Menin's two nuclear localisation signals, 479-497 and 588-608: mapped in 1998, verified residue-for-residue today, annotated by UniProt nowhere - NLS-1 in no structure at all, NLS-2 resolved in one
also called menin NLS-1 · menin NLS-2 · RRRGPRRESKPEEPPPPKK · KKQKVSTPSDYTLSFLKRQRK
Menin carries two separate address labels that tell the cell to send it into the nucleus, both near its far end, and no protein database records either of them. The first has never been seen in any 3D structure and the second has been seen in just one of 69, so anything that cuts menin short leaves the protein stranded outside the nucleus where it cannot work.
The same thing, for a scientist
Guru et al. mapped two independent NLSs in the 610-residue menin protein by EGFP-tagged deletion analysis in HEK-293T: NLS-1 at residues 479-497 (RRRGPRRESKPEEPPPPKK, containing both a monopartite RRRGPRR at 479-485 and a candidate bipartite 484-497) and NLS-2 at 588-608 (KKQKVSTPSDYTLSFLKRQRK, with a candidate monopartite KRQRK at 604-608). Constructs truncated at residue 476 were cytoplasmic; extending to 571 restored nuclear localisation. Both sequences were re-derived from the live UniProt O00255 sequence today and match character for character. **UniProt annotates no NLS on this entry at all** - rule 91 names exactly this omission. Measured across all 69 menin deposits (68 with an identifiable menin chain) on 2026-09-22: **NLS-1 is modelled in none of them**, but **NLS-2 is modelled in 35**, and 3U84 chain A resolves 596-608 - 13 of NLS-2's 21 residues, including the candidate monopartite KRQRK at 604-608 named above, with deposited numbering confirmed against UniProt residue by residue. The other 34 clip only one or two residues past 588. So the correct statement is that NLS-1 exists in no coordinate set and NLS-2 is ordered in exactly one. The 1998 authors' own consequence still stands: of 43 known frameshift/nonsense MEN1 mutations, 39 would truncate menin before both NLSs and strand the product in the cytoplasm.
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.
The mapping, and the readback
| element | residues | sequence in the paper | sequence in live O00255 | match |
|---|---|---|---|---|
| NLS-1 | 479-497 | RRRGPRRESKPEEPPPPKK | RRRGPRRESKPEEPPPPKK | exact |
| NLS-1 monopartite candidate | 479-485 | RRRGPRR | - | within the above |
| NLS-1 bipartite candidate | 484-497 | RRESKPEEPPPPKK | - | within the above |
| NLS-2 | 588-608 | KKQKVSTPSDYTLSFLKRQRK | KKQKVSTPSDYTLSFLKRQRK | exact |
| NLS-2 monopartite candidate | 604-608 | KRQRK | KRQRK | exact |
The exact match at two 19-21 residue strings is strong evidence that Guru et al. used the same 610-residue numbering UniProt displays today, which is not free: menin has a longer 615-residue isoform and a five-residue offset would have shown up here 12.
Deletion logic, from the paper: EGFP-menin constructs truncated at residue 476 or shorter were cytoplasmic; a construct extending to 571 was nuclear, placing NLS-1 in the extra 95 residues. N-terminal deletions then identified NLS-2 within the C-terminal 39 residues. Constructs removing NLS-2 entirely, or either half of its bipartite candidate, still went to the nucleus because they retained NLS-1 - so the two are independent and NLS-1 is sufficient.
Two absences worth recording
UniProt annotates no NLS. The O00255 feature table carries exactly two Regions - 214-390 'Interaction with FANCD2' and 460-552 'Disordered' - and no Motif of any kind 2. The QUEEN_BRIEF's rule 91 uses this precise case as its worked example, and it remains true on the record fetched today. Any pipeline that reads menin's targeting signals out of UniProt gets nothing.
