MTCH2 (Q9Y6C9) review notes

Gene: MTCH2 (mitochondrial carrier homolog 2; synonyms MIMP, HSPC032). Human.
UniProt Q9Y6C9, 303 aa. Paralogue: MTCH1 (Q9NZJ7) — reviewed separately in
genes/human/MTCH1/MTCH1-ai-review.yaml; the two reviews are written to agree.

2026-09-17 update

This file was created as part of an update that folded four papers postdating the
previous version of the review into it: PMID:42308315 (Sci Adv 2026, structure),
PMID:42056306 (Nat Struct Mol Biol 2026, BAX/BAK), PMID:41044057 (Nat Commun 2025,
CPT1) and PMID:40704594 (J Cell Sci 2025, yeast MIM complementation).

1. Core molecular function: outer-membrane protein insertase

Established by Guna et al. 2022 with genome-wide CRISPRi, in vitro insertion into
mitochondria from knockout cells, and reconstitution of purified protein into
proteoliposomes
PMID:36264797. Substrate range is alpha-helical and broad
PMID:36264797; beta-barrels are not MTCH2 substrates.

Structural mechanism (new)

PMID:42308315

PMID:42308315

PMID:42308315

PMID:42308315

This settles the long-running topology disagreement (five TMs, not six) and gives the
activity a physical basis — a groove that lowers the barrier to moving a soluble domain
across the bilayer — rather than leaving "insertase" as a phenotypic label.

2. The carrier question: name asserts what nobody has shown

GOA carries no transporter molecular-function term for MTCH2. The only MF terms in
MTCH2-goa.tsv are GO:0032977 (membrane insertase activity, IDA) and GO:0005515 (protein
binding, IPI x3). That absence is the correct and defensible position.

However, the name "Mitochondrial carrier homolog 2" and the UniProt record push the other
way: KW Transport and DR TCDB; 2.A.29.25.2; the mitochondrial carrier (mc) family
[file:human/MTCH2/MTCH2-uniprot.txt "Belongs to the mitochondrial carrier (TC 2.A.29)
family."]. UniProt's GO cross-references also include an Ensembl-derived
GO:0042775 (mitochondrial ATP synthesis coupled electron transport) that does not appear
in the QuickGO GOA file.

The structural work argues the transport machinery is gone, not merely unused
PMID:42308315 and, importantly, restoring them does not help insertion
PMID:42308315.

No transport assay on purified MTCH2 has ever been published, positive or negative, so
"transport lost" is an inference from structure. Recorded as a knowledge_gap and a
suggested_experiment rather than asserted. This matches how the MTCH1 review handles the
same question.

3. Apoptosis: NOT downstream of the insertase activity

The previous version of this review assumed the apoptotic role was secondary to insertion
("Apoptotic role may be secondary to its insertase activity"; "Likely downstream of
insertase function"). That assumption is refuted.

PMID:42056306

PMID:42056306

Three separate controls make that stick:

  1. Insertase-dead mutant PMID:42056306.
  2. Paralogue control PMID:42056306.
  3. Lipid rescue, pointing at the mechanism
    PMID:42056306.

The authors are appropriately hedged about how absolute this is
PMID:42056306.

It is also distinct from the older tBID-recruitment role
PMID:42056306, and it has downstream
consequences beyond cell death
PMID:42056306.

Curation consequence. GO:0043065 stays KEEP_AS_NON_CORE, but for a different reason
than before: not because it is a knock-on effect, but because it is a regulatory biological
process whose molecular function is undefined. There is nothing to put in
core_functions.molecular_function for it. Recorded as a knowledge_gap (MF_DARK).

4. Lipid metabolism: a real mechanism, not just a knockout phenotype

PMID:41044057

PMID:41044057

This upgrades GO:0055088 (lipid homeostasis) from an unexplained ortholog-transferred
phenotype to a role with a named partner. It is still kept non-core: whether the CPT1
effect survives an insertase-dead MTCH2 has not been tested, so it cannot yet be called a
separate activity. Note the thematic convergence with the LPA/MFN2 fusion axis and the
LPA rescue of BAX/BAK assembly — three phenotypes all pointing at MTCH2 and mitochondrial
lipid handling.

