The KMT2A::MLLT3 chimeric protein: what survives the join on each side, in residue numbers - and the four AF9 modules that do not

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• A peptidomimetic or small molecule aimed at the AF9 AHD hits the same surface on wild-type AF9 and on wild-type ENL, both present in the leukaemic cell, and one of which (ENL) is required for AML maintenance. Fusion selectivity is not available from this surface.
• The AHD is intrinsically disordered and folds only on binding, so a compound optimised against one bound conformation may not engage the others; the primary laboratory reports its own point mutants are not partner-specific.
• Because the fusion carries no AF9 nuclear import signal, any strategy that works by mislocalising AF9 will move wild-type AF9 and leave the oncoprotein where it is.

1. Has AF9's predicted NLS at 295-300 ever been tested experimentally? UniProt annotates it by sequence analysis alone, and no publication in this beat's searches maps AF9 nuclear import to any residue.
2. Where exactly is the AHD N-terminal boundary? Three published constructs (495, 499, 502) differ by seven residues and no structure resolves that edge.
3. Is there any t(9;11) patient whose breakpoint falls between residues 300 and 376, retaining the AF9 NLS but not the oligomerisation run? The recombinome archive could answer this and nobody has asked.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProt Q03164 is KMT2A_HUMAN, primary gene name KMT2A, 3,969 aa, with the menin-binding motif at 6-25, IBM1 123-134, IBM2 147-152, A.T hooks 169-180/217-227/301-309 and CXXC 1147-1195; UniProt P42568 is AF9_HUMAN, primary gene name MLLT3, 568 aa, with a YEATS domain at 1-138, a low-complexity serine run at 149-190 and a nuclear localization signal at 295-300.

   UniProtKB entries Q03164 (KMT2A_HUMAN) and P42568 (AF9_HUMAN) · UniProt 2026
   ▸What this evidence cannot show

   Feature coordinates are curator annotations on the canonical isoform, not measurements in a leukaemic cell. UniProt's call at 149-190 is a compositional-bias (low-complexity) call, which is a statement that the stretch is unstructured rather than an assertion that it is the functional oligomerisation module; the functional assignment comes from the primary report cited in `ana-af9-regulatory-elements-lost-and-kept`. Both accessions annotate alternative isoforms, so a transcript-level claim needs isoform resolution this entry does not provide. Critically, the nuclear localization signal at 295-300 carries evidence code ECO:0000255 - sequence-analysis PREDICTION with no experimental support and no cited publication - so 'AF9's only annotated NLS' means annotated, not demonstrated.

   ▸Evidence that argues against this (1)

   • UniProt's single 'Menin-binding motif (MBM) 6-25' does not match this graph's `gene-kmt2a` entry, which splits the same region into MBM1 4-15 and MBM2 23-40; and UniProt gives CXXC as 1147-1195 where `gene-kmt2a` gives 1147-1242. Where the graph and UniProt disagree neither is automatically right, and the disagreement is a reason to name the source beside every coordinate. UniProt 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In 302 t(11;19) patients 84.1% retain the full ENL YEATS domain, whereas AF9 genomic breaks occur almost exclusively downstream of MLLT3 exon 4 (98.4%, n=442) and exclude the AF9 YEATS domain, with only two MLL-AF9 patients (0.5%) breaking upstream of MLLT3 exon 1 and retaining it.

   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 2023
   PMID 36435883DOI
   ▸What this evidence cannot show

   Read verbatim from the PMC HTML body (PMC11246743, 61,252 characters of stripped text) on 2026-09-20 after `fullTextXML` returned HTTP 500. The AF9 analysis is a supplementary figure (Fig. S1) in a paper whose subject is ENL, and the t(9;11) breakpoints are mined from a previously published clinical cohort rather than collected here. The percentages are of genomic breakpoints, not of expressed protein: no patient's fusion protein was blotted.

   ▸Evidence that argues against this (1)

   • The same laboratory's review states the boundary in residue rather than exon terms - approximately 92% of MLL-AF9 patients break at or after AF9 residue 376 - and this graph's existing `subtraction-mllt3-yeats` entry is built on that 92% figure. The two numbers are not in conflict but they are not the same measurement, and every residue-level claim in this beat (K339 lost, S395 retained, NLS lost) rests on the 92%/376 statement rather than on the 98.4%/exon-4 one. Hu H et al. 2023
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The AF9 region that mediates partner binding was defined as aa 495-568 by earlier biochemistry and used as aa 499-568 for structural work; in the DOT1L-AF9 NMR structure AF9 forms three helices and a beta hairpin around the partner peptide while its C-terminus (563-568) remains unstructured.

