Oncofusion

start / graph / risk-a-menin-degrader-is-not-a-pocket-inhibitor-for-the-hr-scaffold

Riskmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q34-menin-surface-map

A menin degrader removes six surfaces; a pocket inhibitor blocks one and touches a second - and in a child receiving TOP2 poisons that difference is a repair-competence difference nobody has measured

also called degrader versus inhibitor HR risk · menin HR scaffold risk

Menin does several separate jobs, and one of them is helping cells repair broken DNA accurately. A drug that blocks one pocket on menin leaves most of the other jobs alone; a drug that destroys the whole protein removes all of them, and children with this leukaemia are also given drugs that deliberately break DNA.

The same thing, for a scientist

Menin is now reported as the scaffold of an H3K36me3-dependent homologous-recombination relay: it reads gamma-H2A.X through residues 371-400, recruits SETD2 to write H3K36me3, and LEDGF reads that mark and recruits CtIP for end resection. This beat maps six menin surfaces with residue coordinates - the MBM cavity, the acidic shelf, the SvID, the gamma-H2A.X motif, the nucleosome grip and the LEDGF patch - and the five clinical/tool compounds contact only the first two. A degrader removes all six, in every cell, for as long as it is dosed. The exposed population is concrete: the standard backbone in KMT2A-rearranged infant and paediatric leukaemia includes etoposide, doxorubicin, daunorubicin and mitoxantrone, all of which create TOP2-mediated double-strand breaks, and this graph already holds fifteen backbone drug entries. The nearest thing to a measurement is PMID 31590682, in which the HDAC inhibitor chidamide COMBINED with the tool menin-MLL inhibitor MI-3 disrupts the CHK1/CHK2 checkpoint and DNA repair 'likely via homologous recombination, but not NHEJ' in human MLL-rearranged AML - a synergy design at subtoxic single-agent doses that does not isolate the menin-inhibitor contribution and uses no clinical compound. Beyond that, no study has measured gamma-H2A.X resolution, H3K36me3 at breaks, RAD51 foci, an HR/NHEJ reporter or menin-gamma-H2A.X co-IP under revumenib, ziftomenib, bleximenib or enzomenib as a single agent, in any system. The risk is therefore unquantified in both directions, and the inhibitor-versus-degrader distinction is currently being made without a single-agent repair measurement for either modality.

Go deeper
What each modality removes, surface by surface
menin surfaceresiduescontacted by revumenib / ziftomenib / DS-1594b?removed by a degrader?
MBM1 cavity (KMT2A, JUND)~135-182, 238-249, 276-285, 319-326, 341, 359-363yes, this is the drug siteyes
acidic shelf / MBM2 counter-surface363-378yes - E366 2.6-3.7 A, V371 2.3-4.3 A (not bleximenib: 8.0 / 10.7 A)yes
gamma-H2A.X motif371-400only its first residueyes
SvID / SUV39H1360-445; critical residues D418, W436no - 9.9-16.6 Ayes
nucleosome grip (H3/H4/H2B/DNA)353-362, 428-442, 556-587noyes
LEDGF/PSIP1 patch91-105, 209-212no - 22.8-37.9 Ayes

All distances measured 2026-09-22 from the deposited coordinates (ana-q34-menin-surface-geometry). A pocket inhibitor is a one-and-a-half-surface intervention. A degrader is a six-surface intervention. This graph has a degrader dossier (target-phase2-degrader-modality-dossier) that does not contain this argument, because the HR scaffold was published after it.

Why this is a paediatric-specific risk and not a general one

Three things compound in this population and in no other:

  1. The backbone is genotoxic by design. Etoposide, doxorubicin, daunorubicin and mitoxantrone all poison TOP2 and generate protein-linked double-strand breaks. Menin inhibitors are being combined with chemotherapy in frontline paediatric protocols, not used as monotherapy in relapse only.
  2. KMT2A rearrangement is itself a TOP2-associated event. This graph holds an entire mechanism arm on the proposition that chromatin state at the KMT2A breakpoint cluster region gates rearrangement. An agent that impairs accurate HR while TOP2 breaks are being made is an agent that shifts repair toward the error-prone end - the same argument risk-an-h3k9-axis-inhibitor-given-with-a-top2-poison-removes-a-repair-licence already makes for the H3K9 axis, now with menin as the node.
  3. The exposure is long. Menin inhibitors are given for months, and a second malignancy in a cured infant is the outcome that matters over a forty-year horizon.

