The ENL YEATS acetyl-reader inside KMT2A::MLLT1: what it reads, what it does, and why both control arms carry it

The ENL YEATS domain binds histone H3 acetylated at K9, K18 and K27, with crystallographic and ChIP evidence that endogenous ENL co-localises with H3K27ac and H3K9ac at promoters of actively transcribed leukaemia-essential genes 3. In the fusion context the affinity was measured directly by biolayer interferometry against an immobilised biotinylated H3K27ac peptide: K_d = 66 uM for the wild-type domain, with F59A and Y78A abolishing the interaction 1. A fluorescein-labelled derivative of the chemical probe SGC-iMLLT binds wild-type ENL YEATS at K_d 0.26 uM and loses affinity against every mutant in the order delta21-26 (4.6 uM) > Y78A (8.82 uM) > F59A (>8 uM) > triple (>8 uM), so the probe and the histone occupy the same groove 1.

Note that this is a micromolar interaction with a single modified peptide. Avidity in a nucleosomal and multivalent context is not measured by it, and the number should not be read as the affinity of the fusion protein for chromatin.

Mice receiving MLL-ENL-transduced marrow died at a median of 66 days; mice receiving MLL-ENL-deltaYEATS (ENL 5-371 removed, AHD retained) failed to develop leukaemia at all, even though the construct still supported third-round serial replating ex vivo 1. The reader-dead point mutants did not phenocopy that: they extended latency, and the extension was transplantable into secondary recipients 1.

Extreme limiting dilution analysis on primary leukaemic cells put the difference on a stem-cell axis: 1/108 cells (95% CI 1/62-1/190) for MLL-ENL against 1/355 (95% CI 1/198-1/637) for the F59A/Y78A leukaemias, with no difference in bone-marrow homing of CD45.2 donor cells 1.

The mechanistic split is the useful part. FLAG-ChIP showed reduced binding of MLL-ENL-deltaYEATS at the Eya1 and Meis1 loci, and PAF1 occupancy mirrored the fusion, i.e. deletion costs occupancy. The point mutants held occupancy and lost output: H3K9ac, H3K79me2 and H3K4me3 fell in proportion to Eya1/Meis1 transcript 1. A Hoxa9-promoter luciferase reporter was unchanged by any of the mutants, so the defect is not a general loss of transactivation capacity 1.

ENL reaches PAF1 through the same YEATS fold. In the fusion, all four YEATS constructs tested (F59A, Y78A, delta21-26, triple) still co-immunoprecipitated endogenous and exogenous PAF1, which the authors read as non-overlapping, non-exclusive binding sites 1.

That sits against the prior study from which the delta21-26 deletion was taken, which reported H3 and PAF1 competing for ENL, showed that PAF1 recruitment determines the transforming potential of MLL-ENL, and proposed that the KMT2A moiety supplies a constitutive PAF1 tether letting the fusion escape the H3 competition 4. The two accounts are not reconcilable as stated, and the 2023 paper says so explicitly: its BLI finding that delta21-26 also disrupts H3K27ac binding contrasts with the earlier report, which used a different biochemical assay 1.

A 2026 chemical-genetic screen adds a third constraint: in-frame deletions of PAF1 confer resistance to ENL/AF9 YEATS inhibitors, and in both sensitive and resistant isogenic lines the inhibitors downregulate ENL target genes equally - so target-gene suppression is not sufficient to explain the anti-leukaemic effect 5. Any model in which YEATS inhibition works purely through the Eya1/Meis1 axis is under-determined.

Both arms of the clean control comparison - 153 patients breaking 5' of MLLT1 but 3' of ACER1, and 101 breaking inside MLLT1 intron 1 - deliver a partner contribution beginning at MLLT1 exon 2, and therefore both carry the complete YEATS domain 2. That is the reason the comparison is worth running: the reader module, which this entry shows is the single largest known determinant of MLL-ENL leukaemic stem-cell frequency, is held constant while reciprocal-allele status varies.

The corollary is a warning. The 15.9% of t(11;19) patients whose break falls downstream of MLLT1 exon 2 (partial or absent YEATS) are a third group, not a subset of either arm, and pooling them into either arm would mix the beat's variable of interest with the strongest known modifier of the phenotype 2. KOPN-8, the cell line most of this fusion's pharmacology runs in, sits in that third group.

The paper reports the non-retaining 15.9% two ways that do not agree. The results text says 'partial inclusion (6.0%) or exclusion of the YEATS domain (9.9%)'; the Figure 1C legend says partial n=22/302 (7.3%) and absent n=26/302 (8.6%) 2. Both pairs sum to 15.9% and the headline 84.1% (n=254/302) is unaffected, but the split between partial and absent differs by about four patients each way depending on which sentence is read. Anyone building a three-way stratification on this paper should use the figure-legend counts, which carry explicit denominators.