Oncofusion

start / graph / gene-nr1h3

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

NR1H3 / LXR-alpha - the best-observed of the three MEN1 sterol partners, with both directions curated in two species, and menin as its corepressor

also called LXRA · LXR-alpha · Liver X receptor alpha · NR1H3 · Oxysterols receptor LXR-alpha

LXR-alpha is a sensor that detects cholesterol and turns on the genes that build fat. Menin - the protein every menin inhibitor is aimed at - binds it and shuts it off.

The same thing, for a scientist

NR1H3/LXR-alpha (Q13133, 447 aa) is an oxysterol-activated nuclear receptor that drives lipogenic and cholesterol-efflux transcription. Menin binds it and represses it [E2]. The IntAct record is **four non-spoke physical-association rows plus one spoke row**, in **two species**, with **both directions curated in mouse** (Fig 4c menin as bait, Fig 4d LXR-alpha as bait) [E1] - which makes it the second-best-observed pair in this beat's twelve after PSIP1-RBBP8, and not the single weak row the dispatch brief recorded. The same figure set puts NCOA1/SRC-1 and EP300 on LXR-alpha, which is the coactivator competition a corepressor model requires [E1]. The direction replicates independently in colorectal tissue with a genetic arm [E3]. The mechanism is written once, in `mechanism-menin-represses-lxr-driven-lipogenesis`; this entry carries the protein.

Go deeper
Why the reciprocal mouse pair matters more than the record count

Most of this interactome is one direction of one assay. Here the curator recorded menin-as-bait and LXR-alpha-as-bait from two different panels of the same paper 1. A reciprocal co-immunoprecipitation is the standard internal control for an antibody artefact, and it is present in only a small minority of the pairs this graph has written.

The cross-species arm adds the second control: the same pair in human HepG2 and in mouse. Neither is a leukaemia system, so this is orthogonality of model, not of disease.

What LXR-alpha is, and why a leukaemia graph should care beyond the interaction

LXR-alpha heterodimerises with RXR and activates ABCA1, ABCG1, SREBF1/SREBP-1c, FASN and SCD-1. It is one of the two oxysterol sensors that couple intracellular sterol load to transcription, and synthetic LXR agonists are an active therapeutic class.

The reason to hold this node is not the interaction, it is the direction. If menin represses LXR-alpha, a menin inhibitor is an LXR de-repressor, and that makes a licensed drug class in paediatric leukaemia a partial LXR agonist in every tissue that expresses menin. Nobody has measured it - see hyp-menin-inhibition-derepresses-lxr-target-genes-in-patients.

