Oncofusion

start / graph / gene-foxm1

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

FOXM1 - a one-panel MEN1 co-IP from a 4,119-row atlas, attached to a protein that is independently a validated MLL-rearranged leukaemia stem-cell dependency

also called FOXM1 · Forkhead box protein M1 · FKHL16 · HFH11 · MPP2 · WIN · Trident · FOXM1B · FOXM1C

FOXM1 keeps leukaemia stem cells alive and quiet, and blocking it kills them - that part is solid. The claim that it physically touches menin rests on a single blot in a supplementary figure of a study about something else.

The same thing, for a scientist

FOXM1 (Q08050, 763 aa) is a proliferation-associated forkhead transcription factor and, in this disease, a **validated dependency**: Foxm1 is required for survival, quiescence and self-renewal of MLL-AF9-transformed leukaemia stem cells in vivo, acts by binding and stabilising beta-catenin, and its inhibition suppresses leukaemogenic potential and induces apoptosis of primary LSCs from MLL-rearranged AML patients [E2]. That paper is already cited three times in this graph. **What is new and weak is the interaction.** IntAct records **one MEN1-FOXM1 row** from Li et al. 2015, a 4,119-row TAP/MS atlas of 56 transcription factors, and it is one of only 53 non-spoke rows in that whole publication, annotated `figure legend (sf3a)` with intact-miscore 0.4 [E1]. **The string MEN1 and the word menin appear nowhere in the paper's 84,072 characters of main text** [E3], and `MEN1[tiab] AND (FOXM1 OR "forkhead box M1")[tiab]` returns **0** on PubMed [E4]. So the interaction is a validation panel nobody discussed and nobody has repeated, attached to a protein that matters here for a completely different reason.

Go deeper
Two claims of very different strength, kept apart
claimevidencegrade
FOXM1 is required for MLL-AF9 LSC survival, quiescence and self-renewal, acting through beta-cateninin vivo mouse MLL-AF9, plus primary human MLL-rearranged AML LSCs in vitro and in xenograft 2strong
FOXM1 physically associates with meninone non-spoke pull-down row, supplementary figure 3a of a TF atlas, HEK293T, miscore 0.4, never discussed, never repeated 13weak

These must not be merged. A reader who sees 'FOXM1 is a menin partner and a menin-inhibitor-relevant LSC dependency' has been handed a mechanism that nobody has evidence for. The dependency is real and does not run through menin as far as anyone has shown.

The beta-catenin route, which is where this node actually connects

Foxm1 upregulation activates Wnt/beta-catenin by directly binding beta-catenin and stabilising it against degradation, preserving LSC quiescence and promoting self-renewal in MLL-rearranged AML 2. This graph holds gene-ctnnb1, and the same beat that wrote this entry also found that DOT1L and MLLT10 are beta-catenin/TCF7L2 coactivators in intestinal crypts (ana-the-dot1l-tcf4-partner-is-tcf7l2-not-tcf4).

So beta-catenin is now reachable from two independent directions in the KMT2A interactome - through FOXM1 in LSCs and through DOT1L/AF10 in intestine - and neither route has been tested against the other. That is a connection this beat did not go looking for.

What the atlas row can and cannot support

Li et al. 2015 purified 56 transcription factors plus 37 FOX-family members by tandem affinity purification, ran 214 purifications and reported 2,156 high-confidence interactions 3. Of the 4,119 IntAct rows from it, 4,065 are spoke-expanded associations and 53 are non-spoke physical associations - the small validated set 1. MEN1-FOXM1 is in the small set, which is the strongest thing that can be said for it.

Against that: MEN1 is not a transcription factor and is a surprising bait in a TF atlas. IntAct records MEN1 twice in the whole publication - once as a FOXB1 prey (spoke) and once as the FOXM1 bait (non-spoke). Whether MEN1 was a genuine bait cell line or whether the curator assigned the roles from a figure could not be resolved, because menin is not mentioned in the main text at all.

