Prime, wash, divide, rechallenge - the one experiment that would turn 'DOT1L provides transcriptional memory' into a measurement, and it doubles as a dosing-schedule study
also called the DOT1L memory washout experiment · does silencing persist after DOT1L inhibitor withdrawal
A 2026 paper says DOT1L gives leukaemia cells a memory of which genes should be switched on, and testing memory means treating cells, taking the drug away, letting them divide, and treating again. That paper did the take-the-drug-away part - for a different drug, with a timecourse - and never did the treat-again part, which is the step that makes it a memory experiment.
The same thing, for a scientist
`concept-transcriptional-memory-two-meanings-one-disease` establishes that the inducible-memory sense carries a specific protocol: prime, wash, allow cell division, re-stimulate, measure **kinetics** [E1]. Neville et al.'s conclusion - DOT1L as 'a component of transcriptional memory co-opted in leukaemia' [E2] - is stated in that vocabulary, and no washout-rechallenge experiment appears in this graph or in this beat's reading of the surrounding literature. Design: treat a KMT2A-rearranged line with a DOT1L inhibitor to steady state, **withdraw the drug**, allow a defined number of divisions, then **re-treat**, measuring at each stage (i) H3K79me2 recovery, (ii) H2AK119ub at the loci that gained it, (iii) transcription of the affected targets, and (iv) the **rate** of re-silencing on the second exposure against the first. The pharmacology makes this unusually tractable: H3K79 methylation has no established eraser and is lost by dilution through division, so the washout arm is a clean, division-counted recovery rather than an enzymatic decay [E3]. **Two readouts, one experiment.** The mechanistic readout is whether memory in the inducible sense exists. The clinical readout is whether **intermittent dosing is better or worse than continuous dosing at matched exposure** - and the two possible answers point in opposite directions, which is why this is worth doing rather than assuming.
Two opposite predictions, both consistent with the published framing
Prediction A - memory favours the tumour. If the fusion-target programme retains a memory of its active state, then on drug withdrawal the targets re-activate faster than they were silenced, and a drug holiday is worse than continuous exposure at the same total dose. This is the reading the phrase 'transcriptional memory co-opted in leukaemia' most naturally suggests, and it is an argument against dose interruption.
Prediction B - memory favours the drug. If the silencing is actively deposited by PRC1.1 rather than passively lapsing - which is what mechanism-menin-inhibitor-silencing-is-prc1-1-deposition-not-only-eviction argues for the menin arm - then a deposited H2AK119ub mark is itself heritable through division and the silencing could outlast the drug. That is an argument for intermittent dosing, and it is an equally faithful reading of the same biology.
The two predictions are distinguishable by a single quantity: whether, at matched division number after washout, the affected loci are more or less repressed than untreated controls. This is not a subtle experiment and it has not been done.
It also matters clinically right now. Menin inhibitors are dosed with interruptions in practice - differentiation syndrome management, cytopenias, QTc holds - so the schedule question is not hypothetical, and the two drugs' schedules are set by toxicity rather than by any chromatin argument.
Why H3K79 methylation makes this the cleanest memory experiment available in this disease
Most chromatin-memory experiments are confounded by the mark being actively removed, so a washout measures an enzymatic decay of unknown rate. H3K79 methylation has no established demethylase in this graph's assessment - mechanism-h3k79-demethylation-contested records the one contested claim - and is therefore lost principally by replication-coupled dilution 3. That has three consequences that favour this design:
- The washout is a clock. Mark loss is a function of division number, so the memory phase can be defined in divisions rather than in hours, which is exactly what the inducible-memory literature requires 1.
- The 'progressive' clause becomes testable. Neville et al. state that enhanced PRC1.1 activity arises from the progressive loss of H3K79 methylation 2. Progressive loss on the way down predicts progressive gain on the way back up, with PRC1.1 activity tracking it in reverse. That is a strong, shape-level prediction that a washout time course would either show or not.