5. MTCH1 vs MTCH2 and the yeast complementation asymmetry

PMID:40704594

This is awkward for a framing in which MTCH2 is the insertase, but it does not contradict
MTCH2 insertase activity. The authors attribute the failure to toxicity and mislocalisation
in yeast, not to inactivity:

PMID:40704594

PMID:40704594

MTCH2's insertase activity rests on reconstitution with purified protein in proteoliposomes,
which a heterologous yeast growth assay simply does not address. Conversely, MTCH1 partially
rescues MTCH2 loss in human cells
PMID:42308315. The honest reading: both paralogues are insertases, they are
partially redundant, and they are not interchangeable in every host — which is exactly how
the MTCH1 review reads it too.

6. Annotation actions changed in this update

Term Before After Why
GO:0005739 mitochondrion (IBA, HTP) MARK_AS_OVER_ANNOTATED ACCEPT A correct parent of a correct location; node placement not in dispute. Matches MTCH1. Also clears the missing-propagation_review warning.
GO:0016020 membrane (IBA) MARK_AS_OVER_ANNOTATED, no metadata same action + propagation_review WITH/FROM inspected: MGI:1929260, PANTHER:PTN001324730, UniProtKB:Q9Y6C9. Granularity objection only. MTCH2's own appearance in its own WITH/FROM is expected, not circular.
GO:0005515 protein binding (IPI x3) REMOVE MARK_AS_OVER_ANNOTATED Project guidance; these are experimental interaction records. Matches MTCH1.
GO:0043065 positive regulation of apoptotic process KEEP_AS_NON_CORE ("downstream of insertase") KEEP_AS_NON_CORE (reason rewritten) Insertase-independence established; kept non-core only because the MF is undefined.
GO:0055088 lipid homeostasis KEEP_AS_NON_CORE ("pleiotropic") KEEP_AS_NON_CORE (reason rewritten) Direct CPT1 interaction supplies a mechanism.
GO:0010635 regulation of mitochondrial fusion KEEP_AS_NON_CORE ("MFN insertion") KEEP_AS_NON_CORE (reason rewritten) Published mechanism runs through MFN2 and LPA, not through inserting the fusion machinery.
GO:7770059 alpha helical protein insertion into OM absent NEW MTCH1 already carries it; the substrate restriction is exactly what was shown for MTCH2.
GO:0005515 (PMID:32296183, P56378-2) missing from review added Second GOA row that had been omitted.
GO:0005634 nucleus (HDA) REMOVE REMOVE (unchanged) Sperm-nucleus fraction; mitochondrial carry-over. Left alone — the new literature does not bear on it.

Also fixed: five supporting_text entries attributed to the Falcon deep-research file were
paraphrases beginning "Falcon synthesis supports…" and appear nowhere in that file. They
have been replaced with verbatim quotations.

7. Open questions

Carried into suggested_questions / suggested_experiments / knowledge_gaps:

  1. Should the "Transport" keyword and the "carrier homolog" name be retired, or is transport
    merely unassayed?
  2. What is the molecular function behind the BAX/BAK effect — lipid scrambling, lipid
    transfer, something else?
  3. Why does MTCH1 but not MTCH2 complement the yeast MIM complex; do the paralogues have
    non-identical substrate ranges in human cells?
  4. Is the CPT1 effect separable from insertase activity?

Earlier entry, carried over from the PAINT no-IBA project

Recorded on main before this update; preserved verbatim below. Where the two
passes disagreed, the merge notes in the review YAML say which was taken and why.

2026-09-04 — Finishing pass on ai-review.yaml (PAINT no-IBA project)

Completed the review pass over all 22 existing_annotations entries; status
moved INITIALIZED → COMPLETE (validation clean, no warnings).

Key points and changes: