AnalysisstrongQ31-interfaces-structural

What was searched for a structure and what came back empty: the KMT2A interface layer enumerated by accession pair, with the graph's own unmapped interfaces listed

also called structural interface enumeration · PDB absence ledger · Q31 enumeration

This is the bookkeeping behind the structural interface entries: every protein pair that was searched in the world's structure database, what came back, and what came back with nothing. The short version is that the parts of KMT2A the leukaemia fusion throws away are extremely well solved, and the parts it keeps are not.

The same thing, for a scientist

Twenty-four accession-pair and single-accession queries were run against the RCSB search API on 2026-09-20 using exact_match on rcsb_polymer_entity_container_identifiers.reference_sequence_identifiers.database_accession, and every returned entry used in a graph entry was then fetched from data.rcsb.org and read back title-by-title and entity-by-entity. Peptide and domain spans were assigned by aligning the fetched entity sequences to UniProt sequences fetched the same day, never from recall. The enumeration records three things the beat could not put in an interface entry: the queries and their counts, the pairs that return nothing, and the interfaces this graph asserts with no residue-level mapping anywhere.

Go deeper

Queries run and counts returned

All on 2026-09-20, RCSB search API, exact_match on reference_sequence_identifiers.database_accession. An empty result set is returned as HTTP 204 with no body 1.

query	entries
Q03164 (KMT2A)	60
O00255 (MEN1)	69
O00255 + Q03164	7
P61964 (WDR5) + Q03164	15
Q15291 (RBBP5) + P61964	20
Q9UBL3 (ASH2L) + Q15291	27
Q9C005 (DPY30) + Q9UBL3	9
Q9UHB7 (AFF4)	9
Q9UHB7 + P50750 (CDK9)	5
Q9UHB7 + O00472 (ELL2)	1
Q9UHB7 + P55199 (ELL)	0
P42568 (AF9) + P51825 (AFF1)	1
P42568 + Q9UHB7	0
P51825 (AFF1), alone	1
Q03111 (ENL)	26
Q03111 + P51825	0
Q03111 + Q9UHB7	0
Q9H6P5 (TASP1)	10
Q9H6P5 + Q03164	0
P55199 (ELL), alone	1
P55199 + O00472	0
O60563 (CCNT1) + P50750	28
Q9UNP9 (PPIE/CyP33)	40
O00472 (ELL2)	4

What has no structure at all

The ENL AHD, with anything. All 26 ENL entries are YEATS-domain work; none contains an AF4-family partner. This graph's account of how KMT2A::MLLT1 reaches P-TEFb has no structural basis in ENL 1.
ELL with any AFF scaffold. P55199 appears in exactly one PDB entry in the world. The t(11;19) partner in KMT2A::ELL is structurally almost untouched, and everything asserted about it is modelled from ELL2 1.
AFF1, except once. P51825 appears in one entry, the single-chain AFF1/AF9 chimera 2LM0. There is no AFF1 structure with AFF4, P-TEFb, ELL or ELL2 5.
Taspase1 with its substrate. Ten taspase1 entries including small-molecule co-crystals, none with KMT2A 1.
Any KMT2A fusion protein. All 60 Q03164 entries are isolated domains, short peptides, or C-terminal fragments in complex; none spans a breakpoint 1.
The retained KMT2A fragment together with WRAD. Every cryo-EM reconstruction of the MLL1 core complex contains only a C-terminal MLL1 fragment 1.

The pattern: the structural literature covers the half of KMT2A the fusion discards, in great depth, and the half it keeps thinly and mostly as short peptides.

Interfaces this graph asserts with no residue-level mapping anywhere

Audited across all 17 partner_interface entries existing before this beat, by scanning every field of each entry for a residue token (single-letter or three-letter code followed by a position). Entries with no residue identified on either side, anywhere in the entry:

interface-aff1-af4int-af9-enl-ahd - the index fusion's only documented elongation interface, and no residue is named on either side
interface-menin-jund-pocket - a motif range for JunD, nothing on menin
interface-lsd1-gfi1-snag
interface-cks1-cks2-kmt2a-cell-cycle-adaptor
interface-jak1-mllt10-fusion-kinase-contact-no-site - the entry title states the gap honestly
interface-aff1-npm1-proximity and interface-aff1-rbm15-proximity - proximity labelling, so no contact is claimed
interface-fyrn-fyrc-heterodimer-of-the-reciprocal-product - descriptive residue fields, no positions

And interfaces with residues but no affinity anywhere in the graph before this beat: every one except interface-menin-kmt2a-mbm1 and interface-af9-ahd-dot1l-three-sites-2mv7. Of 17 entries, 2 carried a Kd.

