Degrade DOT1L, inhibit DOT1L, or leave a dead enzyme in place - run all three against the same cells and the same marrow
also called degrader versus inhibitor head to head · catalytic-dead knock-in in KMT2A-rearranged cells
The argument about whether the DOT1L drug failed because it was the wrong kind of drug has never actually been tested in leukaemia cells. This experiment does the three comparisons side by side: remove the protein, block its chemistry, or replace it with a chemically dead copy, and see which one the leukaemia minds most - and which one healthy bone marrow minds most.
The same thing, for a scientist
The central question of this beat - refuted target or wrong modality - has a three-arm answer and nobody has run it in this disease. **Arm A: genetic.** In a KMT2A::MLLT3 line and a KMT2A::AFF1 line, compare (i) DOT1L knockout, (ii) catalytically dead DOT1L knock-in, (iii) AF9-binding-deficient DOT1L knock-in, (iv) wild-type control. This genotype series exists in mouse ESCs [E1], in mouse embryos [E2] and in mouse marrow with an MLL-AF9 line [E3], and **has never been run as a complete series in a human KMT2A-rearranged leukaemia line**. **Arm B: pharmacological, matched.** Titrate a DOT1L PROTAC [E4] and pinometostat to **matched time-integrated target coverage** - degradation depth x time for the degrader, H3K79me2 depletion depth x time for the inhibitor - rather than to matched concentration, and compare viability, differentiation and fusion-target expression. **Arm C: the toxicity arm nobody has published.** Run both agents at their efficacious exposure against primary human CD34+ progenitors in colony assays and against murine marrow in vivo, because adult haematopoiesis requires DOT1L's catalytic activity and a degrader removes it [E3]. Cost tier: medium. Timeline: 9-15 months.
Design, with the controls that decide whether it is interpretable
Cell systems. At least one myeloid (KMT2A::MLLT3, e.g. MOLM-13-class) and one lymphoid (KMT2A::AFF1, e.g. SEM-class) line, plus a KMT2A-germline leukaemia line as the specificity control. The lymphoid arm is not optional: the graph's own DepMap synthesis finds the larger raw dependency difference in the lymphoid half (-0.310 against -0.186), and every mechanistic DOT1L study in this beat's reading used a myeloid fusion.
The matching step is the experiment. Comparing a degrader to an inhibitor at matched concentration answers nothing, because they have different pharmacology; comparing at matched target coverage over time is what isolates modality. For the inhibitor, coverage is H3K79me2 depletion measured by calibrated ChIP; for the degrader, it is residual DOT1L protein by quantitative immunoblot or targeted mass spectrometry. Both curves integrated over the same window.
Controls that must be present.
- A VHL-null or VHL-low line as a degrader-specific negative control, since the published first-in-class degrader is VHL-recruiting and its inactivity in such a line would be trivially explained 5.
- An inactive epimer of the PROTAC to separate degradation from target binding.
- Readouts at multiple time points spanning several cell divisions, because H3K79me2 has no eraser and is lost by dilution, so a 72-hour endpoint measures pharmacology rather than biology.
Primary readouts. Viability and colony-forming capacity; HOXA9/MEIS1 and FLT3 expression separately, since these mark the two dose regimes; H3K79me1/me2/me3 by calibrated ChIP at fusion targets and genome-wide.
What each outcome means
| outcome | reading |
|---|---|
| KO more lethal than catalytic-dead, in leukaemia cells | catalysis-independent function is real in this disease; hyp-pinometostat-wrong-modality supported |
| KO and catalytic-dead equivalent | catalysis is the whole contribution; the pinometostat result was about dose, not modality; the hypothesis is falsified |
| degrader beats inhibitor at matched coverage | modality matters pharmacodynamically |
| degrader beats inhibitor only at unmatched exposure | the advantage is duration of effect, not modality - which is still useful and is a different claim |
| degrader more toxic to CD34+ colonies at equi-efficacious exposure | the predicted cost of the modality, and the reason to prefer interface inhibition |
The fourth row is the one most likely to be reported as the third if the matching step is skipped, which is why the matching step is written as the core of the design rather than as a refinement.
