GPR158 review notes

Gene: human GPR158 (UniProt Q5T848), class C orphan-derived GPCR, HGNC:23689.
Reviewed as part of a contested-function batch (see also GPR75, GPR25).

Why this gene is contested

Three competing models of what GPR158 "is" coexist in the 2017-2026 literature:

1. Osteocalcin (OCN) receptor — the bone-brain axis model.
2. Metabotropic glycine receptor (mGlyR) — the 2023 deorphanization.
3. Ligand-independent RGS7-Gβ5 anchor / synaptic organizer — the oldest and,
   by structural standards, the best-established role.

Crucially these are not mutually exclusive: the mGlyR model is built on the RGS7
model, because glycine binding is read out as inhibition of the receptor-bound
RGS7-Gβ5 complex, not as Gα activation.

Model 3 (RGS7-Gβ5 anchoring) — the structural backbone

GPR158 was first characterised as an RGS-anchoring receptor:
PMID:22689652

and then as an allosteric enhancer of RGS7 catalytic activity:
PMID:25792749

Two independent 2021 cryo-EM structures nailed the architecture and the GPR158-RGS7
interface:
PMID:34793198 and
PMID:34815401

The same 2019 study that provides the GOA enzyme activator activity IDA explicitly
reports that GPR158 does not behave as a canonical G-protein-activating GPCR:
PMID:31189666 and
PMID:31189666

This is the most reproducible, most mechanistically resolved thing GPR158 does, and it
is the reason GO:0008047 enzyme activator activity (IDA, two independent papers) is
treated here as core.

Model 2 (metabotropic glycine receptor) — the 2023 deorphanization

PMID:36996198 with a defined binding site and a defined
effector:
PMID:36996198 and
PMID:36996198, with a cellular readout
PMID:36996198

This has independent functional replication in a different laboratory and a different
brain region (Aceto et al., Università Cattolica del Sacro Cuore, nucleus accumbens):
PMID:38884814

A 2026 hippocampal study reports functional metabotropic glycine responses in CA3
pyramidal cells, Gi/o-dependent, though it does not itself manipulate GPR158:
PMID:42347714

A contemporaneous commentary frames the deorphanization as accepted in the field:
PMID:37321907

UniProt has adopted it as the recommended protein name ("Metabotropic glycine receptor").
Position taken: GO:0160079 is core, and the direction of residual doubt is about
whether glycine is the only or dominant physiological input, not about whether the
binding/response is real.

Model 1 (osteocalcin receptor) — real signal, weaker molecular grounding

The OCN model rests on mouse genetics and electrophysiology rather than on direct
high-affinity binding to a defined pocket:
PMID:28851741 (Gpr158 mediates osteocalcin's regulation
of cognition, J Exp Med 2017).

Reviews continue to state it flatly:
PMID:40337551

But UniProt itself hedges this one ("may also act as a receptor for osteocalcin ...
By similarity"), and no GO term for osteocalcin binding/receptor activity is carried in
GOA for this gene. No new term is proposed here: the claim is a candidate second
ligand, and asserting it in a machine-readable slot would overstate the evidence.
It is recorded in suggested_questions instead.

The 2025 ligand-independent module (Dev Cell)

PMID:40393451 and
PMID:40393451

This is a third effector arm (RGS7-Gβ5, and now PLCXD2), and no ligand is invoked.
It does not contradict the glycine model; it reinforces the general picture that GPR158
is an atypical receptor that works by holding and gating intracellular effectors rather
than by activating Gα. It is the main reason GPR158's synapse-organisation annotations
(GO:0050807) are kept as genuine rather than demoted.

The nucleus annotations (GO:0005634)

Two EXP annotations (PMID:23451275, PMID:30855200) from one group, in trabecular
meshwork / ocular cells:
PMID:23451275

UniProt records the location but flags it as mechanistically unexplained: "Trafficks
between the nucleus and the cell membrane; it is unclear how a multi-pass membrane
protein can traffick between the nucleus and the cell membrane (PubMed:23451275)."

Per project rules an experimental annotation is not removed from an abstract reading.
Action taken: KEEP_AS_NON_CORE on all three GO:0005634 rows (one IEA from the
SubCell mapping plus the two EXP rows), with the caveat recorded, because it is
cell-type-restricted, unreplicated outside this group, and irreconcilable with the
dimeric 7TM postsynaptic architecture that the structural work established.

Curation position taken

• GO:0008047 enzyme activator activity (IDA ×2, PMID:36996198 and PMID:31189666) →
  ACCEPT, core. Allosteric activation of RGS7-Gβ5 GAP activity. Best-supported MF.
• GO:0160079 G protein-coupled glycine receptor activity (IDA) → ACCEPT, core.
  Independent replication exists; the reason field records that the ligand assignment
  is one of three competing models.
• GO:0004930 GPCR activity and GO:0004888 transmembrane signaling receptor activity
  → MODIFY to GO:0160079. Correct in kind but under-specific, and the canonical
  "activates an associated G protein" mechanism these terms imply is explicitly not
  what GPR158 does (PMID:31189666).
• GO:0007186 GPCR signaling pathway → MODIFY to GO:0008277 regulation of GPCR
  signaling pathway, which is what GPR158 actually does via RGS7-Gβ5.
• GO:0008277, GO:0072659 (recruiting RGS7-Gβ5 to the membrane), GO:0050807 →
  ACCEPT.
• GO:0005886, GO:0045211, GO:0098839, GO:0016020 → ACCEPT (locations).
• GO:0042734 presynaptic membrane → KEEP_AS_NON_CORE (UniProt: "Mainly localizes
  to the postsynaptic membrane, with a small portion to the presynaptic membrane").
• GO:0005634 nucleus → KEEP_AS_NON_CORE (see above).
• GO:0001956, GO:0007420, GO:0050890 → KEEP_AS_NON_CORE (organismal/distal
  phenotypes, not what the protein does).

Unresolved

• Is osteocalcin a direct GPR158 ligand, or does GPR158 sit downstream of OCN action
  by another route? No direct-binding structure exists.
• Does glycine occupancy of the Cache domain account for the synaptic-organiser and
  PLCXD2 phenotypes, or are those genuinely ligand-independent?
• The nuclear pool: artefact of overexpression/antibody, a proteolytic fragment, or real?