Does menin hold FANCD2 in a KMT2A-rearranged cell, and does revumenib let go? Nobody has asked — zero records over every menin-inhibitor name
also called menin FANCD2 coIP under menin inhibitor · FANCD2 displacement experiment
The single experiment that would decide whether menin's Fanconi connection matters in leukaemia has never been done: pull menin out of a KMT2A-rearranged leukaemia cell, see whether FANCD2 comes with it, then repeat with the drug on board. This entry writes down the experiment and records, with the exact queries, that the literature contains nothing.
The same thing, for a scientist
**The absence first, because it is the finding.** A PubMed title/abstract search over every menin-inhibitor name in use — `revumenib`, `ziftomenib`, `SNDX-5613`, `KO-539`, `MI-503`, `MI-2`, and the phrase `"menin inhibitor"` — against `FANCD2` or `Fanconi` returns **0 records** [E1]. Broadening the drug set against `"DNA repair"` or `"DNA damage"` returns **1 record**, a therapy-related AML case report in which revumenib appears as treatment and DNA damage as aetiology; title-checked count **0** [E2]. And `menin AND (co-immunoprecipitation OR coimmunoprecipitation) AND (FANCD2 OR NF-kappaB)` returns **0** [E3]. **No menin inhibitor has ever been assayed against any Fanconi protein, in any system.** **The experiment.** Endogenous co-immunoprecipitation of menin and FANCD2 in three KMT2A-rearranged human lines (SEM and MV4-11 for KMT2A::AFF1, MOLM-13 for KMT2A::MLLT3) against a non-rearranged comparator, in four arms: untreated, 2 Gy ionising radiation, revumenib at saturating occupancy, and revumenib plus radiation. Reciprocal pulls both ways, endogenous antibodies only, with a KMT2A MBM1 blot on the same membrane as the positive displacement control that the pocket inhibitor is guaranteed to produce. Read-out at 4 h, before transcriptional output moves. **What each outcome means.** No co-IP in any arm kills `hyp-menin-fancd2-bridge-explains-the-fa-selectivity-signal` outright. A co-IP that revumenib does not touch kills the hypothesis' pharmacological consequence while leaving the biology. A co-IP that revumenib displaces is the JunD result over again in a new partner and would be the most consequential finding this beat could produce.
Design
| element | choice | why |
|---|---|---|
| lines | SEM, MV4-11 (KMT2A::AFF1), MOLM-13 (KMT2A::MLLT3) | the three menin-dependent lines this graph already uses; SEM is the B-lineage arm, which matters because the FA signal is stronger in the lymphoid half |
| comparator | a menin-expressing, non-KMT2A-rearranged leukaemia line | separates 'menin binds FANCD2' from 'menin binds FANCD2 in fusion-driven cells' |
| arms | untreated; 2 Gy IR; revumenib at saturating occupancy; revumenib + IR | the 2003 report says the association is gamma-irradiation-enhanced, so an unirradiated arm alone could produce a false negative |
| antibodies | endogenous both directions, menin-pull and FANCD2-pull | the 2003 evidence is largely FLAG-tagged over-expression; repeating that would add nothing |
| positive control | KMT2A (or fusion) MBM1 displacement on the same membrane | proves the compound reached its target in that lysate; without it a negative is uninterpretable |
| negative control | an abundant nuclear protein that should not move | guards against a lysate-quality artefact |
| timing | 4 h | before transcriptional output changes, so a loss of association is displacement rather than downstream protein loss |
| second read-out | FANCD2 short/long form ratio on the same blots | asks in passing whether menin occupancy changes K561 monoubiquitination, which nobody has ever looked at |
Cost tier: low. Three lines, one compound, four arms, antibodies that all exist commercially. This is a few weeks of one person's time and it settles a question the project has been circling for four cycles.
Why the absence is real and not a search failure
Three separate queries, each written to fail loudly rather than quietly, and the exact strings are in the evidence items.
- Every clinical and tool menin-inhibitor name currently in the literature was enumerated, including the pre-approval code names
SNDX-5613andKO-539, because a drug's literature is split across its code name and its INN and a query on one alone under-counts. MI-2was included despite being a short, word-like token precisely because the brief's alias-precision rule cuts both ways: a short alias inflates counts. It returned nothing to inflate.- The
"DNA repair"/"DNA damage"broadening was run to catch a paper that touches the topic without naming FANCD2. It returned one record and title-checking removed it 2.
