Our investigation of synapse formation in the HSN neurons found that a pair of Ig superfamily proteins, SYG-1 and SYG-2, are critical for specification of synapses. SYG-2, expressed by the surrounding guidepost cells, binds directly to SYG-1 on the HSN neuron and concentrates it to the short stretch of the axon where synapses form. SYG-1 binds directly to the WAVE regulatory complex (WRC) through a conserved sequence motif in its cytoplasmic domain. The WRC is required to build a local F-actin patch, which in turn recruits an adaptor protein, NAB-1/neurabin. NAB-1 binds to both F-actin and AZ scaffolding proteins SYD-1, SYD-2, and ELKS-1. SYD-1, SYD-2, and ELKS-1 form a cytomatrix meshwork that recruits functional AZ proteins and synaptic vesicles. These studies showed that synaptic adhesion molecules trigger synapse formation by organizing major cytoskeleton components and recruiting presynaptic components.

 Although multiple other cell-adhesion and secreted molecules have been found to stimulate the assembly of synapses, the contribution of signals that negatively regulate synaptic development is not well understood. We examined synapse formation in the C. elegans cholinergic motor neuron DA9, whose en passant presynapses are restricted to a specific segment of its axon. We report that a signaling pathway composed of the Wnts lin-44 and egl-20, the Wnt receptor lin-17/Frizzled, and the cytoplasmic effector dsh-1/Dishevelled defines the subcellular location of DA9 presynapses by inhibiting their assembly in regions of the axon proximal to the sources of Wnt. LIN-44/Wnt is secreted from the tail hypodermis and localizes LIN-17/Frizzled to a subdomain of the DA9 axon that is devoid of presynaptic specializations (Figure 1). Conversely, the ectopic overexpression of LIN-44 in cells adjacent to DA9 is sufficient to expand LIN-17 localization in the DA9 axon and concomitantly displace the assembly of presynaptic terminals. These results demonstrate a novel role for Wnt signaling in inhibiting synapse formation and suggest that morphogenetic signals can spatially regulate the patterning of synaptic connections by subdividing regions of the axon into discrete domains.

The long axonal projection of DA9 far exceeds the anterior border of the synaptic domain, suggesting that inhibitory mechanisms might restrict the synaptic formation along the distal axon. We found that the fasciculating sister axon DA8 tiles its synaptic domain with DA9 by inhibiting synaptogenesis in the DA9 distal axon. This inhibition is mediated by two transmembrane semaphorins and their plexin receptor, PLX-1. PLX-1 forms a concentrated patch at the distal synapse-free axon, delineating the distal synaptic border. Together with the results on Wnts and another project that shows that UNC-6/netrin can also inhibit synapse formation, these findings argue strongly that inhibitory cues from the environment shape synaptic connections in vivo (Figure 1).

Once the extrinsic cues determine the regions of synapse formation on axons, AZ proteins and synaptic vesicles (SVs) aggregate into presynaptic specifications with defined size. We found that AZ proteins and SVs are co-trafficked by the UNC-104/KIF1A motors from the cell body to the synaptic domain. Along the axon shaft, the trafficking organelles make numerous stops en route to the synapses. At these stop sites, mobile trafficking organelles frequently join the stationary packets (capture); the packets also split into mobile puncta (dissociation). The balance between capture and dissociation rates dictates the size of the puncta. The conserved small G protein ARL-8 is localized on synaptic vesicles and is critical for the number, size, and location of synapses because it regulates the capture and dissociation rates. ARL-8 directly interacts with the kinesin motor UNC-104/KIF1A. ARL-8 also antagonizes the synapse assembly activity of the AZ proteins. By tuning the motor protein and the aggregation activity of the AZ proteins, ARL-8 promotes axonal transport and reduces the size of presynaptic specializations.

 In C. elegans, the sensory neuron PVD establishes stereotypical, highly branched dendrite morphology. We used genetic approaches to identify a tripartite ligand-receptor complex of membrane adhesion molecules, which is both necessary and sufficient to effect spatially restricted growth and branching of PVD dendrites. SAX-7/L1CAM and MNR-1 function as a ligand complex at defined locations in the surrounding hypodermal tissue, and DMA-1 acts as the cognate receptor on PVD. Mutations in this complex lead to dramatic defects in the formation, stabilization, and organization of the dendritic arbor. Ectopic expression of SAX-7 and MNR-1 generates a predictable, unnaturally patterned dendritic tree in a DMA-1–dependent manner. Both in vivo and in vitro experiments indicate that all three molecules are needed for interaction and that binding requires the FnIII domains of SAX-7, leucine-rich repeats of DMA-1, and an unknown domain of MNR-1.