USP22 — a KMT2A-rearranged dependency that works through MEIS1 protein stability, and is not an interactor of the fusion

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• MEIS1 is required for normal definitive haematopoiesis and vascular development; an agent that destabilises MEIS1 protein systemically could carry haematopoietic and vascular toxicity that a transcriptional inhibitor targeted to leukemic chromatin would not.
• USP22 deubiquitinates histone H2B among other substrates, so a USP22 inhibitor is not a MEIS1-specific agent and its on-target effects extend well beyond this axis.
• A KMT2A-rearranged mean gene effect of -0.130 is a weak dependency in absolute terms. Resistance by simple incomplete target engagement is the most likely failure mode, before any acquired mechanism.
• If MEIS1 abundance is buffered by a second deubiquitinase, USP22 loss would be compensated; no paralogue redundancy test is reported.
• The mark cannot be used as a pharmacodynamic biomarker for a USP22 inhibitor in this disease, because the genetic and chemical perturbations move H2Bub in opposite directions (PMIDs 27132940, 34936860). Anyone dosing to an H2Bub readout would be titrating against a signal whose sign depends on the modality.
• The SAGA DUB module counteracts heterochromatin silencing and is an enhancer of position-effect variegation (PMID 18206972). Inhibiting it is therefore predicted to INCREASE heterochromatic silencing genome-wide - a global chromatin consequence that has nothing to do with the MEIS1 rationale and has never been assessed for these compounds.

1. Does USP22 contact KMT2A::AFF1 at all? A single co-IP from a KMT2A::AFF1 B-ALL line would settle whether this is a dependency entry or an interactome entry, and no published experiment answers it.
2. Does USP22 bind MEIS1 directly, or does it act on MEIS1 through the SAGA DUB module on chromatin? The ChIP co-occupancy is compatible with either.
3. Would USP22 inhibition retain activity against a MEN1 gatekeeper-mutant line? Acting on MEIS1 protein rather than MEIS1 transcription predicts yes, and that prediction is directly testable in the published revumenib-resistant models.
4. The DepMap KMT2A-rearranged group is AML-heavy while the discovery was in B-ALL. Is the selectivity larger in B-ALL lines specifically?
5. Which chemical USP22 inhibitor was used, and what is its selectivity across the DUB family?
6. Does USP22 loss change what a KMT2A-rearranged blast IS, or only how fast it grows? Every readout in the founding study is growth, protein stability or target-gene expression. An immunophenotype or differentiation readout after USP22 depletion would convert an inferred cell-state claim into a measured one.
7. USP22 stabilises MEIS1 protein while menin inhibitors act on the transcription of the same programme. Do the two act additively on MEIS1 protein level, and does USP22 depletion delay menin-inhibitor resistance? Both drugs converge on one output by different routes, which is the shape that usually produces synergy.
8. Does USP22 erase H2AK119ub in a leukaemic cell? The founding characterisation names H2A as well as H2B as substrates of the SAGA DUB module (PMID 18206972), and H2AK119ub is the mark this graph's PRC1.1/USP7 arm is built on - so two supposedly separate polycomb and SAGA arms of this graph may share a substrate and nobody has checked.
9. Is H2Bub altered at MEIS1-co-occupied loci when USP22 is depleted in a KMT2A-rearranged line? The ChIP establishes the address and the mark has never been measured there, so the chromatin half of the mechanism is entirely inferred.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   A genome-wide CRISPR/Cas9 screen in a KMT2A-rearranged B-ALL cell line identified the SAGA-complex deubiquitinase USP22 as a regulator of MEIS1 protein stability; USP22 depletion impaired growth and proliferation in KMT2A-rearranged acute leukemia models; ChIP revealed cooperative binding of USP22 and MEIS1 at oncogenic target genes; and genetic or chemical USP22 inhibition caused MEIS1 polyubiquitination, proteasomal degradation and downregulation of target-gene expression.

