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

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.

• Atxn7l3-null mouse embryos are already abnormal at E7.5 - earlier than Usp22-null embryos - so the node with the better selectivity signal is also the one with the more severe developmental requirement.
• ATXN7L3 licenses three enzymes with many non-histone substrates; inhibiting it is not a clean H2Bub intervention.
• The direction of benefit is unknown. Raising H2Bub is as easily a pro-leukaemic intervention as an anti-leukaemic one on the evidence this beat assembled, because H2Bub licenses the very DOT1L activity the disease runs on.
• A rank-290 hit from 11 AML-weighted lines has, in this project's own history, repeatedly failed to survive better annotation.

1. Is the rank-290 selectivity real? The cheapest test is an ATXN7L3 knockout in matched KMT2A-rearranged and non-rearranged B-ALL lines with the same guides, reading out proliferation and H2Bub.
2. Why does ENY2 show nothing when ATXN7L3 shows a top-2% signal, if they are partners in the same module? Either the module model is wrong, or one of the two DepMap measurements is.
3. Does raising H2Bub by ATXN7L3 loss help or harm a KMT2A-rearranged cell? Nothing in this beat answers the direction question, and it is the question on which the whole therapeutic reading turns.
4. Has anyone ever measured H2Bub at HOXA9/MEIS1 after ATXN7L3 depletion? Global westerns dominate this literature and the locus-level behaviour is unknown.

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

   ATXN7L3 and ENY2 coordinate the activity of multiple H2B deubiquitinases: depletion of ATXN7L3 or ENY2 increases global H2Bub whereas ablation of the catalytic USP22 reduces it, and USP27X and USP51 are H2Bub deubiquitinases that function independently of SAGA and compete with USP22 for ATXN7L3 and ENY2 for activity.

   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 2016
   PMID 27132940DOI
   ▸What this evidence cannot show

   Abstract read in full; full text not read. Not leukaemia, not KMT2A-rearranged. Global bulk H2Bub measurements only.

   ▸Evidence that argues against this (1)

   • Depleting ATXN7L3 raising global H2Bub while depleting USP22 does not is evidence that ATXN7L3 serves several DUBs, and it also means ATXN7L3 loss is not a clean probe of any one of them. Any phenotype attributed to the SAGA DUB module through ATXN7L3 depletion is attributable to the whole ATXN7L3-dependent DUB set. Molecular Cell 2016
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Atxn7l3-/- mouse embryos show developmental delay already at E7.5 while Usp22-/- embryos are normal at that stage and die at E14.5; global H2Bub is strongly increased in Atxn7l3-null embryos and derived cell lines but only slightly affected in Usp22-null embryos; and the strong H2Bub increase in Atxn7l3-/- mESCs and MEFs does not correspond to more than modest changes in Pol II occupancy and shows no change in Pol II elongation.

   Histone H2Bub1 deubiquitylation is essential for mouse development, but does not regulate global RNA polymerase II transcription. · Wang F, El-Saafin F, Ye T, Stierle M, Negroni L, Durik M, Fischer V, Devys D, Vincent SD, Tora L · Cell Death Differ 2021
   PMID 33731875DOI
   ▸What this evidence cannot show

   Abstract read in full; full text not read. Mouse development; no leukaemia. The negative elongation finding is global, and the authors state the ATXN7L3-related DUB activity regulates only a subset of genes.

   ▸Evidence that argues against this (1)

   • These are germline developmental phenotypes in mouse embryos, and the entry uses them to rank importance among DUB components. Embryonic lethality timing measures how early a gene becomes indispensable in development, which is a poor proxy for whether a leukaemic cell needs it — and the same paper reports that H2Bub1 deubiquitylation is essential for development WITHOUT regulating global Pol II transcription, which undercuts the transcriptional reading. Cell Death & Differentiation 2021
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

   In the project's genome-wide DepMap KMT2A-selectivity scan, ATXN7L3 has a KMT2A-rearranged mean gene effect of -0.446 against -0.258 in other leukaemia lines, a difference of -0.188, rank 290 of 17,728, on a pan-essentiality of -0.336 across 1,208 lines; its module partner ENY2 shows a difference of +0.015 at rank 11,080 on a pan-essentiality of -0.313.

   DepMap CRISPR (Chronos) genome-wide KMT2A-selectivity scan of 17,728 protein-coding genes, 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines vs 43 other leukaemia lines · Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project (analyses/depmap-kmt2a-selective-scan-full) · Database and derived analysis 2026
   ▸What this evidence cannot show

   Eleven KMT2A-rearranged lines, AML-weighted, none infant-derived; the scan has not been permutation-tested for this gene. ENY2's null result in the same module is an unexplained internal inconsistency and is the main reason to discount the ATXN7L3 signal.