Facts and figures
uniprotQ08050
length AA763
gene symbolFOXM1
protein nameForkhead box protein M1 (Forkhead-related protein FKHL16, HNF-3/fork-head homolog 11, Trident)
uniprot resolutionaccession:Q08050, reviewed human, one record; primary gene name read back as FOXM1 (entry FOXM1_HUMAN), synonyms FKHL16, HFH11, MPP2, WIN; length 763 aa checked against UniProt 2026-09-21.
symbol collision checkdata/symbol-collisions.json greped 2026-09-21 for FOXM1: no row existed. A row IS proposed in this beat's memo - the UniProt synonym WIN collides with the WDR5 WIN motif and with WIN-site inhibitors, which are a live topic in this graph.
baitMEN1 (O00255)
shown for fusionMLL-AF9 in mouse and primary human MLL-rearranged AML for the DEPENDENCY; nothing in KMT2A::AFF1 or B-lineage, and NO fusion system at all for the menin interaction.
fetched2026-09-21
interaction with men1
{
  "source_pmid": "25609649",
  "source_class": "proteome-scale TAP/MS atlas of 56 transcription factors; this row is one of its 53 non-spoke validation rows",
  "n_intact_records": 1,
  "n_spoke_expanded": 0,
  "n_source_publications": 1,
  "intact_acs": [
    "EBI-10690476"
  ],
  "intact_miscore": 0.4,
  "figure": "supplementary figure 3a",
  "system": "Homo sapiens HEK293T embryonic kidney cell",
  "publication_row_composition": {
    "total": 4119,
    "spoke_expanded_association": 4065,
    "non_spoke_physical_association": 53,
    "adp_ribosylation": 1
  },
  "men1_rows_in_publication": 2,
  "menin_mentions_in_main_text": 0,
  "independent_corroboration": 0,
  "shared_pdb_entries": []
}
independent role in this disease
{
  "claim": "required for MLL-AF9 leukaemia stem cell survival, quiescence and self-renewal; acts by binding and stabilising beta-catenin",
  "pmid": "32066721",
  "system": "murine MLL-AF9 in vivo plus primary human MLL-rearranged AML LSCs",
  "already_cited_in_graph": [
    "cell-state-quiescence",
    "exp-cell-cycle-sorted-menin-sensitivity",
    "hyp-menin-sensitivity-tracks-cell-cycle-state"
  ]
}
How this could bite us back
  • FOXM1 inhibitors exist and 'menin partner plus LSC dependency' reads like a combination rationale. The dependency is strong and the menin link is one undiscussed supplementary panel; combining them on paper would be a mechanism this graph invented.
  • FOXM1 is required for proliferation in normal tissue too, including regenerating liver and intestinal epithelium. An FOXM1-directed agent in an infant is an essential-function liability on every proliferating compartment.
  • The LSC evidence is myeloid MLL-AF9. This project's most-documented synthesis failure is importing exactly that system into infant B-lineage KMT2A::AFF1, and this entry is a prime candidate for it.
Still open
  1. Does menin bind FOXM1 in a leukaemic cell, and does a menin inhibitor disrupt it? If yes, the FOXM1 LSC dependency and the menin dependency are the same axis and the drug already engages it; if no, they are two independent dependencies that happen to sit in the same cells.
  2. Is the FOXM1-beta-catenin route active in KMT2A::AFF1 B-lineage disease? Every FOXM1 measurement in this disease is myeloid MLL-AF9 or myeloid patient LSC.
  3. Beta-catenin is now reachable from FOXM1 in LSCs and from DOT1L/MLLT10 in intestinal crypts. Are these one Wnt axis in the KMT2A interactome or two unrelated observations? Nobody has put them in the same experiment.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct's REST record holds one MEN1-FOXM1 row, EBI-10690476, typed MI:0915 physical association, detection 'pull down', host HEK293T, no expansion method, intact-miscore 0.4, annotated 'figure legend (sf3a)', from PMID 25609649 - which contributes 4,119 rows in total, of which 4,065 are spoke-expanded associations and only 53 are non-spoke physical associations; MEN1 appears in exactly two rows of that publication.

    What this evidence cannot show

    One database on one day, REST endpoint (brief rule 93). The non-spoke flag means the curator read a directed validation panel rather than splitting an n-ary deposition - it does not mean the pair was shown to be direct. intact-miscore 0.4 is low. MEN1 being a bait in a transcription-factor atlas is unexplained and could not be verified from the paper.

    Evidence that argues against this (1)
    • The non-spoke flag here is misleading as a quality signal. In this publication 4,065 of 4,119 rows are spoke-expanded associations and 53 are non-spoke physical associations, so the flag separates the atlas's validation panels from its screen output - not a good experiment from a bad one. A focused paper that deposits an n-ary complex gets spoke-flagged while this atlas panel does not, which is the one place in this beat's thirteen pairs where the spoke flag and the source class disagree. EMBL-EBI IntAct 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In mouse MLL-AF9 leukaemia in vivo and in primary human MLL-rearranged AML leukaemia stem cells, Foxm1 is required for LSC survival, quiescence and self-renewal, acts by directly binding beta-catenin and stabilising it against degradation, and FOXM1 inhibition suppresses leukaemogenic potential and induces LSC apoptosis in vitro and in xenograft mice.

    PMID 32066721DOI
    What this evidence cannot show

    Murine MLL-AF9 myeloid leukaemia for the in vivo arm, with primary human MLL-rearranged AML LSCs for the translational arm - no B-lineage system and no KMT2A::AFF1 (brief rule 73), which is this project's commonest documented generalisation failure. Already cited three times in this graph, for cell-cycle quiescence rather than for the interactome (brief rule 82); this entry adds the partner node, not a new reading.

