Oncofusion

start / graph / ana-q139-neither-single-agent-comes-close-to-degrading-the-fusion-and-only-the-combination-does

Analysismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q139-fusion-degron

Scored against the same DMSO in the same batch, degrading the fusion leaves 1.0% of its chromatin sites, the DOT1L+menin combination leaves 10.7%, the menin inhibitor alone leaves 51.7% and the DOT1L inhibitor alone 75.5% - and in a second independent batch the menin inhibitor again beats the DOT1L inhibitor on the fusion's own channel

also called the drug-versus-degron ladder · 1.0 / 10.7 / 51.7 / 75.5 · only the combination phenocopies degradation · how far short do the drugs fall

Destroying a leukaemia-driving protein outright clears 99% of its spots on DNA, the two available drugs together clear about 89%, and either drug alone clears far less - roughly half for one and a quarter for the other. This is the first time this project has been able to say how far short the drugs fall of simply removing the protein, because the comparison needed a yardstick nobody had.

The same thing, for a scientist

**The design.** Batch 200908 of GSE173599 contains HA ChIP at 4 days in HCB1 against a single shared DMSO, in five arms — **degron, DOT1L inhibitor, menin inhibitor, both inhibitors, control** — so degron and drug are scored against the same reference in the same sequencing run. **The channel is HA, i.e. the fusion itself**, not a proxy. | arm, 4 d | peaks | top-decile summits retained | |---|---:|---:| | DMSO | 11,044 | 100% | | **dTAG-VHL 500 nM (degrade)** | 320 | **1.0%** | | **EPZ-5676 250 nM + VTP-50469 100 nM** | 572 | **10.7%** | | VTP-50469 100 nM (menin) | 1,558 | **51.7%** | | EPZ-5676 250 nM (DOT1L) | 2,365 | **75.5%** | **A second, independent batch (1912x2) replicates the drug ordering**, with no combination arm: degron 10.3%, menin inhibitor 76.9%, DOT1L inhibitor 79.4%. **Three things fall out, in descending order of confidence.** 1. **Only the combination approaches degradation.** 10.7% against the degron's 1.0%; each single agent leaves five to seven times as much fusion on chromatin as the combination does. This reproduces, from the deposited peak tables and independently of the authors' own normalisation, their stated conclusion that *"only the combination of EPZ-5676 and VTP-50469 induced MLL::AF9 chromatin loss with similar kinetics as MLL::AF9 degradation"* [E1]. 2. **The menin inhibitor outperforms the DOT1L inhibitor on the fusion's own channel, in both batches** (51.7% vs 75.5%; 76.9% vs 79.4%). This is the same ordering the sibling beat found on the MLL1 channel in MOLM13, now in a different cell line, on a different antibody, against the fusion's own tag. 3. **Even the combination does not equal removal.** A tenfold gap between 10.7% and 1.0% is the measured distance between the best available pharmacology and deleting the protein. **The second batch is the caution.** There the menin/DOT1L separation is 76.9% vs 79.4% — **2.5 points, well inside anything this deposit can resolve** — while in batch 200908 it is 23.8 points. The *ordering* replicates; the *magnitude* does not, which is this project's standing finding appearing again, now between batches of one experiment rather than between cell lines.

Go deeper
Why a degron reference changes what a drug number means

Until now every displacement figure in this corpus has been a percentage of a control, with no upper anchor. "VTP-50469 removes 58% of MLL1 sites" invites the question out of how much removable? and nothing answered it.

The degron supplies the anchor. Against it, VTP-50469 at 100 nM for 4 days achieves roughly half of what deleting the protein achieves, and EPZ-5676 at 250 nM achieves about a quarter. Those are the numbers a translational reader wants, and they are considerably less flattering than the drug-versus-DMSO framing.

One dose caveat that matters. The DOT1L arm here is 250 nM, an order of magnitude below the 1 µM used in the MOLM13 2x2. The DOT1L inhibitor is therefore being judged at a low exposure in this ladder, and its poor showing is partly a dose statement.

H3K27ac, the same ladder, as context and not as a control

The same batch design exists for H3K27ac at 4 days in HCB1, and under summit containment nothing moves at all: degron 99.9%, DOT1L 100.0%, menin 100.0%, combination 99.8%. Under bin overlap the same arms read 90.8 / 94.8 / 90.0 / 89.3%.

Read as context, not as a control. H3K27ac responds to both drugs on its own account, so agreement does not validate the HA result and disagreement would not invalidate it. What it does show is that the HA collapse is not a generic loss of ChIP signal across the experiment — in the same cells, same day, same batch, an abundant histone mark's peak calls are completely unmoved.

