Human LRFN2 encodes SALM1, a type-I single-pass membrane protein. The comparative
mouse/human family study reports that Lrfn proteins are glycoproteins spanning the
plasma membrane with extracellular N termini and that Lrfn2's C-terminal tail binds
PSD95 PDZ domains [PMID:16828986 "Lrfn1-5 commonly encode glycoproteins spanning the
plasma membrane, with their N-terminus located on the extracellular side.";
"C-termini of Lrfn1, Lrfn2 and Lrfn4 were bound by PDZ domains of postsynaptic
protein PSD95"]. The original SALM1 work used rat brain and cultured rat hippocampal
neurons; it places SALM1 in synaptic-membrane and postsynaptic-density fractions and
also in axons and dendrites PMID:16495444. These rodent studies support conservation of
the human protein's architecture and synaptic role, but they are not direct human
neuronal assays.
Rat-brain co-immunoprecipitation showed SALM1 association with PSD-95, SAP102, and
SAP97, while neuronal overexpression recruited NMDA receptors and PSD-95 to dendritic
puncta [PMID:16495444 "SALM1 interacts with PSD-95, synapse-associated protein 102
(SAP102), and SAP97 based on coimmunoprecipitation of detergent-solubilized brain.";
"Overexpression of SALM1 in 14 DIV neurons recruits NMDA receptors (NR) and PSD-95 to
dendritic puncta."]. The Reactome human model captures two possible routes to the
NMDA-receptor complex: direct extracellular SALM1-GRIN1 association and indirect
recruitment through PSD-95 [Reactome:R-HSA-8849906 "SALM1 can directly interact with
the extracellular domain of the NR1 subunit of NMDA receptor or indirectly by
binding to PSD-95"]. This event is an orthology/conservation-based human model; its
decisive primary assays were not performed in human neurons.
Mouse Lrfn2 knockout analysis supports a central role in excitatory synapse
maturation: mutant synapses had smaller PSDs, lower AMPA/NMDA ratios, and enhanced
LTP in one line PMID:28604739. The same study found that AMPAR surface
expression depends on Lrfn2-PSD-95 association PMID:28604739. A second mouse knockout study instead reported enhanced
NMDAR transmission but suppressed NMDAR plasticity, plus fewer inhibitory synapses
and reduced inhibitory transmission PMID:29798891. The apparently divergent plasticity
directions should remain visible rather than be synthesized into a single precise
effect.
LRFN2 also participates in AMPA-receptor recycling through sorting nexin 27. A
purified LRFN2 tail peptide directly bound the SNX27 PDZ domain with micromolar
affinity PMID:34251337. In rat
neurons, LRFN2 associated with AMPARs, and LRFN2 knockdown decreased surface AMPAR,
synaptic activity, and hippocampal LTP PMID:34251337.
SALM1 is not exclusively postsynaptic. In mouse hippocampal neurons it was found at
both pre- and postsynaptic membranes, and depletion on either side impaired
neurexin/neuroligin-mediated excitatory synapse formation PMID:31368584. Presynaptic SALM1 promotes
F-actin/PIP2-dependent cis clustering of neurexin PMID:31368584. This is not a direct SALM1-neurexin biochemical
interaction and should not be described as a trans-synaptic SALM1 ligand pair.
SALM1 also directly associates through its PDZ-binding tail with CASK in the
CASK/Mint1/Lin7b presynaptic organizer PMID:31368584.
SALM1-3 form complexes in rat brain, but the family-association study found no
trans-cellular association for SALM1-3; only SALM4 and SALM5 formed homophilic trans
associations in its assay [PMID:18227064 "SALMs 1-3 strongly co-immunoprecipitated
with each other"; "Both SALMs 4 and 5 formed homophilic, but not heterophilic
associations, whereas no trans associations were formed by the other SALMs."]. The
Reactome LRFN2 event should therefore be understood as a cis SALM complex
[Reactome:R-HSA-8849900 "SALM1, SALM2, and SALM3 form homo- and heteromeric complexes
in a cis manner."].
The classic mixed-culture presynaptic-induction phenotype belongs to SALM3/LRFN4 and
SALM5/LRFN5, not SALM1/LRFN2 PMID:20410109. This does not contradict the
distinct presynaptic SALM1 mechanism in PMID:31368584, which acts by cis-regulating
neurexin clustering in the SALM1-expressing neuron.
A yeast two-hybrid screen using the SALM1 extracellular domain recovered RTN3A1 and
mapped association to the conserved LRR domain PMID:19681166. Brain immunoprecipitation detected an
RTN3C-sized band with SALM1-4, while a separate 90-kDa signal was selective for
SALM2/3 PMID:19681166. Reactome consequently
describes SALM1's interaction as relatively weak compared with SALM2/3
[Reactome:R-HSA-8849882 "RTN3 ... tightly associates with SALM2 and SALM3 ... and
interacts relatively weakly with SALM1 and SALM4."]. Evidence for a trafficking role
is suggestive, not a demonstrated LRFN2 molecular function.
SALM family overexpression and perturbation in cultured rat hippocampal neurons
support a neurite-outgrowth phenotype, with SALM1-3 PDZ-binding domains required for
most measured effects PMID:18585462. This is a family-level
in-vitro phenotype and is weaker evidence for a human LRFN2 core function than the
orthologous knockout synapse data.
In mouse retina, LRFN2 is selectively localized at cone terminals with DBC signalplex
members PMID:38408870. Lrfn2 loss reduced the
photopic ERG b-wave only at the brightest intensities, indicating impaired cone-to-
cone-DBC transmission PMID:38408870. The responsible molecular partner and
whether LRFN2 is pre-, post-, or bidirectionally positioned at this retinal synapse
remain unresolved.