Single-cell CUT&Tag on p279 — is the double-positive blast one cell state or two clones?
One patient's leukaemia appeared to have the driver protein on both HOXA and IRX1 at once. That could mean two different groups of cells mixed together, or one cell doing both. Only looking cell by cell can tell.
The same thing, for a scientist
p279 carries KMT2A-N at HOXA9/HOXA10 (2.92x) **and** at IRX1 (6.08x), both antibodies agreeing, in the deepest library pair in the deposit [E1]. Bulk CUT&RUN cannot distinguish (i) a mixture of an iALL-HOXA9 clone and an iALL-IRX1 clone from (ii) genuine co-occupancy in single nuclei. The distinction decides whether the published **mutual exclusivity** is a clone-level property that bulk averaging obscures, or a population-level tendency that individual cells violate — and it decides whether subgroup-directed therapy treats a patient or a fraction of a patient.
Design
Material. Cryopreserved p279 (GSM8000492-96 source material) if available, or any KMT2A::AFF1 blast independently shown to be bulk double-positive by the same promoter assay. p279 is the natural target because it is the deepest sample and the one patient who subsequently lineage-switched.
Assay. Single-cell CUT&Tag against KMT2A-N, ideally multiplexed with H3K4me3 on the same cells. Both N antibodies should be run on split aliquots, not pooled: they are different reagents and this beat treated them separately throughout for that reason.
Readout. Per-cell fragment counts in the IRX1 gene body and the HOXA9/HOXA10 20 kb window, as a two-dimensional per-cell scatter. Pre-declare the null: the expected joint distribution if occupancy at the two loci is independent given per-cell depth. Cells are sparse in scCUT&Tag, so a per-cell zero at either locus is uninformative individually — the claim must rest on the shape of the joint distribution across thousands of cells, not on classifying individual cells.
Controls. MEIS1, CDK6 and PAX5, which are occupied in every patient in this beat (5.4-40.4x) and should therefore be positive in essentially all cells — this is the per-cell sensitivity control, and without it a bimodal-looking scatter is indistinguishable from a depth artefact.
Outcomes
- Two separate populations → the subgroups are clones, bulk assignment is a mixture estimate, and the mixture reading that currently rescues mechanism (A) in
hyp-irx1-is-a-second-oncoprotein-address-in-infant-t411stands. Clinically it means a patient can carry both subgroups and relapse from the untreated one. - One population, both loci occupied → mutual exclusivity is a property of transcript output, not of binding, and the expression-versus-occupancy distinction this beat drew becomes the central fact about the subgroup rather than a caveat.
- A gradient with no structure → report it. A continuum at single-cell level would match the continuum this beat found at bulk level across patients (outcome B2 of the pre-registration), and would mean 'subgroup' is a threshold on a gradient at every scale examined so far.
| design | single-cell CUT&Tag for KMT2A-N (both antibodies on split aliquots), ideally multiplexed with H3K4me3, on p279 or another bulk double-positive KMT2A::AFF1 blast |
|---|---|
| readout | per-cell joint distribution of fragment counts at IRX1 gene body versus HOXA9/HOXA10 20 kb, against a pre-declared depth-matched independence null |
| controls | MEIS1, CDK6, PAX5 as per-cell sensitivity controls (occupied in all seven bulk patients); per-cell depth as a covariate in every comparison |
| cost tier | high |
| timeline | 9-12 months |
| critical dependency | availability of viable cryopreserved p279 material |
- If subgroups are clones within one patient, then any subgroup-stratified trial design assigns a patient to one arm while the other clone goes untreated - and the trial would read as a subgroup-specific failure rather than as a mixture problem.
- Is viable p279 material still available? The samples were deposited from Texas Children's Hospital in 2023-24 and the beat has no access information.
- Would scCUT&Tag have the per-cell sensitivity to call a 2-6x promoter enrichment at all? The MEIS1/CDK6/PAX5 controls are what would answer that, and if they fail the experiment reports unmeasured.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
p279 carries KMT2A-N at HOXA9/HOXA10 (2.92x) and at IRX1 (6.08x) with both N antibodies agreeing, in the deepest library pair in the deposit (P(zero in both by chance) = 0.019).
q145 re-analysis of GSE252378: IRX1 and HOXA oncoprotein occupancy across seven KMT2A::AFF1 patients (analyses/q145-irx1-in-hoxa-negative-infants) · Iqbal Lab oncofusion knowledge graph, beat q145 · internal analysis over deposited GSE252378 bigWigs 2026What this evidence cannot show
n = 1 patient, and the one selected for having lineage-switched, so not a random draw. Bulk CUT&RUN cannot distinguish mixture from co-occupancy, which is the entire premise of this experiment. Every number is a WITHIN-SAMPLE share: an interval's mean coverage divided by that same library's mean promoter coverage, so 1.0 = an average promoter in that library. **GSE252378 has no spike-in and no drug arm**, so no between-sample absolute statement is licensed; the only cross-sample quantity used is a ratio of two loci measured inside one library, where the unknown per-library scale cancels.
Evidence that argues against this (1)
- One deposit, one lab, one assay; seven patients, no RNA-seq, no menin track in any KMT2A::AFF1 sample. Janssens DH et al. 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The iALL-HOXA9 and iALL-IRX1 subgroups are defined by statistically significant mutually exclusive expression across patient cohorts.
HOXA9/IRX1 expression pattern defines two subgroups of infant MLL-AF4-driven acute lymphoblastic leukemia · Symeonidou V, Ottersbach K · Exp Hematol 2021What this evidence cannot show
A population-level statistical statement over two reanalysed cohorts; it was never a claim that an individual cell cannot express both, and this experiment tests a proposition the paper did not assert.
Evidence that argues against this (1)
- If the original cohorts contain patients expressing both, the mutual exclusivity is already softer than this graph has recorded it, and the experiment's motivation is correspondingly weaker. Symeonidou V et al. 2021
the experiment that would resolve mixture versus co-occupancy
- tests →IRX1 is a second address for the KMT2A::AFF1 oncoprotein, not merely the gene that turns HOXA off
removes or confirms the mixture reading that currently rescues the hypothesis
whether mutual exclusivity holds at single-cell occupancy
whether the state is a clone or a cell-intrinsic program