GRID1 (GluD1, glutamate receptor ionotropic delta-1) — review notes

UniProt: Q9ULK0. Human, 1009 aa, 3 TM segments + re-entrant pore loop, ATD + LBD
(S1/S2) + TMD + long CTD — the canonical iGluR architecture. UniProt places it in
the "glutamate-gated ion channel (TC 1.A.10.1) family, GRID1 subfamily".

The contested question: is GluD1 an ion channel, and what is the agonist?

This is the whole review. GluD1 has the architecture of an ionotropic glutamate
receptor but does not behave like one, and the field currently holds several
mutually incompatible positions at once. GOA has annotated most of them.

Position 1 — it is not a canonical ligand-gated channel; it is a scaffold/transducer

Position 2 — the pore is real and carries current in native neurons, but it is gated indirectly

Position 3 — the agonist is GABA (2023), or acetylcholine (2026)

My position

  1. Core = trans-synaptic organizer / signal transducer. This is the only role with
    converging, independently replicated, mechanism-level evidence (Dai 2021 chimeras;
    Piot 2023 requirement for trans-synaptic anchoring; Ryu 2011 presynaptic
    differentiation; Liu 2020 thalamostriatal circuit). Family membership is not function.
  2. Core = GABA receptor, non-ionotropic. Direct structural + functional evidence
    (Piot 2023), and the GO term's definition fits.
  3. Non-core, but real = a cation conductance in native neurons (Gantz 2020, Copeland
    2023), tonic and/or augmented by Gq-coupled receptors. Not directly transmitter-gated.
  4. Unresolved = direct transmitter gating (GO:1904315). One 2026 report (ACh)
    supports it; the standing literature (PNAS 2024) and the GABA paper itself argue
    against it. UNDECIDED is the honest action.
  5. Wrong = the AMPA-receptor annotations. GluD1 does not bind glutamate at all
    (UniProt: "does not bind glutamate as a primary ligand"), so GO:0004971 AMPA glutamate receptor activity (definition: "exhibits fast gating by glutamate") and
    GO:0032281 AMPA glutamate receptor complex are simply false, and the
    inter-ontology-inferred GO:0035235 ionotropic glutamate receptor signaling pathway
    falls with them.

Why the AMPA IBAs are a propagation failure (not just a quibble)

GO:0004971 and GO:0032281 on GRID1 both come from PANTHER node
PTN000438081. Every donor in the WITH/FROM is an AMPA receptor subunit
(UniProtKB:P42261 = human GRIA1, P42263 = GRIA3, P48058 = GRIA4, MGI:95808 = mouse
Gria1, plus the rat Gria orthologues RGD:61862/61863/621531/70958). No delta-receptor
donor appears. The IBD was therefore placed on the AMPA-receptor clade, and GRID1 —
a member of the delta branch, which diverged before the AMPA/kainate/NMDA split —
should not be inside it. The same node also seeded GO:0043197 dendritic spine, which
happens to be true of GluD1 for independent reasons and so is harmless.

Contrast GO:1904315, which comes from the much broader node PTN001826301 whose
donors span AMPA, kainate, NMDA and delta (UniProtKB:O43424 = human GRID2). That node
placement is a defensible judgement about the ancestral iGluR; the question there is
whether the delta branch retained direct transmitter gating, which is exactly the open
question. Hence REMOVE for the first pair, UNDECIDED for the second.

The GOA incoherence, itemised

GOA currently asserts that one protein is, simultaneously:
* a GABA receptor (GO:0016917, IDA, PMID:38060673) — defensible, non-ionotropic;
* an AMPA glutamate receptor (GO:0004971, IBA) — false, mis-propagated;
* a G-protein-coupled receptor (GO:0099530, IDA, PMID:27276689) — structurally
impossible; the GPCR in that paper is mGlu1;
* a transmitter-gated ion channel (GO:1904315, IBA/IEA/ISS) — unresolved;
* a generic ligand-gated ion channel (GO:0015276, IEA from InterPro) — the specific
claim refuted by PMID:39052831;
* a signaling receptor regulator (GO:0030545, IDA, PMID:34135511) — correct, and the
best single MF summary of what GluD1 actually does.

GO:0099538 synaptic signaling via neuropeptide (IDA from PMID:27276689, plus ISS and
IEA rows) is a third oddity: no neuropeptide appears anywhere in the GluD1 literature.
The GO definition ("Cell-cell signaling to or from a synapse, mediated by a peptide")
is loose enough that a curator could have reached it for cerebellin-mediated signalling,
but GO:0099557 says that precisely and is already on the gene, so these rows are
re-pointed rather than kept.

GO:0005515 protein binding rests on two IntAct/BioPlex rows whose sole partner is
UniProtKB:P68871 = haemoglobin subunit beta, a textbook affinity-purification
contaminant and not a plausible partner for a postsynaptic receptor.
GO:0070062 extracellular exosome is HDA from a urinary exosome proteome
(PMID:19056867) — GluD1 is a brain protein.

Non-core biology worth keeping

Validation / provenance housekeeping