   Degree of recruitment of DOT1L to MLL-AF9 defines level of H3K79 Di- and tri-methylation on target genes and transformation potential. · Kuntimaddi A, Achille NJ, Thorpe J, et al. · Cell Rep 2015
   PMID 25921540DOI
   ▸What this evidence cannot show

   Read in full text from the PMC HTML page (PMC4426023) on 2026-09-20. The AHD boundary is an operational construct boundary chosen for NMR behaviour, not a domain boundary UniProt annotates - UniProt annotates no domain at all in this region of P42568, only disorder to residue 475. Any figure drawing 'AHD 499-568' is drawing a construct.

   ▸Evidence that argues against this (1)

   • This graph's `concept-mllt1-versus-mllt3-the-paralogue-pair-that-breaks-in-opposite-places` records the AF9 AHD as 502-568 and the ENL AHD as 494-559, from a different source. Three published boundaries for the same module (495, 499, 502) differ by up to seven residues, so a mutational claim near the AHD's N-terminal edge is boundary-dependent. Kabra A et al. 2022
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The AF9 ANC1 homology domain is intrinsically disordered and recruits multiple transcription factors through coupled folding and binding, retaining significant dynamic behaviour that may facilitate exchange between disordered partners.

   Leukemia fusion target AF9 is an intrinsically disordered transcriptional regulator that recruits multiple partners via coupled folding and binding. · Leach BI, Kuntimaddi A, Schmidt CR, et al. · Structure 2013
   PMID 23260655DOI
   ▸What this evidence cannot show

   Isolated AHD peptide and isolated partner peptides in vitro. Nothing in this work is done on a fusion protein, in a leukaemic cell, or at endogenous stoichiometry, and the AHD is reported to have a propensity to aggregate at the concentrations used.

   ▸Evidence that argues against this (1)

   • A dynamic hub that exchanges partners is also the profile of a protein whose apparent interactions are concentration-dependent artefacts of in vitro reconstitution. The same laboratory reports that AF9 point mutants designed against one partner are not completely specific for that partner, so 'the AHD binds X' claims are quantitative rather than categorical. Kuntimaddi A et al. 2015
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   Conditionally expressed MLL-AF9 and MLL-AF4 chimeric proteins localise to nucleoplasm and nucleolus in the same pattern as an amino-terminal MLL fragment and as endogenous MLL, indicating that the amino terminus of MLL confers a common subnuclear target to MLL chimeras.

   The amino terminus targets the mixed lineage leukemia (MLL) protein to the nucleolus, nuclear matrix and mitotic chromosomal scaffolds. · Caslini C, Alarcon AS, Hess JL, et al. · Leukemia 2000
   PMID 11069025DOI
   ▸What this evidence cannot show

   Conditional overexpression in U937 and HeLa, not endogenous fusion in a t(9;11) line or a patient sample, read out by immunofluorescence and immunoelectron microscopy in 2000. The paper does not test whether the AF9 moiety contributes anything, because it has no AF9-only comparator and no NLS mutant.

   ▸Evidence that argues against this (1)

   • A later study found that the MLL-AF4 fusion protein can itself alter the subnuclear localisation of AF9, so partner localisation in an MLL-rearranged cell is not simply dictated by the MLL N-terminus and can be redistributed in trans. Erfurth F et al. 2004

• part of →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)

  residue-level architecture of the chimeric protein

• related to →MLLT3 / AF9 — a famous fusion partner that is not a dependency, and a hub domain that is

  P42568 coordinates, fetched and read back 2026-09-20

• related to →KMT2A / MLL1 — the gene that breaks

  Q03164 coordinates, fetched and read back 2026-09-20

• analogous to →The KMT2A::MLLT1 chimeric protein: exactly what survives the join on each side, in residue numbers

  same 5-prime half, opposite half of the paralogous partner

• supports →MLLT3 (AF9): the YEATS acetyl-lysine reader is discarded in ~92% of MLL-AF9 patients — same shape as PZP, but the causal test has never been run

  adds the NLS at 295-300 to the list of discarded AF9 modules

• supports →Four regulatory elements on AF9, and the fusion keeps exactly one - the phospho-switch that controls SEC binding

  a fifth discarded element: the only annotated AF9 NLS

• ← supportsWhere t(9;11) actually breaks inside MLLT3: 98.4% downstream of exon 4, and the two patients in 442 who kept the reader

  the breakpoint that sets the architecture

• ← part ofThe retained AF9 surface in residue detail: one disordered 70-mer, three DOT1L sites, two of them high-affinity and simultaneously occupiable

  the only AF9 module the fusion retains

• ← depends onWhere KMT2A::MLLT3 sits in the nucleus: two papers, both overexpression, both from before 2005 - and the fusion carries no AF9 import signal

  the predicted AF9 NLS at 295-300 is on discarded sequence

• ← part ofBCOR, not CBX8, is the AHD contact MLL-AF9 cannot lose - and the same two partners bind ENL far more tightly than AF9

  what the retained AHD actually does