IRX1 is a second address for the KMT2A::AFF1 oncoprotein, not merely the gene that turns HOXA off
Either the leukaemia protein has moved house from HOXA to IRX1 in half these patients, or IRX1 is something the cell switched on that then silenced HOXA. Both fit what we can see, and telling them apart changes which drugs should work.
The same thing, for a scientist
Two mechanisms explain the same anti-correlation and this beat cannot separate them. **(A) Second-address:** the oncoprotein is redirected to IRX1 in the HOXA-negative state, making IRX1 a bona fide fusion target and the subgroup a matter of where the fusion binds. **(B) Kühn's direction:** IRX1 is switched on by something else, binds the MLL-AF4 complex non-competitively and shuts HOXA down, making the occupancy at IRX1 a consequence rather than a cause [E1]. The chromatin data are compatible with both: the oncoprotein **is** at IRX1 at 2.7-6.4x mean promoter, with ENL and H3K4me3, in the HOXA-dead states [E2] — which (A) predicts and (B) does not forbid. Prior: **0.35** for (A) as the dominant mechanism.
What each mechanism predicts that the other does not
| observation | (A) second address | (B) IRX1 represses the fusion |
|---|---|---|
| oncoprotein at IRX1 | required | permitted, not required |
| IRX1 protein at HOXA9/HOXA10 in patient blasts | not required | required |
| menin retained in the complex where HOXA is dead | permitted | required (Kühn's co-IP) |
| a HOXA-negative, IRX1-negative patient | problem — where is the fusion? | fine, IRX1 is just one route to HOXA-off |
| menin-inhibitor sensitivity in the IRX1 state | retained (fusion still working, elsewhere) | reduced (output already suppressed) |
TB920488 is the observation that already bites (A): HOXA-low, IRX1-zero at adequate depth, positive controls intact. Under (A) its oncoprotein must be at a third address nobody has named.
The evidence currently favouring each
For (A): the oncoprotein really is at IRX1 with ENL and H3K4me3 in p318 and p454 2; the anti-correlation across seven patients is significant (Spearman -0.714, exact p = 0.029) 2; and IRX1 is focal rather than cluster-wide, which looks like a targeted address rather than a regional state change.
For (B): Kühn's co-IP and ChIP show IRX1 binding the MLL-AF4 complex and repressing HOXA, and IRX1 binds HOXA promoters only when MLL-AF4 is present 1; p279 is double-positive, which (A) reads as a mixture and (B) reads naturally as an intermediate; and the whole clinical rationale for the subgroup — a stem-cell, quiescence-associated relapse phenotype — is (B)'s story 1.
Neither explains why IRX1 is on at all. Both mechanisms take IRX1 expression as given.
| statement | In the HOXA-negative half of infant t(4;11) ALL the KMT2A::AFF1 oncoprotein occupies IRX1 as a genuine alternative target, rather than IRX1 merely being an upstream repressor of the HOXA program. |
|---|---|
| origin | data-pattern |
| prior probability | 0.35 |
predictions
[ "IRX1 CUT&RUN in a primary iALL-IRX1 blast will NOT show IRX1 protein at HOXA9/HOXA10 (mechanism B predicts it will).", "Menin CUT&RUN in an IRX1-high KMT2A::AFF1 blast will show menin co-occupying IRX1 with the oncoprotein.", "Degrading or inhibiting the fusion in an IRX1-high blast will reduce IRX1 transcription (B predicts little effect on IRX1).", "Single-cell CUT&Tag on p279 will show HOXA-occupied and IRX1-occupied nuclei as separate populations if the double-positive is a mixture." ]
experiments
[ "exp-menin-cut-and-run-in-an-irx1-high-infant-t411-blast", "exp-irx1-cut-and-run-in-primary-t411-blasts", "exp-single-cell-cut-and-tag-on-the-double-positive-blast" ]
analyses
[ "ana-q145-irx1-is-fusion-occupied-exactly-where-hoxa-is-not" ]
falsifiers
[ "IRX1 CUT&RUN in a primary iALL-IRX1 blast shows IRX1 protein enriched >=2-fold over IgG at the HOXA9 and HOXA10 promoters while the oncoprotein at IRX1 is below 1.5x mean promoter. That is mechanism (B) with (A) absent, and this hypothesis is refuted.", "A survey of >=10 HOXA-negative KMT2A::AFF1 patients finds fewer than half with oncoprotein at IRX1 above 2x mean promoter. TB920488 is already one such patient; a majority would make IRX1 a minority route rather than the second address.", "Fusion degradation in an IRX1-high blast leaves IRX1 transcription unchanged at 48 h while HOXA-subgroup controls collapse - IRX1 is then upstream of the fusion, not downstream of it.", "Single-cell CUT&Tag on p279 shows HOXA and IRX1 occupancy in the SAME nuclei rather than in separate populations, which does not refute (A) but removes the mixture reading that currently rescues it." ]
- Under (A) a menin inhibitor should still work in the IRX1 subgroup but its pharmacodynamic biomarker (HOXA9/MEIS1 collapse) is unavailable, so an effective drug could be abandoned for lack of a readout.
