Oncofusion

start / graph / gene-set

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q37-histone-chaperones

SET / TAF-I beta - a histone chaperone that binds the retained N-terminus of MLL, co-activates HOXA with it, and is itself a HOXA-activating fusion partner that recruits DOT1L

also called SET · TAF-I · TAF-Ibeta · TAF-1beta · I2PP2A · PHAPII · IGAAD · SET-CAN · template-activating factor I · Q01105

SET is a histone chaperone nobody in this project had written down, and it turns out to stick to the exact piece of the MLL protein that every KMT2A fusion keeps - and the two together switch on the same HOXA genes the fusion is famous for. When SET itself gets fused to a different partner it causes leukaemia the same way.

The same thing, for a scientist

SET (Q01105, 290 aa), also called TAF-I beta, is an H2A-H2B histone chaperone and INHAT component. It entered this beat **only because a generic-phrase query found what every symbol query missed** - `"histone chaperone" AND (MLL OR KMT2A)` returns 2 records where `HIRA AND (MLL OR KMT2A)` returns 0. Three results, none of them in this graph: 1. **The amino-terminal region of MLL (MLLN) interacts with TAF-I beta/SET** through TAF-I beta's acidic region; GST-MLLN pulls down both TAF-I beta and histones; and **MLLN and TAF-I beta synergistically upregulate Hoxa9 transcription and co-immunoprecipitate on chromatin containing the Hoxa9 promoter** [E1]. **MLLN is exactly what a KMT2A fusion retains.** 2. **SET-NUP214**, a recurrent fusion in paediatric T-ALL arising from del(9)(q34.11q34.13), **binds the promoter regions of specific HOXA genes where it interacts with CRM1 and DOT1L**; siRNA against it abolishes HOXA expression, inhibits proliferation and induces differentiation in LOUCY [E2]. 3. **SET-NUP214 interacts with MLL via the SET acidic region and the two cooperatively enhance the HOXA10 promoter**; neither the SET region alone nor the NUP214 region alone is sufficient [E3]. So the same acidic region of the same histone chaperone is the contact point for MLL in both directions, at the same gene cluster, in the same disease family - and it recruits this graph's principal drug target.

Go deeper
Why this is the most important thing in the beat and why it is still weak evidence

What it means if it holds. The graph's mechanism entries have KMT2A fusions activating HOXA through menin/LEDGF targeting plus DOT1L and SEC-driven elongation. This adds a fourth element at the same promoters - a histone chaperone bound to the retained N-terminus, contributing transcriptional activation, and independently able to pull DOT1L to HOXA promoters when it is itself the fusion. That is a testable alternative route to HOXA activation that does not run through the menin interface.

What is actually shown. Very little, by modern standards:

  • 1 is a 2005 FEBS Letters short report: yeast two-hybrid, GST pull-down, a reporter assay and a co-IP. The transcription result is promoter activity, and the co-IP is on 'chromatin containing the Hoxa9 promoter region', which is a ChIP-like readout described in one sentence.
  • 3 is a reporter assay plus mapping - again promoter activity, in a 2021 paper in a specialist journal.
  • 2 is a real Blood paper with patient data, but its subject is SET-NUP214 in T-ALL, not KMT2A.
  • None of these uses a KMT2A fusion protein. MLLN is the wild-type N-terminal fragment, not a fusion.

Nobody has asked whether a KMT2A fusion binds SET. That is a co-IP.

What this would do to the menin-interface hypothesis

This graph's therapeutic centre of gravity is the menin-KMT2A interface and the drugs that block it. A second, independent contact from the retained N-terminus to a transcriptional coactivator at the same target promoters is a resistance hypothesis by construction: if SET contributes to HOXA activation through a surface a menin inhibitor does not touch, then menin inhibition would leave part of the output intact.

The menin-binding motifs MBM1/MBM2 sit in the first ~45 residues of KMT2A. The SET contact is mapped only to 'MLLN', an N-terminal region, in a 2005 paper - so whether the two contacts overlap is unknown and is answerable from the existing constructs. If they overlap, a menin inhibitor might displace SET too and this is not a resistance route. If they do not, it is one.

