Where t(9;11) actually breaks inside MLLT3: 98.4% downstream of exon 4, and the two patients in 442 who kept the reader
also called AF9 breakpoint distribution · MLLT3 breakpoints · t(9;11) breakpoints
In 442 patients with this translocation, the chromosome broke past AF9's fourth exon in all but a handful, throwing the reader module away every time. Exactly two patients kept it, and nothing at all is known about those two.
The same thing, for a scientist
Verified verbatim from the PMC body of PMID 36435883 on 2026-09-20 after `fullTextXML` returned HTTP 500: *'In contrast to ENL, AF9 genomic breaks occur almost exclusively downstream of exon 4 (98.4%, n=442) and exclude the AF9 YEATS domain from the resultant MLL-AF9 fusion proteins... Only two MLL-AF9 patients (0.5%) had breakpoints upstream of MLLT3 exon 1, resulting in YEATS domain inclusion in the MLL-AF9 fusion product'* [E1]. Three things follow that are worth separating. **First**, the figure is a *genomic* breakpoint distribution mined from the Frankfurt DCAL recombinome archive, not a protein measurement [E1][E2]. **Second**, the two YEATS-retaining patients are not the mirror image of the YEATS-losing majority: they broke **upstream of MLLT3 exon 1**, which is the *spliced-fusion* architecture, where a truncated KMT2A lands 5-prime of an intact partner gene and no reciprocal allele is expressed [E3]. So the only two t(9;11) patients known to carry an AF9 YEATS domain are also the only two known to be reciprocal-negative, and the two properties cannot be separated in a cohort of two. **Third**, the denominators differ between the two published statements of this fact: 98.4% of 442 is 435 patients, 0.5% of 442 is 2, and the residual ~1.1% (about 5 patients) is unaccounted for in the sentence. The review from the same laboratory states the same distribution in residue terms - approximately 92% of MLL-AF9 patients break at or after AF9 residue 376 [E2] - and that is a materially looser bound.
The exact sentence, and what it does and does not say
The AF9 analysis is Supplementary Figure S1 of a paper about ENL 1. That matters for how much weight it can carry: it is a control comparison, not the study's subject, and its underlying data are a re-analysis of a clinical archive published elsewhere 23.
What it establishes: in the archive, essentially every t(9;11) breakpoint falls in a position that excludes MLLT3 exons 1-4 from the fusion transcript.
What it does not establish: that the corresponding fusion protein lacks the YEATS fold in those patients. No patient's MLL-AF9 protein is blotted anywhere in this literature, for the sufficient reason that no junction-specific antibody exists for any KMT2A fusion. The inference from genomic breakpoint to protein architecture runs through an assumed reading frame and an assumed splice, and it is an inference this whole beat depends on.
The two patients, and why they are not a clean natural experiment
The obvious question this beat was sent to ask is: what do the 0.5% look like? The answer in the primary text is one clause - they 'had breakpoints upstream of MLLT3 exon 1' - and nothing else. No age, no lineage, no outcome, no karyotype, no follow-up 1.
The structural point is that breaking upstream of exon 1 of the partner gene is the definition of a spliced fusion in the recombinome nomenclature: the truncated KMT2A gene is recombined 5-prime of an intact partner, transcription reads through, and the partner contributes from its own exon 1 3. The recombinome reports this architecture in 2 of 612 KMT2A::MLLT3 cases 3, which is the same order as the two YEATS-retaining patients and very probably the same two.
That has a consequence for how these two can ever be used. A spliced fusion produces no reciprocal allele 3. So in the only two t(9;11) patients who retain the YEATS domain, YEATS retention and reciprocal-negative status are perfectly confounded. Anyone proposing to look these patients up as the human version of a YEATS add-back experiment is proposing a comparison in which two variables move together in a sample of two.
The KMT2A side of the same break
On the KMT2A side, t(9;11) sits in the major breakpoint cluster region (KMT2A intron 9 to intron 11), like the other common partners 3. One partner-specific observation is worth recording because it is a confound rather than a curiosity: in the recombinome cohort the shift of KMT2A breakpoints towards intron 11 in adult MLLT3 patients was 'clearly linked to therapy-induced leukemia', and the same shift was not seen for ELL, EPS15 or SEPTIN6 3.