Facts and figures
uniprotQ13133
length AA447
gene symbolNR1H3
protein nameOxysterols receptor LXR-alpha (Liver X receptor alpha, Nuclear receptor subfamily 1 group H member 3)
uniprot resolutionaccession:Q13133, reviewed human, one record; primary gene name read back as NR1H3 (entry NR1H3_HUMAN), synonym LXRA; length 447 aa checked against UniProt 2026-09-21.
symbol collision checkdata/symbol-collisions.json greped 2026-09-21 for NR1H3, NR1H2 and LXR: no rows existed. A row IS proposed in this beat's memo, because LXR is a two- or three-letter token that resolves to two distinct genes and most of the literature names the pathway rather than the gene.
baitMEN1 (O00255); mouse Men1 (O88559)
heterodimer partnerRXR
shown for fusionNO - nobody has looked in any KMT2A fusion, in any lineage.
fetched2026-09-21
subnuclear localizationUNPUBLISHED for LXR-alpha in any cell type. The one nucleolar result in this receptor family (PMID 19356704) is about LXR-BETA / NR1H2 and must not be attached here.
localization shown for fusionNO - nobody has localised LXR-alpha in any KMT2A-rearranged cell, in any lineage; the bounded query returned 0. Kept separate from this entry's existing shown_for_fusion, which concerns the MEN1-NR1H3 interaction.
interaction with men1
{
  "source_pmid": "25962847",
  "source_class": "focused mechanistic study",
  "n_intact_records": 5,
  "n_spoke_expanded": 1,
  "n_source_publications": 1,
  "reciprocal_direction_curated": true,
  "species": [
    "Homo sapiens (HepG2)",
    "Mus musculus"
  ],
  "figures": {
    "Fig 4b": "human HepG2, menin bait",
    "Fig 4c": "mouse, menin bait",
    "Fig 4d": "mouse, LXR-alpha bait (reciprocal)",
    "SF5A": "human, spoke-expanded with NCOA1 and EP300"
  },
  "coactivators_competed_with": [
    "NCOA1 / SRC-1",
    "EP300"
  ],
  "brief_stated_record_count": 1,
  "rest_record_count": 5,
  "shared_pdb_entries": []
}
target genes
[
  "SREBF1 / SREBP-1c",
  "FASN",
  "SCD-1",
  "ABCA1",
  "ABCG1"
]
localization
{
  "subcellular": "Nucleus, with a cytoplasmic call that must be read carefully. UniProt Q13133: Nucleus (ECO:0000255 PROSITE-ProRule PRU00407 - a SEQUENCE RULE, plus ECO:0000269 PubMed:25661920) and Cytoplasm on ECO:0000250 BY SIMILARITY TO MOUSE Nr1h3 (Q9Z0Y9). HPA immunofluorescence: main Nucleoplasm, Cytosol additional, reliability Approved.",
  "EVIDENCE_GRADE_WARNING": "Two of the three UniProt supports here are NOT human experiments on this protein. ECO:0000255 PRU00407 is the nuclear-receptor DNA-binding-domain profile rule - a prediction from sequence. ECO:0000250 Q9Z0Y9 is inference from the MOUSE orthologue. The brief's standing instruction is to record a bounded absence rather than infer from a homolog; here UniProt has done the inferring, and the entry records that the cytoplasmic pool of HUMAN LXR-alpha rests on mouse data. HPA's independent Nucleoplasm-plus-Cytosol call is the only human measurement supporting both compartments, and it is at HPA's weaker 'Approved' tier.",
  "subnuclear": "UNPUBLISHED. No subnuclear compartment has been reported for LXR-alpha in any cell type. The bounded subnuclear query returned 3 records, none of which is an NR1H3 subnuclear study.",
  "THE_NEAR_MISS_THAT_MUST_NOT_BE_MISFILED": "PMID 19356704 is titled 'Liver X receptor BETA with mutations in the activation function-2 region is excluded from the NUCLEOLUS'. It is a genuine subnuclear result about a liver X receptor - and it is about NR1H2 (LXR-beta, P55055, 460 aa), NOT NR1H3 (LXR-alpha, Q13133, 447 aa). The collision registry's LXR row exists for exactly this: thirteen residues apart, the literature says 'LXR' for both, and the menin corepressor result this entry is built on is specifically LXR-ALPHA. This record is logged so that the next reader does not re-find it and attach a nucleolar-exclusion result to the wrong receptor.",
  "how_shown": "PROSITE profile rule (sequence); mouse orthologue inference; one human experimental citation (PMID 25661920); HPA antibody immunofluorescence",
  "rule_91_note": "PMID 25661920, the single human ECO:0000269 citation, is titled 'CCAR2 negatively regulates nuclear receptor LXR-alpha by competing with SIRT1 deacetylase' - an acetylation-regulation paper. Any localisation data in it is incidental to its subject.",
  "tissue_of_evidence": "liver, macrophage and transfected-cell systems; HPA's non-haematopoietic immunofluorescence panel",
  "node": "localization-partner-baseline-compartment-is-often-miscalled"
}
cell type
{
  "haematopoietic_answer": "MONOCYTES AND MACROPHAGES. HPA Blood Atlas classes NR1H3 as Immune cell enhanced, Detected in many, with intermediate monocyte at 13.3 nTPM the enriched population; RNA tissue-enhanced in liver at 84.3 nTPM. LXR-alpha's canonical function - the oxysterol sensor driving cholesterol efflux through ABCA1/ABCG1 - is a macrophage function, and the bounded haematopoietic query returns a macrophage-dominated literature.",
