First systematic enumeration of the KMT2A fusion interactome, and the diff against this graph

Full text read via PMC. The fusions tested are KMT2A::EP300 in transfected 293 cells and endogenous fusion KMT2A in the HB line — not KMT2A::AFF1. 2004 mass-spectrometry depth. The WRAD negative is a co-IP negative in two fusion contexts, which is real evidence but not exhaustive; later work may have shown WRAD association with fusions by other means, and this analysis did not attempt to settle that literature. **The universal is false and the qualification matters.** The SET domain and its WRAD contacts are lost from the **direct 5' product** only. Two classes keep them: the **reciprocal der(4) product AFF1::KMT2A**, which carries KMT2A exons 12-37 including the WIN motif and SET domain and is expressed in a large minority of t(4;11) infants; and **NUP98::KMT2A** (inv(11)(p15q23)), which deletes the exon-1 menin motif while keeping WIN and SET, and is menin-inhibitor resistant. Both counterexamples were already in this graph when the universal was written — a linking failure, not a knowledge gap. Corrected 2026-09-20 after the Q23 adjudication; verified by the coordinator against both abstracts.

• The enumeration's headline - that the WRAD negative on fusion MLL is 'the sharpest correction this enumeration produced' - depends on a 2004 co-IP negative bearing more weight than co-IP negatives bear. Lambert et al. ran BioID and a chromatin-optimised AP-MS protocol on the same baits and found they 'identified few shared interaction partners and enriched for distinct biological processes', with BioID specifically recovering lower-abundance proteins. On that calibration, Yokoyama's negative licenses 'ASH2L, WDR5 and RBBP5 are not stable, co-purifying members of the fusion complex' and does not license 'they are not fusion interactors'. The discrepancy report against gene-wdr5, gene-rbbp5, gene-ash2l and gene-dpy30 should be softened to that wording. Read: abstract in full, not full text. Lambert JP et al. 2014
• The deeper problem with the cross-beat discrepancy is that it is partly a category error. Whether WRAD coprecipitates with the fusion is a different question from whether WRAD matters in KMT2A-rearranged leukemia, and the answer to the second is yes. Thiel et al. show that menin recruits both wild-type MLL and MLL-AF9 to HOX loci, that wild-type MLL controls MLL-AF9-induced histone methylation, and that the wild-type Mll allele is required for MLL-AF9 leukemogenesis and for maintenance of transformed cells. Every KMT2A-rearranged patient retains an intact allele carrying the full WRAD-binding SET module. Deleting or demoting the WRAD entries on the basis of a co-IP negative would remove biology the disease demonstrably depends on. Read: abstract in full, not full text. Thiel AT et al. 2010
• And the coupling is mechanical, not just correlative. Milne et al. map an essential MLL1 recruitment domain that spans the CXXC domain AND the PHD fingers and is controlled by PAF1C contact and H3K4me2/3 reading, and report that MLL1 fusion proteins, lacking the PHD fingers, 'require prebinding of a wild-type MLL1 complex and CXXC domain recognition of DNA for stable HoxA9 association'. The enumeration's positional table classifies contacts as 'in the fusion' or 'not in the fusion'; this result says a third category exists and is load-bearing - contacts on the discarded fragment that the fusion depends on in trans. Read: abstract in full, not full text. Milne TA et al. 2010
• The practical consequence, and the sharpest test of the enumeration's 'redirect away from the wild-type complex literature' recommendation: MM-401, an inhibitor of the MLL1-WDR5 interaction, specifically blocks proliferation of MLL-rearranged cells with cell-cycle arrest, apoptosis and myeloid differentiation and without general toxicity to normal bone marrow or non-MLL cells, and does not affect other MLL-family HMTs. A KMT2A-selective agent already exists that was derived from a WRAD contact this enumeration classifies as outside the fusion interactome. Read: abstract in full, not full text. Cao F et al. 2014
• Symmetrically for MOF: the enumeration places KAT8 on the discarded fragment, correctly on Dou 2005's mapping, but Valerio et al. show conditional Mof deletion reduces tumour burden and prolongs survival in MLL-AF9 leukemia, that MOF's HAT activity is required (catalytically dead rescue fails), and that the MYST inhibitor MG149 is antiproliferative in the human KMT2A-rearranged lines MOLM13 and NOMO1. Read from the full text, two qualifications the abstract hides and that this graph should carry: Mof loss does not downregulate Hoxa9 or Meis1, and Kasumi-1, U937 and K562 are MG149-sensitive too - so MOF is a genuine requirement of KMT2A-rearranged leukemia but a non-selective one. Read: FULL TEXT via PMC5501293. Valerio DG et al. 2017
• The reciprocal AFF1::KMT2A product is the exact mirror of the one this entry describes: **WRAD-competent and menin-blind**. In the subset of t(4;11) infants expressing it, the cell contains a second KMT2A-derived oncoprotein with an intact methyltransferase module that a menin inhibitor cannot engage by construction. Any claim about what 'the fusion' retains has to say which one. Oncogene 2021

