Oncofusion

start / graph / gene-sumo1

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q78-interactome-absent

SUMO1 - not a complex partner of LEDGF but a modifier of it, and the interactome row cannot tell you which

also called SUMO1 · SMT3C · SMT3H3 · UBL1 · Small ubiquitin-related modifier 1 · GAP-modifying protein 1 · SUMO-1

SUMO1 is a small tag cells stick onto other proteins to change what they do. It is stuck onto LEDGF - the protein that anchors the leukaemia fusion to DNA - and removing the tag makes LEDGF more active and more abundant.

The same thing, for a scientist

SUMO1 (P63165, **101 aa**) is a ubiquitin-like modifier conjugated to lysine side chains. It appears in the PSIP1 interactome with **four IntAct records, none spoke-expanded**, all from Ishihara et al. 2012 in lymphatic endothelial cells: three `MI:0914 association` co-immunoprecipitations and one `MI:0403 colocalization` by fluorescence imaging [E1]. **That set is a category distinct from every other partner in this beat.** A SUMO record is a covalent modification, not a complex; the 'interaction' is an enzyme-substrate relationship with a transient thioester intermediate, and a colocalization row asserts only that two signals overlap. LEDGF is constitutively SUMOylated; deSUMOylation by SENP1, or the K364R mutation, **enhances LEDGF's transactivation capacity and its cellular abundance**, increases DNA binding and HSP27 transcription, and improves cell survival, while SUMO1 overexpression reverses it [E2]. An independent group mapped four sites - **K75, K250, K254 in the shared N-terminus and K364 in the p75-exclusive C-terminus** - and found SUMOylation extends nothing about localisation or chromatin binding but shortens LEDGF's half-life and lowers its transcriptional activity [E3]. So the tag is a **negative regulator of LEDGF dose**, and LEDGF dose is what the KMT2A fusion depends on.

Go deeper
Why a SUMO row is not a partner row

concept-what-a-curated-interaction-row-actually-claims distinguishes what different MI types assert. SUMO adds a case that concept does not cover:

  • The pair is covalent and transient. A SUMO-substrate co-IP catches an isopeptide-linked conjugate, not an association equilibrium. There is no Kd, no interface in the usual sense, and no stoichiometry to report.
  • The colocalization row (MI:0403) asserts even less. Two fluorescent signals overlapping in a nucleus is not a contact; SUMO1 is in every nucleus.
  • The functional direction is what matters, and it is inhibitory. More SUMO1 on LEDGF means less LEDGF activity and less LEDGF protein 23.

So counting SUMO1 as a PSIP1 'interactor' in a coverage metric is a category error of the same family as counting a protein-DNA ChIP record as a protein-protein interaction - which is exactly what cycle 43's REST recount caught when hoxa9_human_gene appeared at the top of a partner list. This one is subtler because SUMO1 is a protein.

The consequence for the fusion, stated as a prediction

The KMT2A fusion reaches chromatin through LEDGF's IBD, and the amount of LEDGF available is therefore a dose term in the fusion's tethering. Two published facts make that dose adjustable:

  1. SUMOylation shortens LEDGF's half-life and lowers its transcriptional activity; K-to-R mutation of the acceptor sites extends the half-life and raises the activity 3.
  2. SENP1 deSUMOylation raises LEDGF abundance and transactivation; SUMO1 overexpression lowers both 2.

The untested prediction is therefore: raising SUMO1 conjugation on LEDGF should lower fusion occupancy at HOXA, and it would do so without touching the PWWP cage or the IBD - a different pharmacological surface from anything this project currently considers. Nobody has measured LEDGF SUMOylation in any leukaemia cell, and K364 is in the p75-exclusive C-terminus, which is the isoform that carries the IBD.