  "grade": "Expression atlas plus a very large macrophage functional literature that was not enumerated here. No dedicated LXR-alpha experiment in a leukaemic cell was found.",
  "why_this_matters_here": "This entry exists because menin is a COREPRESSOR of LXR-alpha (PMID 25962847). The cell type in which LXR-alpha actually works is the monocyte/macrophage - so the on-target off-tumour consequence of a menin inhibitor for this axis would land in the myeloid compartment and in lipid handling, not in the leukaemic blast. Nobody has measured it."
}
cell state
{
  "state": "STEROL-LOADED. LXR-alpha is ligand-activated by oxysterols, so its active state is a lipid state of the cell rather than a differentiation state. See cell-state-lipid-pufa-dependence for the adjacent metabolic entry in this graph.",
  "bounded": "No NR1H3 cell-state result in any leukaemia or KMT2A-fusion system."
}
localization bounded absence
{
  "date": "2026-09-21",
  "tool": "NCBI E-utilities esearch, db=pubmed",
  "shape": "gene clause AND localisation clause AND KMT2A/MLL-fusion clause, one [tiab] tag per term (brief rule 94), every cytogenetic term quoted (brief rule 95)",
  "querytranslation_fusion_clause": "\"KMT2A\"[Title/Abstract] OR \"MLL\"[Title/Abstract] OR \"MLL1\"[Title/Abstract] OR \"MLL-AF4\"[Title/Abstract] OR \"MLL-AF9\"[Title/Abstract] OR \"MLL-ENL\"[Title/Abstract] OR \"KMT2A-rearranged\"[Title/Abstract] OR \"MLL-rearranged\"[Title/Abstract] OR \"t 4 11\"[Title/Abstract] OR \"t 9 11\"[Title/Abstract] OR \"MLL fusion\"[Title/Abstract]",
  "count": 0,
  "controls": {
    "fusion_clause_alone": 8743,
    "gene_clause_alone": 4765
  },
  "reading": "Counted in TITLES AND ABSTRACTS only; supporting figures were not searched (brief rule 80)."
}
subnuclear localization CORRECTED 2026 09 21
{
  "was": "UNPUBLISHED for LXR-alpha in any cell type. The one nucleolar result in this receptor family (PMID 19356704) is about LXR-BETA / NR1H2 and must not be attached here.",
  "correction": "Recorded as UNPUBLISHED in any cell type. That is wrong: PMID 19356704 (Cell Biol Int 2009, 'Liver X receptor beta with mutations in the activation function-2 region is excluded from the nucleolus', verified live by me) imaged CFP-LXR-alpha as the COMPARATOR arm and reports significantly LESS nucleolar signal than LXR-beta. That is a low measurement, not an absence.",
  "what_stays_right": "The refusal to transfer the LXR-beta AF-2 mutant result to LXR-alpha was correct and is untouched. Only the 'never measured' framing changes.",
  "found_by": "Q89, while writing gene-nr1h2; verified and written by the king"
}
How this could bite us back
  • LXR agonists are themselves a therapeutic class and a menin inhibitor is predicted here to act as a partial one. In a paediatric marrow, LXR activation touches cholesterol efflux, innate immunity and lipogenesis - none of it measured in any menin-inhibitor trial.
  • Four records look like independent support and are one paper. The reciprocal mouse pair is a good internal control and is not independent replication of anything.
  • Grading this partner on record count alone would have put it fourth in the beat; grading it on assay quality puts it second. Neither ranking is a measurement of biological importance in this disease, where it is zero.
Still open
  1. Is menin a corepressor of LXR-alpha in a haematopoietic cell? Every measurement is hepatic or intestinal, and LXR agonists are under separate study in AML, so the sign in a blast cannot be predicted from the liver.
  2. Does the menin-LXR-alpha contact use the MBM cavity that revumenib occupies? That single question decides whether a clinical menin inhibitor displaces LXR-alpha directly.
  3. Does menin compete with NCOA1/SRC-1 and EP300 on LXR-alpha quantitatively, or is the coactivator co-precipitation incidental? The spoke-expanded rows cannot distinguish a ternary complex from mutual exclusion.
  4. LXR-beta is excluded from the nucleolus when its AF-2 is mutated (PMID 19356704) and nobody has asked the same question of LXR-alpha, the paralogue menin actually corepresses. Does LXR-alpha have a nucleolar pool, and does menin binding change it?
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct's REST record holds five MEN1-NR1H3 rows from PMID 25962847: human HepG2 non-spoke (Fig 4b), mouse non-spoke with menin as bait (Fig 4c), mouse non-spoke with LXR-alpha as bait (Fig 4d), a human spoke-expanded row (SF5A), and alongside them spoke rows placing NCOA1 and EP300 on LXR-alpha.