   USP22 is a novel vulnerability regulating MEIS1 protein abundance and gene transcription in KMT2Ar acute leukemia. · Bröchtel S, Schneider C, Müller M, Villinger C, Plenge T, Gupta MK, Kesel LI, Scheich S, Wolf S, Cakir AY, Häupl B, Buettner F, Knapp S, Serve H, Stegmaier K, Perner F, Oellerich T, Cremer A · Blood 2026
   PMID 42418681DOI
   ▸What this evidence cannot show

   **Abstract only** — full text not read. The abstract names one KMT2A-rearranged B-ALL line for the screen without specifying it, does not state which KMT2A fusion that line carries, and does not name the chemical inhibitor used. Critically, it reports no test of whether USP22 contacts a KMT2A fusion protein, so the classification of USP22 as a dependency rather than an interactor reflects what was shown, not a negative result. Cooperative ChIP occupancy is co-localisation on chromatin and is not evidence of a direct USP22-MEIS1 protein-protein interaction. **Resolved favourably:** the unspecified line is **SEM**, and GSE337426 records `genotype: KMT2A::AFF1` in a **structured field** - the only deposition in the whole sweep to state the index fusion explicitly - paired with MV4;11. **The inhibitor clause of the original caveat still stands.** Resolved 2026-09-19 by the Q18b unstated-systems sweep from the study's deposition, not from its text; bounded as a statement about the deposited data rather than about the whole study.

   ▸Evidence that argues against this (2)

   • The DepMap panel gives a weaker picture than a genome-wide screen hit suggests. Across 1,208 lines USP22 has a mean gene effect of +0.089 and is depleted in 0.25% of them, and even in KMT2A-rearranged lines the mean is only -0.130. A single-line CRISPR screen hit that validates in that line and its relatives can still be a small effect in absolute terms, and the absolute effect here is small. The selectivity is the interesting part; the magnitude is not. Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project 2026
   • One submitter-level caution carries over: in both lines the `chip antibody` fields are swapped between the MEIS1 and USP22 samples, so no claim should be built on that field. The genotype field, which is what this resolution uses, is unaffected. NCBI Gene Expression Omnibus 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

   USP22 has DepMap mean gene effect +0.0886 across 1,208 lines with 0.25% of lines depleted, mean -0.1295 in 11 KMT2A-rearranged leukemia lines against +0.0815 in 43 other leukemia lines, a difference of -0.211; MEN1 computed in the same lines has a difference of -0.866.

   DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API, with KMT2A-rearrangement status from Cellosaurus; per-gene recomputation for this beat using the method of analyses/depmap-kmt2a-selective-scan/run.py · Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project · Database and derived analysis 2026
   ▸What this evidence cannot show

   DepMap release version not stated by the API. Eleven KMT2A-rearranged lines against 43 other leukemia lines; this difference has not been permutation-tested in this entry. DepMap's KMT2A-rearranged group is dominated by AML and biphenotypic lines, so it is not a clean read-out for the B-ALL context in which USP22 was discovered.

   ▸Evidence that argues against this (1)

   • +0.0886 overall with 0.25% of lines depleted is a null; −0.1295 against +0.0815 is a 0.21 difference computed from eleven index lines, unsplit by lineage. This is the weakest class of evidence on which to build the H2Bub direction question the entry's beat turns on. Oncofusion knowledge graph (internal analysis) 2026
3. E_genetic_vs_chemical_directionstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Removing USP22 genetically and inhibiting it chemically move its own substrate mark in OPPOSITE directions. Atanassov 2016, in the authors' words: 'ablation of USP22, the enzymatic center of the DUBm, leads to a reduction, rather than an increase, in global H2Bub1 levels' - because ATXN7L3 redistributes to the SAGA-independent deubiquitinases USP27X and USP51. Morgan 2022, on a USP22-selective macrocycle: 'cells treated with peptide had increased levels of H2B monoubiquitination'. This entry's dependency evidence (E1) is a genome-wide CRISPR screen - a genetic depletion - so on the chromatin axis it models the opposite of what a drug would do.

