'Transcriptional memory' means two different things in this literature, and both of them are used about MLL leukaemia
also called transcriptional memory · epigenetic memory · priming and re-exposure · cell-of-origin memory
When a 2026 paper says DOT1L provides 'transcriptional memory', it means genes respond faster the second time they are switched on. When a 2017 paper says MLL leukaemia stem cells have 'transcriptional memory', it means they still carry the gene programme of the healthy cell they came from. Those are different ideas with the same name, and both are used about the same disease.
The same thing, for a scientist
The term has two established, non-overlapping technical senses. **Sense 1 - inducible memory.** A cell responds to a previously experienced signal faster, more strongly and more sensitively on re-exposure [E1]; the canonical mammalian demonstration is interferon-gamma priming of the GBP cluster, where permissive marks are maintained through division in the absence of transcription and reactivation is accelerated [E2]. It is measured by a **prime, wash, divide, re-stimulate** protocol with kinetics as the readout. **Sense 2 - cell-of-origin memory.** A transformed cell retains the transcriptional programme of the normal cell it arose from; in MLL leukaemia this determines whether the leukaemic stem cell is beta-catenin-dependent, and the memory gene is Hoxa9 [E3]. It is measured by transforming **different sorted progenitors** with the same fusion and comparing their programmes. Neville et al. use **sense 1** when they describe DOT1L as 'a component of transcriptional memory co-opted in leukaemia' [E4], and this graph carries the phrase in 15 entries, all tracing to that paper. **Nothing in this graph has ever run a sense-1 experiment**, and the distinction matters because the two senses imply opposite drug schedules: sense-1 memory predicts that intermittent dosing allows rapid re-activation, sense-2 memory predicts nothing about dosing at all.
What a sense-1 claim commits you to measuring
'DOT1L provides transcriptional memory' is an operational claim with a specific protocol behind it, and the graph should hold it to that standard rather than treating the phrase as a metaphor for epigenetic stability.
The minimum protocol, from the IFN-gamma system 2:
- Prime. Deliver the signal; confirm target induction.
- Wash and let the cells divide. Memory is defined by surviving mitosis in the absence of ongoing transcription. This is the step that separates memory from a slowly decaying response, and it is the step that is usually skipped.
- Re-stimulate. Measure induction kinetics, not endpoint level. The memory phenotype is faster and steeper, not necessarily higher at plateau.
- Show a chromatin correlate maintained through the memory phase in the absence of transcription. In the GBP system these are H3K4me1, H3K14ac and H4K16ac, with KAT7 required for the H3K14ac deposition and for accelerated reactivation 2.
Translated to this disease, the DOT1L version would be: treat a KMT2A-rearranged cell with a DOT1L inhibitor, withdraw it, let the cells divide, and ask whether the previously repressed targets re-activate faster than naive loci when H3K79 methylation is restored - with H3K79me and H2AK119ub tracked through the washout. No such experiment exists in this graph or, as far as this beat could find, in this literature. exp-h3-3-turnover-at-fusion-targets-under-dot1l-inhibition and exp-h3k79-state-resolved-icechip-dose-series both measure the mark's kinetics under drug; neither includes a washout-and-rechallenge arm, which is the part that would make a memory claim testable.
Why the collision is a real hazard and not a semantic quibble
A PubMed search for "transcriptional memory"[tiab] AND (leukemia OR leukaemia) on 2026-09-20 returns 5 records. Two of them are about MLL leukaemia and they use the term in the two different senses: Neville et al. in sense 1 4 and Siriboonpiputtana et al. in sense 2 3. There is no third sense and no paper that reconciles them.
The hazard is the same shape as the identifier traps in this project's brief, transposed from a symbol to a concept. A reader who finds mechanism-dot1l-prc1-antagonism asserting that DOT1L is 'a component of transcriptional memory' and then finds the 2017 cell-of-origin paper will reasonably connect them, and the connection is false: Hoxa9-mediated cell-of-origin memory and H3K79me-mediated resistance to Polycomb takeover are different phenomena with different measurements and different therapeutic implications. The graph should say which sense it means every time it uses the phrase, and this entry exists so that there is somewhere to point.