NLS-1 is in no structure; NLS-2 is in one. This entry previously said the largest modelled span reaches residue 587 (8GPN) and that both signals are outside every coordinate set. The second half is wrong, and it was wrong in the title as well as here. Every one of the 69 menin deposits was re-measured on 2026-09-22 5: 35 reach past 587, the maximum verified residue is 608 - menin's own C-terminus, in 3U84 chain A, which models 596-608 - and 8GPN's 587 is its own maximum, not the field's. NLS-1 at 479-497 does hold: it falls inside the 460-552 disordered region that crystallography constructs replace, and 0 of 68 menin chains model a single residue of it.
The trap in that measurement, which nearly inverted it. Eight deposits - the 9WKU / 9WKV / 9WKW / 9WKX / 9WN9 / 9WNA / 9WNI / 9WNJ series - carry 10 or 11 residues numbered inside 479-497, and none of them is menin NLS-1. In 9WKU the chain agrees with UniProt up to residue 457 and then diverges completely: deposited 479-481 is SAI where menin is RRR, the first three residues of the RRRGPRR monopartite signal. A survey keyed on auth_seq_id alone would have reported NLS-1 as present in eight structures. Every count above therefore required the deposited residue type to match O00255 at that position. All 35 NLS-2 observations pass that test; all 8 apparent NLS-1 observations fail it.
| partner a | gene-men1 |
|---|---|
| partner b | gene-men1 |
| partner b note | this is not a two-protein interface; the importin that reads these signals has never been identified, so the partner is unassigned and `partner_b` repeats menin rather than naming a protein this beat cannot evidence |
| motif a | NLS-1, menin 479-497 (RRRGPRRESKPEEPPPPKK); NLS-2, menin 588-608 (KKQKVSTPSDYTLSFLKRQRK) |
| motif b | unidentified - no importin alpha/beta has been assigned to menin in any source read here |
| method that mapped it | EGFP-tagged deletion constructs scored by fluorescence localisation in HEK-293T, plus immunofluorescence and subcellular fractionation of endogenous menin |
| uniprot annotation status | ABSENT. O00255 carries no NLS feature, no Motif and no Domain. This is the worked example in QUEEN_BRIEF rule 91 and it is still true on the record fetched 2026-09-22. |
| localization | Nucleus, primarily - established by immunofluorescence, subcellular fractionation and EGFP tagging in the same 1998 paper |
| localization shown for fusion | false |
| druggability | A basic-residue import signal is not a small-molecule target. It is, however, the handle for the alternative modality: an agent that mislocalises menin rather than blocking its pocket would remove every menin surface at once from chromatin, which is the degrader-like phenotype without degradation. |
| shown for fusion | NO for any fusion, and nobody has looked. Menin's import has never been examined in a KMT2A-rearranged cell, under a menin inhibitor, or in any leukaemia. |
| clinical consequence from source | 39 of 43 catalogued MEN1 frameshift/nonsense mutations truncate menin before both NLSs; the four most distal retain NLS-1 and would be predicted nuclear. The authors found no phenotypic difference between those four and the rest. |
| relationship to existing entries | New. `localization-menin-has-no-subnuclear-compartment-and-nobody-has-looked` records the compartment gap; this entry records the targeting SEQUENCES, which are a different fact, and it supplies the coordinates the graph needs if it ever models what a truncating MEN1 mutation does. `mechanism-pka-menin-s487-chromatin-release` concerns Ser487, eight residues from NLS-1, and the two have never been considered together. |
interface residues a
[ "menin 479-485 RRRGPRR (monopartite candidate)", "menin 484-497 (bipartite candidate)", "menin 604-608 KRQRK (monopartite candidate)" ]
interface residues b
[]
domains
[
{
"name": "NLS-1",
"start": 479,
"end": 497
},
{
"name": "NLS-2",
"start": 588,
"end": 608
}
]structures
[
{
"pdb": "3U84",
"chain": "A",
"models": "menin 596-608, 13 of NLS-2's 21 residues",
"sequence_observed": "SDYTLSFLKRQRK",
"includes": "the candidate monopartite KRQRK at 604-608 this entry nominates",
"numbering_check": "deposited auth_seq_id 596-608 matches UniProt O00255 596-608 character for character, so the numbering is not offset here",
"segments_in_chain": "1-386, 402-459, 520-528, 534-539, 548-581, 596-608",
"measured": "2026-09-22, king, from coordinates"
},
{
"note": "34 further deposits model 1-2 residues at the 588 edge only: 4GPQ, 4X5Y, 5DB0, 5DB2 and 5DD9 model 2; 4GQ3, 4GQ4, 4GQ6, 4OG4, 4OG5, 4OG6, 4OG7, 4OG8, 4X5Z, 5DB1, 5DB3, 5DDC, 5DDD, 5DDE, 6BXH, 6BXY, 6BY8, 6O5I, 6OPJ, 8VA5, 8VA6, 9C4S, 9C4T, 9C4U, 9C4V, 9C4W, 9C4X, 9C4Y and 9C4Z model 1. Only 3U84 resolves the signal in any useful sense."