Subtraction check. Before writing this absence the graph's own citations were searched for each of these PMIDs. The only near-miss is PMID 31590682, the chidamide plus MI-3 paper, which was already absent from the graph and is now cited in gene-fancd2 and hyp-menin-fancd2-bridge-explains-the-fa-selectivity-signal. It does not perform this experiment: it assays CHK1, CHK2, HR and NHEJ in a combination setting and never blots a Fanconi protein.
What the absence is not. It is not evidence that the interaction does not exist in leukaemia. Nobody has looked, and the standing rule applies: an absence of testing is not a negative result.
| design | Endogenous reciprocal co-immunoprecipitation of menin and FANCD2 in SEM, MV4-11 and MOLM-13 plus a non-KMT2A-rearranged comparator line, in four arms: untreated, 2 Gy ionising radiation, revumenib at saturating occupancy, revumenib plus IR. Lysates harvested at 4 h. KMT2A MBM1 displacement blotted on the same membrane as the target-engagement control. |
|---|---|
| readout | presence and intensity of FANCD2 in the menin pull and of menin in the FANCD2 pull, normalised to input; secondarily the FANCD2 short/long (unmodified/monoubiquitinated) ratio on the same blots |
| controls | target engagement via MBM1 displacement on the same membrane; an abundant nuclear protein that should not move; IgG pulls; a non-rearranged comparator line; input blots for every arm |
| cost tier | low |
| timeline | 4-6 weeks with existing lines, antibodies and compound |
| status | planned |
decision rule
{
"no_coIP_in_any_arm": "kills hyp-menin-fancd2-bridge-explains-the-fa-selectivity-signal outright; the 2003 result would be mouse-fibroblast-specific",
"coIP_present_revumenib_no_effect": "the association is real but pharmacologically inert; the hypothesis loses its only consequence for the drug",
"coIP_present_revumenib_displaces": "the JunD result in a new partner; would require the menin inhibitor programme to ask what else the pocket was holding"
}counted absences — the numbers behind the chart
[
{
"query": "(revumenib[tiab] OR ziftomenib[tiab] OR \"SNDX-5613\"[tiab] OR \"KO-539\"[tiab] OR \"MI-503\"[tiab] OR \"MI-2\"[tiab] OR \"menin inhibitor\"[tiab]) AND (FANCD2[tiab] OR Fanconi[tiab])",
"raw": 0,
"title_checked": 0
},
{
"query": "(\"menin inhibitor\"[tiab] OR revumenib[tiab] OR ziftomenib[tiab]) AND (\"DNA repair\"[tiab] OR \"DNA damage\"[tiab])",
"raw": 1,
"title_checked": 0
},
{
"query": "menin[tiab] AND (co-immunoprecipitation[tiab] OR coimmunoprecipitation[tiab]) AND (FANCD2[tiab] OR \"NF-kappaB\"[tiab])",
"raw": 0,
"title_checked": 0
}
]- A positive result here is bad news for the class, not good: it would mean the approved drug releases a DNA-repair effector in children who also receive genotoxic chemotherapy.
- A negative result removes a hypothesis but does not remove the unexplained co-dependency, so the project should decide in advance whether it will publish the negative or quietly keep looking.
- Three immortalised lines cannot speak for primary infant blasts; a clean negative in SEM and MV4-11 still leaves the patient-derived question open.
- Should this be run as a single blot alongside the already-designed JunD displacement experiment, adding FANCD2 and RelA to the same membranes for almost no extra cost?
- Is 4 h the right window, or does a gamma-irradiation-enhanced association need a damage time-course rather than a single point?
- If the association exists but is not displaced by revumenib, is there any menin ligand that would displace it, and would such a compound be a useful tool or a liability?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
No published study connects any menin inhibitor to any Fanconi protein. A PubMed title/abstract search over revumenib, ziftomenib, SNDX-5613, KO-539, MI-503, MI-2 and 'menin inhibitor' against FANCD2 or Fanconi returns 0 records.
