Scn9a (NaV1.7) — Heterocephalus glaber (naked mole-rat), UniProt G9DCX3

Research journal for the GO annotation review. Every assertion carries provenance.

0. What this accession is

G9DCX3 is the TrEMBL entry for the SCN9A mRNA (EMBL JF912494, protein AEV53348.1)
submitted by Smith, Omerbasic, Anirudhan, Lechner, Lapatsina and Lewin — the authors of
the Science paper that established naked mole-rat acid insensitivity. The UniProt record
cites that paper as its only literature reference, so this is the sequence physically tied
to the acid-block literature rather than a genome-pipeline model.

The record is not a complete protein. It carries
[file:HETGA/Scn9a/Scn9a-uniprot.txt "Flags: Fragment;"] with NON_TER at both residue 1
and residue 1884.

My own alignment (see Scn9a-bioinformatics/RESULTS.md) places the fragment at
human Q15858 residues ~23-1901, i.e. missing 22 residues of the N-terminal cytoplasmic
head and 87 residues of the distal cytoplasmic C-terminal tail, at 92.1% identity within
matched blocks. Everything structurally required for channel function is present: the
UniProt feature table lists four complete Pfam PF00520 ion-transport (S1-S6) domains
(105-390, 724-955, 1171-1446, 1495-1750), the PF11933 cytoplasmic domain (515-674), the
PF06512 sodium-ion-transport-associated domain (962-1167), 20 transmembrane helices, and
the PF24609 SCN5A-like C-terminal IQ motif (1862-1883).

Curation consequence. The fragment status does not undermine any molecular-function or
cellular-component claim about the channel core — the pore, all four voltage sensors and the
inactivation-gate region are in the entry. It would undermine a claim about the distal
C-terminal tail (e.g. specific C-terminal protein-interaction or trafficking motifs), and
none is annotated. The UniProt CAUTION line ("Lacks conserved residue(s) required for the
propagation of") is a UniRule feature-propagation notice and should be read against this
truncation, not as a statement that the channel core is defective.

Family assignment is unambiguous:
[file:HETGA/Scn9a/Scn9a-uniprot.txt "PANTHER; PTHR10037:SF221; SODIUM CHANNEL PROTEIN TYPE 9 SUBUNIT ALPHA; 1."]
and [file:HETGA/Scn9a/Scn9a-uniprot.txt "Belongs to the sodium channel (TC 1.A.1.10) family."]

1. The conserved function: this is a real voltage-gated sodium channel

UniProt's rule-based function statement is the standard NaV alpha-subunit description:
[file:HETGA/Scn9a/Scn9a-uniprot.txt "Mediates the voltage-dependent sodium ion permeability of"]
excitable membranes, with catalytic activity
[file:HETGA/Scn9a/Scn9a-uniprot.txt "Reaction=Na(+)(in) = Na(+)(out)"] and
[file:HETGA/Scn9a/Scn9a-uniprot.txt "SUBCELLULAR LOCATION: Cell membrane"]
[file:HETGA/Scn9a/Scn9a-uniprot.txt "Multi-pass membrane protein"].

Crucially, the naked-mole-rat literature treats this protein as a functional channel
whose gating is altered, not as a dead or pseudo-channel. Smith et al. cloned it and
measured proton block of the expressed channel
PMID:22174253.
A channel that can be blocked is a channel that conducts. The naked mole-rat also shows
PMID:34476892,
which requires normal nociceptor action-potential initiation and hence a working NaV1.7.

The general statement that this class of channel is what fires action potentials is stated
plainly in the full-text review:
PMID:32206859
and, for this subunit specifically,
PMID:32206859.

So the projected core annotations (voltage-gated sodium channel activity; sodium ion
transmembrane transport; membrane depolarization during action potential; neuronal action
potential; plasma membrane; voltage-gated sodium channel complex) are not merely transferred
from mouse — they are positively confirmed by naked-mole-rat work on this very protein.

2. The species-specific adaptation: enhanced proton block

The organismal phenotype and the mechanism:

• PMID:22174253
• PMID:22174253
• PMID:34476892
• PMID:31992138

Note the inversion of the usual logic. In every other vertebrate examined, acid excites
nociceptors through proton-gated channels
PMID:22174253;
naked-mole-rat ASICs and TRPV1 behave normally, and the change is downstream, in the
channel that would have to fire the action potential.

The motif — verified, not recalled

The abstract-only primary papers (PMID:22174253, PMID:31147513) do not give residue
numbers. The full-text review does describe the motif, in words rather than coordinates:
PMID:32206859
and
PMID:32206859
The review states that the ancestral trio is KKV (positively charged) in mouse and human and
becomes EKD or EKE in subterranean African mole-rats.

Harms et al., who tested the motif experimentally, locate it precisely:
PMID:28939386.

