Mitochondrion–ER Membrane Tethering — Obsoletion & Replacement (GO:1990456)

Bottom line: Mitochondria and the ER are held together at contact sites by
tether proteins such as the yeast ERMES complex and mammalian PDZD8. GO has
obsoleted the process term GO:1990456 mitochondrion-endoplasmic reticulum
membrane tethering because it describes a molecular function, and points
curators to GO:0140474 mitochondrion-endoplasmic reticulum membrane tether
activity (OLS, checked 2026-09-26). We listed the 30 upstream annotations
by group (SGD, UniProt and FlyBase done; CACAO, ComplexPortal, MGI
and dictyBase pending) and the three mappings to re-point, and found five
reviews in this repo that carry the old term: human VMP1 and CALM1 (one IDA
row each from PMID:28890335, both ACCEPT) and mouse Calm1, Calm2 and Calm3
(electronic or similarity rows, KEEP_AS_NON_CORE). Scoped, not yet started:
none of those rows has been revisited since the obsoletion, and none of the
canonical tethers (PDZD8, MFN2, VAPB, RMDN3, ERMES subunits) is reviewed here.
The impact table below reports the mouse calmodulin actions loosely; the
reviews mark them KEEP_AS_NON_CORE.

Overview

A GO obsoletion proposal will retire GO:1990456 mitochondrion-endoplasmic
reticulum membrane tethering (a BP term). The rationale matches the parallel
ER–PM and peroxisome–chloroplast tether obsoletions: the activity is more
appropriately captured at the molecular function level by the existing
GO:0140474 mitochondrion-endoplasmic reticulum membrane tether activity.

This project tracks the impact on AI Gene Review and queues the canonical
mitochondrion–ER tether proteins for review.

Upstream tickets

• Annotation tracker: geneontology/go-annotation#6397
• Ontology ticket: geneontology/go-ontology#31875

Obsoletion plan (per upstream)

Affected upstream groups (from issue body)

Mappings to be reviewed

Impact on this repo

The obsoletion intersects existing reviews in the following way:

These reviews will need a refresh once the obsoletion lands — for VMP1 and the
calmodulin family the GO:1990456 annotation should be reassessed against
GO:0140474 (or, for calmodulins, likely REMOVED because the evidence does
not show that calmodulin itself is a mitochondrion–ER tether). Human VMP1 and
CALM1 carry UniProt IDA annotations from PMID:28890335. Mouse Calm1/2/3 carry
computational or similarity-propagated annotations (IEA/ISO/ISS), not Reactome
TAS annotations.

Human CALM2 is not currently present in this repository, and human CALM3 does
not currently have a GO:1990456 row in either GOA or the review YAML.

No canonical mito–ER tether genes (ERMES components, PDZD8, MFN2,
VAPB/PTPIP51) are currently reviewed in this repository.

Scope

• Organisms: human (PDZD8, MFN2, VAPB, PTPIP51), S. cerevisiae (ERMES
  components: MMM1, MDM10, MDM12, MDM34), Dictyostelium (5 dictyBase
  annotations still pending), mouse (1 MGI annotation pending).
• GO branches: BP (the obsoleted term itself) and MF GO:0140474. Both
  belong to the membrane contact site (MCS) branch.
• Type of fix: terminological in GO; biology is well established. Reviews
  should evaluate whether the MF replacement (tether activity) is the better
  core-function term and add it where the underlying assay supports it.

Candidate genes for initial review

Verify each with just fetch-gene <organism> <gene> before starting; do not
add files without confirming the UniProt accession from the UniProt API.

Mammalian mito–ER tethers

1. PDZD8 (human) — PDZ domain-containing protein 8. Direct target of the
   InterPro2GO mapping (IPR039275). Vertebrate functional analog of yeast
   Mmm1; established mito–ER tether (Hirabayashi et al. 2017, Science).
2. MFN2 (human) — Mitofusin-2. Long-debated mito–ER tether at MAMs
   (de Brito & Scorrano 2008, Nature), with substantial counter-evidence that
   MFN2 can instead act as a negative regulator of ER–mitochondria proximity
   (Filadi et al. 2015, PNAS).
3. VAPB (human) — VAMP-associated protein B/C. Forms a tether with
   PTPIP51 at MAMs (Stoica et al. 2014, Nat Commun).
4. PTPIP51 / RMDN3 (human) — partner of VAPB at MAMs.

Yeast ERMES complex (S. cerevisiae)

1. MMM1 — ER-anchored ERMES SMP-domain subunit. MMM1 is a sister ERMES
   subunit and should not be treated as the direct HAMAP MF_03102 target.
2. MDM10 — outer mitochondrial membrane β-barrel; ERMES + TOM-assembly
   bridge. Direct target of HAMAP MF_03102.
3. MDM12 — cytosolic SMP-domain subunit bridging Mmm1 and Mdm34.
4. MDM34 — outer mitochondrial membrane SMP-domain subunit.

Yeast SGD has already migrated its annotations upstream; review here would
serve to capture the canonical tether MF for these reference proteins rather
than to fix outstanding GOA rows.

Other organisms

1. The 5 dictyBase pending annotations are likely to be the Dictyostelium
   ERMES homologs — review opportunistically once the gene set is confirmed.

Proposed approach and priority

1. Refresh existing affected reviews (human VMP1 and CALM1; mouse
   Calm1/2/3) to revisit GO:1990456 rows once the obsoletion lands. The
   calmodulin rows likely should be REMOVED rather than rerouted, but the
   rationale should distinguish the human UniProt IDA annotation from the mouse
   computational/similarity-propagated annotations.
2. Anchor new reviews on PDZD8 + MFN2 as the highest-impact
   mammalian tethers, and on MMM1 + MDM12 for the yeast ERMES anchor
   pair.
3. Once GO:0140474 is the canonical MF term, audit other genes in this
   repository whose core_functions could legitimately gain it (e.g. through
   the proposed_replacement_terms mechanism).

Status

• Upstream FlyBase, SGD, and UniProt updates are done; CACAO (1),
  ComplexPortal (9), MGI (1), and dictyBase (5) experimental annotations are
  still pending.
• Mappings (InterPro2GO IPR039275, HAMAP MF_03102, UniRule UR000106107) all
  need to be re-pointed to GO:0140474.
• No reviews in this repo are blocked by the obsoletion; the impact is a
  refresh queue (human VMP1/CALM1 plus mouse Calm1/2/3) and a forward-looking
  review queue (PDZD8, MFN2, VAPB, PTPIP51, ERMES components).
• Recommend starting with PDZD8 as the anchor review, since it is both
  the InterPro2GO mapping target and the most recently mechanistically
  characterized mammalian mito–ER tether.
• 2026-09-26 — OLS lists GO:1990456 as obsolete ("this term represents a
  molecular function"), pointing to GO:0140474. The five affected reviews
  still carry their pre-obsoletion actions (human VMP1 and CALM1 ACCEPT;
  mouse Calm1/2/3 KEEP_AS_NON_CORE). The CALM1 row is an IDA from
  PMID:28890335, but its review text is generic Reactome/TAS boilerplate, so
  it needs a real assessment when refreshed. The local
  cache/ontologies/go.tsv still records GO:1990456 as live, so validation
  does not flag these five reviews yet.

Slides

• Slides (Marp source: MITO_ER_TETHERING_OBSOLETION-slides.md) — AI generated