WDR5 WIN site and KMT2A Arg3765: measured at 1.7 uM, three co-crystals, and absent from every leukaemic fusion - plus the MLL3 peptide that binds it 51-fold tighter
also called WIN motif · WDR5-interaction motif · MLL1 3762-3773 · Arg3765
Normal KMT2A docks onto WDR5 by pushing a single arginine into a hole on WDR5's surface, and that docking is what builds its enzyme. Every leukaemia fusion is cut about 2,400 amino acids before that arginine, so the fusion never docks at all - and the same hole binds a related protein, MLL3, fifty times more tightly.
The same thing, for a scientist
The KMT2A (MLL1) WIN motif docks Arg3765 into the arginine-binding channel of WDR5's WD40 beta-propeller, the same channel that reads histone H3 Arg2. Three independent co-crystals define it: 3EG6 (1.72 A, MLL1 3762-3773), 3EMH (1.37 A, MLL1 3764-3776) and 4ESG (1.70 A, MLL1 3755-3771); all three peptide spans were mapped onto the fetched UniProt Q03164 sequence in this cycle rather than taken on trust. By ITC the 12-residue and 19-residue MLL1 WIN peptides bind WDR5 at Kd 1.7 uM and the 14-residue peptide at 2.8 uM; full-length MLL1 protein binds at 120 nM. Crucially for anyone reading WIN-site pharmacology as KMT2A pharmacology, the MLL3 WIN peptide binds the same site at 54 nM, 5- to 51-fold tighter than MLL1's. The motif begins at residue 3755-3762 depending on construct, roughly 2,350 residues C-terminal to the KMT2A breakpoint cluster region, so it is absent from KMT2A::AFF1 and from every other der(11) fusion protein.
Residues on both sides
KMT2A (Q03164): the WIN core is G3762-S3763-A3764-R3765-A3766-E3767, with flanking residues V3768-H3769-L3770-R3771-K3772-S3773 contributing affinity. The residue assignment here is arithmetic on the fetched UniProt sequence: the 3EG6 peptide GSARAEVHLRKS aligns uniquely at 3762-3773, which puts Arg at 3765 15.
WDR5 (P61964): the peptide inserts into the central channel of the seven-bladed WD40 propeller - the same channel used by histone H3 Arg2 - which is why the interaction was originally described as MLL binding the histone-H3-binding pocket 3.
A construct detail worth recording because it bites: the WDR5-RBBP5-MLL1 ternary structure 3P4F uses an MLL1 peptide HGAARAEVHL, which aligns to 3761-3770 with an S3763A substitution. Structures of this motif are not all wild-type peptide.
The affinity table, all from one ITC study
| ligand | Kd | method |
|---|---|---|
| MLL1 WIN 12-mer (3762-3773) | 1.7 uM | ITC |
| MLL1 WIN 19-mer (3755-3773) | 1.7 uM | ITC |
| MLL1 WIN 14-mer (3758-3771) | 2.8 uM | ITC |
| full-length MLL1 protein | 120 nM | quoted by Dharmarajan et al. from earlier work |
| MLL3 WIN 14-mer | 54 nM | ITC |
The last row is the one that matters for interpretation. Dharmarajan et al. state that the other human SET1-family WIN peptides bind WDR5 5- to 51-fold more tightly than MLL1's, with MLL3 the tightest 1. A WIN-site ligand is therefore competing against MLL1 at its weakest family member. This is independent structural support for the graph's existing concept-a-wdr5-result-is-not-automatically-a-kmt2a-axis-result.
Lost in the fusion, and what that does to the WIN-site rationale
Residue 3755 is about 2,350 residues C-terminal to the KMT2A breakpoint cluster region, so no der(11) KMT2A fusion protein contains the WIN motif. Consistently, the founding biochemical survey of the MLL complex found that oncogenic MLL fusions retain menin but not the WRAD subunits 4.
That makes every WIN-site inhibitor a wild-type-allele drug in a KMT2A-rearranged cell, not a fusion drug - and the chemical series that reaches picomolar affinity at this site was developed and profiled against MYC-driven biology rather than against MLL1 2. The graph's subtraction-kmt2a-win-wrad-interface already records the loss; this entry adds the residue-level geometry, the measured affinities and the paralogue-selectivity problem.
