GTT3 (YEL017W / P39996) — curation notes

Journal for the AI GO-annotation review of Saccharomyces cerevisiae GTT3, an
understudied ("dark") gene. Provenance is recorded inline as
[PMID:xxxx "verbatim quote"] or [file:... "quote"].

Identity (from UniProt P39996, GTT3-uniprot.txt)

• Standard name GTT3; systematic YEL017W; UniProtKB P39996 (GTT3_YEAST),
  337 aa, 38.2 kDa. RecName "Glutathione transferase 3".
• PE 1 (evidence at protein level) — detected in proteomics (phospho-MS,
  PaxDb/PeptideAtlas), so the protein is genuinely expressed
  [GTT3-uniprot.txt "PE 1: Evidence at protein level"; "Present with 1300 molecules/cell in log phase SD medium."].

Domain / topology reasoning (inline, from GTT3-uniprot.txt)

• Family/domain signatures are GTT3-specific and named "putative": Pfam
  PF27945 (GTT3_N), PANTHER PTHR41807 ("GLUTATHIONE TRANSFERASE 3"),
  InterPro IPR038872 — whose InterPro name is "Put_GTT3" = putative GTT3
  [GTT3-uniprot.txt "InterPro; IPR038872; Put_GTT3." / "Pfam; PF27945; GTT3_N; 1."].
  These are not the canonical soluble-GST signatures (no thioredoxin/GST N-
  and C-terminal domain PROSITE/Pfam hits, no PF00043/PF02798). So the "glutathione
  transferase" name is an annotation-by-name, not a domain-supported catalytic call.
• Membrane topology is the strongest structural fact. UniProt models a large
  N-terminal cytoplasmic region (1-239) then two transmembrane helices
  (TRANSMEM 240-260 and 314-336) flanking a short perinuclear-space loop (261-313)
  [GTT3-uniprot.txt "TRANSMEM 240..260"; "TRANSMEM 314..336";
  "TOPO_DOM 261..313 /note=Perinuclear space"]. Multi-pass membrane protein.
  Canonical cytosolic GSTs (Gtt1/Gtt2, Gto1-3) are soluble, so a tail-anchored /
  multi-pass topology is unusual for a GST and argues against classic soluble-GST
  catalysis.
• The cytoplasmic N-terminus contains two disordered/low-complexity regions
  (66-95, 107-132; polar-residue biased) carrying multiple Cdk1/phospho sites
  (Ser-66/72/99/116) [GTT3-uniprot.txt "MobiDB-lite" REGIONs; MOD_RES phosphoserines
  from PMID:17330950, PMID:19779198]. Phospho-regulation is consistent with a
  regulated membrane protein but does not by itself assign a molecular function.

Localization (KNOWN, high-throughput / experimental)

• Nuclear membrane / multi-pass membrane protein — UniProt SUBCELLULAR LOCATION,
  from the Huh et al. global GFP screen
  PMID:14562095.
• Nuclear periphery (HDA) — SGD annotation GO:0034399 from the SWAp-Tag N'-GFP
  library screen [PMID:26928762; localization category "Nuclear periphery" is one of the
  assigned classes: GTT3-goa.tsv row GO:0034399 HDA PMID:26928762].
• Membrane (IBA) — GO:0016020, phylogenetic (GO_Central) inference.
• Two orthogonal high-throughput localizations agree on nuclear-envelope / perinuclear
  membrane. nuclear periphery (a nuclear-lumen-proximal region) and nuclear membrane
  are consistent with a nuclear-envelope multi-pass protein.

Molecular function (NOT known)

• GOA carries GO:0003674 molecular_function, ND (no data) — SGD explicitly records
  no molecular-function data [GTT3-goa.tsv GO:0003674 ND GO_REF:0000015].
• Deep research: no direct enzymatic assay, substrate specificity, catalytic mechanism,
  or localization study for Gtt3p was found
  [file:yeast/GTT3/GTT3-deep-research-falcon.md "no direct enzymatic assay, substrate
  specificity, or catalytic mechanism for the specific YEL017W protein was retrieved"].
• The GST-family literature (Gtt1/Gtt2/Gto1-3) is about the paralogs, not GTT3;
  Gtt1/Gtt2 are soluble ER/mitochondrial CDNB-active GSTs induced by oxidative stress
  [file:...falcon.md "Gtt1 and Gtt2 ... catalyze the enzymatic conjugation of glutathione
  (GSH) with a multitude of ... toxic compounds"]. GTT3 has the distinct Put_GTT3/GTT3_N
  domains and does not share these signatures.

Biological process (NOT known)

• GOA carries GO:0008150 biological_process, ND [GTT3-goa.tsv].
• GTT3 is not co-regulated with the oxidative-stress GSTs: in a frataxin-deficient
  Friedreich's-ataxia yeast model its transcription is essentially unchanged
  [file:...falcon.md "GTT3 transcription was not detectably altered in the
  frataxin-deficient oxidative-stress model, unlike some other glutathione-related genes"],
  whereas Gtt1/Gtt2 are induced by H2O2 / cumene hydroperoxide. This weakens the
  assumption that GTT3 belongs to the same oxidative-stress detox pathway.
• The only deletion phenotype found is indirect: a SCRaMbLE synthetic-chromosome-V study
  where a YEL017W-containing deletion set increased prodeoxyviolacein output, mechanism
  unresolved and confounded by co-deletions [file:...falcon.md "the exact mechanism by
  which YEL017W deletion increases PDV productivity was not elucidated ... difficult to
  attribute the phenotype solely to loss of GTT3"]. Not curatable as a specific GTT3 role.

Caveats about the falcon report

• The falcon report states "285 amino acids" — this is wrong (UniProt P39996 is
  337 aa). I do not propagate the length claim.
• Falcon could not retrieve the Huh et al. full text; I rely on the cached UniProt
  SUBCELLULAR LOCATION + GOA HDA rows for localization, which are independent of falcon.

Curation decisions (summary)

• GO:0016020 membrane (IBA) — ACCEPT (broad but correct; multi-pass TM helices).
• GO:0031965 nuclear membrane (IEA, SubCell) — ACCEPT (Huh GFP + SubCell mapping;
  agrees with nuclear-periphery HDA). Core location.
• GO:0034399 nuclear periphery (HDA, PMID:26928762) — ACCEPT (experimental HT
  localization; do not overrule a curator's HDA call).
• GO:0003674 molecular_function ND — ACCEPT. No demonstrated activity; explicitly
  do NOT infer "glutathione transferase activity" (GO:0004364) — the name is
  annotation-by-analogy, domains are Put_GTT3-specific, no substrate demonstrated.
• GO:0008150 biological_process ND — ACCEPT. No specific process demonstrated;
  do NOT infer oxidative-stress/glutathione-metabolism from the paralogs.

Knowledge gap (primary deliverable)

Gtt3 is MF-dark and BP-dark with a solid CC: its subcellular home (nuclear-envelope /
perinuclear multi-pass membrane) is established by two orthogonal screens, but its
molecular activity, substrate, and biological role are entirely unknown. The
"glutathione transferase" name rests on a family/name assignment (PTHR41807 / Put_GTT3),
not on any demonstrated GSH-conjugation activity; the GTT3-specific domains and the
multi-pass membrane topology distinguish it from the soluble Gtt1/Gtt2/Gto GSTs, and it
is not co-regulated with them under oxidative stress. Whether Gtt3 is even a bona fide
glutathione transferase is unresolved.