The menin small-molecule pocket as a coordinated surface: four clinical-stage compounds, 21-26 contact residues each, and bleximenib sitting on a different part of it from the other three

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.

• If a clinically meaningful part of the menin-inhibitor effect comes from the fingers-lobe rim rather than from KMT2A displacement, then optimising a next-generation compound purely on KMT2A-displacement potency selects against the very contacts that contribute, and a more potent compound can be a less effective one.
• The converse risk is a toxicity one. Contacting V371 at 2.3-4.3 A in a cell that needs homologous recombination - a normal haematopoietic progenitor, a gut crypt - is an on-target liability that dose reduction does not separate from efficacy, because it is the same molecule in the same pocket.
• Four compounds sharing a 15-residue core means a resistance mutation in that core plausibly cross-resists all four, and this graph already carries `resistance-cross-resistance-between-menin-inhibitors`. Switching within the class after failure is a weaker move than the number of available compounds suggests.
• Bleximenib's smaller footprint is not automatically the safer one. A compound that makes fewer contacts may need higher exposure for the same occupancy, and nothing here measures potency.

1. Do bleximenib and revumenib differ in any DNA-damage readout at matched target occupancy? Nothing has been measured. See `hyp-the-menin-inhibitor-class-splits-at-val371`.
2. Is the DS-1594b footprint's small NTD/palm-lobe count (3, against 5-8 for the other compounds) chemistry or is it the 2.60 A resolution? Re-deriving the footprint from a higher-resolution structure of the same chemotype would answer it.
3. Does the M327I resistance mutation - which this graph records for the class - lie in or near any of these footprints? Residue 327 is not in any of the five contact lists measured here, which is itself worth an entry, since a resistance mutation outside every contact list is an allosteric or folding claim rather than a steric one.
4. Does any compound contact W436 or D418, the two SUV39H1-binding-sensitive positions? None of the four does; the nearest published approach is 9.9 A. A compound that did would be a chemical tool for separating menin's SUV39H1 function, and none exists.

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

   Menin residues within 4.5 A of the bound ligand, computed per chain pair with every residue number verified against UniProt O00255, number 23 for revumenib in 7UJ4, 26 for ziftomenib in 8VA6, 21 for bleximenib in 9WKU, 21 for DS-1594b in 8IG0 and 29 for the KMT2A MBM1 peptide in 4GQ6; fifteen residues are common to all five; bleximenib's footprint reaches only to residue 363 and places 23.8% of its contacts in menin's fingers lobe against 43.5%, 46.2% and 47.6% for the other three compounds and 17.2% for the peptide.

   analyses/q36-menin-residue-axis/run.py - per-chain-pair ligand contact footprints and lobe assignment on RCSB deposits 7UJ4, 8VA6, 9WKU, 8IG0 and 4GQ6, run 2026-09-22 · RCSB PDB, computed 2026
   ▸What this evidence cannot show

   `established` for the geometry and nothing else - these are distances in deposited coordinates, which is about as direct as a measurement gets, but a 4.5 A cutoff is a convention and residues at 4.6 A are excluded by fiat. Resolutions differ (1.55-2.60 A) and rim contacts are the least reliable part of the lowest-resolution structure. 9WKU renumbers residues 458-489, which is outside the pocket; its numbering was read back residue-by-residue below 458 and is clean apart from a single engineered position at residue 5. Each footprint is from one deposit; no attempt was made to average over multiple structures of the same compound.

   ▸Evidence that argues against this (1)

   • A contact list is not a binding determinant. Alanine-scanning would very likely show that most of these 21-26 residues contribute little, and the acidic-shelf contacts specifically are to backbone and side chains at the rim where burial is partial. This entry records what touches what, and nothing about energetics. Huang J et al. 2012
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The cavity that all four compounds occupy is the same one that accepts the KMT2A MBM1 motif and the JunD menin-binding motif, established by isothermal titration calorimetry and co-crystal structures showing MLL1 residues 6-25 bind menin at 82 nM with Phe9 nested in a deep hydrophobic pocket formed by the thumb and palm, and JunD residues 27-47 binding the same site.

