The menin acidic shelf, Glu366-Asp370-Val371: the counter-charge that MBM2 binds, resolved in 3U88 after all - and the one patch where the drug pocket, the SvID and the gamma-H2A.X motif all touch
also called menin E366/D370 acidic residues · MBM2 counter-surface · menin 363-378 shelf
There is a small negatively charged ledge on menin, just outside the pocket the drugs sit in. Two arginines from KMT2A land on it, and the same ledge is the first part of two other menin regions that other papers describe as separate. It is the single place where four different stories about menin overlap.
The same thing, for a scientist
Menin residues 363-378 form a shallow acidic shelf at the rim of the central MBM cavity. Four independent lines converge on it and none of the four source papers cites the others. (1) Huang et al. state that MLL1 Arg24/Arg25 are disordered in the 21-mer MBM co-crystal 3U85 yet cost 21-fold when charge-reversed, that the pocket's acidic residues are positioned to meet them, and that mutating those menin acidic residues also reduces binding; UniProt curates those residues as E366 and D370. (2) In 3U88, whose KMT2A chain is modelled at residues 6-15, 23-39 and 107-135, KMT2A Arg24 is 2.9-3.2 A from menin Asp370 and 3.0-3.5 A from Glu366, and KMT2A Leu27 packs against menin 371/374/375/378 - so MBM2 IS resolved in a deposit, which this graph's `interface-menin-kmt2a-mbm2` entry records as not being the case. (3) The same shelf is 2.3-4.3 A from ziftomenib, revumenib, DS-1594b and M-89, and 2.6-4.0 A from the JUND peptide. (4) Menin 360-371 is the N-terminal end of the SvID and 371-378 is the N-terminal end of the gamma-H2A.X motif. The shelf is therefore the physical reason the 'distinct surfaces' framing keeps failing.
No three-dimensional model is available for this protein, so the map above is the whole picture. That is itself informative: AlphaFold does not model chains this long, and the parts that have been crystallised are held under the entries for the individual domains.
MBM2 is resolved, in 3U88, and the graph says it is not
interface-menin-kmt2a-mbm2 states that "no deposit in the PDB is annotated as a menin-MBM2 complex", that "Arg24/Arg25 are reported disordered", and asks as an open question whether MBM2 can be crystallised at all. The first clause remains true as a statement about annotation. The biology clause is not.
Measured in 3U88 (X-ray, 3.0 A, two menin copies A/B with KMT2A chains M/N) 2:
| menin | KMT2A | min heavy-atom distance, copy A | copy B |
|---|---|---|---|
| Glu366 | Thr15 | 2.7 A | 2.7 A |
| Glu366 | Arg24 | 3.0 A | 3.5 A |
| Asp370 | Arg24 | 3.2 A | 2.9 A |
| Val371 | Thr15 | 3.7 A | 4.1 A |
| Val371 | Arg24 | 4.0 A | 3.9 A |
| Val371 | Leu27 | 3.7 A | 4.2 A |
| 374 / 375 / 378 | Leu27 | 4.1 / 3.7 / 4.3 A | 4.2 / 3.9 / 5.0 A |
The KMT2A chain in 3U88 is modelled at residues 6-15, 23-39 and 107-135, with the sequence read back against Q03164 (R6 W7 R8 F9 P10 A11 R12 P13 = RWRFPARP). In 3U85, the 21-mer MBM peptide deposit from the same paper, only KMT2A 5-13 is modelled - which is where the "disordered" statement comes from, and it is a statement about the short peptide, not about the protein.
So the 26-fold-weaker second motif of the menin grip has a structure, in the deposit this graph already cites, and its counter-surface is the acidic shelf. Reported to the coordinator as an amendment owed to interface-menin-kmt2a-mbm2 (the fields structures_absent, structure_details.3U88 and the third open question).
Four descriptions of one patch
| source | what it calls menin 363-378 | year | cites the others? |
|---|---|---|---|
| Huang et al. 1 | the acidic residues of the MBM-binding pocket that meet MLL1 24-RR-25 | 2012 | – |
| Yang et al. 3 | the N-terminal boundary of the SvID, a surface distinct from the MLL cavity | 2013 | cites the menin/MLL structure, not these residues |
| Lian et al. 4 | the start of the Val371-Gln400 gamma-H2A.X motif, in the Palm and Fingers | 2026 | cites Yang et al. as ref 7 in a three-reference bundle about H3K9me3, and never names SUV39H1, the SvID, MBM, the cavity or any menin inhibitor |
| five inhibitor co-crystals 2 | pocket wall, 2.3-4.3 A from bound drug | 2018-2026 | – |
The coordinator's brief said neither source paper mentions the other. That is half right and the half that is wrong is worth knowing: PMID 42578372 does cite PMID 23579270, as reference 7, inside the bundle "[6-8]" supporting the sentence that menin interacts with epigenetic writers to orchestrate H3K4me3, H3K9me3 and H3K36me3. It cites it and discusses nothing about it. The 2013 paper of course cannot cite the 2026 one.