   ▸Evidence that argues against this (1)

   • ENY2, the obligate partner adaptor in the same SAGA DUB module, shows no KMT2A selectivity whatsoever (difference +0.015, rank 11,080 of 17,728). If the ATXN7L3 signal reflects a real requirement of the module in KMT2A-rearranged cells, ENY2 should show it too. Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project 2026
4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   A controlled PubMed eutils search on 2026-09-19 found no indexed literature connecting ATXN7L3 to MLL or KMT2A leukaemia; ATXN7L3 appears in the SAGA/USP22 literature and in a developmental-disorder report, and the USP22-SAGA search that surfaced it returned 42 records of which none concerns KMT2A-rearranged leukaemia apart from the MEIS1-stability paper already held in this graph as gene-usp22.

   NCBI Entrez Programming Utilities (esearch/efetch), PubMed, searches run 2026-09-19 with [Title/Abstract] field tags · US National Library of Medicine; searches run by this project · Database 2026
   ▸What this evidence cannot show

   This absence was established from a USP22/SAGA-anchored search rather than an ATXN7L3-anchored one with both halves counted, so it is weaker than the negative search reported in `gene-ube2a`. Treat as 'not found', not as 'does not exist'.

   ▸Evidence that argues against this (1)

   • A counted absence is only as good as its symbol set, and ATXN7L3 publishes under SCA7-related and SAGA-component descriptions as well as its approved symbol. This project has repeatedly found that bounded, well-formed absence claims are the ones that turn out wrong, and the search as described does not record which aliases were covered. Molecular Cell 2016
5. E_q95_atxn7l3_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   ATXN7L3's UniProt nuclear call has two evidence codes of very different worth and the experimental one is about the PARALOGUE: Q14CW9 carries 'Nucleus' on ECO:0000255 HAMAP-Rule MF_03047, a family rule rather than a measurement, plus ECO:0000269 PubMed 27601583 - a paper whose subject is ATXN7L3B, which shares 74% identity with the N-terminal region of ATXN7L3, and whose finding is that ATXN7L3B LOCALIZES IN THE CYTOPLASM while competing with ATXN7L3 for ENY2 binding.

   UniProtKB Q14CW9 (AT7L3_HUMAN), SUBCELLULAR LOCATION comment with evidence codes and sources plus the feature table, fetched live 2026-09-22 · UniProt Consortium · Databases 2026
   ▸What this evidence cannot show

   A statement about the database record. ECO:0000255 with a HAMAP rule is the weakest evidence class this graph records for a compartment - the same class as EP400's PROSITE-ProRule nuclear call in tranche 5 - and here it sits beside an experimental code borrowed from work on a different gene product.

   ▸Evidence that argues against this (1)

   • AND THIS IS A RULE-91 INVERSION: THE APPARENTLY MISDIRECTED CITATION IS THE BEST ONE IN THE RECORD. A paper about the paralogue is exactly the right source for this compartment, because the paper's whole point is a compartment CONTRAST - ATXN7L3B is cytoplasmic and cannot support DUB activity toward H2Bub1, ATXN7L3 is nuclear and can, and the two compete for ENY2, so overexpressing the cytoplasmic one RAISES H2Bub1 while overexpressing the nuclear one lowers it. A curator citing it for ATXN7L3's nucleus is citing a paper that establishes the nuclear/cytoplasmic split as the functional variable. The defect that remains is the other code: a HAMAP family rule contributes nothing and makes the annotation look doubly supported. Li W et al. 2016
6. E_q95_atxn7l3_loc2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   ATXN7L3's functional address is the SAGA DEUBIQUITINASE MODULE and its documented chromatin effect is on heterochromatin rather than on a compartment: ATXN7L3, USP22 and ENY2 are the three subunits forming the deubiquitination module of the 2 MDa GCN5-containing TFTC/STAGA complexes, that module is an enhancer of position-effect variegation in Drosophila, it is required as a cofactor for full transcriptional activity by nuclear receptors, and its deubiquitinase activity is necessary to COUNTERACT HETEROCHROMATIN SILENCING.

   A TFTC/STAGA module mediates histone H2A and H2B deubiquitination, coactivates nuclear receptors, and counteracts heterochromatin silencing. · Zhao Y, Lang G, Ito S, Bonnet J, Metzger E, Sawatsubashi S, Suzuki E, Le Guezennec X, Stunnenberg HG, Krasnov A, Georgieva SG, Schüle R, Takeyama K, Kato S, Tora L, Devys D · Mol Cell 2008
   PMID 18206972DOI
   ▸What this evidence cannot show

   Position-effect variegation is a genetic readout in Drosophila and 'counteracting heterochromatin silencing' is a functional statement, not a localisation - nobody has imaged ATXN7L3 at heterochromatin or anywhere else in any organism. The module assignment is biochemical and is solid; the compartment inside the nucleus is unmeasured.

