Oncofusion

start / graph / exp-q147-swap-the-aff1-c-terminus-onto-kmt2a-af9-and-score-irx1-occupancy

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q147-hoxa-irx1-across-partners

Swap the AFF1 C-terminus onto a KMT2A-AF9 backbone and score IRX1 occupancy: the direct test of partner causality

Every KMT2A fusion keeps the same front half of the protein; only the back half differs. Building a fusion with an AF9 back half and an AFF1 back half in the same cells, and asking which one puts the protein on IRX1, tests the partner explanation directly.

The same thing, for a scientist

**The logic.** The KMT2A N-terminal fragment retained in all six fusions is identical, and it carries the menin interface, the LEDGF interface and the CXXC domain. **An address present only in KMT2A::AFF1 therefore cannot be explained by the menin interface alone** - it must come from the partner C-terminus or from something it recruits [E1]. That is a causal claim and it is testable by construction rather than by cohort. **Design.** In one cell background - a KMT2A-r line or human CD34+ HSPC - express, from a single-copy dox-inducible locus at matched protein level: 1. KMT2A-AFF1 (positive arm), 2. KMT2A-AF9, 3. KMT2A-AF9 with the AFF1 C-terminus in place of AF9, 4. KMT2A-AFF1 with the AF9 C-terminus in place of AFF1, 5. KMT2A N-terminal fragment alone (no partner), 6. an AFF1 construct with the pSer/AF4 C-terminal homology domain deleted. **Match the protein level and verify it** - the deposit paper's central finding is that oncoprotein LEVEL redirects the programme, so an unmatched-level comparison measures level and not partner [E2]. This is the single most likely way for the experiment to fail silently. **Readout.** KMT2A-N1/N2 CUT&RUN, spike-in calibrated, at the IRX1 gene body and the HOXA9-HOXA10 span, with MEIS1/CDK6 as positives and the eight-locus width-matched panel as the floor. **Pre-declared decision rule.** IRX1 occupancy appearing with construct 3 and disappearing with construct 4 => **the AFF1 C-terminus is sufficient and necessary for the IRX1 address**. IRX1 occupancy in none of them at matched level => the address requires something the cell background supplies (cell state), and the cohort confound in GSE252378 is the real story.

Go deeper
The honest weakness of this design

A chimaera is not a patient. An ectopic fusion in a cell line background has repeatedly failed to reproduce primary-sample chromatin behaviour in this disease, and the cell-of-origin dependence of KMT2A-r biology is one of its best-established features. So a negative result in constructs is weak evidence about patients, while a positive result - IRX1 occupancy following the AFF1 C-terminus - is strong, because it would demonstrate sufficiency.

The asymmetry should be stated in the grant, not discovered at the analysis stage. If the constructs are all IRX1-negative the experiment has not refuted the partner hypothesis; it has failed to test it.

Facts and figures
backgroundsingle KMT2A-r line or human CD34+ HSPC, single-copy dox-inducible
critical requirementmatched and verified oncoprotein protein level across constructs
readoutspike-in-calibrated KMT2A-N1/N2 CUT&RUN at IRX1 gene body and HOXA9-HOXA10 span
decision ruleIRX1 occupancy appearing with the AFF1 C-terminus and disappearing without it demonstrates sufficiency and necessity of the partner
evidence asymmetrya positive is strong, a negative is weak - construct systems have repeatedly failed to reproduce primary KMT2A-r chromatin behaviour
constructs
[
  "KMT2A-AFF1",
  "KMT2A-AF9",
  "KMT2A-AF9 with AFF1 C-terminus",
  "KMT2A-AFF1 with AF9 C-terminus",
  "KMT2A N-terminal fragment alone",
  "KMT2A-AFF1 with the AF4 C-terminal homology domain deleted"
]
tests hypotheses
[
  "hyp-q147-the-irx1-address-requires-the-aff1-partner-not-the-menin-interface"
]
analyses
[
  "ana-q147-irx1-fusion-occupancy-is-confined-to-kmt2a-aff1"
]
How this could bite us back
  • If the AFF1 C-terminus is what creates the IRX1 address, then the therapeutic handle is the AFF1 side of the fusion and not menin - which is a much harder target, and it would mean the IRX1 subgroup is precisely the group a menin inhibitor helps least.
  • A construct system that reproduces the IRX1 address would also be the first tractable model of it, since SEM has no IRX1 signal. That is the upside and it is worth the weak-negative risk.
Still open
  1. Which AFF1 region? The AF4 C-terminal homology domain, the pSer region and the ALF domain are the candidates, and a domain-resolution version of this experiment is the follow-on.
  2. Would the swap have to be done in an infant-derived background? If the IRX1 address needs a fetal or infant pro-B chromatin state, an adult CD34+ background will be negative for a reason unrelated to the partner.
Every source behind this entry (2)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    All six KMT2A fusions measured in GSE252378 retain the same KMT2A N-terminal fragment, recognised by the same two N-terminal antibodies, and differ only in the C-terminal partner; IRX1 occupancy occurs in 3 of 8 KMT2A::AFF1 samples and 0 of 18 samples carrying the other five partners.

    What this evidence cannot show

    The inference from 'only AFF1 shows it' to 'the AFF1 C-terminus causes it' is exactly the step this experiment exists to test rather than assume, because partner, institution and cell state are confounded in the cohort. Every number is a WITHIN-SAMPLE share: an interval's mean coverage divided by that same library's mean coverage over all 19,404 hg38 RefSeq promoters, so 1.0 = an average promoter in that library. GSE252378 has no spike-in, so no between-library 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)
    • Partner identity is also correlated with lineage and with age at diagnosis in KMT2A-r disease - AFF1 is predominantly infant B-ALL and AF9/AFDN predominantly AML - so a partner difference in a cohort is never cleanly a partner difference. Janssens DH et al. 2024
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The deposit's own paper reports that at high expression levels the oncoproteins preferentially activate either a pro-B-cell-enriched ALL programme or a haematopoietic-stem-cell-enriched AML programme, that the fusion-partner-specific binding patterns over these gene sets correlate with each mutation's prevalence in ALL versus AML, and that at reduced levels the oncoproteins redirect to GMP genes.

    MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nat Commun 15:9341 2024
    PMID 39472576DOI
    What this evidence cannot show

    This is the paper's central claim and it cuts both ways for this experiment: it supports the existence of genuine partner-specific binding patterns, and it establishes that oncoprotein LEVEL alone redirects the programme - so any construct comparison at unmatched level is uninterpretable.

    Evidence that argues against this (1)
    • If partner-specific binding patterns are themselves downstream of the cell type in which each partner's disease arises, then matching the cell background across constructs - which this design does deliberately - removes the very variable that produces the difference in patients. Janssens DH et al. 2024
Connected to