   ATXN7L3 and ENY2 Coordinate Activity of Multiple H2B Deubiquitinases Important for Cellular Proliferation and Tumor Growth · Atanassov BS, Mohan RD, Lan X, et al. (Dent SYR, senior author) · Molecular Cell 62(4):558-71 2016
   PMID 27132940DOI
   ▸What this evidence cannot show

   Both abstracts fetched and read on 2026-09-19; the quoted sentences are verbatim. The divergence is mechanistically coherent rather than a contradiction - knockout removes a scaffold and frees ATXN7L3 for competing DUBs, while a catalytic inhibitor blocks turnover with the scaffold intact - so neither result is wrong. Note also that the KMT2A-specific paper behind E1 reports the MEIS1 effect for BOTH genetic and chemical inhibition, so the MEIS1 readout itself may be robust to this; what diverges is the H2Bub consequence and therefore everything downstream of the mark.

   ▸Evidence that argues against this (1)

   • This is a caution about preclinical validation, not a reason to drop the target. If the MEIS1 degradation phenotype holds under chemical inhibition - and the source paper says it does - then the therapeutic rationale survives intact. But the chromatin consequences do not transfer: a genetic model predicts LOWER H2Bub and a drug delivers HIGHER, which matters because this graph separately holds that raising H2Bub is what resolves bivalency and permits differentiation, while lowering it removes DOT1L's licence. A USP22 inhibitor may therefore be doing something useful for a reason its own genetic validation did not predict, and the H2Bub-direction experiment this graph proposes is the way to find out which. Morgan MT et al. 2022
4. E_morgan_macrocyclestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   USP22-selective cyclic peptide inhibitors exist and are cell-active: a screen of a combinatorial macrocycle library gave a top hit highly specific for USP22 against a panel of 44 other human DUBs, and treated cells showed increased H2B monoubiquitination.

   Potent macrocycle inhibitors of the human SAGA deubiquitinating module · Morgan MT, et al. · Cell Chemical Biology 29(4):544-554.e4 2022
   PMID 34936860DOI
   ▸What this evidence cannot show

   Abstract only. Selectivity is reported against a 44-DUB panel, which is not proteome-wide, and cell activity is demonstrated by the H2Bub readout rather than by an anti-leukaemic effect - these compounds have not been tested in KMT2A-rearranged cells.

   ▸Evidence that argues against this (1)

   • That these compounds exist is what makes the direction problem urgent rather than academic: the only selective cell-active USP22 chemistry available pushes H2Bub UP, so anyone acting on this entry's genetic rationale would be testing the opposite chromatin perturbation from the one the rationale was built on. Atanassov BS et al. 2016
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In KMT2A-rearranged acute leukaemia, USP22 is chromatin-bound and co-occupies critical oncogenic target genes with MEIS1 by chromatin immunoprecipitation, where it protects MEIS1 from polyubiquitination and proteasomal degradation; genetic depletion impairs growth of KMT2Ar acute leukaemia models.

   USP22 is a novel vulnerability regulating MEIS1 protein abundance and gene transcription in KMT2Ar acute leukemia. · Bröchtel S, Schneider C, Müller M, Villinger C, Plenge T, Gupta MK, Kesel LI, Scheich S, Wolf S, Cakir AY, Häupl B, Buettner F, Knapp S, Serve H, Stegmaier K, Perner F, Oellerich T, Cremer A · Blood 2026
   PMID 42418681DOI
   ▸What this evidence cannot show

   ChIP gives a genomic address, not a spatial compartment: it establishes that USP22 is on chromatin in the right disease and says nothing about which part of the nucleus. The primary screen was in a KMT2Ar B-ALL line, which is the correct lineage for this graph's index fusion - unusually good, since most of this graph's chromatin evidence is murine myeloid (brief rule 73). The abstract does not state the fusion partner of the screening line.

