Sort KMT2A-rearranged blasts by cell-cycle state, then measure menin-inhibitor response — the one-week experiment nobody has run
also called Hoechst/Pyronin Y menin sensitivity · cycle-state-stratified menin inhibitor assay
Separate leukaemia cells into the ones that are dividing and the ones that are resting, then give each group the same drug and see which responds. It is a standard sorting technique and a standard drug assay, and putting the two together would say whether this drug class can touch the cells that cause relapse.
The same thing, for a scientist
Tests `hyp-menin-sensitivity-tracks-cell-cycle-state`. Primary KMT2A-rearranged blasts or PDX-derived cells are separated by Hoechst 33342 and Pyronin Y into G0 (2N DNA, low RNA), G1 (2N DNA, higher RNA) and S/G2M, then exposed to a menin inhibitor at matched concentration and duration. The readout is deliberately three-layered so that failure to respond can be attributed: **target engagement** (MEIS1, HOXA9, PBX3 transcript suppression), **differentiation** (CD11b, CD14, CD15, morphology) and **death** (annexin V, viability). A G0 fraction showing preserved target suppression but blunted differentiation and death localises the block downstream of the drug's mechanism and points to a combination that supplies the missing step; a G0 fraction showing blunted target suppression points instead to drug access or chromatin state. Equal responses across fractions retire the hypothesis, which is a useful result rather than a wasted experiment.
Design
Material. Primary KMT2A-rearranged blasts (AML and B-ALL, ideally infant-derived) and PDX-derived cells, with at least one KMT2A::AFF1 and one KMT2A::MLLT3 source so that partner-dependence can be seen if it exists.
Sort. Hoechst 33342 (DNA) with Pyronin Y (RNA) resolves G0 from G1 within the 2N population, which single-parameter DNA stains cannot do. Sort G0, G1 and S/G2M. Verify the sort with Ki-67 and CDKN1B/p27 staining on an aliquot.
Exposure. Matched concentration and duration of revumenib or VTP50469 across fractions, plus vehicle. Duration must be short enough that sorted G0 cells have not re-entered cycle — the reversibility of dormancy on removal from the niche is the principal threat to this design, so a re-sort aliquot at the end of exposure is mandatory to document how much of the G0 fraction is still in G0.
Readouts, in this order:
- Target engagement — MEIS1, HOXA9, PBX3 transcripts by qPCR; menin and fusion occupancy by CUT&RUN on the larger fractions.
- Differentiation — CD11b, CD14, CD15 by flow; cytospin morphology.
- Death — annexin V and viability.
Controls. Vehicle in each fraction; a non-KMT2A-rearranged line to confirm assay specificity; a cycling KMT2A-rearranged line as a positive control for the drug lot; and an unsorted bulk arm so the result can be compared with how sensitivity is conventionally measured.
Extension if the ex vivo design fails. If dormancy does not survive the sort, repeat in vivo: dye-dilution-marked PDX cells treated in the mouse, with label-retaining and label-diluted fractions analysed directly without culture.
Why it is worth running now
Paediatric menin-inhibitor trials are enrolling, and combination design decisions are being made. Two of the combinations under discussion have opposite predicted interactions with cell-cycle state: a CDK4/6 inhibitor pushes cells toward G1 and possibly G0, while a niche-disrupting or mobilising agent would push them the other way. There is currently no data on which direction is right, and the assumption of additivity is being made by default.
The experiment is also cheap by the standards of this graph's experiment nodes: no new model construction, no knock-in, no animal cohort in the first instance. It uses reagents and assays that every leukaemia laboratory already has.
| design | Hoechst 33342 plus Pyronin Y sorting of primary KMT2A-rearranged blasts and PDX-derived cells into G0, G1 and S/G2M, followed by matched menin-inhibitor exposure in each fraction, with vehicle, unsorted-bulk, non-rearranged-line and cycling-line controls; in vivo dye-dilution extension if dormancy does not survive the sort. |
|---|---|
| readout | Three-layered and ordered: (1) target engagement by MEIS1/HOXA9/PBX3 transcript suppression and CUT&RUN occupancy on larger fractions; (2) differentiation by CD11b/CD14/CD15 and cytospin morphology; (3) death by annexin V and viability. Plus a mandatory re-sort aliquot documenting G0 retention through exposure. |
| controls | Vehicle per fraction; unsorted bulk arm for comparison with conventional sensitivity measurement; non-KMT2A-rearranged line for specificity; cycling KMT2A-rearranged line as drug-lot positive control; Ki-67 and CDKN1B/p27 staining to verify the sort. |
| cost tier | low |
| timeline | 2-4 months for the ex vivo arm given access to primary material; 6-9 months if the in vivo dye-dilution extension is required |
| principal threat | dormancy is reversible on removal from the in vivo niche, so sorted G0 cells may leave the state during exposure; the re-sort aliquot exists to detect this and the in vivo extension exists to escape it |
tests hypotheses
[ "hyp-menin-sensitivity-tracks-cell-cycle-state" ]
discriminating outcomes
[
{
"observation": "G0 shows preserved MEIS1/HOXA9 suppression but blunted differentiation and death",
"reading": "block is downstream of target engagement; a combination must supply the missing step"
},
{
"observation": "G0 shows blunted target suppression",
"reading": "drug access or chromatin state differs in G0; a different problem"
},
{
"observation": "no difference across fractions",
"reading": "hypothesis retired; menin inhibitors reach the residual compartment"
},
{
"observation": "G0 more sensitive",
"reading": "clinical implication inverts; favours extending rather than intensifying therapy"
}
]- If the result is that dormant cells resist, the obvious countermeasure is to drive them into cycle, which would also drive normal haematopoietic stem cells into cycle and risk exhausting an infant's stem-cell pool.