This beat added a measured or bounded affinity to nine further interfaces.

Bounded rejections - what was read and not used

PMID 26886794 (Li et al., Nature 2016), PDB 5F6L, 7W6A and the MLL1 SET-domain series (2W5Y, 2W5Z, 5F5E, 6U9K/M/N/R, 7U5V). On-topic and high quality: a minimised RBBP5-ASH2L heterodimer activates all MLL-family methyltransferases. Cited in interface-wdr5-rbbp5-vdv but not given its own entry, because the crystallised MLL1 constructs carry activating N3861I/Q3867L substitutions and the whole module is C-terminal to every breakpoint. It would be a wild-type-allele entry with an engineered protein.
The 28 CDK9-cyclin T1 entries. Kinase-cyclin structures, mostly inhibitor co-crystals. Dropped: they describe the kinase, not its recruitment by a fusion.
PDB 2W5Y/2W5Z (MLL1 SET with SAH). Dropped for the same reason as above.
PDB 5SVH - a designed MLL/c-Myb chimeric KIX ligand whose deposition title claims nanomolar affinity. Recorded in interface-kmt2a-tad-crebbp-kix but with confidence: weak, because RCSB lists the primary citation as To Be Published with no PMID and no DOI, so the affinity claim is unverifiable.
PDB 3HPG - visna virus integrase with the LEDGF IBD. Noted as counter-evidence only; a non-human lentivirus is too indirect for its own entry.
The four AFF4-P-TEFb-Tat co-crystals (4OGR, 4OR5, 5L1Z, 6CYT). All add HIV-1 Tat to the same core; recorded in the data payload, not given separate entries.

Papers found, on topic, and unreadable after the full route test (Europe PMC core returns no pmcid, which terminates routes 2, 3 and 4 because all three require one): PMID 21220120 (Avdic, MLL1 core complex assembly - the source of the 3P4F structure and the only likely home of a WDR5-RBBP5 Kd), PMID 16216576 (Khan, taspase1 crystal structure), PMID 33784495 (Nagaratnam, catalytically active taspase1), PMID 29276034 (Xu, CXXC DNA selectivity), PMID 16253272 (De Guzman, KIX cooperativity), PMID 18829459 (Patel, WDR5-MLL1 peptide). Each is cited for what its title, abstract and deposited structures establish, and for nothing more.

Method notes that cost time and should not cost it again

RCSB returns HTTP 204 with an empty body for a zero-result search. A naive json.loads raises a JSONDecodeError that looks like a network failure rather than a real negative. Four of this beat's most important absences arrived that way.
Fetch the polymer entities, not just the entry. Three of the structures central to this disease (2MTN, 6EMQ, 2LM0) and one more (2KU7) are single-chain chimeras annotated to two UniProt accessions. Nothing in the entry title says so; only the entity list does.
Align the deposited sequence, do not trust the deposit's residue numbering. 7UD5's MLL1 chain is annotated to a cDNA clone (B4DIJ7) rather than to Q03164. The 4RIQ ASH2L peptide numbers from a 534-residue isoform, 94 lower than UniProt Q9UBL3. 3P4F's MLL1 WIN peptide carries an S3763A substitution. 4GQ6's MLL peptide carries C5A.
The RCSB primary-citation title is not the PubMed title. For PMID 16260736, RCSB stores '...HIV-1 integrase and LEDGF/p75' and PubMed gives '...HIV-1 integrase and transcriptional coactivator p75'. Storing the RCSB version would have produced a title mismatch of exactly the shape this project's checker exists to find.
Route 4 was needed repeatedly. fullTextXML returned HTTP 500 for 22665483, 20961854, 25305204, 25456412, 31147444, 23260655, 20541251, 20010842 and 16990798; efetch db=pmc recovered five of those and the PMC HTML page recovered the rest.