And the counterweight, stated fairly: if menin's HR role is real, impairing HR in a leukaemic blast is a therapeutic mechanism, not only a toxicity - it predicts synergy with PARP inhibition and with the TOP2 backbone itself. The 2026 paper's own peptide tool suppresses HR and potentiates NHEJ. This entry does not claim the direction is bad; it claims the direction is unmeasured and that the modality choice currently being made ignores it.

Facts and figures
kindessential-function
severityunquantified in both directions. The mechanism is published, the exposed population is the standard one, and the measurement has never been made in any system - so this is a named uncertainty rather than a graded hazard.
mitigationMeasure it. The cheapest sufficient readout is a gamma-H2A.X foci resolution time course plus RAD51 foci in a KMT2A-rearranged line (MV4-11 or SEM) with and without 1 uM revumenib, plus or minus 2 Gy - two days of work with reagents every leukaemia laboratory has. Then a DR-GFP/EJ5-GFP HR-versus-NHEJ reporter pair under drug. If either moves, the combination schedules with TOP2 poisons in frontline protocols need a rationale they do not currently have.
exposed populationchildren with KMT2A-rearranged ALL or AML receiving a menin inhibitor concurrently with or sequentially after TOP2 poisons and anthracyclines; infants in particular, because cumulative exposure and remaining lifespan both run against them
absence classnever-measured
relationship to existing entries`exp-hr-competence-under-menin-inhibition-in-kmt2a-rearranged-blasts` (interactome10) already proposes the measuring experiment and already reasoned that the drug pocket 'is not obviously the Val371-Gln400 surface - so the prediction is no'. THIS BEAT CHANGES THAT PREMISE: revumenib is 4.3 A, ziftomenib 2.3 A and DS-1594b 3.1 A from Val371, and bleximenib is 10.7 A. That entry's arm 3 (pocket-occupancy-dominant versus degradation-dominant agent) is the right design and now has a structural reason to expect a compound-level difference within the pocket arm too. Reported to the coordinator as an amendment owed; not edited here.
affects
[
  "drug-revumenib",
  "drug-ziftomenib",
  "drug-bleximenib",
  "drug-enzomenib",
  "drug-etoposide",
  "drug-doxorubicin",
  "drug-daunorubicin",
  "drug-mitoxantrone"
]
modality comparison
{
  "pocket_inhibitor_removes": "the MBM cavity (KMT2A and JUND), plus contact with the acidic shelf at 363-371 for revumenib, ziftomenib and DS-1594b but NOT bleximenib",
  "degrader_removes": "all six mapped surfaces plus both nuclear localisation signals, in every cell in the body, for as long as it is dosed",
  "the_surfaces_only_a_degrader_removes": [
    "gamma-H2A.X motif 371-400 (HR)",
    "SvID 360-445 / D418, W436 (H3K9me3 delivery)",
    "nucleosome grip 353-362, 428-442, 556-587 (H3K79me2 reading)",
    "LEDGF patch 91-105, 209-212 (chromatin anchoring)",
    "SETD2 contact (NTD + Palm, no coordinates published)"
  ],
  "why_this_is_not_symmetric": "menin's tumour-suppressor functions are LOSS-of-protein phenotypes - MEN1 syndrome, the Men1-null lung carcinogenesis in the HR paper - so a degrader phenocopies the germline disease in a way a pocket inhibitor does not"
}
what has never been measured
[
  "gamma-H2A.X foci resolution under any CLINICAL menin inhibitor as a single agent, in any system",
  "menin-gamma-H2A.X co-immunoprecipitation under any menin inhibitor",
  "H3K36me3 at double-strand breaks under any menin inhibitor",
  "an HR-versus-NHEJ reporter pair under a clinical menin inhibitor as a single agent",