| partner a | gene-wdr5 |
|---|---|
| partner b | gene-kmt2a |
| motif b | KMT2A WIN motif, core 3762-3767 (GSAREA numbering G3762 S3763 A3764 R3765 A3766 E3767), crystallised spans 3755-3776 |
| motif a | WDR5 central arginine-binding channel of the WD40 beta-propeller (the histone H3 Arg2 channel) |
| kd NM | 1700 |
| fate in fusion | lost - the motif starts about 2,350 residues C-terminal to the KMT2A breakpoint cluster region and is absent from every der(11) KMT2A fusion protein |
| paralogue selectivity | MLL1 is the WEAKEST SET1-family binder at this site; MLL2, MLL3, MLL4, SETD1A and SETD1B WIN peptides bind 5- to 51-fold tighter, MLL3 at 54 nM |
| druggability | validated as a ligandable pocket (picomolar WIN-site inhibitors exist) but irrelevant to the fusion protein and non-selective within the SET1 family |
| resolution note | PDB titles, methods, resolutions and entity sequences fetched from data.rcsb.org 2026-09-20; residue spans derived by aligning those sequences to the UniProt Q03164 sequence fetched the same day. |
interface residues b
[ "KMT2A R3765 (inserted arginine)", "KMT2A A3764", "KMT2A A3766", "KMT2A H3769" ]
hot spot residues
[ "KMT2A Arg3765" ]
kd measurements — the numbers behind the chart
[
{
"ligand": "MLL1 3762-3773 (12-mer)",
"kd_nM": 1700,
"method": "ITC",
"source_pmid": "22665483"
},
{
"ligand": "MLL1 3755-3773 (19-mer)",
"kd_nM": 1700,
"method": "ITC",
"source_pmid": "22665483"
},
{
"ligand": "MLL1 3758-3771 (14-mer)",
"kd_nM": 2800,
"method": "ITC",
"source_pmid": "22665483"
},
{
"ligand": "full-length MLL1 protein",
"kd_nM": 120,
"method": "quoted by Dharmarajan et al. from prior work",
"source_pmid": "22665483"
},
{
"ligand": "MLL3 WIN 14-mer (paralogue comparator)",
"kd_nM": 54,
"method": "ITC",
"source_pmid": "22665483"
}
]structures
[ "3EG6", "3EMH", "4ESG", "3P4F" ]
structure details — the numbers behind the chart
[
{
"pdb": "3EG6",
"title_fetched": "Structure of WDR5 bound to MLL1 peptide",
"method": "X-ray",
"resolution_A": 1.72,
"kmt2a_span": "3762-3773",
"fetched": "2026-09-20"
},
{
"pdb": "3EMH",
"title_fetched": "Structural basis of WDR5-MLL interaction",
"method": "X-ray",
"resolution_A": 1.37,
"kmt2a_span": "3764-3776",
"fetched": "2026-09-20"
},
{
"pdb": "4ESG",
"title_fetched": "X-ray structure of WDR5-MLL1 Win motif peptide binary complex",
"method": "X-ray",
"resolution_A": 1.7,
"kmt2a_span": "3755-3771",
"fetched": "2026-09-20"
},
{
"pdb": "3P4F",
"title_fetched": "Structural and biochemical insights into MLL1 core complex assembly and regulation.",
"method": "X-ray",
"resolution_A": 2.35,
"kmt2a_span": "3761-3770, with an S3763A substitution in the deposited peptide",
"fetched": "2026-09-20"
}
]- Essential function: WDR5 is pan-essential across DepMap lineages and sits in NSL and other complexes, so WIN-site occupancy has a toxicity floor that is not about MLL1 at all.
- Target mismatch: the fusion does not contain this motif, so a WIN-site drug cannot be a fusion-directed agent however good its chemistry.
- Paralogue rescue: MLL2/MLL3/SETD1A bind the same channel 5- to 51-fold tighter and would be displaced last, so partial occupancy preferentially removes the weakest binder, MLL1.
- Readout confound: WIN-site inhibitors kill through ribosome protein gene suppression in at least one well-characterised series, so an anti-leukaemic effect is not evidence that the MLL1 axis was engaged.
- Does any WIN-site inhibitor discriminate MLL1 from MLL3 at the 51-fold affinity gap the peptide data predict? No published compound is profiled against all six SET1-family WIN peptides.
- If the WIN motif is absent from the fusion, does a WIN-site drug in a KMT2A-rearranged cell act on the wild-type allele, on ribosome protein genes, or on neither in a way that matters?
- Is the 120 nM full-length MLL1 affinity - 14-fold tighter than the peptide - carried by a second WDR5 contact outside the WIN motif, and if so is that second contact also lost in the fusion?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
MLL1 WIN motif peptides bind WDR5 with Kd 1.7 uM (12- and 19-mer) and 2.8 uM (14-mer) by ITC, while the MLL3 WIN peptide binds at 54 nM and the SET1-family peptides bind 5- to 51-fold more tightly than MLL1's; PDB 4ESG is the WDR5-MLL1 WIN binary co-crystal at 1.70 A.