Facts and figures

script	none - queries issued directly against search.rcsb.org and data.rcsb.org
status	run
date run	2026-09-20
result summary	Six interface classes central to KMT2A-rearranged leukaemia have no experimental structure; eight of the graph's seventeen pre-existing partner_interface entries name no residue anywhere; and two of seventeen carried an affinity.

queries

unreadable after all four routes

inputs

[
  "RCSB search API v2",
  "RCSB data API v1",
  "UniProtKB REST",
  "Europe PMC REST",
  "NCBI efetch/esummary"
]

queries — the numbers behind the chart

[
  {
    "q": "Q03164",
    "n": 60
  },
  {
    "q": "O00255",
    "n": 69
  },
  {
    "q": "O00255+Q03164",
    "n": 7
  },
  {
    "q": "P61964+Q03164",
    "n": 15
  },
  {
    "q": "Q15291+P61964",
    "n": 20
  },
  {
    "q": "Q9UBL3+Q15291",
    "n": 27
  },
  {
    "q": "Q9C005+Q9UBL3",
    "n": 9
  },
  {
    "q": "Q9UHB7",
    "n": 9
  },
  {
    "q": "Q9UHB7+P50750",
    "n": 5
  },
  {
    "q": "Q9UHB7+O00472",
    "n": 1
  },
  {
    "q": "Q9UHB7+P55199",
    "n": 0
  },
  {
    "q": "P42568+P51825",
    "n": 1
  },
  {
    "q": "P42568+Q9UHB7",
    "n": 0
  },
  {
    "q": "P51825",
    "n": 1
  },
  {
    "q": "Q03111",
    "n": 26
  },
  {
    "q": "Q03111+P51825",
    "n": 0
  },
  {
    "q": "Q03111+Q9UHB7",
    "n": 0
  },
  {
    "q": "Q9H6P5",
    "n": 10
  },
  {
    "q": "Q9H6P5+Q03164",
    "n": 0
  },
  {
    "q": "P55199",
    "n": 1
  },
  {
    "q": "P55199+O00472",
    "n": 0
  },
  {
    "q": "O60563+P50750",
    "n": 28
  },
  {
    "q": "Q9UNP9",
    "n": 40
  },
  {
    "q": "O00472",
    "n": 4
  }
]

structural absences

[
  "ENL (Q03111) AHD with AFF1 or AFF4 - zero entries",
  "ELL (P55199) with AFF4 or ELL2 - zero entries; P55199 appears in one entry in the whole PDB",
  "AFF1 (P51825) with any SEC partner - one entry total, and it is a single-chain AFF1/AF9 chimera",
  "taspase1 (Q9H6P5) with KMT2A - zero entries",
  "any KMT2A fusion protein or breakpoint-spanning construct - zero entries",
  "the retained KMT2A N-terminal fragment together with WRAD - zero entries"
]

graph interfaces with no residue mapping

[
  "interface-aff1-af4int-af9-enl-ahd",
  "interface-menin-jund-pocket",
  "interface-lsd1-gfi1-snag",
  "interface-cks1-cks2-kmt2a-cell-cycle-adaptor",
  "interface-jak1-mllt10-fusion-kinase-contact-no-site",
  "interface-aff1-npm1-proximity",
  "interface-aff1-rbm15-proximity",
  "interface-fyrn-fyrc-heterodimer-of-the-reciprocal-product"
]

graph interfaces carrying a kd before this beat

[
  "interface-menin-kmt2a-mbm1",
  "interface-af9-ahd-dot1l-three-sites-2mv7"
]

chimeric deposits flagged

[
  "2MTN (MLL1 + LEDGF IBD)",
  "6EMQ (MLL1 111-160 + LEDGF IBD)",
  "2LM0 (AFF1 + AF9 AHD)",
  "2KU7 (MLL1 PHD3 + CyP33 RRM)",
  "5SVH (MLL TAD + c-Myb)",
  "3U88 (spliced MLL1 construct)"
]

numbering traps found

[
  "ASH2L: papers number from a 534-residue isoform; UniProt Q9UBL3 is 628; offset exactly 94 (paper L513 = Q9UBL3 L607)",
  "7UD5 MLL1 chain annotated to cDNA clone B4DIJ7, not Q03164",
  "3P4F MLL1 WIN peptide carries S3763A",
  "4GQ6 MLL peptide carries C5A",
  "RCSB primary-citation title for PMID 16260736 differs from the PubMed title"
]