The pocket lining, for reference
Menin residues within 4.5 A of the bound species, measured per chain pair 2:
- revumenib (7UJ4, 1.96 A), 23 residues: 155, 177-182, 238, 241, 242, 276, 278, 285, 319, 322, 323, 325, 326, 330, 341, 363, 366, 367, 371
- KMT2A MBM1 12-mer (4GQ6, 1.55 A), 29 residues: 135-139, 153-155, 177-182, 238, 241, 242, 244, 246-249, 278, 279, 282, 285, 319, 322, 323, 359, 363
The two lists share 16 residues and differ at both ends: the peptide reaches further into the 135-155 arm, the drug reaches further onto the 325-371 shelf. That asymmetry is the structural statement behind every "does the drug displace partner X" question in this graph, and it has not previously been written down here in residue form.
| partner a | gene-men1 |
|---|---|
| partner b | gene-kmt2a |
| motif a | menin acidic shelf, residues 363-378, at the rim of the central MBM cavity; the acidic pair is Glu366 and Asp370 |
| motif b | KMT2A MBM2, residues 23-40, whose Arg24, Arg25 and Leu27 are the contacting residues resolved in 3U88 |
| kd NM | 1400 |
| kd note | 1,400 nM is the fluorescence-polarisation Kd of the isolated KMT2A 23-40 (MBM2) peptide reported by Grembecka et al., carried here from `interface-menin-kmt2a-mbm2` for context; no affinity has been measured for the menin side of this shelf. |
| ligand note | revumenib, ziftomenib, DS-1594b and M-89 all contact the shelf (E366 at 2.6-5.3 A); bleximenib does not reach it (E366 8.0 A, V371 10.7 A). |
| druggability | Already drugged, incidentally. Three of the four clinical compounds extend onto this shelf; one does not. That is a measurable selectivity axis that nobody has used, and it is the concrete version of the partner-selectivity idea in `hyp-partner-selective-menin-ligand-separates-mandates`. |
| fate in fusion | KMT2A MBM2 (aa 23-40) lies ~1,360 residues N-terminal to the breakpoint cluster region, so the fusion retains it exactly as it retains MBM1 |
| shown for fusion | The KMT2A side is shown only with isolated KMT2A fragments and bacterially expressed menin, never with a fusion protein. No fusion co-structure of any kind exists. |
| relationship to existing entries | Complements `interface-menin-kmt2a-mbm1` and `interface-menin-kmt2a-mbm2`, which describe the grip from the KMT2A side and carry no menin-side residues; and `interface-menin-jund-pocket`, which describes the same cavity for JUND. It CORRECTS one factual field in `interface-menin-kmt2a-mbm2` (MBM2 is resolved in 3U88) and supplies the menin-side residue lists none of the three had. Reported to the coordinator; those entries were not edited. |
interface residues a
[ "menin E366", "menin D370", "menin V371", "menin 374", "menin 375", "menin 378", "menin 363", "menin 367" ]
interface residues b
[ "KMT2A T15", "KMT2A R24", "KMT2A L27" ]
domains
[
{
"name": "menin acidic shelf (MBM2 counter-surface)",
"start": 363,
"end": 378
}
]structures
[ "3U88", "3U85", "3U86", "4GQ6", "7UJ4", "8VA6", "8IG0", "6E1A", "9WKU" ]
structure details
[
{
"pdb": "3U88",
"title_fetched": "Crystal structure of human menin in complex with MLL1 and LEDGF",
"method": "X-ray",
"resolution_A": 3,
"kmt2a_modelled_residues": "6-15, 23-39, 107-135",
"finding": "MBM2 (KMT2A 23-39) is ORDERED here; R24 contacts menin D370 and E366",
"fetched": "2026-09-22"
},
{
"pdb": "3U85",
"title_fetched": "Crystal structure of human menin in complex with MLL1",
"method": "X-ray",
"resolution_A": 3,
"kmt2a_modelled_residues": "5-13",
"finding": "the deposit behind the 'R24/R25 disordered' statement - the peptide simply does not extend that far in the model",
"fetched": "2026-09-22"
}
]pocket lining menin residues
{
"revumenib_7UJ4_within_4_5A": [
155,
177,
178,
179,
180,
181,
182,
238,
241,
242,
276,
278,
285,
319,
322,
323,
325,
326,
330,
341,
363,
366,
367,
371
],
"kmt2a_mbm1_12mer_4GQ6_within_4_5A": [
135,
136,
137,
138,
153,
154,
155,
177,
178,
179,
180,
181,
182,
238,
241,
242,
244,
246,
247,
248,
249,
278,
279,
282,
285,
319,
322,
323,
359,
363
],
"note": "16 shared; the drug reaches further onto the 325-371 shelf, the peptide further into the 135-155 arm",
"measured": "2026-09-22",
"script": "analyses/q34-menin-surface-geometry/run.py"
}ligands
[ "drug-revumenib", "drug-ziftomenib", "drug-bleximenib" ]
- This patch is where a selectivity argument could be built and also where one could quietly fail. Three clinical compounds engage it and one does not, so any statement of the form 'menin inhibitors do/do not disturb partner X' that is made at class level is unsupported at residue level for at least one member.