- The recovery arm is informative by itself. Whether H2AK119ub leaves as H3K79me returns - and with what lag - measures the antagonism's reversibility, which nothing in the published account addresses.
exp-h3k79-state-resolved-icechip-dose-series and exp-h3-3-turnover-at-fusion-targets-under-dot1l-inhibition already exist in this graph and both measure the mark under drug. Neither has a withdrawal arm. This experiment is the missing half and should share their reagents.
| design | Prime-washout-rechallenge in a KMT2A-rearranged line. Phase 1: DOT1L inhibitor to steady-state H3K79me2 loss, with a dense time course. Phase 2: drug withdrawal, with divisions counted (CFSE or equivalent) rather than hours, sampling at defined division numbers. Phase 3: re-treatment at the same dose, measuring the RATE of re-silencing against the phase-1 rate. Parallel arm: continuous dosing at matched total exposure, as the schedule comparator. |
|---|---|
| readout | H3K79me2 by spike-in-normalised CUT&RUN; H2AK119ub at the loci that gained it in phase 1; nascent transcription of the affected targets; and the ratio of second-exposure to first-exposure silencing rate, which is the memory quantity |
| controls | untreated cells carried through the same number of divisions, because the memory comparison is against division-matched naive loci and not against time-matched controls; naive loci within the same cells as an internal control; spike-in normalisation throughout, since the DOT1L arm of this biology is reported as a genome-wide change and ratio normalisation would erase it |
| primary discriminating quantity | at matched division number after washout, are the affected loci MORE repressed than untreated controls (silencing outlasts the drug - favours intermittent dosing) or LESS (memory favours the tumour - argues against dose interruption) |
| why h3k79 makes this tractable | no established demethylase, so mark loss is principally replication-coupled dilution and the washout is a division-counted clock rather than an enzymatic decay of unknown rate |
| testable shape level prediction | if PRC1.1 activity rises with PROGRESSIVE H3K79me loss, it should fall with progressive H3K79me recovery, with a measurable lag; the shape of the recovery curve is as informative as its endpoint |
| cost tier | med |
| timeline | 4-6 months |
| status | planned |
| relationship to existing entries | exp-h3k79-state-resolved-icechip-dose-series and exp-h3-3-turnover-at-fusion-targets-under-dot1l-inhibition both measure the mark UNDER drug and neither has a withdrawal arm; this is the missing half and should share their reagents |
| clinical relevance | menin and DOT1L inhibitors are dosed with interruptions in practice - differentiation syndrome, cytopenias, QTc holds - so the schedule question is live and is currently settled by toxicity alone |
tests
[ "concept-transcriptional-memory-two-meanings-one-disease" ]
what already exists measured cycle 57
{
"series": "GSE260456 (RNA-seq subseries of PMID 41634341, the paper this experiment tests)",
"washout_timecourse_that_exists": [
"Control - VTP50469 300nM 8h and 8+24w",
"Control - VTP50469 300nM 48h, 48+24w, 48+48w, 48+72w",
"MOLM13 Control - VTP50469 96h 500nM, 96+48w, 96+96w",
"MOLM13 PCGF1/BCOR KO VTP50469 96h, 96+48w, 96+96w",
"washout DMSO controls in Control, PCGF1 KO, MOLM13 Control and MOLM13 PCGF1/BCOR KO, n=3"
],
"why_it_does_not_close_this_experiment": "It is a MENIN-inhibitor washout, not a DOT1L-inhibitor washout, and it has NO RECHALLENGE ARM: 0 of 180 sample titles contain both a drug name and 'washout', so the '+Nw' points are recovery timepoints and not second exposures.",
"what_it_does_change": "The recovery-kinetics half of the design exists for one drug, with a genetic axis (PCGF1 and PCGF1/BCOR knockout) this entry never proposed. So the experiment is cheaper than costed and should be re-scoped to the re-treatment arm and to the DOT1L inhibitor, with the existing menin-inhibitor recovery curves as the comparator rather than as a gap.",
"a_free_analysis_this_makes_possible": "The same matrices allow asking whether the serine/one-carbon/ASNS programme from hyp-menin-inhibition-collapses-the-serine-asns-programme-and-should-resensitise-kmt2a-r-all-to-asparaginase RECOVERS after menin-inhibitor washout - which would distinguish a durable reprogramming from a reversible pharmacological effect. Not run here."