Structural basis for WDR5 interaction (Win) motif recognition in human SET1 family histone methyltransferases. · Dharmarajan V, Lee JH, Patel A, et al. · J Biol Chem 2012What this evidence cannot show
All affinities are isolated synthetic peptide to full-length WDR5 in 300 mM NaCl at 20 C; peptide affinity is not protein affinity, as the paper's own quoted 120 nM for full-length MLL1 shows. The MLL1 H3769Y peptide was insoluble in ITC buffer and is missing from the comparison. The full text was reached only by the fourth route (PMC HTML page, 74,125 characters) after fullTextXML returned HTTP 500 and efetch db=pmc returned no body element.
Evidence that argues against this (1)
- A WIN-site inhibitor with picomolar affinity for WDR5 displaces WDR5 from chromatin and kills cancer cells through ribosome protein gene suppression rather than through an MLL1-dependent mechanism, so occupancy at this interface does not establish that the MLL1 arm is the arm being drugged. Aho ER et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
WIN-site chemical matter reaching picomolar affinity for WDR5 exists, and its cellular mechanism is displacement of WDR5 from ribosome protein gene chromatin.
Displacement of WDR5 from Chromatin by a WIN Site Inhibitor with Picomolar Affinity. · Aho ER, Wang J, Gogliotti RD, et al. · Cell Rep 2019What this evidence cannot show
The cell models are MYC-driven and not principally KMT2A-rearranged; the paper establishes the site is ligandable, not that ligating it reports on MLL1.
Evidence that argues against this (1)
- Dharmarajan et al. show MLL1 is the weakest SET1-family binder at this site, so a compound that displaces MLL1 easily may leave MLL2/MLL3/SETD1A engaged - the opposite of a clean MLL1 probe. Dharmarajan V et al. 2012
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
MLL binds WDR5 through the same surface pocket WDR5 uses to read histone H3, defined in a 1.37 A co-crystal (PDB 3EMH).
WDR5 interacts with mixed lineage leukemia (MLL) protein via the histone H3-binding pocket. · Song JJ, Kingston RE · J Biol Chem 2008What this evidence cannot show
A 13-residue peptide in a crystal; the paper does not address the fusion protein, which does not contain the motif.
Evidence that argues against this (1)
- Patel et al., publishing the same interface in the same journal and year, emphasise the arginine-channel geometry rather than histone mimicry; the two framings imply different selectivity expectations for a competitive ligand. Patel A et al. 2008
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Oncogenic MLL fusion proteins retain menin but not the WRAD subunits WDR5, RBBP5 and ASH2L.
Leukemia proto-oncoprotein MLL forms a SET1-like histone methyltransferase complex with menin to regulate Hox gene expression. · Yokoyama A, Wang Z, Wysocka J, et al. · Mol Cell Biol 2004What this evidence cannot show
This is the founding survey and its fusion-versus-WRAD conclusion rests on deletion constructs rather than on a co-IP of an endogenous fusion; the graph's `concept-founding-wrad-negative-is-a-deletion-mutant-not-a-fusion` records that limitation and it applies here too. Read via the PMC HTML page (PMC480881).
Evidence that argues against this (1)
- Because the WIN motif is simply absent from the fusion by position, the biochemical result is over-determined: even a flawed co-IP could not have shown retention. The genuine open question is not whether the fusion binds WRAD directly but whether WRAD reaches it as a second-shell passenger. Dharmarajan V et al. 2012
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
PDB 3EG6 is a 1.72 A co-crystal of WDR5 with an MLL1 peptide whose sequence aligns to Q03164 residues 3762-3773, placing the docking arginine at 3765.
PDB 3EG6: Structure of WDR5 bound to MLL1 peptide · Patel A, Dharmarajan V, Cosgrove MS · RCSB Protein Data Bank 2008What this evidence cannot show
Title, method, resolution and both polymer entities fetched from data.rcsb.org on 2026-09-20; the residue assignment is this cycle's alignment of the fetched entity sequence to the fetched UniProt Q03164 sequence, not a number copied from a paper or from another graph entry.
Evidence that argues against this (1)
- The 3P4F ternary structure uses an MLL1 WIN peptide carrying an S3763A substitution, so peptide identity differs between deposits and a residue-level claim should name the entry it comes from. Dharmarajan V et al. 2012
MLL1 WIN peptide in the WDR5 arginine channel, Kd 1.7 uM
KMT2A WIN motif 3762-3767, Arg3765
- supports →Lost from the KMT2A half: the WIN motif and the whole WRAD interface (aa 3762-3767, Arg3765)
residue-level geometry and measured affinity for the interface that entry records as lost
MLL3 WIN peptide binds 51-fold tighter than MLL1's at the same site
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
absent from the index fusion by position
the structural half of that distinction
the two grips share one propeller and are co-crystallised in 3P4F
the SET1-family WIN affinity ranking
the competing WDR5 route: a WIN-motif protein interface rather than a histone-tail interface
acts on the histone SUBSTRATE, not on the WIN contact - the two can be intact and permissive at once
orthogonal to the WIN contact: the gate acts on the histone tail and does not require the fusion to have retained anything