GAL3ST1 (Q99999) — review notes

Reviewed alongside SULT1B1 as the "negative control" pair of the contested-functions batch.
Headline finding: the 2026 histone-tyrosine-sulfotransferase claim has NOT entered GOA for
GAL3ST1, and the established sulfatide-synthase annotations are sound.

1. Established function: galactosylceramide 3'-O-sulfotransferase (cerebroside sulfotransferase, CST)

GAL3ST1 is the single gene responsible for sulfoglycolipid synthesis in mammals. It is a type II
single-pass Golgi membrane protein that transfers sulfate from PAPS to the 3-OH of the non-reducing
terminal beta-galactose of galactolipids, producing sulfatide (3-O-sulfogalactosylceramide) and
seminolipid.

• Purification from human renal cancer cells, EC 2.8.2.11:
  PMID:8830034
• Acceptor specificity (this is what grounds both the sphingolipid and the glycerolipid BP
  annotations):
  PMID:8830034
  PMID:8830034
  GalAAG (galactosyl alkylacylglycerol) and GalDG (galactosyl diacylglycerol) are glycerolipids, so
  the GO:0046486 glycerolipid metabolic process IDA is correctly grounded, not a stray.
• cDNA cloning and heterologous expression:
  PMID:9030544
  PMID:9030544
• Physiology, from the mouse Cst (Gal3st1) knockout — one gene makes both sulfoglycolipids, and its
  loss causes paranodal/myelin and spermatogenic failure:
  PMID:11917099
  PMID:11917099
  PMID:11917099
  This is the experimental basis behind the GO:0042552 myelination IBA (WITH/FROM MGI:MGI:1858277 =
  mouse Gal3st1), so that IBA rests on a real, strong descendant annotation and is accepted.
• UniProt places the protein in the Golgi apparatus membrane as a single-pass type II membrane
  protein (TRANSMEM 15..35, TOPO_DOM 36..423 lumenal), with PATHWAY "Lipid metabolism; sphingolipid
  metabolism".

Two legacy TAS rows I removed

Both are 2003 PINC-era TAS annotations traced to PMID:9030544, and both misread the paper:

• GO:0005886 plasma membrane — 9030544 reports that the product sulfatide "was expressed on the
  transformed cells", i.e. the glycolipid reaches the cell surface. The enzyme is a Golgi type II
  membrane protein. Wrong entity.
• GO:0006487 protein N-linked glycosylation — derives from "contains two potential N-glycosylation
  sites". GAL3ST1 is a substrate of N-glycosylation, not a participant in the glycosylation
  machinery. Classic substrate/enzyme inversion.

2. The 2026 histone-sulfation claim — and why GOA has not taken the bait

PMID:41686426
PMID:41686426

Why this is hard to accept as it stands:

1. It conflicts with the enzyme's own cell biology. GAL3ST1 is a type II Golgi membrane protein
   whose catalytic domain faces the Golgi lumen. The claimed reaction is on nascent histone H3 in
   the cytosol. The paper does not resolve how a lumenal catalytic domain reaches a cytosolic
   substrate; it reports an in vitro assay with purified His-GAL3ST1 and immunoblot readout
   ("Purified recombinant His-GAL3ST1 protein was incubated with purified histone H3 in HST buffer
   (50 mmol/L Tris-HCl and 15 mmol/L MgCl2, pH 7.5) at 37°C... H3Y99sulf levels were determined by
   immunoblotting"), plus AlphaFold/HADDOCK docking.
2. The mark itself is disputed. The H3Y99sulf modification was defined in PMID:36805701 and
   formally challenged in PMID:40890505, which reannotated the original raw MS data, showed the
   spectra fit a phosphotyrosine peptide far better than a sulfotyrosine one, and found no
   sulfation of recombinant histone H3.2 by recombinant SULT1B1 in an assay that did sulfate T3:
   PMID:40890505
3. The detection reagent is the one under question. The refutation's specific warning is that the
   anti-H3Y99sulf antibody was validated only on an N-terminal histone peptide array and that CUT&Tag
   signal could reflect phosphorylation:
   PMID:40890505
   PMID:41686426's H3Y99sulf readouts are immunoblot/CUT&RUN with that class of antibody.
4. The refutation is not cited. I grepped the full text of PMID:41686426 for "Youssef" and
   "refute": no hits. The paper cites PMID:36805701 (its ref 13) as settled background.
5. It also conflicts with the original attribution. PMID:36805701 assigns the writer role to
   SULT1B1; PMID:41686426 assigns it to GAL3ST1 and notes SULT1B1 levels were unchanged
   PMID:41686426. So there are now two incompatible enzyme assignments for a mark whose
   existence a third paper denies.

GOA status check (2026-09-17): GAL3ST1-goa.tsv contains no histone, chromatin, nucleus,
cytosol or protein-tyrosine-sulfotransferase annotation of any kind; every row is
galactolipid/sulfotransferase/Golgi. The 2026 claim has not been curated. Recorded here as the
finding, not as an annotation change.

3. Is there a GO term for this at all?

Yes — and that is what makes the absence informative rather than a vocabulary gap.
GO:0008476 protein-tyrosine sulfotransferase activity and GO:0006478 peptidyl-tyrosine sulfation
both exist. A QuickGO query (goId=GO:0008476, goUsage=exact, taxonId=9606, retrieved 2026-09-17)
returns 24 human rows, all of them TPST1 (O60507) or TPST2 (O60704) and their isoform accessions.
Neither GAL3ST1 nor SULT1B1 is annotated to it. The ontology could express the disputed claim; GO
has declined to.

4. Review decisions summary

Core: GO:0001733 galactosylceramide sulfotransferase activity in the Golgi membrane
(GO:0000139), driving sulfoglycolipid (sulfatide/seminolipid) biosynthesis and, downstream,
myelination. Accepted essentially unchanged. Small cleanups: two generic parents
(GO:0008146 sulfotransferase activity, GO:0016020 membrane, GO:0009247 glycolipid biosynthetic process) modified to specific terms; two legacy PINC TAS rows (GO:0005886 plasma membrane,
GO:0006487 protein N-linked glycosylation) removed as misreadings of PMID:9030544.
No annotation was added or changed on account of the 2026 histone-sulfation claim.