That is the cheapest high-value experiment this beat identifies and it is recorded in exp-does-kmt2a-aff1-bind-set-taf1.

Facts and figures
uniprotQ01105
gene symbolSET
length AA290
chaperone classH2A-H2B chaperone / INHAT component (TAF-I beta)
kmt2a contactinteracts with the amino-terminal region of MLL (MLLN) via the SET/TAF-I beta acidic region; MLLN is the portion retained in every KMT2A fusion
own fusionSET-NUP214 (SET-CAN), del(9)(q34.11q34.13) or t(9;9)(q34;q34), recurrent in paediatric and adult T-ALL, HOXA-activating, DOT1L- and CRM1-interacting
symbol hazardthe bare symbol SET is unusable as a PubMed token - it collides with the SET methyltransferase domain, with SETD/SETDB symbols and with the English word
subnuclear localizationnucleoplasm, avoiding the nucleolus (UniProt Q01105 note, describing the granzyme-A-induced nuclear pool). No body, speckle or condensate assignment exists. The exclusion is worth recording explicitly because localization-nucleolus-kmt2a-fusion holds the only published images of a KMT2A fusion inside a compartment, and SET - which binds the retained MLL N-terminus - is annotated as staying out of it.
shown for fusionNO for KMT2A, and the query is unusable in its bare form. Measured 2026-09-21 with NCBI esearch: (SET/TAF-I[tiab] OR TAF-Ibeta[tiab] OR I2PP2A[tiab] OR SET oncoprotein[tiab] OR SET protein[tiab]) AND (MLL[tiab] OR KMT2A[tiab] OR MLL-AF4[tiab] OR MLL-AF9[tiab] OR mixed lineage leukemia[tiab]) AND (localiz*[tiab] OR localis*[tiab] OR subcellular[tiab] OR nuclear body[tiab] OR nuclear bodies[tiab] OR speckle*[tiab] OR condensat*[tiab] OR nucleolar[tiab] OR immunofluorescence[tiab] OR compartment*[tiab]) returns 0; the disambiguated gene clause alone returns 288 and gene AND fusion returns 1. The bare token cannot be used: SET[ti] AND (TAF-I[ti] OR nuclear[ti] OR cytoplasmic[ti] OR granzyme[ti] OR translocation[ti]) returns 184 records whose first page is a Calceolariaceae bait set, a quantum-computing paper, a Nile tilapia ribosomal gene set and a Bignoniaceae probe set - so the name list, not the query mechanics, is the instrument here (brief rules 70 and 72). Positive control through the identical query shape with (MLLT1[tiab] OR ENL[tiab]) substituted returns 23. Zero in titles and abstracts; supporting figures not searched (brief rule 80). SET HAS been localised in the context of its OWN fusion, SET::NUP214, which is a different lesion in a different lineage.
epigenetic readoutAN INHIBITOR OF A WRITER - the readout is NEGATIVE, and that is the honest answer for this protein rather than a gap. SET / TAF-I beta (Q01105, 290 aa) is an H2A-H2B histone chaperone and, in the same molecule, a subunit of INHAT (INHibitor of Acetyl Transferases). The founding characterisation purified a multiprotein cellular complex that POTENTLY INHIBITS the histone acetyltransferase activity of p300/CBP and PCAF, mapped a novel acetyltransferase-inhibitory domain that BINDS HISTONES AND MASKS THEM FROM BEING ACETYLTRANSFERASE SUBSTRATES, and showed that endogenous INHAT subunits including the Set/TAF-I beta oncoprotein associate with chromatin in vivo and block coactivator-mediated transcription when transfected [E_q97_epi1]. So SET's epigenetic readout is the SUPPRESSION of H3/H4 acetylation by substrate masking - it writes nothing, erases nothing, and reads nothing in the reader-domain sense; it hides the substrate. THE COLLISION WITH THIS TRANCHE IS THE POINT, AND IT IS TESTABLE. gene-crebbp, amended in the same tranche, is the p300/CBP acetyltransferase whose H3K27ac output the MLL-r-BET complex reads at critical loci (PMID 38693103), and SET is a documented inhibitor of exactly that enzyme's nucleosomal