In SEM, H3K4me3 responds to menin displacement the way a partially redundant mark should: selectively at menin-bound sites, by about a quarter, and not at all under a DOT1L inhibitor - which bounds how much of a menin inhibitor's lethality H3K4me3 can be carrying
Menin inhibitors kill this leukaemia. If they worked by wiping out a chemical mark at the genes the fusion controls, we would expect that mark to collapse. In the one dataset that measures it in an infant-derived cell line, it drops by roughly a quarter, only where menin was - a real, specific effect, but far too small to be the whole story of how the drug works.
The same thing, for a scientist
Synthesis of this beat's measurements in SEM (t(4;11) KMT2A::AFF1), GSE159608. **(1) Site-selective.** VTP-50469 lowers the within-library read-share ratio between menin-enriched and H3K4me3-matched menin-free sites to 0.766 of DMSO (95% CI 0.667-0.870); at KMT2A N-terminus-enriched sites, 0.871 (CI 0.762-0.985). **(2) Not a class effect of being treated.** EPZ-5676, a selective DOT1L inhibitor at the same dose and duration, gives 1.010 (CI 0.956-1.064) and 0.990 (CI 0.950-1.031). **(3) Partial.** A quarter, not a collapse - and the pre-registered 1.5-fold threshold was not met. **(4) The global picture is unmeasurable here**, because the deposit is compositionally normalised with no spike-in, and the two VTP libraries independently hold 2.5-fold less of their reads in peaks. So the mechanism statement is about *redistribution between site classes*, which is what these data can support, and explicitly not about absolute level.
Why partial is the expected answer, and why that is not a rescue
Most H3K4me3 in any cell is deposited by SETD1A/SETD1B and wild-type KMT2A complexes at CpG-island promoters that carry no fusion. A menin inhibitor acts on the menin-KMT2A interface and should leave those untouched 2. So a small, site-restricted effect is what the model predicts, and the 23% figure is consistent with the menin-interface hypothesis rather than a problem for it.
The part that is a problem is the inference chain people usually draw: menin inhibitor -> fusion leaves chromatin -> H3K4me3 collapses at target genes -> genes switch off -> cells differentiate and die. If step three is a 23% change, either the transcriptional consequence is far more sensitive than the mark, or H3K4me3 is not the transmitting step at all and is a correlate of a change carried by something else - elongation, H3K79me2, or the physical presence of the complex. This dataset cannot distinguish those, and the entry does not pretend to. 13
What would make this a stronger mechanism claim
One experiment: spike-in-normalised H3K4me3 and menin CUT&Tag in SEM under VTP-50469, at 6, 24 and 72 h, with an IgG arm under drug. It would convert 'redistribution between site classes' into an absolute statement, tie the H3K4me3 change to the menin loss that is supposed to cause it, and separate a lagging mark from a partial one. See the two experiment entries attached to this node. The entire current answer rests on two libraries.
Where this sits relative to the field's own claim
The depositors report from these data that aberrant enrichment of H3K4me3 in gene bodies is menin-inhibitor sensitive 3. That is a claim about the shape of the domain, not its height at promoters, and it is compatible with everything here. We could not test it: a gene body-to-promoter ratio at 5 kb bins is meaningless for genes shorter than a bin, and ours returned 1.000 for GAPDH by construction. Our measurement and theirs are different statistics on the same libraries and neither confirms nor contradicts the other.
| system | SEM, human, t(4;11) KMT2A::AFF1, infant-derived |
|---|---|
| intervention | VTP-50469 30 uM 72 h vs DMSO; EPZ-5676 30 uM 72 h as comparator |
| readout | H3K4me3 AutoCUT&Tag, within-library menin-bound / menin-free read-share ratio |
| what is not measurable here | absolute H3K4me3 level (compositional normalisation, no spike-in); gene-body domain shape (5 kb bins too coarse) |
| n libraries carrying the whole result | 2 |
effect size
{
"VTP_over_DMSO": 0.766,
"CI95": [
0.667,
0.87
],
"EPZ_over_DMSO": 1.01,
"CI95_EPZ": [
0.956,
1.064
]
}- If H3K4me3 carries only a quarter of the signal, a pharmacodynamic assay built on it will report failure in responding patients and success in non-responders - the worst possible error for a dose-finding trial.
- Resistance: MEN1 pocket mutations restore menin binding without altering H3K4me3 machinery, so this readout cannot distinguish a resistant clone from a sub-therapeutic dose.
- Compensation: SETD1A/B and wild-type KMT2A supply most promoter H3K4me3 and are unaffected. Any strategy to deepen the H3K4me3 effect necessarily engages them, and they are essential in normal haematopoiesis.
- The whole claim rests on two libraries whose fraction of reads in peaks is 2.5-fold below their own controls. If those libraries are simply weak, the 23% is an artefact of imperfect cancellation and the true effect could be zero.
- Is 23% the steady state or the 72-hour snapshot of a slow decay? H3K4me3 half-life at promoters is long relative to menin displacement.
- Does the effect concentrate at a subset of menin-bound sites, and if so are those the fusion-driven ones? 130 bins and two libraries cannot answer this.