}
]structures absent
[ "NLS-1 at 479-497: modelled in 0 of 68 menin chains across all 69 deposits (measured 2026-09-22). CORRECTED: this slot previously read 'no menin deposit models past residue 587 (8GPN is the maximum); NLS-2 at 588-608 is outside every coordinate set'. That was false for NLS-2 - see data.structures - and 587 is 8GPN's own maximum, not the maximum across deposits, which is 608.", "NLS-1 at 479-497 falls inside the UniProt-annotated disordered region 460-552 and is absent from every crystallisation construct measured" ]
ligands
[]
nls structural coverage measured 2026 09 22
{
"deposits_queried": 69,
"chains_identified_as_menin": 68,
"method": "RCSB search for experimental entries referencing UniProt O00255, then every mmCIF parsed for observed polymer residues; a position counts only when the deposited residue type also matches O00255 at that number",
"nls1_479_497": {
"deposits_modelling_it": 0,
"deposits_with_residues_merely_NUMBERED_in_the_window": 8,
"why_the_difference": "the 9W** series diverges from menin after residue 457 while continuing menin numbering"
},
"nls2_588_608": {
"deposits_modelling_at_least_one_residue": 35,
"best": "3U84 chain A, 596-608, 13 of 21",
"all_35_pass_the_residue_identity_check": true
},
"max_verified_menin_residue_in_any_deposit": 608
}nls2 lattice test cycle 65
{
"question": "Is NLS-2 ordered in 3U84 because it folds, or because the crystal holds it?",
"answered": "2026-09-22, cycle 65, king, from the deposited coordinates",
"method": "gemmi NeighborSearch over 3U84 (space group P 4-1, cell 139.9 x 139.9 x 54.1 A) with full crystal symmetry, 4.0 A cutoff, amino-acid residues only. Contacts normalised per atom because a Lys has more atoms than a Gly. Thresholds for the verdict were fixed in the script before the numbers were read.",
"result": {
"mean_B_segment_596_608": 81.13,
"mean_B_chain_A_body": 45.46,
"B_ratio": 1.78,
"symmetry_contacts_per_atom_segment": 0.485,
"symmetry_contacts_per_atom_body": 0.166,
"symmetry_contact_ratio": 2.92,
"every_intermolecular_contact_is_to_a_symmetry_mate": true,
"worst_residue": "Arg607 makes 23 symmetry contacts on its own",
"internal_control": "chain B of the same crystal stops at residue 581 and models none of NLS-2, so the segment is ordered in one copy of two"
},
"VERDICT": "SUPERSEDED IN CYCLE 66 - DO NOT QUOTE THIS FIELD ALONE. It read: \"LATTICE-STABILISED. The ordered conformation of menin 596-608 in 3U84 is a crystal-packing feature, not evidence of a folded nuclear localisation signal in solution.\" That verdict was reached by comparing a chain TERMINUS against the chain's INTERIOR, which is the wrong baseline. Measured against other menin C-terminal segments, 3U84 is at the 37.5th percentile for symmetry contact (median 5.37) and the 94.4th for B-factor ratio (median 1.40). REVISED: weakly ordered, not lattice-stabilised. See SUPERSEDED_BY_ITS_OWN_NULL_CYCLE_66 in this same block.",
"what_this_kills": "the inference the king drew last cycle from his own correction - that because NLS-2 is modelled somewhere, it is in principle a structure-guided target. It is not, on this evidence.",
"what_this_does_NOT_kill": "the correction itself. NLS-2 IS modelled in 3U84, and this entry's original claim that both signals lie outside every structure ever solved remains false. A lattice-stabilised conformation is still a conformation that was observed.",
"reproducible": "analyses/menin-nls2-lattice-test/run.py",
"SUPERSEDED_BY_ITS_OWN_NULL_CYCLE_66": {