activity. This entry independently establishes that the MLL N-terminus - the piece every KMT2A fusion retains - binds TAF-I beta through its acidic region, that GST-MLLN pulls down TAF-I beta together with histones, and that MLLN and TAF-I beta synergistically upregulate Hoxa9 [E1]. Put those together and the prediction is uncomfortable for the tidy account: the retained MLL N-terminus recruits a CBP/p300 INHIBITOR to the same promoters at which this graph argues CBP/p300-deposited H3K27ac is read. Nobody has measured H3K18ac or H3K27ac at HOXA in the presence or absence of SET in any MLL system, so whether the chaperone and the INHAT functions are separable at those loci is open. WHAT IS NOT KNOWN: no source read for this entry measures any histone mark - acetylation, methylation or ubiquitylation - at any locus in an MLL or KMT2A-fusion background with SET manipulated. The INHAT result is HeLa-scale biochemistry, not leukaemia, and it long predates the mark-specific vocabulary, so it reports bulk H3/H4 acetylation rather than named residues. SEARCH HAZARD, and it is in data/symbol-collisions.json under SET-CAN / CAN: SET is ordinary English and an unqualified symbol query is useless here. Every query for this entry used SET/TAF-Ibeta, SET oncoprotein or histone chaperone as the operand, never bare SET.
target genes
[
  "HOXA9",
  "HOXA10"
]
localization
{
  "subcellular": "SWITCHABLE, and that is the finding rather than a hedge. UniProt Q01105 places SET in the cytosol and on the endoplasmic reticulum - the SET complex is ER-associated - and has it move rapidly to the nucleoplasm after granzyme A cleavage; immunofluorescence in the SET::NUP214 work shows wild-type SET/TAF-Ibeta in the nucleus; SUMOylation at K68 drives cytoplasmic retention and K68R keeps it nuclear; and in FLT3-wild-type AML cells SET works at the ER and the plasma membrane on FLT3 trafficking. None of the three UniProt locations carries an evidence code, so the curated list is unattributed and the experimental papers are the load-bearing sources.",
  "subnuclear": "NUCLEOPLASM, EXPLICITLY EXCLUDING THE NUCLEOLUS. UniProt's note states that after granzyme A cleavage SET is found in the nucleoplasm 'avoiding the nucleolus'. That is a rare thing in this graph - a positively stated subnuclear EXCLUSION - and it matters, because the nucleolus is the one compartment in which a KMT2A fusion has actually been photographed. No nuclear body, speckle or condensate has ever been assigned to SET.",
  "how_shown": "immunofluorescence comparison of SET, NUP214 and the SET-NUP214 fusion with an NES-reporter export assay (PMID 15239126); SUMO-site mutagenesis with compartment readout in HEK-293, neurons and mice (PMID 30767764); immunofluorescence, co-immunoprecipitation and RNA immunoprecipitation for the FLT3/HuR/ER axis in AML cells (PMID 37190162); UniProt Q01105 SUBCELLULAR LOCATION comment and note.",
  "systems": "transfected cultured cells, HEK-293 and rodent neurons, and FLT3-wild-type versus FLT3-ITD AML lines; no KMT2A-rearranged system has ever been used to localise SET",
  "node": "localization-set-nup214-crm1-nuclear-bodies"
}
cell type
{
  "expressed_in": "Not resolved to a haematopoietic cell type in this entry. SET is broadly expressed; the compartment, not the level, is what varies.",
  "required_in": "Not established in KMT2A-rearranged leukaemia. The functional cell-type result here is a different one: in FLT3-WILD-TYPE AML cells SET is required for normal FLT3 delivery to the plasma membrane, and in FLT3-ITD cells that interaction is largely lost [E_q66_loc3]. So the cell type in which SET has a demonstrated job is the FLT3-wild-type myeloid blast, and this graph's interest in SET comes from a physical contact with the MLL N-terminus [E1] that has never been tested for requirement in any cell.",