- If MBM2 contributes real affinity through this shelf, then a compound that occupies the shelf competes against a tighter ligand than the MBM1 peptide used in most competition assays, and reported biochemical potencies understate what cellular displacement requires. That is the same argument `interface-menin-kmt2a-mbm2` makes, now with a structure behind it.
- The shelf is shared with JUND at 2.6-4.0 A, so any chemistry designed to gain affinity here gains it against the resistance-driving partner too.
- Resistance: the known MEN1 resistance substitutions cluster in the pocket (M327I, G331R, T349M, all with deposited structures). A compound that extends onto the shelf adds contacts at 363-371, residues where no resistance mutation has yet been reported - which is an opportunity and also a prediction that can be wrong.
- Does bleximenib, which stops 8 A short of the shelf, differ from revumenib and ziftomenib in displacing the intact KMT2A N-terminal fragment rather than the MBM1 peptide? A competition FP with MLL 1-160 rather than MLL 4-15 would show it and is a one-week experiment with existing reagents.
- Is the Arg24-Asp370 pair a salt bridge or a rotamer at 3.0 A? A higher-resolution menin structure with a KMT2A chain longer than 15 residues does not exist, and nothing else can settle it.
- Why has no group re-refined or re-annotated 3U88 as an MBM2 complex in fourteen years, given that the affinity argument for a second motif was published in 2010 and the coordinates have been public since 2012?
- Does the shelf move? Every structure here is menin plus a short ligand. If 366-378 reorganises on nucleosome or gamma-H2A.X engagement, a distance measured in a peptide co-crystal does not bound what a drug can perturb.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Huang et al. report that the C-terminal arginine pair of MLL1 MBM, Arg24-Arg25, is disordered in the menin-MBM peptide crystal yet costs 21-fold in affinity when substituted with two glutamates; that the acidic residues of the MBM-binding pocket are positioned to make electrostatic contact with them; and that mutation of those menin acidic residues also decreased binding - which UniProt curates as the menin E366A/D370A double substitution reducing KMT2A interaction.
The same pocket in menin binds both MLL and JUND but has opposite effects on transcription. · Huang J, Gurung B, Wan B, Matkar S, Veniaminova NA, Wan K, Merchant JL, Hua X, Lei M · Nature 2012What this evidence cannot show
Full text read on 2026-09-22 (41,215 characters, via efetch db=pmc after Europe PMC fullTextXML returned HTTP 500). The body text does NOT name residues 366 or 370; it says 'these menin acidic residues' and points at Supplementary Fig. 6, which was not retrievable. The identification of those residues as E366 and D370 therefore comes from UniProt's curated mutagenesis features, i.e. a curator who read that supplementary panel - rule 91's exact situation, recorded rather than smoothed over. The paper's own thumb/palm/fingers domain boundaries are likewise in a supplementary figure and are consequently not quoted anywhere in this beat.
Evidence that argues against this (1)
- The E366A/D370A effect is described as 'reduced', and the charge-reversal on the KMT2A side (R24E/R25E) is a harsher perturbation than an alanine pair, so the shelf could be a soft electrostatic steering surface rather than a discrete sub-site. That reading is compatible with everything measured here and would mean a drug touching Val371 is touching a low-energy periphery. Grembecka J et al. 2010
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
In PDB 3U88 the KMT2A chain is modelled at residues 6-15, 23-39 and 107-135 (identities read back against Q03164), and KMT2A Arg24 lies 2.9-3.2 A from menin Asp370 and 3.0-3.5 A from menin Glu366 in both crystallographic copies, while KMT2A Leu27 packs at 3.7-5.0 A against menin 371, 374, 375 and 378 - so MBM2 is ordered and its menin counter-surface is the acidic shelf. In the same measurement, menin 363/366/367/371 lie within 4.5 A of revumenib (7UJ4), ziftomenib (8VA6), DS-1594b (8IG0) and M-89 (6E1A), and menin 366/370/371 lie within 4.0 A of the JUND peptide (3U86); in 3U85, the 21-mer MBM peptide deposit, only KMT2A 5-13 is modelled.
