The normal human P00918/CA2 seed contains 89 distinct source assertions from 89 GOA records and 74 source references. It contains no structured alternative-products list, tested-product qualifier or negated assertion. These machine fields are preserved. The normal UniProt, GOA, publication and Reactome files are immutable. Two pre-existing publication caches (PMID:14675051 and PMID:19056867) were preserved despite fuller fetched variants; family exports were not imported over the existing PTHR18952 records.
One normal Falcon deep-research attempt with the configured perplexity-lite fallback failed because DNS resolution was unavailable. No provider output was produced or imitated. This assessment uses the normal primary-source caches, selected original-paper access, UniProt, the source GOA, and the documented ontology checks.
CAII is a soluble zinc-dependent carbonate dehydratase. Its own active site catalyzes reversible CO2/bicarbonate interconversion; its physiological consequences include acid-base equilibration and the substrate supply coupled to epithelial transport. The core is therefore one carbonate-dehydratase/pH-regulation unit, rather than a collection of inhibitor, receptor or transporter functions. The official GO:0004089 definition describes the reversible bicarbonate-dehydration reaction, and GO:0051453 covers intracellular pH regulation.
The many source assertions for this same enzyme activity were assessed individually. The available abstracts distinguish kinetic mutants, substrate/metal structures, inhibitor complexes, comparative isozyme panels and human deficiency studies. Acceptance of the well-established activity does not mean that each paper's full experiment was independently read. In particular, PMID:17705204 and PMID:4994926 are bibliographic-only caches, PMID:8485129 explicitly truncates its abstract at 250 words, and the CAI-mutant focus of PMID:7758465 or CA XIII focus of PMID:18618712 does not establish misattribution of the curated CAII annotation. These limitations are recorded in their reference reviews; the activity itself has independent direct human evidence.
The actual cached Methods and catalytic Results in PMID:18942852 identify recombinant human CAII variants expressed in bacteria, enzyme purification, isotope exchange and wild-type comparisons. The catalytic and Discussion passages were read, but crystallographic images and supplementary data were not. PMID:19170619 uses a human CAII A65S/N67Q construct as a CAIX active-site mimic with wild-type CAII controls. The complete abstract and construct/purification Methods were read; the cancer biology of CAIX is not assigned to native CAII. The human CAII structures and mutational kinetics, not a domain label alone, support zinc-dependent catalysis.
PMID:3126084 explicitly reports hydrolysis of 4-nitrophenyl acetate by human CAII variants. This substrate is compatible with GO:0004064, but does not establish an independent major physiological aryl-ester pathway. PMID:10550681 and PMID:11015219 report cyanamide hydration and structural intermediates in human CAII. These secondary activities are retained as non-core. The cyanamide term's reaction is conventionally written in the reversible dehydration direction; that is not a contradiction of the measured hydration direction.
The exact three seeded partners are SLC4A7/NBC3 Q9Y6M7 (PMID:14736710), SLC4A4/kNBC1 Q9Y6R1 (PMID:15218065), and SLC26A6 isoform 4 Q9BXS9-4 (PMID:15990874). The first two canonical caches are abstract-only. The third contains actual main text: the complete available Results, Discussion and relevant construct, binding, immunoprecipitation and transport Methods were read. The SLC26A6 experiments use human SLC26A6a/S+Q and CAII in HEK293 cells. They support the reported interaction and reversible recruitment of endogenous CAII to the cytosolic face of the plasma membrane. Images and supplementary experiments were not independently inspected.
Direct binding and substrate-channeling models require assay-specific care. The later primary PMID:17090540(https://pubmed.ncbi.nlm.nih.gov/17090540/) reports no binding of purified SLC4-A1/A4/A8 C-terminal peptides to CAII, while reproducing orientation-dependent binding of GST fusions. That is a genuine challenge to direct SLC4A4 C-terminal recognition, but is not proof against all full-length transporter association, and did not test SLC4A7 or SLC26A6. The normal source was requested before final integration. The review distinguishes observed association from a universal transport-metabolon mechanism, and does not assign transmembrane transport activity to CAII. In addition, CAII changes the kinetics of acid-base equilibration itself, which matters when pH change is the transport readout.
In PMID:15990874, angiotensin receptor activation displaces CAII from SLC26A6 through a PKC-dependent mechanism; phosphorylation of SLC26A6 at Ser553 is proposed rather than directly established. The enzyme's demonstrated contribution is carbonate chemistry coupled to transport; its release is a regulated effector mechanism. The GO:0038166 definition describes angiotensin-triggered GPCR signal transduction. The draft treats a separate signaling-process assignment as overextended, while retaining the directly supported transport-regulation context. This is based on the actual experiment, not an argument from the title or abstract omission.