  "expressed_in_lineage_note": "The SET::NUP214 literature this entry draws on is paediatric and adult T-lineage ALL [E2, E3]. The index disease is infant B-lineage KMT2A::AFF1, so every SET result reaching this graph carries a lineage difference, and in the SET::NUP214 case an age difference too.",
  "how_shown": "immunofluorescence, co-immunoprecipitation and RNA immunoprecipitation in FLT3-WT versus FLT3-ITD AML lines, with HuR inhibition and forced SET nuclear retention (PMID 37190162)",
  "nodes": [
    "cell-type-monoblast",
    "cell-type-early-t-precursor"
  ]
}
cell state
{
  "maintains_state": "not established. No experiment shows SET maintaining a cell identity in any leukaemia. Recorded as a bounded absence.",
  "regulates_state": "Receptor-availability state, in AML: SET nuclear retention reduces FLT3 at the plasma membrane, and midostaurin increases both membrane FLT3 and SET/FLT3 binding [E_q66_loc3]. That is a state of the cell's signalling input rather than of its identity, and it is the only SET result in this entry with a drug attached.",
  "state_dependent_localisation": "SET's OWN compartment is the state variable. K68 SUMOylation retains SET in the cytoplasm; granzyme A cleavage during cytotoxic attack drives it into the nucleoplasm [E_q66_loc2]. A cell under immune attack and a cell under SUMO stress hold SET in opposite compartments, and neither switch has been measured in a leukaemic blast.",
  "how_shown": "SUMO-site mutagenesis with compartment readout (PMID 30767764); membrane FLT3 quantification under HuR inhibition, forced SET nuclear retention and midostaurin (PMID 37190162)",
  "nodes": []
}
How this could bite us back
  • SET is an abundant, multifunctional nuclear protein - a PP2A inhibitor, an INHAT component and a granzyme-A DNase inhibitor as well as a histone chaperone - so any intervention hits at least four pathways and a phenotype could not be attributed to chaperoning.
  • Inhibiting SET's PP2A-inhibitory function is an established oncology idea pointing in the OPPOSITE therapeutic direction (restoring PP2A activity), so the two rationales for touching this protein disagree about what a good drug would do.
  • If SET does contribute to HOXA activation alongside menin, then it is also a ready-made resistance route: a cell under menin inhibition that retains a SET-dependent contribution to HOXA output does not need any mutation to keep part of its programme.
  • SET's compartment is switchable by at least three independent mechanisms - K68 SUMOylation, granzyme A cleavage and FLT3 genotype (PMIDs 30767764, 37190162, plus the UniProt Q01105 note). Any SET-directed agent, including the PP2A-reactivating chemotypes that name SET as their target, is acting on a protein whose location depends on the cell's immune exposure and its FLT3 status, and no published assay reports SET's compartment alongside a drug response.
  • Inhibiting SET's nuclear retention has a documented on-target cost in AML: SET nuclear retention REDUCES FLT3 at the plasma membrane (PMID 37190162). Pushing SET into the nucleus therefore changes receptor availability, which is an effect on the signalling input of the cell and not a neutral relocation.
  • The SET token is not usable as a search term, which is a delivery problem for the literature rather than for a drug: a title-scoped query on SET returns plant bait sets, quantum-computing papers and fish gene sets. Any automated evidence sweep on this gene will silently mix them in unless it uses SET/TAF-I, TAF-Ibeta or I2PP2A.