  "H3K9me3 under a PHARMACOLOGICAL menin inhibitor (QUEEN_BRIEF rule 72 records that the genetic arm has been measured and the pharmacological arm has not)"
]
what HAS been measured and bounds the absence
{
  "pmid": "31590682",
  "what": "chidamide (HDAC inhibitor) COMBINED with MI-3 (tool menin-MLL inhibitor) in human MLL-rearranged AML cells disrupts the DNA damage checkpoint at CHK1/CHK2 - not their upstream kinases ATR and ATM - and disrupts DNA repair likely via homologous recombination but not NHEJ",
  "why_it_does_not_close_the_question": "a two-drug synergy design at subtoxic single-agent doses, with an HDAC inhibitor that has its own documented DNA-damage effects, using a tool compound rather than a clinical menin inhibitor, and with no single-agent menin-inhibitor HR arm reported in the abstract",
  "read": "abstract only, fetched 2026-09-22",
  "already_in_this_graph": "exp-hr-competence-under-menin-inhibition-in-kmt2a-rearranged-blasts records this under PARTLY_ANSWERED_cycle_62"
}
How this could bite us back
  • The measurement that would settle this could show the risk is real, in which case the frontline combination schedules in this disease need a rationale they do not currently have; or it could show menin inhibition suppresses HR in the blast, in which case the combination is better than anyone argued and PARP inhibition becomes a rational partner. Both outcomes are worth more than the current silence.
  • A repair deficit shows up as a second malignancy decades later, which is the one endpoint no trial in this disease is powered for and the one that matters most in an infant.
  • This entry's own premise is vulnerable: the HR scaffold is one paper in lung and fibroblast systems and could fail to replicate in a blood cell. The correct response is to run the KMT2A-rearranged experiment, not to assume in either direction.
  • Degrader enthusiasm in this target class is driven by the resistance mutations in the pocket (M327I, G331R, T349M all have deposited structures), which a degrader would bypass. That argument is sound and this entry does not contradict it - it names the cost that argument does not currently price.
Still open
  1. Does revumenib change gamma-H2A.X foci resolution kinetics in a KMT2A-rearranged line? Two days of work, and it is the single measurement that would move this entry from speculation to a graded risk.
  2. Is impaired HR a therapeutic mechanism here rather than a toxicity? If menin inhibition suppresses HR in the blast, the combination with TOP2 poisons is rational rather than risky - and the same measurement answers both readings.
  3. Do the four clinical compounds differ? Bleximenib does not reach the acidic shelf and the other three do, which is a structurally specified prediction of a class difference that the trials will never be designed to detect.
  4. Would a menin degrader be tolerable at all? It phenocopies germline MEN1 loss across every tissue, and this graph already holds five risk entries on menin loss in beta cells, bone, cranial neural crest, endocrine late effects and multiorgan hyperplasia.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Menin scaffolds an H3K36me3-dependent homologous-recombination relay: Men1 ablation in mice causes spontaneous lung carcinogenesis with severe genomic instability (1,439 somatic alterations across 328 genes in six tumours, including 800 CNVs), menin recognises H2A.X through residues 371-400 with the interaction markedly promoted by DNA damage, menin recruits SETD2 to write H3K36me3 at damage sites, LEDGF reads it and recruits CtIP for resection activating RPA2/Rad51, and restoring H3K36me3 pharmacologically rescues impaired HR in Men1-deficient cells.