Supported generic binding is retained as non-core under the project's standing user instruction. The contested SLC4A4 edge remains uncertain pending assay reconciliation. No generic annotation is removed merely for being uninformative, and no finer molecular activity is invented from the name of a binding partner.
The soluble cytoplasmic/cytosolic location has direct human support and agrees with all five Reactome contexts. The first two Reactome summaries explicitly describe cytosolic catalysis; other summaries identify the reaction or inhibitor context without exposing a distinct compartment field. Their GOA cytosol assertions are retained on independent evidence, without inventing unseen Reactome metadata.
For PMID:14675051, the cached abstract explicitly includes CAII in mouse and human kidney studies. An institutional copy of the original paper provided Methods and early Results, and indexed passages provided mouse CAII localization captions. The human CAII fine-localization panel was not inspected; later page retrieval failed. The broad presence of CAII in the study does not establish that every apical ClC5 signal is also an apical CAII signal. The fine apical source assertion remains uncertain rather than being labelled a wrong-gene citation.
PMID:19056867 is abstract-only locally, but the original authors' NHLBI Urinary Exosome Protein Database explicitly lists CA2/NP_000058, nine peptides and references 1 and 2, the latter being this study. This target-specific observation supports non-core detection in the urinary vesicle fraction. It does not establish dedicated secretion, vesicle biogenesis, or a principal extracellular catalytic role. The full spectra were not reanalyzed.
The inherited assertions are assessed as phylogenetic or orthology claims, not by donor count. A target appearing among its own IBA sources is legitimate. PTHR18952 entries/metadata alone do not expose the ancestral IBD placements. Fine apical/myelin claims require their own donor evidence; family membership and a CAII-focused physiological expectation are insufficient. The directly relevant mouse study PMID:20150244(https://pubmed.ncbi.nlm.nih.gov/20150244/) reports CAII-dependent pH regulation during Gly-Sar absorption. Its normal cache was requested before final integration; this does not by itself prove that it was the exact source used in every transferred GOA row.
No NEW annotation is proposed. In particular, no new disease-development, inhibitor-response, transporter, secretion or osteoclast process is manufactured from necessity evidence. The existing pH and transport-regulation assertions already cover the established functional context, and proposed ancestor/descendant duplicates would not add coverage. No target match was recovered from the local GO-CAM index during this bounded check; absence of a local model is not evidence of a missing annotation. The alternative-product and source evidence fields remain as seeded.
Final action counts, normal validation results and any updated primary-source access will be appended after the remaining bounded consultation and normal reference recovery. This paragraph records the draft stage; it is not a claim of canonical application or campaign closure.
The final proposed decisions cover all 89 source objects: 67 ACCEPT, 13 KEEP_AS_NON_CORE, 4 MODIFY, 4 UNDECIDED and 1 MARK_AS_OVER_ANNOTATED. No NEW assertion is introduced. All four broad cytoplasm assertions are consistently refined to the already-established cytosol term. The four unresolved claims are the contested SLC4A4 interaction, one myelin-sheath transfer and two fine apical-location assertions. Their unresolved status does not question the independently established soluble catalyst or the supported transporter-specific associations.
The two additional papers were recovered by the unchanged normal publication fetcher. PMID:17090540 remains abstract-only; its complete abstract identifies the pure-domain/GST and SPR comparisons discussed above. PMID:20150244 supplies actual Methods, Results and Discussion, all read for the CAII context: the CAII-mutant mouse duodenal experiment measures increased epithelial acidification during Gly-Sar exposure. The corresponding short-circuit-current comparisons chiefly concern Pat-1 rather than directly measuring CAII-specific peptide flux. Thus the inherited dipeptide-regulation assertions are retained as contextual regulation through pH handling, without asserting a human intestinal experiment, peptide-transporter activity or an exact reconstruction of the historical donor citation. Figure images were not inspected.
A bounded consultation checked the eleven propagation/orthology assertions, another checked the single core and angiotensin/transport distinction, and a small source-join check confirmed the corrected inhibitor/proton-transfer summaries for PMID:17251017, PMID:17330962 and PMID:17346964. These consultations do not imply that every primary paper's full text was read. For angiotensin signaling, the actual reason is that AT1a/PKC acts on SLC26A6 and displaces the carbonate enzyme, rather than a general requirement for a separate molecular function in every process. The GO signaling definition includes downstream regulation, which was considered explicitly.
The limited evidence quotations are exact excerpts from the immutable publication caches. Aggregate counts include repeated uses; no publication exceeds 25 words across the proposed YAML and these notes. The notes add no verbatim publication quotations. Source assertions, products and old caches remain unchanged, and this completed scientific proposal still awaits independent peer review and canonical application.
