ST6GAL1 (human, P15907) review notes

Identity and core enzymology

Localization

Subunit / dimerization

CD75 caution

Biology (downstream / context-specific)

GOA annotation assessment summary (35 annotations)

Core molecular function: GO:0003835 beta-galactoside alpha-2,6-sialyltransferase activity. Many
duplicate rows (IBA, IEA, 3x TAS-Reactome, 3x EXP, 1 IDA) — all ACCEPT/KEEP; the EXP/IDA-backed ones
are the strongest. GO:0008373 sialyltransferase activity (parent) is correct but over-general -> MODIFY
to the specific GO:0003835.

Process: GO:0097503 sialylation and GO:0006487 protein N-linked glycosylation are accurate and
specific (alpha-2,6-sialylation is the terminal capping step of complex N-glycan biosynthesis) ->
ACCEPT. GO:0009101 glycoprotein biosynthetic process is a vaguer parent -> KEEP_AS_NON_CORE.
GO:0006054 N-acetylneuraminate metabolic process (IDA, PMID:23999306) -> ACCEPT as a correct, slightly
broad sibling describing sialic-acid utilization.

GO:0016266 protein O-linked glycosylation via N-acetylgalactosamine (TAS Reactome sTn / O-glycan
termination): ST6GAL1 is primarily an N-glycan enzyme but Reactome documents it adding sialic acid to
Tn antigens forming sTn (O-GalNAc) and terminating Core1/2 O-glycans. Keep but as non-core (minor /
context-specific activity relative to the dominant N-glycan role).

GO:0019082 viral protein processing (2x TAS Reactome, SARS-CoV-2 3a/spike sialylation): this is a
generic host-glycosylation step the Reactome curators slotted into viral life-cycle pathways. The MF
(sialylation) is real but "viral protein processing" mis-frames a housekeeping glycosylation step as a
dedicated viral function -> MARK_AS_OVER_ANNOTATED.

Locations: GO:0005794 Golgi apparatus (IBA), GO:0000139 Golgi membrane (IDA PMID:20378551 + 5x TAS),
GO:0032580 Golgi cisterna membrane (IEA) -> ACCEPT (core). GO:0005576 extracellular region (IEA + ISS
from rat ortholog P13721) -> KEEP_AS_NON_CORE (the soluble secreted form is real but its serum/extrinsic
activity is secondary to the Golgi-resident core function).

GO:0042803 protein homodimerization activity (IDA, PMID:20378551) -> ACCEPT (well supported by BiFC;
biologically real, organizes the enzyme in the Golgi). Not the core catalytic MF but a genuine MF.

GO:0005515 protein binding (IPI, PMID:16237761): from a yeast two-hybrid screen for hepatocyte proteins
binding HCV F protein; "1 colony was sialyltransferase" among 36 colonies. Generic, uninformative
protein-binding with weak Y2H evidence and no biological follow-up for ST6GAL1. Per curation guideline
to avoid generic "protein binding" -> REMOVE.

Falcon integration (2026-06-21)

Used from Falcon report (after verifying against UniProt / cached pubs):
- Core enzymology framing (GT29, trans-Golgi, CMP-Sia donor, alpha-2,6 linkage on complex N-glycans) —
matches UniProt and PMID:23999306. USED.
- Soluble/secreted form via BACE1 shedding; serum source; extrinsic sialylation — consistent with
UniProt "Secreted" + Note. USED as background; supports KEEP_AS_NON_CORE for extracellular region.
- IgG Fc N297 sialylation and B-cell/CD22 immune-regulatory framing — consistent with UniProt FUNCTION
and cached PMID:25733881/27872474. USED as biological context (KEEP_AS_NON_CORE territory).
- Context-specific cancer/inflammation substrate biology (EGFR, TNFR1, TLR4, integrin beta1, CSF1R) —
USED only as downstream-context rationale, not as separable core functions. Treated as KEEP_AS_NON_CORE
conceptually; none correspond to a current GOA row here, so they inform notes/questions, not actions.

Rejected / not used as citations:
- Falcon repeatedly cites "EC 2.4.99.1" for ST6GAL1. REJECTED: UniProt/IUBMB current EC for ST6GAL1 is
EC 2.4.3.1 (the 2.4.99.x sialyltransferase EC class was renumbered to 2.4.3.x). Did not propagate the
stale EC number.
- All Falcon citations use opaque internal keys (e.g. "sankaranarayanan2023computationalstudieson",
"lau2024sialicacidin") rather than PMIDs/DOIs, and several are preprints (bioRxiv, Authorea) with 0
citations. REJECTED as direct citations: none were added to the YAML references; only claims that I
could independently anchor to UniProt or a cached PMID were used.
- Falcon's neuronal / synaptic-remodeling / pyroptosis / autophagy / developmental-cell-death
sections: Falcon itself concludes "no direct evidence" for these. AGREED and REJECTED for annotation —
none proposed, consistent with the absence of such GOA rows.
- Falcon's long table of context-specific protein substrates (MET, HER2, VEGFR, PODXL, ICAM1, ALCAM1,
ECE1, etc.) is mostly review-/preprint-level and rated Moderate/Limited even by Falcon. NOT used to
author any GO term or substrate claim.