Oncofusion

start / graph / interface-menin-suv39h1-svid-360-445

Interfacemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q34-menin-surface-map

The SvID, menin 360-445: a truncation boundary the field reads as a domain, whose functional residues D418 and W436 are 10-20 A from every menin drug - but whose first ten residues ARE the drug pocket

also called SvID · SUV39H1 interaction domain · menin 360-445

Menin hands the repressive enzyme SUV39H1 to DNA using a stretch of itself called the SvID. This graph has been using that stretch as proof that a menin drug cannot disturb the enzyme - but the first ten residues of the SvID are part of the drug pocket, so the proof needs to be about individual residues, not about the whole stretch.

The same thing, for a scientist

Yang et al. (Cell Death Dis 2013) mapped the menin-SUV39H1 contact by an N-terminal truncation series in 293T: menin lacking the first 360 residues still binds SUV39H1, menin lacking the first 445 does not, and they named the intervening 360-445 window the SvID. Of four MEN1 patient missense mutants inside it - A411P, L413P, D418N, W436R - only D418N and W436R lose binding. The authors concluded, from their truncations plus the then-new menin/MLL co-structure, that distinct domains mediate SUV39H1 and MLL binding. **That conclusion is right about the residues that matter and wrong about the window they drew around them.** Measured here over all 69 menin deposits: D418 and W436 sit 15.5-20.0 A from every bound menin inhibitor and peptide, on helices 407-424 and 434-445, and W436 instead sits 7.5-8.1 A from histones H3 and H4 in the menin-nucleosome cryo-EM structure 8GPN - i.e. on the nucleosome-binding face. But menin 363, 366, 367 and 371, the SvID's own N-terminal edge, line the MBM cavity at 2.6-4.3 A from ziftomenib, revumenib, DS-1594b, the JUND peptide and the KMT2A chain, and Huang et al. report that the E366A/D370A double substitution reduces KMT2A binding. The SvID therefore BEGINS inside the drug pocket and ENDS on the nucleosome face.

What it looks like

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.

Go deeper
What the 2013 mapping establishes, exactly

Quoting the source rather than paraphrasing it, because the graph has been carrying the paraphrase: "The N-terminal truncation mutants maintained their interaction with SUV39H1 until they lost almost more than a half of the protein from the N-terminus, whereas further deletion beyond amino acid 360 (dN445) failed to interact with SUV39H1, indicating that the region encompassing amino acids 360-445 (SvID; SUV39H1 interaction domain) is involved in SUV39H1 interaction." 1

Three things follow that the name "domain" obscures:

  1. It is an N-terminal truncation series, so it establishes necessity, not sufficiency. No construct consisting of 360-445 alone was shown to bind SUV39H1. A 86-residue window between the last binding construct and the first non-binding one is a bracket, not a domain.
  2. The only residue-level evidence is at 418 and 436. A411P and L413P, also inside the window, do not lose binding - so two of the four patient mutants tested are negatives, and they bracket the positives.
  3. The authors' own "distinct domains" sentence is an inference from a figure comparison, not a competition experiment. There is no SUV39H1-versus-MLL competition assay, no affinity, and no structure of menin with SUV39H1 anywhere.
Where the SvID actually is, in three dimensions

Measured over all 69 O00255 deposits with auth numbering proved against the live UniProt sequence at eight positions per deposit 3:

menin residuein SvIDmin distance to revumenib (7UJ4)to ziftomenib (8VA6)to KMT2A 1-160 (3U88)to histone H3 (8GPN)
I360boundary7.36.25.18.5
E366yes2.92.62.77.9
D370yes7.15.82.9-
V371yes4.32.33.714.0
A411yes16.516.317.824.5
L413yes13.012.412.8-
D418yes16.616.117.517.0
W436yes10.910.011.27.5
G445boundary20.519.616.420.8

And independently, UniProt curates from Huang et al. that the menin E366A/D370A double substitution reduces KMT2A interaction 2 - two SvID residues whose only published functional assignment is to the MLL contact.

So the SvID spans two different surfaces of menin. Its first ~12 residues are pocket wall. Its last ~20 residues (428-445, including W436) are the nucleosome grip that contacts H3, H4 and H2B in 8GPN. In between sits the helix 407-424 carrying D418, which faces neither.

