Experiment: build the missing KMT2A::AFF1 degron and titrate it — what fractional occupancy does the program actually need?
also called KMT2A-AFF1 dTAG · fusion degron titration · occupancy-response curve
Nobody has made a version of the KMT2A-AFF1 fusion that can be switched off on demand in a real leukemia cell. Building one and removing it in controlled amounts would tell us how much of the fusion a drug actually has to block.
The same thing, for a scientist
Knock an FKBP12(F36V) degron tag into the endogenous KMT2A::AFF1 allele in a t(4;11) line (SEM or RS4;11) and, if feasible, in an infant PDX. Titrate degrader dose to establish a fractional-degradation-versus-response curve, reading out fusion and menin chromatin occupancy (CUT&RUN), nascent transcription and pausing index (PRO-seq) at fusion targets, H3K79me2/H3K4me3 decay kinetics, and self-renewal/differentiation. This fills three documented gaps at once: there is no KMT2A::AFF1 degron in the indexed literature (all kinetic work is MLL::AF9 or ENL); no H3K79me2 decay half-life has ever been reported; and no dose-response relationship exists between fusion occupancy and program output, which is the number that determines whether interface disruption can work.
Why this experiment and not another1 deeper
Three separate nodes in this beat end at the same missing measurement.
mechanism-fusion-dependency-reversibility-kinetics: the 15-minute elongation result and the reversibility-of-established-disease result are both from other fusions (MLL::AF9 degron; tet-off MLL-ENL) 12.mechanism-dot1l-h3k79-methylation: the H3K79me2 decay half-life after fusion removal is unreported anywhere, and it sets the floor on how fast any mark-directed drug can act.mechanism-hoxa9-meis1-dose-requirement: the output program is buffered, and no dose-titration study exists, so 'how complete must eviction be?' is unanswered.
A single degron allele, titrated, answers all three in the right fusion and the right disease.
The method is established and its failure mode is documented: the original FKBP12(F36V)/CRBN dTAG system 3, with VHL-recruiting dTAG-1 molecules as the documented fallback for oncoproteins that resist CRBN-mediated degradation 4. Building the fallback into the plan from the start is the difference between a six-month and an eighteen-month project.
Design specifics that determine whether the result is interpretable
Tagging. CRISPR knock-in at the endogenous fusion allele, not lentiviral overexpression. Overexpression invalidates both kinetic and condensate readouts. Confirm mono-allelic tagging of the derivative chromosome and unchanged fusion protein level by immunoblot before starting.
Controls. (i) A tagged wild-type KMT2A allele in the same line, to separate fusion-specific from KMT2A-general effects. (ii) Degrader in an untagged parental line, for off-target degrader effects. (iii) A KMT2A::MLLT3 degron line as the bridge to the published MLL::AF9 data, so any discrepancy can be attributed to partner rather than to method.
Titration. At least six degrader doses spanning ~10-95% degradation, with degradation quantified by immunoblot or targeted MS at each dose, at a fixed time point. The deliverable is an occupancy-response curve, not a binary on/off comparison.
Readouts, in kinetic order. 0-60 min: PRO-seq pausing index at fusion targets; menin and fusion CUT&RUN. 2-24 h: nascent and steady-state RNA. 24-96 h: calibrated ICeChIP for H3K79me2, H3K4me3, H3K27ac (calibrated, because relative enrichment gave misleading directions in this exact system 5). 4-14 days: differentiation markers, colony replating, and in PDX, leukemia burden.
Primary endpoints. (1) Half-maximal fusion degradation required for half-maximal loss of self-renewal. (2) H3K79me2 decay half-life. (3) Whether pausing index returns toward the pre-leukemic state or merely falls.
