Human BAP1 is UniProt Q92560, HGNC:950. The review preserves all 96 source annotation objects, including evidence, references, supporting entities and the source term label on the parkin-localization row. Decisions comprise 44 ACCEPT, 47 KEEP_AS_NON_CORE, four MODIFY and one UNDECIDED. No NEW annotation is proposed. Two core functions distinguish histone from nonhistone deubiquitination.
The generic protein-binding assertions are retained as non-core when the partner relationship is supported. Co-purification is not described as proof of an isolated binary interaction. The BAP1–BRAF assertion from PMID:35512704 remains undecided because its exact construct, variant and pair-level assay were not resolved. This is an evidence-access limit, not a claim that the curator assigned the wrong gene. The P38398-5 supporting entity on the original BRCA1 interaction is a partner isoform, not a BAP1 isoform.
The four refinements specify cysteine-type deubiquitinase activity instead of generic peptidase activity, transcriptional regulation instead of two broad expression terms, and protein deubiquitination instead of protein modification. The InterPro-derived ubiquitin-dependent protein-catabolism assertion is retained as non-core. UCH-family membership alone does not establish target-protein breakdown, but HCFC1 accumulation after BAP1 depletion supports a contextual turnover role. The authors' proposed connection to proteasomal degradation remains tentative. One supported context can justify a process annotation; BAP1 need not execute degradation generally. This evidence-based retention supersedes the preliminary over-annotation proposal.
The original PR-DUB study directly reconstituted both human BAP1–ASXL1 and fly Calypso–Asx. Its human enzyme removes nucleosomal H2A ubiquitin, while the Hox-repression genetics are fly experiments. A paper centered on a fly complex therefore provides valid human catalytic evidence. PMID:20436459(https://pubmed.ncbi.nlm.nih.gov/20436459/)
Purified human BAP1 is stimulated by ASXL DEUBAD domains. Nucleosomal H2AK119 ubiquitin is distinguished from H2AK13/15 ubiquitin; isolated peptide assays do not preserve all nucleosome specificity. Earlier recruitment and oligomerization measurements should not be generalized beyond their constructs and conditions. PMID:26739236(https://pubmed.ncbi.nlm.nih.gov/26739236/)
The later human BAP1–ASXL1 substrate-bound structure resolves DNA, histone and ubiquitin contacts. The structural substrate contains Xenopus histones, while the enzyme is human; separate biochemical assays test cleavable substrates. The resolved 1:1 assembly is not imposed as the universal stoichiometry of every native complex. The earlier bidentate study contains a fly structure and direct human biochemical assays, with the human oligomer modeled rather than structurally resolved. PMID:37556531(https://pubmed.ncbi.nlm.nih.gov/37556531/), PMID:30258054(https://pubmed.ncbi.nlm.nih.gov/30258054/)
BAP1-mediated H2A deubiquitination has context-dependent transcriptional outcomes. Human HAP1 knockout and catalytic rescue support protection of active genes from PRC1-dependent silencing. Tagged interaction profiling in that study used HeLa cells; the HAP1 genetic and chromatography experiments are distinct. FOXK2-target studies separately support repression and phosphorylation-dependent recruitment. These observations motivate a direction-neutral transcription-regulation core. PMID:30664650(https://pubmed.ncbi.nlm.nih.gov/30664650/), PMID:25451922(https://pubmed.ncbi.nlm.nih.gov/25451922/), PMID:24748658(https://pubmed.ncbi.nlm.nih.gov/24748658/)
Local PAINT data contain positive ancestral assertions for BAP1-family catalytic activity and heterochromatin formation. The latter is retained as non-core; the complete family topology/MSA was not reconstructed. Calypso grounding and a short donor list are not reasons to reject an IBA. Human BAP1 appearing among experimental descendants is valid grounding, not circular evidence.