So an adult t(9;11) breakpoint distribution is partly a distribution over aetiology, and a study stratifying adult MLL-AF9 patients by KMT2A intron will be partly stratifying them by whether they had prior topoisomerase-II-targeting chemotherapy.
What would settle the YEATS question in humans, and why the ENL side is the better place to look
The t(9;11) arm cannot answer it: n=2, confounded.
The t(11;19) arm can. In the same cohort, 15.9% of 302 MLL-ENL patients break downstream of ENL exon 2, giving 7.3% partial YEATS inclusion (n=22) and 8.6% no YEATS inclusion (n=26) 1. That is 26 patients whose MLL-ENL fusion is architecturally MLL-AF9-like, against 254 who are not, in one archive, with clinical annotation already attached. The comparison this beat exists to make - does an AHD-only fusion behave differently from a YEATS-plus-AHD fusion in a human being - has a denominator of 26, not 2, and it is on the ENL side of the pair.
The existing experiment entry exp-yeats-cis-resupply-kmt2a-mllt3 already proposes that arm. This entry's contribution is the reason to prefer it: the AF9 arm of the natural experiment does not exist.
| fusion | fusion-kmt2a-mllt3 |
|---|---|
| partner | gene-mllt3 |
| translocation | t(9;11)(p21.3;q23.3) |
| cohort n | 442 |
| breaks downstream of partner exon4 PCT | 98.4 |
| breaks upstream of partner exon1 n | 2 |
| breaks upstream of partner exon1 PCT | 0.5 |
| unaccounted PCT | 1.1 |
| yeats retained in fusion PCT | 0.5 |
| partner exon boundary | 4 |
| partner residue boundary approx | 376 |
| kmt2a side | major breakpoint cluster region, KMT2A intron 9 to intron 11 |
| kmt2a intron11 shift in adults | linked to therapy-induced leukaemia in the recombinome cohort |
| spliced fusion cases | 2 |
| spliced fusion denominator | 612 |
| spliced fusion PCT | 0.33 |
| what is known about the two yeats retaining patients | nothing beyond breakpoint position; no age, lineage, karyotype, treatment or outcome is reported |
| protein level confirmation | false |
| junction specific antibody exists | false |
paralogue comparison — the numbers behind the chart
{
"MLLT1_spliced_n": 241,
"MLLT1_denominator": 460,
"MLLT1_yeats_retained_pct": 84.1,
"MLLT1_yeats_absent_n": 26,
"MLLT1_yeats_partial_n": 22,
"MLLT1_cohort_n": 302
}confounded variables in those two
[ "YEATS retention", "reciprocal-negative (spliced) architecture" ]
- If a YEATS-domain inhibitor (SGC-iMLLT and successors) is ever positioned for KMT2A-rearranged disease, this distribution says it has no fusion target in 99.5% of t(9;11) patients; it would be acting on wild-type AF9 and ENL only.
- Stratifying a trial by 'YEATS-retaining KMT2A fusion' would enrol essentially no t(9;11) patients, so a partner-agnostic eligibility criterion and a YEATS-based one select almost disjoint populations.
- Who are the two YEATS-retaining t(9;11) patients? Age, lineage, karyotype and outcome exist in the source archive and have never been published.
- What is in the unaccounted ~1.1% of the 442 - breakpoints inside MLLT3 exons 1-4 giving partial YEATS inclusion, as happens for 7.3% of ENL patients?
- Is the 2-of-612 spliced count and the 2-of-442 YEATS-retaining count the same two patients? The two numbers come from the same archive and nobody has stated whether they are the same cases.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
AF9 genomic breaks in t(9;11) occur almost exclusively downstream of MLLT3 exon 4 (98.4%, n=442), excluding the AF9 YEATS domain from the fusion protein, and only two MLL-AF9 patients (0.5%) had breakpoints upstream of MLLT3 exon 1 and so retained the YEATS domain; in the same work 84.1% (254/302) of t(11;19) patients retain the full ENL YEATS domain, 7.3% (22/302) retain it partially and 8.6% (26/302) lose it entirely.