- A null result from the ex vivo arm cannot be trusted, because the state being tested may not survive the assay, so the experiment must be designed from the outset to report how much dormancy it retained rather than only whether it saw an effect.
- If dormant cells prove equally sensitive, then clinical resistance is chromatin-mediated rather than compartment-mediated, and no scheduling or combination change will address it.
- What fraction of a primary KMT2A-rearranged sample is in G0? If it is under one percent, the sorted fraction may be too small for a CUT&RUN occupancy readout and the design has to drop to transcript and flow readouts only.
- Is there a menin-inhibitor exposure short enough to precede G0 exit but long enough to produce a measurable differentiation response? The two requirements may not be simultaneously satisfiable ex vivo.
- Should the same design be run with a CDK4/6 inhibitor pre-treatment arm, to test directly whether pushing cells into G1 reduces menin-inhibitor response?
- Would a matched diagnosis-versus-MRD comparison of G0 signature scores in patients on menin-inhibitor therapy answer the clinical question faster than any laboratory sort?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Menin-MLL inhibition displaces menin from protein complexes, inhibits chromatin occupancy of MLL at select genes, and leads to changes in gene expression, differentiation and apoptosis, with dramatic reductions in leukaemia burden in MLL-rearranged AML and ALL patient-derived xenografts.
A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia · Krivtsov AV, Evans K, Gadrey JY, Eschle BK, Hatton C, Uckelmann HJ, et al., Armstrong SA · Cancer Cell 2019What this evidence cannot show
Cell-cycle state was not a variable; the differentiation readout is at the population level. The study's year-long disease-free survival in ALL PDX mice is evidence that the residual compartment can be cleared in that model, which sets the bar this experiment has to explain. Read: abstract only.
Evidence that argues against this (1)
- The same study's durable disease-free survival in MLL-rearranged ALL PDX mice argues that no substantial drug-refractory compartment survived, which would make this experiment's expected positive result hard to reconcile with an established in vivo finding. If the experiment shows a resistant G0 fraction ex vivo but PDX mice are cured, the ex vivo assay rather than the biology may be at fault. Read: abstract only (same paper). Krivtsov AV et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A rare dormant, treatment-resistant, stem-like subpopulation in ALL patient-derived xenografts resembles primary cells isolated at minimal residual disease, and its resistance and dormancy are reversible: the cells become treatment-sensitive and resume proliferation once dissociated from the in vivo environment.
Characterization of Rare, Dormant, and Therapy-Resistant Cells in Acute Lymphoblastic Leukemia · Ebinger S, Ozdemir EZ, Ziegenhain C, Tiedt S, Castro Alves C, Grunert M, et al., Jeremias I · Cancer Cell 2016What this evidence cannot show
PDX models not selected for KMT2A rearrangement. This evidence simultaneously motivates the experiment and identifies its principal methodological threat, which is why the design mandates a re-sort aliquot to document how much of the G0 fraction remains in G0 through exposure. Read: abstract only.
Evidence that argues against this (1)
- If reversal on removal from the niche is fast, the ex vivo arm of this design cannot work at all, and a null result would be uninterpretable rather than informative. The paper's own conclusion — that patients might benefit from strategies that release MRD cells from the niche — concedes that the dormant state is defined by the environment, which means any assay performed outside that environment is measuring something else. Read: abstract only (same paper). Ebinger S et al. 2016
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
FOXM1 is required for survival, quiescence and self-renewal of MLL-AF9 leukaemia stem cells in vivo and its inhibition kills primary LSCs from MLL-rearranged AML patients, establishing that a quiescence-dependent leukaemia-stem-cell compartment exists in this specific disease rather than in leukaemia generally.
FOXM1 regulates leukemia stem cell quiescence and survival in MLL-rearranged AML · Sheng Y, Yu C, Liu Y, Hu C, Ma R, Lu X, et al., Qian Z · Nature Communications 2020What this evidence cannot show
Mouse MLL-AF9 model with primary patient LSC validation; quiescence scored by cell-cycle distribution rather than label retention, so the compartment this experiment would sort may not be exactly the compartment FOXM1 sustains. Read: abstract only.
Evidence that argues against this (1)
- The MLL-AF9 leukaemia stem cell was originally characterised as retaining the immunophenotype and global expression profile of a normal granulocyte-macrophage progenitor, a proliferative compartment, with only a limited self-renewal programme re-activated. If most LSCs cycle, the G0 fraction this experiment sorts may be too small to yield interpretable numbers from a clinical sample, and the experiment could fail on cell numbers rather than on biology. Read: abstract only. Krivtsov AV et al. 2006
constrains the design: the proposed phase-stratified sensitivity assay must sort or elutriate, because a chemical block does not work in these lines
the sorting experiment that discriminates the outcomes
measures menin-inhibitor response as a function of cell-cycle exit
target engagement readout on the menin-KMT2A axis