Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Neuron. Author manuscript; available in PMC 2010 December 10.
Published in final edited form as:
PMCID: PMC2796257

Rab3 Dynamically Controls Protein Composition at Active Zones


Synaptic transmission requires the localization of presynaptic release machinery to active zones. Mechanisms regulating the abundance of such synaptic proteins at individual release sites are likely determinants of site-specific synaptic efficacy. We now identify a novel role for the small GTPase Rab3 in regulating the distribution of presynaptic components to active zones. At Drosophila rab3 mutant NMJs, the presynaptic protein Bruchpilot, calcium channels, and electron-dense T-bars are concentrated at a fraction of available active zones, leaving the majority of sites devoid of these key presynaptic release components. Late addition of Rab3 to mutant NMJs rapidly reverses this phenotype by recruiting Brp to sites previously lacking the protein, demonstrating that Rab3 can dynamically control the composition of the presynaptic release machinery. While previous studies of Rab3 have focused on its role in the synaptic vesicle cycle, these findings demonstrate an additional and unexpected function for Rab3 in the localization of presynaptic proteins to active zones.

Keywords: Rab3, Drosophila, active zone, synaptic plasticity, presynaptic release machinery, neuromuscular junction


Individual neurons can form thousands of discrete synaptic connections with their postsynaptic partners. Each synapse comprises tightly apposed pre- and postsynaptic membranes, a postsynaptic cluster of neurotransmitter receptors, and a presynaptic complex of proteins that promotes neurotransmitter release. For a synapse to function, the proper complement of proteins must localize to the presynaptic release machinery, and the protein composition at the release site is a likely determinant of its synaptic efficacy (Fejtova and Gundelfinger, 2006). While the general properties of synapses formed by a single axon are similar, the release probability of such synapses can vary dramatically (Pelkey and McBain, 2007). This presynaptic heterogeneity is likely due to mechanisms that control synapse specific plasticity and may represent one aspect of the molecular basis of learning and memory. Thus, identifying mechanisms that control the protein composition and presynaptic release properties of individual synapses will provide insights into plasticity mechanisms in the brain.

The Drosophila neuromuscular junction (NMJ) is an excellent system for identifying mechanisms that regulate the protein composition of individual active zones. A single Drosophila motoneuron and single muscle cell form an NMJ comprising hundreds of individual release sites (Atwood et al., 1993), or presynaptic active zones, each apposed to a postsynaptic glutamate receptor (GluR) cluster (Petersen et al., 1997). Each release site is akin to a single mammalian central nervous system synapse, and like CNS synapses, there is heterogeneity in their release properties (Marrus et al., 2004). Drosophila contains orthologs of all of the major vertebrate presynaptic proteins with the exception of Bassoon and Piccolo (Owald and Sigrist, 2009). Among these, Bruchpilot (Brp), the Drosophila ortholog of CAST, plays an essential role in organizing the presynaptic release machinery (Kittel et al., 2006; Wagh et al., 2006). This role is similar in mammals where CAST acts as a molecular scaffold within the cytomatrix at the active zone, interacting with Piccolo, Bassoon, Rim1α, and α-liprins/SYD-2 (Ohtsuka et al., 2002; Takao-Rikitsu et al., 2004) and in C. elegans where the Brp homologue ELKS-1 acts with SYD-2/α-liprin to promote the assembly of presynaptic active zone components (Dai et al., 2006; Patel and Shen, 2009). In Drosophila, Bruchpilot localizes to every active zone, but its distribution is heterogeneous, and the abundance of Brp at an active zone appears to correlate with the release probability of that site (Marrus et al., 2004; Schmid et al., 2008). Brp is not required for active zone formation per se, but is an integral component of T-bars (Fouquet et al., 2009), electron-dense active zone specializations that likely promote transmitter release, and is required for the continuous accumulation of Ca2+-channels at active zones during synapse maturation (Fouquet et al., 2009; Kittel et al., 2006). These findings with Brp imply that mechanisms exist to 1) ensure that Brp is present at each release site and 2) regulate the level of Brp at each site. Such mechanisms would likely impact site-specific release probability by controlling the protein composition of the release machinery at each site. To identify such mechanisms, we performed a large-scale genetic screen to identify genes required for the proper localization of Brp to active zones.