   ▸Evidence that argues against this (1)

   • THE MODULE ASSIGNMENT IS THE THING THAT HAS TO BE KEPT CLEAN, BECAUSE THREE NEIGHBOURING ENTRIES IN THIS GRAPH SIT ON THE OTHER SIDE OF THE SAME COMPLEX AND ONE OF THEM HAS THE SAME KIND OF EVIDENCE PROBLEM. SGF29 (tranche 4) is a tandem-Tudor H3K4me2/3 reader whose human nuclear call is ECO:0000250 from RAT; KAT2A (tranche 5) is the acetyltransferase; TRRAP (this tranche) is the scaffold whose nuclear residence needs GCN5. All three are the HAT side. ATXN7L3 is the DUB side, with USP22 and ENY2, acting on H2Bub1 rather than on histone acetylation - and this graph also holds the WRITER side of the same mark, RNF20/RNF40 with the adaptor WAC, amended in this same tranche. So ATXN7L3 and WAC are the eraser-adaptor and writer-adaptor of ONE mark and neither has a compartment measurement; that is the coherent gap, and it would be invisible if SAGA were treated as a single object. Zhao Y et al. 2008
7. E_q95_atxn7l3_loc3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   Bounded 2026-09-22: ATXN7L3 has no compartment measurement in any system beyond the paralogue-contrast paper, and its entire literature is 28 records. The gene clause returns 28; with a localisation clause 4; with a subnuclear clause 1, which is the TFTC/STAGA DUB-module paper and contains no compartment; against the nine-term KMT2A/MLL fusion clause 1, and with a localisation or subnuclear clause added, 0. The paralogue ATXN7L3B returns 10 records and the DUB-partner query (USP22, ENY2, H2Bub, deubiquitin*) returns 14.

   NCBI E-utilities esearch db=pubmed; every count stored with the querytranslation read back, in data.localization_bounded_absence · NCBI E-utilities · Databases 2026
   ▸What this evidence cannot show

   Titles and abstracts only; supporting figures not searched (rule 80). At n = 28 this is one of the smallest literatures in the whole corpus, so the zeros are cheap to trust and equally cheap to overturn - a single paper would change every number here. Retired or alternate symbols were checked and there are none: UniProt lists no gene synonyms for ATXN7L3 at all, which is unusual and removes the alias route entirely.

   ▸Evidence that argues against this (1)

   • THE ABSENCE IS REAL AND THE ENTRY'S OWN HEADLINE MAKES IT MORE INTERESTING THAN A TYPICAL ZERO, WHICH IS ALSO AN ARGUMENT FOR CAUTION. This entry is titled for being a KMT2A-selectivity hit nobody in this field has looked at, and it is the shared adaptor that actually sets H2Bub levels - so the compartment question here is not decorative: if ATXN7L3 sets the global H2Bub level and H2Bub is deposited co-transcriptionally on gene bodies, then the quantity that matters is occupancy along genes, and a nuclear-versus-cytoplasmic answer would not address it. The measurement to ask for is an ATXN7L3 or USP22 ChIP in a KMT2A-rearranged cell, not an immunofluorescence. Recording a localisation absence and calling it the gap would point the next beat at the wrong experiment. Li W et al. 2016

• 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 (Wang 2021, PMID 33731875): the strong global H2Bub increase in Atxn7l3-null mouse ES cells and MEFs changed Pol II occupancy only modestly and Pol II elongation not at all, so raising the mark may be transcriptionally near-silent and leave the direction argument with no readout

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

  limiting adaptor for three H2Bub deubiquitinases; its loss strongly raises the mark

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

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

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

  ATXN7L3/ENY2 are the rheostat rather than the catalytic subunit

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

  shared adaptor of the SAGA DUB module

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

  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)

• related to →WAC — the adaptor that hands RNF20/RNF40 to RNA polymerase II, and the only member of this axis whose knockdown alone killed MLL-rearranged ALL cells

  ERASER-ADAPTOR AND WRITER-ADAPTOR OF ONE MARK, and neither has a compartment measurement. ATXN7L3 is the SAGA DUB-module adaptor that sets H2Bub1 levels; WAC is the adaptor that hands RNF20/RNF40 to Pol II to deposit it. Both amended in this tranche; WAC has nuclear speckles by marker colocalisation, ATXN7L3 has nothing subnuclear at all (1 record, containing no compartment) (Q95 2026-09-22)

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

  Same complex, opposite modules, and comparable evidence weakness: SGF29's human nuclear call is ECO:0000250 from RAT (tranche 4), ATXN7L3's carries a HAMAP FAMILY RULE plus one ECO:0000269 that is a paper about the paralogue ATXN7L3B. In ATXN7L3's case the apparently misdirected citation is the better one - the paper's point is that ATXN7L3B is cytoplasmic and ATXN7L3 nuclear, and they compete for ENY2 (Q95 2026-09-22)

• ← part ofUSP22 — a KMT2A-rearranged dependency that works through MEIS1 protein stability, and is not an interactor of the fusion

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