CHST3 (HGNC:1971; Q7LGC8; C6ST/C6ST1) was unchecked and unowned at the authenticated main snapshot. The genuine isolated normal intake produced 29 assertions, 14 original references and no alternative-products field. All original machine fields, order, qualifiers, supporting entities and reference identities are retained. Source snapshots, normal PANTHER output and original cache variants remain immutable. The authored artifact is provenance, not independent experimental evidence; its file references have no self-quotes.
Actual Falcon research timed out after 90 seconds and the configured Perplexity-lite fallback returned HTTP 401 insufficient_quota. The exact command/results are in the packet. No provider-named report was invented. Manual primary research supplies the evidence below.
PMID:9883891(https://febs.onlinelibrary.wiley.com/doi/10.1016/S0014-5793(98)01532-4), Methods 2.3–2.5 and Results 3.2, tests a human 49–479 CHST3 protein-A fusion expressed in COS-1 cells and captured on IgG-Sepharose. Mock pSVL transfection controls lack the recovered activity. Sulfate incorporation, chondroitinase digestion, HPLC and sulfatase analyses identify GalNAc C6 sulfation in GlcA-linked units. Prior 4-sulfation and IdoA-linked contexts are not equivalent substrates. This is affinity-recovered soluble enzyme, not an assay of native full-length membrane topology. The official indexed full article was independently read; its normal cache remains abstract-only. The local 403 page is explicitly a failed access artifact.
The same Results 3.2 paragraph reports recombinant utilization of corneal keratan sulfate as data not shown. This corroborates the earlier experimental keratan annotation, despite incomplete access to PMID:9714738(https://pubmed.ncbi.nlm.nih.gov/9714738/). Keratan capability is retained as secondary; the observation does not map a GlcNAc sulfation site or establish dominance over dedicated keratan enzymes in vivo.
PMID:15215498(https://pmc.ncbi.nlm.nih.gov/articles/PMC454181/), full Functional Studies and recombinant Results/Table 4, independently uses the corresponding soluble human construct and shows R304Q loss of activity. Fibroblast and urine disaccharide analysis links the enzyme defect to altered chondroitin 6-sulfation. The PAPS-region structural discussion uses homologous sulfotransferase models, not a CHST3 crystal structure. Skeletal disease is consequential biology; no additional developmental NEW process is inferred from necessity alone.
Rows 4/5 remain UNDECIDED: GO0001517 specifically names GlcNAc, whereas the directly resolved human reaction modifies GalNAc and the separate keratan result does not settle residue identity. Reactome2046222 genuinely includes CHST3 in the GlcNAc catalyst set. Its three cited papers were fetched normally and read at their available abstract scope; bounded publisher attempts did not recover full target-specific evidence. This supports a precise unresolved question, not a wrong-gene or miscitation assertion from incomplete evidence.
Row 8 preserves obsolete GO0006044. The official reason is an unnecessary grouping term and no replacement is supplied. No lateral process replacement is invented from the term’s obsoletion. Exact current PTN002473209/PTN002581850 source records and donor identities are audited, but historical PAINT node placement/MSA are not reconstructed. Donor counts and legitimate human self-evidence are not treated as weaknesses. MGI/RGD identities were resolved through exact UniProt crossreferences after initial id-search queries returned no hits; those query results are not biological negatives.
The remaining broad, correct sulfur/carbohydrate and sulfotransferase assertions are refined to the measured chondroitin reaction or biosynthetic step. The disease Reactome event describes loss of normal function and is not a claim that all mutant products retain catalysis. Golgi localization retains its original evidence scope; the recombinant soluble constructs do not independently image the native compartment. One principal GO0008459/GO0050650 core is used, with Golgi membrane location. No NEW annotation or process is proposed; the authenticated GO-CAM index has no target hit.
All three original PMID caches were absent from the pinned main and are genuine normal additions. The full 15215498 cache, abstract 9714738/9883891 caches, three additional Reactome-cited abstract caches, five event caches and normal source products are separately authenticated. Existing main Reactome2018682/3595175 and shared ontology/GO-CAM context are protected, not overwritten. The final applicability guard refreshes these identities before handoff.
Primary quotation accounting includes all YAML/notes repeats. Notes contain no primary quotations. YAML quotes are short literal anchors from 15215498 and 9883891; the final audit supplies exact per-source totals. Every original assertion has an ordinary review; the three UNDECIDED actions represent unresolved biological/source scope, not pending work or campaign holds.
The established reactions modify GalNAc C6 on chondroitin and, as curated in UniProt Q7LGC8, Gal C6 on keratan. Those positive findings do not constitute a negative GlcNAc assay. In the actual Wiley article, PMID:9883891, Methods 2.3–2.5 and Results 3.2 describe human soluble protein-A fusion assays. Tables 1–2 and enzymatic/HPLC product analysis resolve GalNAc-containing substrates. The separate statement that the enzyme “efficiently utilized corneal keratan sulfate” is explicitly data not shown. It supports keratan utilization, not a displayed GlcNAc-negative experiment.
