Oncofusion

start / graph / mechanism-q130-dot1l-itself-is-evicted-in-proportion-to-eviction-depth-so-the-mark-is-following-its-writer

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q130-does-the-eviction-gradient-exist-in-h3k79me2

In human MOLM13, DOT1L's own occupancy under a menin inhibitor runs 0.1364 / 0.3083 / 0.3555 / 0.3913 across MLL1-N eviction-depth quartiles at Spearman -1.000 - the writer is lost in proportion to eviction depth, and its mark follows with a 2.9-fold smaller span

also called DOT1L follows the fusion · the writer is graded too · why H3K79me2 tracks eviction depth

The enzyme that puts the activating tag on is itself pulled off the DNA in proportion to how much fusion protein the drug removed. So the tag's pattern is most simply explained by the enzyme's: where the fusion goes, the enzyme goes, and the tag decays behind it.

The same thing, for a scientist

**Measured in the same MOLM13 cells, the same bins, the same contrast** as the H3K79me2 gradient [E1]: | MLL1-N eviction quartile | Q1 most | Q2 | Q3 | Q4 least | Spearman | span | aggregate | |---|---:|---:|---:|---:|---:|---:|---:| | **DOT1L occupancy** VTP/DMSO | **0.1364** | 0.3083 | 0.3555 | 0.3913 | **-1.000** | -0.2549 | 0.2773 | | H3K79me2 VTP/DMSO | 0.7746 | 1.0601 | 1.0995 | 1.1421 | **-1.000** | -0.3675 | 0.998 | | INPUT (control) | 0.9497 | 0.9740 | 0.9603 | 0.9377 | +0.400 | +0.0120 | 0.9542 | **DOT1L occupancy collapses far more than its mark does** - aggregate 0.2773 against 0.998 - and it collapses *most* where the fusion is most evicted [E1]. **The simplest mechanism consistent with this is that the mark follows the writer.** DOT1L is recruited to MLL-fusion sites through the AF9/ENL-AF4/AFF4 axis, the menin inhibitor evicts the fusion, DOT1L leaves with it in proportion, and H3K79me2 - a stable mark diluted by replication rather than actively removed - decays behind it at a lag. **That lag is what the smaller mark span measures**: a 2.9-fold difference between the writer's aggregate depletion (0.2773) and the mark's (0.998) over three days [E1][E2]. **And the murine timecourse is consistent with a decay behind the writer** rather than with an instantaneous loss: the gradient span grows monotonically over 4, 8, 16 and 24 h and is larger still at 48 h, so nothing has plateaued in any measurement available [E3].

Go deeper
What this explains and what it does not

Explains: why an active mark shows an eviction gradient at all, and why its span is smaller than its writer's. Explains: why the DOT1L inhibitor produces a large, flat effect in the murine arm - blocking catalysis removes the mark everywhere it is being maintained, regardless of where the fusion is.

Does not explain the Q4 gain. H3K79me2 at the LEAST-evicted quartile goes UP - 1.1421 in human, 1.1865 in mouse - while DOT1L occupancy there is still down at 0.3913. A mark rising where its writer has fallen by 60% is not a writer-follows-fusion story, and this entry does not have one. The candidate explanations are a within-sample-share artefact (the share must go somewhere), a difference in turnover, or genuine redistribution of residual DOT1L activity - and nothing here separates them.

Does not establish causation. The writer and the mark are correlated across four strata; n = 4, and a DOT1L degron or a washout arm would be needed to order them in time.