analyses/q34-menin-surface-geometry/run.py - menin surface geometry over all 69 O00255 PDB deposits · Q34-menin-surface-map · Oncofusion Knowledge Graph 2026What this evidence cannot show
Computed 2026-09-22 against live RCSB mmCIF and live UniProt. 3U88 is 3.0 A, which is low enough that individual side-chain rotamers in a solvent-exposed salt bridge should be treated as a model rather than a measurement; the Arg24-Asp370 pairing is consistent between the two independent copies, which is the available internal control but is not independent evidence. Distances are minimum heavy-atom, hydrogens absent, computed per chain pair so no crystal-packing contact between copies is counted. The 3U88 KMT2A entity is a 75-residue ENGINEERED chain that splices three separated segments of Q03164, so its residues 15 and 23 are adjacent in the construct and not in the protein - a single-chain construct models the bound state rather than measuring association.
Evidence that argues against this (1)
- This graph's own `interface-menin-kmt2a-mbm2` reaches the opposite conclusion from the same deposit, recording that presence in the 3U88 construct 'is not the same as being resolved' and that Arg24/Arg25 are disordered. That entry is right about 3U85 and about the annotation, and the disagreement is entirely about whether anyone looked at 3U88's coordinates - which is the check this entry ran. RCSB Protein Data Bank depositors 2026
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The SvID's N-terminal boundary is menin residue 360, set by an N-terminal truncation series in 293T, and the authors inferred from that mapping plus the menin/MLL co-structure that distinct domains mediate SUV39H1 and MLL binding.
Menin mediates epigenetic regulation via histone H3 lysine 9 methylation. · Yang YJ, Song TY, Park J, Lee J, Lim J, Jang H, Kim YN, Yang JH, Song Y, Choi A, Lee HY, Jo CH, Han JW, Kim ST, Youn HD, Cho EJ · Cell Death Dis 2013What this evidence cannot show
The inference is not wrong about SUV39H1 - D418 and W436 really are 10-17 A from every drug - but it is wrong about the window, because 360-371 is pocket wall. Overexpression co-IP, no affinity, no structure, no competition assay. Full text read on 2026-09-22.
Evidence that argues against this (1)
- Huang et al. published the menin/MLL structure in the year before, and it is the structure Yang et al. cite; the E366A/D370A result establishing that two SvID residues serve the MLL contact is in that same paper's supplementary data. The information needed to refute the distinct-domains inference was available at the moment it was made. Huang J et al. 2012
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
PMID 42578372, which defines the menin Val371-Gln400 gamma-H2A.X motif, does cite PMID 23579270 - as reference 7, inside the bundle [6-8] supporting a sentence about menin interacting with epigenetic writers to orchestrate H3K4me3, H3K9me3 and H3K36me3 - and it never uses the words SUV39H1, SvID, MBM, cavity, pocket, revumenib or menin inhibitor anywhere in its full text.
The tumor suppressor menin is a key scaffold mediator for homologous recombination repair. · Lian A, Zhu X, Jin B, Gu G, Xu B, Zhang L, Lin X, Zhu X, Jin G · Nucleic Acids Res 2026What this evidence cannot show
Absence established by case-insensitive regex over the 96,530-character Europe PMC fullTextXML flattened text, fetched 2026-09-22. That covers body, figure legends and reference titles but not supplementary material or figure images, so it bounds what the paper SAYS and not what its authors knew. The citation of ref 7 is a bundle citation and carries no discussion.
Evidence that argues against this (1)
- A bundle citation is weak evidence of engagement, but it is not zero: the group's own prior paper (ref 8, PMID 39817905) maps a menin-SETD2 contact to the same NTD and Palm region, so this group has mapped two partners to overlapping menin territory and has not remarked on the overlap either. Zhu X et al. 2025
menin 363-378, the acidic shelf at the rim of the MBM cavity
supplies the menin counter-surface for MBM2 and shows KMT2A 23-39 is ordered in 3U88
menin E366/D370 meet KMT2A Arg24; menin 371-378 pack against KMT2A Leu27
the same shelf is 2.6-4.0 A from the JUND peptide in 3U86
revumenib contacts menin E366 at 2.9 A and V371 at 4.3 A in 7UJ4
bleximenib does NOT reach the shelf - E366 8.0 A, V371 10.7 A in 9WKU
a measured 8 A spread across the class at one residue is the concrete selectivity axis that hypothesis predicts
the shelf is the physical reason the distinct-surfaces framing fails
places E366/D370/V371 on the fingers side of the deepest seam in menin, not in the palm