   ▸Evidence that argues against this (1)

   • USP22 is not exclusively nuclear. In non-small-cell lung cancer USP22 enhances mitochondrial biogenesis by stabilising MYBBP1A, which places a functional pool of the enzyme outside the SAGA/chromatin compartment. An entry that writes USP22 as a chromatin protein full stop will mispredict what a USP22 inhibitor does to a cell's metabolism. Sun Y et al. 2026
6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   A genome-wide CRISPR/Cas9 screen in a KMT2A-rearranged B-cell acute lymphoblastic leukaemia cell line identified the deubiquitinase USP22 as a regulator of MEIS1 protein stability; genetic depletion of USP22 impaired cellular growth and proliferation in KMT2A-rearranged acute leukaemia models, USP22 and MEIS1 bind cooperatively at critical oncogenic target genes, and genetic or chemical inhibition of USP22 led to polyubiquitination and proteasomal degradation of MEIS1 with downregulation of its target genes.

   USP22 is a novel vulnerability regulating MEIS1 protein abundance and gene transcription in KMT2Ar acute leukemia. · Bröchtel S, Schneider C, Müller M, Villinger C, Plenge T, Gupta MK, Kesel LI, Scheich S, Wolf S, Cakir AY, Häupl B, Buettner F, Knapp S, Serve H, Stegmaier K, Perner F, Oellerich T, Cremer A · Blood 2026
   PMID 42418681DOI
   ▸What this evidence cannot show

   Cell lines and cell-line-derived models; the specific KMT2A fusion partner and the identity of the lines are not given in the abstract, so this cannot be attributed to KMT2A::AFF1 specifically, and there is no infant or primary-patient arm. The readouts are growth, proliferation, protein stability and target-gene expression - there is no differentiation or immunophenotype measurement, so the cell-STATE reading is an inference from the MEIS1 programme. Read: abstract only.

   ▸Evidence that argues against this (1)

   • The requirement for USP22 in NORMAL haematopoiesis is not established and should not be assumed from the SAGA complex. An in vivo genome-wide CRISPR knockout screen in haematopoietic stem and progenitor cells identified Tada2b and Taf5l as the SAGA members regulating haematopoiesis, with loss causing a buildup of immature cells in the marrow, altered H3K9ac and H2Bub, interferon upregulation, reduced mitochondrial activity and increased megakaryocyte-progenitor commitment - and Usp22 was not among the hits reported. Writing a complex-level normal-tissue phenotype onto USP22 would be the error rule 64 exists to prevent. Shankar A et al. 2026
7. E_q97_epi1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The 2 MDa metazoan GCN5-HAT-containing TFTC/STAGA complexes contain a histone H2A and H2B deubiquitinase activity; three additional subunits - ATXN7L3, USP22 and ENY2 - form the deubiquitination module, and that module is an enhancer of position-effect variegation, counteracting heterochromatin silencing, as well as coactivating nuclear receptors.

   A TFTC/STAGA module mediates histone H2A and H2B deubiquitination, coactivates nuclear receptors, and counteracts heterochromatin silencing. · Mol Cell 2008
   PMID 18206972
   ▸What this evidence cannot show

   Abstract fetched and read in full 2026-09-21; full text not read. Human/Drosophila biochemistry and a position-effect-variegation assay - no haematopoietic system and no leukaemia (rule 73). The deubiquitinase activity is attributed to the MODULE (ATXN7L3 + USP22 + ENY2), not to USP22 in isolation, which is the correct reading and is consistent with PMID 27132940's later finding that ATXN7L3 is the shared adaptor. CITATION-REUSE NOTE: my dispatch brief stated this PMID was 'uncited corpus-wide'. It is cited three times - gene-atxn7l3 E5, gene-meis1 E11/E12, gene-trrap E5 - and git log -S shows all three entered in cycle 52 itself, in commits 88e9f91 (Q94) and 57c6447 (Q95). The premise was true when measured and went stale inside its own cycle.