unreadable after all four routes — the numbers behind the chart

[
  {
    "pmid": "21220120",
    "title_asserts": "Structural and biochemical insights into MLL1 core complex assembly",
    "why": "Europe PMC core returns no pmcid, which terminates routes 2-4",
    "title": "Structural and biochemical insights into MLL1 core complex assembly.",
    "authors": "Avdic V, Zhang P, Lanouette S, et al.",
    "venue": "Structure",
    "year": 2011,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "16216576",
    "title_asserts": "Crystal structure of human Taspase1, a crucial protease regulating the function of MLL",
    "why": "no pmcid",
    "title": "Crystal structure of human Taspase1, a crucial protease regulating the function of MLL.",
    "authors": "Khan JA, Dunn BM, Tong L",
    "venue": "Structure",
    "year": 2005,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "33784495",
    "title_asserts": "Structural insights into the function of the catalytically active human Taspase1",
    "why": "no pmcid",
    "title": "Structural insights into the function of the catalytically active human Taspase1.",
    "authors": "Nagaratnam N, Delker SL, Jernigan R, et al.",
    "venue": "Structure",
    "year": 2021,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "29276034",
    "title_asserts": "DNA Sequence Recognition of Human CXXC Domains and Their Structural Determinants",
    "why": "no pmcid",
    "title": "DNA Sequence Recognition of Human CXXC Domains and Their Structural Determinants.",
    "authors": "Xu C, Liu K, Lei M, et al.",
    "venue": "Structure",
    "year": 2018,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "16253272",
    "title_asserts": "Structural basis for cooperative transcription factor binding to the CBP coactivator",
    "why": "no pmcid",
    "title": "Structural basis for cooperative transcription factor binding to the CBP coactivator.",
    "authors": "De Guzman RN, Goto NK, Dyson HJ, et al.",
    "venue": "J Mol Biol",
    "year": 2006,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  },
  {
    "pmid": "18829459",
    "title_asserts": "Structure of WDR5 bound to mixed lineage leukemia protein-1 peptide",
    "why": "no pmcid",
    "title": "Structure of WDR5 bound to mixed lineage leukemia protein-1 peptide.",
    "authors": "Patel A, Dharmarajan V, Cosgrove MS",
    "venue": "J Biol Chem",
    "year": 2008,
    "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
  }
]

How this could bite us back

A counted structural absence is only as good as its query: all counts here are bounded by UniProt cross-annotation and exclude computed models, so a design decision taken on 'no structure exists' should re-run the search first.
Six of the structures underpinning this field are engineered single chains, so any affinity or geometry inferred from them overstates how stable these complexes are in trans.
The best-solved interfaces in this disease sit in the half of KMT2A the fusion discards, which makes structure-guided design here systematically biased toward wild-type-allele targets.

Still open

Is there a WDR5-RBBP5 dissociation constant in PMID 21220120? The paper is closed and the number is a one-experiment gap in the WRAD assembly chain.
Would a homology model of the ENL AHD on the AF9 template, validated by a single fluorescence-anisotropy measurement of ENL AHD to an AF4 peptide, be enough to close the largest structural absence in this beat?
Does any structure genomics or AlphaFold-derived resource contain a model of a KMT2A breakpoint-spanning construct, and would such a model be worth recording given that the retained fragment is largely intrinsically disordered?

Every source behind this entry (5)

