LEDGF IBD to KMT2A: TWO motifs, not one - IBM1 at 124-133 and IBM2 at 147-152, and the Kd this graph recorded as missing is 14.7 uM
also called LEDGF IBM1 · LEDGF IBM2 · MLL IBD-binding motifs · PSIP1-KMT2A affinity
LEDGF does not grab the KMT2A fragment at one place but at two, and on its own the grip is weak; menin tightens it roughly tenfold. Both grips sit in the piece of KMT2A that every leukaemic fusion keeps.
The same thing, for a scientist
Murai et al. mapped the LEDGF/p75 IBD contact on MLL1 to two separate motifs by NMR: IBM1, well ordered over MLL1 residues 124-133, and IBM2, residues 147-152, both resolved in the MLL-IBD solution structure PDB 2MTN (construct: MLL1 110-160 fused to LEDGF 337-442; IBD ordered 349-428). The binary MLL1(1-160)-IBD interaction has Kd = 14.7 +/- 0.7 uM by NMR titration; preforming the menin-MLL1(1-160) complex increases IBD affinity about tenfold to Kd = 1.4 uM by ITC. F129A (IBM1) and F148A or F148A/L149A (FLAA, IBM2) each degrade the interaction, and the single-motif view this graph previously carried - KMT2A 140-160 with hotspots F148/F151, from the 2MSR structure - captures IBM2 only. Both motifs lie N-terminal to residue 160 and are therefore retained intact in KMT2A::AFF1 and in every other der(11) KMT2A fusion.
What is new here relative to the graph's existing LEDGF entries
interface-psip1-ibd-kmt2a states that no published dissociation constant could be verified for the LEDGF IBD-KMT2A interaction and calls it the largest quantitative gap in the partner dossier. That is now closed. Murai et al. 2014 report both a binary and a ternary Kd for exactly this interaction 1. The same paper also shows the interface is bipartite, which the 2MSR-derived description (KMT2A 140-160) could not show because that construct contains only IBM2 2.
The hotspot list also needs care. From the fetched UniProt Q03164 sequence, residues 147-152 are Q-F-L-G-F-G, so F148 and F151 are both inside IBM2 and the existing entry's hotspot pair is not wrong; but the mutant Murai et al. actually used to break IBM2 in cells is F148A/L149A, and the motif they define is 147-152, not 140-160 1.
Residues, on both sides
KMT2A (Q03164, retained in the fusion): IBM1 = 124-133, with F129 the mutated hotspot; IBM2 = 147-152 (Q-F-L-G-F-G), with F148, L149 and F151. Both coordinates were checked against the UniProt Q03164 sequence fetched in this cycle rather than copied from another entry 1.
LEDGF/PSIP1 (O75475): the IBD, ordered as residues 349-428 in 2MTN. The pocket that receives IBM2 is lined by L363, I403 and F406 1. The 2MSR construct places the same IBD surface against KMT2A 140-160 2.
Four structures, three of them tethered
| PDB | contents | chains |
|---|---|---|
| 2MSR | LEDGF IBD + KMT2A 140-160 | two, bimolecular |
| 2MTN | KMT2A 110-160 fused to LEDGF 337-442 | one, chimeric |
| 6EMQ | KMT2A 111-160 fused to the LEDGF IBD | one, chimeric |
| 3U88 | menin + engineered KMT2A + LEDGF IBD | three, but the KMT2A chain is spliced |
The only bimolecular structure contains only IBM2. Every structure that spans both motifs is a tethered construct. That is a bounded statement about what this interface's structural record can support: connectivity and geometry, yes; the free energy of bipartite association, no.
Why menin tightens it - and what that means for a menin inhibitor
IBD binds the preformed menin-MLL1 complex about tenfold better than it binds MLL1 alone (1.4 uM against 14.7 uM) 1. So the clamp is cooperative in one direction: menin occupancy raises LEDGF affinity. The untested inverse is whether a clinical menin inhibitor, by evicting the MBM1 peptide, also drops LEDGF occupancy by that same order - which would make menin inhibitors partly LEDGF-axis drugs by accident. No published experiment measures LEDGF occupancy on chromatin under revumenib, ziftomenib or bleximenib.