    What this evidence cannot show

    One database on one day, REST endpoint (brief rule 93). Five rows from ONE publication are five curated experiments. The dispatch brief for this beat recorded ONE record for NR1H3; the discrepancy is in the counting, not the data. The mouse rows are curated against the mouse accessions O88559 (Men1) and Q9Z0Y9 (Nr1h3), so a query restricted to the human accession sees fewer rows than exist.

    Evidence that argues against this (2)
    • Five is a property of the endpoint, not of the literature, and the endpoint penalty is not uniform across the partners this beat compares. Measured on the same day: IntAct REST returns 132 records for Q13133 where PSICQUIC id:Q13133 returns 40 (30%), while for Q13309/SKP2 the same comparison is 365 against 300 (82%) and for O60477/BRINP1 it is 35 against 9 (26%). Any ranking of partners by record count - and this entry's summary does rank this pair 'second-best-observed in this beat's twelve' - is partly a ranking of how well each accession happens to be reachable by the query form. EMBL-EBI IntAct and PSICQUIC 2026
    • One of the five is not the row this claim describes. Re-enumerated from the endpoint on 2026-09-21, the MEN1-NR1H3 rows are EBI-11362117 (human O00255|Q13133, f4b), EBI-11362135 and EBI-11362145 (mouse Q9Z0Y9|O88559, f4c and f4d), the mouse spoke row EBI-11364871 (Q9Z0Y9|O88559, sf5a) and EBI-11364853 (Q13133|O88559, F5A) - which is HUMAN NR1H3 against MOUSE Men1, taxa 9606/10090, not 'a human spoke-expanded row'. A cross-species row is a curator's reconstruction of a blot run in one system, and the two spoke rows are different records rather than one. IntAct MI scores across the five are 0.35-0.5. EMBL-EBI IntAct 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In HepG2 hepatoma cells and mouse liver (no haematopoietic system), menin physically interacts with LXR-alpha, inhibits its transcriptional activity, and reduces its target genes SREBP-1c, FASN and SCD-1; the authors conclude menin is a novel corepressor of LXR-alpha.

    PMID 25962847DOI
    What this evidence cannot show

    Abstract read this session; full text is Elsevier and route 5 (DOI resolution) returns HTTP 403, so the blots themselves were not seen and the figure attributions come from IntAct's curator annotations. HepG2 hepatocellular carcinoma cells and mouse liver - no haematopoietic system, no KMT2A fusion, no leukaemia of any kind (brief rule 73). Transient transfection and overexpression throughout.

    Evidence that argues against this (1)
    • The same group reports menin as a co-activator of PPAR-alpha in the same tissue, so menin's sign on lipid transcription is partner-dependent and a single-receptor reading overstates the net direction. Cheng P et al. 2011
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Independently, menin inhibition upregulated the LXR target genes ABCG1 and ABCA1 and reduced cellular cholesterol in colorectal cancer lines and in Men1-floxed;Vil1-Cre mouse colonic epithelium.

    PMID 37627154DOI
    What this evidence cannot show

    Abstract only; colorectal lines and mouse colonic epithelium, no haematopoietic system (brief rule 73). Replicates the direction of regulation, not the physical interaction - the evidence is RNA-Seq and a cholesterol assay.

