start / graph / ana-q136-the-menin-inhibitor-evicts-dot1l-from-fusion-target-genes-while-dot1l-inhibitors-leave-it-there

AnalysismoderateQ136-menin-vs-dot1l-inhibitors

In the same deposit, the MENIN inhibitor removes DOT1L from all five DMSO DOT1L peaks at canonical target loci (p = 0.0432) while all three DOT1L inhibitors keep 5/5 - and the VTP DOT1L library still calls ACTB, GAPDH and EEF1A1, which is the internal positive control the MEN1 arm lacks

also called GSE149183 DOT1L ChIP head-to-head

The enzyme DOT1L is brought to leukaemia genes by a complex that menin holds together. Blocking menin knocked DOT1L off all of those genes, while blocking DOT1L's own enzyme activity left it sitting exactly where it was.

The same thing, for a scientist

The DOT1L ChIP arm of GSE149183 (GSM4491616-GSM4491620, MOLM13, 4 d, hg38, same shared input, same MACS2 settings) is a second, independently prepared library that asks the same question about a different subunit [E1]. Five distinct DMSO DOT1L peaks touch the fifteen pre-specified canonical menin-MLL target loci. **VTP-50469 recovers 0 of 5; EPZ5676, compound 10 and compound 11 each recover 5 of 5.** Strength-matched permutation p = 0.0432 for VTP against its own genome-wide recovery rate of 0.392; p = 1.0 for each DOT1L inhibitor [E2]. The zero is invariant at 0, 2, 5, 25 and 100 kb flanks [E2]. **This arm carries the positive control the MEN1 arm does not.** The VTP-50469 DOT1L sample still calls peaks at three housekeeping promoters - ACTB 6.33-fold, GAPDH 5.18-fold, EEF1A1 3.45-fold - while calling nothing at any of the fifteen target loci. A globally failed ChIP cannot produce that pattern. It also calls **1,695 peaks, more than the DMSO arm's 1,139**, with a maximum fold-enrichment of 84.92 against DMSO's 62.87 [E2]. **Caveat that must travel with this arm and does not apply to the MEN1 arm**: the DOT1L libraries are NOT depth-matched. DOT1L DMSO is 17,750,246 fragments and DOT1L VTP-50469 is 11,225,637, a ratio of 0.632, and the per-sample MACS fragment-size estimate ranges 822-1112 bp across this arm. The strength-matched null uses each arm's own recovery rate, which absorbs a uniform depth effect, but n = 5 is small and p = 0.0432 is the weakest number in this beat [E2]. **Direction of the effect at individual loci**, DMSO to VTP to EPZ to compound 10 to compound 11: HOXA9/HOXA10 2.56, none, 4.39, 5.08, 7.47; MEIS1 3.67, none, 5.65, 4.93, 6.82; PBX3 4.85, none, 8.32, 9.41, 13.92; SENP6 3.37, none, 6.17, 6.03, 9.68. Under every DOT1L inhibitor the enrichment at these loci is HIGHER than DMSO - see `hyp-q136-catalytically-inhibited-dot1l-is-retained-or-trapped-at-fusion-target-genes` for why that reading is confounded here and what would settle it [E2].

Go deeper

Why two arms matter more than two replicates would

Each condition is n = 1, so nothing in this deposit has a within-condition error bar. What it has instead is two separately immunoprecipitated libraries, against two different proteins, agreeing: VTP-50469 gives 0/19 on MEN1 and 0/5 on DOT1L at the same pre-specified loci, and the three DOT1L inhibitors give 36/57 and 15/15. Two independent antibodies failing at exactly the same fifteen loci in the same drug arm, and at no other loci, is not a plausible technical coincidence. It is weaker than replication and it is not nothing.

This is the predicted geometry, which is a check and not a discovery

Menin tethers the KMT2A N-terminus, the fusion recruits DOT1L through its partner, and the corpus already holds ana-q118-menin-and-dot1l-leave-together-h3k79me2-falls-and-mll1-n-stays and mechanism-q130-dot1l-itself-is-evicted-in-proportion-to-eviction-depth-so-the-mark-is-following-its-writer. The value of this measurement is not that DOT1L leaves with menin - that was known - but that the same experiment shows DOT1L inhibitors do not move DOT1L off those loci at all, which is the comparison nobody had made.