  • SET's epigenetic output is INHIBITORY, so an agent that removes SET would RAISE p300/CBP-dependent histone acetylation - the opposite direction to A-485, the p300/CBP inhibitor that gene-crebbp records as MLL-r-selective in this same tranche (PMID 38693103). A SET-directed and a CBP-directed agent are therefore predicted to antagonise each other on this mark, and nobody has tested the combination.
  • This entry already records that SET is a PP2A inhibitor, an INHAT component and a granzyme-A DNase inhibitor as well as a chaperone. Adding the acetylation arm makes the attribution problem worse, not better: a phenotype after SET loss could be phosphatase release, histone unmasking, chaperone failure or apoptotic sensitisation, and no current tool distinguishes them.
Still open
  1. Does a KMT2A fusion protein - KMT2A::AFF1 specifically - bind SET/TAF-I beta? Only the wild-type N-terminal fragment has been tested, in 2005, and the experiment is a co-immunoprecipitation.
  2. Does the SET contact surface on the KMT2A N-terminus overlap the menin-binding motifs MBM1/MBM2? If it does not, SET is a menin-inhibitor-independent route to HOXA activation and therefore a resistance hypothesis.
  3. SET-NUP214 recruits DOT1L to HOXA promoters. Does SET do so in a KMT2A-rearranged cell, and would that make SET a second, non-menin handle on the same DOT1L-dependent programme?
  4. Why has nobody followed the 2005 result in twenty years? Recording the absence of follow-up is not the same as recording a refutation, and this beat found no refutation.
  5. Where is SET in a KMT2A::AFF1 blast? This entry's founding result is that SET binds the retained MLL N-terminus and co-activates HOXA with it - a nuclear claim - while the curated default compartment for SET is the cytosol and the ER. The two have never been reconciled in one cell, and one fractionation in an index line would do it.
  6. Is the nucleolar EXCLUSION real in leukaemia? SET is annotated as avoiding the nucleolus, and the nucleolus is the one compartment where a KMT2A fusion has been photographed. If SET and the fusion occupy mutually exclusive subnuclear space, the SET-MLL contact has to happen somewhere else, and that location is unknown.
  7. Does K68 SUMOylation of SET occur in leukaemic blasts, and does it change under menin inhibition? A single switchable residue controlling the compartment of an MLL-binding chaperone is a cheap thing to measure and nobody has.
  8. Does SET inhibit CBP/p300 acetylation at HOXA in an MLL-fusion cell, or is the chaperone function the only one operating there? This graph now holds both halves - MLLN binds TAF-I beta and synergistically activates Hoxa9 (E1), and SET/INHAT inhibits p300/CBP (PMID 11163245) - and they point in opposite transcriptional directions at the same promoters.
  9. Can the chaperone and INHAT activities of SET be separated by mutation? The acidic region carries the MLL contact and the histone-masking domain is described as a distinct acetyltransferase-inhibitory domain; a separation-of-function allele would make the HOXA question answerable in one experiment.
  10. Is H3K27ac at HOXA altered when SET is depleted in an MLL-rearranged line? No acetylation measurement at any locus with SET manipulated exists in any MLL or KMT2A system.
Every source behind this entry (9)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The amino-terminal region of MLL (MLLN) interacts with TAF-I beta/SET, the acidic region of TAF-I beta is essential for that binding, GST-MLLN pulls down both TAF-I beta and histones, and MLLN and TAF-I beta synergistically upregulate transcription of Hoxa9 and co-immunoprecipitate in chromatin containing the Hoxa9 promoter region.