    Evidence that argues against this (2)
    • A menin inhibitor can produce exactly this readout without menin. Phillips et al. 2019 show that MI-2 kills glioma cells largely independently of its ability to target menin: its direct molecular target is lanosterol synthase, and the consequence is suppressed cholesterol synthesis, generation of the endogenous LXR ligand 24,25-epoxycholesterol, and cholesterol depletion. 'Menin inhibition upregulated ABCG1 and ABCA1 and reduced cellular cholesterol' is the phenotype an MI-series lanosterol-synthase inhibitor produces by off-target chemistry. The Nyul abstract does not name the compound, so the pharmacological arm of this replication cannot currently be attributed to menin at all; only the Men1-floxed;Vil1-Cre genetic arm can. Phillips RE et al. 2019
    • The gut result may not be an LXR-alpha result. Cao et al. 2026 place the intestinal menin-LXR contact on LXR-BETA, with HDAC1 recruitment and Ces1g as the target gene, and report that MI-463 recapitulates intestinal-epithelium-specific Men1 deletion. LXR-alpha and LXR-beta share ligands, RXR and most target genes including ABCA1 and ABCG1, so an ABCG1/ABCA1 readout in colonic epithelium does not identify which paralogue was de-repressed, and this entry's use of the colorectal work as independent support for the LXR-ALPHA arm assumes an assignment the readout cannot make. Cao X et al. 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    UniProt Q13133 supports LXR-alpha's nuclear location with a PROSITE sequence rule (ECO:0000255 PRU00407) plus one human experimental citation, and supports its cytoplasmic location ONLY by similarity to the mouse orthologue Q9Z0Y9 (ECO:0000250) - so neither compartment rests on a dedicated human localisation experiment.

    What this evidence cannot show

    This is a deliberately negative evidence item about evidence grade. ECO:0000255 is prediction and ECO:0000250 is homology inference; the brief's rule is to record a bounded absence rather than infer from a homolog, and a reader who sees only the word 'Cytoplasm' in the record would not know it came from a mouse. The one human coded citation (PMID 25661920) is an acetylation paper, not a localisation study (rule 91).

    Evidence that argues against this (1)
    • The evidence-grade criticism is overstated as written. HPA's independent HUMAN antibody immunofluorescence reports both Nucleoplasm and Cytosol, so the cytoplasmic pool is not solely a mouse inference even though UniProt's citation for it is ECO:0000250. The defensible claim is narrower than the one this item makes: UniProt's RECORD lacks a human experimental citation for the cytoplasmic compartment, while a human measurement of that compartment does exist outside UniProt. Human Protein Atlas 2026
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Human Protein Atlas v25.1 assigns NR1H3 main location Nucleoplasm with Cytosol additional at reliability Approved, and classes it Immune cell enhanced with intermediate monocyte at 13.3 nTPM and liver-enhanced tissue expression at 84.3 nTPM.

    What this evidence cannot show

    Reliability Approved is HPA's weaker tier. This is nonetheless the only HUMAN measurement supporting both a nuclear and a cytoplasmic pool for LXR-alpha, since UniProt's cytoplasmic call is mouse inference. Blood Atlas values are RNA.

    Evidence that argues against this (1)
    • Agreement between two weak sources is not strong evidence, and both of the sources for LXR-alpha's localisation are weak in different ways. HPA's call sits at 'Approved', its lower reliability tier, and UniProt's only human coded citation is an acetylation-regulation study about CCAR2 competing with SIRT1 - a paper whose subject is not where the receptor is. Two non-dedicated measurements pointing the same way is the minimum this field should accept, not a reason to raise confidence. J Steroid Biochem Mol Biol 2015
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The one nucleolar observation in the liver X receptor family - that activation-function-2 mutants are excluded from the nucleolus - is about LXR-BETA and not LXR-alpha, and must not be transferred to NR1H3.

    PMID 19356704
    What this evidence cannot show

    Recorded here as a NEGATIVE control on misattribution, not as evidence about NR1H3. LXR-beta is NR1H2/P55055 (460 aa) and LXR-alpha is NR1H3/Q13133 (447 aa) - thirteen residues apart, inside the length-check limit, and the literature calls both 'LXR' (collision registry). Read at title level.

    Evidence that argues against this (1)
    • Refusing the LXR-beta result may be too strict, and the counter-argument is worth keeping alive. The activation-function-2 region is conserved between the two paralogues, so a nucleolar-exclusion mechanism operating through AF-2 is a legitimate HYPOTHESIS for LXR-alpha rather than an irrelevance - the reason not to write it down is that the experiment was done on the other receptor, not that the biology is known to differ. The correct handling is the one this entry adopts for the record and should also adopt in reasoning: unresolved, and cheap to resolve. UniProt Consortium 2026
Connected to