Protein Localization to ER Exit Site — Obsoletion

IN_PROGRESS OBSOLETIONFLAGSHIP

Species: human, yeast

Genes: LRRK2 BCAP31 YET2

Protein Localization to ER Exit Site — Obsoletion

Bottom line: GO has obsoleted the process term GO:0070973 protein
localization to endoplasmic reticulum exit site, because curators had used it
for four different things: COPII coat assembly (SEC16A), ER-to-Golgi transport
(MIA3, GBF1), regulation of that transport (LRRK2) and ER quality control
(Bcap29/Bcap31). There is no single replacement, so we mapped each of the 10
curated annotations to the term the upstream plan proposes (GO:0048208,
GO:0006888, GO:0060628, GO:0034976, or deletion) and checked which reviews in
this repo use the term. Two reviews annotate it. Human BCAP31 already marks
its IBA row MARK_AS_OVER_ANNOTATED, so its review needs no change. A third
reviewed gene, yeast YET2 (a BCAP29/BCAP31-family member descending from the
same PANTHER node PTN000294723), carries the same IBA in its cached UniProt
record but not in its GOA file or review; the obsoletion removes it. Human LRRK2 had
two GO:0070973 rows marked ACCEPT (IEA GO_REF:0000120 and IMP PMID:25201882). Both
now MODIFY to GO:0060628 regulation of ER to Golgi vesicle-mediated transport,
fixed in #3241 (merged).
The obsoletion has now landed (OLS lists GO:0070973 as obsolete, with
context-dependent replacements), so the "OPEN" ticket status below is out of
date.

The term also carried about 14,400 IEA and IBA annotations driven by one
UniRule mapping, so the cleanup reaches much further than the curated count
suggests.

Overview

A GO obsoletion proposal will obsolete the biological-process term
GO:0070973 protein localization to endoplasmic reticulum exit site
(defined as "A process in which a protein is transported to, or maintained in, a
location at an endoplasmic reticulum exit site"). The upstream rationale
(ValWood, go-ontology#19122) is that the term says nothing about the process
itself and conflates several distinct biological roles — it has been used for
COPII coat assembly factors, anterograde ER→Golgi transport cargo, regulators of
that transport, and ER quality-control proteins. Because the existing
annotations sit in different pathway contexts, a single blanket
replaced_by cannot be applied; each annotation needs an individually chosen
process term.

This term sits alongside the related "protein exit from endoplasmic reticulum"
family (GO:0032527 and its regulation children GO:0070861/0070862/0070863,
plus GO:1904211) that the same ontology ticket proposes to obsolete; that
sibling set is reviewed in go-annotation#6172. GO:0070973 was added to the same
ticket because it has the same conflation problem.

Upstream tickets

Obsoletion plan (per upstream)

There is no single replaced_by. The proposed per-annotation remapping
(ValWood, go-annotation/go-ontology#19122 comment, 2026-05-21) is:

Accession Symbol Source Organism Proposed action
O15027 SEC16A UniProt Human → GO:0048208 COPII vesicle coat assembly (syn. "COPII vesicle coating")
P34643 sec-16A.1 WB C. elegans → GO:0048208 COPII vesicle coat assembly
P48415 SEC16 SGD S. cerevisiae → GO:0048208 COPII vesicle coat assembly
Q5JRA6 MIA3 UniProt Human → GO:0006888 ER to Golgi vesicle-mediated transport
Q5S007 LRRK2 UniProt Human → GO:0060628 regulation of ER to Golgi vesicle-mediated transport
Q5S006 Lrrk2 UniProt Mouse delete — redundant; already carries GO:0060628 from the same paper (PMID:25201882)
Q57WC1 Tb03.28C22.610 GeneDB T. brucei delete — already carries GO:0006888 ER to Golgi vesicle-mediated transport
Q61334 Bcap29 MGI Mouse → ER quality control, likely GO:0034976 response to endoplasmic reticulum stress (or descendant)
Q61335 Bcap31 MGI Mouse → ER quality control, likely GO:0034976 response to endoplasmic reticulum stress (or descendant)
Q92538 GBF1 UniProt Human → vesicle-mediated transport or a descendant

ValWood notes that for the forward (ER-export) direction generally,
GO:0090114 COPII-coated vesicle budding (synonym "ER exit") is often the
best replacement, since it is the COPII vesicle that drives exit from the ER.

Term labels verified in OLS on 2026-05-29:

Affected experimental / curated annotations

Retrieved from the QuickGO annotation API on 2026-05-29 (goId=GO:0070973,
restricted to the named accessions). The curated (non-IEA) records:

# Source Accession Symbol Organism Evidence Qualifier Reference
1 UniProt O15027 SEC16A Human IMP involved_in PMID:28442536
2 WB P34643 sec-16A.1 C. elegans IMP involved_in PMID:21478858
3 SGD P48415 SEC16 S. cerevisiae IMP involved_in PMID:22675024
4 GeneDB Q57WC1 Tb03.28C22.610 T. brucei IMP acts_upstream_of_or_within PMID:24612401
5 UniProt Q5JRA6 MIA3 Human IMP involved_in PMID:28442536
6 UniProt Q5S006 Lrrk2 Mouse IMP involved_in PMID:25201882
7 UniProt Q5S007 LRRK2 Human IMP involved_in PMID:25201882
8 MGI Q61334 Bcap29 Mouse IGI acts_upstream_of_or_within PMID:15187134
9 MGI Q61335 Bcap31 Mouse IGI acts_upstream_of_or_within PMID:15187134
10 UniProt Q92538 GBF1 Human IMP involved_in PMID:17956946

The upstream group tally is GeneDB 1, MGI 2, SGD 1, WB 1, ~UniProt 7~ (the
UniProt portion is struck through in the issue, i.e. already handled by UniProt).
The minor count difference vs. the 10 rows above (UniProt tally of 7 vs. 5
distinct UniProt rows here) is likely spreadsheet rows for annotation extensions
or duplicates; flagged for the curator's awareness.

