Raise H2Bub in KMT2A-rearranged cells with the existing USP22 macrocycles and see which way the disease moves
also called USP22 macrocycle dose-response in SEM/MV4;11 · H2Bub direction test
Selective molecules that raise the H2B ubiquitin tag already exist and get into cells. Put them on KMT2A-fusion leukaemia lines, measure the tag, the downstream mark, the leukaemia genes and whether the cells grow or mature, and the direction question is answered.
The same thing, for a scientist
A direct test of `hyp-h2bub-direction-is-unknown-in-kmt2a` using published tool compounds rather than new chemistry. Morgan et al. reported macrocyclic cyclic-peptide USP22 inhibitors, selective against a 44-DUB panel, cell-permeable, and shown to raise cellular H2B monoubiquitination [E1]. Treat SEM (KMT2A::AFF1 B-ALL) and MV4;11 with a dose range, with a non-KMT2A B-ALL line and a normal CD34+ progenitor as comparators, and read out in causal order: H2BK120ub by western and by CUT&RUN; H3K79me2 at HOXA9 and MEIS1 and genome-wide; fusion target-gene expression; differentiation markers; proliferation and apoptosis. Because genetic USP22 ablation is reported to *lower* global H2Bub while chemical inhibition raises it [E2], the experiment must include a USP22 knockout arm — the discrepancy between the two perturbations is itself informative and would otherwise confound interpretation.
Why this is the highest value-per-unit-effort experiment in the beat
Everything else this beat proposes needs a reagent that does not exist: a separation-of-function RTF1 allele, a CUT&RUN-grade H2BK34ub antibody, a degron knock-in. This one needs compounds that have been published, cell lines this field already keeps, and assays that are routine.
It also resolves the beat's central ambiguity in a single readout. The three outcomes map cleanly:
| Outcome | Reading |
|---|---|
| H2Bub up, H3K79me2 at HOXA9/MEIS1 up, proliferation unchanged or increased | the 'lower' model is right and USP22 inhibition is contraindicated in this disease |
| H2Bub up, differentiation markers induced, proliferation down despite H3K79me2 up | the 'raise' model is right; bivalency resolution dominates and this is a candidate strategy |
| H2Bub up, nothing else moves | the H2Bub layer is not rate-limiting in KMT2A-rearranged cells and this beat's whole premise is a chromatin footnote - a legitimate and publishable negative |
The third outcome is the one the graph most needs, because the beat's premise currently rests on pathway position rather than on measurement in this disease.
What bounds this experiment, added in the 2026-09-19 coverage repair. The outcome table above assumes the phenotype reports on H2Bub. It will not. Broechtel et al. show that in KMT2A-rearranged leukaemia, genetic or chemical USP22 inhibition polyubiquitinates MEIS1 and degrades it, downregulating its targets and impairing growth (PMID 42418681, counter-evidence on 1) - so the proliferation and differentiation rows of that table will move regardless of which way the mark goes, and only the chromatin read-outs stay interpretable. Second, the available pharmacology is one-directional: every USP22 tool compound raises H2Bub, so the 'lower H2Bub' arm of the hypothesis is untestable here. The agent that lowers it in this disease is panobinostat, via the RNF20/RNF40/WAC axis (PMID 28690313, counter-evidence on 2) - and the graph's own trial entry records NCT00723203, phase 2 panobinostat in relapsed/refractory ALL/AML, terminated for lack of efficacy, which is already a partial answer on the downward arm. A genuine direction test needs both arms and a MEIS1-protein read-out as a covariate.
| design | Dose-response treatment of SEM (KMT2A::AFF1 B-ALL) and MV4;11 with published macrocyclic USP22 inhibitors, alongside a non-KMT2A B-ALL line and normal CD34+ progenitors, with parallel USP22 knockout and ATXN7L3 knockdown arms. |
|---|---|
| readout | H2BK120ub (western, CUT&RUN); H3K79me2 at HOXA9/MEIS1 and genome-wide; fusion target-gene expression by RNA-seq; differentiation markers by flow cytometry; proliferation and apoptosis. |
| controls | Scrambled macrocycle; non-KMT2A B-ALL line; normal CD34+ progenitors; genetic USP22 knockout arm to expose the chemical-versus-genetic discrepancy; ATXN7L3 knockdown as the orthogonal way to raise H2Bub. Added 2026-09-19: MEIS1 protein level by western in every arm, as the covariate that would otherwise be mistaken for an H2Bub-direction effect (PMID 42418681); and a downward arm - panobinostat or RNF20/WAC knockdown - because every USP22 tool compound moves H2Bub upward only. |
| cost tier | low |
| timeline | 6-10 weeks once compound is in hand. |
| prerequisite | Establish that the macrocycles can be obtained. This beat did not verify availability and the experiment is otherwise blocked. |
| bounded conclusion | as designed the experiment tests only the upward direction, and its growth readout is driven by MEIS1 degradation rather than by H2Bub level |
tests hypotheses
[ "hyp-h2bub-direction-is-unknown-in-kmt2a" ]
- If the 'lower' model is right, this experiment will show a compound class making KMT2A-rearranged cells worse - which is the point of running it before anyone proposes USP22 inhibition in this disease, and the reason the graph's existing gene-usp22 rationale should not be acted on without it.
