CEP57 is a microtubule-binding centrosomal organizer with a separable protein-anchoring role. The review preserves all 69 normal seeded assertions, five UniProt products and 40 original reference identifiers/titles. Seven genuine normal publication caches add mechanistic context. No NEW annotation is needed: the existing microtubule-binding, cytoskeletal-adaptor and spindle-organization assertions already cover the principal activities.
The most informative anchoring experiment is not a generic depletion phenotype. Human CEP57 binds purified PCNT PACT fragments, recruits PACT to ectopic microtubule-associated CEP57, and supports centrosomal loading. A PCNT–CEP57 chimera probes that physical route. The primary description, “anchor for the PACT domain,” captures this performed work PMID:30804344, Results/Fig. 4 and subsequent rescue experiments. This does not establish a fixed native ternary bridge or an intrinsic motor activity. The paper explicitly uses human isoform-a NP_055494.2; the five current products are retained without extrapolating identical behavior to all of them.
Direct microtubule evidence has different construct boundaries. The 2008 work uses mouse CEP57 fragments and bovine-brain tubulin; pronounced bundling contributes to its turbidity signal PMID:18294141(https://pmc.ncbi.nlm.nih.gov/articles/PMC4351815/). The 2016 study supplies independent human HeLa-cDNA-derived C-terminal microtubule co-sedimentation PMID:26743940(https://pubmed.ncbi.nlm.nih.gov/26743940/). The 2024 study examines full-length tagged human CEP57 condensation but conducts most detailed reconstitution with engineered Cep57S, residues 56–264 joined to 334–433. Tubulin enrichment and a lower assembly threshold in PEG/taxol-containing assays are genuine measured activities; they are not a demonstration of native full-length processive polymerase activity. Full-length WT versus KR/FF cellular rescue is a separate experiment, and the mutations affect multiple interaction/assembly properties PMID:38857398, Figs. 1–6 and purification/assembly Methods.
The September 2026 architecture article DOI:10.1038/s41467-026-77795-4 was accessible only as an authentic publisher abstract during this review. It is consistent with a structural CEP57 bridge but is not used to assert construct identity, atomic interfaces or controls absent from the accessible text. No publication cache or PMID was invented for it.
The physical kinetochore-attachment IBA remains UNDECIDED. The frog donor Q5FWP9 is currently named cep57l1; that fact does not prove PAINT node misplacement. Its original study explicitly addresses attachment in Xenopus egg extracts. The human 2016 paper reports checkpoint effects but no significant attachment defect. The 2019 human study obtains different checkpoint results under its conditions while acknowledging some kinetochore staining. These observations distinguish physical attachment from checkpoint signaling and PCM integrity; neither donor count nor legitimate target self-inclusion is used to reject an ancestral assertion. Complete PTN001064019/PTN001456386 history was not reconstructed PMID:17803911(https://pubmed.ncbi.nlm.nih.gov/17803911/), PMID:26743940(https://pubmed.ncbi.nlm.nih.gov/26743940/), PMID:30804344(https://pubmed.ncbi.nlm.nih.gov/30804344/).
FGF2 trafficking is retained as a distinct contextual role. Actual publisher Main/Figs. 2–5 and Methods show human-library interaction discovery, mouse NIH3T3 depletion and bovine endothelial ligand experiments. The binding-deficient ligand retains selected FGFR/ERK/p70S6K outputs while losing nuclear-associated accumulation and most mitogenicity; an NLS fusion partly rescues the latter. The source therefore supports refining the existing receptor-signaling assertion to protein localization to nucleus, retaining nuclear/perinuclear ambiguity. It does not prove universal nonparticipation in receptor signaling or identify a nuclear-pore motor, importin or dynactin bridge. The inspected Methods do not unambiguously identify the species of every purified construct PMID:12717444, full primary.
The two spermatid-development transfers are over-annotated at the inspected source scope. The exact mouse Q8CEE0 donor uses IEP from PMID:12954732. The original author-uploaded Methods/Results measure developmental expression, enriched germ-cell populations, vitamin-A deficiency and RARalpha-null backgrounds, plus localization and interaction. They do not demonstrate a CEP57-performed developmental step or a CEP57-specific germ-cell perturbation. This is a limit of that evidence chain, not a claim that CEP57 cannot function in spermatids. Mouse full-length GFP localization in CV-1 cells supports the retained broad nuclear/cytoplasmic locations; the source's inconsistent NLS numbering is not propagated PMID:12954732, original author copy.