    Synergistic action of MLL, a TRX protein with template activating factor-I, a histone chaperone. · Shimoyama T, Kato K, Miyaji-Yamaguchi M, Nagata K · FEBS Letters 2005
    PMID 15670842DOI
    What this evidence cannot show

    Abstract read from efetch 2026-09-20; full text not read (no PMCID in the record). **A short report from 2005 using yeast two-hybrid, GST pull-down and a transcription assay.** The authors' own closing sentence is 'these results suggest', and this entry grades it moderate for that reason. The MLL fragment is the wild-type N-terminal region, **not a KMT2A fusion protein**; the transcription readout is Hoxa9 promoter activity; and nothing here is a measurement in a leukaemic cell. Twenty years old with, on this beat's searching, no direct follow-up in a KMT2A-rearranged system.

    Evidence that argues against this (1)
    • SET/TAF-I beta is an abundant acidic nuclear protein that binds histones and many basic proteins non-specifically, and the interaction is mapped to its acidic region - the part most likely to produce charge-driven artefacts in a pull-down. A 2005 two-hybrid-plus-GST result on an acidic domain is exactly the evidence class this project treats as a reason to look rather than a reason to conclude. Cigdem S et al. 2021
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    A cryptic recurrent deletion del(9)(q34.11q34.13) produces a conserved SET-NUP214 fusion in paediatric T-cell acute lymphoblastic leukaemia and in the LOUCY cell line; SET-NUP214 binds in the promoter regions of specific HOXA genes where it interacts with CRM1 and DOT1L, which may transcriptionally activate specific HOXA cluster members; siRNA against SET-NUP214 abolished HOXA expression, inhibited proliferation and induced differentiation in LOUCY but not in other T-ALL lines. The HOXA-high T-ALL subgroup also includes MLL-rearranged and CALM-AF10 patients.

    PMID 18299449DOI
    What this evidence cannot show

    Abstract read from efetch 2026-09-20; full text not read (PMC2343598 exists). Established within the reported cohort: 3 of 5 HOXA-high patients among 92 T-ALL samples, plus one cell line. **The functional work is one cell line, LOUCY, and the authors note the siRNA effect was not seen in other T-ALL lines.** The DOT1L interaction is stated as part of a promoter-bound complex and the causal language in the abstract is 'may transcriptionally activate'. T-ALL, not B-ALL, and not a KMT2A fusion.

    Evidence that argues against this (1)
    • The HOXA-high grouping that unites SET-NUP214 with MLL-rearranged and CALM-AF10 T-ALL is an expression-cluster grouping, and convergence on a transcriptional output is weak evidence for a shared molecular mechanism - this graph has repeatedly found that phenotypic convergence between fusion families hides different machinery. Song J et al. 2023
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    SET-NUP214 interacts with MLL via the SET acidic region of SET-NUP214, and SET-NUP214 and MLL cooperatively enhance the promoter activity of the HOXA10 gene; neither the SET region alone nor the NUP214 region alone sufficiently enhanced the HOXA10 promoter.

    SET-NUP214 and MLL cooperatively regulate the promoter activity of the HoxA10 gene. · Cigdem S, Saito S, Nishikata D, et al. · Genes to Cells 2021
    PMID 34320268DOI
    What this evidence cannot show

    Abstract read from efetch 2026-09-20; full text not read (no PMCID). **Promoter-reporter activity in transfected cells, not endogenous HOXA10 transcription**, and the interaction mapping uses fragments. Graded moderate: the domain mapping is informative, the transcriptional claim is a reporter result. Same laboratory lineage as E1 (Nagata group), so E1 and E3 are not independent.

    Evidence that argues against this (1)
    • A reporter assay measures what a promoter fragment does in a plasmid, not what a chromatinised locus does in a blast. The 2008 Blood study found the SET-NUP214 dependency of HOXA expression in exactly one of several T-ALL lines, so even the endogenous effect is context-restricted. Van Vlierberghe P et al. 2008
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    SET is UniProt Q01105 (SET_HUMAN), protein SET, 290 amino acids, also named template-activating factor I (TAF-I), phosphatase 2A inhibitor I2PP2A, PHAPII, HLA-DR-associated protein II and inhibitor of granzyme A-activated DNase; the reviewed record lists no alias gene symbols.

    What this evidence cannot show

    Fetched 2026-09-20 with `gene_primary` requested and read back per rule 62: exactly one reviewed human record with primary gene name SET. **The symbol `SET` is the worst retrieval token in this beat** - it collides with the SET methyltransferase domain (which KMT2A itself carries), with SETD/SETDB gene names, with the English word, and with 'position sets' in mathematics. A phrase query built on it returned 1,421 records of which the first ten titles were chemistry, mathematics and botany. Every count in this entry was title-checked and the protein was searched under TAF-I and SET-NUP214 rather than under the bare symbol.

