TMEM63A review notes

Why this gene was selected

Reviewed as the paralog partner of TMEM63B in a contested-function pair: mechanically
activated cation channel vs mechanically activated lipid scramblase. The two paralogs were
reviewed together so the reasoning is consistent, but they were not assumed to behave
identically — and they do not.

Position taken

For TMEM63A, only the mechanosensitive cation channel activity is core. The scramblase
claim, which is core for TMEM63B, is not established for TMEM63A in cells; GOA carries the
opposite assertion (NOT|enables GO:0017128, IDA, PMID:39716028) and the 2026 reconstitution
work that does report TMEM63A lipid translocation explicitly agrees that it does not happen
in the cellular context at rest.

Channel function (core, uncontested)

Family-level founding demonstration: PMID:30382938.

Human genetics: PMID:31587869.

Structure and biophysics, monomeric with a single restricted pore:
PMID:37543036 and
PMID:37543036.

In vivo lung physiology, alongside TMEM63B:
PMID:38127458 with the localisation
PMID:38127458.

The 2026 Neuron paper is a purely channel-framed account in the cell type where TMEM63A matters
most: PMID:41483808 with the loss-of-function phenotype
PMID:41483808. (Abstract-only in the cache;
full_text_available: false.)

The scramblase question for TMEM63A — the NOT annotation is sound

GOA carries NOT|enables GO:0017128 phospholipid scramblase activity, IDA, PMID:39716028.
That paper deliberately compared paralogs in a scramblase-null background:
PMID:39716028 — i.e. the activity tracked with TMEM63B orthology, not with
family membership.

The 2026 Nat Commun paper appears at first to contradict this, since it reports lipid
translocation by purified hTMEM63A: PMID:41617699. But the same paper states the cellular result agrees with the NOT annotation:
PMID:41617699. Only when
force is actively applied does cellular PS exposure appear:
PMID:41617699.

So the two results are reconcilable, and the reconciliation is about gating state, not
identity: TMEM63A retains the structural capacity to scramble, but under resting cellular
conditions it does not. The NOT annotation is therefore accepted as written, with the
condition-dependence recorded in the review reason and the open question moved to
suggested_questions. This is not treated as a straight contradiction of the IDA.

Mechanistically, the V→M latch variants make the same point from the other direction: the
TMEM63A V53M disease substitution creates constitutive scrambling
PMID:40480214 — which only makes sense if WT TMEM63A does not normally scramble.

The authors of that structural work leave the WT question open:
PMID:40480214.

Term-choice problems found in the existing annotation set

GO:0003676 nucleic acid binding (IEA, InterPro IPR035979) — remove

This is a fold-only inference. The InterPro matches for O94886 include
SSF54928 RNA-binding domain, RBD → IPR035979 RNA-binding domain superfamily and
IPR012677 Nucleotide-binding alpha-beta plait domain superfamily, both of which are
structural-superfamily hits on the CSC1/OSCA1-like cytosolic domain (IPR027815), which
adopts an RRM-like alpha-beta plait fold. There is no report of nucleic acid binding by any
OSCA/TMEM63 protein, and TMEM63A is an 11-TM integral membrane channel of the lysosomal and
plasma membranes. Notably the paralog TMEM63B does not carry this annotation, so the
pipeline is not even internally consistent. This is precisely the class of demonstrably wrong
electronic inference that REMOVE exists for.

GO:0005227 calcium-activated cation channel activity — wrong gating stimulus

Identical issue to TMEM63B; see TMEM63B-notes.md for the full argument. In brief, the GO
definition requires the channel to open when a calcium cation has been bound, whereas
TMEM63A opens in response to membrane stretch and is Ca2+-permeable. The InterPro family
behind the IEA is literally named "Calcium permeable stress-gated cation channel 1-like"
(IPR045122) and is mapped to the calcium-activated term. MODIFY → GO:0140135 +
GO:0005262 on both the IBA and the IEA.

GO:0008381 mechanosensitive monoatomic ion channel activity — correct but under-specific

TMEM63A conducts cations (Ca2+, and the family conducts Na+/K+/Cs+); GO:0140135
mechanosensitive monoatomic cation channel activity is available and is what TMEM63B
already carries from an IDA. MODIFY all four rows (IEA, IDA x2, IMP) to GO:0140135.

Localisation

Native TMEM63A is principally lysosomal/endolysosomal, with plasma-membrane presence in some
cell types (and in heterologous over-expression, which is how the currents are recorded).
This is stated in the 2026 reconstitution paper's own framing and matches the GOA set
(lysosomal membrane: IDA PMID:39716028, IDA PMID:38127458, IDA PMID:20957757, HDA
PMID:17897319). The lysosome-organization annotation transfers by ISS from Drosophila
Tmem63 (UniProtKB:Q6NP91), the single fly ortholog — kept as non-core.

HPA GO:0034451 centriolar satellite (IDA, GO_REF:0000052) is a single-antibody
immunofluorescence call that is incompatible with the established multi-pass
endolysosomal/plasma-membrane topology and is not corroborated by any focused study; marked as
over-annotated rather than removed, since it is an experimental-code annotation.

GO:0070062 extracellular exosome comes from urinary-exosome shotgun proteomics
(PMID:19056867) — a high-throughput co-purification term with no functional content here.