Spatial Regulation of an E3 Ubiquitin Ligase Directs Selective Synapse Elimination

See allHide authors and affiliations

Science  17 Aug 2007:
Vol. 317, Issue 5840, pp. 947-951
DOI: 10.1126/science.1145727


Stereotyped synaptic connectivity can arise both by precise recognition between appropriate partners during synaptogenesis and by selective synapse elimination. The molecular mechanisms that underlie selective synapse removal are largely unknown. We found that stereotyped developmental elimination of synapses in the Caenorhabditis elegans hermaphrodite-specific motor neuron (HSNL) was mediated by an E3 ubiquitin ligase, a Skp1–cullin–F-box (SCF) complex composed of SKR-1 and the F-box protein SEL-10. SYG-1, a synaptic adhesion molecule, bound to SKR-1 and inhibited assembly of the SCF complex, thereby protecting nearby synapses. Thus, subcellular regulation of ubiquitin-mediated protein degradation contributes to precise synaptic connectivity through selective synapse elimination.

Synapse elimination is a developmental hallmark of neural circuit refinement (1). In the vertebrate neuromuscular junction, the neurotransmitter acetylcholine (ACh) drives elimination of postsynaptic ACh receptor clusters (2, 3). However, little is known about the molecular machinery that eliminates presynaptic specializations, or the mechanisms that selectively eliminate certain synapses while sustaining others.

We investigated synapse elimination in the egg-laying motor neuron of C. elegans, the HSNL. In adult animals, the HSNL connects to its targets, the vulval muscles and VC motor neurons, via a cluster of synapses localized exclusively to the vulval region, the primary synapse region (PSR). These synapses were visualized with the vesicle marker synaptobrevin fused to yellow fluorescent protein (SNB-1::YFP) (4) (Fig. 1). To determine whether synapses specifically localize to the PSR as a result of synapse elimination, we followed the development of synapses in individual live animals. At early stages of development, we observed that additional SNB-1::YFP puncta formed immediately anterior to the vulva, the secondary synapse region (SSR) (Fig. 1, B and E). Stereotyped elimination of these puncta gave rise to the synaptic pattern present in adult animals (Fig. 1, A to I). The average number of SSR puncta per animal and the percentage of animals with SSR puncta gradually decreased as the animals matured (Fig. 1J). In addition to SNB-1, these SSR puncta contained synaptic vesicle protein RAB-3/rab3 (5), active zone protein SYD-2/liprin (6) (Fig. 1, K to P), and presynaptic protein UNC-10/RIM (fig. S1). Thus, SSR puncta probably represent bona fide presynaptic structures.

Fig. 1.

Synapse elimination at the SSR in the HSNL. The vulval epithelial cells are used as a staging criterion for early L4 (A to C), mid-L4 (D to F), and young adult (G to I) animals. SNB-1::YFP reveals SSR puncta in early and mid-L4 animals [(B) and (E)], which are absent in adult worms (H). Asterisks mark the vulva; arrows and brackets indicate SNB-1::YFP puncta at the SSR and PSR, respectively; arrowheads mark the HSNL cell body. (J) Quantification of SNB-1::YFP puncta in the SSR; error bars, SEM. (K to S) RAB-3 colocalizes with SYD-2 at synapses in both the PSR and SSR [enlarged view in (K) to (P)]. All images are lateral views, anterior to the left and ventral down. Scale bar, 20 μm.

The immunoglobulin superfamily (IgSF) protein SYG-1 (homologous to NEPH1 and IrreC in mouse and Drosophila, respectively) is an essential determinant of synaptic specificity in the HSNL (4). In syg-1(ky652) null mutants, the SNB-1 puncta at the SSR failed to be eliminated and instead persisted into adulthood (Fig. 2A), whereas synapses at the PSR were greatly diminished (4). A weak allele of syg-1, wy2, displayed SSR puncta in 100% of mid-L4 animals; however, only 47.3% of wy2 adults showed SSR puncta (Fig. 2A). Thus, the level of functional SYG-1 may influence synapse elimination. To test this idea, we generated animals with varying dosages of SYG-1. Heterozygous syg-1(ky652) animals exhibited a level of SSR puncta intermediate to those of homozygous and wild-type animals. Furthermore, overexpression of full-length SYG-1 resulted in a lower frequency of SSR puncta relative to the wild type (Fig. 2B). Thus, the level of SYG-1 in the HSNL directly correlates with the extent of SSR puncta elimination.

Fig. 2.

SYG-1 is required for SSR synapse elimination. (A) Quantification of SSR puncta in a null syg-1(ky652) allele and a weak syg-1(wy2) allele. (B) SYG-1 dosage effect on the SSR SNB-1::YFP puncta. **P < 0.01; n ≥ 100. (C to K) SSR SNB-1::YFP puncta are located anterior to the developing vulva (dashed boxes) and do not overlap with SYG-1::CFP. Asterisks indicate the vulva.