}- If silencing outlasts the drug, that is an argument for intermittent dosing - and intermittent dosing is also the condition under which resistant subclones are least suppressed, so a schedule conclusion from this experiment must not be taken in isolation from resistance dynamics.
- If memory favours the tumour, the conclusion is that dose interruptions are harmful, which conflicts with the toxicity management that currently requires them. The experiment could produce an answer that cannot be acted on.
- The experiment measures a cell line's chromatin, and drug holidays in patients also involve marrow recovery, immune reconstitution and clonal competition - none of which a washout in culture models.
- A negative result - no memory in the inducible sense - would not refute the underlying mechanism, only the framing, and this graph has already repeated the framing in 15 entries. The finding would require going back and rewording them rather than retracting anything.
- Does the published paper already contain a washout arm? The abstract does not say and the full text was not reachable by this beat; that should be checked before the experiment is funded, since it changes this from a first test to a replication.
- How many divisions are needed for H3K79me2 to return to baseline after inhibitor withdrawal? This has not been reported and it sets the entire sampling schedule.
- Does H2AK119ub leave as H3K79me returns, and with what lag? Reversibility of the antagonism is untouched by the published account.
- Is the schedule answer the same for a menin inhibitor, where the H2AK119ub gain is subset-restricted rather than genome-wide? The two drugs could have opposite optimal schedules for the same mechanistic reason.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The operational template for an inducible transcriptional-memory claim: a subset of targets retains a mitotically heritable memory of prior exposure and is hyperactivated on re-exposure, with permissive chromatin marks established during priming and selectively maintained through cell division in the absence of transcription.
Heritable maintenance of chromatin modifications confers transcriptional memory of interferon-gamma signaling. · Mikulski P, Tehrani SSH, Kogan A, Abdul-Zani I, Shell E, James L, et al. · Nat Struct Mol Biol 2025What this evidence cannot show
A non-malignant human system and an unrelated signalling pathway; the marks carrying memory there are acetylation and H3K4me1 rather than H3K79me. It supplies the experimental form - prime, divide, rechallenge, measure kinetics - and not a prediction about this disease. The analogy between a signal being withdrawn and a drug being withdrawn is not exact: a drug washout restores an enzyme, a signal withdrawal removes a stimulus.
Evidence that argues against this (1)
- In that system the repressive mark behaves counterintuitively - H3K27me3 transiently accumulates during active transcription and is depleted only during the memory phase - so a naive expectation that repressive marks track repression would have been wrong there. The same could hold for H2AK119ub here, and the experiment must therefore measure the mark rather than infer it from expression. Neville D et al. 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Enhanced PRC1.1 activity arises specifically from the progressive loss of DOT1L-mediated H3K79 methylation, independent of MLL-fusion displacement or transcriptional repression, driven by direct biochemical antagonism between H3K79 methylation and PRC1 activity; the authors establish DOT1L as a component of transcriptional memory co-opted in leukaemia.
DOT1L provides transcriptional memory through PRC1.1 antagonism. · Neville D, Ferguson DT, Heikamp EB, Lai Z, Magor GW, Lam C, et al. · Nat Cell Biol 2026What this evidence cannot show
Abstract-level; whether a washout-rechallenge experiment was performed is not determinable from the abstract, so this entry does not assert that the memory claim is untested - only that no such experiment is visible to this beat and that the graph has never recorded one. If the paper contains one, this experiment becomes a replication with a dosing-schedule extension, which is still worth running.
Evidence that argues against this (1)
- An independent 2026 study in GCB-DLBCL reports DOT1L inhibition causing LOSS of H2AK119ub1 at PRC1/2-target promoters, the opposite direction. If the sign is context-dependent, the washout experiment could show recovery kinetics that do not match either published account, and the design must be able to report an unexpected direction rather than only a magnitude. Gobel C et al. 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
PRC1 activity and H2AK119 monoubiquitination are required to prevent differentiation of leukaemic cells and to sustain the leukaemogenic potential of Hox-activating and non-Hox-activating fusions, establishing that the repressive arm of this axis is functionally consequential rather than a passive mark.
