Oncofusion

start / graph / exp-catalytic-dead-versus-degrader-test-of-class-iia-hdac-function-in-kmt2a-r-all

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q39-acetylation-cycle

Catalytic-dead versus degrader: a four-arm test that resolves the HDAC7 direction conflict and supplies the pharmacodynamic marker this drug class has never had

also called the class IIa scaffold test · HDAC7 catalytic-dead rescue · PD-resolved HDAC experiment

Two groups disagree about whether to add or remove HDAC7 in infant leukaemia. This experiment settles it by separating two things that are always confused: whether the enzyme's chemistry matters, or whether the protein just needs to be physically present. It also adds the one measurement that every HDAC drug study in this disease has skipped - checking whether the drug actually did anything to the chromatin it was supposed to.

The same thing, for a scientist

A four-arm perturbation of HDAC7 in KMT2A::AFF1 infant ALL cells, each arm separating **scaffold from catalysis**, read out with a **locus-resolved acetylation panel** that has never been run in this disease. Arms: (1) **shRNA/CRISPR knockout** - removes protein and catalysis; (2) **re-expression of wild-type HDAC7** in a low-HDAC7 t(4;11) line - restores both; (3) **re-expression of a catalytically dead HDAC7** (the class IIa catalytic tyrosine already substituted to histidine in the human enzyme makes this construct straightforward) - restores scaffold only; (4) **LMK-235 or a class IIa degrader** - blocks catalysis only, or removes scaffold only, respectively. **The decisive comparison is arm 2 versus arm 3.** If catalytic-dead HDAC7 reproduces wild-type's differentiation and anti-proliferative benefit, the function is scaffolding and `hyp-the-class-iia-hdac-direction-conflict-is-scaffold-versus-catalysis` is supported; if it fails, that hypothesis is dead and class IIa catalysis is real. **The readout is the second half of the value and is independent of the hypothesis.** Every arm is read with ChIP or CUT&RUN for **H3K9ac, H3K27ac and H2A.Z acetylation at KMT2A::AFF1 target loci** (HOXA9, MEIS1, PROM1, and the paper's own HDAC7 locus), plus MEF2D occupancy, plus global histone acetylation. **No paper in this disease has measured acetylation at a fusion target locus under any HDAC perturbation** - the writer half was measured at a locus by the Tip60 work and the eraser half never was. That panel, once it exists, is the pharmacodynamic assay that all four published HDAC mechanisms and both registered trials lacked.

Go deeper
Design

Cells. Two KMT2A::AFF1 infant ALL lines (e.g. SEM, RS4;11) plus at least one KMT2A-rearranged line with a different partner, and - this is the arm this graph's standing rules require - a lineage-matched non-KMT2A B-ALL comparator. Without it the experiment cannot distinguish a fusion-directed effect from a B-lineage one, which is the failure mechanism-kat8-mof-requirement-is-myeloid-not-fusion-specific exists to prevent and which every class IIa paper in this disease has committed.

Arms, each versus its own control:

armprotein presentcatalysiswhat it isolates
1. CRISPR knockoutnonototal loss
2. WT re-expression in a low-HDAC7 lineyesyesfull restoration
3. Catalytic-dead re-expressionyesnoscaffold alone
4a. LMK-235yesblockedcatalysis alone
4b. class IIa degradernonoscaffold removal, pharmacologically

Readouts, in ascending cost:

  1. Viability, apoptosis, cell cycle, differentiation markers (CD19, CD15, myeloid switch).
  2. Locus-resolved acetylation: CUT&RUN for H3K9ac, H3K27ac and acetyl-H2A.Z at HOXA9, MEIS1, PROM1 and HDAC7 itself; plus global histone acetylation western.
  3. MEF2D CUT&RUN, the adaptor prediction's direct test.
  4. RNA-seq, to ask whether the affected programme is the fusion's or the lineage's.
  5. In vivo: arms 1, 3 and 4b in a KMT2A::AFF1 PDX, with the PD panel sampled at trough.

Controls that this beat's literature shows are needed: a class I selectivity check on LMK-235 at the concentrations used (global histone hyperacetylation would reveal off-target class I engagement and would explain its in vitro activity without any class IIa catalysis); and a target-engagement measurement in vivo, whose absence is why three published class IIa failures cannot be interpreted.

Why to read PMID 40722046 before running any of this

The single cheapest step is not an experiment. The de Barrios paper's HDAC7 rescue experiments may already have used a catalytic-dead construct, and the abstract does not say. The full text is open access at PMC12305908 and was not read in this beat.

