The retained AF9 surface in residue detail: one disordered 70-mer, three DOT1L sites, two of them high-affinity and simultaneously occupiable
also called AF9 AHD-DOT1L interface · 2MV7 · AHD DOT1L
The only part of AF9 the cancer protein keeps is a floppy 70-residue tail that folds into shape only when something grabs it. DOT1L grabs it in three different places, and two of those are strong enough that two copies of the tail can hold one DOT1L at once.
The same thing, for a scientist
KMT2A::MLLT3 retains approximately AF9 residues 499-568, the ANC1 homology domain, which is intrinsically disordered in isolation and folds on binding [E2]. Structural work on that surface gives a level of residue detail available for almost no other KMT2A partner contact. **DOT1L presents three separate AF9-binding motifs: Site 1 (aa 628-653), Site 2 (aa 863-878) and Site 3 (aa 877-900)** [E1]. Site 1 is weak (Kd approximately 2,000 nM) and Site 2 binds about 17-fold more weakly than Site 3; Sites 2 and 3 are separated by a four-residue spacer, forming a repeat motif (aa 863-900) that **size-exclusion chromatography shows can bind two AF9 molecules simultaneously** [E1]. The NMR solution structure of the highest-affinity complex, DOT1L Site 3 with AF9, is **PDB 2MV7**; AF9 forms three helices and a beta hairpin around a DOT1L beta strand, with the AF9 C-terminus (563-568) unstructured even when bound, and the complex superposes on the earlier AF4-AF9 structure **PDB 2LM0** with a backbone RMSD of 1.4 A [E1][E2]. Structure-guided AF9 mutants separate the sites: **D546R** abolishes binding to both high-affinity sites, **D544R** hits Site 3 much harder than Site 2 [E1]. Two facts constrain how far this can be pushed therapeutically: **ENL binds the same DOT1L peptides with similar affinity to AF9** [E1], and the authors state explicitly that their AF9 point mutants are **not completely specific for a single protein partner** [E1].
The three sites and their affinities
| DOT1L motif | residues | affinity |
|---|---|---|
| Site 1 | 628-653 | weak, Kd approximately 2,000 nM |
| Site 2 | 863-878 | high, but about 17-fold weaker than Site 3 |
| Site 3 | 877-900 | highest |
All three form structurally similar complexes with AF9 - the NMR spectra are nearly superimposable - so AF9 folds essentially the same way around each 1. Sites 2 and 3 are four residues apart and together make a repeat motif; a size-exclusion profile of AF9 with the 863-900 repeat is consistent with two AF9 proteins bound to one DOT1L repeat 1.
That last point is the one with mechanistic consequences that nobody has drawn out. In a KMT2A::MLLT3 cell there are at least two AHD-bearing species present - wild-type AF9 and the fusion - and a DOT1L molecule with two adjacent high-affinity sites could in principle hold one of each. Whether it does is unmeasured.
The structures, and what they are not
2MV7 is the NMR solution structure of DOT1L Site 3 (aa 877-900) bound to AF9; 2LM0 is the AF4 peptide bound to the same AF9 surface, determined earlier by the same laboratory 12. The two superpose at 1.4 A backbone RMSD, and DOT1L and AF4 share a consensus hydrophobic motif 1.
A database fact is opposable on scope, and this one should be. Neither entry contains any KMT2A sequence. Both are short synthetic partner peptides bound to an isolated AF9 fragment expressed as an MBP fusion. The density - or here, the restraint set - covers a 70-residue disordered peptide and a 24-residue partner motif. It is excellent for designing a competitor of that peptide-peptide contact, and it says nothing about the geometry of the same contact inside a 1,970-residue chimera loaded onto chromatin.
Why this surface is a hard drug target, in the primary authors' own words
Two statements from the primary work bound every therapeutic claim built on this interface.
- The paralogue binds the same peptides just as well. 'As ENL, another fusion partner with MLL, is highly homologous to AF9, we tested binding of each of these DOT1L peptides with ENL which yielded similar binding affinities for ENL as seen for AF9' 1. So a DOT1L-competitive AHD ligand is an AF9-and-ENL ligand.
- Their own mutants are not partner-specific. The authors write that binding measurements show these and other previously described AF9 or ENL mutations 'are not completely specific for a single protein partner' 1, and they make the methodological point directly: co-immunoprecipitation alone does not establish that a mutation has no effect on an interaction.
The second point is a correction aimed at the field's own published mutants, and it applies to any entry in this graph that treats an AF9 point mutant as a clean separation-of-function reagent.
