COG4 is an integral, noncatalytic subunit of the conserved oligomeric Golgi tethering complex. One protein-complex-scaffold core captures its multiple COG-subunit contacts and its coordination of Sly1/SCFD1 with syntaxin 5. The original human-subunit assembly analysis supports this refinement of the COG7-source binding assertion; the other positive COG pairs remain separate. The official scaffold definition and its three specific descendants were checked. Ubiquitin-ligase, MAP-kinase and centriole scaffolds do not describe this complex. No NEW molecular function or biological process is added. [PMID:15047703; PMID:19536132]
The original 2004 paper uses human constructs, including COG4 AI634312/SCOD1M4, expressed in rabbit reticulocyte lysate. Pairwise reciprocal co-immunoprecipitation supports contacts with COG1, COG2, COG5 and COG7. The lysate may contain bridging factors, so these experiments do not establish purified binary affinities. The assembly experiments support a structural role without making COG4 the sole bridge responsible for every subunit contact. PMID:15047703
The 2009 source uses human COG4 and STX5, recombinant rat rSly1, and endogenous HeLa material. Purified COG4 residues 1–231 bind GST-rSly1. The E53/71A mutant loses Sly1 association while preserving STX5 binding and incorporation into the COG complex; it does not restore retrograde trafficking as wild type does. The independent bounded consultation inspected original supplementary Figure S7: wild-type COG4 supports association of rSly1 with an STX5 mutant that cannot bind Sly1 normally, whereas the COG4 binding mutant does not. This is positive coordination of distinct partners. Autonomous SNARE zippering, membrane fusion and glycosyltransferase activity are not established. PMID:19536132
The human HEK293T COG4 knockout and replacement experiments connect complex function to Golgi architecture and glycosylation. Electron microscopy reveals cisternal defects even when some Golgi-marker distributions remain perinuclear. The toxin-transport assays cross several compartments and have receptor-binding limitations; they are interpreted together with the source-specific recruitment and retrograde-transport experiments. COG4 performs structural assembly and recruitment, rather than catalyzing the glycan reactions whose distribution is disturbed. [PMID:27066481; PMID:19536132]
Human fibroblast fractionation supports a large cytosolic COG4 pool and a smaller Golgi-associated pool. These are complementary locations. The nine Reactome assertions preserve their actual complex-location chains: Golgi COG4 entity 6808722 and TGN entity 6813868 both resolve to Q9H9E3. The source-evidence index retains the ContentService URLs, response hashes and literal entity/reference projections for these links. The pathway source explicitly leaves joint GARP/COG capture of one vesicle uncertain, so these location assertions do not establish that mechanism or assign every reaction operation to COG4. PMID:19494034
All three fetched human products are preserved and independently reconstructed against official sequences: 785 residues, 337 residues and 712 residues. The short product changes positions 331–337 and lacks 338–785. The 712-residue product lacks residues 1–73, including the E53/E71 region tested in the Sly1 experiments. The full recruitment mechanism has not been demonstrated for either alternative product; missing sequence is not itself an experiment proving inactivity.
Biallelic COG4 deficiency and the recurrent dominant Saul-Wilson variant have distinct clinical and trafficking contexts. The p.R729W/deletion fibroblasts retain complex formation despite reduced COG4 and disturbed trafficking. COG4 overexpression itself perturbs traffic, complicating rescue interpretation. The C-terminal structure and HeLa work support a glycosylation-related domain role without intrinsic glycosyltransferase chemistry. The Saul-Wilson paper reports p.Gly516Arg; its original coordinates are preserved rather than silently equated with a different current sequence coordinate. No developmental or disease process is manufactured from these phenotypes. [PMID:19494034; PMID:19651599; PMID:30290151]
The campaign instruction in projects/CLINGEN_MENDELIAN.md retains supported experimental generic-binding observations as non-core when a narrower class is not established. Genericity or uncertain native colocalization does not make a positive interaction false. All original machine source objects remain intact. The exact IntAct/MINT projections contain 58 matched export records across the 22 binding assertions, drawn from 156 raw 42-field records. Duplicate cross-database exports remain identified rather than counted as independent experiments.