The therapeutic consequence is concrete. Sense 1 predicts that a drug holiday is dangerous: if the silenced targets retain a memory of their active state, they re-activate faster than they were silenced, and intermittent dosing is worse than continuous dosing at equal total exposure. Sense 2 predicts nothing about scheduling and instead predicts that response differs by cell of origin - which, for an infant disease arising in a fetal progenitor, is a different and equally testable claim.
| graph usage | the phrase appears in 15 node files as of 2026-09-20, all tracing to PMID 41634341 and therefore all in sense 1 |
|---|---|
| count method | case-insensitive search for 'transcriptional memory' over all 1,510 node JSON files, 2026-09-20 |
| pubmed bound | "transcriptional memory"[tiab] AND (leukemia[tiab] OR leukaemia[tiab]) returns 5 records, 2026-09-20; two concern MLL leukaemia and they use opposite senses; querytranslation read back and unremarkable |
| therapeutic consequence of sense 1 | if targets silenced by a menin or DOT1L inhibitor retain memory of their active state, intermittent dosing permits faster re-activation than continuous dosing at equal total exposure - a schedule question, and a testable one |
| therapeutic consequence of sense 2 | response differs by cell of origin, which for an infant disease arising in a fetal progenitor is a different claim entirely and is not about scheduling |
| what would test sense 1 here | DOT1L inhibitor on, then washout with cell division, then rechallenge, measuring re-silencing kinetics at previously affected loci against naive loci, with H3K79me and H2AK119ub tracked throughout |
senses
[
{
"sense": 1,
"name": "inducible / signal-response memory",
"definition": "faster, stronger, more sensitive response on re-exposure to a previously experienced signal",
"minimum_protocol": "prime -> wash -> allow cell division -> re-stimulate -> measure induction KINETICS, with a chromatin correlate maintained through the memory phase in the absence of transcription",
"canonical_mammalian_system": "IFN-gamma priming of the GBP cluster (PMID 40186025)",
"used_in_this_disease_by": "PMID 41634341",
"run_in_this_graph": false
},
{
"sense": 2,
"name": "cell-of-origin memory",
"definition": "retention by a transformed cell of the transcriptional programme of the normal cell it arose from",
"minimum_protocol": "transform DIFFERENT sorted progenitor populations with the same fusion; compare programmes and dependencies",
"canonical_system": "LSK-derived versus GMP-derived MLL cancer stem cells (PMID 28978671)",
"used_in_this_disease_by": "PMID 28978671",
"memory_gene": "Hoxa9; downstream mediator Prmt1"
}
]- If the sense-1 reading is right, a drug holiday - which is how differentiation-syndrome management and dose interruptions actually work in the clinic - is the exact condition under which silenced targets would return fastest, and the standard of care would be working against the mechanism.
- A memory claim that has not been tested by washout and rechallenge is currently unfalsifiable as stated, and this graph has repeated it in 15 entries. Repetition is not corroboration.
- Attributing a sense-2 result to a sense-1 mechanism, or the reverse, would produce a confident and wrong therapeutic inference - scheduling advice from a cell-of-origin finding, or cell-of-origin stratification from a kinetics finding.
- Did Neville et al. actually run a prime-washout-rechallenge experiment, or is the memory framing an interpretation of a maintenance result? The abstract does not say and the full text was not reachable by this beat; this should be resolved before the graph repeats the claim again.
- Does a DOT1L-inhibitor washout in a KMT2A-rearranged cell show accelerated re-activation of previously silenced targets? That is the sense-1 experiment and nobody appears to have run it.
- If sense-1 memory is real here, is intermittent menin- or DOT1L-inhibitor dosing worse than continuous dosing at matched exposure? That is a clinically consequential question that follows directly from the vocabulary and has never been posed.
- Do the two senses interact? Cell-of-origin memory sets which genes are active, and inducible memory governs how fast they return - an infant fetal-progenitor-derived leukaemia could differ from an adult one in both, and no study has measured either in infant disease.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review
Transcriptional memory, in the inducible sense, is the property that allows cells to respond to previously experienced signals in a faster, stronger and more sensitive manner, and its formation involves interplay between transcription factors and repressive chromatin.
Transcriptional memory formation: Battles between transcription factors and repressive chromatin. · Zhao Z, Zhu B · Cell Syst 2024What this evidence cannot show
A short commentary on another group's synthetic-biology work, not a primary report - cited only as a definitional map, which is the use this graph's quality bar permits for a review. It does not itself measure anything and does not address leukaemia.
