Hypothesis: the ceiling on menin-interface disruption is cell-state plasticity, not target mutation
also called non-genetic menin resistance hypothesis · state-escape hypothesis
We think the main reason blocking menin will fail is not that the drug stops fitting — it is that the leukemia cell can shift into a different state where it no longer needs the fusion sitting on its genes. If true, the drug needs a partner that locks the cell state.
The same thing, for a scientist
Three independent published observations converge on a non-genetic, cell-state escape route in KMT2A-rearranged leukemia: heavily pretreated KMT2A::AFF1 samples are menin-inhibitor-resistant with sustained on-target transcriptional effect and no prior class exposure; BET-inhibitor resistance in MLL-AF9 emerges from leukaemia stem cells, is not efflux-mediated and cross-resists genetic BET depletion; and BET resistance operates by WNT/beta-catenin-mediated restoration of MYC. We hypothesise that a shared plastic state — plausibly WNT-high, KMT2C/D-low, and/or enriched at the bivalent/lineage-plastic fusion binding sites identified by single-cell CUT&Tag — renders the output program independent of fusion chromatin occupancy. Prediction: single-agent interface disruption is structurally insufficient, and efficacy requires pairing with an agent that prevents the state transition.
Why we think this rather than something simpler1 deeper
The straightforward model of menin-inhibitor failure is gatekeeper mutation, and that model is correct as far as it goes 1. But it cannot explain the observation that matters most for infant KMT2A::AFF1 disease: samples from the same patients, later in their treatment course and never exposed to the drug class, were markedly less sensitive — while transcriptomics showed sustained on-target efficacy in the resistant cells. RAS-pathway and TP53 comutations were present but neither was sufficient to confer resistance in vitro 2.
So the drug engages its target, the transcriptional consequence of engagement occurs, and the cells survive anyway. That is a definitionally downstream, non-genetic mechanism.
The BET literature supplies a worked precedent of exactly this shape in the same fusion class. Resistance emerged from leukaemia stem cells, was not efflux- or metabolism-mediated, and cross-resisted genetic BET knockdown — so it is not about the drug at all 3. And independently, resistant leukemias were uniformly characterised by rapid restoration of MYC transcription through WNT signalling and a focal MYC enhancer 4. Two groups, same journal issue, converging on WNT.
The mechanistic bridge we propose: a subset of KMT2A oncofusion binding sites carry bivalent chromatin with cell-to-cell heterogeneity suggestive of lineage plasticity 5. If that heterogeneous subpopulation is the reservoir, then the state escape has a measurable chromatin signature and a measurable cell of origin.
What would make this wrong
The hypothesis is genuinely falsifiable, which is why it is worth stating. It dies if the upfront-resistant KMT2A::AFF1 samples turn out to carry an unrecognised genetic lesion that is sufficient on its own; if single-cell profiling shows resistant cells are transcriptionally and epigenetically identical to sensitive ones; if WNT inhibition has no effect on menin-inhibitor sensitivity; or if fusion occupancy is in fact lost rather than sustained in the resistant cells (i.e. the 'sustained on-target' inference from transcriptional signatures is wrong when measured directly by ChIP).
Note that the last of these is the cheapest experiment and also the most likely to embarrass the hypothesis, which is a good reason to do it first.