Phosphorylation sits on top of this
Sharma et al. show the IBM class is a CK2-gated phospho-switch, raising IBD affinity 2- to 10-fold, with KMT2A S136 and S142 as the acceptors 3. S136 and S142 lie between IBM1 (124-133) and IBM2 (147-152) - in the linker joining the two motifs - which is consistent with a switch that changes how the two grips are presented rather than one that sits in either pocket. The graph's interface-psip1-ibd-kmt2a-phosphoswitch carries this and should be read alongside.
| partner a | gene-psip1 |
|---|---|
| partner b | gene-kmt2a |
| motif a | LEDGF/PSIP1 integrase-binding domain (IBD), ordered 349-428 in PDB 2MTN |
| motif b | two IBD-binding motifs on KMT2A: IBM1 = 124-133, IBM2 = 147-152 |
| kd NM | 14700 |
| fate in fusion | retained - both IBM1 and IBM2 lie before residue 160, roughly 1,240 residues N-terminal to the breakpoint cluster region |
| druggability | No small molecule against the IBD is in the clinic. The IBD's best-characterised ligand is HIV-1 integrase, which binds about a thousandfold tighter than the menin-MLL complex does and displaces it with IC50 301 nM in AlphaLISA - so the pocket is demonstrably ligandable by a protein, and the affinity a small molecule must beat is only micromolar. |
| resolution note | All PDB titles, methods, resolutions and polymer entities in this entry were fetched from data.rcsb.org on 2026-09-20 and the peptide sequences aligned to UniProt Q03164/O75475 sequences fetched the same day. No identifier was written from recall or copied from a sibling entry. |
interface residues a
[ "PSIP1 L363", "PSIP1 I403", "PSIP1 F406" ]
interface residues b
[ "KMT2A F129 (IBM1)", "KMT2A F148 (IBM2)", "KMT2A L149 (IBM2)", "KMT2A F151 (IBM2)" ]
kd measurements — the numbers behind the chart
[
{
"ligand": "MLL1 1-160 + LEDGF IBD (binary)",
"kd_nM": 14700,
"error_nM": 700,
"method": "NMR chemical-shift titration",
"source_pmid": "25305204"
},
{
"ligand": "menin-MLL1(1-160) complex + LEDGF IBD (ternary)",
"kd_nM": 1400,
"method": "ITC",
"source_pmid": "25305204"
},
{
"ligand": "HIV-1 integrase + LEDGF IBD (comparator, quoted by Murai et al. from their ref 27)",
"kd_nM": 10,
"method": "quoted comparator",
"source_pmid": "25305204"
}
]structures
[ "2MTN", "2MSR", "6EMQ", "3U88" ]
structure details
[
{
"pdb": "2MTN",
"title_fetched": "Solution structure of MLL-IBD complex",
"method": "NMR",
"resolution_A": null,
"contents": "single chimeric chain annotated to Q03164 + O75475; MLL1 110-160 fused to LEDGF 337-442",
"fetched": "2026-09-20"
},
{
"pdb": "2MSR",
"title_fetched": "Solution structure of LEDGF/p75 IBD in complex with MLL1 peptide (140-160)",
"method": "NMR",
"resolution_A": null,
"contents": "Q03164 21-mer (mapped to 140-160 against the UniProt sequence) + O75475 88-mer",
"fetched": "2026-09-20"
},
{
"pdb": "6EMQ",
"title_fetched": "Solution structure of the LEDGF/p75 IBD - MLL1 (aa 111-160) complex",
"method": "NMR",
"resolution_A": null,
"contents": "single chimeric chain annotated to O75475 + Q03164, MLL1 111-160 fused to the LEDGF IBD",
"fetched": "2026-09-20",
"note": "deposited by the Sharma et al. CK2 phospho-switch study, PMID 29997176"
},
{
"pdb": "3U88",
"title_fetched": "Crystal structure of human menin in complex with MLL1 and LEDGF",
"method": "X-ray",
"resolution_A": 3,
"contents": "MEN1 550-mer + engineered Q03164 75-mer (maps to 5-15 + 23-39 + 109-153) + O75475 89-mer",
"fetched": "2026-09-20"
}
]ligands
[]
corrections owed elsewhere
[ "interface-psip1-ibd-kmt2a states no Kd could be verified for this interaction; Kd = 14.7 uM (binary, NMR) and 1.4 uM (with menin, ITC) are published in PMID 25305204.", "interface-psip1-ibd-kmt2a and target-phase2-psip1-ibd-dossier describe the KMT2A determinant as a single motif at 140-160; it is bipartite (IBM1 124-133, IBM2 147-152).", "interface-psip1-ibd-kmt2a lists structures 2MSR, 3U88, 2N3A, 2B4J, 3F9K, 3HPG but not 6EMQ, the LEDGF IBD-MLL1 111-160 complex from PMID 29997176." ]
- Resistance: the interface is bipartite, so a ligand that occupies only the IBM2 pocket leaves IBM1 free and may be out-competed by avidity rather than by mutation.