Facts and figures
scriptanalyses/what-leaves-when-the-fusion-is-degraded/run.py
statusrun
result summaryDegron 1.0% < combination 10.7% < menin inhibitor 51.7% < DOT1L inhibitor 75.5% retained, one shared DMSO, one batch. Ordering of the two single agents replicates in an independent batch (76.9% vs 79.4%) but the magnitude does not. H3K27ac in the same design is unmoved in every arm.
preregistrationanalyses/what-leaves-when-the-fusion-is-degraded/PREREGISTRATION.md
inputs
[
  "GSE173599 GSM5271664/GSM5271667/GSM5271669/GSM5271670/GSM5271672 (HA 4 d, batch 200908)",
  "GSE173599 GSM5271663/GSM5271666/GSM5271668/GSM5271671 (HA 4 d, batch 1912x2)",
  "GSE173599 GSM5271653-GSM5271657 (H3K27ac 4 d, batch 201007)"
]
How this could bite us back
  • DELIVERY AND EXPOSURE: the combination needs both agents at target simultaneously for four days to reach 10.7%. Two chromatin drugs with different pharmacokinetics, both requiring sustained target engagement, is a demanding delivery problem and the single-agent numbers show what partial coverage buys - very little.
  • TOXICITY: DOT1L and menin are both required in normal haematopoiesis. A combination that approaches total fusion removal is also approaching total inhibition of two housekeeping chromatin functions, and the degron arm has no equivalent normal-cell cost because it removes only the transgene.
  • RESISTANCE: a drug regimen that leaves half the fusion's sites occupied (either single agent) is selecting under a partial block, which is the classic setting for escape. The degron shows the target is fully removable in principle; the drugs show it is not being removed in practice.
Still open
  1. What does the ladder look like at therapeutic exposures - revumenib or ziftomenib rather than VTP-50469, and EPZ-5676 at 1-5 uM rather than 250 nM? The tool-compound doses here are not the clinical ones and this project has already been burned once by publishing a negative that was a fact about a weak tool compound.
  2. Does the residual 10.7% under combination treatment sit at the SAME loci as the residual 1.0% under degradation? If the drugs leave fusion at a different subset than incomplete degradation does, the combination is not simply a weaker version of removal.
  3. Is the ladder the same in a KMT2A::AFF1 line? Nothing in this deposit is AFF1, and the ordering-transfers-magnitude-does-not rule means the ratios here should not be carried across.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Only the combination of EPZ-5676 and VTP-50469 induces MLL::AF9 chromatin loss with kinetics similar to MLL::AF9 degradation; single agents have more modest effects on MLL::AF9 occupancy.

    MLL::AF9 degradation induces rapid changes in transcriptional elongation and subsequent loss of an active chromatin landscape. · Olsen SN, Godfrey L, Healy JP, Choi YA, Kai Y, Hatton C, Perner F, Haarer EL, Nabet B, Yuan GC, Armstrong SA · Molecular Cell 2022
    PMID 35245435DOI
    What this evidence cannot show

    The authors' conclusion is drawn from normalised ChIP-seq signal and RT-qPCR/RNA-seq time courses across human and murine FKBP12-MLL::AF9 systems and MOLM13; the ladder here is peak-call presence in one human system. The two agree, but they are not the same measurement and the agreement is corroboration rather than replication. EPZ-5676 is at 250 nM in this ladder, well below the 1 uM used elsewhere in the same deposit, so the DOT1L arm's weakness is partly a dose effect.

    Evidence that argues against this (1)
    • The second independent batch (1912x2) puts the menin and DOT1L inhibitors 2.5 points apart (76.9% vs 79.4%) rather than 23.8 points apart. If that batch were the only one available the conclusion 'the menin inhibitor outperforms the DOT1L inhibitor' would not be supportable at all. The ordering replicates; the effect size does not. NCBI Gene Expression Omnibus 2022
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    In batch 200908 at 4 days, HA top-decile summit retention is 1.0% under dTAG-VHL, 10.7% under EPZ-5676 250 nM plus VTP-50469 100 nM, 51.7% under VTP-50469 alone and 75.5% under EPZ-5676 alone, all against one shared DMSO arm.

    What this evidence cannot show

    n = 1 library per arm, no replicates within the batch, no spike-in, no measurable noise floor. HCB1 is KMT2A::MLLT3, so the ladder is partner-imported and does not speak to KMT2A::AFF1. HA reads the tagged transgene only. The 4 d timepoint means these are steady-state drug effects in a population that has had four days to change composition, not acute occupancy changes.

    Evidence that argues against this (1)
    • A four-day drug exposure selects and differentiates the population, so a lost peak call can reflect a changed cell state rather than displacement of the fusion in a constant cell. The degron arm at 3 h does not have this problem; the 4 d drug arms do, and they are the ones carrying the pharmacological comparison. Olsen SN et al. 2022
Connected to