- What is the corresponding number for H3K79me2 in the same cells? The deposit has no H3K79 readout in SEM at all, which is striking for a series that includes a DOT1L inhibitor arm.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In SEM H3K4me3 CUT&Tag, VTP-50469 lowers the menin-bound/menin-free within-library share ratio to 0.766 of DMSO while EPZ-5676 leaves it at 1.010.
sem-arm-identity-from-signal: run3.py, run4_bootstrap.py (this repository) · Q140-sem-menin-arm · Oncofusion knowledge graph 2026What this evidence cannot show
Two libraries per arm, one time point, one dose, one cell line, 5 kb bins. The interval is a genomic bootstrap and not a biological one.
Evidence that argues against this (1)
- A 23% effect is within the range that a residual, non-uniform library-quality difference could produce if the VTP libraries' extra background were not perfectly uniform across the genome. The within-library class ratio cancels a UNIFORM background change exactly; a background that is itself enriched at accessible sites would not fully cancel, and menin-bound sites are accessible. The IgG control value inside the class (0.076) bounds this but does not eliminate it. Janssens DH et al. 2021
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
VTP-50469 displaces menin from protein complexes and inhibits MLL chromatin occupancy at select genes, with selective responsiveness of MLL-rearranged lines; loss of MLL binding leads to changes in gene expression, differentiation and apoptosis.
A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia · Krivtsov AV, Evans K, Gadrey JY, Eschle BK, Hatton C, Uckelmann HJ, et al., Armstrong SA · Cancer Cell 36:660-673.e11 2019What this evidence cannot show
The word 'select' is the authors' own and is the point: occupancy loss is not genome-wide, which is the premise of the site-selective measurement here. Their experiments are ChIP-seq in MLL-r lines and PDX, not in SEM.
Evidence that argues against this (2)
- The cited chain runs occupancy loss to expression change to differentiation and apoptosis. This entry measures a histone mark and infers a position in that chain, but the deposit contains no expression, differentiation or viability readout for these libraries, so every link except the first is imported from other systems at other doses.
- If loss of MLL binding causes differentiation and apoptosis, then a 72-hour exposure samples cells partway through that process, and a 23% mark change in surviving cells may reflect which cells survived rather than what the drug did to chromatin.
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The dataset's own publication reports that aberrant enrichment of H3K4me3 in gene bodies is sensitive to menin inhibitors.
Automated CUT&Tag profiling of chromatin heterogeneity in mixed-lineage leukemia · Janssens DH, Meers MP, Wu SJ, Babaeva E, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Genetics 53:1586-1596 2021What this evidence cannot show
Read from the abstract. We did not retrieve the figure-level analysis, so we cannot say which libraries or which statistic supports it, and we therefore neither claim agreement nor disagreement with it.
Evidence that argues against this (1)
- The publication's sensitive quantity is gene-body H3K4me3 enrichment; this entry's is promoter-level H3K4me3 partitioned by menin occupancy. Agreement in direction between two different quantities in the same deposit is reassuring but is not the independent confirmation the citation is being used to supply.
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
EPZ-5676 is a selective DOT1L inhibitor (Ki 80 pM, 37,000-fold selectivity) that reduces H3K79 methylation and MLL-fusion target gene expression and selectively kills MLL-translocated lines.
Potent inhibition of DOT1L as treatment of MLL-fusion leukemia · Daigle SR, Olhava EJ, Therkelsen CA, Basavapathruni A, Jin L, Boriack-Sjodin PA, et al., Pollock RM · Blood 122:1017-1025 2013What this evidence cannot show
Target engagement at 30 uM for 72 h in SEM is not verified within this deposit; no H3K79 readout exists in the SEM drug arms. The null result for EPZ-5676 here is therefore a weaker negative control than it appears.
Evidence that argues against this (2)
- The DOT1L arm is used here as the specificity control that makes the VTP result interpretable, and its enzymatic selectivity figures do not establish target engagement at 30 uM for 72 h in SEM. Without an H3K79me2 readout in these arms - and the deposit has none - the null is uninterpretable as a control.
- Pinometostat's clinical record is the relevant counterweight to its in vitro selectivity: single-agent activity in KMT2A-rearranged leukaemia was limited, so 'selective and potent' has not translated into the biological dominance that using it as a contrast implies.
The 0.766 and 1.010 values are the mechanism statement's primary evidence
Co-distribution plus a drug arm in one cell line is what makes the mechanism statement more than correlation
The mechanism concerns displacement of menin from the KMT2A interface
States what H3K4me3 can and cannot report about menin eviction
Bounds how much of the drug's action H3K4me3 can be transmitting
The same interface is the target of the approved menin inhibitor, so the biomarker caution transfers
No H3K79 readout exists in the SEM drug arms, which is why the DOT1L comparator's engagement is unverified
- related to →Plain-English: chromatin and epigenetics — the sticky notes on the DNA, and which ones matter here
Layman entry point for what a histone mark is and why a partial change matters
The hypothesis is about the residual the mechanism entry measures