Rather than assert residue numbers from memory, I checked the sequences directly
(Scn9a-bioinformatics/div_ploop_motif.py, results in
Scn9a-bioinformatics/RESULTS.md). Anchoring on the invariant flanking residues of the
domain IV extracellular P-loop (D[CS][DN]P...HPG):

[file:HETGA/Scn9a/Scn9a-bioinformatics/RESULTS.md "The naked mole-rat NaV1.7 sequence G9DCX3 carries EKE at fragment positions 1698-1700, where human Q15858 carries KKV at 1718-1720, in the domain IV extracellular P-loop."]

This reproduces the published motif exactly, in the sequence this review is about, and
locates it in the topology: taking the UniProt feature table at face value, domain IV S5
ends at 1641 and S6 begins at 1717, so 1698-1700 sits in the distal half of the
extracellular P-loop between them. Guinea pig — the closest sampled hystricomorph
relative — retains KV with only a conservative K→H at the first position, so EKE is not a
hystricomorph-wide feature. The one hibernator I sampled retains KKV; I did not attempt to
reproduce the broad hibernator survey of
PMID:24352952,
which reports the convergence, and which independently confirms that
PMID:24352952

The motif is a sequence correlate; the physical mechanism is still open

I initially wrote that a +2 to -2 swap next to the pore is "exactly the kind of change
expected to raise local proton affinity". That is a just-so story, and the one paper that
actually tested it says otherwise. Harms et al. put the naked-mole-rat motif into the human
channel and confirmed the effect —
PMID:28939386
and
PMID:28939386 —
but they also found that the simple electrostatic account does not generalise:
PMID:28939386,
PMID:28939386,
and
PMID:28939386

Two consequences for this review. First, the wording throughout describes EKE as the
sequence correlate of enhanced proton block, not as a pore-occluding site. Second, the
Harms observation that the EKE substitution also reduces channel function in the human
chimera is a caution against reading "the naked mole-rat channel behaves normally except
for pH" too literally; the in-vivo argument for retained function rests on the animal's
normal heat and mechanical nociception, not on the chimera's biophysics.

The convergent-evolution framing is repeated in the multi-species rodent screen:
PMID:31147513

3. Does the acid-block property justify an extra GO annotation?

This was the substantive question for this review. I searched properly rather than assuming.

Terms checked (QuickGO /ontology/go/terms/<id>/complete, plus keyword search):

• GO:0160128 pH-gated monoatomic ion channel activity and its sodium child
  GO:0160125 pH-gated sodium channel activity. Definition of GO:0160125: "A gated
  channel activity that enables the transmembrane transfer of a sodium ion by a channel that
  opens in response to a change in pH." These describe a channel that opens on
  acidification — that is the ASIC/TRPV1 behaviour naked-mole-rat NaV1.7 does not have.
  Using them here would assert the opposite of the published mechanism. GO:0160128 replaces
  the older GO:0044736; no relevant secondary id.
• GO:1905150 regulation of voltage-gated sodium channel activity (BP) — its only child is
  GO:1905152 (positive regulation). GO:1905151 "negative regulation of voltage-gated sodium
  channel activity" is obsolete (obsoleted 2025-04-08, comment: "This term was obsoleted
  because it represents a molecular function"), with no replaced_by and no consider
  term. In any case a regulation term describes the regulator; the regulator here is a
  proton, not a gene product.
• GO:0016248 channel inhibitor activity / GO:0019871 sodium channel inhibitor
  activity — defined as "Binds to and stops, prevents, or reduces the activity of a
  sodium channel", i.e. the property of the inhibitor, not of the inhibited channel.
• GO:1901691 proton binding — GO explicitly comments on GO:0160128 that pH effects on
  transporters work "by protonation of specific residues in the protein, and not by H+
  binding", so proton binding is the wrong model and is discouraged for exactly this case.
• BP alternatives (GO:0010447 response to acidic pH, GO:0071468) describe a cell's or
  organism's response process; they do not capture a property of the channel, and asserting
  involved_in response to acidic pH for a channel whose contribution is to fail to fire
  would be misleading without a NOT/negative-regulation framing GO cannot currently express.

Conclusion: there is a genuine ontology gap, and the obsoletion note on GO:1905151
("represents a molecular function") shows GO itself moved the concept to the MF branch
without creating the MF term. I have therefore filed it under proposed_new_terms rather
than forcing an existing term. I did not add a speculative annotation.

4. The falcon deep-research report

A falcon report (Scn9a-deep-research-falcon.md, Edison Scientific Literature, 15
citations) was generated while this review was in progress. It is naked-mole-rat-specific,
unlike the affinage record, and it agrees with the conclusions I had already reached
independently, which is a useful check rather than a source of them:

• It reaches the same identity and fragment caveat:
  [file:HETGA/Scn9a/Scn9a-deep-research-falcon.md "However, G9DCX3 is explicitly a sequence fragment, and no retrieved study directly characterized that exact deposited fragment as a complete channel."]
• It reaches the same molecular function and location:
  [file:HETGA/Scn9a/Scn9a-deep-research-falcon.md "NaV1.7 is not an enzyme. It is an electrogenic ion channel whose transported substrate is principally Na+."]
  and
  [file:HETGA/Scn9a/Scn9a-deep-research-falcon.md "NaV1.7 performs its transport function in the plasma membrane."]
• It independently confirms the localisation gap I had flagged:
  [file:HETGA/Scn9a/Scn9a-deep-research-falcon.md "the retrieved literature did not provide a dedicated naked-mole-rat immunolocalization or spatial-transcriptomic study uniquely mapping G9DCX3"]
• It rules out the alternatives I did not annotate:
  [file:HETGA/Scn9a/Scn9a-deep-research-falcon.md "The retrieved evidence does not establish that G9DCX3 has unique catalytic activity, unusual ion substrate specificity, or a structural scaffolding role."]