We find that the small GTPase Rab3 functions to influence the distribution of Brp and other crucial presynaptic active zone components to release sites. In the absence of Rab3, key constituents of the presynaptic release machinery are concentrated at a fraction of available sites, resulting in the formation of a small number of super sites with enhanced release probability and a larger number of sites devoid of key presynaptic release proteins. Rab3 can rapidly recruit Brp to active zones, demonstrating that the protein composition of the release machinery is under dynamic control and that Rab3 is well-positioned to participate in synapse-specific plasticity mechanisms. Whereas previous studies have implicated Rab3 in the cycling and docking of synaptic vesicles (Sudhof, 2004), here we report a novel role for Rab3 in influencing the protein composition of the presynaptic release apparatus at individual active zones.


rup mutants display defects in active zone development

To identify mechanisms that control the molecular composition of individual release sites, we have screened through a collection of Drosophila mutants for those with defects that differentially affect presynaptic active zones within an NMJ. We performed an anatomical genetic screen on a collection of ~1500 lines that carry unique insertions of transposable elements in or near genes on the second chromosome. We dissected 3rd-instar homozygous mutant larvae from each line and stained for the presynaptic active zone protein Bruchpilot (Brp) and the essential glutamate receptor subunit DGluRIII (Marrus et al., 2004). We imaged the immunostained NMJs with fluorescence microscopy and identified mutants with altered active zones, including changes in Brp puncta size, number, or intensity, as well as those with defects in the apposition of presynaptic Brp and postsynaptic DGluRIII puncta. Within this group, we identified one line, P{SUPor-P}KG07292, that has a novel active zone phenotype (Figure 1A). In this mutant there is a dramatic loss of Brp-positive active zones, yet the morphology and number of DGluRIII clusters appears grossly normal. As such, most GluR clusters are unapposed to a Brp-positive active zone. The remaining Brp puncta are apposed to GluR clusters, and these Brp puncta are significantly larger than in wild type. Due to the large number of unapposed GluR clusters, we named this mutant running-unapposed (rup).

Figure 1
rup mutant NMJs contain fewer and larger Bruchpilot puncta

While Brp morphology is altered in rup, the gross morphology of the mutant NMJ is normal and the synaptic terminal area is not significantly different than in wild type (wt, 274 ± 8 μm2; rup, 278 ± 11 μm2; p>0.7; n = 20). Staining with an antibody against the vesicular glutamate transporter (DVGLUT) demonstrates that synaptic vesicles are distributed throughout the NMJ, and co-staining for the postsynaptic scaffolding protein Discs-large (Dlg) reveals that the presynaptic terminal is apposed to the postsynaptic specialization across its length (Figure S1). Hence, the presence of unapposed GluR clusters is not due to synaptic retraction (Eaton et al., 2002) of synaptic boutons or branches; instead, affected synapses distribute in a salt-and-pepper pattern throughout the synaptic terminal, suggesting that the defect occurs at the level of individual synapses. Glutamate receptors co-localize with the serine-threonine kinase Pak at the Drosophila NMJ (Albin and Davis, 2004; Rasse et al., 2005). Pak distribution appears normal in the rup mutant, and like GluR clusters, most Pak clusters are unapposed to Brp-positive active zones (Figure 1B). Hence, in rup postsynaptic morphology is relatively normal and the primary morphological defect is likely presynaptic. Despite their abnormal active zones, rup mutant animals are viable and fertile.

rup is a mutant allele of rab3

To investigate the mechanism underlying the defective active zones in rup, we wished to identify the responsible gene. Although rup was found in a collection of insertional mutants, the phenotype does not map to the P{SUPor-P}KG07292 transposable element. Instead, rup is a second-site mutation fortuitously present on the chromosome. We roughly mapped rup by meiotic recombination to position 43–48 on the right arm of the second chromosome, and identified a deficiency chromosome (Df(2R)ED2076) that fails to complement the mutation. This deficiency deletes 26 predicted genes in the region between 47A10-47C1. The list of candidates was narrowed to 21 by complementation testing with known null mutants in the region. We then sequenced the coding regions of candidate genes, ultimately identifying a five base pair deletion near the 3-prime end of the rab3 gene. This deletion throws rab3 out of frame and would lead to a deletion of the last 35 amino acids of the protein, including the final CXC motif that in other systems is required for lipid modification, the binding of Rab3 to synaptic vesicles, and proper Rab3 localization (Iwasaki et al., 1997; Johnston et al., 1991).