| design | CRISPR knock-in of an FKBP12(F36V) degron tag at the endogenous KMT2A::AFF1 allele in SEM and RS4;11, plus an infant KMT2A::AFF1 PDX if tagging is feasible; six-plus degrader doses spanning approximately 10-95% degradation with degradation quantified at each dose; VHL-recruiting dTAG-1 as the documented fallback if CRBN recruitment fails. |
|---|---|
| readout | Fusion and menin chromatin occupancy by CUT&RUN; pausing index and nascent transcription by PRO-seq at fusion targets; calibrated ICeChIP for H3K79me2, H3K4me3 and H3K27ac with time-resolved decay fitting; differentiation markers, serial colony replating, and leukemia burden in PDX. |
| cost tier | high |
| timeline | 9-15 months (3-5 months to validated knock-in clones; 3 months titration and genomics; 3-6 months PDX arm) |
| compute needs | PRO-seq, CUT&RUN and ICeChIP alignment plus decay-kinetic modelling — suitable for the Explorer Slurm cluster |
primary endpoints
[ "Fractional fusion degradation required for half-maximal loss of self-renewal (the occupancy-response EC50)", "H3K79me2 decay half-life after fusion removal (never previously reported)", "Whether pausing index returns toward the pre-leukemic paused state or merely declines" ]
controls
[ "Degron-tagged wild-type KMT2A allele in the same line, to separate fusion-specific from KMT2A-general effects", "Degrader applied to untagged parental cells, for off-target degrader effects", "A KMT2A::MLLT3 degron line as a bridge to the published MLL::AF9 dataset", "Tagged-but-untreated cells, to exclude a tagging artefact" ]
tests hypotheses
[ "hyp-cell-state-plasticity-limits-interface-disruption", "hyp-methylome-of-origin-sets-fusion-target-set" ]
fills gaps
[ "no KMT2A::AFF1 degron exists", "no H3K79me2 decay half-life reported", "no fusion-occupancy dose-response curve exists" ]
- A degron reports what happens when the protein is destroyed; a drug reduces one function of a protein that remains present. The occupancy-response curve is therefore an upper bound on what an inhibitor achieves at matched occupancy.
- If the EC50 for program collapse turns out to require greater than 90% degradation, the result would argue against interface inhibitors generally and in favour of a fusion degrader — a genuinely unwelcome answer that the lab should commit in advance to acting on.
- Tagging the endogenous allele can alter fusion stability or partner binding, and the tagged-untreated control is the only guard against a whole project resting on an artefact.
- Can the endogenous fusion allele be tagged without disturbing the derivative chromosome's structure, given the breakpoint architecture? This is the main technical risk and should be scoped before committing.
- Is the reciprocal AFF1::KMT2A allele also expressed in the chosen line, and should it be tagged separately? If it contributes, an untagged reciprocal product confounds the interpretation.
- Does the occupancy-response curve differ between HOXA-high and HOXA-low backgrounds? No HOXA-low model currently exists, which is itself a finding worth reporting.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Degron-tagged MLL::AF9 degradation changes transcriptional elongation at a core subset of target genes within 15 minutes with subsequent loss of the active chromatin landscape, and combined DOT1L plus menin inhibition resembles degradation by releasing the oncoprotein from chromatin globally — establishing the assay design and the comparator dataset for this experiment.
MLL::AF9 degradation induces rapid changes in transcriptional elongation and subsequent loss of an active chromatin landscape · Olsen SN, Godfrey L, Healy JP, et al. (Armstrong SA, senior author) · Molecular Cell 82(6):1140-1155.e11 2022What this evidence cannot show
MLL::AF9 in cell lines, with no titration and no reported mark half-life — the two gaps this experiment is designed to fill.
Evidence that argues against this (1)
- The comparator dataset is MLL::AF9 in a myeloid system and the phenotype it produces is not the one KMT2A::AFF1 ALL produces. Adriaanse et al. showed KMT2A-r ALL undergoes rapid dose-dependent death under menin inhibition while KMT2A-r AML responds slowly with myeloid differentiation. A titration designed around the AML kinetics may therefore sample the wrong time window and score the wrong terminal readout in ALL. Adriaanse FRS et al. 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a doxycycline-repressible MLL-ENL model, leukemic cells that had acquired additional genetic abnormalities remained dependent on fusion expression in vitro and in vivo and its ablation caused regression of established leukemias, showing that complete fusion removal is curative in a mouse model and setting the optimistic bound this titration will calibrate against.
Acute myeloid leukemia induced by MLL-ENL is cured by oncogene ablation despite acquisition of complex genetic abnormalities · Horton SJ, Walf-Vorderwulbecke V, Chatters SJ, et al. (Williams O, senior author) · Blood 113(20):4922-4929 2009What this evidence cannot show
Complete transcriptional shutdown rather than graded degradation; KMT2A::MLLT1 in mouse.