    The tumor suppressor menin is a key scaffold mediator for homologous recombination repair. · Lian A, Zhu X, Jin B, Gu G, Xu B, Zhang L, Lin X, Zhu X, Jin G · Nucleic Acids Res 2026
    PMID 42578372DOI
    What this evidence cannot show

    A549 lung adenocarcinoma, Men1-knockout MEFs and Men1-floxed mice. No haematopoietic cell, no B-lineage cell, no KMT2A fusion and no leukaemia appear in the study; the word 'leukemia' appears only in the introduction's framing of menin-MLL biology. Genetic ablation is not pharmacological inhibition, and the entire HR phenotype is a LOSS-OF-PROTEIN phenotype - which is precisely why it maps onto a degrader and not onto a pocket inhibitor without further evidence. Full text read from Europe PMC fullTextXML (96,530 characters) 2026-09-22.

    Evidence that argues against this (1)
    • The H3K36me3-dependent HR pathway was established without menin: SETD2-dependent H3K36me3 was shown to be required for HR in 2014 and LEDGF was shown to promote DNA-end resection and HR in 2012. So the relay exists independently of the menin arm, and a menin inhibitor that leaves menin protein in place may leave the relay intact even if genetic menin loss breaks it. Daugaard M et al. 2012
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    Of six menin surfaces with residue coordinates, the five measured menin ligands (revumenib, ziftomenib, bleximenib, DS-1594b, M-89) contact only the MBM cavity and, for three of the five, the acidic shelf at 363-371; the SvID's critical residues are 9.9-19.5 A away, the nucleosome grip residues are not contacted at all, and the LEDGF patch is 22.8-37.9 A away.

    What this evidence cannot show

    Geometry, not pharmacology. 'Not contacted' in a crystal does not establish 'not perturbed in a cell', and the LEDGF case shows why: the menin-KMT2A-LEDGF assembly is cooperative across 23-38 A. No co-IP of any of these partners has ever been run under a menin inhibitor. Computed 2026-09-22 from live RCSB mmCIF and live UniProt.

    Evidence that argues against this (1)
    • Menin inhibitors evict the menin complex from target chromatin, and this graph already records a mechanism by which removing menin from a locus removes its delivered partner without any change at the partner's binding site (PKA phosphorylation at Ser487 reduces SUV39H1 and HDAC1 at the Ins1 promoter). A surface can be spared and its function lost, which makes the surface map an incomplete guide to the pharmacology. UniProt Consortium 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PDB 7UJ4 (revumenib, 1.96 A), 8VA6 (ziftomenib, 1.57 A), 9WKU (bleximenib, 1.9 A) and 8IG0 (DS-1594b, 2.6 A) are the deposited co-structures of the clinical menin-inhibitor chemotypes, and menin resistance substitutions M327I, G331R, G331D and T349M all have deposited structures (9C4T, 9C4S, 9C4V, 9C4U, 8VA5) - so the pocket is the best-characterised part of menin by a wide margin and the other five surfaces have none of that attention.

    PDB 8VA6: Menin in complex with Ziftomenib (KO-539) · RCSB Protein Data Bank depositors · RCSB Protein Data Bank 2026
    What this evidence cannot show

    Entry titles, methods, resolutions and chemical-component identities fetched from data.rcsb.org on 2026-09-22. A PDB entry existing is not a structure suitable for design and an entry count tracks crystallisability and industrial attention, not biology - which is the point being made here: the asymmetry of structural coverage across menin's surfaces is a fact about the field's priorities.

    Evidence that argues against this (1)
    • There is no deposited structure of any menin degrader with menin, and no menin degrader has entered clinical trials in this disease, so the degrader arm of this risk is a statement about a modality rather than about an agent. Treating it as an imminent clinical choice would overstate it. RCSB Protein Data Bank 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The combination of the HDAC inhibitor chidamide with the tool menin-MLL interaction inhibitor MI-3 is synergistic in human MLL-rearranged AML cells in vitro and in vivo and disrupts the DNA damage checkpoint at CHK1 and CHK2 rather than at ATR and ATM, as well as DNA repair likely via homologous recombination but not non-homologous end joining - so a menin-directed agent has already been associated with impaired HR in an MLL-rearranged cell, in combination.

    PMID 31590682DOI
    What this evidence cannot show

    ABSTRACT ONLY, fetched 2026-09-22; the full text was not opened by this beat. Author list truncated at eight names as returned by efetch. This is a COMBINATION result at subtoxic single-agent doses with an HDAC inhibitor that has its own DNA-damage pharmacology, using MI-3 rather than a clinical compound, in adult MLL-r AML lines - so it does not isolate the menin-inhibitor contribution and it is not in a paediatric or B-lineage system. The HR statement is hedged in the abstract itself ('likely via homologous recombination').

    Evidence that argues against this (1)
    • The same abstract reports a sharp increase in ROS and loss of mitochondrial membrane potential, so an apparent repair defect in this design can be downstream of oxidative stress and apoptosis rather than a direct effect on a menin scaffold. A repair readout taken in dying cells is not a repair measurement. Lian A et al. 2026
Connected to