    Evidence that argues against this (1)
    • The same symbol collision means this entry's own literature searching is likely incomplete in the opposite direction: a precision-driven search on 'TAF-I' and 'SET-NUP214' will miss papers that use the bare symbol correctly. Recall and precision fail in opposite directions here and both produce a confident wrong number. US National Library of Medicine 2026
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    As of 2026-09-20, the generic-phrase query '("histone chaperone"[tiab] OR "histone chaperones"[tiab]) AND (MLL OR KMT2A)' returns 2 records, of which one is the MLL/TAF-I beta paper; the equivalent symbol queries for HIRA, ASF1A/ASF1B and SUPT6H against MLL or KMT2A each return 0, and '"chromatin assembly factor 1"[tiab] AND (MLL OR KMT2A)' returns 0.

    What this evidence cannot show

    All query translations captured; none contained a [UID] term. All returned titles fetched in full. **This is the finding, not an incidental number:** a beat enumerated from a gene list would have concluded that no histone chaperone has ever been connected to MLL, and the generic phrase found the one that has. The project already logged this failure for NSL/KANSL and MED23; this is a third instance.

    Evidence that argues against this (1)
    • The phrase route has its own blind spot: a paper describing a chaperone function without using the words 'histone chaperone' - which is most of the older literature, including papers that say 'nucleosome assembly protein' or 'template activating factor' - is invisible to it. Neither route is complete and the honest conclusion is that this layer has never been systematically searched in this disease. Shimoyama T et al. 2005
  6. E_q66_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Wild-type SET/TAF-Ibeta is NUCLEAR and its own fusion protein keeps that compartment while wrecking the export machinery: immunofluorescence shows SET/TAF-Ibeta in the nucleus and CAN/Nup214 at the nuclear envelope, whereas the majority of SET/TAF-Ibeta-CAN/Nup214 is localised IN THE NUCLEUS; the fusion interacts with hCRM1 and causes aberrant intracellular localisation of hCRM1, and in cells expressing the fusion a nuclear-export-signal-bearing protein accumulates in the nucleus, an export block partially rescued by overexpressing hCRM1.

    PMID 15239126DOI
    What this evidence cannot show

    Transfected cultured cells with overexpressed constructs, not primary leukaemia, and SET::NUP214 is an adult T-lineage lesion rather than the index infant B-lineage KMT2A::AFF1 - two lineage steps away from this graph's disease. The wild-type SET compartment is established here by immunofluorescence in the same overexpression setting, so it is a comparison within one experiment rather than an independent measurement of endogenous SET. Read: abstract only. Identifiers copied field by field from the efetch record fetched 2026-09-21, ArticleId anchored per brief rule 81.

    Evidence that argues against this (1)
    • UniProt's curated reading of SET's compartment is the opposite way round: it places SET in the cytosol and on the endoplasmic reticulum by default - the SET complex is associated with the ER - and has SET move rapidly TO the nucleus only after granzyme A cleavage during cytotoxic-lymphocyte attack. On that reading nuclear SET is the induced state, not the resting one, and none of the three locations carries an evidence code, so the annotation is itself unattributed. UniProt Knowledgebase, reviewed human record, fetched 2026-09-21 2026
  7. E_q66_loc2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    SET's nucleus-to-cytoplasm partition is switched by a post-translational mark: SET is SUMOylated predominantly at K68, that modification induces its cytoplasmic RETENTION, and blocking it by K68R keeps SET nuclear - in HEK-293 cells the SUMOylated cytoplasmic form inhibits PP2A and drives tau hyperphosphorylation, and K68R rescues the phenotype in vivo.

    SET SUMOylation promotes its cytoplasmic retention and induces tau pathology and cognitive impairments. · Qin M; Li H; Bao J; Xia Y; Ke D; Wang Q; Liu R; Wang JZ; Zhang B; Shu X; Wang X · Acta Neuropathol Commun 2019
    PMID 30767764DOI
    What this evidence cannot show

    HEK-293 cells, rat primary hippocampal neurons and mice: a neurodegeneration system with no haematopoietic or leukaemic arm at all, and the downstream readout is tau, which is irrelevant here. What transfers is only the mechanism - a single SUMO site partitions SET between compartments - and whether K68 SUMOylation operates in a blast is entirely untested. Read: abstract only.