| design | Three-arm comparison in human KMT2A-rearranged leukaemia lines of both lineages. Arm A: isogenic DOT1L genotype series (knockout, catalytically dead knock-in, AF9-binding-deficient knock-in, wild type). Arm B: DOT1L PROTAC versus pinometostat titrated to MATCHED time-integrated target coverage - residual DOT1L protein for the degrader, calibrated H3K79me2 depletion for the inhibitor. Arm C: both agents at equi-efficacious exposure against primary human CD34+ colony-forming assays and murine marrow in vivo. |
|---|---|
| readout | viability and colony-forming capacity; HOXA9/MEIS1 and FLT3 expression scored separately as markers of the two dose regimes; H3K79me1/me2/me3 by calibrated internal-standard ChIP at fusion targets and genome-wide; residual DOT1L protein by quantitative immunoblot or targeted mass spectrometry |
| controls | KMT2A-germline leukaemia line for specificity; VHL-null or VHL-low line as a degrader-specific negative control; inactive PROTAC epimer to separate degradation from binding; time course spanning several cell divisions because the mark is lost by dilution and has no eraser |
| cost tier | med |
| timeline | 9-15 months |
| decisive outcome | If DOT1L knockout is more lethal than a catalytically dead knock-in in a KMT2A-rearranged line, catalysis-independent function is real in this disease and the pinometostat result is a modality result. If they are equivalent, it is a dose result. |
| why not done yet | the four-genotype knock-in series has been built in mouse ESCs and in mouse embryos but not in a human KMT2A-rearranged line, and the two degrader papers report efficacy without a matched-coverage comparator |
cell systems
[ "KMT2A::MLLT3 (myeloid)", "KMT2A::AFF1 (lymphoid)", "KMT2A-germline control", "primary human CD34+ progenitors" ]
tests hypotheses
[ "hyp-pinometostat-wrong-modality", "hyp-menin-inhibition-spares-endogenous-h3k79-unit" ]
- If the degrader wins on efficacy and loses on marrow toxicity, the experiment resolves the mechanism and does not resolve the clinical question, which is whether any DOT1L modality has a usable window.
- Matching on target coverage rather than concentration is the correct design and is also the step most likely to be dropped under time pressure, at which point the experiment produces a publishable superiority result that answers nothing.
- Are the published DOT1L PROTACs available to external laboratories, and if not, is MS2133's chemistry reproducible from the published route?
- Can a catalytically dead knock-in be made cleanly at the endogenous DOT1L locus in a KMT2A::AFF1 line, or does the size of the locus force an overexpression design with the dosage caveats that carries?
- What is the right unit of matched target coverage when the two agents act on different quantities - protein abundance versus mark depth? The experiment proposes integrating each over time, but the two integrals are not obviously commensurable.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In mouse embryonic stem cells, DOT1L knockout and catalytically dead DOT1L knock-in give different phenotypes: glial specification defects and SEC-sensitised elongation defects appear in the knockout and not in the catalytic-dead cells, establishing that the knockout-versus-catalytic-dead comparison is informative and technically achievable.
DOT1L-controlled cell-fate determination and transcription elongation are independent of H3K79 methylation. · Cao K, Ugarenko M, Ozark PA, et al. · Proc Natl Acad Sci U S A 2020What this evidence cannot show
Abstract read from the eutils record; full text not read. ESCs, not leukaemia. Cited as precedent for the genotype series rather than for its result. Building the same series in a human leukaemia line requires knock-in editing at a 1,537-residue locus, which is the main technical risk in this design.
Evidence that argues against this (1)
- A catalytically dead knock-in retains SAM binding and the intact catalytic fold, so it does not test every enzymatic contribution and a null result in that arm would not fully exclude catalysis-dependent function. The arm bounds the question rather than closing it. Jang S et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A Dot1l methyltransferase-mutant mouse separates from the Dot1l knockout in primitive erythropoiesis, and EPZ-5676 administered in vivo minimally affects erythropoiesis, showing the same separation-of-function design is informative in a haematopoietic setting.