What this changes for the graph's protective argument

gene-suv39h1 currently states: "It does not follow that a menin inhibitor competes SUV39H1 off menin: on this mapping the inhibitor's pocket and the SvID are different surfaces."

The conclusion survives; the reason does not. The pocket and the SvID are not different surfaces - they overlap at menin 360-371. What is true, and is the sentence the graph should carry, is narrower and stronger: the two SvID residues whose mutation is known to cost SUV39H1 binding, D418 and W436, are 10.0-16.6 A from every clinical menin inhibitor and are not part of the pocket in any of the five inhibitor co-crystals measured. That is a residue-level claim, it is directly checkable, and it does not depend on the word "domain".

This matters because the graph reuses the domain-level form. Anything that inherits "the SvID is a different surface from the MLL cavity" inherits a statement that is false at the SvID's own boundary. Reported to the coordinator rather than edited here.

Facts and figures
partner agene-men1
partner bgene-suv39h1
motif aSvID, menin 360-445 (O00255 610-aa numbering), defined as the bracket between a binding dN360 construct and a non-binding dN445 construct
motif bnot mapped; no region of SUV39H1 has been assigned to the menin contact in any source read here
interface residues a noteThese are the two patient missense positions whose substitution costs SUV39H1 binding. A411P and L413P, inside the same window, do NOT. No alanine scan and no structure exists, so 'interface residue' here means 'mutation-sensitive position', which is a weaker claim.
kd statusNEVER MEASURED for menin-SUV39H1 by any method.
method that mapped itN-terminal truncation series by co-IP in 293T, plus four patient missense mutants tested by in-vitro-translation pulldown
ligand noteNo clinical menin inhibitor comes within 9.9 A of D418 or W436. All of them come within 2.6-8.0 A of E366, which is inside the SvID as drawn.
druggabilityNot assessable as a target and not intended as one. The value of this window is as a NEGATIVE control region for menin-inhibitor selectivity, and that use requires quoting the residues rather than the window.
shown for fusionNO for any fusion, in either direction. The menin-SUV39H1 axis is 293T overexpression plus four endocrine replications (mouse islet tumour, human NET lines, rat Ins1, human fetal pancreatic progenitors). No KMT2A-rearranged cell, no leukaemia, no B-lineage cell.
epigenetic readoutdelivers the H3K9me2/me3 writer SUV39H1 to menin-occupied promoters; loss of menin lowers global H3K9 methylation in Men1-/- MEFs
cell typeHEK293T (mapping); Men1-/- mouse embryonic fibroblasts (global marks); mouse GBX2 promoter work in MEFs
relationship to existing entries`gene-suv39h1` E3 already carries the 2013 mapping and is the entry this one is anchored to. This entry adds what that one could not: the three-dimensional position of the window relative to the drug pocket and the nucleosome, measured rather than inferred, and the finding that the window's N-terminal boundary is pocket wall. It therefore WEAKENS the reason given in `gene-suv39h1`'s 'menin connection' section while leaving that section's conclusion standing on a narrower basis. Reported to the coordinator; not edited here.
domains
SvID (SUV39H1 interaction…360.0SvID (SUV39H1 interaction domain): 360SvID N-terminal edge that…360.0SvID N-terminal edge that lines the MBM cavity: 360SvID helix carrying D418407.0SvID helix carrying D418: 407SvID C-terminal helix car…434.0SvID C-terminal helix carrying W436, on the nucleosome face: 434
interface residues a
[
  "menin D418",
  "menin W436"
]
interface residues b
[]
domains — the numbers behind the chart
[
  {
    "name": "SvID (SUV39H1 interaction domain)",
    "start": 360,
    "end": 445
  },
  {
    "name": "SvID N-terminal edge that lines the MBM cavity",
    "start": 360,
    "end": 371
  },
  {
    "name": "SvID helix carrying D418",
    "start": 407,
    "end": 424
  },
  {
    "name": "SvID C-terminal helix carrying W436, on the nucleosome face",
    "start": 434,
    "end": 445
  }
]
structures
[]
structures absent
[
  "no PDB deposit contains menin with SUV39H1",
  "no PDB deposit contains SUV39H1 with any menin-family protein",
  "SUV39H1 itself has one deposit in this graph's record, 3MTS, and it does not contain menin"
]
distance from drug site angstrom
{
  "D418_to_revumenib_7UJ4": 16.57,
  "D418_to_ziftomenib_8VA6": 16.08,
  "D418_to_DS1594b_8IG0": 16.07,
  "D418_to_M89_6E1A": 15.52,
  "D418_to_bleximenib_9WKU": 19.52,
  "W436_to_revumenib_7UJ4": 10.86,
  "W436_to_ziftomenib_8VA6": 9.96,
  "W436_to_DS1594b_8IG0": 9.9,
  "W436_to_M89_6E1A": 11.05,
  "W436_to_bleximenib_9WKU": 14.25,
  "E366_to_ziftomenib_8VA6": 2.6,
  "V371_to_ziftomenib_8VA6": 2.33,
  "script": "analyses/q34-menin-surface-geometry/run.py",
  "measured": "2026-09-22"
}
ligands
[
  "drug-revumenib",
  "drug-ziftomenib",
  "drug-bleximenib"
]
How this could bite us back
  • The protective argument this graph makes from this interface is only as good as its narrowest form. If it is stated as 'the SvID is a different surface from the drug pocket', it is false at residues 360-371; if it is stated as 'D418 and W436 are 10-17 A from every clinical inhibitor', it holds. A reader who quotes the first form will over-claim safety.
  • Distance is not non-competition. Menin's pocket ligands are large and the protein has documented conformational plasticity; an allosteric effect on a helix 10 A away is not excluded by a distance measurement and has never been tested by a co-IP of endogenous SUV39H1 under drug.
  • The whole axis is endocrine. Every replication of menin-SUV39H1 is in islet, NET or pancreatic material, so importing it into a leukaemia argument is the partner-import failure this graph documents as its commonest synthesis error.
  • A menin degrader removes this surface along with everything else, so any argument that depends on SUV39H1 delivery surviving therapy is an argument for the inhibitor and against the degrader.
Still open
  1. Does a menin dN360 construct - the last one that still binds SUV39H1 - fold at all? Menin is a compact single domain and removing 360 N-terminal residues removes the entire thumb and most of the palm, so a truncation series through a globular protein may be reporting stability rather than interface.
  2. Is the SvID sufficient? Nobody has expressed menin 360-445 alone and asked whether it binds SUV39H1. That is the experiment that converts a bracket into a domain.
  3. Do D418N and W436R lose SUV39H1 binding because they remove a contact or because they destabilise helices 407-424 and 434-445? A thermal-shift assay on the two mutants would separate these in an afternoon and has never been reported.
  4. Does W436R alter nucleosome binding? W436 is 7.5 A from H3 in 8GPN, so a patient mutation catalogued as a SUV39H1-binding mutant may in fact be a nucleosome-binding mutant, and the two have never been tested side by side.
Every source behind this entry (4)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In 293T cells, an N-terminal truncation series maps the menin-SUV39H1 interaction to menin residues 360-445, which the authors name the SvID; of four MEN1 patient-derived missense mutants inside that window - A411P, L413P, D418N and W436R - only D418N and W436R are impaired in SUV39H1 binding, and the authors infer from their mapping plus the published menin/MLL co-structure that distinct domains mediate SUV39H1 and MLL binding, the MLL surface being the central cavity.