"what_was_wrong": "The cycle-65 test compared menin 596-608 against the INTERIOR of the same chain and read 2.92x the symmetry-contact rate as evidence of lattice stabilisation. Chain termini are more solvent-exposed and more inter-molecularly contacted than interiors everywhere, so that is the wrong baseline: it measures being a terminus, not being held by this particular crystal.",
"the_right_baseline": "every other menin chain's own C-terminal segment, length-matched to 13 residues, across all 69 deposits referencing O00255 - 74 chains measured, 73 with a non-zero body rate so that a ratio exists",
"result": {
"symmetry_contact_ratio_3U84": 2.92,
"symmetry_contact_ratio_median_of_72_others": 5.37,
"symmetry_contact_ratio_range": "0.00 to 9.44",
"symmetry_contact_percentile_of_3U84": 37.5,
"b_factor_ratio_3U84": 1.79,
"b_factor_ratio_median_of_72_others": 1.4,
"b_factor_ratio_range": "0.87 to 2.14",
"b_factor_percentile_of_3U84": 94.4
},
"REVISED_VERDICT": "NOT lattice-stabilised, on this evidence. 3U84's C-terminal segment makes FEWER symmetry contacts relative to its own chain body than the median menin C-terminus (37.5th percentile, below a median of 5.37). What IS exceptional is its mobility: the B-factor ratio is at the 94.4th percentile. The segment is weakly ordered rather than held by the crystal.",
"what_this_does_and_does_not_change": "It does NOT revive the therapeutic inference withdrawn in cycle 65 - a 94th-percentile-B terminus is not a design-ready conformation either way - so the practical guidance ('do not design on this conformation') stands. What changes is the stated REASON, from crystal packing to intrinsic disorder, and a mechanism asserted without its null is exactly what this project keeps telling agents not to write.",
"caveat_on_the_comparison": "The comparators are not the same window. Most menin deposits stop near residue 587, so a typical chain's last 13 residues are roughly 576-588 while 3U84's are 596-608. The baseline is therefore position-analogous and length-matched but not residue-matched, and a C-terminus that reaches 20 residues further into a disordered tail may not be comparable to one that stops at the ordered boundary. This is a real limit on the comparison and it cuts both ways.",
"reproducible": "analyses/menin-c-terminus-lattice-null/run.py",
"measured": "2026-09-22, king, cycle 66"
}
}- Any menin-directed degrader or mislocaliser acts upstream of every surface in this beat at once. That is the point of such a modality and also its whole risk: it removes the MBM grip, the SvID, the gamma-H2A.X motif, the nucleosome grip and the LEDGF patch in one step, in every tissue.
- Truncating MEN1 mutations are not equivalent to pocket blockade. A cytoplasmic truncated menin has lost all nuclear functions, so germline MEN1 phenotypes are a poor guide to what a pocket inhibitor does and this graph should not read the endocrine tumour spectrum as a direct prediction of drug toxicity.
- Delivery: NLS-1 is disordered and absent from all 68 menin chains, so no structure-guided approach to it is possible. NLS-2 has been observed once, in 3U84 at 596-608, and is still not a design-ready conformation - but the reason has been revised twice. Cycle 64 called it 'in principle a structure-guided target' (withdrawn). Cycle 65 called the conformation lattice-stabilised. Cycle 66 tested that against other menin C-termini and found 3U84 at the 37.5th percentile for symmetry contact - BELOW median - and the 94.4th for B-factor ratio. So the segment is WEAKLY ORDERED rather than crystal-held. Neither signal is a structure-guided target today; the honest difference between them is that one has been seen once, poorly, and the other has never been seen at all.
