Only a handful of loci lose fusion occupancy under a menin inhibitor, and they are the same few whether menin is drugged or deleted: MEIS1, MEF2C, JMJD1C - not HOXA, not MYB
also called the menin-sensitive locus set · MEIS1 MEF2C JMJD1C · HOXA retains fusion occupancy
When the drug is given, the fusion protein leaves only a few genes. Those few are named in the original paper and they do not include the HOX genes, which are the ones everybody talks about. Deleting menin genetically hits the same short list.
The same thing, for a scientist
The locus set that loses fusion occupancy under menin inhibition is small, reproducible across pharmacological and genetic menin removal, and **does not include the genes the field's model would predict**. Drugged: more than 2-fold MLL1n loss at **MEIS1, MEF2C and JMJD1C**; no loss at the **HOXA cluster, MYB or GAPDH**, in both MOLM13 and RS4;11 [E1]. Deleted: MEN1 CRISPR knockout followed by CUT&RUN identifies **39 genes losing more than 2-fold MLL1n occupancy, 20 of them shared with the drug**, again with dramatic loss at MEF2C, MEIS1 and JMJD1C and none at HOXA, MYB or GAPDH [E1]. The same three loci, plus PBX3, are the tracks chosen to illustrate MLL1 loss in the clinical-resistance paper [E2]. **The authors' own conclusion is that only a subset of MLL-target genes is sensitive to inhibition of menin by either genetic deletion or pharmacologic perturbation** [E1]. **Two consequences that the field's summary sentences lose.** First, **HOXA retains fusion occupancy while HOXA transcription falls** - so HOXA silencing under a menin inhibitor is not caused by the fusion leaving the HOXA locus. Something else does it: loss of menin-stabilised spreading, loss of DOT1L, PRC1.1-deposited H2AK119ub, or arriving MLL3/4-UTX. Second, the sensitive set is **small enough to enumerate and heterogeneous in function** - a homeobox cofactor, a MADS-box transcription factor and a JmjC demethylase - which is a poor fit for a model in which menin governs a coherent programme and a good fit for a model in which menin governs a particular chromatin configuration at a few unusually menin-dependent promoters.
The set, and what is conspicuously not in it
| locus | fusion occupancy under drug | transcription under drug |
|---|---|---|
| MEIS1 | lost (>2-fold) | falls |
| MEF2C | lost (>2-fold) | falls |
| JMJD1C | lost (>2-fold) | falls |
| PBX3 | shown as a track in the resistance paper | falls |
| HOXA cluster | retained | falls |
| MYB | retained | falls |
| GAPDH | retained | unchanged |
The HOXA row is the important one. HOXA9 and the HOXA cluster are the canonical output of this fusion and the pharmacodynamic biomarker of the class, and the fusion does not leave them.
Why 39 genes is a small number
Kerry et al. define 4,427 MLL-AF4 binding sites in SEM, corresponding to roughly 3,000 gene targets 3. Complete genetic removal of menin costs the fusion detectable occupancy at 39 genes. Even allowing for the different cell line, the different fusion partner, a 2-fold threshold and a TSS-anchored window, that is on the order of 1% of the fusion's target genes.
That ratio is the quantitative form of the finding in mechanism-vtp-50469-does-not-globally-displace-the-fusion-from-chromatin, and it is why the residual-occupancy question is not about a minority subset that survives - it is about the 99% that never moved.
| n genes lost genetic | 39 |
|---|---|
| n shared drug and genetic | 20 |
| fusion target denominator sem | 4,427 binding sites / ~3,000 gene targets |
| approximate fraction sensitive | order of 1% of fusion target genes |
| shown for fusion | KMT2A::MLLT3 (MOLM13) and KMT2A::AFF1 (RS4;11) with a shared-epitope antibody; KMT2A::AFDN (ML2) for the genome-wide pattern |
loci lost
[ "MEIS1", "MEF2C", "JMJD1C" ]
loci illustrated in resistance paper
[ "MEIS1", "PBX3", "MEF2C", "JMJD1C" ]
loci retained
[ "HOXA cluster", "MYB", "GAPDH" ]
- MEIS1 is the class's standard pharmacodynamic biomarker, and it is one of the three loci where the fusion actually leaves - so the biomarker is measuring the exception, not the rule, and will look like complete target engagement while 99% of fusion occupancy is untouched.
- A combination designed to 'reinforce eviction at HOXA' is designed against an event that does not happen; HOXA silencing must be attacked through whatever does silence it.
