Oncofusion

start / graph / concept-reciprocal-product-of-t-9-11-mllt3-kmt2a

Explainermoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q24-reciprocal-product

t(9;11): the reciprocal product is made, is engineered into every genome-editing model, and has never been separately tested

also called AF9-MLL · MLLT3-KMT2A reciprocal · der(9) product

The second most common version of this leukaemia also makes a mirror-image fusion. When scientists build the translocation from scratch in human stem cells, they build both halves - and then attribute everything that happens to the famous half.

The same thing, for a scientist

TALEN-mediated genome editing generated t(9;11) chromosomal translocations encoding MLL-AF9 **and reciprocal AF9-MLL fusion products** in CD34+ human cord blood cells; the translocated clones showed elevated MLL target gene expression, high surface CD9, increased colony-forming ability, and produced myeloid leukaemias in immunocompromised mice within 48 weeks with no secondary pathogenic mutations detectable by targeted exome and whole-genome RNA sequencing, which the authors read as demonstrating the genetic sufficiency of the t(9;11) translocation for leukaemia development from human HSPCs [E1]. Read carefully, that is a sufficiency statement about **the translocation**, not about MLL-AF9: both products were present throughout. The same logic applies to the CRISPR-engineered t(4;11) model, where both reciprocal fusion proteins are concomitantly present [E2]. So the most convincing human models of KMT2A-rearranged leukaemia in existence are double-fusion models whose results are routinely attributed to the single direct fusion. This is a live attribution problem, not a hypothetical one, and it is cheap to fix: the editing platform can delete or silence the der(9) allele after transformation, and nobody has reported doing so.

Go deeper
The attribution gap in one sentence

An engineered balanced translocation is the gold-standard human model precisely because it is physiological - one event, native loci, native regulation - and that is exactly why it cannot attribute the phenotype to one of the two products it creates.

What is known about the t(9;11) reciprocal specifically

Very little. Searching 'AF9-MLL' returns 29 records, of which most are multiplex RT-PCR panels and unrelated MLL-AF9 biology; the retired-symbol searches for this partner ('MLLT3-KMT2A' returns 2, 'AF10-MLL' returns 0) show how thin the dedicated literature is. There is no frequency figure for reciprocal expression in t(9;11) comparable to the 45% in t(4;11), no functional study of an AF9-MLL protein, and no knockdown. The one mechanistic anchor is architectural: the KMT2A portion of the reciprocal product is the same in every major-cluster translocation, so the PHD/bromodomain cassette and the SET domain are present here too 3.

Facts and figures
translocationt(9;11)(p21-22;q23)
partnerMLLT3 (AF9)
reciprocal productAF9-MLL
present in engineered modelstrue
reciprocal expression frequency in patientsnot reported
dedicated functional studies0
attribution problemthe best human models of KMT2A leukaemia are balanced translocations carrying both products, and results are attributed to the direct one
pubmed counts 2026 09 19
AF10-MLL0.000AF10-MLL: 0MLLT3-KMT2A2.000MLLT3-KMT2A: 2AF9-MLL29.0AF9-MLL: 29
pubmed counts 2026 09 19 — the numbers behind the chart
{
  "AF9-MLL": 29,
  "MLLT3-KMT2A": 2,
  "AF10-MLL": 0
}
How this could bite us back
  • If engineered balanced-translocation models are the most physiological human systems and they cannot attribute phenotype to one product, then the strongest evidence for menin-dependence in human cells is systematically ambiguous.
  • A retroviral single-fusion model avoids the ambiguity but introduces expression-level and integration artefacts, so there is currently no system that is both physiological and attributable.
Still open
  1. What fraction of t(9;11) patients express the reciprocal AF9-MLL transcript? No figure exists.
  2. In the TALEN t(9;11) model, does deleting the der(9) allele after transformation change the leukaemia? The platform exists and the experiment has not been reported.
  3. Has an AF9-MLL protein ever been detected, in any system?
  4. Do any of the many MLL-AF9 mouse models carry a reciprocal product? Retroviral MLL-AF9 models do not, which makes them the cleanest single-product systems and worth contrasting explicitly with the engineered human ones.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    TALEN-mediated genome editing generated t(9;11) chromosomal translocations encoding MLL-AF9 and reciprocal AF9-MLL fusion products in CD34+ human cord blood cells; translocated clones showed elevated MLL target gene expression, high surface CD9 and increased colony-forming ability, and transplantation induced myeloid leukaemias within 48 weeks with no secondary pathogenic mutations by targeted exome and whole-genome RNA sequencing, which the authors interpret as genetic sufficiency of the t(9;11) translocation for leukaemia development from human HSPCs.

    PMID 29650777DOI
    What this evidence cannot show

    Translocation efficiency was very low and clones were selected by cytokine-driven monoclonal outgrowth and immortalisation, so the leukaemogenic clones are a selected population; the sufficiency claim is for the translocation as a whole and the paper does not separate the contributions of the two products.

    Evidence that argues against this (1)
    • A t(4;11) study that removed the reciprocal transcript from established human lines found no phenotype, so presence of a reciprocal product in a model does not establish that it contributes to the phenotype. Kumar AR et al. 2011
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CRISPR/Cas9-mediated t(4;11) translocation in primary human haematopoietic stem and precursor cells likewise produces both reciprocal fusion proteins concomitantly, and the resulting cells show plasticity to differentiate into either myeloid or lymphoid lineage.

    PMID 41145673DOI
    What this evidence cannot show

    Establishes the same double-product confound in the index fusion's best human model; the paper does not attempt to separate the two products either.

    Evidence that argues against this (1)
    • An earlier human model expressing the reciprocal fusion alone in cord-blood CD34+ cells produced no leukaemia within 36 weeks, so the reciprocal product is not sufficient by itself in human cells even though it is present in the engineered models. Prieto C et al. 2017
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt Q03164 places PHD1-4, the bromodomain, FYRC, the WIN motif and the SET domain downstream of the major breakpoint cluster region, so these modules are carried by the reciprocal product of any major-cluster balanced KMT2A translocation regardless of partner.

    UniProt Q03164 (KMT2A_HUMAN), feature table · UniProt Consortium · UniProtKB 2026
    What this evidence cannot show

    Fetched 2026-09-19, primary gene name read back as KMT2A. The partner-independence of the KMT2A portion is an architectural inference; whether the reciprocal transcript is actually produced is partner-specific and geometry-dependent.

    Evidence that argues against this (1)
    • Partner geometry can abolish the reciprocal product entirely: about half of KMT2A::MLLT1 cases break upstream of MLLT1 exon 1 and produce no reciprocal transcript, so the KMT2A portion being constant does not mean a reciprocal protein exists. Meyer C et al. 2023
Connected to