CPAP (formerly CENPJ; HGNC:17272, Q9HC77) binds tubulin and couples its incorporation to centriole construction. The strongest integrated description is controlled centriole elongation, with a tubulin-binding molecular activity, centriolar location and membership in a CPAP–STIL procentriole assembly complex. Replication, elongation and localization annotations describe related aspects of that work; they are not independent enzymes.
The original aster assay combines human centrosomes with bovine tubulin. It supports CPAP contribution to astral nucleation but does not show that purified CPAP alone nucleates microtubules. Native gamma-tubulin co-IP/cosedimentation supports association with a broad gamma-tubulin complex. The paper explicitly leaves precise composition unresolved, so the small-complex source assignment is generalized while gamma-tubulin binding remains positive non-core context (PMID:11003675(https://pmc.ncbi.nlm.nih.gov/articles/PMC86375/)).
The direction of microtubule effects depends on what was tested. Human PN2-3(311–422) and high full-length expression inhibited assembly in the original experiments (PMID:15047868(https://pmc.ncbi.nlm.nih.gov/articles/PMC420094/): "this 112-residue CPAP also binds to tubulin dimers"). Later engineered CPAPmini/CPAPlong reconstitution showed slower growth with stabilization, whereas human PN2-3EX121–1060 modestly enhanced growth in another configuration. Full-length cellular experiments connect these tubulin contacts to centriole length, but neither a universal polymerase nor a universal depolymerase label captures the evidence (PMID:27219064(https://pubmed.ncbi.nlm.nih.gov/27219064/); PMID:27306797(https://pubmed.ncbi.nlm.nih.gov/27306797/)).
Human CPAP coiled-coil residues 897–1056 form a dimer. That result is distinct from dilute G-box monomers, zebrafish G-box/STIL crystals and a proposed higher-order molecular-ruler model (PMID:24076405(https://pmc.ncbi.nlm.nih.gov/articles/PMC3824074/)). The public evidence index supplies a short original Results anchor with the exact construct and section.
CPAP is a substrate of PLK2/PLK4; phosphorylation modulates its assembly contribution rather than giving CPAP kinase activity (PMID:20531387(https://pubmed.ncbi.nlm.nih.gov/20531387/): "PLK2 phosphorylates the S589 and S595 residues of CPAP"). Actual STIL interaction, endogenous complex and CPAP depletion/overexpression experiments support centriole replication and elongation. Cell-cycle synchronization in the STIL paper does not itself establish the CDK-activating G1/S signaling pathway (PMID:22020124(https://pmc.ncbi.nlm.nih.gov/articles/PMC3243611/)). The 2007 Plk4 study supplies positive human localization/assembly Results; the independently read author thesis is separately identified rather than called the journal paper (PMID:17681131(https://doi.org/10.1016/j.devcel.2007.07.002); author thesis).
CEP135 and STIL binding is supported by defined assays, with human and zebrafish constructs distinguished (PMID:23511974(https://pmc.ncbi.nlm.nih.gov/articles/PMC3630357/); PMID:25385835(https://pubmed.ncbi.nlm.nih.gov/25385835/)). CPAP loss impairs CEP295 recruitment, but the source did not detect strong CPAP–CEP295 co-IP and suggested a CEP135-mediated route; the process annotation does not become a direct adaptor/binding assertion (PMID:27185865(https://pubmed.ncbi.nlm.nih.gov/27185865/)).
The two spindle-regulation assertions citing PMID:17576815 remain unresolved. Its complete main text assays SIL/STIL in zebrafish and human cells, not CPAP. That is a source-specific mismatch; it does not disprove a CPAP spindle contribution. Aster nucleation and centriole construction alone do not resolve the precise regulatory relations (original paper).
Positive partner-class refinements retain the original assays and all source fields. EPB41 is a cytoskeletal component; CEP135, CEP152, STIL and BCCIP are cytoskeletal regulators; YWHAG is a 14-3-3 protein; PLK2 and TNKS are enzymes. The GFI1B and CIC screen pairs narrow to RNA polymerase II DNA-binding transcription-factor binding. Those classifications do not assign partner catalytic/transcriptional activities to CPAP or establish native encounters from screens.
The source-specific distinctions matter: BioID reports proximity; OpenCell reports IP-MS associations; HuRI reports binary two-hybrid interactions; separate small-scale studies supply direct or co-complex evidence. All 41 selected interaction records retain their 42 literal fields, and 82 assayed participant ROGIDs match the official displayed sequences. Fragment and mutation fields still constrain interpretation. The original OpenCell YWHAG–CENPJ row groups several CPAP identifiers, including Q9HC77-2, and cannot resolve individual product activity. The LYST domain-specific binding row remains undecided because its actual target/domain mapping was not recovered (PMID:11984006(https://pubmed.ncbi.nlm.nih.gov/11984006/)).
CPAP also increases STAT5-dependent reporter output in the tested human-cell system. This genuine coactivator result is retained as non-core, with the human CPAP/mouse STAT5 and reporter context explicit; upstream kinase activation and endogenous target genes remain unestablished (PMID:12198240(https://doi.org/10.1210/me.2002-0108): "CPAP was found to augment Stat5-mediated transcription.").
The official products reconstruct exactly to 1338 and 1086 amino acids. Product 2 retains PN2-3 but replaces 1073–1086 and lacks 1087–1338, removing part of the STIL-interaction region. Sequence retention is not proof of equivalent centriole construction, gamma-complex association or STAT5 function. No product-specific GO assertion is manufactured from those sequence features.