   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, et al. · Nature 2012
   PMID 22327296DOI
   ▸What this evidence cannot show

   That paper's pocket description is qualitative - 'formed by the thumb and palm' - with no residue boundaries for either domain anywhere in its main text. The coordinate parse in `ana-q36-menin-residue-axis` puts 10-12 of the small-molecule contacts on the fingers side of the deepest seam in the protein, so the canonical 'thumb and palm' description of the pocket is incomplete by that parse's boundaries. Read full text here via the PMC HTML route.

   ▸Evidence that argues against this (1)

   • If the coordinate parse's palm|fingers boundary is placed differently - and the one published table of menin domain coordinates would put it near residue 300 rather than 317 - then fewer or more pocket residues fall in the fingers lobe, and the 43-48% figures move. The fingers-lobe fractions are conditional on one parse and should be quoted with it. Nelakurti DD et al. 2020
3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   Bleximenib's nearest approach to menin V371 is 10.72 A and to E366 is 8.02 A in 9WKU, whereas DS-1594b reaches V371 at 3.11 A and E366 at 3.67 A in 8IG0; so the clinical menin-inhibitor class is not uniform at the rim residues that begin the gamma-H2A.X motif.

   analyses/q36-menin-residue-axis/run.py - minimum heavy-atom distances from menin residues 366, 370 and 371 to bound ligands 9N6 (9WKU) and 7IX (8IG0), computed 2026-09-22 · RCSB PDB, computed 2026
   ▸What this evidence cannot show

   Independently reproduces the corresponding numbers in `ana-q34-menin-surface-geometry` (bleximenib-V371 10.72 A, DS-1594b-V371 3.11 A) from a separately written script, which is a check on the code rather than on the structures - both scripts read the same deposits. A single crystal form per compound; a different crystal form or a solution measurement could place the rim differently, and rim side chains are exactly where that is most likely.

   ▸Evidence that argues against this (1)

   • Distance to a residue is not evidence that the residue matters. The three compounds that contact V371 and the one that does not have not been compared in any assay that reads out the gamma-H2A.X motif's function, so the structural difference has no measured phenotype attached to it and may have none. Lian A et al. 2026

• part of →MEN1 / menin — the adaptor the fusion cannot work without

  the small-molecule pocket as a coordinated surface of menin

• ← bindsRevumenib (SNDX-5613, Revuforj) — first approved menin–KMT2A interface inhibitor

  23 menin residues within 4.5 A in 7UJ4

• ← bindsZiftomenib (KO-539, Komzifti) — once-daily menin inhibitor approved in NPM1-mutant AML

  26 menin residues within 4.5 A in 8VA6

• related to →The menin–KMT2A grip (MBM1), the interface the drugs actually block

  shares only 15 of 37 menin residues with the KMT2A MBM1 footprint, Jaccard 0.405

• depends on →A residue axis for menin, derived from coordinates: three spatial lobes, not four named ones, and the sharpest seam in the protein runs through the drug pocket

  lobe assignment of every pocket residue comes from that parse

• related to →Menin-JunD: the same pocket, the same drug, the opposite job

  the same cavity accepts JunD; this entry coordinates the small-molecule footprint specifically

• related to →Bleximenib (JNJ-75276617) — menin inhibitor with a cleaner cardiac profile, and the only one with a 2-year-old floor

  bleximenib does not reach V371 and so does not share the fingers-lobe edge of the footprint

• ← depends onThe menin-inhibitor class splits at Val371: three of four clinical compounds contact the first residue of the gamma-H2A.X motif and bleximenib does not, so DNA-repair effects should separate from KMT2A-displacement effects within the class

  rests on the measured footprints

• ← supportsA residue axis for menin, derived from coordinates: three spatial lobes, not four named ones, and the sharpest seam in the protein runs through the drug pocket

  supplies the lobe decomposition of every pocket residue