Oncofusion

start / graph / exp-q136-spike-in-normalised-men1-and-dot1l-chip-with-replicates-and-both-drug-classes-in-rs4-11

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q136-menin-vs-dot1l-inhibitors

The one experiment that would close this beat: MEN1 and DOT1L ChIP in RS4;11 (KMT2A::AFF1) under DMSO, a menin inhibitor and a target-engaging DOT1L inhibitor, with Drosophila spike-in and two replicates per arm - about 24 libraries, and the depositing lab already runs every component of it

also called the index-fusion menin-versus-DOT1L experiment

Everything found here was measured in the wrong leukaemia and without the internal ruler that would let anyone say how much protein was really lost. The same lab already has all the pieces; running them together in the right cell line would settle it.

The same thing, for a scientist

**Design.** RS4;11 (KMT2A::AFF1, Cellosaurus CVCL_0093) and, as the bridging comparator, MOLM13. Four arms: DMSO, VTP-50469 or revumenib, compound 10 (or EPZ5676 at a dose whose engagement is demonstrated in the same run), and a DOT1L degrader if available. Two biological replicates per arm. ChIP for **MEN1** and for **DOT1L**, plus **H3K79me2** as the engagement control, plus a **matched input per arm** - not one shared DMSO input. **The three things it adds that GSE149183 lacks, in order of importance:** 1. **Drosophila spike-in chromatin in the MEN1 and DOT1L tubes.** This converts every number from a within-sample contrast to an absolute scale and is the single change that matters most. The depositors already run this protocol - for the H3K79me2 samples in this very series - so it is one extra aliquot per tube [E1]. 2. **Two replicates per arm.** GSE149183 is n = 1 throughout, so no noise floor can be measured in it and none may be imported: the same-condition floor in this project spans ~20-fold across arms of a single deposit [E2]. 3. **The index fusion.** Every chromatin result in this thread is KMT2A::MLLT3 or murine MLL-AF9. RS4;11 is in GSE149183 already and was given H3K79me2 only [E1]. **Two design details that are not optional.** - **A matched input per arm.** GSE149183 divides all ten ChIPs by a single DMSO input collected on a different date. That is good for cross-arm comparability and blind to drug-induced accessibility change. - **Report the per-sample fragment-size distribution.** The fragment-size estimate spans 822-1112 bp across GSE149183's ten factor ChIPs and its rank order matches the rank order of a whole apparent effect (see `hyp-q136-catalytically-inhibited-dot1l-is-retained-on-chromatin-at-fusion-target-genes`). Without it, the same confound recurs. **Pre-specified readout, fixed before the experiment:** peak call and spike-in-normalised occupancy at the canonical menin-MLL target loci - HOXA5/7/9/10, MEIS1, PBX3, MEF2C, JMJD1C, CDK6, SENP6, MYB, FLT3, BCL2, SKIDA1, MYC - gene body +/- 5 kb, hg38, with a housekeeping set (ACTB, GAPDH, EEF1A1) as the internal negative. **Not** bulk peak-set overlap, which cannot distinguish two different antibodies in untreated cells [E3].