  • If a catalytic-dead construct was used and worked, arm 3 is already done and the hypothesis is substantially supported before anyone pipettes anything.
  • If it was used and failed, the hypothesis is dead and the experiment should not be run in this form.
  • If wild-type only was used, the experiment stands as designed.

This is recorded as a step rather than assumed away because this project's standing rule is that an absence of testing is not a negative result, and because reading one paper is cheaper than a four-arm cell-line study by several orders of magnitude.

Second-cheapest step, also not an experiment: ask whether any class IIa HDAC degrader exists and has been characterised in any cell type. Arm 4b is only runnable if one does.

Facts and figures
designFour-arm perturbation of HDAC7 in KMT2A::AFF1 infant ALL lines separating scaffold from catalysis - (1) CRISPR knockout, (2) wild-type re-expression, (3) catalytic-dead re-expression, (4a) LMK-235 catalytic inhibition, (4b) class IIa degrader - with a lineage-matched non-KMT2A B-ALL comparator arm.
decisive comparisonarm 2 (wild-type re-expression) versus arm 3 (catalytic-dead re-expression)
readoutViability, apoptosis, cell cycle and differentiation markers; CUT&RUN for H3K9ac, H3K27ac and acetyl-H2A.Z at HOXA9, MEIS1, PROM1 and the HDAC7 locus; MEF2D CUT&RUN; global histone acetylation western; RNA-seq; and in vivo arms in a KMT2A::AFF1 PDX with the PD panel sampled at trough.
cost tiermed
timeline18 months for the in vitro arms; 24-30 months including the PDX arms
step zeroRead the full text of PMID 40722046 (PMC12305908) to determine whether the published HDAC7 rescue construct was catalytically active. This costs one afternoon and may settle arm 3 before anything is run.
step zero point fiveEstablish whether any characterised class IIa HDAC degrader exists; arm 4b is contingent on it.
independent valueThe locus-resolved acetylation panel is worth running regardless of how the scaffold question resolves: no paper in this disease has measured acetylation at a KMT2A-fusion target locus under any HDAC perturbation, and all four published HDAC-inhibitor mechanisms and both registered trials lacked any pharmacodynamic marker.
tests hypothesishyp-the-class-iia-hdac-direction-conflict-is-scaffold-versus-catalysis
controls
[
  "lineage-matched non-KMT2A B-ALL comparator - the arm every published class IIa study in this disease omitted",
  "class I selectivity check on LMK-235 at its in vitro active concentrations (global histone hyperacetylation would reveal off-target class I engagement)",
  "in vivo target engagement, whose absence makes the three published class IIa failures uninterpretable",
  "a KMT2A-rearranged line with a non-AFF1 partner, to separate fusion-general from AFF1-specific"
]
what it settles
[
  "whether class IIa function in this disease is scaffolding or catalysis",
  "whether the two opposing HDAC7 reports are reconcilable",
  "whether the three in vivo class IIa failures were the wrong modality or the wrong target",
  "whether any HDAC perturbation reaches acetylation at fusion target loci at all"
]
How this could bite us back
  • If this experiment supports the scaffold hypothesis, the therapeutic implication is a class IIa degrader in infants - removing a MEF2 adaptor from a child whose cardiac, skeletal-muscle and neuronal MEF2 programmes are all active. The experiment would make the biology clearer and the safety case harder.
  • Arm 3 is the whole experiment, and a catalytic-dead class IIa construct is not as clean as it sounds: the human class IIa catalytic site is already substituted relative to class I, so 'catalytic-dead' means further disabling an enzyme that is arguably already dead, and a negative result in that arm could mean the construct was misfolded rather than that catalysis matters.
  • The lineage-matched comparator arm is the one most likely to be dropped for cost, and it is the one without which the result repeats the exact error that has made every other class IIa paper in this disease uninterpretable for fusion specificity.
  • Even a complete positive result establishes a mechanism, not a therapy. The three class IIa compounds that exist have already failed in vivo here, and a degrader that does not yet exist is a medicinal-chemistry programme, not a next step.
Still open
  1. Does a characterised class IIa HDAC degrader exist? Arm 4b - the arm that would convert this from a mechanistic study into a therapeutic one - depends entirely on the answer.
  2. Which KMT2A::AFF1 target loci should anchor the acetylation panel? HOXA9 and MEIS1 are the obvious choices but are myeloid-weighted; the lymphoid index disease may need PROM1 and the B-lineage programme instead.
  3. Should the comparator arm be a non-KMT2A infant B-ALL or a non-KMT2A paediatric B-ALL? Infant material is scarce and the choice trades statistical power against the lineage-and-age confound this graph keeps meeting.
  4. If catalytic-dead HDAC7 rescues, what is the scaffold actually recruiting in a B cell - NCoR/SMRT-HDAC3, or something else? That would become the next target rather than HDAC7 itself.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Restoring HDAC7 expression in t(4;11) infant B-ALL - via the menin inhibitor MI-538, enhanced by chidamide - drives leukaemic pro-B cells toward a more differentiated state and impairs aberrant proliferation in an HDAC7-dependent manner, with HDAC7 epigenetically silenced by EZH2 and the KMT2A::AFF1 fusion protein and low HDAC7 associated with poor therapeutic response.