The original Golgi IDA remains UNDECIDED because its exact target construct/image and marker were not recovered from PMID:10942595. Later centrosomal evidence cannot disprove a Golgi pool. Conversely, independent human target microscopy corroborates the original centrosome assertions even though the target output from the older proteomics sources was not re-extracted. Human gamma-TuRC co-immunoprecipitation supports the gamma-tubulin association at complex resolution; NEDD1-dependent recruitment does not establish a purified binary gamma-tubulin interface PMID:22508265(https://pubmed.ncbi.nlm.nih.gov/22508265/).
The durable source extract contains 49 literal IntAct records supporting 19 exact source pairs, the two original 2011 supplementary pairs, current donor records and the 22 Reactome event memberships. It uses complete source-object hashes rather than annotation positions as join keys. Original Supplementary Table 2 PMID:21516116, publisher PDF, visually inspected on pages 16–17, contains GCC1–CEP57 (79571–9702) and RALBP1–CEP57 (10928–9702). Screen stages are not independent replications; curated interaction records may share GOA provenance. Features preserve MYC/CEP57 fragments and the GCC1/TFIP11 partner variants. Blank features do not establish WT state. The NS5A source explicitly names translokin in its authentic abstract, while its exact viral strain/truncation linkage remains limited. Supported MYC, HRAS and FGFR3 interactions receive partner-class refinements without importing their activities into CEP57.
Supported broad binding follows the published campaign instruction: retain a biologically correct supported generic binding assertion as KEEP_AS_NON_CORE unless a more informative MF is established. This instruction does not verify uninspected targets. The 22 event-derived cytosol assertions are peripheral compartment context, not 22 independent experiments or activities of CEP57 in every named event.
Normal fetch seeded the source assertions and products. The genuine Falcon attempt timed out; the explicit Perplexity fallback returned an authentication/quota error. No provider report exists and none was written under a provider filename. Manual primary research, exact API records and independent annotation/core consultations supplied the review. Existing source caches remain unchanged; externally recovered full text is distinguished from abstract-only normal caches. Short verbatim excerpts are budgeted across YAML and this journal together, including repeated anchors. All other source references use assay locators and paraphrases.
The gamma-tubulin IEA is now UNDECIDED. The live InterPro IPR025913 entry really carries GO:0043015 and cites PMID:18294141, but that paper uses gamma-tubulin as a positional marker. Independent human CEP57 immunoprecipitation in PMID:22508265 positively recovers gamma-tubulin, GCP2 and NEDD1. Its NEDD1 binding/localization experiments do not establish that NEDD1 is the sole bridge to gamma-tubulin. The unresolved point is the partner-specific inference; complex coassociation remains positive evidence, and purified binary assays are not a universal requirement for a binding annotation. The appended source artifact records the official mapping, source URLs and selected full-text locators.
The FGF2 refinement now uses GO:0006886 intracellular protein transport. PMID:12717444 distinguishes cytoplasmic uptake from subsequent nuclear-associated accumulation and reports predominantly perinuclear signal. The broader destination avoids resolving the authors' nuclear/perinuclear ambiguity by terminology alone. The original IPI/FGF2 source stays intact. The homodimerization refinement explicitly acknowledges convergence with the separately sourced existing homodimerization row.
No Mis12/Mad1 NEW is proposed in this follow-up. PMID:26743940 provides genuine human Mis12 binding and kinetochore/checkpoint experiments, which remain part of the evidence and suggested questions. PMID:30804344 reports a different outcome for Mad1 recruitment and checkpoint duration under its conditions and instead supports centrosomal/PCM mechanisms. These positive binding assays are not discarded; the combined evidence does not yet resolve a stable checkpoint role to assert as a new process, and no extra generic binding row is added solely for coverage.
The 22 Reactome cytosol assertions remain non-core localization context with shared provenance. Their placement does not assign CEP57 the catalytic or regulatory work of each event, and they are not 22 independent experiments. Being a peripheral event participant does not itself refute a cytosolic pool. PMID:21552266 was genuinely fetched through the normal publication workflow and supports the existing MVA disease statement, without assigning its phenotype to a specific mechanism.
Crossref independently resolves DOI:10.1038/s41467-026-77795-4, with online publication on 2026-09-30 and the matching CEP57 title. The artifact preserves its literal DOI/title/date and response hash. No PMID is invented, and this abstract-level discovery still supports no annotation. Earlier provider failures and external-versus-cache access descriptions remain unchanged.
All 69 source assertions, five product records, 47 original reference id/title pairs and the entire earlier notes text are preserved. One normal publication cache and reference are added. No new publication quotation is added; existing whole-file totals (YAML plus the inherited journal) remain PMID:12954732 6 words, PMID:26743940 16, PMID:30804344 21, PMID:12717444 13, and PMID:38857398 8, including repetitions. The earlier journal is inherited verbatim and is not presented as newly authored text. The 25-word authoring limit is not a repository validator rule.