A single ortholog of Drosophila rab3 was previously cloned and demonstrated to be highly conserved (Johnston et al., 1991). We further showed that it is expressed throughout the fly nervous system (DiAntonio et al., 1993); however, no functional studies were performed. To investigate the localization of Rab3 protein and the nature of the mutant allele, we generated a polyclonal antibody to a peptide epitope in the unique C-terminal region of Drosophila Rab3. This antibody stains synaptic terminals of wild type NMJs in a pattern similar to synaptic vesicle markers such as synapsin and DVGLUT (Figures 2A and S2). However, unlike synapsin and DVGLUT, Rab3 staining is further concentrated in a punctate pattern at active zones, as visualized by co-staining with Brp (Figure 2A, inset). This punctate localization of Rab3 at active zones is not observed in brp mutant NMJs (brp69/Df(2R)BSC29) even though the synaptic vesicle-like distribution of Rab3 staining remains (Figure S2C). In addition, the antibody recognizes a single band of the predicted size on immunoblots from wild type larvae (Figure 2B). Both the synaptic staining at the NMJ and the band on the immunoblot are absent in the rup mutant (Figures 2A and 2B), demonstrating that the antibody is specific for Rab3 and that the rup mutant does not express wild type Rab3 protein. Since the mutation in rup is located in the C-terminal region of rab3 just upstream of the epitope, it is possible that a truncated protein could be expressed. While such a mutant protein could have residual function, the active zone phenotype of rup homozygotes and transheterozygotes of rup and Df(2R)ED2076 are similar in terms of the percentage of GluR clusters apposed to Brp (rup, 35 ± 2%; rup/Df(2R)ED2076, 33 ± 1%, p>0.3; n=10) and the average area of individual Brp punctum (rup, 0.60 ± 0.014 μm2; rup/Df(2R)ED2076, 0.69 ± 0.02 μm2, p<0.01; n=10). Therefore, rup behaves as a genetic null or a very strong hypomorph.

Figure 2
The rup active zone phenotype is due to loss of rab3 function

To test whether the active zone phenotype in the rup mutant is due to the absence of wild type rab3, we used the Gal4/UAS system to express transgenic rab3 in the rup background. Expression of UAS-rab3 under the control of a neuronally expressed Gal4 results in the localization of Rab3 to the NMJ. Neuronal expression of the rab3 transgene in the rup mutant background fully rescues the rup active zone phenotype (Figure 2C). In the rup mutant, more than 60% of GluR clusters are unapposed to Brp-positive active zones (Figure 2D, p[double less-than sign]0.001) due to a more than 60% decrease in the number of apparent presynaptic active zones. Furthermore, those Brp puncta that are present are large, with a more than two-fold increase in area compared to wild type (Figure 2E, p[double less-than sign]0.001). Neuronal expression of transgenic rab3 rescues both the total number of Brp-positive active zones (wt, 298 ± 18; rup, 81 ± 5, p[double less-than sign]0.001; rescue, 270 ± 17, p>0.25; n=10) per NMJ and the percentage of GluR clusters apposed to Brp to wild type levels (Figure 2D). Transgenic expression of rab3 also rescues the Brp area phenotype, reducing the average size of Brp puncta to that observed at wild type NMJs (Figure 2E). Hence, the rup phenotype is due to the loss of rab3 function and Rab3 is required in the neuron for normal localization of Brp to active zones. Transgenic expression of a mutated form of rab3 (N134I) that disrupts the guanine nucleotide binding domain (Weber et al., 1996) leads to little or no rescue of the rup phenotype (data not shown), suggesting that the GTPase activity of Rab3 is important for this function.