    HDAC7 induction combined with standard-of-care chemotherapy provides a therapeutic advantage in t(4;11) infant B-cell acute lymphoblastic leukemia. · de Barrios O, Ocon-Gabarro I, Gusi-Vives M, Collazo O, Meler A, Romecin PA, et al. · Biomarker Research 13(1):99 2025
    PMID 40722046DOI
    What this evidence cannot show

    The abstract does not state whether the HDAC7 rescue construct was catalytically active, which is precisely the detail this experiment turns on; the full text (PMC12305908) was not read in this beat and should be read first. The effect is stated to be only partially HDAC7-mediated. MI-538 is a tool compound.

    Evidence that argues against this (1)
    • The opposing result - shRNA knock-down of HDAC4/5/7 inducing apoptosis and cell-cycle arrest in KMT2A-rearranged ALL cells - is the reason this experiment is worth running, and either arm of the conflict could turn out to be the one that does not generalise. Verbeek TCAI et al. 2024
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Three chemically distinct class IIA-selective inhibitors - LMK-235, MC1568 and TMP195 - failed to induce anti-leukaemic effects at the maximum achievable dose in a KMT2A-rearranged ALL xenograft model despite LMK-235 eradicating KMT2A-rearranged ALL cell lines and primary patient samples in vitro, and LMK-235 with venetoclax did not prolong survival.

    Selective inhibition of HDAC class IIA as therapeutic intervention for KMT2A-rearranged acute lymphoblastic leukemia. · Verbeek TCAI, Vrenken KS, Arentsen-Peters STCJM, Castro PG, van de Ven M, van Tellingen O, et al. · Communications Biology 7(1):1257 2024
    PMID 39362994DOI
    What this evidence cannot show

    No target-engagement measurement is reported in vivo, which is exactly why the failure is uninterpretable and why this experiment includes one. The authors' own attribution - compound stability and efficiency - is plausible and untested. Abstract read; PMCID PMC11450098, full text not fetched.

    Evidence that argues against this (1)
    • Three chemically distinct inhibitors failing together is stronger evidence than one failing, but all three are class IIA-selective ACTIVE-SITE inhibitors, so the shared failure is equally consistent with class IIA HDACs mattering through a non-catalytic scaffold function that an active-site inhibitor cannot touch. That is precisely why this experiment proposes a degrader arm, and the negative therefore motivates the design rather than settling it. de Barrios O et al. 2025
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Tip60 is recruited by MLL-AF10 and MLL-ENL to the Hoxa9 locus where it acetylates H2A.Z, establishing that fusion-target-locus-resolved acetylation measurement is technically feasible in this disease and has been done for the writer half.

    Tip60 activates Hoxa9 and Meis1 expression through acetylation of H2A.Z, promoting MLL-AF10 and MLL-ENL acute myeloid leukemia. · Yamagata K, Shino M, Aikawa Y, Fujita S, Kitabayashi I · Leukemia 35(10):2840-2853 2021
    PMID 33967269DOI
    What this evidence cannot show

    Myeloid mouse models (MLL-AF10, MLL-ENL) and not the lymphoid index disease, so the feasibility transfers but the biology does not. Cited here as a methods precedent - acetyl-H2A.Z ChIP at a fusion target locus is an assay that has been performed - rather than as a result about KMT2A::AFF1. Abstract read; no PMCID reported by efetch.

    Evidence that argues against this (1)
    • That locus-resolved acetylation measurement is technically feasible in MLL-AF10 and MLL-ENL does not make it feasible in the index t(4;11) B-lineage system: both are myeloid fusions with different partner biology, and neither experiment was done in infant cells, where cell number is the binding constraint on any ChIP-based readout. Verbeek TCAI et al. 2024
Connected to