The reviewed human protein is UniProtKB P02746, HGNC:1242, the 253-residue complement C1q B chain. The normal source snapshot contains 62 distinct annotation objects and 63 GOA rows: one synapse assertion differs only in source date between two raw records. The review preserves all 62 machine-sourced objects, their qualifiers, references and partner/donor identifiers. The normal seed has no structured alternative-products slot; none is invented.
All 62 decisions have been assessed: 47 ACCEPT, nine KEEP_AS_NON_CORE and six UNDECIDED. No NEW, REMOVE, MODIFY or PENDING assertions are proposed. The unresolved cases concern the C1QB–UBQLN2 interaction record, two ECM proteomics records, a blood-microvesicle record, and two fine synaptic localizations. Completion of the review does not resolve those biological or source-access questions.
Human C1q consists of six A/B/C heterotrimers, each with a collagen-like stem and globular head. Recombinant human B-chain globular modules recognize immunoglobulins, with greater IgG than IgM binding in the reported assays. The independently folded MBP modules were produced in E. coli; the intact protein retains multivalent heterotrimeric organization. Thus B-chain-specific recognition is supported, but isolated-module preferences should not be equated with the complete serum molecule's affinity or geometry (PMID:12847249(https://pubmed.ncbi.nlm.nih.gov/12847249/)). The human C1q scFv study implicates B-chain Arg108/Arg109 while mutagenizing A/C modules; it is not described here as an isolated B-chain mutation experiment (PMID:19006321(https://pubmed.ncbi.nlm.nih.gov/19006321/)).
The first core uses immunoglobulin binding with classical complement activation. Full-length human C1q assembled in HEK293-F cells provides direct evidence for structural participation: mutation of mature B-chain Lys61 interferes with recruitment of the C1r2C1s2 tetramer and activation, without abolishing the tested ligand-recognition properties. This is work performed by the B chain, not a downstream necessity phenotype. Lys61 is mature-chain numbering, corresponding to position 88 in the precursor. C1r and C1s perform proteolysis; C1QB does not acquire their catalytic activity by being in C1 (PMID:23650384(https://pubmed.ncbi.nlm.nih.gov/23650384/)).
Antibody hexamerization, C1–IgG structures and human C1q/C1 binding to antibody-coated bacterial and model surfaces corroborate recognition and activation at the assembled-complex level. The 1988 ligand-mutagenesis paper tests mouse IgG2b, not a human B-chain mutant. These distinctions are preserved in individual reasons (PMID:24626930(https://pubmed.ncbi.nlm.nih.gov/24626930/), PMID:29449492(https://pubmed.ncbi.nlm.nih.gov/29449492/), PMID:34155115(https://pubmed.ncbi.nlm.nih.gov/34155115/), PMID:3258649(https://pubmed.ncbi.nlm.nih.gov/3258649/)). Extracellular cell-surface association means deposited ligand-bound C1q, not a transmembrane B chain. The microbial-surface annotation is supported by actual S. aureus experiments.
The second core represents a contribution to phosphatidylserine binding. Human serum C1q and its heterotrimeric globular region directly recognize PS in biochemical and apoptotic-HeLa-cell experiments. The inspected crystal Results place the principal phosphoserine-binding pocket on the C chain; an A-chain contact involves a neighboring crystal molecule. These experiments do not establish autonomous B-chain PS recognition. The original GOA row is retained unchanged, including its qualifier, while the authored core makes the complex contribution explicit (PMID:18250442(https://pubmed.ncbi.nlm.nih.gov/18250442/)). No new apoptotic-clearance process is added merely because C1q binding accompanies cell removal.
The third core represents a contribution to receptor-ligand activity. Human C1q and globular fragments bind CD33; monocyte experiments demonstrate CD33 ITIM phosphorylation, and whole C1q is required for coordinated CD33/LAIR-1 engagement. This supports signaling by a ligand assembly. It does not show that a purified B chain alone activates either receptor, that C1QB is a kinase, or that activating an inhibitory receptor makes the ligand an antagonist (PMID:28325905(https://pubmed.ncbi.nlm.nih.gov/28325905/)). The distinct human ghB–gC1qR interaction is directly supported by purified-module assays, but gC1qR binding and PBMC proliferation alone do not identify an equivalent receptor-signaling mechanism (PMID:28018340(https://pubmed.ncbi.nlm.nih.gov/28018340/)).
The three cores separate B-module immunoglobulin recognition, contributed lipid recognition, and contributed inhibitory-receptor ligand activity. They describe cooperating functions of C1q rather than three independent B-chain enzymes. The two contribution cores omit an unsupported autonomous molecular function. Complex membership is recorded in in_complex, not anatomical locations.
The standing project instruction takes precedence over the general skill/validator preference to remove generic protein binding: supported interactions remain non-core when no defensible finer MF is established; unverified interactions remain UNDECIDED. All six protein-binding annotations were joined to their exact GOA partner identifiers. Five are retained as non-core and the C1QB–UBQLN2 screen record remains unresolved.
The C1QA partner is independently established by the non-polar interfaces in the C1q globular heterotrimer. A covalent A/B peptide linkage alone would not establish selective noncovalent binding; independent structural evidence supplies that distinction. For PMID:21054788 and PMID:28325905, the precise A/B assay was not isolated in the accessed text, but the same association is independently established, so the review defers to experimental curation rather than contradicting it from an abstract or foregrounded receptor. The latter GOA row is not silently transformed into a CD33 interaction (PMID:12960167(https://pubmed.ncbi.nlm.nih.gov/12960167/), PMID:7240(https://pubmed.ncbi.nlm.nih.gov/7240/)).
