Three routes to one observation: the CUT&RUN panel that says whether menin eviction, H3K36me3 loss or MLL loss takes LEDGF off the fusion's genes
also called LEDGF occupancy monotherapy panel
We know LEDGF leaves the leukaemia's genes when the menin drug is given, but not why. Three different explanations predict different results in one experiment that uses drugs and cell lines that already exist.
The same thing, for a scientist
A 2025 preprint reports that LEDGF chromatin binding falls under revumenib, but the design cannot say why, and three routes are simultaneously live: loss of the tenfold cooperative enhancement when menin leaves the KMT2A N-terminus; loss of H3K36me2/3, the PWWP arm's ligand; and loss of MLL occupancy, on which LEDGF binding is independently known to depend. The discriminating experiment is a CUT&RUN panel across three menin inhibitors and two lineages with the right control arms. Its strongest single arm is the **MEN1 gatekeeper-mutant** arm: those mutations attenuate inhibitor binding without disturbing the menin-KMT2A interaction, so the cooperative-clamp model predicts LEDGF occupancy is preserved under drug in those cells while the H3K36me3 and MLL-occupancy routes predict it still falls.
Arms and predictions
| Arm | Manipulation | Clamp model predicts | H3K36me3 model predicts | MLL-occupancy model predicts |
|---|---|---|---|---|
| 1 | revumenib, 6 h and 72 h | LEDGF down early, tracking menin | LEDGF down only after H3K36me3 falls (late) | LEDGF down tracking MLL |
| 2 | ziftomenib, bleximenib | same as arm 1 | same | same |
| 3 | MEN1 gatekeeper mutant line + revumenib | LEDGF preserved | LEDGF down | LEDGF down |
| 4 | KMT2A IBM2 point-mutant fusion, no drug | LEDGF down at fusion targets | unchanged | unchanged |
| 5 | SETD2-null isogenic pair | unchanged | LEDGF occupancy up (H3K36me2 rises) | unchanged |
| 6 | shMLL (wild-type allele) | unchanged | unchanged | LEDGF down |
Arm 3 is the decisive one and it is also the clinically interesting one, because it asks whether the LEDGF grip survives in the exact population a LEDGF-directed agent would be sold into. Arm 5 exploits a published result: SETD2 loss-of-function, which is prevalent in MLL leukaemias, increases occupancy of H3K36me2, LEDGF and MLL at MLL target genes 3. That gives the H3K36 route a positive prediction rather than only a null one.
Lineage, and why both
The existing measurement is in MOLM-13 and MV4-11, both myeloid. This project's cell-state work found five independent axes on which KMT2A-rearranged disease separates by lineage rather than by fusion, and menin inhibition kills in the lymphoid disease while driving differentiation in the myeloid one. Running the panel in SEM or RS4;11 alongside MV4-11 holds the fusion constant and varies the lineage, which is the only configuration in which a lineage effect on this axis could be seen at all.
Antibody and normalisation are not incidental: the preprint's LEDGF CUT&RUN used Bethyl A300-848A with pAG-MNase and E. coli spike-in 1, and a repeat should use the same reagent so that a disagreement is biological.
Timing is the cheapest discriminator
The three routes differ most in their kinetics. Loss of a protein-protein cooperative term should be immediate on drug-induced menin eviction; loss of a histone mark requires turnover or dilution through replication; loss of MLL occupancy is intermediate. A 1 h / 6 h / 24 h / 72 h series at a single dose separates them at a fraction of the cost of the full matrix, and should be run first.
| design | CUT&RUN for LEDGF, menin, KMT2A-N, H3K36me2, H3K36me3 and Ser2P Pol II across three menin inhibitors, two lineages, a MEN1 gatekeeper-mutant arm, an IBM2 fusion-mutant arm, a SETD2-null isogenic pair and a wild-type-MLL knockdown arm, with a 1/6/24/72 h time course at fixed dose. |
|---|---|
| readout | Spike-in-normalised LEDGF occupancy at HOXA9, MEIS1, CDK6, FLT3 and PBX3, expressed as fold change against DMSO and against the menin occupancy change in the same sample. |
| controls | Bethyl A300-848A LEDGF antibody and E. coli spike-in to match the existing dataset; IgG; menin occupancy as the internal drug-activity control. |
| cost tier | low |
| timeline | 4-6 months |
| decisive arm | MEN1 gatekeeper-mutant line under revumenib |
- A null result in arm 3 has two readings - the clamp model is wrong, or the cooperative term is swamped by PWWP avidity - and the experiment as designed cannot separate them.