- Every USP22 tool compound raises H2Bub, so a design built on them cannot test the half of the hypothesis that says lowering the mark helps - and the one clinical test of the downward arm, panobinostat in NCT00723203, was terminated for lack of efficacy.
- Can any arm of this design attribute a phenotype to H2Bub? USP22 inhibition degrades MEIS1 in KMT2Ar cells (PMID 42418681), so the compound has a large, direction-independent effect on the very transcription factor the read-outs are anchored on.
- Are the macrocycles available? If not, ATXN7L3 knockdown raises H2Bub by a route the same paper validates and is a complete substitute for the direction question, at the cost of losing the pharmacological framing.
- Should primary infant KMT2A::AFF1 blasts be included? They are the right material and the hardest to get; a cell-line answer that does not replicate in primary cells would be the usual outcome of skipping them.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Macrocyclic cyclic-peptide inhibitors of USP22 identified from a combinatorial library are potent, highly specific for USP22 against a panel of 44 other human DUBs, enter human cells, and increase cellular levels of H2B monoubiquitination.
Potent macrocycle inhibitors of the human SAGA deubiquitinating module. · Morgan M, Ikenoue T, Suga H, Wolberger C · Cell Chem Biol 2022What this evidence cannot show
Abstract read in full; full text not read. No potency value, no pharmacokinetics, no in vivo data and no leukaemia data. Whether the compounds are obtainable - synthesised, shared by the authors, or commercially available - was not established by this beat and is the first thing to check.
Evidence that argues against this (1)
- The compounds bound the experiment to one direction, and the graph's own USP22 paper says the phenotype will come from somewhere else anyway. Broechtel et al. found, from a genome-wide CRISPR screen in a KMT2A-rearranged B-ALL line, that USP22 regulates MEIS1 protein stability and that GENETIC OR CHEMICAL USP22 inhibition polyubiquitinates MEIS1, sends it for proteasomal degradation and downregulates its target genes, with impaired growth in KMT2Ar models. So a macrocycle will slow these cells whichever way H2Bub moves, and the proliferation and differentiation read-outs in this design cannot be attributed to the mark. Only the chromatin read-outs (H2BK120ub and H3K79me2 at HOXA9/MEIS1) remain interpretable, and even those sit downstream of a transcription factor the compound is destroying. Brochtel S et al. 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Genetic ablation of USP22 reduces rather than increases global H2Bub, because USP27X and USP51 compete with USP22 for the shared ATXN7L3 and ENY2 adaptors; depletion of ATXN7L3 or ENY2 increases H2Bub.
ATXN7L3 and ENY2 Coordinate Activity of Multiple H2B Deubiquitinases Important for Cellular Proliferation and Tumor Growth. · Atanassov BS, Mohan RD, Lan X, Kuang X, Lu Y, Lin K, McIvor E, Li W, Zhang Y, Florens L, Byrum SD, Mackintosh SG, Calhoun-Davis T, Koutelou E, Wang L, et al. · Mol Cell 2016What this evidence cannot show
Abstract read in full; full text not read. Not leukaemia. The genetic-versus-chemical discrepancy may reflect the difference between removing a protein and inhibiting its catalysis, which is why both arms are required in this design.
Evidence that argues against this (1)
- A one-directional pharmacology cannot test a two-directional question, and the design omits the agent that moves the mark the other way. Every available USP22 tool compound RAISES H2Bub, so the 'lower H2Bub' arm of `hyp-h2bub-direction-is-unknown-in-kmt2a` is untestable in this design. The one clinically available agent shown to DEPLETE H2Bub in MLL-rearranged ALL is panobinostat, through suppression of the RNF20/RNF40/WAC axis, with WAC knockdown phenocopying loss of H2B ubiquitination and causing cell death - and the graph's own trial entry records NCT00723203, phase 2 panobinostat in relapsed/refractory ALL/AML, terminated for lack of efficacy. A direction test that omits the downward arm, and the one clinical result already available on it, answers half the question. Garrido Castro P et al. 2018
Design objection (Broechtel 2026, PMID 42418681): genetic or chemical USP22 inhibition degrades MEIS1 in KMT2A-rearranged cells, so the proliferation and differentiation read-outs will move whichever way H2Bub goes and cannot be attributed to the mark; MEIS1 protein must be measured as a covariate in every arm
Design objection (Garrido Castro 2018, PMID 28690313): every USP22 tool compound raises H2Bub, so the downward arm of the direction hypothesis is untestable in this design; the agent that lowers the mark in MLL-rearranged ALL is panobinostat via the RNF20/RNF40/WAC axis, and NCT00723203 - phase 2 panobinostat in relapsed/refractory ALL/AML - was terminated for lack of efficacy, which is already a partial answer on that arm
dose-response with published selective USP22 macrocycles in KMT2A-rearranged lines
chemical versus genetic perturbation of USP22 move H2Bub in opposite directions