Facts and figures
uniprotP63165
length AA101
gene symbolSUMO1
protein nameSmall ubiquitin-related modifier 1 (SUMO-1, GAP-modifying protein 1, Ubiquitin-like protein UBL1)
uniprot resolutionaccession:P63165, reviewed human, one record; primary gene name read back as SUMO1 (entry SUMO1_HUMAN), synonyms SMT3C, SMT3H3, UBL1; length 101 aa checked against UniProt 2026-09-21.
symbol collision checkdata/symbol-collisions.json greped 2026-09-21 for SUMO1: no row existed. A row IS proposed in this beat's memo - not for the symbol but for the CLASS: a SUMO/ubiquitin/NEDD8 interaction record is a covalent modification and an enzyme-substrate relationship, and counting it as a complex partner in an interactome coverage metric is a category error.
baitPSIP1 (O75475)
functional directionSUMOylation LOWERS LEDGF transcriptional activity and shortens its half-life; deSUMOylation raises both. It does NOT change LEDGF localisation or chromatin binding.
shown for fusionNO - nobody has measured LEDGF SUMOylation in any leukaemia cell or in the presence of any KMT2A fusion.
fetched2026-09-21
subnuclear localizationPML NUCLEAR BODIES ARE THE ONE SUPPORTED SUBNUCLEAR COMPARTMENT, AND THE NUCLEAR-SPECKLE CALL IS NOT EVIDENCE ABOUT SUMO1 AT ALL. UniProt P63165 carries 'Nucleus, PML body' on three ECO:0000269 attributions and 'Nucleus speckle' on a single ECO:0000250 - INFERRED BY SIMILARITY FROM P63166, i.e. from SUMO2. There is no experimental speckle observation for SUMO1 in the record; the compartment was transferred from a paralogue that is only 44% identical to it [E_q91_sumo1_loc1]. Treat SUMO1-in-speckles as unmeasured. The PML-body assignment is itself target-derived - PML and SP100 are SUMO1 substrates and their SUMOylation is what is imaged [E_q91_sumo1_loc4] - and diffusely distributed nuclear SUMO1 can be SEQUESTERED into ectopic ZFHX3/ATBF1 dots, which shows the nuclear pool follows its substrates rather than occupying a compartment of its own.
localization shown for fusionNO - nobody has looked, in any fusion, in any lineage, and the retrieval is bounded rather than assumed: SUMO1[tiab] AND MLL[tiab] and "SUMO-1"[tiab] AND MLL[tiab] both return ZERO against operands of 1,794 and 901, and SUMO1[tiab] AND KMT2A[tiab] returns zero [E_q91_sumo1_loc5]. The closest existing result is not a localisation: the SUMO isopeptidase SENP3 associates with MLL1/MLL2 methyltransferase complexes and deSUMOylates RBBP5, controlling H3K4 methylation at HOX genes [E_q91_sumo1_loc6]. That is a SUMO-pathway contact with the complex this graph is about, on a different subunit, with no compartment attached.
interaction with psip1
{
  "source_pmid": "22748127",
  "source_class": "focused mechanistic study",
  "relationship_class": "covalent post-translational modification, NOT a complex",
  "n_intact_records": 4,
  "n_spoke_expanded": 0,
  "n_source_publications": 1,
  "intact_acs": [
    "EBI-7946621",
    "EBI-7946637",
    "EBI-7946671",
    "EBI-7946713"
  ],
  "interaction_types": {
    "MI:0914 association": 3,
    "MI:0403 colocalization": 1
  },
  "system": "Homo sapiens LEC lymphatic endothelial cell",
  "brief_stated_record_count": 2,
  "rest_record_count": 4,
  "shared_pdb_entries": []
}
sumoylation sites on psip1
{
  "K75": "shared N-terminal region (p75 and p52)",
  "K250": "shared N-terminal region",
  "K254": "shared N-terminal region",
  "K364": "LEDGF/p75-exclusive C-terminal region - the isoform that carries the integrase-binding domain",
  "source_pmids": [
    "20382164",
    "22748127"
  ],
  "modifiers": [
    "SUMO-1",
    "SUMO-3"
  ],
  "deconjugating_enzyme": "SENP1"
}
localization
{
  "THE_QUESTION_HAS_TWO_ANSWERS_AND_THEY_DIFFER": "FREE SUMO1 MONOMER IS PREDOMINANTLY CYTOSOLIC; CONJUGATED SUMO1 IS PREDOMINANTLY NUCLEAR. This is the distinction every SUMO1 localisation statement has to make and mostly does not. The founding measurement is explicit: immunocytochemical staining AND cell fractionation showed that the sentrin (SUMO1) MONOMER is localised predominantly to the CYTOSOL, while p90 and the majority of SENTRINIZED (conjugated) proteins are localised to the NUCLEUS [E_q91_sumo1_loc2]. So 'SUMO1 is nuclear' is a statement about its substrates' compartment, not about where the free modifier is.",
  "CONSTITUTIVE": "CYTOSOL (free monomer) plus NUCLEUS (conjugated pool) plus a specific and large NUCLEAR-ENVELOPE / CYTOPLASMIC-FACE-OF-THE-NUCLEAR-PORE pool. The pore pool is not a generic 'nuclear membrane' call: SUMO1 conjugation is what targets RanGAP1 to the cytoplasmic periphery of the nuclear pore complex, where it associates with RanBP2/NUP358, and that single conjugate is abundant enough to dominate steady-state SUMO1 signal at the envelope [E_q91_sumo1_loc3]. UniProt P63165 (fetched live 2026-09-21) lists Nucleus membrane, Cytoplasm, Nucleus, Nucleus PML body, Nucleus speckle and Cell membrane [E_q91_sumo1_loc1].",
  "system_note": "None of it haematopoietic. The fractionation was in COS cells, the pore result in mammalian tissue-culture and in vitro systems, the PML-body work in HeLa-type cells."
}
cell type
{