| partner a | gene-mllt3 |
|---|---|
| partner b | gene-dot1l |
| motif b | DOT1L Site 1 aa 628-653; Site 2 aa 863-878; Site 3 aa 877-900 |
| interface residues a note | D544 and D546 lie on the AF9 beta-hairpin (beta2) and make direct contacts with DOT1L S882 and K878 respectively; 563-568 is the AF9 C-terminal tail that stays unstructured in the complex. |
| structure note | 2MV7 = NMR, DOT1L Site 3 (877-900) with AF9 AHD. 2LM0 = AF4 peptide with AF9 AHD. Neither contains KMT2A sequence. Identifiers read from the body text of PMID 25921540 (PMC4426023) on 2026-09-20. |
| kd NM | 2000 |
| kd note | 2,000 nM is the WEAK Site 1 value, the only absolute Kd stated in the body text; Site 2 is reported as 17-fold weaker than Site 3 and the absolute Site 2 and Site 3 values are in a figure panel rather than the text, so they are not recorded here. |
| stoichiometry | two AF9 AHD molecules per DOT1L 863-900 repeat, by size-exclusion profile |
| paralogue cross reactivity | ENL binds the same DOT1L peptides with similar affinity to AF9 |
| retained in fusion | true |
| druggability | Peptidomimetic inhibitors of the DOT1L-AF9/ENL contact exist in the literature, but the surface is shared with wild-type AF9 and with ENL, and ENL is required for AML maintenance while AF9 is required for normal HSC self-renewal. There is no reported ligand that distinguishes the fusion's AHD from the wild-type AHD, and none is structurally obvious, because the sequences are identical. |
interface residues a
[ 544, 546, 563, 564, 565, 566, 567, 568 ]
structures
[ "2MV7", "2LM0" ]
separating mutants
[
{
"mutant": "AF9 D546R",
"effect": "abolishes binding to DOT1L Sites 2 and 3; still co-IPs the low-affinity Site 1"
},
{
"mutant": "AF9 D544R",
"effect": "pronounced effect on Site 3, limited on Site 2; still interacts with full-length DOT1L"
}
]ligands
[]
- An AHD-DOT1L competitor cannot distinguish the fusion from wild-type AF9 or from ENL: the retained sequence is identical to the wild-type sequence, so there is no fusion-specific epitope on this surface at all.
- ENL is required for AML maintenance and MLLT3 is required for normal human HSC self-renewal and engraftment, so an inhibitor of this shared surface carries an on-target stem-cell toxicity risk that is not hypothetical.
- Three DOT1L sites with graded affinities means partial occupancy is the likely pharmacological regime, and the same laboratory shows partial recruitment gives partial H3K79 methylation and partial transformation - a shallow dose-response is the expected clinical shape, which is what pinometostat delivered.
- Does one DOT1L molecule bridge the fusion AHD and a wild-type AF9 or ENL AHD in a leukaemic cell? The two-site repeat makes it possible and nobody has looked.
- What are the absolute Kd values for DOT1L Sites 2 and 3? They are in a figure panel, not in the text, and this graph should not carry a number read off a plot.
- Is there any position in the AHD where AF9 and ENL differ enough to build partner-selective chemistry? The paralogue cross-reactivity result says the DOT1L-facing surface is not that position.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
DOT1L has three separate AF9-binding motifs - Site 1 (aa 628-653, Kd approximately 2,000 nM), Site 2 (aa 863-878, about 17-fold weaker than Site 3) and Site 3 (aa 877-900) - which form structurally similar complexes with AF9 aa 499-568; the Site 2/3 repeat can bind two AF9 molecules simultaneously; the DOT1L Site 3-AF9 NMR structure is PDB 2MV7 and superposes on the AF4-AF9 structure PDB 2LM0 at 1.4 A RMSD; AF9 D546R abolishes binding to both high-affinity sites while D544R affects Site 3 more than Site 2; ENL binds the same DOT1L peptides with similar affinity; and the authors state their AF9 and ENL mutants are not completely specific for a single partner.
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 2015What this evidence cannot show
Read in full text from the PMC HTML page (PMC4426023) on 2026-09-20. All affinities are fluorescence polarization on MBP-tagged isolated AF9 AHD against synthetic DOT1L peptides; the AHD is reported to aggregate at these concentrations. The two-AF9 stoichiometry is inferred from a size-exclusion elution profile against protein standards, not from a structure or from analytical ultracentrifugation. The authors note their own Site 2 affinity differs significantly from a previously published measurement of the same contact.
Evidence that argues against this (1)
- The authors' own caution cuts against reading any of these mutants as a clean separation-of-function reagent: DOT1L mutations affect binding to wild-type AF9 and to ENL as well as to MLL-AF9, so a phenotype in a DOT1L-mutant add-back experiment is the sum of effects on all three targets and cannot be ascribed to the fusion contact alone. Kuntimaddi A et al. 2015
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The AF9 ANC1 homology domain is intrinsically disordered, recruits multiple transcription factors through coupled folding and binding, and retains substantial dynamic behaviour in complex, with the structure of the AF9 AHD bound to the elongation factor AF4 showing conserved aliphatic partner residues forming an integral part of the complex's hydrophobic core.
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 2013What this evidence cannot show
Isolated peptides in vitro; no fusion protein, no chromatin, no cell. The dynamic behaviour is measured by NMR relaxation on a construct, and its functional interpretation (facilitating exchange between partners) is the authors' proposal rather than a measurement.
Evidence that argues against this (1)
- A review from the same structural laboratory notes that ENL, not AF9, is the family member shown to undergo phase separation to enhance SEC recruitment, so 'the AHD is a dynamic exchange hub' may describe a property whose functional expression differs between the two paralogues - which matters because the fusion keeps only this module. Kabra A et al. 2022
the AF9 side of the same surface, with the competing DOT1L ligand
the same retained AF9 surface, with partner-conserved aliphatic core
the retained AF9 AHD surface
three DOT1L motifs, two high-affinity, PDB 2MV7
the only AF9 module the fusion retains
same AF9 surface, competing partner
the structural basis of the gradient
same site, competing partner