The two ECM proteomics papers and the microvesicle paper have plausible biological contexts, but the actual target-level C1QB entries and peptide/compartment evidence were not inspected. Secretion by itself does not settle a matrix or vesicle location. The HuRI paper's general two-hybrid methodology likewise does not supply the individual C1QB–UBQLN2 record. These annotations remain unresolved without alleging false positives.
The ten-row independent annotation consultation checked exact IBA/ISS source objects and six PAINT IBD node/term/source joins. Target self-evidence is legitimate, and donor count is not a weakness. The classical-pathway PAINT snapshot includes an additional zebrafish donor absent from the seeded WITH/FROM list; both authentic snapshots are preserved. No tree topology or target-specific loss is inferred from that difference. The full phylogenetic tree/MSA and every donor experiment were not reconstructed.
General synaptic localization and pruning transfers remain non-core. Mouse C1q/C3 perturbation and neuronal-localization work, plus later C1qa-knockout/anti-C1q experiments, support whole-C1q participation in synaptic remodeling. These are not autonomous human B-chain assays. Exact extrinsic-presynaptic-membrane and glutamatergic-synapse localization donor experiments remain unverified, so those two IBA assertions remain UNDECIDED (PMID:18083105(https://pubmed.ncbi.nlm.nih.gov/18083105/), PMID:27033548(https://pubmed.ncbi.nlm.nih.gov/27033548/)).
No NEW assertion is proposed. Existing annotations cover the supported recognition, assembly, signaling and broad process roles. The broad immunoglobulin-binding term in core synthesis summarizes the existing IgG and IgM binding assertions; it is not a redundant NEW row. The exact UniProtKB:P02746 join in the local GO-CAM index has no entry. This observation is not treated as evidence of a missing annotation.
All 26 available complete abstracts among the 27 PMID references were read. PMID:6249812 has metadata only; its original experimental evidence remains unverified. The C1-complex/classical-pathway assertions from that source are retained using independent human C1q experiments, with no claim that the inaccessible paper was inspected. All seven cached Reactome reaction summaries were read. Their extracellular assembly context does not make the B chain the enzyme or inhibitor performing the reaction.
Selected complete source-relevant Methods, Results, Discussion and captions were read from the canonical bodies of PMID:18250442, 24626930, 25037231, 28018340, 28104818, 28325905, 28675934, 32296183, 34155115 and 27033548. Coverage differs by paper, as the reference assessments record. This is not a claim to have read every body, every figure image or the missing supplementary target tables. The canonical PMID:28675934 body is partial despite its full-text flag.
PMID:23650384 is an explicit exception in source location: its preserved canonical cache is abstract-only, while an authenticated, normally fetched alternate body supplied the inspected recombinant-human Methods/Results/Discussion. The alternate is preserved separately under the source transport extraction; no canonical cache was overwritten. The review's full-text-availability field records actual accessed full text, not an assertion that the canonical cache has those sections. Quotes from this reference come only from the canonical abstract, so they remain independently checkable by normal validation.
The normal Falcon attempt with perplexity-lite fallback failed DNS for both providers and produced no provider files. The review uses manual primary-source reading and the documented independent annotation consultation; no provider output or publication body was fabricated. The automated source files and canonical seed remain unchanged during candidate preparation.
Five short exact quotation instances anchor key mechanistic statements. Aggregate quoted words, counting repeated uses, are 12 for PMID:18250442, six for PMID:12847249, eight for PMID:23650384 and seven for PMID:28325905. These notes add no verbatim quotations. Bibliographic identifiers/titles and all source fields are preserved; two existing normal publication caches and the preserved UniProt record are added as explicit review references.
Normal candidate validation, including schema, authored ontology terms, references and GOA checks, passed with no errors and six warnings. Five warnings reflect the explicit project policy retaining supported generic binding. The sixth flags the core's immunoglobulin-binding summary term, which is an ancestor of already present IgG/IgM terms; adding a redundant NEW row would violate the annotation rule. DRAFT is therefore the schema-consistent status even though every source decision has been completed. Independent whole-gene peer review remains pending. Canonical authored application, history and publication have not been performed by this preparation.
Independent whole-gene scientific review passed, and the approved review was applied after checking the original source bytes. All 62 machine assertions and original reference identities remain intact; decisions are 47 ACCEPT, nine KEEP_AS_NON_CORE and six UNDECIDED. The three cores distinguish B-chain immunoglobulin recognition/protease recruitment from complex-contributed phosphatidylserine recognition and receptor-ligand activity. Source caches remain unchanged. DRAFT records the documented advisories and unresolved evidence limits; no NEW assertion was added.
The broad complement-activation assertion is refined to the more specific classical-pathway term already supported by the human recognition and protease-recruitment experiments. This is a MODIFY of an existing source assertion, with its original term/reference/evidence fields preserved, rather than a NEW annotation. The final decisions are 46 ACCEPT, nine KEEP_AS_NON_CORE, six UNDECIDED and one MODIFY. All other annotations, three cores, reference identities, quotation anchors and biological access limits are unchanged. The preceding counts record the completed initial assessment before this specificity refinement.