- Spike-in-normalised CUT&RUN comparisons across drug treatments are sensitive to changes in global chromatin accessibility, which menin inhibitors produce, so an apparent occupancy change can be a normalisation artefact.
- If the answer is that LEDGF occupancy falls for reasons unrelated to the clamp, the entire structural argument for dual menin-plus-LEDGF blockade loses its quantitative basis.
- Is there a published isogenic SETD2-null KMT2A-rearranged pair, or would arm 5 require making one? If it exists the strongest positive prediction in the panel is nearly free.
- Should the panel include the reciprocal measurement - menin occupancy after LEDGF depletion? Cooperativity is symmetric in principle and nobody has looked in that direction.
- Would a degron-based menin depletion be a better comparator than the inhibitors, given that an inhibitor evicts menin from KMT2A while leaving menin in the cell to do everything else it does?
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.preprint
The existing measurement is a CUT&RUN profile of LEDGF, MLL1, menin, LSD1, H3K4me3, H3K36me3 and H3K27me3 in MOLM-13 and MV4-11 under SNDX-5613, ORY-1001 and the combination, using a named LEDGF antibody and E. coli spike-in normalisation.
Co-targeting menin and LSD1 dismantles oncogenic programs and restores differentiation in MLL-rearranged AML. · Tayari MM, Dos Santos HG, et al. · bioRxiv 2025What this evidence cannot show
Preprint; combination design; qualitative reporting. Cited here to define the reagents and baseline this experiment should match.
Evidence that argues against this (1)
- If the preprint's observation does not survive peer review or replication, the premise of this experiment weakens, though the three-route ambiguity it is designed to resolve would remain. Tayari MM et al. 2025
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MEN1 mutations at the drug-menin interface cause clinical resistance to menin inhibition and attenuate inhibitor binding without disturbing the menin-KMT2A interaction, which is what makes arm 3 discriminating.
MEN1 mutations mediate clinical resistance to menin inhibition. · Perner F, Stein EM, et al. · Nature 2023What this evidence cannot show
The gatekeeper mutations were characterised for drug binding and for the menin-KMT2A interaction; LEDGF was not examined in that work, so arm 3 tests an untested extension.
Evidence that argues against this (1)
- If a resistance mutation subtly alters the menin-KMT2A interface geometry rather than leaving it intact, the cooperative enhancement could be lost even though the interaction is retained, and arm 3 would give a false negative for the clamp model. Murai MJ et al. 2014
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
SETD2 loss of function, which is prevalent in MLL leukaemias, increases occupancy of H3K36me2, LEDGF and MLL at MLL target genes, and H3K36me3 is therefore not essential for MLL occupancy or leukaemogenesis; conversely KDM2A overexpression reduces H3K36me2 and reduces LEDGF and wild-type MLL occupancy without affecting fusion occupancy.
ASH1L Links Histone H3 Lysine 36 Dimethylation to MLL Leukemia. · Zhu L, Li Q, et al. · Cancer Discov 2016What this evidence cannot show
Murine transformed HSPCs; the SETD2 statement is the authors' interpretation of their KDM2A and ASH1L data plus the known mutation spectrum, not a SETD2-null isogenic experiment. That is exactly why arm 5 is worth running.
Evidence that argues against this (1)
- If SETD2 loss raises LEDGF occupancy but leaves fusion occupancy unchanged, then the H3K36 arm again turns out to be about wild-type MLL rather than the fusion, and arm 5 would not bear on the drug question. Zhu L et al. 2016
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The cooperative term the clamp model rests on is a tenfold affinity enhancement: LEDGF IBD binds MLL 1-160 at Kd 14.7 uM and the preformed menin-MLL complex at Kd 1.4 uM.
The same site on the integrase-binding domain of lens epithelium-derived growth factor is a therapeutic target for MLL leukemia and HIV. · Murai MJ, Pollock J, et al. · Blood 2014What this evidence cannot show
In vitro, isolated fragments, two different methods for the two constants (NMR titration and ITC).
Evidence that argues against this (1)
- A tenfold affinity change in the low-micromolar range may be entirely absorbed by avidity from the PWWP-nucleosome contact, in which case the clamp model predicts no measurable occupancy change and arm 3 returns a null that is hard to interpret. Koutná E et al. 2023
separates the three causal routes behind the observation
the decisive arm uses gatekeeper-mutant cells
no LEDGF occupancy measurement exists under this drug
no LEDGF occupancy measurement exists under this drug
the lymphoid arm the existing dataset lacks
arm 3 is the discriminating measurement
adds an HRP-2 arm to the proposed panel