- Selectivity: the same IBD surface is used by HIV-1 integrase, POGZ and JPO2, and integrase binds it about a thousandfold more tightly than menin-MLL does; an IBD ligand tuned to displace MLL will be a weak integrase-site ligand and vice versa.
- Cooperation: because menin raises LEDGF affinity about tenfold, combining a menin inhibitor with an IBD ligand may be sub-additive - the menin drug has already weakened the target the second drug is aimed at.
- Essential function: PSIP1 anchors H3K36me-directed targeting generally, not only the KMT2A complex, so loss-of-function toxicity is not confined to the leukaemic complex.
- Does a clinical menin inhibitor reduce LEDGF occupancy at fusion target genes? The tenfold cooperativity predicts it should, and no published experiment measures it.
- Is the IBM1+IBM2 bipartite arrangement present on the fusion protein in cells, or does the AFF1 moiety reorganise the N-terminal fragment? Every affinity number here is from isolated KMT2A fragments.
- Does CK2 phosphorylation at S136/S142 - which sit in the IBM1-IBM2 linker - act by changing the spacing of the two motifs rather than by altering either pocket?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
LEDGF IBD binds MLL1 through two motifs, IBM1 (residues 124-133) and IBM2 (147-152); MLL1(1-160) binds IBD with Kd 14.7 uM, and the menin-MLL1 complex binds IBD about tenfold tighter at Kd 1.4 uM.
The same site on the integrase-binding domain of lens epithelium-derived growth factor is a therapeutic target for MLL leukemia and HIV. · Murai MJ, Pollock J, He S, et al. · Blood 2014What this evidence cannot show
The structure 2MTN is a designed single-chain fusion of MLL1 110-160 to LEDGF 337-442, not a bimolecular complex, chosen to make a weak interaction tractable; intramolecular tethering can stabilise contacts that are marginal in trans. The 14.7 uM is from NMR chemical-shift titration averaged over three residues; the 1.4 uM is ITC on a preformed complex. Neither is measured on a KMT2A fusion protein.
Evidence that argues against this (1)
- Cermakova et al., working on the same interface in the same year, define the MLL1 determinant as a single segment, residues 140-160, and deposited 2MSR on that basis; a reader taking only that paper would not know IBM1 exists. The two descriptions are reconcilable (2MSR's construct contains IBM2 and not IBM1) but the published record is genuinely two-voiced about what the motif is. Cermakova K et al. 2014
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The LEDGF IBD-KMT2A contact was independently characterised as a single KMT2A 140-160 segment and deposited as PDB 2MSR.
Validation and structural characterization of the LEDGF/p75-MLL interface as a new target for the treatment of MLL-dependent leukemia. · Cermakova K, Tesina P, Demeulemeester J, et al. · Cancer Res 2014What this evidence cannot show
NMR structure of a 21-residue peptide with an isolated IBD; no affinity constant is reported in the abstract, and the construct cannot report on residues outside 140-160.
Evidence that argues against this (1)
- Murai et al. show MLL1 fragments upstream of 140 are required for full binding and note that the menin-MLL-LEDGF crystal structure contains no MLL contacts upstream of residue 135, so a 140-160-only description under-reports the interface. Murai MJ et al. 2014
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
IBD-binding motifs are phospho-switches; CK2 phosphorylation of KMT2A S136 and S142 raises IBD affinity, and S136A/S142A reduces MLL1-fusion transformation.
Affinity switching of the LEDGF/p75 IBD interactome is governed by kinase-dependent phosphorylation. · Sharma S, Cermakova K, De Rijck J, et al. · Proc Natl Acad Sci U S A 2018What this evidence cannot show
The 2- to 10-fold affinity statement is made at the level of the IBM class across several IBD partners rather than measured for KMT2A alone; the transformation assay is colony formation with an MLL1 fusion, not the index fusion in patient cells.
