DOCK2 is a hematopoietic RAC guanine-nucleotide exchange factor. The human DOCK2–ELMO1 structure/biochemistry and patient-cell phenotypes support one integrated exchange and RAC-regulation unit, with cytosol as a positively supported functional pool. ELMO1 regulates accessibility of the catalytic machinery: removing its N-terminal region increases exchange in a purified assay but has a different effect on productive signaling in cells. These are assay-dependent regulatory observations, not a second catalytic activity. See PMID:32651375(https://pubmed.ncbi.nlm.nih.gov/32651375/) and PMID:26083206(https://pubmed.ncbi.nlm.nih.gov/26083206/).
The official alternative products are 1830 and 336 residues. The short product lacks the canonical C-terminal DHR2 region; its molecular activity remains unresolved rather than demonstrated absent. Historical full-length constructs and the 431-residue catalytic fragment do not establish equal activity of both natural products. The human 2YIN deposition is GECGD followed by canonical residues 1192–1622. The full-length 6TGC/6TGB sequences contain 524 unresolved positions; all 1306 known residues match, but this is not an exact all-residue sequence check. No unresolved position is treated as a mutation.
GO:0005096 requires acceleration of GTP hydrolysis. The imported electronic DHR2 inference conflates that activity with nucleotide exchange, despite the same InterPro record describing the GEF mechanism. Removing this mapping leaves all supported exchange annotations intact. The mouse T-cell paper supports RAC activation and synapse phenotypes; its inaccessible full text is not characterized as a wrong-gene experiment. The broad signaling annotations are refined to the performed positive RAC-regulation step, with no additional NEW process.
The source record extract preserves 46 literal interaction assertions and 142 matching curated records. The two older RAC1 papers have positive MINT records, including a DOCK2 1115–1680 binding-associated fragment and a separately unspecified construct taxon. They are not rejected because their abstracts foreground DOCK180. Full original access remains limited.
HI-II-14 and HuRI support binary associations with sequence-confirmed reference reagents. They do not establish a native cellular pathway, coexpression or stoichiometry for every pair. Established partner classes justify specific binding terms for enzymes, an E3 ligase, a DNA-binding transcription factor, a signaling scaffold and keratins. No partner activity is assigned to DOCK2 itself. Keratin monomer binding is not automatically binding to an assembled intermediate filament.
The full-length MTUS2 partner and short Q5JR59-3 product are distinct claims: the short sequence lacks the mapped MAPRE1/KIF2C regulatory interaction regions, so its positive association remains generic. PCSK5-2 retains its peptidase domain and catalytic residues, but the two-hybrid assay does not test processing or activity in the secretory pathway. GOLGA6L9 and TENM4 separately carry sufficient-binding-region features; their reference-sequence checksums are not proof of full-length reagents.
Positive generic interactions are retained outside the core where a narrower justified class is not established. This applies the explicit ClinGen project binding instruction. Different partner identities and tested products are different biological claims; a source-specific limit does not negate DOCK2 protein binding in general.
The actual PAINT target lineage reaches the IBD nodes for exchange, small-GTPase binding, cytoplasm, plasma membrane, migration and myoblast fusion. The experimental donor records and exact IBD assertions were checked. A short donor list is not weak evidence, and the target's own experimental annotation is legitimate grounding. Conserved myoblast-fusion evidence is retained contextually without relabeling zebrafish/fly experiments as human DOCK2 assays.
The recovered 2011 original Figure 6 and uptake Methods show positive mouse Dock2 loss/rescue evidence for soluble-antigen and particle uptake. The 2012 addendum limits the proposed ASC-to-Dock2 relationship to some ASC-null mouse lines and leaves the cause unresolved. It explicitly retains the DOCK2 uptake/rescue findings. No particular passenger mutation, retraction or invalidation of every DOCK2 result is inferred. The reference assessment flags the disputed general upstream interpretation while preserving the direct observations.
Reactome resolves the exact human Q92608 participant, including residues 1–1830 and compartment. DOCK2 is a hasMember of a CandidateSet in the RAC2 exchange model, not an experimentally measured RAC2 reaction merely by virtue of model membership. The cytosol annotation is positively supported by the participant record and is not exclusive of regulated membrane association.
The granule-lumen and extracellular assertions have exact target-containing Reactome paths. They are retained as curated contextual pools, with explicit limits because the primary target-proteomics table was not recovered. Jurkat F-actin colocalization supports a cytoskeletal pool but not direct actin binding. The membrane-enriched YTS screen is compatible with peripheral association. The separate B-cell-exosome assertion remains UNDECIDED because the target-specific table could not be recovered; no contamination or misattribution is claimed.
All 77 seeded source dictionaries, including reference IDs, evidence codes, partners and qualifiers, remain unchanged. No NEW annotations are proposed. The review has 9 ACCEPT, 50 KEEP_AS_NON_CORE, 16 MODIFY, one REMOVE and one UNDECIDED decision. One integrated core and both official products are represented. Provisional DRAFT reflects retained generic-binding assertions; final status is determined by normal repository checks.
Normal isolated fetching and publication caching were performed. The genuine Falcon attempt failed authentication and the configured perplexity-lite fallback returned insufficient quota; no provider text was fabricated. Source caches and their flags remain unchanged. Separate original recovery is documented in the source evidence guide, including cases where a full_text_available cache omits Methods. Short exact source anchors occur once each in the review YAML; reference-only support elsewhere avoids duplicating those quotations.
The 11 partner-class refinements were checked again against 33 literal positive records, including validated two-hybrid confirmation, reference-sequence identity and the actual GO definitions. Each narrower term describes binding to the verified class, not the partner's catalytic or regulatory activity. These related screen records do not constitute independent native-cell validations. Keratin binding does not assert actin binding or recognition of assembled intermediate filaments; KRTAP partners are not automatically keratin equivalents. The tested full-length MTUS2 and short product remain distinct, and the PCSK5 product retains its protease domain without demonstrating maturation in the assay. Native coexpression, stoichiometry and physiological function remain unresolved for these pairs. HI-II-14, HuRI, and the literal records and reproduction support these bounded decisions. The project instruction permits an evidence-backed refinement; it does not require inventing a physiological role for a binary interaction.
This follow-up supersedes the preceding retention of the specific-granule-lumen and extracellular-region annotations. Their exact Reactome participant paths remain verified, but the primary neutrophil survey target measurement and topological evidence were not recovered. Both verdicts are now UNDECIDED, consistent with the unresolved exosome detection. Missing conventional signal-peptide annotation does not prove contamination or exclude unconventional release. Distinguishing a true enclosed pool, surface association and co-fractionation requires appropriate purity and protease-protection evidence.
The exchange core preserves the directly supported RAC1 substrate and additionally records human RAC2/P15153. The curated RAC2 reaction places DOCK2 as a hasMember of the catalytic CandidateSet, explicitly distinguishes established exchange factors from candidate-only entries, and models RAC2:GDP/GTP exchange. Its nucleotide complexes resolve to RAC2 entity 195339 and accession P15153; the processed chain spans residues 1–189 of the official 192-residue precursor. Human cellular activation experiments provide independent context. These sources support the scoped RAC2 substrate assignment without being described as an independently read purified RAC2 assay; direct purified human exchange in the existing core was measured on RAC1.
All 77 source dictionaries, both natural product records and all existing quotation text remain unchanged. The updated totals are 9 ACCEPT, 48 KEEP_AS_NON_CORE, 16 MODIFY, 1 REMOVE and 3 UNDECIDED. No NEW annotation or additional catalytic core is introduced. The earlier notes and history are preserved as historical records, and the source guide records the operative evidence and limitations.