  "measured_in": "COS (monkey kidney), HeLa, HEK293 and NT2 for the localisation and abundance work [E_q91_sumo1_loc1][E_q91_sumo1_loc2]; lymphatic endothelial cells for the PSIP1/LEDGF SUMOylation this entry already rests on (its E1).",
  "haematopoietic": "NONE. SUMO1 has never been localised in a haematopoietic cell in anything this beat could retrieve, let alone in a KMT2A-rearranged blast.",
  "note": "SUMO1 is ubiquitously expressed, so 'which cell type' is a weaker question for it than for a lineage factor - but the compartment question is not weaker, because the free/conjugated ratio is cell-state-dependent and has never been measured in a leukaemic cell."
}
cell state
{
  "CELL_CYCLE": "The conjugated pool is cell-cycle-regulated at a named compartment: PML and SP100 are SUMO-1 conjugated during interphase and DE-CONJUGATED during mitosis, and the two - which tightly colocalise in ND10/PML bodies in interphase - BECOME SEPARATED during mitosis [E_q91_sumo1_loc4]. The PML-body SUMO1 signal is an interphase state, not a constant.",
  "STRESS": "Heat shock and phosphatase inhibition both change ND10 structure and the PML modification pattern [E_q91_sumo1_loc4]. A SUMO1 compartment measured in an unstressed line does not transfer to a stressed one.",
  "not_measured": "Nothing about the free/conjugated ratio, the PML-body pool or the pore pool has been measured under menin inhibition or in any fusion-bearing cell."
}
localization bounded absence
{
  "date": "2026-09-22",
  "tool": "NCBI esearch db=pubmed",
  "run_by": "Q91",
  "queries": [
    {
      "query": "SUMO1[tiab]",
      "querytranslation": "\"SUMO1\"[Title/Abstract]",
      "count": 1794,
      "role": "operand control - the instrument can return a one"
    },
    {
      "query": "\"SUMO-1\"[tiab]",
      "querytranslation": "\"SUMO-1\"[Title/Abstract]",
      "count": 901,
      "role": "second spelling, run because hyphenation is a different string"
    },
    {
      "query": "SUMO1[tiab] AND MLL[tiab]",
      "querytranslation": "\"SUMO1\"[Title/Abstract] AND \"MLL\"[Title/Abstract]",
      "count": 0
    },
    {
      "query": "\"SUMO-1\"[tiab] AND MLL[tiab]",
      "querytranslation": "\"SUMO-1\"[Title/Abstract] AND \"MLL\"[Title/Abstract]",
      "count": 0
    },
    {
      "query": "SUMO1[tiab] AND KMT2A[tiab]",
      "querytranslation": "\"SUMO1\"[Title/Abstract] AND \"KMT2A\"[Title/Abstract]",
      "count": 0
    },
    {
      "query": "sumoylation[tiab] AND MLL[tiab]",
      "querytranslation": "\"sumoylation\"[Title/Abstract] AND \"MLL\"[Title/Abstract]",
      "count": 3,
      "note": "the pathway term, run because rule 71 says a pathway is often named by something other than its gene symbol. One of the three is the SENP3/RBBP5 result cited above; none reports a compartment."
    },
    {
      "query": "\"SUMO-1\"[tiab] AND (nucleolus[tiab] OR nucleolar[tiab])",
      "querytranslation": "\"SUMO-1\"[Title/Abstract] AND (\"nucleolus\"[Title/Abstract] OR \"nucleolar\"[Title/Abstract])",
      "count": 22,
      "note": "a nucleolar SUMO1 literature exists and is not empty; it is simply not connected to this disease. Not screened in detail - this entry claims no nucleolar assignment either way."
    }
  ],
  "reading": "Two large literatures with an empty intersection at the level of the modifier's own name. Titles and abstracts only; supporting figures not searched (rule 80), so a SUMO1 immunofluorescence panel inside an MLL paper would not have been seen.",
  "absence_class": "never-measured"
}
How this could bite us back
  • SUMO pathway inhibitors exist and are in trials. Reading SUMO1 as a 'LEDGF partner' would make the SUMO machinery look like a handle on the fusion; the published direction is that SUMOylation LOWERS LEDGF, so a SUMO E1 inhibitor would be expected to RAISE LEDGF and could make fusion tethering worse.
  • SUMO1 conjugates thousands of substrates. Nothing about a SUMO-directed intervention would be selective for LEDGF, let alone for the p75 isoform, except possibly at K364.
  • An interactome coverage figure that counts SUMO1 as a PSIP1 partner is counting a modification as a complex. The same recount that removed protein-DNA ChIP rows from the partner list should look at ubiquitin-like modifiers next.
Still open
  1. Is LEDGF SUMOylated in a KMT2A-rearranged blast, and does the level change under a menin inhibitor? LEDGF dose is a term in fusion tethering and nobody has measured its modification state in this disease.
  2. Does raising SUMO1 conjugation on LEDGF lower fusion occupancy at HOXA? K364R and a SENP1 knockdown are both available reagents and the experiment has never been run in a leukaemia line.
  3. Does the KMT2A fusion's IBD contact occlude K364? Both are in the p75-exclusive C-terminus. If the fusion protects the site, fusion-bound LEDGF would have a longer half-life than free LEDGF, which is a self-reinforcing loop nobody has proposed.
Every source behind this entry (9)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct's REST record holds four PSIP1-SUMO1 records, none spoke-expanded, all from PMID 22748127 in human lymphatic endothelial cells: three typed MI:0914 association by anti-bait co-immunoprecipitation (including one with SUMO1 as the A-side bait) and one typed MI:0403 colocalization by fluorescence imaging.