Evidence that argues against this (1)
- The definition is not universal in the field: the same phrase is used in MLL leukaemia for cell-of-origin programme retention, which has no signal, no re-exposure and no kinetic readout. A definition drawn from one community does not bind another's usage, which is the point of this entry. Siriboonpiputtana T et al. 2017
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A subset of interferon-gamma targets retains a mitotically heritable memory of prior exposure and is hyperactivated on re-exposure: the permissive marks H3K4me1, H3K14ac and H4K16ac are established during priming and selectively maintained on the guanylate-binding protein gene cluster in dividing human cells in the absence of transcription, KAT7 is required for H3K14ac there and for accelerated reactivation, and H3K27 methylation is selectively depleted from primed genes during the memory phase.
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 cell system and a signalling pathway unrelated to KMT2A. Cited as the operational template - it is the clearest published statement of what a sense-1 memory experiment has to contain - and not as evidence about leukaemia. Note that in this system the repressive mark behaves counterintuitively: H3K27me3 transiently ACCUMULATES during active transcription and is depleted only in the memory phase.
Evidence that argues against this (1)
- The mark set that carries memory here is acetylation and H3K4me1, with PRC2 limiting priming - a different chemistry from the H3K79me/H2AK119ub axis proposed for DOT1L. So the template establishes the experimental form and does not support the specific molecular claim, and a DOT1L memory experiment could satisfy this protocol while the underlying mechanism is unrelated. Neville D et al. 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In MLL leukaemia, 'transcriptional memory' is used in a second, distinct sense: memory of the cell of origin predicts AML patient survival and permits beta-catenin-independent transformation in MLL cancer stem cells derived from an HSC-enriched LSK population but not from myeloid-granulocyte progenitors, with Hoxa9 as the key transcriptional-memory gene and Prmt1 as a common downstream mediator.
Transcriptional memory of cells of origin overrides beta-catenin requirement of MLL cancer stem cells. · Siriboonpiputtana T, Zeisig BB, Zarowiecki M, Fung TK, Mallardo M, Tsai CT, et al. · EMBO J 2017What this evidence cannot show
Mouse transformation of sorted progenitor populations, with a human survival association layered on. The usage is metaphorical in the sense that no signal is delivered and no re-exposure occurs; 'memory' here means programme retention across transformation. It is nonetheless the established usage in this exact disease and predates the 2026 sense-1 usage by nine years.
Evidence that argues against this (1)
- Neville et al. use the phrase in the inducible sense within the same disease and without reference to cell of origin, so neither usage has priority in KMT2A-rearranged leukaemia and a reader encountering the phrase cannot infer which is meant. That ambiguity is the finding of this entry rather than a defect of either paper. Neville D et al. 2026
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Neville et al. establish DOT1L as a component of transcriptional memory co-opted in leukaemia and suggest it serves as the missing link balancing the opposing forces of the MLL-Polycomb axis, on the basis that enhanced PRC1.1 activity arises specifically from progressive loss of DOT1L-mediated H3K79 methylation, independent of MLL-fusion displacement or transcriptional repression.
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. The abstract states the memory framing as a conclusion ('establish DOT1L as a component of transcriptional memory') without describing a prime-wash-divide-rechallenge experiment; whether one was performed is not determinable from the abstract, and this entry therefore does not assert that the sense-1 protocol was run, only that the sense-1 vocabulary was used. The load-bearing mechanistic claim - antagonism independent of fusion displacement - stands on its own regardless of the memory framing.
Evidence that argues against this (1)
- A mark that maintains a transcriptional state through division is not the same as one that accelerates reactivation after a signal; the inducible-memory literature distinguishes maintenance from priming explicitly, and H3K79me's slow turnover makes it a good candidate for maintenance and an unproven one for priming. The memory claim could be true in the maintenance sense and untested in the inducible sense. Mikulski P et al. 2025
A THIRD AND LITERAL MEANING: mitotic bookmarking, and the one with a direct menin connection.
the source of the phrase in this graph; it is used there in the inducible sense
'DOT1L provides transcriptional memory' is an operational claim requiring a prime-washout-rechallenge experiment that has not been run here
- related to →KAT7 / HBO1 — an enzyme, a leukaemia-stem-cell dependency, and an inhibitor already in a crystal
KAT7 is required for the H3K14ac that carries inducible memory in the canonical mammalian system
- related to →HOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it
HOXA9 is the memory gene in the cell-of-origin sense of the term in MLL leukaemia
the prime-wash-divide-rechallenge protocol the inducible sense of the term requires