Oncofusion

start / graph / gene-irx1

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

IRX1 — Iroquois homeobox 1, the gene that names the HOXA-negative half of infant t(4;11) ALL

also called IRX-1 · Iroquois homeobox protein 1 · IRXA1

IRX1 is a developmental control gene that is normally switched on in the early embryo, not in blood. About half of babies with the commonest infant leukaemia switch it on instead of the HOXA genes everyone expected, and those babies do worse.

The same thing, for a scientist

🔴 **LINEAGE CAVEAT added cycle 146 (`ana-q158-partner-exclusivity-does-discriminate-and-the-one-rival-that-fired-is-a-lineage-artefact`).** The gene-body AFF1-exclusive class survives a lineage-matched test - 7 against a null of 1.66, p 0.0044, using only the 7 B-lineage AFF1 samples against the 16-sample B-lineage pool - **but IRX1 is not one of its members.** IRX1's three positive samples are `p279`, `p318` and `p454`, and **`p318` is AML**, so within B-lineage IRX1 has 2 of 7 and falls below the k>=3 recurrence bar. **On the most lineage-controlled version of this test IRX1 cannot appear.** Its exclusivity (0 of 18 non-AFF1) is unaffected and holds at every gate tested; what is not lineage-robust is its *recurrence*. Independently, IRX1's own permutation p is 0.0219 unstratified but 0.0772 (patient), 0.1190 (institution, exact) and 0.1756 (B-lineage). IRX1 (chr5:3,595,831-3,601,403, hg38, plus strand; NM_024337) is a TALE-class Iroquois homeobox transcription factor. In t(4;11)/KMT2A::AFF1 infant ALL it defines the *iALL-IRX1* subgroup: patients whose blasts lack the canonical HOXA signature overexpress IRX1 instead, and have a higher relapse rate and worse outcome [E1][E2]. Mechanistically IRX1 protein binds the MLL-AF4 complex at HOXA9/HOXA10 promoters and acts dominant-negatively over its activator function, while directly activating HOXB4 and indirectly EGR1-3 — a stem-cell maintenance and quiescence program [E2]. Whether IRX1 is expressed at all in normal haematopoiesis is genuinely disputed in the primary literature [E1][E3]. This beat adds the first measurement of **oncoprotein occupancy at the IRX1 locus itself** in primary patient material [E4].

Go deeper
Where it sits and what it is

hg38 chr5:3,595,831-3,601,403, plus strand, RefSeq NM_024337, a single NM_ transcript — coordinates read out of the UCSC refGene table, not typed 5. It is the telomeric member of the IRXA cluster on 5p15.33 (IRX4 → IRX2 → IRX1, spanning 1.72 Mb); the paralogous IRXB cluster (IRX3, IRX5, IRX6) is on 16q12.2.

The clinical claim, and exactly what kind of claim it is

Symeonidou & Ottersbach reanalysed two published RNA-seq cohorts (Andersson et al.; Agraz-Doblas et al.) with Kallisto → tximport → DESeq2 and found mutually exclusive expression of HOXA9/HOXA10 and IRX1, naming the groups iALL-HOXA9 and iALL-IRX1 1. iALL-IRX1 blasts are enriched for MYC targets, oxidative phosphorylation, G2M, E2F and MTORC1 — a more proliferative profile — and those patients are diagnosed younger and do worse 1.

This is an expression claim and nothing else. There is no ChIP, no CUT&RUN and no ATAC anywhere in that paper. The graph must not read it as an occupancy claim; see concept-an-expression-subgroup-is-not-an-occupancy-subgroup.

Is IRX1 expressed in normal blood at all? The literature disagrees with itself

Kühn et al. state flatly that "Irx1 and its human counterpart IRX1 are usually not expressed in hematopoietic tissues" 2. Symeonidou, interrogating single-cell RNA-seq of adult bone marrow and fetal liver, found IRX1 "exhibited very little expression in the hematopoietic system", and in mouse gastrulation datasets Irx1 is mesodermal while Hoxa9 and Meis1 sit in hematoendothelial progenitors 1. Nagel & Meyer report the opposite: IRX1 and MEIS1 expressed exclusively in pro-B-cells within normal B-cell development 3.