    Evidence that argues against this (1)
    • A second, independent switch moves SET the other way and is proteolytic rather than post-translational: UniProt records that following cytotoxic-lymphocyte attack and cleavage by granzyme A, SET moves rapidly to the nucleus, where it occupies the nucleoplasm while AVOIDING the nucleolus. Two opposing switches means SET's compartment is a state variable rather than a property, and no study measures either switch in leukaemia. UniProt Knowledgebase, reviewed human record, fetched 2026-09-21 2026
  8. E_q66_loc3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    SET has a job at the endoplasmic reticulum and the plasma membrane in AML, and an oncogenic mutation breaks it: in FLT3-wild-type AML cells SET and FLT3 co-localise and interact, the interaction occurs BEFORE FLT3 glycosylation and depends on HuR binding the FLT3 3'UTR, SET nuclear retention reduces FLT3 at the membrane, and in FLT3-ITD cells SET barely binds FLT3, contributing to its retention in the endoplasmic reticulum.

    Set Protein Is Involved in FLT3 Membrane Trafficking. · Marcotegui N; Romero-Murillo S; Marco-Sanz J; Peris I; Berrozpe BS; Vicente C; Odero MD; Arriazu E · Cancers (Basel) 2023
    PMID 37190162DOI
    What this evidence cannot show

    AML cell lines, so leukaemia but myeloid and FLT3-driven rather than KMT2A-rearranged; the mechanism is a 3'UTR-scaffold model tested by immunofluorescence, immunoprecipitation and RNA immunoprecipitation, and the causal direction rests on perturbations of HuR and of SET's nuclear retention. It is a co-localisation-plus-interaction result, not a quantitative trafficking measurement. Read: abstract only.

    Evidence that argues against this (1)
    • This makes SET's compartment a consequence of the cell's genotype rather than a fixed address, and in the direction that undercuts a simple reading: in FLT3-ITD cells SET hardly binds FLT3 at all, so the same protein has a membrane-trafficking role in one AML genotype and not in another. Any entry recording 'SET is nuclear' or 'SET is cytoplasmic' as a fact is recording one genotype's answer. The nuclear-versus-cytoplasmic partition is itself switchable by a single SUMO site. Qin M; Li H; Bao J; Xia Y; Ke D; Wang Q; Liu R; Wang JZ; Zhang B; Shu X; Wang X 2019
  9. E_q97_epi1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    A purified multiprotein cellular complex, INHAT, potently inhibits the histone acetyltransferase activity of p300/CBP and PCAF; its acetyltransferase-inhibitory domain binds histones and masks them from being acetyltransferase substrates; and endogenous INHAT subunits, which include the Set/TAF-I beta oncoprotein, associate with chromatin in vivo and can block coactivator-mediated transcription.

    PMID 11163245
    What this evidence cannot show

    Abstract fetched and read in full 2026-09-21; full text not read. Human cell biochemistry and transfection, no haematopoietic system and no leukaemia (rule 73), and no KMT2A or MLL arm. The inhibition is reported as HAT activity on histones in bulk; the paper predates residue-resolved mark nomenclature, so it does NOT name H3K18ac or H3K27ac and this entry does not claim it does. SET appears as one named subunit of INHAT, so subunit-level attribution of the masking activity to SET alone is not established by this abstract (the same rule-64 caution that applies to gene-bmi1 in this tranche).

    Evidence that argues against this (1)
    • This entry's own E1 counter-evidence already warns that SET/TAF-I beta is an abundant acidic nuclear protein that binds histones and many basic proteins non-specifically, and that the MLL interaction maps to its acidic region - the part most likely to produce charge-driven artefacts. The SAME objection applies with full force to histone masking: a highly acidic protein that coats basic histone tails will inhibit a HAT assay in vitro whether or not it does so at a locus in a cell. The in vivo chromatin association and the transfection block are what raise this above an artefact, and neither is a locus-resolved measurement. Cell 2001
Connected to