- If the sensitive set is only 39 genes, the therapeutic window of this drug class comes from those genes being disproportionately important, not from broad displacement - which makes any lesion that makes a leukaemia independent of MEIS1, MEF2C or JMJD1C a candidate primary-resistance mechanism.
- What makes MEIS1, MEF2C and JMJD1C menin-dependent when HOXA and MYB are not? If the answer is spreading, those three should be spreading targets and HOXA should not - a testable prediction against the published spreading list in SEM.
- If HOXA transcription falls while fusion occupancy at HOXA is retained, what turns it off? Candidate answers are DOT1L loss, PRC1.1 H2AK119ub deposition and MLL3/4-UTX arrival, and they are separable by genetics.
- Does the 39-gene set differ between B-lineage and myeloid KMT2A-rearranged cells? The set was defined in a myeloid line; the index disease is B-lineage.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Under VTP-50469, MLL1n occupancy fell more than 2-fold at MEIS1, MEF2C and JMJD1C but not at HOXA genes, MYB or GAPDH in MOLM13 and RS4;11; CRISPR deletion of MEN1 followed by CUT&RUN identified 39 genes losing more than 2-fold MLL1n occupancy, 20 shared with the drug, with the same locus pattern - leading the authors to conclude that only a subset of MLL-target genes is sensitive to menin inhibition by either route.
A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia. · Krivtsov AV, Evans K, Gadrey JY, et al. · Cancer Cell 2019What this evidence cannot show
Shared-epitope MLL1n antibody in lines retaining wild-type MLL1, so per-locus attribution to the fusion is inferential except in ML2. A 2-fold threshold on TSS-anchored window signal from single samples; a locus with a 1.9-fold loss is scored as retained. Genetic arm read at day 6 after infection, pharmacological arm at day 3. MOLM13 and RS4;11 are adult-derived.
Evidence that argues against this (1)
- HOXA retention could be an artefact of the highly repetitive, multi-gene HOXA locus and of window-based quantification across a cluster, rather than a real biological retention. A locus-resolved reanalysis with unique-mapping reads would settle it and has not been done. Krivtsov AV et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The clinical-resistance study illustrates MLL1 loss under revumenib with tracks at MEIS1, PBX3, MEF2C, JMJD1C and the HOXA cluster, and reports profound MLL1 loss specifically from promoters showing the greatest menin displacement.
MEN1 mutations mediate clinical resistance to menin inhibition. · Perner F, Stein EM, Wenge DV, et al. · Nature 2023What this evidence cannot show
Representative tracks from three replicates rather than a quantified per-locus analysis; the figure legend lists the HOXA cluster among the displayed loci without stating the direction of change there in the text read. MV4;11 retains wild-type KMT2A.
Evidence that argues against this (1)
- If HOXA does lose MLL1 in MV4;11 under revumenib, the locus set would be drug- or line-dependent rather than fixed, which would weaken the claim that the sensitive set is reproducible. The two papers' HOXA panels should be compared directly by someone with the figures open. Krivtsov AV et al. 2019
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MLL-AF4 occupies 4,427 binding sites in SEM cells corresponding to roughly 3,000 gene targets, which is the denominator against which a 39-gene menin-sensitive set should be read.
MLL-AF4 Spreading Identifies Binding Sites that Are Distinct from Super-Enhancers and that Govern Sensitivity to DOT1L Inhibition in Leukemia. · Kerry J, Godfrey L, Repapi E, et al. · Cell Rep 2017What this evidence cannot show
Different cell line (SEM, infant KMT2A::AFF1) from the lines in which the 39-gene set was measured, so the ratio is approximate and crosses a line, a partner and a lineage. Peak counts depend on calling parameters.
Evidence that argues against this (1)
- Comparing a peak count in one line to a gene-level loss count in another mixes units; a within-line ratio has never been reported, and the 1% figure should be treated as an order of magnitude rather than a measurement. Krivtsov AV et al. 2019
the gene at the third locus in that set, now with an entry and a rank-23 dependency signal
the downstream end is already in that entry: JMJD1C is one of three loci losing fusion occupancy under a menin inhibitor
the menin-inhibitor loss set is small and named, and HOXA9 is not in it for the MLL1-N readout while it is for the menin readout
the named loss set is the exception inside the global retention
one of three loci at which the fusion genuinely loses occupancy under the drug
one of three loci at which the fusion genuinely loses occupancy under the drug
- related to →HOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it
the HOXA cluster retains fusion occupancy while its transcription falls
MYB retains fusion occupancy under the drug
the same short list is produced by pharmacological and genetic menin removal
MEIS1's status as the class biomarker is why the loss set matters clinically