   ▸Evidence that argues against this (1)

   • The module attribution is the counter-argument to reading USP22 as the enzyme for this mark. ATXN7L3 and ENY2 are obligate partners, ATXN7L3 is shared with USP27X and USP51, and this entry already records that removing USP22 genetically LOWERS H2Bub - the opposite of what a simple 'USP22 is the H2Bub eraser' model predicts. The correct statement is that the MODULE erases H2A/H2B ubiquitin and USP22 is its catalytic centre within a redistributable adaptor. Cancer Research 2016

• ← part ofKAT2A / GCN5 - the SAGA catalytic subunit that reaches this disease through its complex rather than through its own symbol, and whose only chemical tool degrades its paralogue too

  both are SAGA subunits - KAT2A in the acetyltransferase module, USP22 in the deubiquitinase module

• regulates →MEIS1 — rate-limiting for MLL leukemia stem cells, a surviving co-dependency, and embryonic-lethal when removed

  USP22 protects MEIS1 from proteasomal degradation; its inhibition polyubiquitinates MEIS1

• ← related toErasing H2Bub: three deubiquitinases share two adaptors, and removing the enzyme the field names lowers the mark instead of raising it

  Counter-evidence (Morgan 2022, PMID 34936860 against Atanassov 2016, PMID 27132940): selective cell-active USP22 macrocycles RAISE H2Bub while genetic USP22 ablation LOWERS it, so only one of the two can be a clean report of USP22's contribution to the mark, and the adaptor-competition model reconciles them only on an assumption neither paper tests

• related to →Erasing H2Bub: three deubiquitinases share two adaptors, and removing the enzyme the field names lowers the mark instead of raising it

  Counter-evidence (Broechtel 2026, PMID 42418681): in KMT2A-rearranged leukaemia, genetic or chemical USP22 inhibition polyubiquitinates MEIS1 and degrades it, impairing growth - so in this disease the compound acts dominantly through a non-histone substrate and H2Bub-level reasoning about it is confounded

• refutes →What this beat looked at and did not write: rejected candidates, with counts for both halves of every search

  Counter-evidence (PubMed eutils re-run 2026-09-19): intersecting USP22[Title/Abstract] with this ledger's own fixed KMT2A term set returns PMID 42418681, a primary KMT2A-specific paper on a SAGA H2B deubiquitinase that the ledger's H2Bub term set missed - and that this graph already cites as gene-usp22 E1. The claim of two primary papers is withdrawn; it is at least three

• related to →Raise H2Bub in KMT2A-rearranged cells with the existing USP22 macrocycles and see which way the disease moves

  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

• regulates →H2BK120ub (H2Bub1) — the ubiquitin tag on histone H2B that licenses H3K79 methylation, and the mark this graph was missing

  catalytic subunit of the SAGA DUB module; ablation lowers rather than raises the mark

• ← bindsATXN7L3 — the shared adaptor that actually sets H2Bub levels, and a KMT2A-selectivity hit nobody in this field has looked at

  adaptor of the SAGA DUB module, shared with USP27X and USP51

• ← testsRaise H2Bub in KMT2A-rearranged cells with the existing USP22 macrocycles and see which way the disease moves

  chemical versus genetic perturbation of USP22 move H2Bub in opposite directions

• regulates →MEIS1 — rate-limiting for MLL leukemia stem cells, a surviving co-dependency, and embryonic-lethal when removed

  deubiquitinates MEIS1; loss causes polyubiquitination and proteasomal degradation

• depends on →Infant acute lymphoblastic leukaemia (ALL diagnosed under 1 year of age)

  genome-wide CRISPR hit in a KMT2A-rearranged B-ALL line; DepMap selectivity -0.211

