MTCH1 (Q9NZJ7) — review notes

Journal for the AI review of human MTCH1 (mitochondrial carrier homolog 1; historical name PSAP,
"presenilin-associated protein"). All assertions carry inline provenance.

1. What the protein is

• SLC25 (mitochondrial carrier) fold, six predicted TM helices, two SOLCAR repeats
  (UniProt Q9NZJ7: DR PROSITE; PS50920; SOLCAR; 2., FT TRANSMEM 94..104 / 156..176 / 210..229 / 255..279 / 323..342 / 372..389), but resident in the outer mitochondrial membrane, not the
  inner membrane where canonical SLC25 carriers sit
  PMID:12377771.
• PANTHER family PTHR10780 (MITOCHONDRIAL CARRIER HOMOLOG), subfamily PTHR10780:SF3
  (MITOCHONDRIAL CARRIER HOMOLOG 1). MTCH2 is SF20 of the same family.
• Long and short isoforms exist and were both used in the 2026 structural study
  (Q9NZJ7-1 = MTCH1-L, Q9NZJ7-2 = MTCH1-S)
  PMID:42308315.

2. The insertase question — three independent lines, all positive for MTCH1

(a) Guna et al. 2022 (the MTCH2 paper), PMID:36264797. The rigorous reconstitution
(purified protein in liposomes) was done for MTCH2, not MTCH1. For MTCH1 the paper shows a
knockdown genetic interaction only:
PMID:36264797
and then generalises:
PMID:36264797.
So the GOA IDA on MTCH1 for GO:0032977 sourced to this paper is, strictly, a proposal plus a
depletion phenotype rather than a direct demonstration on MTCH1. The conclusion turned out to be
right, but the evidence code overstates what this particular paper did for MTCH1. Recorded, not
acted on — the curator had the full text and figures S17, and the claim is now independently
confirmed (below).

(b) Dimogkioka, Elias & Rapaport 2025, J Cell Sci, PMID:40704594 — heterologous complementation
in yeast. This is the paper that makes MTCH1 insertase activity a direct claim about MTCH1 itself:
PMID:40704594
PMID:40704594
PMID:40704594

(c) Stevens et al. 2026, Sci Adv, PMID:42308315 — rescue in human cells, the cleanest
homologous-system test:
PMID:42308315.

Is the "MTCH1 but not MTCH2 rescues yeast" result actually awkward?

Less than it looks. The authors themselves attribute the MTCH2 failure to behaviour of MTCH2 in
yeast, not to absence of insertase activity: MTCH2 was toxic and mislocalised to the ER —
PMID:40704594
— and even when a chimeric N-terminal MTCH1 segment redirected it to mitochondria it still did not
complement
PMID:40704594.
They offer conformational/substrate-recognition differences as the explanation
PMID:40704594.

So the correct reading is not "MTCH2 is not an insertase". MTCH2 insertase activity is
established by reconstitution with purified protein in liposomes (PMID:36264797) and by the 2026
cryo-EM structure-function work (PMID:42308315), neither of which this yeast assay touches. What
the yeast result adds is a positive claim about MTCH1 and a caution that heterologous
complementation reports on compatibility with the host membrane/substrate set as much as on
catalytic capability. Net effect on curation: it strengthens GO:0032977 on MTCH1; it does not
weaken anything on MTCH2.

3. The carrier/transporter question

GOA carries no transporter or carrier GO term on MTCH1 (checked every row of
MTCH1-goa.tsv: the MF terms present are GO:0032977 membrane insertase activity ×2 and
GO:0005515 protein binding ×2 — nothing in the transmembrane-transporter branch). That is the
right call and there is nothing to remove.

The carrier claim survives only in (i) the gene/protein name ("Mitochondrial carrier homolog 1")
and (ii) the UniProt keyword block, which still lists Transport (line 316 of
MTCH1-uniprot.txt) alongside the family statement
CC -!- SIMILARITY: Belongs to the mitochondrial carrier (TC 2.A.29) family. Both are
family-signature inferences from the SLC25 fold, not observations. No transported substrate has
ever been reported for MTCH1 or MTCH2.

The 2026 structural work argues positively against the transporter reading: the MTCH clade has
lost the machinery that makes an SLC25 protein a carrier —
PMID:42308315
— and insertase activity arose from loss of a transmembrane helix that opened a lipid-facing groove
PMID:42308315.
Consistent with this, MTCH1 has only two SOLCAR repeats rather than the canonical three.
The UniProt Transport keyword is the one thing worth flagging to UniProt; the GO side is clean.

4. The old PSAP literature (1999–2002)

MTCH1 was discovered as a presenilin-1 interactor in a yeast two-hybrid screen and mis-assigned a
PDZ-like domain
PMID:10551805.
The paper's closing speculation is the sole basis for two NAS annotations still in GOA
(GO:0009966 regulation of signal transduction; GO:0045161 neuronal ion channel clustering)
PMID:10551805.
No PDZ domain is annotated in the current UniProt entry (the only domain-level features are the two
SOLCAR repeats and six TM helices), and the protein is a multi-pass mitochondrial outer-membrane
protein, not a cytosolic scaffold. Both NAS terms are marked REMOVE.

The proapoptotic claim is better grounded but is an overexpression phenotype
PMID:12377771,
PMID:12377771.
Independent corroboration that expressing MTCH1 kills cells comes, incidentally, from the 2026
methods section
PMID:42308315.
So the phenotype is real and reproducible, but no mechanism links it to a defined MTCH1 molecular
activity, and an alternative reading — that perturbing OMM protein biogenesis secondarily triggers
apoptosis — is untested. Kept as non-core rather than removed.

5. Paralog discipline (explicitly not imported onto MTCH1)

MTCH2 has recently acquired two further asserted functions. Neither has been tested on MTCH1 and
neither is transferred here:
- MTCH2 promotes BAX/BAK self-assembly and apoptotic pore growth [PMID:42056306, Nat Struct Mol Biol 2026].
- MTCH2 modulates CPT1 activity in adipocytes by direct interaction [PMID:41044057, Nat Commun 2025].
Given that MTCH1 has its own (older, weaker) proapoptotic literature, the temptation to merge the
two apoptosis stories is exactly the failure mode to avoid: the MTCH2/BAX work is a distinct
mechanism at a distinct step and says nothing about MTCH1.

6. Decisions