A null search is bounded by query terms and by indexing. A study could exist under wording these queries miss (for example 'APEX2', 'biotin ligase', or a methods paper whose abstract does not name the fusion), and preprints are not indexed in PubMed. The claim should be read as 'not findable by these queries', which is weaker than 'does not exist'.

• 'Nobody has ever done proximity labelling' is true for KMT2A and false for chromatin-bound fusion oncoproteins as a class, and the precedents change what should be expected of the missing experiment. Elzi et al. ran BioID on EWS-FLI1 and the hit they pursued was the cation-independent mannose 6-phosphate receptor, a trans-Golgi/endosome cargo receptor, which led to a lysosomal-turnover mechanism rather than to a chromatin partner. A proximity map of a nuclear fusion can be dominated by compartment and abundance bystanders, so the enumeration's framing of it as the adjudicating experiment for the WRAD question is optimistic. Read: abstract in full, not full text. Elzi DJ et al. 2014
• A second precedent, aimed at exactly the question this enumeration cares about. Kitagawa et al. applied BioID to EWS-ETS fusion transcription factors 'to identify the transcriptional complexes' and recovered AHDC1, previously absent from the EWS-FLI1 interactome, whose knockdown reduced EWS-FLI1 transcriptional activity and lowered BRD4 and BRG1. The canonical partners were reached only indirectly. Feasible, therefore; adjudicating, probably not. Read: abstract in full, not full text. Kitagawa T et al. 2022
• The enumeration's implicit premise - that fusion-oncoprotein complex composition has not been revisited with modern proteomics - is also too strong for the wider KMT2A field. Xu et al. showed by modern AP-MS that NUP98 fusion proteins (NHA9, NHD13, NUP98-NSD1, NUP98-PHF23, NUP98-TOP1) physically interact with MLL1 and with the NSL histone-modifying complex, that NHA9 and MLL1 co-localise at Hox promoters, and that MLL1 is required for NHA9 cell proliferation in vitro and in vivo. Full-length MLL1 - the protein carrying every 'discarded' contact - is itself a fusion-associated, functionally required complex member in a related leukemia. Read: abstract in full, not full text. Xu H et al. 2016
• Search extended for this repair beat on 2026-09-18 with four further query forms against the NCBI PubMed E-utilities - (BioID OR TurboID OR APEX2 OR 'proximity labeling' OR 'proximity labelling' OR 'proximity-dependent biotinylation') AND (KMT2A OR MLL1 OR 'mixed lineage leukemia'); 'proximity labeling fusion oncoprotein interactome leukemia biotin ligase'; 'TurboID leukemia fusion protein nuclear interactome'; 'BioID menin interactome'. All four returned zero indexed articles, so the beat's negative holds under a harder search than the one that produced it. It remains bounded by PubMed indexing and excludes preprints. Oncofusion Knowledge Graph project et al. 2026

DepMap release version not stated by the API. Eleven KMT2A-rearranged lines, dominated by AML rather than the infant B-ALL this project targets. Differences computed on pan-essential genes have little variance to work with and should not be read directionally.

• A selectivity difference expressed as a rank inherits this project's own correction to its instrument: the scan's statistic divides by comparator standard deviation, which ranks NARROWNESS rather than effect size at Spearman −0.92 in the depleted tail. A rank from that scan is not an effect-size ordering, and 167 entries were written before that was measured. MEN1 ranking first on raw effect size is now the supported statement; the ranks quoted here for FLT3, PSIP1 and KMT2A come from the scaled statistic and move when the scale changes — FLT3 from 1st to 2nd, KMT2A from 13th to 5th. Oncofusion knowledge graph (internal analysis) 2026