The rab3rup phenotype is not a generic response to changes in activity

We identified the Drosophila rab3 mutant, which we now name rab3rup, on the basis of abnormal localization of the active zone protein Bruchpilot. Previous genetic analysis of rab3 mutants from C. elegans and mice identified defects in transmitter release rather than structural abnormalities at the active zone (Mahoney et al., 2006; Nonet et al., 1997; Schluter et al., 2006; Schluter et al., 2004). Might the active zone defects in the rab3rup mutant be secondary to defects in transmitter release? We find this unlikely because similar changes in Brp localization have not been reported in any other Drosophila mutant, including those that alter neuronal activity. However, to test directly whether alterations in neuronal activity can phenocopy the rab3rup mutant, we stained for Brp and DGluRIII in a mutant with decreased sodium channel expression and neuronal activity (Wu et al., 1978), a mutant with decreased potassium channel function and increased neuronal activity (Mosca et al., 2005), and in an unc-18 (rop) mutant with decreased transmitter release (Wu et al., 1998). In each case, apposition of Brp and DGluRIII as well as the size of the Brp puncta are indistinguishable from wild type (Figures S3A and S3B). As a further test of the role of activity, we assessed genetic interactions between the rab3rup mutant and the activity mutants. If the rab3rup phenotype were due to a decrease in transmitter release, then the sodium channel and potassium channel mutants would be predicted to enhance or suppress, respectively, the rab3rup phenotype. We generated such double mutants and find that the rab3rup phenotype is similar in all cases (Figures S3C and S3D). Hence, we have no evidence that the abnormal localization of Brp is secondary to functional defects. While Drosophila Rab3 may function in vesicle release, our results suggest that Rab3 may also have a previously undescribed function regulating active zone development.

Rab3 controls the distribution of active zone components among release sites

At a wild type NMJ, Brp is a marker of all active zones; however, in the mutant the large reduction in the number of Brp puncta could either reflect 1) a large decrease in the number of active zones or 2) a change in the distribution of Brp such that it only localizes to a small subset of active zones. To distinguish between these possibilities, we performed an ultrastructural analysis of synapses from wild type and rab3rup NMJs. At the Drosophila NMJ, the active zone is visualized as an electron dense thickening of the membranes where the pre- and post- synaptic membranes are in tight apposition (Atwood et al., 1993). Glutamate receptors cluster opposite these sites (Petersen et al., 1997). In addition, an electron dense T-bar, which likely promotes transmitter release, is often associated with the active zone (Atwood et al., 1993). If there were a large decrease in active zone number in rab3rup, then we would predict a large drop in active zone density. Brp is a component of T-bars (Fouquet et al., 2009), so if Brp were concentrated at only a subset of active zones in rab3rup, then we would predict a change in the distribution of T-bars across active zones.

We examined the ultrastructure of synaptic boutons from muscles 6/7 of third-instar mutant and wild type larvae. We quantified the length of electron dense active zones, the number of active zones per micron of membrane length, the percentage of active zones with a T-bar, and assessed the distribution of T-bars across active zones. Active zone length in the mutant is no different from wild type (wt, 0.69 ± 0.01 μm, n = 314; rab3rup/Df(2R)ED2076, 0.66 ± 0.01 μm, n = 223; p>0.1). In the rab3 mutant, active zone number per micron is mildly reduced (wt, 0.47 ± 0.02, n = 78; rab3rup/Df(2R)ED2076, 0.35 ± 0.02, n = 70; p[double less-than sign]0.001). At the light level, we observe a similar decrease in the total number of DGluRIII clusters per NMJ (wt, 307 ± 18; rab3rup/Df(2R)ED2076, 244 ± 12, p<0.01; n=10) resulting in a moderate reduction of DGluRIII cluster number per μm2 (wt, 1.11 ± 0.07; rab3/Df(2R)ED2076, 0.94 ± 0.03; p<0.05; n = 10). This suggests that in the mutant the GluR clusters unapposed to a Brp-positive puncta are likely apposed to an EM-defined active zone. In addition, the ratio of T-bars to active zones is no different from wild type when averaged over the entire population of active zones (wt, 0.44 ± 0.03 T-bars/AZ, n=314; rab3rup/Df(2R)ED2076, 0.44 ± 0.06 T-bars/AZ, n=223; p>0.9). However, the distribution of T-bars at active zones is dramatically altered (Figures 3A, 3C, and 3D). At wild type NMJs, zero or one T-bar is observed at most active zones, with only a rare sighting of a two T-bar active zone. In contrast, in the rab3rup mutant there is a 30% decrease in the proportion of active zone sections with a T-bar (Figure 3C; p<0.005), and a more than 3-fold increase in the proportion of active zone sections with T-bars that have multiple T-bars (Figure 3D; p[double less-than sign]0.001). The largest number of T-bars we have observed in wild type at a single active zone is two, but in the rab3rup mutant we have observed active zones with three, four, and five T-bars. These results demonstrate that while the number of putative release sites is only mildly reduced in the rab3rup mutant, there is a concentration of the release machinery at a small subset of those sites. Since Brp is a component of T-bars, we suggest that sites observed by light microscopy to have abnormally large concentrations of Brp in the rab3rup mutant are likely those active zones with multiple T-bars. Hence, this ultrastructural analysis suggests that rab3 is required for the distribution of Brp and its associated proteins across the array of active zones at an NMJ.