    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 2013
    PMID 23579270DOI
    What this evidence cannot show

    Overexpression co-IP in 293T with Myc-menin and GFP-SUV39H1, plus in-vitro-translated 35S-menin mutants against partially purified FLAG-SUV39H1 - no endogenous co-IP of the mapped region, no affinity, no structure, no competition with MLL or with any inhibitor. The 360-445 window is a bracket between the last binding truncation and the first non-binding one, so it establishes necessity and not sufficiency. Two of the four patient mutants tested were negative. The 'distinct domains' sentence is an inference from comparing their figure with a published structure, not an experiment. Full text read from Europe PMC fullTextXML (51,918 characters) on 2026-09-22.

    Evidence that argues against this (1)
    • UniProt curates from Huang et al. 2012 that a menin E366A/D370A double substitution REDUCES KMT2A interaction. Residues 366 and 370 are inside the SvID as drawn, so at least two residues of the 'SUV39H1 domain' have a published functional assignment to the MLL contact instead. The 'distinct domains' inference is therefore falsified at the SvID's own N-terminal boundary. UniProt Consortium 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    UniProt O00255 carries nine curated menin mutagenesis features affecting KMT2A binding, all traced to PMID 22327296: A182F, M278W, D285R/E288R/E290R, Y319A, Y323A, and E366A/D370A - so the experimentally mutated menin residues implicated in the KMT2A contact run from 182 to 370 and reach into the SvID window.