| statement | In KMT2A-rearranged leukemia the dominant limit on menin-interface disruption is a reversible cell-state transition that renders the output program independent of fusion chromatin occupancy, rather than mutation of the drug-target interface. |
|---|---|
| origin | data-pattern |
| prior probability | 0.45 |
predictions
[ "In upfront-resistant KMT2A::AFF1 samples, direct ChIP/CUT&RUN will confirm that menin and the fusion ARE evicted from chromatin by drug, yet target-gene output is maintained.", "Resistant cells will be enriched for a WNT/beta-catenin-high transcriptional state and/or for the bivalent, lineage-plastic subpopulation seen by single-cell CUT&Tag.", "Pharmacological WNT inhibition, or agents that enhance fusion eviction (KAT6/KAT7 inhibition, DOT1L inhibition, IKAROS degradation), will restore menin-inhibitor sensitivity in resistant PDX models.", "Resistance will be reversible on drug withdrawal and re-expansion, distinguishing it from a clonal genetic lesion." ]
experiments
[ "exp-kmt2a-aff1-degron-occupancy-titration", "exp-aff1-idr-condensate-separation-of-function" ]
analyses
[ "ana-menin-resistance-cell-state-signature" ]
falsifiers
[ "A recurrent genetic lesion is identified in upfront-resistant samples that is sufficient to confer resistance when introduced alone.", "Single-cell RNA and chromatin profiling shows resistant and sensitive cells are indistinguishable in state.", "WNT inhibition has no effect on menin-inhibitor sensitivity in resistant models.", "Direct occupancy measurement shows the fusion is NOT evicted in resistant cells, meaning resistance is upstream of engagement after all and the 'sustained on-target' inference was an artefact of transcriptional signature analysis.", "Resistance is stably heritable through drug-free passage, which would argue for a genetic or stably epigenetic clonal mechanism rather than plasticity." ]
- If the hypothesis is right, improving menin-inhibitor potency or selectivity has no effect on the dominant failure mode, and a large share of current medicinal-chemistry effort in this field is aimed at the wrong variable.
- State-locking agents (WNT inhibitors, differentiation agents) act on pathways essential to normal stem cells, so the combination that fixes efficacy may create the toxicity.
- A reversible resistance state would make intermittent dosing worse than continuous, inverting a common clinical instinct about epigenetic drugs.
- Is the plastic state the same one that underlies lineage switching under blinatumomab and CAR-T in KMT2A-r? If so, one mechanism would explain resistance to three unrelated modalities, which would make it the highest-value target in the disease.
- Does KMT2C/D loss create the permissive state, as the single-patient observation hints? KMT2C/D are frequently mutated across cancers and their loss is testable by CRISPR in a sensitive t(4;11) line.
- Can the state be locked pharmacologically, and is that a better use of effort than improving interface affinity?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Somatic MEN1 mutations at the drug-menin interface mediate clinical resistance to menin inhibition; the mutants attenuate drug binding but not MLL1 binding and prevent inhibitor-induced eviction of menin and MLL1 from chromatin.
MEN1 mutations mediate clinical resistance to menin inhibition · Perner F, Stein EM, Wenge DV, et al. (Cai SF, senior author) · Nature 615(7954):913-919 2023What this evidence cannot show
Establishes the genetic route, which this hypothesis does not dispute; cited to define what the hypothesis is an alternative to.
Evidence that argues against this (1)
- Interface mutations are the opposite of the plasticity this entry proposes: they are a genetic, on-target, structurally explicable route to resistance in which the pathway is unchanged and only drug binding is lost. Citing them in support of a cell-state hypothesis inverts what they show, and the two mechanisms make different predictions about whether a second-generation inhibitor helps. Nature 2023
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Heavily pretreated KMT2A::AFF1 ALL samples were markedly less menin-inhibitor-sensitive than earlier samples from the same patients with no prior menin-inhibitor exposure, with transcriptomics showing sustained on-target efficacy in resistant cells; RAS-pathway and TP53 comutations were present but neither was sufficient to induce resistance in vitro, and KMT2C/D downregulation was implicated in one patient.
Upfront menin-inhibitor resistance in multiply pretreated leukemias · Mahdavi L, Alikarami F, Goodrow H, et al. (Bernt KM, senior author) · Experimental Hematology 153:105268 2025What this evidence cannot show
Two serial patients plus five PDX models. 'Sustained on-target efficacy' is inferred from transcriptional signatures, not from direct chromatin-occupancy measurement — the single weakest link in this hypothesis and the first thing to test.