    Evidence that argues against this (1)
    • The FOXM1 dependency is established in murine MLL-AF9 myeloid leukaemia, and this project's commonest documented synthesis failure is importing exactly that system into infant B-lineage KMT2A::AFF1. The human arm is primary MLL-rearranged AML LSCs - also myeloid. No B-lineage or KMT2A::AFF1 measurement exists, so the dependency claim is bounded to myeloid disease. Sheng Y et al. 2020
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The source publication for the MEN1-FOXM1 row is a proteome-scale study of soluble and chromatin-associated complexes of 56 transcription factors and 37 FOX-family members, with 214 purifications and 2,156 reported interactions; the strings MEN1 and menin do not appear anywhere in its 84,072 characters of main text as retrieved from Europe PMC.

    PMID 25609649DOI
    What this evidence cannot show

    Full text read this session via Europe PMC fullTextXML (PMC4332150, 84,072 characters). The zero-mention statement is bounded to the MAIN text - the supplementary figure legend that IntAct cites was not retrieved, so menin may well be named there. HEK293T cells; no haematopoietic system, no KMT2A fusion (brief rule 73).

    Evidence that argues against this (2)
    • The bound was tested and the absence does not survive outside the main text. Europe PMC's supplementaryFiles endpoint returns the paper's full supplement (33 files, 63 MB) and MEN1 is an exact cell value in at least three of the supplementary workbooks - msb0011-0775-sd7.xlsx, whose only sheet is named 'protein identification list'; msb0011-0775-sd9.xlsx; and msb0011-0775-sd10.xlsx, whose only sheet is named 'HCIP', the study's high-confidence-interacting-protein list. The string 'menin' appears nowhere in any of them. So the correct statement is not that the paper never names menin, but that it names it ONLY as a symbol in machine-readable tables, one of which is its high-confidence interactor list - which makes the MEN1-FOXM1 row better provenanced than 'a panel nobody discussed' and still entirely undiscussed in prose. Europe PMC 2026
    • The 84,072-character figure is also not the document. The same fullTextXML retrieved today is 206,743 characters of XML and declares 21 <supplementary-material> elements - four figures, twelve tables, a review-process file, two source-data archives, a dataset and a legends file. A character count of the flattened main text is a count of the part of the document the retrieval route happened to flatten, and quoting it gives an absence an air of exhaustiveness the measurement does not have. Brief rule 80's general form applies inside one paper as well as across an index: state which part you searched. Europe PMC 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    As of 2026-09-21, PubMed returns 0 records for MEN1[tiab] AND (FOXM1 OR 'forkhead box M1')[tiab] and 5 for menin[tiab] AND (FOXM1 OR 'forkhead box M1')[tiab], none of which concerns a physical interaction - all five are beta-cell, placental-lactogen or review records - so the menin-FOXM1 interaction has no corroboration in the indexed literature.

    What this evidence cannot show

    Titles and abstracts only; supporting figures not searched (brief rule 80) - and the one record that does exist is itself in a supplementary figure, so this absence bounds the discussion of the interaction rather than its existence. Name list tried: MEN1, menin, FOXM1, forkhead box M1. The retired/alternate symbols HFH11, MPP2, WIN, Trident and FKHL16 were not run as operands. Operand control (brief rule 67): menin[tiab] alone returns 1,264 and FOXM1[tiab] a large literature, so the zero is a real intersection.

    Evidence that argues against this (2)
    • The unrun synonyms were run and the answer does not move, with one deliberately excluded. Re-run 2026-09-21: '(menin OR MEN1 OR multiple endocrine neoplasia type 1)[tiab] AND (FOXM1 OR "forkhead box M1" OR HFH11 OR MPP2 OR Trident OR FKHL16 OR FoxM1b OR FoxM1c)[tiab]' returns 5 - the same five PMIDs - and FOXM1[tiab] AND (MLL OR KMT2A)[tiab] still returns 6. WIN was NOT added as an operand, because data/symbol-collisions.json carries a row for it: WIN is a current UniProt gene synonym of FOXM1 and also this project's own abbreviation for the WDR5-INteraction motif, and WIN[tiab] alone returns 11,443 records. Excluding it is the registry working as intended rather than a gap in the name list. NCBI PubMed 2026
    • The instrument is mislabelled in the same way as the other bounded queries in this beat: PubMed's querytranslation shows 'menin[tiab]' becoming "menin"[Title/Abstract] while the parenthesised '(FOXM1 OR "forkhead box M1")[tiab]' loses its tag and expands to [All Fields]. So the zero for the MEN1 arm is broader than titles and abstracts, and - more to the point here - the interaction IS in the indexed record: PMID 25609649's supplementary HCIP table lists MEN1, and no PubMed query of any shape reaches a spreadsheet. 'No corroboration in the indexed literature' is true of the discussion and untestable for the data. NCBI PubMed 2026
Connected to