• related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  downstream dependency, NOT a demonstrated interactor: no published test of binding to any KMT2A fusion

• part of →SGF29 — a SAGA reader with a Tudor pocket, intermediate essentiality, and the best unexploited profile in this beat

  both are SAGA subunits — SGF29 the reader module, USP22 the DUB module

• related to →Revumenib (SNDX-5613, Revuforj) — first approved menin–KMT2A interface inhibitor

  proposed as a post-translational route to the same MEIS1 output, motivated by menin-inhibitor resistance

• ← related toUSP2 (ubiquitin carboxyl-terminal hydrolase 2) - the ninth commonest KMT2A partner and the only one that is a deubiquitinase

  DIFFERENT GENE, easily confused: USP2 = O75604, 605 aa, 11q23.3, KMT2A fusion partner; USP22 = Q9UPT9, 525 aa, 17p11.2, SAGA DUB module, not a fusion partner. Opposite DepMap verdicts in the same analysis.

• ← testsLineage-split DepMap: neither USP2 nor MAML2 is a dependency in KMT2A-rearranged lines, and USP22 is

  positive control and disambiguation test: USP22 IS depleted in both halves (-1.79 / -0.76), the opposite verdict to USP2 in the same run

• related to →The soluble-nucleoplasm versus chromatin-bound axis - the one localisation readout that keeps a size dimension, and nobody has run it on a KMT2A fusion

  USP22 is chromatin-bound in KMT2A-rearranged leukaemia by ChIP; the soluble fraction has not been measured

• regulates →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia

  USP22 protects MEIS1 from proteasomal degradation and co-occupies oncogenic targets with it in KMT2A-rearranged B-ALL; inferred from the MEIS1 programme, no differentiation readout in the source

• ← depends onMEIS1 — rate-limiting for MLL leukemia stem cells, a surviving co-dependency, and embryonic-lethal when removed

  MEIS1 protein abundance depends on USP22, the catalytic subunit of the SAGA H2A/H2B DEUBIQUITINATION module - so MEIS1 is a non-histone substrate of a histone-directed enzyme. Whether the H2Bub activity is involved in the MEIS1 effect is untested; gene-usp22 carries no epigenetic_readout and does not cite the founding SAGA-DUB characterisation (PMID 18206972).

• ← part ofATXN7L3 — the shared adaptor that actually sets H2Bub levels, and a KMT2A-selectivity hit nobody in this field has looked at

  DUB-MODULE assignment, kept separate from the acetyltransferase side: ATXN7L3 + USP22 + ENY2 form the deubiquitination module of TFTC/STAGA, an enhancer of position-effect variegation in Drosophila whose activity counteracts heterochromatin silencing (PMID 18206972). KAT2A, SGF29 and TRRAP are the HAT side. The modules have different substrates and different compartment evidence (Q95 2026-09-22)

• regulates →H2BK120ub (H2Bub1) — the ubiquitin tag on histone H2B that licenses H3K79 methylation, and the mark this graph was missing

  catalytic centre of the SAGA/TFTC-STAGA DUB module, which deubiquitinates histone H2A and H2B; direction of the cellular effect differs between genetic and chemical loss

• part of →ATXN7L3 — the shared adaptor that actually sets H2Bub levels, and a KMT2A-selectivity hit nobody in this field has looked at

  ATXN7L3, USP22 and ENY2 form the TFTC/STAGA deubiquitination module

• related to →H2AK119ub - the mark every PRC1.1 claim in this graph rests on, named in fifteen entries and written in none until now

  the module's substrate list includes histone H2A, so the SAGA eraser and this graph's PRC1.1 writer may share a substrate - untested in leukaemia

• ← related toThe unnamed KMT2A-rearranged B-ALL line is SEM, and the deposition says KMT2A::AFF1 in a genotype field

  resolves the unnamed KMT2A-rearranged B-ALL line to SEM with an explicit KMT2A::AFF1 genotype field