Sparbier CE, Gillespie A, Gomez J, et al. · Nat Cell Biol 2023What this evidence cannot show
Full text read via Europe PMC fullTextXML (PMC7614190) on 2026-09-20. This is one group's assessment of the 2022-2023 reagent landscape. It is the reason prediction 2 specifies an endogenous tag rather than an antibody, and it is also the reason this hypothesis has not been tested by anyone.
Evidence that argues against this (1)
- A reagent limitation explains why the measurement is missing but is not itself evidence that the KMT2B arm is substantial. The same study's genetic data - KMT2B knockout alone only modestly impairs the phenotype while double knockout impairs it strongly - suggest KMT2B's contribution at these loci is real but smaller than KMT2A's. Chen Y et al. 2019
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Germline heterozygous KMT2B loss of function causes complex progressive childhood-onset dystonia, giving the second menin partner a named paediatric phenotype distinct from any reported menin-inhibitor adverse event.
Mutations in the histone methyltransferase gene KMT2B cause complex early-onset dystonia. · Meyer E, Carss KJ, Rankin J, et al. · Nat Genet 2017What this evidence cannot show
Abstract fetched and read 2026-09-20. **Constitutional haploinsufficiency from conception; not a model for weeks of partial reversible interface occupancy**, exactly as this project's brief says of gnomAD constraint. Prediction 4 in this entry is deliberately labelled the weakest and should not be cited as a safety expectation.
Evidence that argues against this (1)
- Menin inhibitors have been given to children without a reported dystonia signal. Small heavily pretreated cohorts with short follow-up are weak evidence of absence, but the direction of the existing clinical experience is against prediction 4. Mahdavi L et al. 2026
structural possibility only; affinity unmeasured
the hypothesis extends the established two-partner mechanism into untested normal tissue
predicts <10-fold selectivity between the menin-KMT2A and menin-KMT2B interfaces
Kd and IC50 ratios; >100-fold weaker KMT2B binding falsifies the hypothesis
the cellular arm; fusion-bearing versus fusion-negative comparison is the discriminating design