GLT8D1

Reason: Full-length FLAG-tagged human WT GLT8D1 colocalized with a Golgi marker in transfected HEK293T cells; I290M instead accumulated in the ER. This supports the curated Golgi placement, with membrane association corroborated by the predicted UniProt signal anchor. Microscopy does not resolve bilayer topology, and the result is not generalized to every isoform or variant (PMID:34746377).

Reason: The UniProt SL-0134 mapping agrees with WT human Golgi localization and the predicted N-terminal signal anchor. Retain Golgi membrane without treating microscopy as direct bilayer-topology measurement or generalizing to the short isoform.

Reason: PAINT assigns Golgi apparatus activity to eukaryotic AN1/PTN001012071; the target descends from this node. The human entry is legitimately among the experimental descendants used by the curator, alongside mapped Arabidopsis GATL proteins with Golgi IDA records. This supports conserved Golgi activity and agrees with direct GLT8D1 localization; keep the incoming is_active_in qualifier distinct from located_in membrane annotations.

Reason: Golgi membrane (GO:0000139), independently supported elsewhere in this review, entails membrane. Mark this broad term TOO_GENERAL. The YTS proteomics target table and peptides remain inaccessible; this disposition does not verify the original HDA identification, plasma-membrane residence or isoform specificity (PMID:19946888).

Reason: Refine to UDP-galactosyltransferase activity: human 52–371 transferred galactose from UDP-galactose in LC–MS/MS assays. Native substrates, linkage and natural short-isoform function remain unresolved.

Proposed replacements: UDP-galactosyltransferase activity

Reason: PAINT places the positive glucosyltransferase assertion at PTHR13778 AN0/PTN004435860, ancestral to human AN57/PTN002492649. Its experimentally grounded donors are bacterial LPS enzymes P19817 (Salmonella waaJ), P27128 (E. coli waaO) and P27129 (E. coli waaJ). The separate bacteria-restricted LPS-process assertion does not taxonomically restrict this ancestral molecular function. Full-length UDP-glucose assays measure UDP release without a defined acceptor or identified glucosylated product (PMID:35198895). Soluble human 52–371 did not hydrolyze UDP-glucose; the panel measured hydrolysis, not glucose transfer (PMID:38066107). Retain this ancestral annotation as a non-core additional specificity.