These cannot all be right. The disagreement is recorded, unresolved, in dispute-is-irx1-expressed-in-normal-b-cell-development. It matters because it decides whether iALL-IRX1 is an ectopic program (arguing a pre-haematopoietic cell of origin) or a retained pro-B program (arguing an ordinary B-progenitor origin).

What this beat measured that nobody had

Across seven KMT2A::AFF1 patients in GSE252378 scored from the deposited hg38 bigWigs, KMT2A-N (oncoprotein) share at the IRX1 promoter is anti-correlated with KMT2A-N share at HOXA9/HOXA10 (Spearman -0.571 at promoter scale, 4.20th percentile of all 19,404 promoters for anti-correlation with HOXA occupancy) 4. IRX1 is a fusion-occupied locus in the HOXA-dead patients and has zero reads in both N antibodies in the HOXA-high ones. Full result in ana-q145-irx1-is-fusion-occupied-exactly-where-hoxa-is-not. 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.

Facts and figures
symbolIRX1
chromchr5
start3595831
end3601403
strand+
buildhg38
refseqNM_024337
clusterIRXA (5p15.33): IRX4, IRX2, IRX1
paralogous clusterIRXB (16q12.2): IRX3, IRX5, IRX6
classTALE-class Iroquois homeobox transcription factor
subgroup definediALL-IRX1 (infant KMT2A::AFF1 ALL lacking the HOXA signature)
lineage matched class membership
{
  "in_lineage_matched_class": false,
  "reason": "p318 is AML; IRX1 has 2 of 7 B-lineage AFF1 samples, below k>=3",
  "lineage_matched_members": [
    "ELF1",
    "GNAQ",
    "H3C15",
    "HSPA2",
    "MIDN",
    "PAN3",
    "SPAG7"
  ],
  "class_survives_lineage_matching": "7 vs null 1.66, p_upper 0.0044",
  "irx1_exclusivity_unaffected": "0 of 18 non-AFF1, stable at every gate 1.25-3.0",
  "by_entry": "ana-q158-partner-exclusivity-does-discriminate-and-the-one-rival-that-fired-is-a-lineage-artefact"
}
How this could bite us back
  • If iALL-IRX1 blasts are HOXA-independent, the whole menin-inhibitor rationale — which is built on collapsing the HOXA9/MEIS1 program — may be weakest in exactly the subgroup with the worst prognosis.
  • Symeonidou shows PROM1 (CD133) is upregulated in iALL-HOXA9 and lower in iALL-IRX1, so CD133-directed CAR-T could systematically under-treat the higher-risk half.
  • Every preclinical KMT2A::AFF1 model in common use — SEM above all — is an iALL-HOXA9 model, so a drug optimised on them is optimised on the better-prognosis subgroup.
Still open
  1. Is IRX1 expressed in normal human pro-B cells or not? Kühn and Symeonidou say no, Nagel says exclusively yes, and no experiment in the corpus settles it.
  2. Does the fusion oncoprotein occupy IRX1 because IRX1 is being driven by it, or is the occupancy a consequence of IRX1 already being open in the cell of origin? Occupancy cannot distinguish these and no RNA-seq exists for these seven patients.
  3. Kühn's IRX1 ChIP is in HEK293T. Does IRX1 protein sit on HOXA9/HOXA10 in a primary iALL-IRX1 blast?
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Reanalysis of two published infant MLL-AF4 ALL RNA-seq cohorts finds mutually exclusive HOXA9/HOXA10 and IRX1 expression, defining iALL-HOXA9 and iALL-IRX1; iALL-IRX1 is more proliferative, diagnosed younger and has worse outcome.

    PMID 33069783DOI
    What this evidence cannot show

    Reanalysis of two existing datasets (n=22 and n=27), not a new cohort. Entirely transcriptional: no chromatin assay of any kind. The younger age at diagnosis did NOT reach statistical significance in the Andersson cohort, which the paper states explicitly. Full text fetched and read this cycle.