Evidence that argues against this (2)
- Complete oncogene ablation curing established mouse leukaemia sets an optimistic bound that human data have not met: in AUGMENT-101, pharmacological disruption of the same dependency gave CR+CRh in 22.8% of 57 efficacy-evaluable KMT2A-r patients. And Mahdavi et al. found heavily pretreated KMT2A::AFF1 ALL resistant with the drug still on target. Degron-level removal in a mouse should not be read as a prediction of what removing the fusion would do in a pretreated patient. Issa GC et al. 2024
- Mahdavi et al.: multiply pretreated KMT2A::AFF1 ALL samples were far less menin-inhibitor-sensitive than earlier samples from the same patients despite no prior exposure, with sustained on-target transcriptional effect in resistant cells. Dependence on the fusion is therefore not a fixed property of the disease; it declines with disease evolution, which a degron titration in a naive line cannot see. Mahdavi L et al. 2025
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The dTAG system uses an FKBP12(F36V) tag with heterobifunctional degraders to achieve immediate, target-specific degradation at endogenous expression levels.
The dTAG system for immediate and target-specific protein degradation · Nabet B, Roberts JM, Buckley DL, et al. (Bradner JE, senior author) · Nature Chemical Biology 14(5):431-441 2018What this evidence cannot show
Methods provenance. Tagging itself can perturb protein function or stability, which the tagged-untreated control must exclude.
Evidence that argues against this (1)
- The dTAG tag is not inert and CRBN recruitment is not universal. This entry's own E4 records that VHL-recruiting dTAG molecules were developed precisely because CRBN-recruiting dTAG fails to degrade certain recalcitrant oncoproteins; a KMT2A::AFF1 that resists degradation would produce partial knockdown that reads as a shallow dose-response rather than as a technical failure. The titration needs an independent measure of degradation depth at every dose, not only a phenotype. Nabet B et al. 2020
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
VHL-recruiting dTAG molecules (dTAG-1) overcome a limitation of CRBN-recruiting dTAG molecules, which fail to degrade certain recalcitrant oncoproteins — the documented fallback if KMT2A::AFF1 resists CRBN-mediated degradation.
Rapid and direct control of target protein levels with VHL-recruiting dTAG molecules · Nabet B, Ferguson FM, Seong BKA, et al. (Gray NS, senior author) · Nature Communications 11(1):4687 2020What this evidence cannot show
Methods provenance; which E3 ligase works for a given target is empirical.
Evidence that argues against this (1)
- Whether KMT2A::AFF1 is CRBN-recalcitrant is untested, so the VHL fallback is insurance against an unmeasured risk. It carries its own: VHL and CRBN differ in expression across cell types and in the ternary geometries they permit, so switching ligase is not a drop-in control and a degradation difference between the two cannot be attributed to the target. Nature 2012
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Calibrated ICeChIP-seq in a KMT2A::AFF1 line showed H3K79me2 depletion accompanied by paradoxical increases in H3K4me3 and loss of H3K27me3, demonstrating that relative (uncalibrated) ChIP can report misleading directions in exactly this system and that calibrated measurement is required.
Non-canonical H3K79me2-dependent pathways promote the survival of MLL-rearranged leukemia · Richter WF, Shah RN, Ruthenburg AJ · eLife 10:e64960 2021What this evidence cannot show
Single cell line (MV4;11) under DOT1L-inhibitor perturbation; cited as methodological justification for calibrated histone-mark measurement in this design.
Evidence that argues against this (1)
- The calibrated-ChIP result is one experiment, and this beat cites it in three separate entries as if it were three observations. Internal calibration fixes the scale but not the biology of the spike-in: paradoxical mark movements can arise when the standard's occupancy differs between conditions, and a single cell line under one inhibitor for one duration cannot establish which direction marks move in the disease. eLife 2021
a titration design inherits the same construct-versus-allele distinction
consumes the paired occupancy-expression time courses that entry generates
Design counter-evidence: the comparator degron dataset is MLL::AF9/AML (PMID 35245435) and KMT2A-r ALL responds qualitatively differently to loss of the menin-fusion axis (PMID 38892207), so the time window and terminal readout must be re-derived for ALL.