- Which importin reads these signals? No importin has been assigned to menin in twenty-eight years, which is unusual for a protein with 500+ papers, and it means the import step cannot be perturbed specifically.
- Does Ser487 phosphorylation, eight residues from NLS-1 and inside a basic-and-acidic compositional bias, modulate import as well as chromatin release? The two literatures do not meet.
- Do these signals still function in the presence of a pocket inhibitor? A drug-bound menin that cannot enter the nucleus would be a different pharmacology from a drug-bound menin that enters and cannot grip, and nobody has fractionated a treated cell.
- Should UniProt carry these features? The mapping is a clean 1998 deletion series in a canonical journal and the sequences verify exactly; a curation request is a five-minute action with a durable benefit to every downstream pipeline.
- ANSWERED IN CYCLE 65 AND THE ANSWER WAS REVISED IN CYCLE 66 BY ITS OWN NULL. The question was whether NLS-2's ordering in 3U84 is a lattice artefact. Cycle 65 said yes, on 2.92x the symmetry-contact rate of the rest of chain A and B-factors 1.78x the body mean. Cycle 66 compared those ratios against every other menin chain's own C-terminal segment, length-matched, across all 69 deposits: 3U84 sits at the **37.5th percentile** for symmetry contact against a median of 5.37, i.e. BELOW median, and at the **94.4th percentile** for B-factor ratio against a median of 1.40. REVISED VERDICT: not lattice-stabilised - weakly ordered. The practical conclusion is unchanged and the mechanism is different. What remains genuinely open is unchanged too: whether the KRQRK at 604-608 adopts any defined conformation in solution, which needs NMR or a second crystal form.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Menin is a 610-amino-acid nuclear protein with at least two independent nuclear localisation signals, both in its C-terminal quarter: NLS-1 at residues 479-497 (RRRGPRRESKPEEPPPPKK) and NLS-2 at 588-608 (KKQKVSTPSDYTLSFLKRQRK), mapped by immunofluorescence, subcellular fractionation and EGFP-tagged deletion analysis in HEK-293T; constructs ending at residue 476 are cytoplasmic while a construct ending at 571 is nuclear, and constructs lacking NLS-2 remain nuclear because NLS-1 alone suffices. Of 43 known frameshift/nonsense MEN1 mutations, 39 would produce a protein lacking both signals.
Menin, the product of the MEN1 gene, is a nuclear protein. · Guru SC, Goldsmith PK, Burns AL, Marx SJ, Spiegel AM, Collins FS, Chandrasekharappa SC · Proc Natl Acad Sci U S A 1998What this evidence cannot show
Overexpressed EGFP fusions in a single cell line, scored by fluorescence pattern - a deletion-mapping assay that identifies sufficiency for import, not a measurement of importin binding, and no importin has ever been identified for menin. The monopartite-versus-bipartite assignments inside each NLS are the authors' sequence inspection, explicitly hedged in the paper ('could function', 'it is possible that'), and are recorded here as candidates rather than findings. The paper's own negative - that none of the then-known 22 missense and 3 in-frame MEN1 mutations falls in either NLS - is a statement about the 1998 mutation catalogue and is not re-checked here. ROUTE NOTE: Europe PMC fullTextXML returned HTTP 500 for PMC19125 and the pmc.ncbi.nlm.nih.gov page returned 167 characters on a first attempt; the text (28,014 characters) was recovered from pmc.ncbi.nlm.nih.gov on retry with a full browser User-Agent.