Figure 3
T-bar and Cacophony-GFP distributions are altered in the rab3rup mutant

Are other components required for synaptic vesicle release also concentrated at these sites? At wild type synapses, Brp co-localizes with the calcium channel subunit Cacophony and is required for its continued accumulation at active zones (Fouquet et al., 2009; Kittel et al., 2006). In the rab3rup mutant, Cacophony-GFP co-localizes with Brp and is clustered opposite only a small subset of GluR puncta (Figure 3B). We quantified the average intensity of Cacophony-GFP opposite wild type and rab3rup mutant DGluRIII clusters (Figure 3E). Brp positive sites in the rab3rup mutant contain more Cacophony-GFP than wild-type active zones; whereas Brp negative sites contain less Cacophony-GFP than wild type active zones. Together, these results reveal that rab3 influences the distribution of multiple components of the presynaptic release machinery, and in its absence these components become concentrated at a small number of super sites while leaving the majority of sites deficient in proteins required for efficient evoked release.

In addition to active zones, we also analyzed synaptic vesicle distribution in the mutant. In C. elegans rab3 mutants there is a defect in vesicle clustering near release sites (Nonet et al., 1997). At the fly NMJ, we find no difference in the distribution of vesicles between the rab3rup mutant and wild type active zones. The average number of vesicles within 300 nm of the active zone membrane is similar between wild type and the rab3 mutant (wt, 39.4 ± 1.0, n = 32; rab3rup/Df(2R)ED2076, 40.8 ± 1.1, n = 41; p>0.3). Within the first 100 nm from the active zone membrane there is also no significant difference in vesicle number (wt, 9.4 ± 0.4, n = 32; rab3rup/Df(2R)ED2076, 10.4 ± 0.4, n = 41; p>0.1). In addition, we find no difference in vesicle clustering within 300 nm from the membrane at rab3rup/Df(2R)ED2076 mutant active zones that have zero, one, or multiple T-bars (zero, 39.3 ± 1.7, n = 13; one, 40.3 ± 2.30, n = 12; multiple, 42.3 ± 1.8, n = 16). Wild type active zones also show no difference in vesicle number within 300 nm from the membrane regardless of whether there exists zero, one, or multiple T-bars at the active zone (not shown). These results demonstrate that Drosophila Rab3 is not required for vesicle clustering near active zones.

GluRs preferentially localize opposite Brp-positive active zones

Approximately one third of active zones in the rab3rup mutant have increased levels of Brp, and these sites are enriched for calcium channels and, likely, T-bars. Since these active zone components promote synaptic vesicle release, we propose that the Brp-positive active zones have a higher release probability than Brp-negative sites. While it is not possible to measure release probability of individual active zones at the Drosophila NMJ, the localization of glutamate receptors opposite each active zone correlates with release probability. When glutamate receptor levels are reduced, receptors preferentially cluster opposite active zones with high release probability (Marrus et al., 2004). To reduce the levels of glutamate receptors, we expressed a transgenic RNAi construct in muscle that targets the essential DGluRIII subunit. When DGluRIII is limiting in the rab3rup mutant, unapposed GluR clusters are on average 50% smaller than clusters apposed to Brp (Figure 4, p[double less-than sign]0.001). Indeed, even with normal DGluRIII levels, GluR clusters opposite Brp-positive active zones are larger than those opposite Brp-negative active zones (Brp-positive, 1.43 ± 0.06 μm2; Brp-negative, 0.97 ± 0.04 μm2; p[double less-than sign]0.001; n = 10). These data support the hypothesis that Brp-positive active zones have a higher release probability than Brp-negative active zones, and indicate that Rab3 regulates not only the protein composition but also the function of individual active zones within an NMJ.