    Evidence that argues against this (1)
    • Nagel & Meyer's TALE-code assigns MEIS1 (not IRX1) to KMT2A fusions in BCP-ALL, and names IRX2 and IRX3 as the aberrantly expressed IRX genes in BCP-ALL — an IRX1/KMT2A association does not appear in it. Nagel S et al. 2022
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    IRX1 protein binds the MLL-AF4 complex at HOXA9 and HOXA10 promoters, represses HOXA transcription dominant-negatively, and directly activates HOXB4 while inducing EGR1-3.

    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

    The mechanism is established in **HEK293T** cells with doxycycline-inducible Sleeping Beauty transgenes, not in primary leukaemia. ChIP is HEK293T only; the SEM experiment is Q-PCR after IRX1 induction with no ChIP. IRX1 binding at HOXA promoters occurs ONLY when MLL-AF4 is co-expressed, so it is a property of the pair, not of IRX1. Full text fetched and read this cycle.

    Evidence that argues against this (1)
    • Kühn reports one HOXA gene that IRX1 failed to repress. The PMC full text has italicised gene symbols stripped, so WHICH gene that is could not be recovered from the text available to me, and I do not name it rather than guess. Kuhn A et al. 2016
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In normal B-cell development IRX1 and MEIS1 are expressed exclusively in pro-B-cells; the aberrant IRX genes in BCP-ALL are IRX2 (with TCF3 fusions) and IRX3 (with ETV6::RUNX1), and MEIS1 with KMT2A fusions.

    PMID 36233173DOI
    What this evidence cannot show

    Expression only, largely from cell lines used as models plus patient expression cohorts. **Directly contradicts E1 and E2 on whether IRX1 is expressed in normal haematopoiesis.** Notably it does NOT report an IRX1-KMT2A association, so it is not support for the IRX1 subgroup.

    Evidence that argues against this (1)
    • Kuhn states IRX1 is usually not expressed in haematopoietic tissues, and Symeonidou found very little IRX1 in adult marrow or fetal liver scRNA-seq - both directly contradict the exclusive pro-B expression claimed here. Symeonidou V et al. 2021
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Across seven KMT2A::AFF1 patients, oncoprotein (KMT2A-N) occupancy at the IRX1 promoter is anti-correlated with occupancy at HOXA9/HOXA10 (Spearman -0.571, 4.20th percentile of 19,404 promoters).

    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

    Seven patients, one deposit, one lab, one assay. No spike-in and no drug arm: all within-sample shares. p279 and p318 are ONE patient in two states. Three of the eight sample records (TB051939, TB920488, TB986292) are shallow enough that a zero is weak evidence (P(zero in both N antibodies by chance) = 0.14, 0.14, 0.26) and their IRX1 zeros are reported as weakly-measured. GSE252378 contains no RNA-seq, so no patient here can be assigned to the published EXPRESSION subgroup directly.

    Evidence that argues against this (1)
    • One patient (TB920488) is HOXA-low AND IRX1-negative at adequate depth, and one (p279) is positive for BOTH - so the anti-correlation is a population trend across patients and not a within-patient switch, which is a weaker claim than the published mutual exclusivity would suggest. Janssens DH et al. 2024
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IRX1 is at hg38 chr5:3,595,831-3,601,403, plus strand, NM_024337, with one NM_ transcript on a primary chromosome.

    What this evidence cannot show

    Read out of the downloaded refGene table this cycle and printed into analyses/q145-irx1-in-hoxa-negative-infants/coords_from_refgene.tsv. Not typed from memory.

    Evidence that argues against this (1)
    • refGene gives one canonical set of coordinates; a gene whose expressed isoform uses an alternative first exon would have a TSS that refGene does not list as the longest transcript. IRX1 has a single NM_ transcript, so this does not bite here, but the same procedure applied to a multi-transcript gene could place the promoter window wrongly. Janssens DH et al. 2024
Connected to