On top of these curated records, GO:0070973 currently carries ~14,414 total
annotations (QuickGO, 2026-05-29), overwhelmingly IEA (TreeGrafter
GO_REF:0000118; UniProt GO_REF:0000104/0000120) and IBA (GO_REF:0000033).
These will be retired/redirected automatically once the term is obsoleted, but
the volume shows how far a handful of curated annotations has propagated.

Why a blanket replaced_by does not work

The curated annotations fall into biologically distinct classes:

Mappings to review

Impact on this repo

Two affected gene products already have *-ai-review.yaml files that annotate
GO:0070973 (verified 2026-05-29). A later grep -rl GO:0070973 genes/ over
all file types adds a third reviewed gene, yeast YET2, whose only occurrence is
the cached UniProt DR block:

Gene Path Relation to affected set Current handling of GO:0070973 Action when obsoletion lands
LRRK2 (human, Q5S007) genes/human/LRRK2 Directly affected (row 7) Two annotations — IEA (GO_REF:0000120) and IMP (PMID:25201882) — were ACCEPT with summary "LRRK2 regulates Sec16A at ER exit sites" MODIFY → GO:0060628 regulation of ER to Golgi vesicle-mediated transport, matching the upstream Q5S007 plan. Done in #3241 (merged): both rows MODIFY, supporting text now quotes the PMID:25201882 abstract sentence on anchoring Sec16A at ERES
BCAP31 (human, Q61335 is the mouse ortholog) genes/human/BCAP31 Ortholog of affected mouse Bcap31 (row 9); the human record's GO:0070973 is an IBA over-propagation IBA already MARK_AS_OVER_ANNOTATED per go-annotation#6385 (over-propagation from PANTHER PTN000294723; core role is ER QC / translocon chaperone) No change needed — already flagged; obsoletion removes the IBA. The curated mouse Bcap29/Bcap31 records remap to GO:0034976, consistent with the existing "ER quality control" rationale
YET2 (yeast, YMR040W) genes/yeast/YET2 BCAP29/BCAP31-family member; same PTN000294723 node (its GOA file carries three other IBA rows from it) Only in YET2-uniprot.txt:86 (IBA:GO_Central); absent from YET2-goa.tsv, so the review has no row for it No change needed — same over-propagation BCAP31 flags; the obsoletion removes it

The human LRRK2 MODIFY pass is done in #3241 (merged); the
human BCAP31 review is already aligned with the upstream direction, and the
YET2 IBA disappears with the term.

Scope

Candidate genes for initial review

Listed in priority order. The first two already have reviews here and would be
directly exercised by the obsoletion.

  1. LRRK2 (human, Q5S007) — already reviewed; the MODIFY of GO:0070973 →
    GO:0060628 (regulation of ER→Golgi transport) is fixed in #3241 (merged).
  2. BCAP31 (human) — already reviewed and MARK_AS_OVER_ANNOTATED; confirm
    alignment once obsoletion lands.
  3. SEC16A (human, O15027) — clean COPII coat-assembly case → GO:0048208;
    not yet in this repo. A good positive control for the COPII-coating remap.
  4. MIA3 / TANGO1 (human, Q5JRA6) — anterograde transport of bulky cargo →
    GO:0006888; not yet in this repo.
  5. GBF1 (human, Q92538) — Arf-GEF at the ER-Golgi interface → vesicle-mediated
    transport; not yet in this repo.
  6. SEC16 (yeast, P48415) — canonical ER-exit-site organizer; cross-check
    against the human SEC16A handling.

Proposed approach

  1. Track the obsoletion. go-ontology#19122 is OPEN and labeled "ready"; the
    replacement terms already exist in GO (GO:0048208, GO:0090114,
    GO:0006888, GO:0060628, GO:0034976), so no NTR blocks this — unlike the
    mitochondrion-targeting-sequence-binding case.
  2. Apply the LRRK2 MODIFY (GO:0070973 → GO:0060628). Both LRRK2 GO:0070973
    annotations (IEA and IMP) move together; fixed in #3241 (merged).
  3. Confirm BCAP31 stays MARK_AS_OVER_ANNOTATED; the curated mouse
    Bcap29/Bcap31 remap to GO:0034976 is consistent with the human review.
  4. Optionally seed SEC16A / MIA3 / GBF1 reviews as the clean COPII / secretory
    cases, since they exercise the three main replacement terms.
  5. Flag the UniRule mapping (UR001349783) for retargeting/removal to stop
    IEA re-propagation after obsoletion.

Priority

Low–medium. Only ~10 curated annotations, replacement terms already exist, and
the two repo-resident genes (LRRK2, BCAP31) are already reviewed — the LRRK2
MODIFY is in #3241 (merged) and BCAP31 is already aligned. No curator group is
blocked waiting on AI Gene Review. The large IEA/IBA tail (driven by the
UniRule mapping) makes the cleanup more impactful than the small curated count
suggests.

Status

Slides