    What this evidence cannot show

    One database on one day, REST endpoint (brief rule 93). Four records from ONE publication. A SUMO-substrate record is a covalent modification and an enzyme-substrate relationship, not a complex; the MI:0403 colocalization row asserts signal overlap only. Lymphatic endothelial cells - no haematopoietic system, no KMT2A fusion (brief rule 73).

    Evidence that argues against this (2)
    • Re-read at the endpoint on 2026-09-21 the roles are not as described, and the publication holds a fifth row that is nobody's. No record has SUMO1 as the A-SIDE BAIT: EBI-7946637 has PSIP1 on the A side as prey with SUMO1 as the B-side bait (figure f1e), and EBI-7946671 has SUMO1 on the A side as PREY with PSIP1 as bait (f5a/f5b) - the claim's parenthesis merges the A/B ordering with the experimental role. The four rows are annotated to f1d, f1e, f1c and f5a/f5b and all carry intactMiscore 0.43. findInteractions/22748127 also returns a fifth record, EBI-22748127, a two-hybrid-array row between Q8TCE9 and O95994 annotated to 'Supplementary Table 9' and involving neither protein - so a publication-keyed query returns rows that are not the paper's own subject, and 'all from PMID 22748127' is true of these four while the publication's record is not four. EMBL-EBI IntAct 2026
    • A co-immunoprecipitation cannot tell a covalent conjugate from a noncovalent SIM contact, so this entry's central framing - 'the pair is covalent and transient, there is no Kd and no interface in the usual sense' - asserts more than the record supports. SUMO paralogues bind proteins noncovalently through SUMO-interacting motifs, and in HIV integrase, which is LEDGF/p75's canonical IBD ligand, two bona fide SIMs (200-IVDI-203 and 257-IKVV-260) mediate IN-SUMO noncovalent interactions, negatively regulate IN's own SUMOylation, and are REQUIRED for the IN-LEDGF/p75 interaction. An MI:0914 association row between LEDGF and SUMO1 is therefore consistent with a noncovalent SUMO-SIM contact that does have an interface and a Kd, and the category argument this entry builds needs the SUMOylation-site mapping to carry it, not the interaction rows. Zheng Y et al. 2019
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In human lymphatic endothelial cells (no haematopoietic system, no KMT2A fusion), LEDGF is constitutively SUMOylated at K364; deSUMOylation by SENP1 or the K364R mutation enhances LEDGF's transactivation capacity and cellular abundance, increases its DNA-binding activity and HSP27 transcription, and improves cell survival, while SUMO1 overexpression reverses these effects.

