The class I and class II arms of menin inhibition are different mechanisms - the class I effect was tested against viral mimicry and NF-kB and excluded both

If the two arms were one mechanism, then one biomarker, one combination partner and one resistance mechanism would serve both. They are not, and they do not:

The last row is the one that has a patient in it. An MLLT10-rearranged leukaemia has a published rationale for a JAK inhibitor (gene-jak1), and a JAK inhibitor would be expected to remove the interferon-dependent arm and leave the chromatin arm running.

The exclusions are single-cell-line CRISPR depletions in K-562, reported with immunoblot confirmation 1. Three limits, stated plainly:

• They test necessity, not presence. In a cell receiving interferon, STAT1 signalling still raises class I; the finding is that the menin-inhibitor effect does not require it. Both routes operate together in any real setting.
• They are one cell line. No KMT2A-rearranged cell was tested, and neither was any cell in which HERVs had been shown to be induced. Fetsch's HERV induction was measured in AML; Sparbier's exclusion was measured in CML. The two papers may simply be describing two different cells.
• They do not exclude a HERV contribution to class I in a cell where HERVs are strongly induced. A cell can have an MDA-5-independent chromatin route and an MDA-5-dependent interferon route at once, and the relative sizes would be cell-dependent.

What survives is narrow and useful: nobody may write, in this graph, that menin inhibition raises MHC class I by viral mimicry. The one experiment that tested it says otherwise.

Fetsch reports interferon-stimulated gene upregulation as the mechanism of enhanced antigen presentation in KMT2A-rearranged AML 3. Sparbier reports that antigen presentation rises after menin targeting without STAT1 or MDA-5, in K-562 1. Hogeling reports the class II arm as CIITA-dependent and MEIS1-independent, with menin bound at the CIITA locus, and says nothing about interferon signalling as the route 2.

Three papers, three proposed mechanisms, one drug class, overlapping phenotypes. That is not a defect in any of them; it is an unresolved question that has never been posed as one. The discriminating experiment is small: menin inhibitor in a KMT2A-rearranged line, with and without STAT1 knockout, reading class I and class II separately and HERV transcripts alongside. It is written up as exp-mhc-class-i-and-nk-ligand-timecourse-under-menin-inhibition-kmt2a-r.

hyp-menin-inhibition-restores-gvl-only-where-mhc-ii-is-limiting (Q53-transplant, same cycle) argues that because MHC class II is constitutively expressed in the B lineage, the antigen-presentation arm of the Fetsch mechanism has little headroom in B-precursor KMT2A::AFF1 infant ALL, so any post-transplant benefit there must come from the T-cell-intrinsic arm alone. This beat's class II material bears on it in three ways, and they do not all point the same way.

1. Concord on the conclusion. Nothing here contradicts the prediction that MHC-II headroom is small in B-precursor disease. This beat independently flagged the mirror-image concern: the bleximenib class II result was obtained in myelomonocytic cells (MV4-11, MOLM13, primary AML) which express class II constitutively, and no source in this beat reports a lineage-matched control for it 2. Both beats are pointing at the same missing control from opposite ends.

2. Contest on the argument, with a named discriminator. 'Constitutively expressed' is a statement about a flow gate, not about a ceiling - and the bleximenib data show the distinction matters. MV4-11 is constitutively class II-positive and the drug still raised HLA-DR dose-dependently in it, while CIITA-null MV4-11 had significantly lower baseline HLA-DR and did not respond 2. So a constitutively class II-positive cell demonstrably retained headroom. The premise 'B-lineage cells are HLA-DR-positive, therefore there is nothing to gain' does not follow on its own; what would settle it is a quantitative baseline against the achievable maximum - an isotype-referenced MFI and a CIITA-saturating positive control such as IFN-gamma - not a positive/negative call. That measurement is in exp-mhc-ii-headroom-under-menin-inhibition-in-b-precursor-infant-all, and this note is a request that it be read quantitatively rather than as a gate.

3. It does not transfer to class I, and for class I the sign may be opposite. The headroom argument is scoped to class II and rests on constitutive B-lineage expression. Class I is expressed by essentially every nucleated cell, so a naive headroom argument would predict no class I benefit anywhere - yet the class I effect is exactly what PMID 36635503 and PMID 39704147 measure, because the relevant variable is not whether the gene is on but whether its promoter is polycomb-restrained 1. Baseline class I on a KMT2A-rearranged blast has never been measured, so headroom is unknown in both directions.

And the asymmetry matters therapeutically. If class II headroom is absent, the conclusion is nothing to gain. If class I headroom is absent because class I is already high - which is what the MLL-AF9 NK-dependency data suggest (mechanism-mhc-class-i-is-an-nk-brake-in-kmt2a-fusion-leukaemia) - the conclusion is something to lose. Two hypotheses with the same shape and opposite consequences, and only one measurement separates them.

Added after the King's join of this graph onto Zhou's published 9,292-locus menin/H2AK119ub target set in KMT2A-rearranged AML (PMID 39158067).

This cuts two ways at once, and both are useful.

It supports this entry's central claim that the class I and class II arms are different mechanisms - and it supplies independent, orthogonal evidence for it, from a chromatin target list rather than from a depletion experiment.

But it inverts the directionality this entry assumed. The table above was built expecting class II to be the direct arm (menin at CIITA by ChIP-qPCR) and class I to be the indirect one (no menin at MHC-I promoters). Zhou's set says the reverse at the transactivator level: class II's transactivator is menin-free and polycomb-marked, class I's is menin-bound. If that holds, the honest formulation is that both arms are indirect with respect to the presented genes, and the class I arm may have a second, direct route through NLRC5 that no published model includes.

Caveats that keep this from being a conclusion. Occupancy is not function - Sparbier found VTP50469 produced remarkably few significant expression changes despite globally reducing menin occupancy, so a menin-solo NLRC5 need not move under the drug. The system is adult myeloid KMT2A-rearranged AML. And absence from Zhou's called set is absence from a thresholded list, not demonstrated absence of binding. The test is one qPCR: NLRC5 transcript under a menin inhibitor. It has never been run.