Its one genuinely new contribution was Harms et al. 2017 (PMID:28939386), which neither
the affinage record nor my own reading of the cached naked-mole-rat corpus had surfaced.
That paper was not in publications/ and I fetched it (abstract only; no PMC record). It is
the study that tested the EKE motif experimentally, and it is also the study that qualifies
the mechanism — see section 3. Finding it changed the wording of this review, so this is a
real recall win for falcon over affinage on this gene.

Falcon's own recommended annotation set matches what this review concluded, with one
difference: it suggests "regulation of membrane potential" as a process term, which I did
not add, because for a channel that carries the depolarizing current itself GO:0086010
(membrane depolarization during action potential) is already annotated and is more precise.

5. What I could not resolve

• Where the protein is expressed in the naked mole-rat, at subcellular resolution. The
  literature speaks of "naked mole-rat nociceptors" and cutaneous C-fibres, but I found no
  naked-mole-rat immunolocalisation of NaV1.7 to axons, nerve terminals or nodes of Ranvier.
  Human data exist PMID:30795902,
  but I have deliberately not projected those sub-plasma-membrane CC terms onto this species;
  plasma membrane is as specific as the naked-mole-rat evidence supports.
• The domain III variants. PMID:32206859 reports that a functionally equivalent part of
  domain III also carries variants unique to acid-insensitive African mole-rats and states
  their functional importance remains to be tested. I did not attempt to localise or
  interpret them; nothing in this review depends on them.
• Visceral vs cutaneous. PMID:31992138 is about colonic afferents and must not be read as
  a claim about cutaneous acid insensitivity. Its own authors point out that
  PMID:31992138,
  consistent with PMID:32206859.
  I cite it only as independent corroboration of the mechanism, not as evidence about gut
  physiology.

6. What the affinage human-ortholog record missed

The Scn9a-deep-research-affinage-human-ortholog.md file is an affinage record for human
SCN9A (Q15858), retrieved as a conserved-mechanism baseline. It is a rich account of human
NaV1.7: channelopathies, CRMP2 SUMOylation and Nedd4-2 trafficking, endogenous opioid tone in
NaV1.7-null animals, ProTx2/acylsulfonamide structural pharmacology, olfaction.

What it does not contain, and what this review turns on:

1. Nothing about the naked mole-rat, or about proton block of NaV1.7 at all. Its 42
   citation-anchored findings include no naked-mole-rat paper and no pH/proton-block finding.
   PMID:22174253, PMID:24352952, PMID:31147513 and PMID:32206859 — the entire evidence base
   for the species-specific function — are absent. This is expected (affinage is human-only)
   but it means the record could not have driven this review's central call.
2. Its GO grounding collapses to uninformative parents. Its mechanism_profile gives
   molecular_activity as GO:0005215 transporter activity — three levels above
   GO:0005248 and not even channel-specific — for a protein whose defining activity is
   voltage-gated sodium channel activity. Following the brief, none of its GO ids were
   imported; every term in this review was re-grounded from GOA, QuickGO and the literature.
3. Its own self-evaluation is a tie, not a win. The file's frontmatter records
   self_evaluation_pairwise: tie against the curated UniProt reference. That is recorded in
   references[].reference_review.correctness: LOW_QUALITY with an explanatory note, rather
   than VERIFIED.

It was nonetheless useful for one thing: confirming that the conserved (human) NaV1.7 is a
threshold-setting nociceptor channel, which is the background against which the naked-mole-rat
change is a modulation of an intact function rather than a loss of function. No sentence
from it is quoted as evidence anywhere in the review.

7. Review decisions, in one place

All 12 GOA rows are electronic (GO_REF:0000002 InterPro2GO, GO_REF:0000118
TreeGrafter/PANTHER PTN000004433, GO_REF:0000120 combined InterPro + PANTHER + UniRule).
There is no experimental GO annotation for this protein and none for any naked-mole-rat gene
relevant here.

No REMOVE: every projected term describes a function the naked-mole-rat evidence positively
supports. The species-specific difference here is a quantitative modulation of gating
(greatly enhanced proton block), which refines the channel's behaviour rather than abolishing
any annotated function — exactly the case the brief warns against over-reading.

No UNDECIDED: unusually for a naked-mole-rat gene, this protein has a dedicated primary
literature, so nothing was left unresolved among the existing rows.