- Under (B) the IRX1 subgroup has a partly self-suppressed fusion and a stem-cell/quiescence program running on top, so fusion-directed therapy addresses the less important half of the biology.
- Both mechanisms predict that IRX1-high blasts survive HOXA-directed differentiation therapy, which is the shared and actionable part.
- Where is TB920488's oncoprotein? It is HOXA-low and IRX1-negative with intact MEIS1/CDK6/PAX5 occupancy, so a third programme exists and has no name.
- Does the fusion reach IRX1 through the KMT2A CXXC CpG-binding address that this corpus argues about elsewhere? The IRX1 promoter's CpG content was not measured here.
- Is IRX1 expression itself fusion-driven, or is it a property of the cell of origin that the fusion arrived into? Nothing in the literature or this corpus answers it.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
IRX1 binds the MLL-AF4 complex non-competitively, occupies HOXA9/HOXA10 promoters only when MLL-AF4 is present, and dominant-negatively represses HOXA transcription while activating HOXB4 and EGR1-3.
The IRX1/HOXA connection: insights into a novel t(4;11)-specific cancer mechanism · Kuhn A, Loscher D, Marschalek R · Oncotarget 2016What this evidence cannot show
HEK293T with transfected inducible transgenes; one SEM transgene line with Q-PCR only, no ChIP. Overexpression levels need not resemble an iALL-IRX1 blast. Full text fetched and read this cycle.
Evidence that argues against this (1)
- If the dominant-negative effect is a dose artefact of transgene overexpression, mechanism (B) loses its only direct support and (A) becomes the default. Symeonidou V et al. 2021
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Across seven KMT2A::AFF1 patients, oncoprotein occupancy at IRX1 is anti-correlated with occupancy at HOXA9/HOXA10 (Spearman -0.714, exact permutation p = 0.0288), with IRX1 carrying ENL and H3K4me3 in the HOXA-dead states.
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 = 7 patients, one deposit, no RNA-seq, no menin track. Occupancy is symmetric in time and cannot establish direction of causation, which is precisely what separates the two mechanisms. One patient (TB920488) is HOXA-low and IRX1-negative and is a counter-example to the simplest form of (A). 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
- depends on →IRX1 — Iroquois homeobox 1, the gene that names the HOXA-negative half of infant t(4;11) ALL
the hypothesis is about IRX1 as a fusion target
- depends on →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
a second address for the KMT2A::AFF1 oncoprotein
the measurement that generated it
- refutes →IRX1 binds the MLL-AF4 complex non-competitively — menin, LEDGF and DOT1L are all unchanged
Kuhn's direction is the competing mechanism this hypothesis must beat
under this hypothesis the menin interface should still operate, at a different address
proposes the mechanism of the state
whether menin accompanies the oncoprotein to IRX1
the cleanest discriminator between the two mechanisms
removes or confirms the mixture reading that currently rescues the hypothesis