HuRI provides eight target/partner pairs. All their COG4 records specify the 253–785 binding region. Official sequence fingerprints resolve the partners, including the 516-residue CTNNA3 isoform 2 and the 421-residue ANKRD11 fragment X5D778. Whole-length alpha-catenin or ANKRD11 functions are not automatically transferred to these shorter tested products. The tested displayed Alpha SORBS3 product supports the more specific cytoskeletal regulatory protein binding class through its documented actin-organization function; FARSA supports enzyme binding. These class assignments do not require a mapped regulatory or catalytic interface and do not assign those activities to COG4. PSMA1 is a noncatalytic proteasome alpha subunit; its isolated pair does not establish binding to the assembled proteasome. Dataset-wide MAPPIT/GPCA validation is not claimed for each pair. PMID:32296183
BioPlex provides four COG-subunit pairs and MUL1. Exact co-purification records and the verified MUL1 E3-ligase identity support the specific ligase-binding refinement, without showing COG4 ubiquitination or a mitochondrial COG4 mechanism. The separate original U2OS AP-MS network contains the exact COG2-bait to COG4-prey edge; this is direct dataset support rather than image-proximity inference or reuse of a BioPlex comparison. [PMID:33961781; PMID:40205054]
The human and rat Sly1 assertions are kept distinct. The rat accession has an authentic recombinant rSly1 premise even though no exact pair was found in the current export. Endogenous HeLa co-immunoprecipitation supports the human association separately. The original Methods identify the rSly1 provenance, but do not provide a verified sequence of that exact clone. PMID:19536132
All 43 original assertions and three products are preserved. The 44 raw GOA rows contain one duplicated STX5 assertion whose assignment agency and date differ; the source tuple is represented once in the seeded review. All 17 original reference identity pairs remain, with four genuinely fetched supporting references added. Six current donor/target joins contain 18 literal records. Target self evidence is legitimate experimental grounding. Historical PAINT node placement and the family alignment were not inspected, and PTN status remains UNRESOLVED without disputing the retained biological annotation. The selected GO-CAM index contains no Q9H9E3 match; absence is not used to infer a missing function.
The selected PMID:15047703 and PMID:19536132 caches are authentic abstracts. Their original full articles were recovered separately, and the actual source windows were read without replacing those caches. The bounded core consultation additionally read the S7 pixels and separate legend; the owner does not claim that independent image inspection as its own. The consultation's initial report called row 11 COG2 in one phrase; its sealed correction confirms the actual P83436/COG7 source. The source object and recommendation never changed.
The selected HuRI text supplies Results/Discussion and validation legends; no separate full Methods access is claimed. The selected BioPlex cache contains a limited body and abstract; its genuine fuller normal variant is archived separately. The 19651599 and 30290151 normal abstracts support their qualified findings. Original PubMed figure legends were also read for 19651599, but full-body recovery failed. For 30290151, indexed abstract/Introduction access succeeded and subsequent body retrieval encountered a browser challenge; detailed Methods/Results were not read. PMID:11703943 remains abstract-only yeast context. Failed routes are preserved as failures, not as evidence.
The genuine default Falcon research attempt timed out; its configured fallback returned an insufficient-quota error. No provider report was fabricated. Manual primary reading and an independent bounded consultation support the review. Raw primary text remains in TMP; the publication index holds source identifiers, hashes, record projections and concise access scopes. The four short primary anchors occur once each in YAML and are not repeated in these notes or the index. Existing publication, Reactome and GO-CAM bytes are protected. No prior history or canonical file was edited by the owner.
Normal validation and rendering passed with 18 advisories: 17 positive generic-binding retentions follow the explicit campaign policy, and GO:0017119 has the current official label “COG complex” while the validator uses the older “Golgi transport complex” label. The correct current term and source labels are preserved. The initial status check identified the seed's INITIALIZED state; the genuine status command subsequently supplied its calculated DRAFT value after the derived field was cleared in the isolated working copy. No biological decision changed. The whole peer corrected the published host-organism projection from MITAB field 15 to field 28, with all 58 raw records and source objects unchanged.