SYG-1 localizes near the vulva region through its interaction with SYG-2, another IgSF protein expressed in the guidepost epithelial cells (7). To understand the relationship between SYG-1 localization and SSR puncta fate, we examined SYG-1 subcellular localization during synapse elimination. In all developmental stages examined, SYG-1 localized exclusively to the PSR where SNB-1 puncta persisted (Fig. 2, C, F, and I). SNB-1 puncta at the SSR (Fig. 2, D and G) did not overlap with SYG-1 fused to cyan fluorescent protein (SYG-1::CFP) (Fig. 2, C to K).

To understand how SYG-1 promotes elimination of SSR synapses, we performed a yeast two-hybrid screen and found that the SYG-1 intracellular domain bound to SKR-1 (fig. S2). This result was confirmed by coimmunoprecipitation (Fig. 3A). SKR-1 is the ortholog of vertebrate Skp1, a core component of the Skp1–cullin–F-box (SCF) complex (8). SCF complexes are E3 ubiquitin ligases that transfer ubiquitin to target proteins destined for degradation by the proteasome (9). Regulated ubiquitin-proteasome system (UPS) activity controls diverse aspects of neuronal development and function, including neurite outgrowth, synapse growth, and pre- and postsynaptic receptor trafficking (1014). To determine skr-1's role in synapse elimination, we performed a loss-of-function analysis using an RNA interference (RNAi) approach. At both the mid-L4 and adult stages, more skr-1 RNAi–treated animals than control animals displayed SSR puncta (Fig. 3, B and C). RNAi knockdown of CUL-1/Cul1, another obligatory component of the SCF complex (8), produced synapse elimination defects similar to those of skr-1 RNAi–treated animals (fig. S2). Therefore, an SCF complex is essential for proper synapse elimination.

Fig. 3.

The SCFSEL-10 complex is required for synapse elimination in the HSNL. (A) SKR-1 immunoprecipitates the intracellular domain of SYG-1 in 293T cells. (B and C) The SSR puncta are not eliminated completely in skr-1 RNAi–treated animals. (D and E) The sel-10 synapse elimination defect is rescued by expressing sel-10 in HSNs but not in other tissues. Punc-86 is expressed in HSNs and other neurons; Pegl-17, in the vulva epithelium; Plin-11B, in VCs and the vulva epithelium; Psra-6, in PVQ and ASH neurons; Punc-4, in VC neurons. ***P < 0.001, **P < 0.01; NS, not significant; n ≥ 100.

F-box proteins provide target specificity for SCF complexes (8). We next tested candidates to identify the specific F-box protein that regulates synaptic elimination in the HSNL. FSN-1 and LIN-23, two F-box proteins that have been implicated in presynaptic differentiation and postsynaptic receptor trafficking (15, 16), did not affect the HSNL synaptic pattern (fig. S2). We then investigated another F-box protein, SEL-10, which is known to bind to SKR-1 and functions in the Notch signaling pathway as well as in sex determination in C. elegans (17, 18). Interestingly, sel-10 mutants displayed a phenotype similar to that of skr-1 knockdown worms: increased SSR puncta as detected with both SNB-1 and the active zone marker SYD-2 (Fig. 3 and fig. S2). Thus, SEL-10 may serve as a specific F-box protein in synapse elimination in the HSNL.

Next, we sought to identify the site of action for sel-10 in synapse elimination. A transgene containing the sel-10 promoter driving green fluorescent protein (GFP) labeled a subset of neurons including the HSNL, which suggests that sel-10 is expressed in the HSNL (fig. S3). Then we expressed the sel-10 cDNA under the control of various cell-specific promoters and asked which promoters could rescue the mutant phenotype. An unc-86 promoter driving the sel-10 cDNA, which confers expression in HSNs, fully rescued the sel-10 mutant phenotype (Fig. 3, D and E). In contrast, when we used promoters that drove expression in cells adjacent to the HSNL, the sel-10 defect remained (Fig. 3D). Finally, we examined the subcellular localization of SEL-10. SEL-10::GFP was diffusely localized along the entire axon throughout development (fig. S3). Thus, SEL-10 functions cell-autonomously to regulate synaptic elimination and is present at both the PSR and SSR.

Because SCF E3 ubiquitin ligases participate in UPS-mediated protein degradation (8), we asked whether proteasome activity is required for synapse elimination. We found that SSR synapse elimination was drastically compromised in animals treated with lactacystin, a proteasome inhibitor (19), which suggests that UPS-mediated protein degradation is required for synapse elimination (fig. S4).

Because the SCFSEL-10 complex is present at both the PSR and SSR, why does synapse elimination occur specifically at the SSR but not at the PSR? We considered two models. First, the SCF-UPS pathway could exert its effect by regulating the expression of syg-1. SYG-1 would then promote synapse elimination through an unidentified pathway. Alternatively, SYG-1 could regulate SCF complex formation or activity. To distinguish between these models, we examined SYG-1::GFP expression in lactacystin-treated animals and sel-10 mutants. In these worms, the SYG-1::GFP pattern was indistinguishable from that of wild-type animals (Fig. 4, A and B). Furthermore, the SEL-10 WD-repeat domain, which recruits substrates (20), did not bind the intracellular domain of SYG-1 (fig. S4). Therefore, it is unlikely that the SCF complex controls the level or localization of SYG-1.