Facts and figures

script	analyses/menin-inhibitor-against-dot1l-inhibitors-in-one-deposit/run.py
status	run
geo	GSE149183
cell line	MOLM13
fusion	KMT2A::MLLT3
genome build	hg38
timepoint	4 days
replicates	1
n target dmso dot1l peaks	5
vtp depth ratio to dmso	0.632
rows dropped by any filter	0
shown for fusion	KMT2A::MLLT3 in MOLM13 only; nothing in KMT2A::AFF1
absolute scale available	false

target peaks kept

p strength matched

genome wide recovery rate

depth fragments

peaks called

vtp internal positive control

samples

[
  "GSM4491616",
  "GSM4491617",
  "GSM4491618",
  "GSM4491619",
  "GSM4491620"
]

target peaks kept — the numbers behind the chart

{
  "VTP50469": 0,
  "EPZ5676": 5,
  "comp10": 5,
  "comp11": 5
}

p strength matched — the numbers behind the chart

{
  "VTP50469": 0.0432,
  "EPZ5676": 1,
  "comp10": 1,
  "comp11": 1
}

genome wide recovery rate — the numbers behind the chart

{
  "VTP50469": 0.392,
  "EPZ5676": 0.369,
  "comp10": 0.496,
  "comp11": 0.403
}

depth fragments — the numbers behind the chart

{
  "DMSO": 17750246,
  "VTP50469": 11225637,
  "EPZ5676": 13064535,
  "comp10": 16402117,
  "comp11": 11522001
}

peaks called — the numbers behind the chart

{
  "DMSO": 1139,
  "VTP50469": 1695,
  "EPZ5676": 1214,
  "comp10": 1524,
  "comp11": 1782
}

vtp internal positive control — the numbers behind the chart

{
  "ACTB": 6.33,
  "GAPDH": 5.18,
  "EEF1A1": 3.45,
  "n_canonical_target_loci_called": 0
}

How this could bite us back

A menin inhibitor evicting DOT1L means the two drug classes are partly redundant on the DOT1L axis - so a combination may buy less than its mechanism diagram suggests, while adding the toxicity of both.
If DOT1L is lost from target genes under menin inhibition, adding a DOT1L inhibitor after menin-inhibitor failure targets an enzyme that has already left the locus. Sequencing the two drugs in that order is the combination that this result argues against most directly.

Still open

Is DOT1L's departure under VTP-50469 loss of the protein or loss of its anchor? A DOT1L immunoblot at 4 d would separate eviction from degradation, and the deposit has no protein measurement at all.
Does DOT1L leave the same loci in the same order as menin? Both arms are in this deposit and a per-locus rank correlation between the MEN1 and DOT1L losses would test whether they travel as one complex or as two - but with 19 and 5 target peaks the correlation would be underpowered, so it needs a genome-wide site set and a bigger deposit.

Every source behind this entry (3)