    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 2012
    PMID 22327296DOI
    What this evidence cannot show

    RULE 91 APPLIES AND IS NOT WAIVED HERE. These are UniProt's curated features citing PMID 22327296; the E366A/D370A pair does NOT appear in the readable body text of that paper, whose ITC mutant panel is in Supplementary Fig. 6. The body text was fetched and read (41,215 characters, via efetch db=pmc after Europe PMC fullTextXML returned HTTP 500) and confirms the pocket residues Y319, Y323 and M278 by name but not 366 or 370. So the 366/370 assignment is a curator's reading of a supplementary panel that I could not open, and it is recorded at 'strong' rather than 'established' for that reason. The paper's own domain nomenclature - N-terminal domain, thumb, palm, fingers - is defined in the text without numeric boundaries; those are in Supplementary Fig. 4 and were likewise not retrievable, which is why no thumb/palm/fingers coordinates appear anywhere in this entry.

    Evidence that argues against this (1)
    • A double-alanine substitution reported as 'reduced interaction' is a weaker claim than the single-residue losses in the same panel (M278W, complete loss). If the 366/370 effect is small, the overlap between the SvID boundary and the pocket is geometric rather than functional - and the geometry is not in dispute, since the same residues are 2.6-2.9 A from bound ligand. RCSB Protein Data Bank depositors 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    Across the menin deposits, the SUV39H1-critical residues D418 and W436 are observed in 69 of 69 and lie 15.5-16.6 A (D418) and 9.9-11.1 A (W436) from revumenib, ziftomenib, DS-1594b and M-89, and 17.5 A and 11.2 A from the bound KMT2A chain in 3U88; in the menin-nucleosome cryo-EM structure 8GPN, W436 is 7.5 A from histone H3 and 8.1 A from H4 while menin 428-434 and 441-442 contact H3, H4, H2B and nucleosomal DNA. By contrast the SvID's N-terminal residues 363, 366, 367 and 371 are 2.3-4.3 A from bound ligand.

    What this evidence cannot show

    Computed 2026-09-22 against live RCSB mmCIF and live UniProt; minimum heavy-atom distances per chain pair. 8GPN is a canonical H3K79me2 nucleosome containing H2A type 1, NOT H2A.X, so it says where menin grips a nucleosome and nothing about gamma-H2A.X. A distance is not an affinity: 10 A is not a guarantee of non-competition, because allostery and loop closure are not distances. In 8GPN menin's contact with H2A is a single residue (577), versus 12 with H3, 9 with H4 and 16 with H2B.

    Evidence that argues against this (1)
    • No structure exists of menin with SUV39H1, so every statement here about where the SUV39H1 contact sits is a statement about where two patient mutations sit, not about where an interface sits. A single mutation can cost binding by destabilising a fold rather than by removing a contact - and W436R and D418N are exactly the class of substitution (buried Trp to Arg, buried Asp to Asn) where that alternative is live. UniProt Consortium 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A second MEN1 missense mutant inside the SvID, T429K, reduces menin's binding to DAXX and the pair's repression of MME - so the SvID window has at least two partner assignments from patient mutations, SUV39H1 at 418/436 and DAXX at 429.

    PMID 27872097DOI
    What this evidence cannot show

    Cited at ABSTRACT level: the efetch abstract was fetched and read on 2026-09-22 and states the T429K, DAXX-binding and MME results verbatim; the full text was not opened, so nothing here rests on a figure. RECORDED AS A CAUGHT ERROR: the first draft of this citation carried the DOI 10.1158/0008-5472.CAN-16-0655, typed from assumption while the correct value 10.1158/0008-5472.CAN-16-1567 sat in a record fetched minutes earlier. Title, authors, venue and year were all correct, which is exactly the shape the brief warns about - a misassembled citation looks right.

    Evidence that argues against this (1)
    • The more partners a coarse truncation window collects, the less it behaves like a domain and the more it behaves like 'the half of the protein that was left'. Three partners (SUV39H1, DAXX, and 371-390 overlapping the curated FANCD2 region) assigned to one 86-residue stretch by three different coarse methods is a pattern that argues for re-mapping, not for a multi-functional hub. UniProt Consortium 2026
Connected to