Facts and figures
mechanism classwriter eviction and mark decay under pharmacological fusion displacement
systemMOLM13, human KMT2A::MLLT3 AML, GSE127507 (PMID 31821784's deposit)
dot1l aggregate0.2773
h3k79me2 aggregate0.998
absence classmeasured-not-analysed
absence class noteQ120 binned the MOLM13 DOT1L tracks and used them for a four-way ordering of RETAINED occupancy at co-occupied sites. Their eviction-depth profile was never computed.
absence class assigned2026-09-23 (Q130)
epigenetic readoutDOT1L occupancy and H3K79me2, same bins, same contrast, eviction-stratified
dot1l by quartile
[
  0.1364,
  0.3083,
  0.3555,
  0.3913
]
h3k79me2 by quartile
[
  0.7746,
  1.0601,
  1.0995,
  1.1421
]
input by quartile
[
  0.9497,
  0.974,
  0.9603,
  0.9377
]
How this could bite us back
  • If a menin inhibitor already removes 72% of DOT1L's occupancy share at the fusion's sites, the headroom for adding a DOT1L catalytic inhibitor at those exact sites is small - and the DOT1L inhibitor's own effect is flat across eviction depth, so most of what it adds is added where the menin inhibitor already works least. That is an argument for combination and an argument that the combination's extra toxicity falls on cells with no eviction gradient at all.
  • A mark that decays behind its writer sets the pharmacodynamic sampling time, and nothing in either deposit has plateaued by the last timepoint measured. An assay sampled too early reads a smaller gradient than the drug produced.
Still open
  1. Why does H3K79me2 RISE at the least-evicted quartile while DOT1L occupancy there is still down 61%? No mechanism in this entry accounts for that and it is the most interesting unexplained number in the beat.
  2. Does the writer lead the mark in time? Ordering them needs a DOT1L occupancy timecourse, and none exists in either deposit.
  3. Is the mark's slower loss replication-dilution? A replication-timing join would test it and was not attempted.
Every source behind this entry (3)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    In MOLM13 (human KMT2A::MLLT3) under 330 nM VTP-50469 for 3 days, DOT1L occupancy runs 0.1364 / 0.3083 / 0.3555 / 0.3913 across MLL1-N eviction-depth quartiles at Spearman -1.000 with aggregate 0.2773, while H3K79me2 in the same bins runs 0.7746 / 1.0601 / 1.0995 / 1.1421 at Spearman -1.000 with aggregate 0.998 and the INPUT is flat.

    What this evidence cannot show

    Same five weaknesses as the human arm generally: shared MLL1-N epitope, no third culture, no DOT1L-inhibitor arm in this series, rank quartiles, n = 1 per antibody per condition, within-sample share only. The DOT1L antibody is CST 77087S clone 1W4Z and IS named in these records, unlike GSE260742's.

    Evidence that argues against this (2)
    • A correlation between a writer's occupancy and its mark across four strata is n = 4 and does not order them causally; the mark could be lost first and the writer follow. Iqbal Lab oncofusion knowledge graph 2026
    • The system is murine MLL-AF9 (KMT2A::MLLT3) myeloid cells carrying a retrovirally over-expressed FLAG-tagged construct. This graph exists for human infant B-lineage KMT2A::AFF1, and this corpus's commonest documented synthesis failure is generalising a murine MLL-AF9 myeloid result to that disease. Neville D et al. 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Selective DOT1L inhibition kills MLL-rearranged leukaemia cells and depletes H3K79 methylation, establishing that the mark at these loci depends on continuous DOT1L catalytic activity.

    PMID 31821784DOI
    What this evidence cannot show

    Cited for the dependence of the mark on DOT1L activity in MLL-rearranged cells, which the founding menin-inhibitor paper also documents for its own system. It is not cited for anything about eviction-depth stratification, which is not in it.

    Evidence that argues against this (1)
    • Dependence of the mark on the writer's activity does not establish that the mark's spatial pattern follows the writer's occupancy; a freely diffusing enzyme could maintain a pattern set by something else. Iqbal Lab oncofusion knowledge graph 2026
  3. E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.dataset

    The murine timecourse is consistent with a mark decaying behind its writer rather than being removed instantaneously: the eviction-gradient span grows monotonically at -0.1041 / -0.1976 / -0.2139 / -0.3347 over 4/8/16/24 h and is -0.6005 at 48 h.

    What this evidence cannot show

    The timecourse arm was declared batch-affected by Q124 and is pre-declared supporting only in this beat's pre-registration; the 48 h point is from a DIFFERENT culture (NTvKO) than the timecourse, so the five points are not one series. Confidence is weak for both reasons.

    Evidence that argues against this (1)
    • Stitching a 48 h point from one culture onto a 4-24 h series from another is exactly the kind of cross-batch comparison Q124's own criterion K4 tested and found failing. Iqbal Lab oncofusion knowledge graph 2026
Connected to