Oncofusion

start / graph / hyp-q128-three-of-menins-nine-declined-partners-are-e3-ligase-substrate-receptors

Hypothesisspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q128-the-menin-interactome-that-is-not-chromatin

Three of menin's nine declined partners are E3 ligase substrate receptors - either menin turnover is ubiquitin-controlled, or substrate receptors are this beat's artefact signature

also called menin E3 substrate receptor pattern · FBXW7 SPOP ZC3HC1 menin

Three of the nine weak menin partners are proteins whose job is to pick out other proteins for destruction. Either menin is one of the things they destroy, or these proteins simply stick to everything when they are artificially overexpressed.

The same thing, for a scientist

Of the nine IntAct-curated MEN1 partners this beat declined, **three are substrate-recognition subunits of E3 ubiquitin ligases**: FBXW7 (SCF-FBXW7), SPOP (CUL3-RBX1-SPOP) and ZC3HC1/NIPA (an SCF F-box protein) [E1]. The hypothesis has two arms and they make opposite predictions. **Arm A, biological:** menin abundance is controlled by ubiquitin-dependent degradation through one or more of these receptors, which would be a non-chromatin menin function of exactly the kind the beat was dispatched to look for, and would matter because menin protein level sets the dose of inhibitor needed to occupy it. **Arm B, artefactual:** over-expressed substrate receptors bind promiscuously and are systematically enriched in AP-MS and proximity datasets, so three of nine is what a low-specificity assay applied to a cancer-gene library produces. **The two arms are separated by experiments that already have reagents** [E3]. Arm B is currently the better-supported reading, because all four SPOP substrate-cleft mutants (Y87C, F102C, F133L, F133V) left the menin signal unchanged [E2] - which is what a non-substrate looks like.

Go deeper
The pattern, and the null this entry deliberately does not claim to have computed
partnerligaserecognition modulemenin-row verdict
FBXW7SCF (CUL1-SKP1-RBX1)WD40 phosphodegron receptordeclined, 4 BRET alleles, mutation with no effect
SPOPCUL3-RBX1MATH domain degron receptordeclined, 4 BRET alleles, mutation with no effect
ZC3HC1 (NIPA)SCF (CUL1-SKP1-RBX1)F-boxdeclined, 1 BioPlex spoke-expanded row

Rule 84 applies and is being obeyed rather than quoted. Three of nine is not reported here as an enrichment, because the null has not been computed. The nine are not a random draw from the genome: seven come from the OncoPPi v2 library, which is a curated cancer-gene set, and cancer-gene sets are enriched for ubiquitin machinery by construction. The correct null is the proportion of E3 substrate receptors among the OncoPPi v2 library genes that were screened against MEN1, which is computable from Mo et al. Table S1(A) and has not been computed. Until it is, what this entry asserts is a class with a known assay pathology, not an over-representation.

What makes the class worth an entry regardless of the null is that it has a mechanistic reading and an artefactual reading that disagree, and both are cheap to test.

Arm A: menin turnover is ubiquitin-controlled

If true, this is a genuine non-chromatin menin function and it has a direct clinical edge. Menin inhibitors are occupancy drugs: revumenib and ziftomenib compete for the MBM pocket. The concentration needed to occupy menin scales with how much menin there is. A leukaemia with a defective menin-degradation pathway would carry more menin protein and would need more drug for the same effect - an intrinsic-resistance mechanism with a genotype attached.

FBXW7 is the strongest candidate of the three, because it is recurrently mutated in T-lineage acute leukaemia, which overlaps the menin-inhibitor population. The prediction is sharp: menin protein should rise on FBXW7 loss and fall on FBXW7 over-expression, and menin should carry a CDC4-phosphodegron-like motif.

Arm A also predicts something about the existing data that has never been checked: if menin is an SCF substrate, neddylation inhibition (MLN4924/pevonedistat) should stabilise it, and pevonedistat has been run in leukaemia lines with proteomics.

Arm B: substrate receptors are the artefact signature of this dataset

Substrate-recognition subunits are among the best-known promiscuous binders in interaction proteomics. They are built to survey the proteome, they are low-abundance and therefore heavily over-expressed relative to endogenous when tagged, and their captured material is dominated by whatever is abundant.

