P2RX7 review notes

Why this gene was selected, and where the exposure actually is

P2X7 is usually described as a "contested" protein. It is important to be precise about what is
contested, because the GOA exposure follows from it.

Not contested: the molecular function. P2X7 is a homotrimeric, ATP-gated non-selective cation
channel. This was established at the cloning of the human receptor
PMID:9038151, confirmed by single-channel analysis, and shown structurally
PMID:31587896. Even the 2026 review that catalogues the controversy opens by taking the receptor's identity
for granted and disputes only where it is found.

Contested: which cells express it.
PMID:41672132 but
PMID:41672132 The
review frames the problem as methodological:
PMID:41672132

So the curation exposure is in CC and BP, not MF. No MF controversy was manufactured; every
molecular-function row was reviewed on its own merits and the ATP-gated cation channel term is core.

The other side of the neuronal argument

The 2026 literature is not uniform. Two 2026 papers proceed as if neuronal P2rx7 were settled:
PMID:41935187 and PMID:41935187 - with the conditional Setdb1 deletion restricted to excitatory
neurons; and PMID:41881298

The strongest positive evidence is the cell-type-specific seizure study:
PMID:38777288 versus
PMID:38777288 with human data
PMID:38777288

Opposite-direction phenotypes from two cell-type-restricted deletions are hard to explain by
antibody artefact, which is why the neuronal annotations are kept rather than removed.

The corrigendum, PMID:40713463 - what could not be retrieved

That seizure study has a 2025 corrigendum. I could not retrieve what was corrected. Specifically:

• The PubMed record (PMID:40713463) is an Published Erratum type with author list and affiliations but
  no abstract and no correction text. It states only Erratum for Brain Behav Immun. 2024 Aug;120:121-140. doi: 10.1016/j.bbi.2024.05.023.
• Europe PMC returns the record with an empty abstractText and isOpenAccess: N, listing only
  "Subscription required" via DOI.
• https://doi.org/10.1016/j.bbi.2025.07.009 redirects to linkinghub.elsevier.com, which served only a
  "Redirecting" stub; the ScienceDirect article page returned HTTP 403.

So: the corrigendum exists (Brain Behav Immun 2025;129:1040, doi 10.1016/j.bbi.2025.07.009), it applies
to the paper that is the single strongest support for functional neuronal P2X7, and its content is
unknown to this review. That is recorded verbatim in reference_review for PMID:40713463 and raised
in suggested_questions, because a curator leaning on PMID:38777288 needs to read it first. Note also
that the corrigendum author list differs slightly from the original in the rendering of several author
names (e.g. "Arribas Blázquez M" becomes "Blázquez MA", "Menéndez Méndez A" becomes "Méndez AM"),
which is consistent with - but not proof of - an authorship/name correction rather than a data
correction. I am not asserting that; it is an observation about the two PubMed records.

Curation positions taken on the neuronal annotations

The neuronal terms split into two groups that deserve different treatment, and this is the main
finding of the review:

Group 1 - derived mechanically from the molecular function: MARK_AS_OVER_ANNOTATED

• GO:0060079 excitatory postsynaptic potential is IEA GO_REF:0000108 (inter-ontology logical
  inference) with GO:0004931 in the WITH/FROM field.
• GO:0098794 postsynapse is IEA GO_REF:0000108 with GO:0060079 in the WITH/FROM field.

The chain is therefore: P2X7 is an ATP-gated cation channel → it could generate an EPSP → it is
postsynaptic. No neuron was involved at any step, and every P2X receptor inherits the same chain. For
a protein whose neuronal expression is the contested question, deriving a synaptic compartment from its
molecular function is the one inference that must not be allowed to stand in as evidence. Both marked
over-annotated, with the derivation spelled out in reason.

This was checked directly against the WITH/FROM column of P2RX7-goa.tsv, not assumed.

Group 2 - orthology transfers of experimental mouse localisations: KEEP_AS_NON_CORE

GO:0043025 neuronal cell body, GO:0045202 synapse, GO:0031594 neuromuscular junction (and
GO:0009897, GO:0005911) are IEA GO_REF:0000107 from mouse P2rx7, and the mouse records carry
experimental evidence behind each: GO:0043025 from PMID:15964665 and PMID:15978588, GO:0045202 from
PMID:18082965, GO:0031594 from PMID:15713258. These are precisely the antibody-based studies the 2026
review questions, but per project rules an experimental curator call is not overruled from an abstract.
Kept, demoted to non-core, with the full controversy recorded in reason on each.