Source checks preserve all 89 original annotation objects and all 74 original reference IDs/titles. The proposal has 77 references after adding the immutable UniProt record and two normally recovered papers. Its five exact evidence excerpts total at most nine words from any one source, including repeated-use counting. All 75 protected local gene/source/family files remain byte-identical to their source intake pins.
The unmodified normal full validator checked the schema, authored ontology terms, GOA and best practices, but reference validation failed on a dependency parser defect: linkml-reference-validator 0.2.1 treats any three hyphens as a YAML frontmatter delimiter, including the four-hyphen mutation notation inside the valid titles of PMID:1909891 and PMID:1928091. The former title is truncated and the latter causes a YAML parsing exception. The actual titles match their immutable caches. No cache or title was changed to satisfy this failure. A proposed delimiter-line-only fix and bounded regression checks are separate from this scientific review; full validation is not yet reported as passing.
Four advisory warnings remain expected: two supported generic-binding annotations deliberately retained under the user policy, and two IBA decisions lacking structured propagation details because the precise ancestral placement was not inspected. Those warnings justify DRAFT status even after the parser problem is repaired. Completed annotation decisions and campaign completion are separate from unresolved biological evidence and validation readiness.
The delimiter-line compatibility correction has now passed the normal full validation command, including schema, ontology, reference, GOA and best-practices checks. The same four advisory warnings remain, and the review retains DRAFT status. The authentic source caches, their titles and all 89 source annotation objects remain unchanged. The earlier failure above records the actual preceding checkpoint; it is now resolved by a separate tooling change.
The SLC26A6 account has also been made more precise: the PKC-dependent displacement of CAII is experimentally supported, while phosphorylation at Ser553 is a proposed mechanism. This wording correction does not change any annotation action, source object or evidence quotation. Scientific peer review and canonical gene application remain distinct from this successful validation checkpoint.
The exact independently approved review and notes were applied after source-preimage checks. Normal canonical validation, status, targeted rendering and a standard CREATE history validation passed. The review retains DRAFT with 4 advisory warnings, all 89 source assertions, 77 references and one catalytic core; the four unresolved biological decisions remain explicit. All publication, Reactome, UniProt, GOA and quarantined family source bytes are unchanged. The parser correction is a separate tooling change, and remote gene publication remains a separate step.
The two experimentally supported binding assertions are refined from generic protein binding to GO:0044325 transmembrane transporter binding. The official molecular-function definition covers binding to a protein or complex that carries substances across a membrane; its parent is protein binding. The PMID:14736710 abstract reports CAII binding to the cytoplasmic C-terminal domain of human SLC4A7/NBC3, with binding-site mutation and membrane recruitment. The cached Results and Methods of PMID:15990874(https://pmc.ncbi.nlm.nih.gov/articles/PMC1176462/) directly examine the human SLC26A6a/S+Q interaction using binding assays and co-immunoprecipitation. Figures and supplements were not newly inspected. The original partner identifiers, including SLC26A6 isoform 4, are preserved. Binding does not make CAII a transporter or establish a universal metabolon mechanism.
The standing project instruction still permits supported generic interactions to remain non-core when no more informative function is established. Here the concrete transporter-binding evidence warrants MODIFY. The disputed SLC4A4 assertion remains UNDECIDED; the later SLC4A1/A4/A8 peptide study does not directly test the other two partners.
PubMed confirms the identities of PMID:7758465(https://pubmed.ncbi.nlm.nih.gov/7758465/) and PMID:18618712(https://pubmed.ncbi.nlm.nih.gov/18618712/), but their experimental full texts could not be inspected through the attempted publisher routes. The former abstract foregrounds CAI mutants and the latter CA XIII structures. Neither omission establishes that CAII was not assayed. Their reference assessments are therefore UNVERIFIED rather than MISCITED, and they are removed from the reviewer-authored supporting citations for their activity rows. Both original GOA citations remain intact. ACCEPT is retained for the established CAII activity with independent direct support from PMID:18942852 and deference to the original curators on the uninspected experiments. The abstract-only cache flags remain unchanged.
This changes two binding actions and clarifies two activity rationales and two reference reviews. All 89 source assertions, 77 reference identifiers/titles, the five exact evidence quotations, all products fields and the single catalytic core are preserved. The proposed tally is 67 ACCEPT, 11 KEEP_AS_NON_CORE, 6 MODIFY, 4 UNDECIDED and 1 MARK_AS_OVER_ANNOTATED, with no NEW assertion. DRAFT status and the four unresolved biological assessments remain. The original validation checkpoints above remain historical; this follow-up still requires its own validation, scientific peer review, canonical application and publication.
TMP validation of this follow-up passed the normal schema, ontology, GOA, reference and best-practices checks with two pre-existing IBA propagation-detail advisories. Distinct scientific peer review and canonical application/publication are still pending.