Figure 4
GluRs preferentially clusters opposite Brp-positive active zones in the rab3rup mutant when DGluRIII is limiting

Impaired short-term facilitation in the rab3rup mutant

The morphological analysis above suggests that in the rab3rup mutant, a subset of the hundreds of active zones comprising a single NMJ have an elevated release probability. What is the effect of these synapses on synaptic strength at the NMJ? To assess the electrophysiological function of the NMJ, we performed voltage clamp recordings from muscle 6 of segments A3 and A4 at the larval NMJs of WT and rab3rup mutants. A single stimulus in low extracellular calcium evokes a very similar excitatory junctional current (EJC) at wild type and mutant NMJs (EJC amplitude: wt, 10.0 ± 1.0 nA, n=12; rab3rup, 10.8 ± 0.9 nA, n=13; p>0.2). There is also no significant change in the amplitude of spontaneous miniature excitatory currents (mEJC amplitude: wt, 0.85 ± 0.03 nA, n=12; rab3rup, 0.87 ± 0.02 nA, n=13; p>0.3). Therefore, estimates of quantal content, the number of synaptic vesicles released following an action potential, are similar for wild type and rab3rup mutant NMJs (quantal content (EJC/mEJC): wt, 12 ± 1 nA, n=12; rab3rup, 13 ± 1 nA, n=13; p>0.3). Hence, synaptic strength in the rab3rup mutant is normal despite the apparent decrease in the number of sites (n) that contain important components of the presynaptic release machinery. Evoked release could be maintained by increased release from the remaining sites with increased levels of Brp if those sites have an increase in release probability (p). Do those mutant active zones that fire at low calcium exhibit a higher probability of release?

To investigate release probability, we measured short-term facilitation, which varies inversely with p. When a nerve is repeatedly stimulated with a short inter-pulse interval, the subsequent evoked events are often larger because of residual calcium in the presynaptic terminal (Zucker and Regehr, 2002). When probability of release is low, such short-term facilitation is more pronounced. After a short train of stimuli, we measured the amplitude of each EJC and calculated the facilitation index (FI) as the ratio of the fifth to the first EJC. At a stimulation frequency of 10 Hz in 0.40 mM Ca2+, both the wild type and rab3rup NMJ facilitate, but facilitation is significantly decreased in the rab3rup mutant (Figures 5A, 5B, and 5D; p<0.01). Facilitation is rescued to wild type levels when a rab3 transgene is expressed in motor neurons (Figures 5C and 5D; p<0.01). Indeed, when the stimulation frequency is increased to 20 Hz, facilitation is approximately three-fold lower in the rab3rup mutant as compared to wild type (wt, FI=3.3 ± 0.4, n=6; rab3rup, FI=1.4 ± 0.1, n=9; p<0.001). These findings suggest that those release sites that fire in low calcium have a higher p in the rab3rup mutant than in wild type. Since evoked release is not elevated in the mutant, this implies that fewer sites are firing. These electrophysiological findings are consistent with the morphological data, and taken together lead to the model that in the rab3rup mutant approximately one third of release sites have a very high p due to an excess of essential active zone components such as Brp, calcium channels, and T-bars, while the remaining sites lack crucial active zone proteins and so have a low p. This interpretation of the data assumes that the number of release sites, n, is equivalent to the number of active zones. An alternative interpretation is that the number of release sites represents a smaller unit of function, such as a T-bar, in which case the mutant would have a redistribution of rather than decrease in n. If this were the case then p at such sites would be unchanged, and the defect in short-term facilitation would likely reflect a more direct role of Rab3 in the synaptic vesicle cycle.

Figure 5
Short-term facilitation is impaired at rab3rup mutant NMJs

Rab3 rapidly and reversibly regulates the protein composition of active zones

Our findings demonstrate that Rab3 influences the distribution of presynaptic active zone components among potential release sites at Drosophila NMJs. We have shown that the mutant active zone phenotype can be rescued by neuronal expression of a wild type rab3 transgene expressed throughout development, but when is Rab3 required during development for the normal formation of active zones? Understanding the temporal requirement for Rab3 function has implications for the mechanism by which Rab3 functions to distribute active zone components. To address these issues, the rab3 transgene can be expressed late in development, after many mutant active zones have formed but before the formation of the latest developing active zones. There are several possible outcomes. First, there may be no rescue, suggesting that Rab3 is required early in development to establish the fate of active zones that subsequently form. Second, there may be partial rescue, implying that active zones formed after transgenic rab3 expression develop normally, but those that previously formed are not rescued. Since most new boutons are added at the ends of NMJs, in this scenario we would predict that terminal boutons would have fewer unapposed GluR clusters than internal boutons. Finally, there may be complete rescue, with late expression of rab3 able to reverse the effects of early rab3 loss. This would demonstrate that Rab3 can dynamically control the protein composition of active zones.