PubMed counted-absence search, menin inhibitors against FANCD2/Fanconi, run 2026-09-20 · PubMed (NCBI) 2026What this evidence cannot show
Title/abstract search only. It cannot see a supplementary western blot, a figure panel in a paper about something else, or an unpublished negative. Bleximenib, enzomenib and icovamenib were not in this particular string; they are later compounds and the conclusion would not change, but the query as written is the query to quote.
Evidence that argues against this (1)
- A zero from a drug-name query is vulnerable in one direction this project has documented: complex-level or pathway-level framing. A paper that says 'menin inhibition perturbs the DNA damage response' without naming FANCD2 would be invisible to this string, which is why the broader query in E2 was also run. PubMed (NCBI) 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Broadening to 'menin inhibitor' OR revumenib OR ziftomenib against 'DNA repair' OR 'DNA damage' returns 1 PubMed record, which title-checks out: it is a case report of concurrent therapy-related AML and lymph node tuberculosis in which revumenib is part of the treatment regimen and cytotoxic DNA damage is the stated aetiology of the t-AML. Title-checked count: 0.
Concurrent therapy-related acute myeloid leukemia and lymph node tuberculosis following treatment for lung squamous cell carcinoma: a case report and literature review. · Yu Y, Yang D, Chen X, Chen Y · Front Oncol 2026What this evidence cannot show
A single case report, cited here only to document why the one hit in a counted absence was excluded. It contains no menin-inhibitor pharmacology and no repair assay.
Evidence that argues against this (1)
- A title-checked search count is two judgements stacked: the query decides what is reachable, and the reader decides what is 'on-beat'. Both are recorded here, which is the right practice, but neither is a measurement. This project has since established that field-scoped queries can return zero regardless of the literature, and that a symbol-level count misses work indexed under the pathway or the complex rather than the gene - so a small count bounds what was found, not what exists. Shan X et al. 2017
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
A PubMed title/abstract search for menin together with co-immunoprecipitation and either FANCD2 or NF-kappaB returns 0 records, so neither of this beat's two partner axes has a co-IP study indexed at the abstract level beyond the original founding papers.
PubMed counted-absence search, menin co-immunoprecipitation against FANCD2/NF-kappaB, run 2026-09-20 · PubMed (NCBI) 2026What this evidence cannot show
`moderate`, not `strong`. Abstracts rarely name the assay, so a zero here is weak: the 2001 NF-kappaB paper itself says 'we used immunoprecipitation to identify interacting proteins' and would be missed by the hyphenated spellings searched. This query is recorded because it was run and returned nothing, not because a zero from it means much.
Evidence that argues against this (1)
- The founding NF-kappaB paper does report co-immunoprecipitation and is not returned by this query, which demonstrates the query's own recall failure and is the reason its confidence is moderate. Heppner C et al. 2001
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.analysis
The experiment has a designed precedent in this graph: the JunD displacement co-IP under a clinical menin inhibitor was written as a planned experiment for the same reason, that a drug occupying the MBM pocket should release whichever partner uses it, and that has never been shown pharmacologically for any approved compound.
JunD displacement co-IP under a clinical menin inhibitor (planned experiment) · Oncofusion knowledge graph, exp-jund-displacement-coip-under-clinical-menin-inhibitor 2026What this evidence cannot show
An internal planned-experiment entry, not a published result. It is cited to establish that the design pattern and its controls already exist in this project and this experiment should reuse them rather than reinvent them - and, if run, should include FANCD2 and RelA on the same membranes as JunD, which costs almost nothing extra.
Evidence that argues against this (1)
- Citing this graph's own planned experiment as precedent is circular: a design we wrote is evidence about our reasoning, not about the world. The JunD precedent shows the assay is conceivable, not that a menin-FANCD2 co-IP would behave the same way - and the FANCD2 association has a published genetic negative that JunD does not. Marek LR et al. 2008
endogenous co-IP with and without a clinical menin inhibitor; decides the hypothesis
does saturating menin-pocket occupancy displace FANCD2?
- analogous to →Does revumenib displace JunD, and at what concentration relative to KMT2A? (The yes/no half is already answered)
same design; FANCD2 and RELA can be blotted on the same membranes