    PMID 22748127DOI
    What this evidence cannot show

    Abstract read this session. Lymphatic endothelial cells and transfection-based assays throughout; no haematopoietic cell, no KMT2A fusion (brief rule 73). New to this graph (checked with scripts/check-citation-reuse.py). Site mapping in this paper is bioinformatic plus in vitro and in vivo SUMOylation assay; it names K364 only, while the independent study below maps four sites.

    Evidence that argues against this (2)
    • The independent study contradicts this one on a readout that matters here. Ishihara et al. report that deSUMOylation INCREASES LEDGF's DNA-binding activity; Bueno et al., mapping the same modification in a different laboratory, report explicitly that SUMOylation 'was not necessary for their chromatin-binding ability, nor did it affect this activity' and did not change cellular localisation. The two agree on direction for transcriptional activity and half-life and disagree on whether DNA or chromatin association moves at all - which is precisely the property this entry's fusion-occupancy prediction depends on. Bueno MT et al. 2010
    • K364 is one of four acceptor sites, and three of them are outside the region this claim is about. Bueno et al. map K75, K250 and K254 in the N-terminal region SHARED with LEDGF/p52 alongside K364 in the p75-exclusive C-terminus. A K364R single mutant therefore leaves three acceptor lysines intact, so 'LEDGF is constitutively SUMOylated at K364' names one site of four and the K364R phenotype cannot be read as loss of SUMOylation - either K364 dominates, which nobody has shown, or the phenotype is not proportional to total SUMO load. Ishihara's mapping is bioinformatic prediction plus assay and names only the consensus-motif site, which is the site a consensus search would find. Bueno MT et al. 2010
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Independently, LEDGF/p75 and LEDGF/p52 are modified by SUMO-1 and SUMO-3 at four mapped sites - K75, K250 and K254 in the shared N-terminal region and K364 in the p75-exclusive C-terminal region; SUMOylation does not affect LEDGF localisation or chromatin binding, but lysine-to-arginine mutation of the acceptor sites extends LEDGF/p75's half-life and significantly increases its transcriptional activity on the HSP27 promoter.

    PMID 20382164DOI
    What this evidence cannot show

    An earlier, independent study from a different group, which is what makes the K364 site corroborated rather than single-source. The systems are transfected cell lines; no haematopoietic cell and no KMT2A fusion (brief rule 73). The three N-terminal sites lie in regions shared with LEDGF/p52, which does not carry the IBD, so a SUMO-directed intervention would not be p75-selective except at K364.

    Evidence that argues against this (1)
    • This study explicitly reports that SUMOylation does NOT affect LEDGF's cellular localisation or its chromatin-binding ability. So the predicted route to lowering fusion occupancy runs through LEDGF protein level and transcriptional activity, not through displacing LEDGF from chromatin - and a chromatin-occupancy readout would show nothing. Bueno MT et al. 2010
  4. E_q91_sumo1_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB P63165 (SUMO1_HUMAN, primary gene SUMO1, synonyms SMT3C/SMT3H3/UBL1, 101 aa, read back live 2026-09-21) carries SIX subcellular location values whose evidence is unevenly strong: 'Nucleus membrane' (ECO:0000269 PubMed:10574707, 12383504), 'Cytoplasm' (ECO:0000269 PubMed:12383504, 9162015), 'Nucleus' (ECO:0000269 PubMed:24651376, 9162015), 'Nucleus, PML body' (ECO:0000269 PubMed:10574707, 12383504, 22406621), 'Cell membrane' (ECO:0000269 PubMed:19223394) and - critically - 'Nucleus speckle' on a SINGLE ECO:0000250 attributed to P63166, i.e. INFERRED BY SIMILARITY FROM SUMO2. There are no molecule-scoped comments and no annotated NLS in the feature table. The one location note is condition-scoped: SUMO1 is recruited by BCL11A into the nuclear body (by similarity) and, in the presence of ZFHX3, sequestered into nuclear-body-like dots some of which overlap PML bodies (PubMed:24651376).