This official indexed full text was genuinely accessed and read, although the unchanged normal cache contains only the abstract. The failed local HTTP 403 retrieval is a separate receipt. VERIFIED therefore records the authentic primary access and supporting relationship; setting full_text_unavailable would incorrectly describe the actual review access. The source record now gives the direct article URL and section/table locators without publishing local session paths.
Rows 4/5/8 remain UNDECIDED. Reactome 2046222 genuinely includes CHST3 in its GlcNAc-transfer catalyst set; its three linked papers remain abstract-limited in this review, and a target-specific assay has not been verified. The PAINT tree/MSA was not reconstructed, so the extant WITH/FROM list cannot establish a broad or narrow ancestral node. GO:0006044 was obsoleted as an unnecessary grouping, not because CHST3 was shown incapable of GlcNAc metabolism. These are bounded unresolved chemical/source assertions, distinct from the positively established principal activity.
The other PAINT comments now distinguish the supported original broad chemistry or biosynthetic role from uninspected historical placement. Target-specific MODIFY recommendations can converge on already supported terms: in particular, the PMID:9883891 sulfur-metabolism refinement converges on its existing chondroitin biosynthesis annotation. It neither adds a NEW row nor revises the original IDA or IBD source. The principal core terms and all annotation actions are unchanged.
PMID:15215498(https://pmc.ncbi.nlm.nih.gov/articles/PMC454181/), Sulfotransferase Activity of Fibroblasts, reports “no C6ST activity” in the patient cells. Its recombinant Results separately compare R304Q with WT, with matched expression checked by the authors. Short anchors now identify the mutant result and sulfate-transfer step rather than repeating a weak activity fragment. Whole-article figure pixels and complete numeric table bodies were not newly inspected.
All blanket annotation-to-artifact pointers have been removed. The compact artifact retains checkable official URLs, hashes and source-specific access boundaries, without raw database dumps, session paths or duplicate primary quotations. The previous artifact is preserved verbatim in the follow-up receipt. The earlier notes and history remain intact; this append supersedes their quotation-accounting and artifact-layout descriptions. Primary quotations are counted across YAML, the complete notes and the compact artifact, including repetitions; the artifact adds none.
The two GO:0001517 assertions remain UNDECIDED because the available evidence does not settle the GlcNAc-specific reaction. Positive GalNAc and Gal acceptors are established; their presence does not test exclusion of another acceptor. Reactome 2046222 genuinely includes CHST3 in its catalyst set, but the linked target-specific assays remain unverified at the available abstract scope. The genuine external full access to PMID:9883891 remains distinct from its unchanged abstract cache; the prior Methods, Results and table-read scope is unchanged.
The current PTHR10704 membership file places human CHST3 and the mouse, rat, chicken and ray orthologs in SF60. The five listed human GlcNAc-activity sources belong to SF3, SF4, SF5, SF13 and SF40, while the listed chondroitin-activity sources belong to SF60. This is a verified current subfamily partition. It does not locate the historical IBD assertion relative to those subfamilies, establish when that function arose, or demonstrate target-specific loss. The chemical uncertainty and the historical placement uncertainty are separate; neither a donor count nor a subfamily list supplies the missing phylogenetic inference.
Row 8 now recommends REMOVE for GO:0006044. The live official record says the grouping was created in error and is unnecessary, with no replacement. This withdraws an obsolete annotation target independently of the substrate question. The machine identifier and source fields remain intact; no replacement process or negative GlcNAc chemistry is asserted. This supersedes the earlier row-8 UNDECIDED disposition without rewriting its historical notes.
The keratan rows now cite the genuine fetched UniProt Q7LGC8 record directly. Its FUNCTION and EC sections attribute the secondary Gal/keratan reaction to PMID:9714738. The short source anchors make that curated residue assignment checkable locally; the primary citations and their actual access limits remain. The patient-cell row also points directly to the existing full-text fibroblast result in PMID:15215498, alongside the recombinant assay evidence already cited. No additional primary access or experiment is claimed.
The description is biological prose, while source attribution and propagation limits remain here and in the relevant reviews. The compact JSON source record is retained; changing its serialization would not improve the evidence. Obsolete bookkeeping is removed, and the new live ontology locator is recorded. Existing broader-to-specific target recommendations still reflect the measured chondroitin chemistry without altering original IBD source assertions.
All 29 machine assertions, the original 18 reference identities, the absence of product objects, the molecular core and every source cache are preserved. One fetched-source reference is added. Both previous histories and the complete earlier notes remain intact. As already corrected by the first follow-up, the historical claim that notes contained no quotations is superseded: aggregate source totals across all authored YAML, the complete notes and the artifact are PMID:15215498 = 23 words, PMID:9883891 = 10 words and fetched UniProt Q7LGC8 = 21 words. The artifact contains no repeated anchors.