Evidence that argues against this (1)
- Menin has documented conditional cytoplasmic behaviour that a constitutive two-NLS model does not explain: PKA phosphorylation at Ser487 - which sits eight residues C-terminal to NLS-1 and inside the 460-552 disordered region - increases menin's binding to nuclear actin/myosin IIa and sequesters it from a target promoter. Import signals and chromatin residence are different problems, and this graph's `mechanism-pka-menin-s487-chromatin-release` holds the second one. UniProt Consortium 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
The live UniProt O00255 record is 610 aa, its residues 479-497 read RRRGPRRESKPEEPPPPKK and its residues 588-608 read KKQKVSTPSDYTLSFLKRQRK - matching the 1998 mapping exactly - and it annotates no nuclear localisation signal, no Motif and no Domain feature, carrying only two Regions (214-390 'Interaction with FANCD2' and 460-552 'Disordered').
UniProtKB O00255 (MEN1_HUMAN), full feature table · UniProt Consortium · UniProt Knowledgebase 2026What this evidence cannot show
Fetched 2026-09-22. An absent annotation is a statement about UniProt's curation, not about the protein; UniProt does not systematically annotate NLSs that rest on deletion mapping rather than on a matched consensus or a structure. The 610-residue form is the DISPLAYED isoform (O00255-2) and a longer isoform is described, so 'menin is 610 aa' is a statement about which isoform a source means.
Evidence that argues against this (1)
- A second study frames menin's NLSs as coordinating gene expression rather than merely import - 'Tumor suppressor menin: the essential role of nuclear localization signal domains in coordinating gene expression' (PMID 16449969, Oncogene 2006) - which if correct means these windows are not only address labels and that deleting them is not a clean localisation-only perturbation. That paper was identified but not read in this beat. [SECOND CAUGHT ERROR IN THIS BEAT: the first draft of this citation listed the authors as 'La P, Yang Y, Karnik SK, et al.' - the first surname was right and the next two were invented. Corrected against efetch on 2026-09-22. Both of this beat's citation errors were in fields typed while writing rather than copied from an open record, which is the exact motion QUEEN_BRIEF warns about.] La P et al. 2006
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.computation_on_public_coordinates
Across all 69 experimental PDB deposits that reference UniProt O00255, 68 carry an identifiable menin chain. NLS-1 (479-497) is modelled in NONE of them. NLS-2 (588-608) is modelled in 35, and 3U84 chain A resolves residues 596-608 - 13 of 21, with the observed sequence SDYTLSFLKRQRK matching UniProt character for character, including the candidate monopartite KRQRK at 604-608. The maximum verified menin residue in any deposit is 608, the protein's own C-terminus, not 587. Eight deposits (the 9WKU / 9WKV / 9WKW / 9WKX / 9WN9 / 9WNA / 9WNI / 9WNJ series) carry 10-11 residues NUMBERED inside 479-497 whose sequence is not menin's - 9WKU has SAI where menin has RRR at 479-481 - so every count here required the deposited residue type to match O00255 at that position.
Observed-residue survey of all 69 menin (UniProt O00255) PDB deposits, with a residue-identity guard against non-corresponding deposited numbering · king, Oncofusion knowledge graph, cycle 64 · analyses/menin-nls-structural-coverage/run.py 2026What this evidence cannot show
Observed-residue counts from deposited coordinates only; a residue present in the construct but disordered is indistinguishable here from one absent from the construct, so 'not modelled' means 'not resolved', not 'not present'. The identity guard rejects a position when the deposited residue type differs from UniProt, which would also reject a genuine engineered point mutant at that position - conservative in the direction that matters, since it can only undercount NLS coverage. 3U84's 596-608 is a single chain in a single crystal form and its 34 companions clip only one or two residues past 588.
Evidence that argues against this (1)
- The 3U84 observation does not make NLS-2 a solved structure in any useful sense. One deposit of 69 resolves it, and the same crystal form's chain B stops at 581, so even within 3U84 the signal is ordered in one copy and not the other - which is what a lattice-stabilised conformation looks like. The entry's practical claim, that menin's targeting signals are structurally uncharacterised, survives; only the absolute form of it does not.