The actual current PANTHER tree/MSA places the human target below PTN000817566 and the target sequence matches the official 1338-residue product. The current IBD GAF has five positive assertions there; a separate loss node is outside the target path. Display annotations with a NOT suffix were not used to invent a target loss. Human self-evidence in WITH/FROM is legitimate descendant support. Current donor records and historical node evidence are kept distinct.
All three current ARBA rules genuinely match the official CPAP FunFam: ARBA00088571 branch 8 also requires Mammalia, and ARBA00092134 branch 2/ARBA00088865 branch 4 require Metazoa. The broad organelle-assembly prediction is refined to the experimentally supported centriole process. Current rule matches do not reconstruct historical versions. The 23 Reactome cytosol rows have actual target paths through modeled physical entities; CPAP is not assigned the catalytic activity of every enclosing event.
The manually assembled evidence index preserves all 82 original source objects, complete current GOA records, products, literal interaction/donor/rule evidence, typed Reactome paths, public source URLs and short excerpt locators. It distinguishes selected caches from separately recovered original main text and from independently read source consultations. Full-text flags on protected machine caches were not rewritten.
Important remaining limits are the complete journal Methods for 17681131, full target/domain evidence for 11984006, the exact 21399614 supplementary target table, complete 16713569/20852615 Methods, and complete 32296183/35271311 Methods in this reading. Attempted PDFs/spreadsheets or pages that returned HTML challenges are recorded as failures, not as usable scientific data. Positive original evidence or current curated records are retained where they independently establish the asserted scope; unavailable material is never treated as negative evidence.
The isolated-centrosome experiment in PMID:11003675(https://pmc.ncbi.nlm.nih.gov/articles/PMC86375/) supports microtubule nucleation by a microtubule organizing center (GO:0051418). Molt4 centrosomes were isolated after nocodazole treatment and combined with bovine brain tubulin; antibody controls tested the centrosomal contribution. An aster shape does not establish the spindle-pole context in the definition of an astral microtubule. This supersedes the earlier astral-nucleation wording above. Both original source assertions remain present with the same more accurate replacement; neither asserts that purified CPAP is an autonomous nucleator.
Broad gamma-tubulin-complex membership remains supported by reciprocal endogenous SiHa co-immunoprecipitation and cosedimentation in gamma-tubulin-enriched Molt4 fractions (original Results and Figure 7). The paper explicitly interprets a native complex while leaving its precise composition unexamined. The broad GO:0000930 definition permits a gamma-tubulin-containing multiprotein complex without requiring the specific small-complex composition. This supports generalization of the small-complex term, while leaving canonical TuRC membership, stoichiometry and direct subunit contact unresolved. Binding and complex membership are distinct judgments, supported here by the combined native experiments rather than the paper title alone.
The single core now explicitly names centriole replication, with its elongation step retained in the description and existing annotations. CPAP's tubulin recognition, assembly-partner interactions and localization within nascent centrioles explain this integrated construction role (PMID:17681131(https://doi.org/10.1016/j.devcel.2007.07.002); PMID:22020124(https://pmc.ncbi.nlm.nih.gov/articles/PMC3243611/)). Broader centrosomal localization and positive-regulation terms need not become separate cores. Human RPE1 tubulin-binding-mutant experiments and fly Sas-4 rescue also support the retained cilium-assembly annotation. The proposed control of tubulin availability at the ciliary base remains distinct from unproved direct action at the axonemal tip (PMID:27306797(https://pmc.ncbi.nlm.nih.gov/articles/PMC4912634/)).
The annotation evidence now includes concise checkable anchors for phospho-CPAP procentriole localization (PMID:20531387, Figure 6 and Methods) and CPAP-induced elongation in the CEP295-depletion experiment (PMID:27185865, Results and Figure S2 description). The centrosome annotation citing PMID:21399614 has an independent PMID:11003675 target-localization anchor; that addition does not claim recovery of the still-unverified supplementary CPAP table.
Positive binding decisions retain their assay limits and verified partner identities. A cytoskeletal regulator, enzyme or transcription-factor partner class describes the interacting protein, not an activity conferred on CPAP. A proximity or association result is not promoted to a purified direct-binding measurement, and lack of separate replication alone does not negate the recorded positive assay.
Centriole elongation (GO:0061511) is restored as the single specific core process. The actual GO relation places elongation part of centriole replication (GO:0007099), so both need not be listed in this compact core. Human CPAP-myc overexpression and STIL-depletion experiments directly test elongation of newly assembled procentrioles (PMID:22020124, Results and Figure 9 caption); human RPE1 CPAP tubulin-binding mutants and live elongation measurements further connect tubulin recognition with controlled growth (PMID:27306797, Results). This supersedes the preceding paragraph's rationale for replacing elongation with the broader replication term. Broader procentriole construction remains in the biological description and accepted annotations.
Ciliary basal body (GO:0036064) is also included as a core location. The existing Ensembl inference traces to mouse Q569L8's exact MGI IDA record for this compartment (PMID:23213448; QuickGO record UniProtKB:Q569L8!208284128), with the Ensembl identifier independently mapped to that mouse protein. The donor record was inspected; the mouse primary full text was not. The separate human RPE1 and fly Sas-4 experiments in PMID:27306797 support a conserved centriole/basal-body construction role, and its Discussion locates the proposed tubulin-pool mechanism at the ciliary base rather than demonstrating action at the axonemal tip. The basal-body definition distinguishes this ciliated-cell context from centriole, despite their structural continuity. No new localization assay or equivalent activity of both human splice products is asserted.