    What this evidence cannot show

    A curated composite over five papers in five cell systems, none haematopoietic. The speckle row is the one to discard for SUMO1: an ECO:0000250 is not a measurement of this protein, and SUMO1 and SUMO2 share only 44% sequence identity (PubMed:12383504) - a similarity this project would not accept for transferring a compartment between paralogues.

    Evidence that argues against this (1)
    • The argument that these six rows describe SUMO1 is weaker than the count suggests, and the weakness runs in one direction: FOUR of the six values are supported by papers whose subject is a SUMO1 TARGET rather than SUMO1. PubMed:10574707 is about PML and SP100 modification and ND10 composition; PubMed:24651376 is about ATBF1/ZFHX3's nuclear localisation and its SUMOylation. Where the modifier is imaged in those papers it is imaged as a proxy for the modified protein. Against this stands the one paper that asked the question directly and got a DIFFERENT answer for the free protein - cytosol, not nucleus - which is why this entry splits the field rather than averaging it. Kamitani T et al. 1997
  5. E_q91_sumo1_loc2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In COS cells transfected with haemagglutinin-tagged sentrin (SUMO1) constructs, immunocytochemical staining AND cell fractionation together showed that the SENTRIN MONOMER is localised PREDOMINANTLY TO THE CYTOSOL, whereas p90 and the majority of sentrinized (conjugated) proteins are localised to the NUCLEUS. Changing the conserved C-terminal Gly-Gly to Gly-Ala abolished the high-molecular-mass conjugates while leaving monomer and p90 - the internal control that the nuclear signal is conjugation-dependent.

    PMID 9162015DOI
    What this evidence cannot show

    1997, COS cells, epitope-tagged and over-expressed, bulk rather than quantitative fractionation, and 'predominantly' is not quantified. It is nonetheless the only experiment retrieved that separates the two species, and no later work contradicts it in anything this beat could reach.

    Evidence that argues against this (1)
    • This does not license the reverse over-reading - that free SUMO1 is EXCLUDED from the nucleus. A later study of ATBF1 explicitly describes DIFFUSELY DISTRIBUTED NUCLEAR SUMO1 that can be sequestered into ectopic ATBF1 nuclear-body-like dots, and sequestration requires a free nuclear pool to sequester. The honest statement is a gradient - monomer enriched in cytosol, conjugates enriched in the nucleus - not a partition. Both are over-expression experiments in epithelial cells and neither measures the endogenous ratio in any cell of interest here. Sun X et al. 2014
  6. E_q91_sumo1_loc3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The single largest named SUMO1 pool is at the CYTOPLASMIC PERIPHERY OF THE NUCLEAR PORE COMPLEX, and it is there as a conjugate: mammalian RanGAP1 is highly concentrated at the cytoplasmic face of the NPC where it associates with the 358-kDa Ran-GTP-binding protein RanBP2, and that interaction REQUIRES the ATP-dependent post-translational conjugation of RanGAP1 with SUMO-1, described in this paper as a novel 101-amino-acid ubiquitin-related protein.

    PMID 9019411DOI
    What this evidence cannot show

    This is the founding SUMO-1 paper and it is a statement about RanGAP1's address rather than SUMO1's - but it is why the steady-state SUMO1 signal at the nuclear envelope exists, and an entry recording 'Nucleus membrane' from UniProt without naming RanGAP1 has recorded the shadow rather than the object. Not haematopoietic; no fusion anywhere near it.

    Evidence that argues against this (1)
    • This cuts against reading UniProt's 'Nucleus membrane' row as a property of SUMO1. If most envelope-associated SUMO1 is one conjugate held by one receptor, then depleting or mutating RanGAP1 would move the SUMO1 envelope signal without changing anything about SUMO1 - and a SUMO1 compartment map is really a map of which substrates are abundant in that cell. For this graph there is a concrete consequence: SUMO1's envelope pool sits on the same structure as NUP214, the subject of a sibling entry in this tranche, and neither entry may be read as evidence about the other. Kraemer D et al. 1994
  7. E_q91_sumo1_loc4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The PML-body SUMO1 signal is an INTERPHASE state rather than a constant: PML and SP100, both SUMO-1 conjugated during interphase, become DE-CONJUGATED during mitosis, and the two proteins - which normally tightly colocalise in ND10/PML bodies in interphase cells - BECOME SEPARATED during mitosis. Interphase cells treated with phosphatase inhibitors or subjected to heat shock also show structural changes in ND10 accompanied by an altered PML modification pattern.