Evidence that argues against this (1)
- Neither Murai et al. nor Cermakova et al. required phosphorylation to measure IBD binding to unmodified bacterially expressed MLL1 fragments, so phosphorylation tunes rather than gates this contact. Murai MJ et al. 2014
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
PDB 2MTN is a solution NMR structure of an MLL1-LEDGF IBD single-chain fusion construct, deposited by the Murai et al. study.
PDB 2MTN: Solution structure of MLL-IBD complex · Murai MJ, Pollock J, He S, et al. · RCSB Protein Data Bank 2014What this evidence cannot show
Fetched from the RCSB entry and polymer-entity APIs on 2026-09-20; the entry carries one polymer entity annotated to both O75475 and Q03164, which is the signature of a chimeric construct rather than a two-chain complex. No resolution applies (NMR).
Evidence that argues against this (1)
- 2MSR, from the other group, is a genuine two-chain complex (separate Q03164 and O75475 entities), so where the two structures disagree the bimolecular one is the more conservative model. Cermakova K et al. 2014
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The menin-MLL1-LEDGF ternary crystal structure 3U88 resolves KMT2A on the LEDGF side of the clamp only from residue ~109 onward.
The same pocket in menin binds both MLL and JUND but has opposite effects on transcription. · Huang J, Gurung B, Wan B, et al. · Nature 2012What this evidence cannot show
3.0 A resolution; the deposited MLL1 chain is a 75-residue engineered construct that splices the menin-binding N-terminus to the LEDGF-binding region, so residue numbering in that entry does not describe a continuous natural fragment. Verified against the fetched entity sequence, which maps to Q03164 5-15, 23-39 and 109-153. INTERACTION PROVENANCE (Q76, 2026-09-21, live IntAct REST), and it is a warning about reading database flags rather than about this interface: IntAct's only two records for the PSIP1-KMT2A pair both come from this PMID, and BOTH are flagged psi-mi:MI:1060 spoke expansion - because IntAct spoke-expands an n-ary deposition whatever produced it, and 3U88 is ternary. One of the two (EBI-15969639) is nonetheless typed MI:0407 DIRECT INTERACTION with detection method X-RAY DIFFRACTION, and the other (EBI-15969504) is MI:0915 physical association by molecular sieving. A graph-wide census this cycle classified pairs as unobserved when every record carried MI:1060; THIS PAIR was its single false positive, 1 of 27, and the correction is that MI:1060 describes the representation, not whether anybody looked. The interface is observed at atomic resolution and the flag says nothing to the contrary. See ana-two-thirds-of-the-curated-interactome-is-a-row-nobody-observed and concept-what-a-curated-interaction-row-actually-claims.
Evidence that argues against this (1)
- Murai et al. note explicitly that this crystal structure lacks any MLL contacts upstream of residue 135, so it cannot be used as evidence about IBM1. Murai MJ et al. 2014
- E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
A third solution structure of this interface exists: PDB 6EMQ, the LEDGF/p75 IBD with MLL1 residues 111-160, deposited by the CK2 phospho-switch study.
PDB 6EMQ: Solution structure of the LEDGF/p75 IBD - MLL1 (aa 111-160) complex · Sharma S, Cermakova K, De Rijck J, et al. · RCSB Protein Data Bank 2018What this evidence cannot show
Fetched from data.rcsb.org 2026-09-20. Like 2MTN it is a single chimeric chain annotated to both O75475 and Q03164, so it is a tethered construct; its MLL1 span 111-160 covers both IBM1 and IBM2, unlike 2MSR's 140-160.
Evidence that argues against this (1)
- Three of the four structures of this interface are engineered single chains and the only bimolecular one, 2MSR, contains only IBM2 - so no unlinked structure of the full bipartite interface exists. Sharma S et al. 2018
LEDGF IBD 349-428 receives KMT2A IBM1 and IBM2
KMT2A IBM1 124-133 and IBM2 147-152
supplies the Kd that entry records as unobtainable, and shows the motif is bipartite
CK2 sites S136/S142 lie in the IBM1-IBM2 linker
both motifs retained in the index fusion
menin occupancy raises IBD affinity about tenfold (14.7 uM to 1.4 uM)
supplies residue-level and affinity detail the dossier lacked
integrase occupies the IBM2 site and displaces the menin-MLL complex, IC50 301 nM
the search behind the LEDGF affinity correction
the tenfold cooperativity predicts this direction
the hand-check case: both IntAct records are flagged MI:1060 and one is MI:0407 direct interaction by x-ray diffraction, so the naive flag-only test misclassified this interface