The strongest single piece of evidence for Arm B is already in the data and it is a negative. IntAct records four SPOP alleles against MEN1 - Y87C, F102C, F133L, F133V - and all four are typed mutation with no effect 2. Those are the MATH-domain substrate-binding hotspot residues; they are the substitutions that abolish SPOP's grip on its degrons. A signal that survives four independent disruptions of the substrate-binding cleft is not substrate recognition. The same logic applies, less cleanly, to the four FBXW7 WD40 arginines (R465, R479, R505), which are the phosphodegron-contacting residues and also changed nothing.

So the data currently favour Arm B, and this entry says so rather than leaving the two arms balanced for rhetorical symmetry.

Facts and figures
statementThree of menin's nine declined IntAct partners are E3 ubiquitin ligase substrate receptors. Either menin abundance is ubiquitin-regulated through one of them (Arm A, a genuine non-chromatin menin function with a resistance implication), or substrate receptors are the promiscuity signature of the over-expression assays these rows come from (Arm B, currently better supported).
origindata-pattern
prior probability0.2
absence classmeasured-not-analysed
absence class noteTwo of the four falsifiers are pure re-analysis of already-published, already-deposited data: Mo et al. Table S1(A) inside PMC9597701 for the library composition, and the BioPlex 3.0 supplementary networks for menin's generic recovery by E3 baits. Neither has been run. The wet-work arm (FBXW7-null menin immunoblot) is separately never-measured.
predictions
[
  "Arm A: menin protein rises at least 1.5-fold by quantitative immunoblot in an FBXW7-null isogenic derivative relative to its FBXW7-wild-type parent, in a human leukaemia line, with no change in MEN1 transcript by qPCR.",
  "Arm A: menin protein rises at least 1.5-fold after 6 hours of 1 uM pevonedistat in the same line.",
  "Arm A: O00255 contains at least one match to the CDC4 phosphodegron consensus (T/S)P..(T/S/E/D) in a region predicted disordered.",
  "Arm B: menin is recovered as a prey in a large fraction of tagged E3 substrate-receptor pull-downs in BioPlex, not preferentially in FBXW7, SPOP and ZC3HC1.",
  "Arm B: the proportion of E3 substrate receptors among the OncoPPi v2 library genes screened against MEN1 is comparable to 3/9."
]
experiments
[
  "exp-q128-is-menin-an-scf-fbxw7-substrate-one-week-with-existing-reagents"
]
analyses
[
  "ana-q128-all-nine-unwritten-menin-partners-are-two-papers-and-neither-paper-is-about-menin"
]
falsifiers
[
  "ARM A DIES IF: menin protein in an FBXW7-null human leukaemia line is within 1.2-fold of its isogenic FBXW7-wild-type parent by quantitative immunoblot (n>=3, loading-normalised), AND is within 1.2-fold after 6 h of 1 uM pevonedistat in the same line. Both conditions must hold; either alone leaves a route open.",
  "ARM A DIES IF: a scan of O00255 against the CDC4 phosphodegron consensus returns no match in a region with AlphaFold pLDDT below 50. This is a desk falsifier, costs nothing, and should be run before anything else.",
  "ARM B DIES IF: the proportion of E3 ubiquitin ligase substrate receptors among the genes screened against MEN1 in Mo et al. Table S1(A) is below 10%, making 3 of 9 a genuine over-representation at a two-sided binomial p below 0.05. THIS IS THE NULL RULE 84 REQUIRES AND IT HAS NOT BEEN COMPUTED - the entry is incomplete until it is.",
  "ARM B DIES IF: menin is recovered as a prey in fewer than 10% of tagged E3 substrate-receptor baits in the BioPlex 3.0 293T network, i.e. its recovery by ZC3HC1 is specific rather than generic. This is computable from the published BioPlex supplementary tables with no wet work."
]
How this could bite us back
  • If Arm A is right, menin protein level is a resistance variable nobody measures. Menin inhibitor trials stratify on fusion partner and on MEN1 mutation status; none reports baseline menin protein abundance, so a patient with stabilised menin would be indistinguishable from one with a resistance mutation at the point of failure.
  • If Arm A is right, combining a menin inhibitor with a neddylation or proteasome inhibitor would RAISE menin levels and could antagonise the menin inhibitor - the opposite of the combination logic anyone would reach for. Bortezomib and pevonedistat are both used in haematological malignancy.
  • If Arm B is right, the graph should expect more E3 substrate receptors in every future IntAct-derived gap list, and treating each as a discovery would consume beats indefinitely. A class-level rejection rule is cheaper than nine more entries.
Still open
  1. What fraction of the OncoPPi v2 library are E3 ubiquitin ligase substrate receptors? Until this is computed the 3-of-9 pattern has no null and rule 84 is unsatisfied.
  2. Is menin a generic prey in E3-receptor pull-downs across BioPlex, or specifically in ZC3HC1's? This is a supplementary-table query and it would settle Arm B for the AP-MS half of the evidence.
  3. Is there any published measurement of menin protein half-life in any cell? If menin is long-lived, Arm A is much less interesting even if a ligase contacts it.
  4. Does menin protein level vary between KMT2A-rearranged patient samples enough to matter for inhibitor occupancy? This is the clinically load-bearing version of the question and is independent of which arm is right.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Three of the nine IntAct-curated MEN1 partners adjudicated in this beat are E3 ubiquitin ligase substrate-recognition subunits: FBXW7 (Q969H0, SCF F-box/WD40), SPOP (O43791, CUL3-BTB/MATH) and ZC3HC1 (Q86WB0, SCF F-box, also known as NIPA). All three are declined on evidence quality, and all three are nuclear with experimental evidence, so compartment does not separate them from menin.