Evidence that argues against this (1)
- Heavily pretreated samples differ from earlier samples from the same patients in more than cell state: they have been through multiple regimens, they carry emergent RAS and TP53 lesions that the entry notes, and they are drawn from patients selected by having survived to be sampled again. Sustained on-target transcriptomic efficacy shows the drug still works on chromatin; it does not establish that a reversible state, rather than selection of a genetically distinct clone, is what changed. Nature 2023
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
BET-inhibitor resistance in MLL-AF9-immortalised murine HSPCs emerged from leukaemia stem cells, was cross-resistant to JQ1 and to genetic BET knockdown, was not efflux- or metabolism-mediated, and was in part a consequence of increased WNT/beta-catenin signalling, with negative regulation of WNT restoring sensitivity in vitro and in vivo.
BET inhibitor resistance emerges from leukaemia stem cells · Fong CY, Gilan O, Lam EYN, et al. (Dawson MA, senior author) · Nature 525(7570):538-542 2015What this evidence cannot show
Murine MLL-AF9 immortalised clones under in-vitro selection; a BET inhibitor is not a menin inhibitor, so this is a precedent for the escape category rather than evidence about menin.
Evidence that argues against this (1)
- The plasticity evidence this hypothesis leans on is from BET inhibitors in murine MLL-AF9 models, and the one resistance mechanism actually recovered from HUMAN patients on a menin inhibitor is the opposite kind: somatic MEN1 mutation at the drug-protein interface, which restores menin and MLL1 chromatin occupancy rather than moving the cell to a fusion-independent state. Until an equivalent human series documents plastic escape under menin-inhibitor pressure, the hypothesis is extrapolating across both species and drug class, and the null it must beat is 'target mutation dominates'. Perner F et al. 2023
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Resistant leukemias under BET inhibition are uniformly characterised by rapid restoration of MYC transcription via activation and recruitment of WNT signalling components and a focal MYC enhancer, and PRC2 suppression promotes BET-inhibitor resistance by remodelling pathways that restore MYC rather than by altering BRD4-dependent transcripts.
Transcriptional plasticity promotes primary and acquired resistance to BET inhibition · Rathert P, Roth M, Neumann T, et al. (Vakoc CR, Zuber J, senior authors) · Nature 525(7570):543-547 2015What this evidence cannot show
MLL-AF9;NrasG12D compound model; the convergence with the independent WNT result is the strength here, not either study alone.
Evidence that argues against this (1)
- This is BET-inhibitor resistance in MLL-AF9;Nras-G12D mouse AML and in human leukaemias under BET inhibition, which is a different drug acting on a different node. WNT-mediated restoration of MYC is specific to a target whose proximate output is MYC transcription; menin inhibition is not, and transposing the escape route assumes the two share a bottleneck the beat has not shown. Oncofusion knowledge graph (internal analysis) 2026
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A subset of KMT2A oncofusion binding sites carry bivalent H3K4me3/H3K27me3 chromatin signatures and single-cell CUT&Tag reveals cell-to-cell heterogeneity at these sites suggestive of lineage plasticity, providing a candidate chromatin substrate for a plastic escape state.
Automated CUT&Tag profiling of chromatin heterogeneity in mixed-lineage leukemia · Janssens DH, Meers MP, Wu SJ, et al. (Ahmad K, Henikoff S, senior authors) · Nature Genetics 53(11):1586-1596 2021What this evidence cannot show
The link from these heterogeneous sites to drug resistance is entirely our inference; the paper does not study resistance. This is the speculative joint of the hypothesis.
Evidence that argues against this (1)
- Bivalent signatures at a subset of sites and cell-to-cell heterogeneity in single-cell CUT&Tag are what any sparse single-cell chromatin assay produces: coverage per cell is low, so absence of a mark in a cell is mostly sampling. Heterogeneity "suggestive of" plasticity is the weakest admissible evidence for the central claim of this entry. eLife 2021
predicts incomplete eviction where the plasticity hypothesis predicts complete eviction with maintained output
- ← testsExperiment: map what stays on chromatin when the menin inhibitor has done everything it can do
the same experiment discriminates between the two competing accounts of menin-inhibitor failure
a recorded threat to the lab's core hypothesis
explains the sustained-on-target-effect-with-no-benefit observation that hypothesis is built on
- ← supportsAdaptive, non-genetic resistance — the leukemia keeps the fusion and rewires how it is maintained
direct clinical-model evidence that cell-state plasticity defeats interface disruption