E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
GSE149183 deposits DOT1L ChIP in MOLM13 at 4 d under DMSO, VTP-50469, EPZ5676, compound 10 and compound 11 (GSM4491616-GSM4491620), MACS2 peak tables against the same single shared input BAM as the MEN1 arm.
NCBI GEO GSE149183 - MEN1, DOT1L and H3K79me2 ChIP-seq in MOLM13 and RS4;11 under DMSO, VTP-50469, EPZ5676, compound 10 and compound 11; 31 samples, Homo sapiens; SubSeries of GSE149185, BioProject PRJNA627612, SRA SRP258055 · NCBI Gene Expression Omnibus 2020
What this evidence cannot show
n = 1 per condition. The DOT1L arm's libraries span 11.2-17.8 M fragments, a 1.58-fold range, so this arm is NOT depth-matched the way the MEN1 arm is.
Evidence that argues against this (1)
- Within the DOT1L arm, peak count runs OPPOSITE to depth (Spearman -0.8 across the five samples), which means depth is not the dominant driver of peak count here - but it also means something else is, most plausibly the 822-1112 bp per-sample fragment-size estimate, and that confound is unresolved. NCBI Gene Expression Omnibus 2020
E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Re-analysis: VTP-50469 recovers 0/5 DMSO DOT1L peaks at the 15 canonical loci (strength-matched permutation p = 0.0432) against 5/5 for each of EPZ5676, compound 10 and compound 11; the VTP DOT1L sample nonetheless calls ACTB (6.33), GAPDH (5.18) and EEF1A1 (3.45) and 1,695 peaks overall.
analyses/menin-inhibitor-against-dot1l-inhibitors-in-one-deposit (PREREGISTRATION.md, run.py, results.json) - this repository, re-analysis of the ten deposited GSE149183 MACS2 peak tables · Oncofusion knowledge graph 2026
What this evidence cannot show
n = 5 target peaks. This is the smallest site set in the beat and p = 0.0432 would not survive any correction for the eight drug-by-factor comparisons made. It is reported as directional support for the MEN1 result, not as an independent significant finding.
Evidence that argues against this (1)
- The three housekeeping promoters are a weak control set: for the MEN1 arm the same three loci give no usable control (MEN1/ACTB is 3.54 in DMSO and is lost under VTP, EPZ and compound 11 alike), so the internal positive control exists for the DOT1L arm and NOT for the MEN1 arm. The MEN1 arm's equivalent assurance is its 1,101 called peaks and its 88.19-fold maximum. Oncofusion knowledge graph 2026
E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
MOLM-13 carries KMT2A::MLLT3, so the DOT1L recruited to these loci is recruited through an AF9/ENL-family partner, not through AFF1.
Cellosaurus CVCL_2119 (MOLM-13): Sequence variation - Gene fusion KMT2A + MLLT3 (MLL-AF9); FLT3-ITD heterozygous · Cellosaurus (SIB) 2026
What this evidence cannot show
The recruitment route differs between KMT2A::MLLT3 and KMT2A::AFF1 in the literature, so this arm's geometry is not automatically the index fusion's.
Evidence that argues against this (1)
- The corpus holds mechanism-kmt2a-aff1-dot1l-recruitment-route as a separate entry precisely because the routes differ; importing this result to KMT2A::AFF1 would be partner-imported evidence of the kind this project keeps finding under its load-bearing claims. Cellosaurus (SIB) 2026

Connected to

tests →DOT1L — the target where excellent chemistry met an indirect interaction
DOT1L ChIP occupancy at the same 15 loci under the same four drugs
supports →First deposit to put both drug classes on the MEN1 channel in one experiment: VTP-50469 leaves ZERO of the 19 DMSO menin peaks at 15 canonical menin-MLL target loci, while EPZ5676 keeps 14, compound 10 keeps 14 and compound 11 keeps 8 - strength-matched permutation p = 0.0003, and the zero is unchanged at every flank from 0 to 100 kb
a second, independently immunoprecipitated library gives the same 0-at-target-loci result for VTP-50469 (0/5, p = 0.0432) with an internal housekeeping control
related to →Pinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
EPZ5676 leaves DOT1L on 5/5 target peaks - inhibiting the enzyme does not remove it from chromatin
supports →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
DOT1L leaves fusion target loci under a menin inhibitor in a second system and a second assay
related to →MEN1 / menin — the adaptor the fusion cannot work without
the menin inhibitor, not the DOT1L inhibitors, is what removes DOT1L from these loci
← depends onResolving cycle 83's threat: the MLL1 channel moves under both drug classes and is therefore not diagnostic, but the MEN1 channel separates them completely - so 'menin displacement' remains a specific readout of menin-interface engagement and the corpus must stop treating the two channels as one
the DOT1L channel gives the same separation in a second library
← depends onUnder all three DOT1L inhibitors, DOT1L's fold-enrichment at canonical target loci is HIGHER than DMSO (PBX3 4.85 to 13.92, SENP6 3.37 to 9.68) - either catalytically dead DOT1L is retained or trapped on chromatin, or the reading is an artefact of a 1.35-fold spread in MACS fragment-size estimates, and this deposit cannot tell which
the observation it is built on, and the arm whose confound prevents calling it a result
← related toA per-sample parameter nobody reads: MACS's fragment-size estimate varies 822-1112 bp across ten ChIPs from one sequencing run, and within the DOT1L arm peak count correlates +0.7 with fragment size and -0.8 with library depth - so the obvious confound is not the operating one, and the non-obvious one is
the DOT1L arm spans 1.58x in depth and 1.35x in fragment size, which is why it is reported as directional support only