GO:0019233 sensory perception of pain got the same treatment - well supported as an organismal
phenotype, but mechanistically attributed largely to microglial P2X7, so it sits inside the cell-type
question rather than outside it.

A second, independent dispute found inside the GOA itself

Does the P2X7 pore dilate? GOA carries GO:0046931 pore complex assembly three times (ARBA IEA,
IMP PMID:25651887, IDA PMID:9038151). The permeabilisation is real:
PMID:9038151

But single-channel analysis of the human receptor argues against dilation of the P2X7 pore itself:
PMID:17483156

and the dye-uptake pathway was assigned to a different protein:
PMID:17036048

So the phenotype is P2X7-dependent while the pore may not be P2X7. KEEP_AS_NON_CORE, with both sides
cited in supported_by on the annotation itself - the same treatment TMEM175's proton-channel terms got
in this project. This is also why GO:0015748 organophosphate ester transport (ARBA) is marked
over-annotated: it reads a transport activity off the ligand.

Scrambling: P2X7 is the trigger, ANO6 is the scramblase

GO:0017121 plasma membrane phospholipid scrambling is IDA on P2X7 from
PMID:25651887 with the decisive control
PMID:25651887

P2X7 supplies the calcium; ANO6 flips the lipids. KEEP_AS_NON_CORE, and
positive regulation of plasma membrane phospholipid scrambling is proposed as a new term, because
GO has no regulation child under GO:0017121 and so annotates trigger and executor identically. This
is the same distinction that had to be drawn for TMC1 in this batch ("necessary for" vs "performs"),
approached from the opposite direction.

Other findings

• GO:0001530 lipopolysaccharide binding (IEA + ISS): MARK_AS_OVER_ANNOTATED. It traces to a
  putative LPS-binding motif in the rodent C-terminal tail, a region the full-length structures assign
  to the C-cys anchor and the cytoplasmic ballast. Biologically it collapses a two-signal pathway:
  TLR4 is the LPS receptor (signal 1), P2X7 is the ATP sensor (signal 2).
• GO:0005267 potassium channel activity and GO:0005272 sodium channel activity (both IDA,
  PMID:17483156): MODIFY to GO:0004931. The recordings are real - the study substituted the charge
  carrier systematically - but the terms name selective channels and P2X7 is non-selective. The fluxes
  themselves remain recorded as GO:0071805 and GO:0035725.
• GO:0005524 ATP binding: kept non-core, not removed. Literally true, but the site is
  extracellular and the protein neither hydrolyses nor uses ATP, so on a surface receptor the term
  invites the wrong reading. Complicated by the structures, which found a genuine cytoplasmic
  guanosine nucleotide site PMID:31587896.
• GO:0042802 identical protein binding was ACCEPTed, not swept up with bare protein binding:
  for an obligate homotrimer, self-association is a real and mechanistically loaded statement.
• GO:0002931 response to ischemia (NAS, PMID:12849743): the weakest evidence code in the set, and
  the cited abstract is about P2X2 and P2X4 PMID:12849743.
  Flagged MARK_AS_OVER_ANNOTATED and MISCITED rather than removed, because the full text was not
  available.
• Reactome:R-HSA-139855 is titled "P2X1-mediated entry of Ca++ from plasma" and is used as a TAS
  source for a P2RX7 plasma-membrane annotation. The localisation is right; the event is about a
  different subtype. Recorded as WRONG_IDENTIFIER in reference_review.
• PMID:17299767 ("Involvement of P2X4 and P2Y12 receptors in ATP-induced microglial chemotaxis")
  is likewise used only for a plasma-membrane TAS on P2RX7 - harmless, but a poor reference choice.
• Nine bare GO:0005515 protein binding rows: MARK_AS_OVER_ANNOTATED per project guidance.
• GO:0005737 cytoplasm (ISS): MARK_AS_OVER_ANNOTATED. A two-transmembrane surface receptor.

Qualifier note

No NOT| qualifiers appear anywhere in P2RX7-goa.tsv - checked directly. All 106 GOA rows are
positive assertions.

PMID verification

Every PMID cited was checked against PubMed metadata or the cached record before use.