Maintenance of leukemic cell identity by the activity of the Polycomb complex PRC1 in mice. · Rossi A, Ferrari KJ, Piunti A, Jammula S, Chiacchiera F, Mazzarella L, et al. · Sci Adv 2016What this evidence cannot show
Cited to justify measuring H2AK119ub as a functional quantity in the washout rather than as an inert readout. It is mouse genetics of both catalytic subunits and says nothing about kinetics, reversibility or drug withdrawal. The claim that H3K79 methylation lacks an established eraser and is lost by replication-coupled dilution is this graph's own assessment, carried in mechanism-h3k79-demethylation-contested, and is not supported by this citation.
Evidence that argues against this (1)
- The same PRC1 activity that is required for leukaemia maintenance here is the activity whose LOSS confers menin-inhibitor resistance, so 'PRC1 is functionally consequential' does not fix the direction of its consequence. A washout experiment must therefore report the transcriptional outcome, not only the mark. Zhou X et al. 2024
- E_k57_washout_existsstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
A WASHOUT EXPERIMENT DOES EXIST, in the source paper's own data, and this entry's statement that none appears is wrong. GSE260456, the RNA-seq subseries of PMID 41634341, contains a washout timecourse across 180 samples: 'Control - VTP50469 300nM 8h' and '8+24w'; 'Control - VTP50469 300nM 48h', '48+24w', '48+48w' and '48+72w'; 'MOLM13 Control - VTP50469 96h 500nM', '96+48w' and '96+96w'; the same 96h/96+48w/96+96w series in 'MOLM13 PCGF1/BCOR KO'; plus dedicated washout controls 'Control - Washout DMSO', 'PCGF1 KO Washout DMSO', 'MOLM13 Control - Washout DMSO' and 'MOLM13 PCGF1/BCOR KO Washout DMSO' at n=3 each. TWO THINGS NARROW IT, and they are what preserve this experiment's value. FIRST, it is a MENIN-inhibitor washout: every '+Nw' timepoint is VTP50469, and every SGC0946 (DOT1L inhibitor) sample in the series is a single fixed exposure with no washout timepoint. SECOND, there is NO RECHALLENGE ARM anywhere in the 180 titles - no title contains both a drug name and 'washout', and the '+Nw' samples are recovery timepoints rather than second exposures. The memory-defining re-stimulation was not performed.
What this evidence cannot show
Established from SAMPLE TITLES, not from the paper's figures or methods: the titles are the submitter's own labels and 'w' is inferred to mean washout from the '48+24w' construction and the presence of matching 'Washout DMSO' controls. The full text was not read, so it is possible that a re-treatment was performed and not deposited, or deposited under a title this census could not recognise - which would make the surviving half of this entry's claim wrong too. The absence of a rechallenge arm is therefore a bounded absence over deposited sample titles, not over the paper. Exposure durations and the cell system behind the unprefixed 'Control -' labels were not resolved here.
Evidence that argues against this (2)
- This is the third time in three cycles that this project has proposed an experiment whose early or partial arm was already sitting in a dataset it cites. The cycle-55 flat-HOXA result, the H3K27me3 timecourse entry, and now this. The common cause is not the individual entries: it is that a proposal is written from the paper's TEXT and never checked against the paper's DEPOSITION, and GEO sample titles are the cheapest check available.
- The narrowing could be read as rescuing the entry and should not be. The existing washout timecourse is for the MENIN inhibitor, and this graph's own cycle-56 work showed menin and DOT1L inhibition move overlapping gene sets in the same experiment - so a menin-inhibitor recovery curve is partially informative about the DOT1L question the entry poses, which makes the remaining novelty smaller than 'no such experiment exists' implies.
the drift-robust observation belongs to that experiment premise: MEIS1 does not rebound within 72-96h of withdrawal, while Actb drifts UP 1.38 log2 in the same samples
the prime-wash-divide-rechallenge protocol the inducible sense of the term requires
- related to →Pinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
doubles as a dosing-schedule study: is intermittent dosing better or worse than continuous at matched exposure
no established eraser means the washout is a division-counted clock rather than an enzymatic decay
measures whether the repressive mark leaves as H3K79me returns, and with what lag