The ENL YEATS epigenetic reader domain critically links MLL-ENL to leukemic stem cell frequency in t(11;19) Leukemia. · Hu H, Saha N, Yang Y, et al. · Leukemia 2023What this evidence cannot show
Read verbatim from the PMC HTML body (PMC11246743) on 2026-09-20 after Europe PMC `fullTextXML` returned HTTP 500. The AF9 data are a supplementary figure in a paper about ENL and are a re-analysis of a previously published clinical archive. The percentages describe genomic breakpoints; no fusion protein was measured. The stated 98.4% and 0.5% do not sum to 100% of 442 and the residual ~1.1% is not described.
Evidence that argues against this (1)
- The same laboratory's review states the boundary in residue terms as approximately 92% of MLL-AF9 patients breaking at or after AF9 residue 376 - a looser bound than 98.4% downstream of exon 4, and the one on which every residue-level retained/lost claim in this graph actually rests. Hu H et al. 2023
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The clinical breakpoint archive mined for these distributions is the Frankfurt DCAL MLL recombinome series, which determined genomic breakpoints within MLL and the involved translocation partner genes in 2,345 acute leukaemia patients and reported an age-dependent breakpoint shift within MLL.
The MLL recombinome of acute leukemias in 2017. · Meyer C, Burmeister T, Groger D, et al. · Leukemia 2018What this evidence cannot show
A diagnostic referral archive, not a population sample: cases reach it because a centre sent material for breakpoint characterisation, so partner and age composition reflect referral practice. The archive characterises the genomic breakpoint, not the expressed protein.
Evidence that argues against this (1)
- Referral-based ascertainment can bias a partner-specific breakpoint distribution if different partners are referred for different reasons - for example if a partner is preferentially characterised when standard RT-PCR fails. The recombinome papers do not report the referral indication per case, so this cannot be excluded from within the data. Meyer C et al. 2023
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In 3,401 KMT2A-rearranged patients, spliced fusions - where a truncated KMT2A is recombined upstream of an intact partner gene so that only the direct fusion transcript and no reciprocal one is expressed - occur in about 50% of MLLT1 cases (241 of 460) but in only 2 of 612 MLLT3 cases; and the shift of KMT2A breakpoints towards intron 11 in adult MLLT3 patients was clearly linked to therapy-induced leukaemia.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023What this evidence cannot show
Read in full text from the PMC HTML page (PMC10169636) on 2026-09-20. Spliced-fusion status is inferred from the genomic breakpoint position relative to the partner's first exon, not from an assay showing that no reciprocal transcript is made in that patient. The 2-of-612 figure is a count in a referral archive.
Evidence that argues against this (1)
- This graph's `concept-reciprocal-product-of-t-9-11-mllt3-kmt2a` records that reciprocal AF9-MLL expression frequency in patients is 'not reported', so the inference that 610 of 612 t(9;11) patients express a reciprocal product is an inference from genomic architecture and not a measurement of transcript. Meyer C et al. 2023
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ENL, but not its paralogue AF9, is required for disease maintenance in acute myeloid leukaemia, so the two YEATS proteins differ in wild-type requirement as well as in fusion architecture.
ENL links histone acetylation to oncogenic gene expression in acute myeloid leukaemia. · Wan L, Wen H, Li Y, et al. · Nature 2017What this evidence cannot show
CRISPR depletion in AML cell lines and mouse models; a negative result for AF9 in a screen is a statement about those lines and that guide set, not a proof of dispensability everywhere.
Evidence that argues against this (1)
- MLLT3 is required for human haematopoietic stem cell self-renewal and engraftment, so 'AF9 is dispensable' is a statement about leukaemic maintenance in established lines and is false for normal HSCs - which matters directly if anyone proposes targeting AF9 rather than the fusion. Calvanese V et al. 2019
breakpoint distribution within the partner gene
refines 92%/residue-376 to 98.4%/exon-4 in 442 patients
establishes that the AF9 arm of the human natural experiment has n=2 and is confounded
2 of 612 t(9;11) are spliced, so almost all are reciprocal-capable
the breakpoint that sets the architecture
verified the 98.4/442/exon-4 figures from the primary text
- ← related tot(9;11) is the reciprocal-POSITIVE arm: 2 of 612 patients are spliced, against 241 of 460 for t(11;19)
same two patients, possibly
- ← related toMLLT3 is the commonest partner in therapy-induced KMT2A leukaemia: 49 of 142, and 47 of those are t-AML
the KMT2A-side confound behind the partner-side distribution
0.51% predicted, 2 of 442 observed