HCFC1 is an established BAP1 partner and substrate. The HCF-binding motif and HCFC1-N/Kelch experiments must be distinguished from assays of HCFC1-C in another study. WT versus C91A and linkage-restricted ubiquitin support K48-linked substrate deubiquitination in human cells; they do not establish unrestricted cleavage of every free K48 chain. Substrate stabilization is not a universal consequence. PMID:19815555(https://pubmed.ncbi.nlm.nih.gov/19815555/), PMID:19188440(https://pubmed.ncbi.nlm.nih.gov/19188440/)
An ER-associated BAP1 pool binds and deubiquitinates ITPR3, supporting receptor abundance and calcium-dependent cell-death responses. The original work includes primary human fibroblasts/mesothelial cells and transfected HEK293 assays. Its biochemical reaction uses immunopurified proteins; minimal isolated binary sufficiency is not claimed. BAP1 is neither the calcium channel nor an integral ER membrane protein. The nonhistone core's nuclear location concerns HCFC1, while its ER-membrane location concerns ITPR3. PMID:28614305(https://pubmed.ncbi.nlm.nih.gov/28614305/)
Wild-type BAP1 is principally nuclear with a regulated cytoplasmic pool. Localization-signal mutants, UBE2O-dependent monoubiquitination, BAP1 autodeubiquitination and TNPO1-dependent import help distinguish catalytic competence from substrate access. BAP1 is import cargo, not the transport receptor. PMID:18757409(https://pubmed.ncbi.nlm.nih.gov/18757409/), PMID:24703950(https://pubmed.ncbi.nlm.nih.gov/24703950/), PMID:35446349(https://pubmed.ncbi.nlm.nih.gov/35446349/)
BRCA1/BARD1 interaction and inhibition of its ligase activity are distinct from deubiquitination of either component. The newer Reactome disruption event explicitly records catalytic-independent inhibition; the broader chain-removal statement in the older event is not used to invent a substrate assignment. All eight source Reactome summaries were read; their nucleoplasmic source annotations agree with direct localization evidence. PMID:19117993(https://pubmed.ncbi.nlm.nih.gov/19117993/), Reactome R-HSA-9700998
Mouse Bap1 loss supports embryonic and hematopoietic phenotypes. Orthology-derived lineage, proliferation and homeostasis annotations remain contextual, rather than separate human catalytic functions. The correct identity is PMID:22878500, DOI 10.1126/science.1221711, PMCID PMC5201002; an earlier scratch fetch receipt mistyped the DOI suffix. Official PubMed and PMC agree on the corrected identity. The indexed abstract and Figures 1–4 captions were read. PMID:22878500(https://pubmed.ncbi.nlm.nih.gov/22878500/)
The parkin-recruitment screen includes BAP1 among reconfirmed candidates in publisher Extended Data Figure 5. This supports the retained recruitment phenotype without making BAP1 a mitochondrial import factor. The source GO identifier and obsolete-labeled term were preserved; no replacement was guessed. Germline neurodevelopmental evidence likewise does not directly establish every developmental GO process transferred from mouse. PMID:24270810(https://pubmed.ncbi.nlm.nih.gov/24270810/), PMID:35051358(https://pubmed.ncbi.nlm.nih.gov/35051358/)
Several caches are abstract-only. Existing PMID:33961781 and PMID:35512704 caches contain incomplete article sections despite their full-text flags; neither flag was equated with a complete-paper read. Pair-level supplementary interaction tables were not independently inspected. Reference reviews in the YAML record source-specific boundaries. Primary Results/Methods and figure captions were read selectively where needed; no raw sequencing, microscopy images, cryo-EM maps or entire supplement reanalysis is claimed.
The normal source-fetch attempt failed before writing gene files. Verified recovery supplied the original machine-generated sources. The default Falcon research attempt and configured perplexity-lite fallback could not start because the required offline dependency was unavailable; neither produced a research report. This document records manual research, and no provider-labeled report was manufactured. An independent annotation reviewer assessed 30 rows and checked the remaining 66 plus the two-core plan. Source files and existing publication caches were preserved.
