Neurons are compartmentalized into functionally distinct domains—axons and dendrites—designed for transmission and processing of information, which underlies every aspect of brain function. Neuronal signaling is predicated on mechanisms that control the targeting of proteins to specific subcellular domains as well as the recruitment of ion channels to dendritic spines, which serve as primary sites of synaptic communication.

SNAREs are a family of proteins involved in mediating membrane fusion events of trafficking vesicles that deliver protein and lipid cargo to specific intracellular destinations, but their role in neuronal polarity and selective protein targeting are unknown. Using a combination of optical and molecular approaches, we have now determined that plasma membrane SNAREs, syntaxins 3 and 4, respectively, are expressed in hippocampal neurons and that polarized expression of syntaxin 3 confers specificity to axonal protein trafficking. Syntaxin 3 localizes to the axonal plasma membrane, particularly to axonal tips, whereas syntaxin 4 localizes to the somatodendritic plasma membrane. Mutating residues in the conserved N-terminal targeting motif causes mislocalization of syntaxin 3, resulting in simultaneous mistargeting of the axonal protein NgCAM, but not of the somatodendritic transferrin receptor. RNA interference-mediated knockdown of endogenous syntaxin 3 leads to partial mistargeting of NgCAM, demonstrating that syntaxin 3 is critical to axonal targeting.

Scaffolding proteins at postsynaptic sites are crucial for organizing ion channels into signaling complexes essential for synaptic transmission. Inward rectifier potassium Kir2.3 channels are enriched at spines, when expressed in cultured hippocampal neurons, but the mechanisms underlying channel localization are unclear. We used mutational analysis to determine how protein-interacting motifs in the C-terminus of Kir2.3 influence its localization. We show that Kir2.3 channels traffic to spines and that synaptic localization requires a previously identified C-terminal PDZ (postsynaptic density-95/Discs large/zona-occludent-1)-binding motif. We also identify a polyproline motif in the C-terminus of Kir2.3 that associates with Homer-1 and Homer-2 in brain. Deleting the PDZ-binding motif abolishes binding of PSD-95 with Kir2.3 and dramatically reduces Kir2.3 synaptic localization, whereas mutating residues in the polyproline motif has little effect on channel localization. Using RNA interference, we show that PSD-95 is necessary for localizing Kir2.3 to spines.