Fig. 4.

SYG-1 inhibits SCFSEL-10 assembly. (A) SYG-1::GFP expression is not altered in sel-10 mutants. Arrows indicate SYG-1::GFP; asterisks mark the vulva. Scale bar, 20 μm. (B) Quantification of SYG-1::GFP intensity. NS, not significant; n = 20. (C) Coimmunoprecipitation of SKR-1 and SEL-10 is reduced in the presence of SYG-1 intracellular domain. (D) Quantification of the SKR-1–SEL-10 interaction. (E) Effects of SCF activity and SYG-1 on PSR synaptic intensity. Depletion of SCF or overexpression of SYG-1 intracellular domain enhances PSR synapses; n = 20. (F) Overexpression of skr-1 reduces the SSR synapses, whereas overexpression of SYG-1 intracellular domain or the F-box domain of SEL-10 causes an increase in SSR synapses; n ≥ 100. **P < 0.01, ***P < 0.001.

To investigate whether SYG-1 regulates the formation of the SCFSEL-10 complex, we assayed the interaction of SEL-10 with SKR-1 by coimmunoprecipitation. The binding between SKR-1 and SEL-10 was significantly weaker in the presence of the SYG-1 intracellular domain (Fig. 4, C and D), which suggests that the SYG-1–SKR-1 interaction inhibited the assembly of SCF complexes. Because SYG-1 protein was strictly localized to the PSR (Fig. 2, C, F, and I), whereas SCFSEL-10 was present at both the PSR and SSR (fig. S3) and synapse elimination only occurred at the SSR, we hypothesized that SCFSEL-10 degraded synaptic components and its activity was attenuated at the PSR because of the presence of SYG-1 (fig. S5). Furthermore, in syg-1 mutants, PSR synapses are significantly reduced and SSR synapses are increased (4). One explanation for this observation is that the SCF complex may be limiting within the HSNL; a high level of SCF activity at the PSR causes depletion of SCF at the SSR, and vice versa. Therefore, in the wild type, SYG-1 localizes at the PSR and interferes with the assembly of the SCF complex by binding to SKR-1. As a result, SCF activity is low at the PSR and high at the SSR, leading to synapse stabilization at the PSR and synapse elimination at the SSR (fig. S5). In the absence of SYG-1, SCF activity increases at the PSR, leading to synapse elimination, whereas at SSR the SCF activity is low and synapses persist.

To validate this hypothesis, we tested three additional predictions suggested by the model. First, if the SCF is rate-limiting for synapse degradation, then increasing the SCF complex levels should enhance synapse elimination at both the SSR and PSR sites. Indeed, overexpressing SKR-1 protein resulted in reduced PSR (Fig. 4E) and decreased occurrence of SSR synapses (Fig. 4F). Conversely, overexpressing a dominant negative form of SEL-10 that contained only the F-box domain inhibited synapse elimination and led to more SSR puncta (Fig. 4F).

Second, if SYG-1 protects synapses from degradation through its inhibition of SCF, a form of SYG-1 that is diffusely localized should inhibit SCF assembly and perturb synapse elimination at both the SSR and PSR. Consistent with our model, the expression of a freely diffusing SYG-1 intracellular domain construct increased the PSR puncta intensity in syg-1 mutants (Fig. 4E and fig. S6) and the SSR synapses in wild-type animals (Fig. 4F).

Third, if the SCF complex also functions in synapse elimination at the PSR, we predict that inhibiting SCF activity should suppress the diminished PSR phenotype of syg-1 mutants. Indeed, in syg-1 mutant animals, compromising SCF by skr-1 RNAi or a sel-10 mutation significantly enhanced PSR synaptic intensity (Fig. 4E). These three lines of evidence support a model in which SYG-1 protects the synapse at the PSR by antagonizing the function of SCF, and SYG-1 promotes synapse elimination by making more SCF available distantly at the SSR. It is also possible that SYG-1 acts through unknown mechanisms to regulate SCF activity at the SSR, which would not require SCF activity to be limiting.

Synaptic circuit assembly is a highly dynamic process of concurrent formation and elimination of synapses during development (21). The UPS is widely used in many cellular processes, including axon pruning, dendrite pruning, and synapse development (15, 2226). Our findings reveal that the UPS also mediates local elimination of synapses without obvious neurite loss, a process that requires precise control of UPS activity. One way to explain the spatial regulation of UPS activity is through local inhibition of the SCF complex by synaptic specificity molecules such as SYG-1. This ensures that synapses are stabilized at appropriate sites but removed from inappropriate sites. A recent study showed that the proteasome could function locally within distal dendrites of vertebrate neurons (27). Our results provide a molecular link between spatial regulation of ubiquitin-mediated protein degradation and selective synapse elimination; similar local protection mechanisms may also underlie vertebrate neural circuit formation.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


References and Notes

View Abstract

Navigate This Article