E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Twenty-four accession-pair and single-accession queries were run against the RCSB search API on 2026-09-20; the pairs ENL+AFF1, ENL+AFF4, AF9+AFF4, AFF4+ELL, ELL+ELL2 and TASP1+KMT2A each returned an empty result set, and AFF1 and ELL each appear in exactly one PDB entry.
RCSB Protein Data Bank search and entry APIs (data.rcsb.org, search.rcsb.org) · RCSB PDB · RCSB PDB 2026
What this evidence cannot show
Every count is bounded by UniProt cross-annotation: a deposit whose entity lacks a reference_sequence_identifier for the accession is invisible to this query, and predicted or computed structure models are excluded by construction. Counts are as of 2026-09-20 and will drift. The statement that all 26 ENL entries are YEATS-domain structures is an inference from their deposition pattern, not an entity-by-entity check.
Evidence that argues against this (1)
- Sequence homology between ELL and ELL2, and between the AF9 and ENL AHDs, means a homology model is available for several of these absences - so 'no structure' is a statement about experimental evidence, not about whether a usable structural hypothesis exists. Leach BI et al. 2013
E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
All peptide and domain spans recorded across this beat were assigned by aligning fetched RCSB polymer-entity sequences to UniProt sequences fetched on the same day.
UniProtKB sequence retrieval (rest.uniprot.org) · UniProt Consortium · UniProt 2026
What this evidence cannot show
Alignment was exact substring matching with a 15-mer fragment fallback for tagged or substituted constructs; a construct differing from the canonical sequence at several positions yields an approximate span, which is flagged as approximate wherever it occurs (2AGH is the one such case).
Evidence that argues against this (1)
- Exact substring matching cannot detect an isoform difference that happens to preserve the matched fragment; the ASH2L 94-residue offset was caught only because the paper's numbering failed to match at all. RCSB PDB 2026
E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The largest quantitative gap recorded in this graph's partner dossier - no verifiable Kd for the LEDGF IBD-KMT2A interaction - is closed by a published measurement of 14.7 uM binary and 1.4 uM in the menin complex.
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 2014
PMID 25305204 DOI
What this evidence cannot show
NMR titration and ITC on isolated fragments; not measured on a fusion protein. The point recorded here is that the number exists, not that it describes the fusion.
Evidence that argues against this (1)
- The graph's existing entry reached the opposite conclusion after its own search, so the gap was a retrieval failure rather than a literature absence - which is itself a caution about how counted absences are generated in this project. Murai MJ et al. 2014
E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A 2023 solution-structure study determined the trimolecular Cyp33 RRM / MLL1 PHD3 / H3K4me3 complex, settling a 2010 disagreement this graph had recorded one-sidedly.
RNA binding induces an allosteric switch in Cyp33 to repress MLL1-mediated transcription. · Blatter M, Meylan C, Clery A, et al. · Sci Adv 2023
PMID 37075125 DOI
What this evidence cannot show
The ternary deposit 7ZEZ uses a helix-deleted isoform-3 Cyp33 RRM construct; no wild-type ternary structure exists.
Evidence that argues against this (1)
- Because the ternary structure uses a truncated RRM, a defender of the mutual-exclusivity model could argue the wild-type helix prevents the arrangement - which is precisely the allosteric mechanism the paper proposes, so the objection does not rescue the 2010 reading. Blatter M et al. 2023
E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
MLL1 is the weakest SET1-family binder of the WDR5 WIN site, with paralogue peptides binding 5- to 51-fold more tightly and MLL3 at 54 nM against MLL1's 1.7 uM.
Structural basis for WDR5 interaction (Win) motif recognition in human SET1 family histone methyltransferases. · Dharmarajan V, Lee JH, Patel A, et al. · J Biol Chem 2012
PMID 22665483 DOI
What this evidence cannot show
Peptide ITC in 300 mM NaCl; the full-length MLL1 affinity quoted in the same paper is 120 nM, so peptide rankings may not transfer to proteins.
Evidence that argues against this (1)
- No published WIN-site inhibitor has been profiled against all six SET1-family WIN peptides, so the selectivity consequence of this ranking is predicted rather than demonstrated. Avdic V et al. 2011

Connected to

supports →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
the search behind the LEDGF affinity correction
supports →The ENL AHD to AFF1 contact: asserted throughout this graph, structurally unsolved, and represented only by an AF9 chimera measured at KD 0.17 nM
the queries establishing that no ENL AHD structure exists
supports →The ELLbow: AFF4 301-351 wrapped over the ELL2 occludin domain at 2.0 A, Kd 86 nM - and no structure of ELL itself, the partner in t(11;19)
the queries establishing that ELL appears in one PDB entry
supports →Taspase1 and its two KMT2A cleavage sites: an enzyme-substrate interface with four apo structures, one co-crystal with a small molecule, and no structure with MLL at all
zero taspase1-KMT2A entries
supports →PHD3 and the CyP33 RRM: Kd 7-15 uM, and a 2023 ternary structure that settles the 2010 disagreement - H3K4me3 and the RRM CAN be held at once, and the real switch is RNA
the 2023 arbitration of the 2010 disagreement
supports →WDR5 WIN site and KMT2A Arg3765: measured at 1.7 uM, three co-crystals, and absent from every leukaemic fusion - plus the MLL3 peptide that binds it 51-fold tighter
the SET1-family WIN affinity ranking
related to →AFF1 AF4_int motif to AF9/ENL AHD: the index fusion's only documented elongation interface, and it competes with DOT1L
named in the audit as asserting an interface with no residue on either side
related to →Enumerating the WRAD-versus-KMT2A-fusion literature: what was searched, found, rejected and could not be read
the structural counterpart to that enumeration
supports →Papers we know exist, know matter, and cannot read
six on-topic papers unreadable after the full four-route test
supports →KMT2A 747-755 on HLA-B*07:02: a phosphoneoantigen from the half of KMT2A every fusion keeps, with seven crystal structures and a cognate TCR
found by reading back all 60 KMT2A-containing PDB entries rather than by topic search