Go deeper
What each outcome would mean
  • Menin cleared by the menin inhibitor and retained under the DOT1L inhibitor, in RS4;11: the MOLM13 result generalises to the index fusion and menin displacement is a usable, fusion-general pharmacodynamic readout.
  • Menin also cleared by the DOT1L inhibitor in RS4;11: cycle 83's threat becomes true on the menin channel too, in the fusion that matters, and the specificity argument for the menin interface is in real trouble. This is the outcome that would most change what this project believes.
  • Menin not cleared by either in RS4;11: the B-lineage index fusion holds menin differently from KMT2A::MLLT3, which would be a bigger finding than the one this beat reports and would explain several partner-imported results at once.
Facts and figures
designRS4;11 (KMT2A::AFF1) and MOLM13 (KMT2A::MLLT3); four arms - DMSO, VTP-50469 or revumenib, compound 10 or EPZ5676 at a demonstrated target-engaging dose, and a DOT1L degrader if available; 2 biological replicates per arm; ChIP for MEN1, DOT1L and H3K79me2 plus a matched input per arm; Drosophila dm6 spike-in chromatin in every tube; 4 d treatment.
readoutpeak call and spike-in-normalised occupancy at 15 pre-specified canonical menin-MLL target loci (gene body +/- 5 kb, hg38) with ACTB/GAPDH/EEF1A1 as internal negatives; per-sample fragment-size distribution reported; genome-wide analysis secondary and labelled exploratory
controlsmatched input per arm (not one shared DMSO input); H3K79me2 spike-in ratio as the DOT1L-inhibitor engagement control at the exact dose used; DMSO-versus-DMSO replicate contrast to measure the noise floor IN THIS ARM rather than importing one
n libraries approx24
cost tiermed
timelineone ChIP-seq campaign, roughly 6-8 weeks including sequencing
closes
[
  "absence-q136-the-index-fusion-line-in-this-deposit-carries-h3k79me2-only-with-no-menin-chip-and-no-menin-inhibitor-arm",
  "absence-q136-the-dm6-spike-in-table-covers-only-h3k79me2-so-the-factor-chips-have-no-absolute-scale"
]
pre specified loci
[
  "HOXA5",
  "HOXA7",
  "HOXA9",
  "HOXA10",
  "MEIS1",
  "PBX3",
  "MEF2C",
  "JMJD1C",
  "CDK6",
  "SENP6",
  "MYB",
  "FLT3",
  "BCL2",
  "SKIDA1",
  "MYC"
]
internal negative loci
[
  "ACTB",
  "GAPDH",
  "EEF1A1"
]
How this could bite us back
  • The experiment is designed to test a pharmacodynamic readout, and a readout that saturates at 50 nM cannot support dose-finding. The design should include a sub-saturating menin-inhibitor dose precisely so that the endpoint's dynamic range is characterised before anyone proposes it for a trial.
Still open
  1. Should the menin arm be VTP-50469 (comparable to every existing deposit) or revumenib (the approved drug)? Running both doubles the menin arms and is the only way to know whether the preclinical tool compound and the clinical drug give the same chromatin readout - which nothing in this corpus establishes.
  2. Is a tagged MEN1 knock-in in RS4;11 worth the extra year? It would remove the epitope-masking ambiguity that no antibody-based design can, and that ambiguity is the single largest unaddressed weakness in this beat's conclusion.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    GSE149183 contains RS4;11 H3K79me2 with dm6 spike-in and MOLM13 MEN1/DOT1L ChIP without it, from one lab in one study - so every component of the proposed experiment is already running in that lab, just never in combination.

    What this evidence cannot show

    That the components exist separately does not mean combining them is trivial: the MEN1 and DOT1L ChIPs are on hg38 and the spike-in H3K79me2 pipeline on hg19, so the analysis pipeline would need unifying too.

    Evidence that argues against this (1)
    • A spike-in normalises for global chromatin recovery, not for antibody efficiency, so it does not make a MEN1 ChIP under drug quantitatively comparable if the drug alters epitope accessibility - which a menin-pocket inhibitor plausibly does. An orthogonal tagged-MEN1 readout is the stronger design and is more work. NCBI Gene Expression Omnibus 2020
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    Every factor-ChIP sample in GSE149183 is labelled REP1 and no condition has a second replicate, so no within-condition variance exists in that arm.

    What this evidence cannot show

    None; this is read directly from the deposited file names and sample records.

    Evidence that argues against this (1)
    • The two factor arms partially substitute for replication - two separately immunoprecipitated libraries against two proteins agree at the same loci in the same drug arm. That is weaker than replication and it is not nothing. NCBI Gene Expression Omnibus 2020
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In GSE149183 the bulk peak-set overlap statistic cannot distinguish a MEN1 ChIP from a DOT1L ChIP in the same untreated cells (0.357, inside the 0.310-0.487 range of all 45 pairs), so a pre-specified locus readout is required rather than a genome-wide overlap.

    What this evidence cannot show

    Shown for this deposit's peak calls, which are broad and near-threshold; a sharper peak caller or a narrower factor might behave differently.

    Evidence that argues against this (1)
    • Pre-specifying loci forfeits discovery: a drug effect concentrated outside the canonical target set would be invisible. The proposed design therefore keeps the genome-wide analysis as a secondary, explicitly exploratory readout. Oncofusion knowledge graph 2026
Connected to