    Cell cycle regulation of PML modification and ND10 composition. · Everett RD, Lomonte P, Sternsdorf T, van Driel R et al. · J Cell Sci 1999
    PMID 10574707DOI
    What this evidence cannot show

    The paper images PML and SP100 and infers the SUMO-1 state from electrophoretic mobility; it is not a SUMO1 localisation study, and UniProt nevertheless cites it as the ECO:0000269 experimental support for BOTH 'Nucleus membrane' and 'Nucleus, PML body' on the SUMO1 record. A RULE-91 INSTANCE: the evidence code correctly says a curator saw an experiment, and the experiment is about the substrates.

    Evidence that argues against this (1)
    • A reader could object that cell-cycle deconjugation is a detail. It is not, for this graph specifically, because this project holds cell-state entries that turn on what can and cannot be measured in a cycling blast population. If SUMO1's only supported subnuclear compartment exists only in interphase and dissolves in mitosis, a bulk immunofluorescence of an asynchronous leukaemic culture averages two different SUMO1 distributions - and there is no published image of SUMO1 in any leukaemic cell to check that against, per data.localization_bounded_absence. Kamitani T et al. 1997
  8. E_q91_sumo1_loc5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    SUMO1 has never been localised in, or in the presence of, any KMT2A fusion. Measured 2026-09-22 with querytranslation read back on every line: SUMO1[tiab] = 1,794 and "SUMO-1"[tiab] = 901 as operand controls; SUMO1[tiab] AND MLL[tiab] = 0; "SUMO-1"[tiab] AND MLL[tiab] = 0; SUMO1[tiab] AND KMT2A[tiab] = 0; sumoylation[tiab] AND MLL[tiab] = 3.

    What this evidence cannot show

    Bounded by ABSTRACTS, not by papers (rule 80): a SUMO1 immunofluorescence panel inside an MLL paper's supporting figures would be invisible to this. Two spellings and the nominalised pathway term were run; the SUMO2/SUMO3 paralogues and the retired synonyms SMT3C, SMT3H3 and UBL1 were not separately enumerated.

    Evidence that argues against this (1)
    • The zero is real but it is the wrong zero to quote as 'SUMO is unstudied in this complex'. Running the pathway term instead of the gene symbol - the move rule 71 prescribes - returns a paper that puts a SUMO enzyme inside the machine this project is about: SENP3 associates with MLL1/MLL2 complexes, deSUMOylates RBBP5 and controls H3K4 methylation at HOX genes. The honest claim is narrow - nobody has imaged SUMO1 in a fusion-bearing cell - and the broad claim that SUMO biology is absent from the KMT2A literature is false. This is the rule-72 shape: a symbol list returning zero while the mechanism is published under another name. Nayak A et al. 2014
  9. E_q91_sumo1_loc6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The SUMO-specific isopeptidase SENP3 is associated with MLL1/MLL2 (KMT2A/KMT2B) methyltransferase complexes and catalyses deSUMOylation of RBBP5; SENP3 thereby controls H3K4 methylation, and the paper frames MLL1/MLL2 complexes - which contain menin as an additional component alongside WDR5, RBBP5, ASH2L and DPY-30 - as particularly important for the activation of HOX genes.

    PMID 24930734DOI
    What this evidence cannot show

    Wild-type MLL1/MLL2, not a fusion; the phenotype studied is osteogenic differentiation, not leukaemia; and the SUMOylated subunit is RBBP5, not SUMO1's own compartment. Cited here as the nearest SUMO-to-KMT2A-complex contact and as the counter-example to a symbol-based absence, not as fusion evidence.

    Evidence that argues against this (1)
    • It is tempting to read this as 'SUMO regulates the KMT2A complex, therefore SUMO1 matters to the fusion'. Two things block that. The paper does not establish that SUMO1 rather than the SUMO2/3 branch is the modifier on RBBP5; and, decisively for this graph, RBBP5 and ASH2L contact the C-terminal SET-domain region of KMT2A that the fusion does NOT retain, so a WRAD-directed SUMO switch acts on the wild-type allele and the unrearranged paralogue rather than on KMT2A::AFF1. The mechanism is real and it is on the wrong side of the breakpoint. Mahajan R et al. 1997
Connected to