Oncofusion

start / graph / finding-hoxb4-is-not-a-fusion-target-in-primary-infant-kmt2a-aff1

Analysismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q145-irx1-in-hoxa-negative-infants

HOXB4, the one direct IRX1 target with ChIP evidence, carries no oncoprotein in any of seven infant patients

IRX1's best-established direct target gene is HOXB4. In seven babies' leukaemias the cancer-driving protein was not on HOXB4 at all — consistent with what the original paper said, that HOXB4 belongs to IRX1 and not to the fusion.

The same thing, for a scientist

Kühn identifies **HOXB4 as a direct IRX1 target** — the only one confirmed by ChIP enrichment in that paper, with 3.5-fold transcriptional activation on IRX1 induction and activation of the downstream HOXB4 targets — and states explicitly that **HOXB4 is not a target gene of MLL-AF4** [E1]. Measured across all seven KMT2A::AFF1 patients, KMT2A-N occupancy at the HOXB4 promoter is **0.00-0.93x** mean promoter, never clearing 1.0 in any patient with either antibody [E2]. This is a small, clean, independent corroboration of a published negative in primary patient material, and it is one of the few places in this beat where my measurement and the literature test the same proposition.

Go deeper
Why a confirmed negative is worth an entry

The two mechanisms on offer for iALL-IRX1 make different predictions about HOXB4. If the fusion simply has a second address at IRX1 and drives an IRX1 program, one might expect the fusion also to appear at the IRX1-activated genes. It does not: HOXB4 stays cold even in the three patients where IRX1 is strongly occupied (p279 0.01/0.32, p318 0.00/0.04, p454 0.93/0.57). So whatever activates HOXB4 in an iALL-IRX1 blast, the KMT2A::AFF1 oncoprotein is not sitting on it — exactly Kühn's model, in which IRX1 protein, not the fusion, is the factor at HOXB4.

This is a prediction-matching negative, not an absence of data: all seven patients have adequate HOXB4 coverage in at least one library, and the positive controls MEIS1/CDK6/PAX5 read 5.4-40.4x in the same libraries.

What it does not show

It does not show that HOXB4 is inactive — no H3K4me3 claim is made, because HOXB4 activity was not the question and the activity tracks exist for only three patients. It does not show IRX1 protein is at HOXB4 in these patients, because there is no IRX1 CUT&RUN in this deposit or, as far as this beat could find, in any primary t(4;11) material at all. That experiment is registered as exp-irx1-cut-and-run-in-primary-t411-blasts.

Facts and figures
locusHOXB4 promoter, chr17:48,574,350-48,582,350 (TSS +/- 4 kb, hg38, minus strand)
positive controls same librariesMEIS1/CDK6/PAX5 at 5.4-40.4x
verdictnot a KMT2A::AFF1 oncoprotein target in primary infant material, concordant with Kuhn
kmt2an range all patients
[
  0,
  0.93
]
in irx1 positive patients
{
  "p279": [
    0.01,
    0.32
  ],
  "p318": [
    0,
    0.04
  ],
  "p454": [
    0.93,
    0.57
  ]
}
How this could bite us back
  • If HOXB4 drives a stem-cell maintenance program in these blasts independently of the fusion, then targeting the fusion or its menin interface leaves that program running - which is Kuhn's own explanation for why IRX1 patients relapse.
Still open
  1. Is HOXB4 transcribed in iALL-IRX1 patient blasts? Kuhn predicts yes via IRX1; no patient expression data in this deposit can say.
  2. Does IRX1 protein occupy HOXB4 in a primary t(4;11) blast, as it does in HEK293T? No IRX1 CUT&RUN exists in primary material.
Every source behind this entry (2)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    HOXB4 is a direct IRX1 target confirmed by ChIP enrichment at its promoter, is activated 3.5-fold on IRX1 induction along with its downstream targets, and is not a target gene of MLL-AF4.

    The IRX1/HOXA connection: insights into a novel t(4;11)-specific cancer mechanism · Kuhn A, Loscher D, Marschalek R · Oncotarget 2016
    PMID 27175594DOI
    What this evidence cannot show

    HEK293T with inducible transgenes; the ChIP is IRX1 antibody (Abcam ab72642) at the HOXB4 promoter with an intron-1 internal control. The 'not a target of MLL-AF4' statement is asserted in the paper's figure legend; MLL-AF4 ChIP at HOXB4 is not shown separately. Full text fetched and read this cycle.

    Evidence that argues against this (1)
    • If HOXB4 were a weak MLL-AF4 target below the paper's detection, the stated negative would be an assay-sensitivity claim rather than a biological one, and my own low-but-nonzero p454 value (0.93x) is not inconsistent with that. Kuhn A et al. 2016
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    KMT2A-N occupancy at the HOXB4 promoter is 0.00-0.93x mean promoter across all seven KMT2A::AFF1 patients and both N antibodies, including the three patients where IRX1 is strongly occupied.

    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 2026
    What this evidence cannot show

    A negative in a promoter-scale window. HOXB4 sits in the HOXB cluster on chr17 where neighbouring promoters were not separately profiled, so a shifted regulatory element would be missed. 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. A second cohort profiled by a different group could reverse any of this. Janssens DH et al. 2024
Connected to