The BRCA1–BARD1 interaction experiments support ubiquitin ligase inhibitor
activity (GO:1990948). PMID:19117993 reports inhibition of ligase-dependent
ubiquitination even with catalytic-mutant BAP1. Its indexed publisher Methods
and Discussion were read separately from the normal abstract-only cache; the
complete paper, figures and supplements were not inspected. Three generic
binding rows with canonical BRCA1 or BARD1 are refined using that evidence,
including the older PMID:9528852 BRCA1 interaction with the later paper as an
additional reference. The BRCA1-5 partner row remains non-core because the
particular partner isoform was not established in those inhibition assays.
Other supported generic interactions retain their non-core classifications.
The nonhistone catalytic core now uses endoplasmic reticulum (GO:0005783).
PMID:28614305 reports fractionation, immunofluorescence and immunogold evidence
for an ER-associated pool, and an artificially ER-targeted BAP1 construct.
Those selected Results and Methods support the organelle assignment; they do
not require assigning endogenous BAP1 to the ER lipid bilayer. Lack of integral
membrane topology alone would not exclude peripheral membrane association.
The HCFC1 details in the protein-turnover and K48-deubiquitination reasons were
checked against original indexed Results, Figure 5 caption and Discussion in
PMID:19188440(https://pmc.ncbi.nlm.nih.gov/articles/PMC2663315/), with relevant
Methods also inspected. The K48-restricted ubiquitin/WT-versus-C91A experiment
and precursor/processed-HCFC1 accumulation are present in those passages.
Proteasomal turnover remains the authors' interpretation, not a universal
mechanism demonstrated for all BAP1 substrates. The normal cache remains
abstract-only; article images and supplements were not read.
The gene-expression replacement is a reassignment to a regulatory process,
not a purported is_a refinement. The original source IDs are unchanged.
The official GO mature-cell differentiation pattern maps GO:0043363 to
CL:0000562, nucleate erythrocyte. This is a resulting mature cell identity, not
merely a nucleated intermediate. The adult-mouse hematology Results and
relevant Methods in the normal full-text cache of PMID:22878500 describe
anemia and erythroid dysplasia, including increased nucleated erythroid cells.
That evidence supports generalizing the electronic annotation to erythrocyte
differentiation (GO:0030218). It does not establish production of the mature
nucleate cell type or a direct human differentiation mechanism. The primary
identity is DOI 10.1126/science.1221711, PMCID PMC5201002. The complete paper
and supplements were not re-read for this bounded follow-up.
The parkin-recruitment phenotype remains non-core; the source's obsolete-
labelled GO:1903749 identifier and label are preserved. No new process or
molecular-function assertion is added. The proposal preserves all 96 source
objects and all 39 references, with 44 ACCEPT, 43 non-core, eight MODIFY and
one UNDECIDED; the two catalytic cores remain distinct.
The 15 remaining generic protein-binding annotations are deliberately retained
as KEEP_AS_NON_CORE under the user-supplied ActionEnum. That instruction
reserves REMOVE for an annotation unlikely to be correct on the combined
evidence; it permits KEEP_AS_NON_CORE for a supported annotation that does not
represent a core function. This takes precedence over the general review
recommendation to remove uninformative protein-binding terms. These rows
retain evidence-supported interaction context, not an additional core
molecular function or a claim of isolated binary binding from every assay.
A supported, more specific replacement has not been established for each
remaining exact assay, partner and partner-isoform assertion. Assigning an
adaptor or inhibitor activity solely from a physical association would add a
functional claim the experiment does not establish. In contrast, three
canonical BRCA1/BARD1 binding rows were already changed to MODIFY because
independent inhibition experiments support GO:1990948. The BRCA1-5 partner
row retains its explicit isoform-assay limit. The unresolved BRAF assertion
remains UNDECIDED and is separate from these 15 retained interactions. The
remaining generic-binding advisories are therefore acknowledged consequences
of this deliberate instruction precedence, rather than unreviewed rows.
The earlier notes sentence assigning the nonhistone core an ER-membrane
location is superseded by the first follow-up and this clarification. The
current core uses endoplasmic reticulum (GO:0005783), reflecting the
ER-associated BAP1 pool without requiring localization to the lipid bilayer.
This notes-only clarification changes no annotation, core function, reference,
source object or prior history record.