TMEM65 (Q6PI78) — review notes

1. What the protein is

TMEM65 is a small (240 aa) polytopic protein of the mitochondrial inner membrane (IMM). It carries an
N-terminal mitochondrial targeting sequence that is cleaved on import, followed by a matrix-exposed
soluble segment and three transmembrane helices.

• Mitochondrial import and IMM integration:
  PMID:24765583
• Independent confirmation in human fibroblasts, plus a pathogenic splice variant:
  PMID:28295037
• Both 2025 mitochondrial-calcium papers also place it in the IMM (GOA carries is_active_in GO:0005743 IDA from each).

A separate, earlier line of work reported Tmem65 as a cardiac intercalated-disc / plasma-membrane
protein that stabilises connexin 43:
PMID:26403541
This is the origin of the plasma membrane, intercalated disc, cardiac conduction,
regulation of cardiac conduction and cardiac ventricle development annotations (plus their
ISS/Ensembl-orthology projections). It sits uneasily with the IMM consensus — a protein with a
cleaved MTS that fractionates with the IMM is not obviously also a sarcolemmal protein — and none of
the mitochondrial studies reproduce it. Treated here as real but non-core and unreconciled, not
removed: the underlying data are experimental (IDA), and the two localisations are not formally
mutually exclusive.

2. The contested molecular function

Everything below concerns mitochondrial Na+/Ca2+ exchange (mito-NCX), the Na+-driven route by which
matrix Ca2+ is exported. Three groups independently found that TMEM65 is required for it. They
disagree about what TMEM65 does.

Position A — TMEM65 is the exchanger (Zhang et al., Nat Cell Biol 2025)

PMID:40691517

The strongest arguments are the heterologous-reconstitution ones:
- PMID:40691517 — in Sf9 mitochondria, which have a uniporter but no native
mito-NCX.
- PMID:40691517
- Tissue distribution matches mito-NCX activity where NCLX's does not:
PMID:40691517
- Conclusion asserted flatly: PMID:40691517

Position B — TMEM65 is a required regulator of NCLX (Garbincius et al., Nat Metab 2025)

PMID:40200126

The decisive experiment is epistatic: TMEM65 has no effect in cells lacking NCLX.
PMID:40200126
PMID:40200126

The field has not resolved it

A Cell Metab commentary (Garbincius & Elrod — note: same lab as Position B, so not a neutral referee,
but the methodological point is concrete) frames the dispute and identifies the specific gap:

PMID:41061666

PMID:41061666

and, even-handedly:
PMID:41061666

So the 45Ca2+ proteoliposome accumulation assay — the one experiment that would make TMEM65 a
transporter rather than an accessory factor — is contested on the grounds that accumulation could
reflect binding rather than translocation. No electrophysiology on purified TMEM65 exists.

3. GOA carries both sides

Confirmed directly in TMEM65-goa.tsv (QUALIFIER column):

NOT|enables  GO:0005432 calcium:sodium antiporter activity  IDA  PMID:40200126  UniProt
enables      GO:0005432 calcium:sodium antiporter activity  IDA  PMID:40691517  UniProt

Two IDA annotations of the same term with opposite polarity, both from UniProt, both dated
2025-11-06. GO has deliberately annotated the disagreement rather than picking a winner. The stub
already carries negated: true on the PMID:40200126 row — verified, no fix needed.

Decision: UNDECIDED for both GO:0005432 rows. Accepting the positive IDA while a NOT IDA of
the same term co-exists would assert a resolution that does not exist; removing either would overrule
a curator who read a full text I have not. UNDECIDED is what the enum is for. The same reasoning
propagates to GO:0035725 sodium ion transmembrane transport, which GOA derives automatically from
GO:0005432 by inter-ontology inference (GO_REF:0000108) and which is therefore only as sound as the
contested MF.

What is not contested: that TMEM65 is required for mitochondrial Ca2+ export, and that it acts
at the IMM. GO:0099093 calcium export from the mitochondrion (IDA from both papers) and
GO:0005743 mitochondrial inner membrane are accepted as core.

4. Cross-reference: the counterion is contested for NCLX too

Both 2025/2026 cryo-EM structures of NCLX note it lacks the canonical Na+ sites — which is also
Position A's structural argument against NCLX:
PMID:40691517
PMID:40931067

Note on provenance: PMID:40691517 (TMEM65 = mito-NCX) and PMID:40931067 (NCLX = H+/Ca2+ exchanger)
share senior authors (Feng L., Tsai M.-F.). They are one internally consistent model — TMEM65 does
mito-NCX, NCLX does something else — rather than two independent confirmations. That matters when
weighing how much of the field has actually moved. See SLC8B1-notes.md.

5. Other annotations

• GO:0005515 protein binding IPI (PMID:32296183, HuRI binary interactome, 37 partners collapsed
  into one stub row) — bare protein binding, uninformative as a molecular function.
• GO:0005739 mitochondrion (IBA, HTP PMID:34800366) — correct, less specific than the IMM
  annotations the gene already carries.
• Cardiac terms (GO:0003231, GO:0061337, GO:1903779) — organismal phenotypes arising from the
  Cx43/intercalated-disc work and its orthology projections; real but distal to the molecular
  function, and in any case downstream of whichever mitochondrial role turns out to be correct.

6. Open questions recorded in the review

1. Does purified TMEM65 translocate charge? (Electrophysiology on reconstituted protein, or
   simultaneous bilayer Ca2+/Na+ reporters, would settle Position A vs B.)
2. Why does TMEM65 require NCLX in fibroblasts (PMID:40200126) but suffice in Sf9 mitochondria and
   liposomes (PMID:40691517)?
3. Does CGP-37157 bind TMEM65, NCLX, or both?
4. Is the intercalated-disc/Cx43 localisation a genuine second pool, or an artefact?