    What this evidence cannot show

    The class assignment is from UniProt protein names and family annotation, not from a systematic ligase-component database join. The null - what fraction of the screened library are E3 substrate receptors - has NOT been computed, so no enrichment is claimed and rule 84 is not satisfied for any such claim. ZC3HC1's F-box role and its nuclear-basket role are separate literatures and this entry does not adjudicate which is primary.

    Evidence that argues against this (1)
    • Three of nine is entirely unremarkable under a plausible null. The human genome encodes roughly 70 F-box proteins and a comparable number of BTB adaptors, and a curated cancer-gene library over-samples them. Reporting this as a pattern without the denominator is precisely the error rule 84 documents costing this project four cycles, and the only defence offered here is that the entry states the null is uncomputed. IntAct (EMBL-EBI) 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct curates four MEN1-SPOP records from PMID 35512704 carrying SPOP substitutions p.Tyr87Cys, p.Phe102Cys, p.Phe133Leu and p.Phe133Val, and four MEN1-FBXW7 records carrying FBXW7 p.Arg465Cys, p.Arg465His, p.Arg479Gln and p.Arg505Cys. Every one of the eight is typed 'mutation with no effect'. These are the MATH-domain and WD40 substrate-binding hotspot residues of the two receptors.

    What this evidence cannot show

    'Mutation with no effect' is IntAct's curator reading of the paper's comparison, not the paper's own phrase, and the underlying supplementary table was not parsed. The inference that an unchanged signal across substrate-cleft mutants argues against substrate recognition assumes the assay could have detected a change, which a single BRET readout at miscore 0.27 may not have been powered to do.

    Evidence that argues against this (1)
    • A substrate can be recognised through more than one surface, and some SCF and CUL3 substrates bind outside the canonical cleft. An unchanged signal across cleft mutants is therefore evidence against canonical degron recognition, not against association. It also does not exclude menin being a ligase COFACTOR or scaffold rather than a substrate, which would predict exactly this pattern. IntAct (EMBL-EBI) 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Both arms are testable with existing reagents and no new material: menin protein level by immunoblot in an FBXW7-null versus FBXW7-wild-type isogenic pair, menin level under neddylation inhibition, and a sequence scan of O00255 against the CDC4 phosphodegron consensus. The corresponding experiment entry is exp-q128-is-menin-an-scf-fbxw7-substrate-one-week-with-existing-reagents.

    What this evidence cannot show

    'Existing reagents' is an assertion about general availability of FBXW7-null isogenic lines, pevonedistat and a validated anti-menin antibody, not a verified inventory of any particular laboratory. No supplier, catalogue number or cell-line identifier is given here, so the cost estimate is a class estimate.

    Evidence that argues against this (1)
    • A negative result from the proposed experiment would not distinguish Arm B from a real but conditional interaction - one that only occurs under a stress, a cell cycle phase or a modification state not present in an unperturbed isogenic pair. The experiment is cheap and its negative is weak, which should be said before it is run rather than after. IntAct (EMBL-EBI) 2026
Connected to