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.computation_on_public_coordinates
Menin residues 596-608, the only part of either NLS ever resolved, are held by crystal packing in 3U84 rather than folded. With full symmetry at a 4.0 A cutoff the segment makes 0.485 intermolecular contacts per atom against 0.166 for the rest of chain A (2.92x), and every one of those contacts is to a symmetry mate rather than to chain B in the asymmetric unit. Mean B is 81.13 against a chain body of 45.46 (1.78x), and Arg607 alone accounts for 23 symmetry contacts. Chain B of the same crystal stops at residue 581 and models none of NLS-2, so the signal is ordered in one of the two copies present.
Crystal-contact and B-factor analysis of the modelled menin C-terminus in PDB 3U84, with full space-group symmetry · king, Oncofusion knowledge graph, cycle 65 · analyses/menin-nls2-lattice-test/run.py 2026What this evidence cannot show
A contact census is not a free-energy calculation: a segment can make many lattice contacts and still adopt a native-like conformation, and the 2.92x ratio bounds the opportunity for packing to dictate the fold rather than proving that it did. B-factors conflate genuine mobility with refinement decisions and differ between structures, so only the within-chain comparison is used here. The 4.0 A cutoff is conventional and unoptimised. An absent NMR or second-crystal-form comparison is what would actually settle the solution conformation.
Evidence that argues against this (1)
- The honest counter-reading is that this test cannot distinguish 'ordered because of packing' from 'ordered enough to be caught by packing'. A fully disordered tail cannot be modelled at all, even against a lattice - 34 other menin deposits clip at 588 and none of them resolves it, despite also being crystals with contacts. So the fact that 3U84's packing COULD capture this segment is weak evidence that the segment has some conformational preference. The verdict is that the observed conformation should not be used for design, not that the region is structureless.
- E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.computation_on_public_coordinates
Menin 596-608 in 3U84 is NOT unusually lattice-contacted for a menin C-terminus. Across all 69 deposits referencing UniProt O00255, 74 menin chains were measured and each chain's own last contiguous modelled segment was taken, capped at 13 residues to length-match 3U84's 596-608. 3U84 chain A's per-atom symmetry-contact ratio against its own chain body is 2.92, against a median of 5.37 for the other 72 chains (range 0.00 to 9.44) - the 37.5th percentile. Its B-factor ratio of 1.79 against a median of 1.40 (range 0.87 to 2.14) is the 94.4th percentile. The six highest symmetry-contact ratios are 8.30 to 9.44 and all belong to segments around 576-589.
C-terminal-segment lattice-contact and B-factor null across all menin PDB deposits, with full space-group symmetry · king, Oncofusion knowledge graph, cycle 66 · analyses/menin-c-terminus-lattice-null/run.py 2026What this evidence cannot show
The comparators are position-analogous and length-matched but NOT residue-matched: most menin deposits stop near 587, so a typical chain's last 13 residues are about 576-588 while 3U84's are 596-608. A terminus reaching 20 residues further into a disordered tail may not be comparable to one stopping at the ordered boundary. B-factors also conflate mobility with refinement decisions and are only compared within a chain. Contact counts at a 4.0 A cutoff bound opportunity, not energy.
Evidence that argues against this (1)
- The percentile cuts against the cycle-65 verdict but does not vindicate the cycle-64 inference either, and it is worth stating why the practical answer did not move. A 94.4th-percentile B-factor ratio means this segment is among the most mobile C-termini in the menin corpus, and a mobile terminus modelled in one chain of two - chain B stops at 581 - is not a conformation to design against whatever holds it there. The correction is to the mechanism and to my own reasoning, not to the guidance.
the resource-citation case: three pairs at 0.019-0.048 because a PDB deposit has no abstract to match
menin NLS-1 479-497 and NLS-2 588-608
the targeting sequences exist and are verified; the compartment does not
- related to →PKA phosphorylates menin at Ser487 and pulls it off chromatin - a kinase input on the drug target itself
Ser487 lies eight residues C-terminal to NLS-1, inside the same disordered region
a degrader also removes both nuclear localisation signals, which no pocket ligand touches
3U84 models menin 596-608 as a helix, so NLS-2 is NOT outside every coordinate set; the NLS-1 claim stands