Primitive Erythropoiesis in the Mouse is Independent of DOT1L Methyltransferase Activity. · Malcom CA, Ratri A, Piasecka-Srader J, et al. · Front Cell Dev Biol 2021What this evidence cannot show
Abstract read from the eutils record; full text not read. Embryonic, and the mouse dies by E13.5 in both genotypes, so this precedent does not extend to adult marrow. The degree of H3K79 methylation reduction achieved by EPZ-5676 in vivo is not stated, which is exactly the matching problem this experiment is designed to solve.
Evidence that argues against this (1)
- Adult marrow gives the opposite assignment - an enzymatically inactive DOT1L fails to rescue sustained haematopoiesis - so the embryonic precedent may not predict the toxicity arm, and the experiment must include adult marrow rather than relying on developmental data. Grigsby SM et al. 2021
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In MLL-AF9 cells a DOT1L mutant with a defective AF9 interaction site loses recruitment and leukaemic growth, and in adult mouse marrow wild-type and PPI-deficient DOT1L but not an enzymatically inactive mutant rescue sustained haematopoiesis, so three of the four genotypes in this design already exist in mouse systems.
Elucidating the Importance of DOT1L Recruitment in MLL-AF9 Leukemia and Hematopoiesis. · Grigsby SM, Friedman A, Chase J, et al. · Cancers (Basel) 2021What this evidence cannot show
Abstract read from the eutils record; full text not read. Mouse and MLL-AF9 only; the constructs are expressed against an endogenous background rather than knocked in, and expression levels are not stated. The lymphoid arm of this experiment has no precedent at all.
Evidence that argues against this (1)
- Because the existing series is built on overexpressed constructs rather than knock-ins, its assignment of function could be dosage-dependent, and a knock-in series in human cells might not reproduce it. That is a reason to run the experiment, and also a reason not to treat the existing result as settled. Cao K et al. 2020
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
DOT1L PROTAC degraders exist with DC50 values of 0.33 micromolar and 5 nanomolar, show MLL-rearrangement-dependent viability effects, retain activity against menin-inhibitor-resistant cells and produce complete regression in an orthotopic leukaemia model, but no comparison against a catalytic inhibitor at matched target coverage is reported.
Development of DOT1L-Targeted Protein Degraders for Treating MLL-r Leukemia. · Quan S, Unno K, Gupta DG, et al. · J Med Chem 2026What this evidence cannot show
Abstract read from the eutils record; full text not read; 2026, single report. The availability of these compounds to an external laboratory is unknown and is a practical precondition for arm B.
Evidence that argues against this (1)
- A degrader with a 5 nM DC50 and event-driven pharmacology may simply sustain target suppression longer than an infused reversible inhibitor, which would produce a superiority result that says nothing about catalysis-independent function - the exact confound the matched-coverage design exists to remove. Chory EJ et al. 2019
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The first-in-class DOT1L PROTAC requires binding to the E3 ligase VHL and acts through the ubiquitin-proteasome system, so VHL status is a determinant of degrader activity.
Discovery of the first-in-class DOT1L PROTAC degrader. · Yim H, Sun R, Xu Z, et al. · Eur J Med Chem 2025What this evidence cannot show
Abstract read from the eutils record; full text not read. The E3 ligase of the second degrader series is not stated in its abstract, so the VHL control applies with certainty only to MS2133.
Evidence that argues against this (1)
- If the second series uses a different E3, a VHL-null control would be the wrong negative control for it, and a single shared control across compounds would be misleading. The control must be chosen per compound once the ligase is known. Quan S et al. 2026
knockout versus catalytic-dead in a KMT2A-rearranged line is the decisive comparison
degrader versus catalytic inhibitor at matched time-integrated target coverage