To determine if late expression of UAS-rab3 rescues the mutant phenotype, we used the GeneSwitch Gal4 system, in which gene expression is induced by addition of RU486 (Osterwalder et al., 2001). We allowed rab3rup mutant larvae that contain both the UAS-rab3 and ELAVGeneSwitch-Gal4 transgenes to develop for three days without exposure to RU486. At this point, larvae were in the early third-instar stage and exhibited the rab3rup mutant active zone phenotype of enlarged Brp puncta that are apposed to only a fraction of DGluRIII clusters (Figure 6A). We then exposed these larvae to food containing RU486 and allowed them to develop for the next 6, 9, 24, or 48 hours, respectively, and then dissected larvae at these time points and stained for Brp and DGluRIII. Within 6 hours of RU486 exposure, sufficient UAS-rab3 transcription, translation, and transport occur such that a low level of Rab3 can be visualized filling some but not all NMJs, indicating that Rab3 protein is just beginning to reach the NMJs by 6 hours. Surprisingly, even at such an early stage of Rab3 localization to the NMJ, several small, likely newly formed Brp puncta can be visualized among enlarged mutant Brp puncta, slightly increasing the percentage of DGluRIII clusters apposed to Brp. Three hours later, Rab3 protein is detected at all NMJs and many more small Brp puncta are observed among the larger mutant puncta such that approximately two-thirds of DGluRIII clusters are apposed to Brp as compared to one-third in the mutant (Figure 6B). The average Brp puncta area at 9 hours after RU486 exposure approaches that of wild-type (Figure 6C). Cumulative probability plots showing the distribution of Brp puncta area at 9 hours indicate that the decreased average area is due to both the addition of small Brp puncta and a reduction in the size of the original enlarged population of mutant Brp puncta (Figure 6D). By twenty-four hours, the apposition phenotype is fully rescued and most Brp puncta are the size of normal wild type active zones, although a few enlarged Brp puncta are still present. Following forty-eight hours of transgenic rab3 expression, Brp puncta are of normal size and the mutant active zone phenotype is fully rescued. These results indicate that active zones are remarkably plastic and that their protein composition can be rapidly and reversibly modified by rab3.

Figure 6
Rab3 rapidly and reversibly regulates Brp distribution

Brp levels modify the rab3rup phenotype

In the analyses above, we have utilized Brp as a marker of the presynaptic active zone, but studies of the brp mutant indicate that Brp is actually a central organizer of the release apparatus (Kittel et al., 2006). As such, might the rab3rup phenotype be due to changes in Brp itself? We first investigated whether the active zone phenotype could be due to decreased Brp expression levels in the rab3rup mutant. Comparing total Brp intensity at the NMJs of mutant and rescued mutant larvae shows no difference in Brp levels (rab3rup/Df(2R)ED2076, 467 ± 18 a.u.; rab3rup/Df(2R)ED2076 with rab3 rescue, 493 ± 30 a.u.; p>0.40, n = 10), so the decrease in the number of Brp puncta cannot be explained by lower levels of Brp at the NMJ. Furthermore, directly decreasing Brp levels in a wild type background with a Brp RNAi construct (Wagh et al., 2006) produces a phenotype distinct from the rab3rup phenotype (Figure 7A). Knockdown of Brp results in the formation of significantly fewer Brp puncta such that approximately 50% of DGluRIII clusters are no longer apposed to Brp puncta (Figure 7C). However, unlike in the rab3rup mutant, the clusters of Brp that form are much smaller than in wild type (Figure 7D). A similar result is obtained when Brp levels are reduced in the rab3rup mutant; there is a large decrease in the number of Brp puncta formed, leaving only 14% of DGluRIII clusters apposed to Brp (Figures 7B and 7C). These mutant Brp puncta are smaller than the puncta in rab3rup mutants under normal Brp levels, however they are still larger than WT puncta (Figure 7D). Hence the decrease in the number of Brp puncta is not an artifact due to small puncta being lost in the noise. Instead, these findings demonstrate that the levels of Brp can directly affect the likelihood of forming a Brp puncta at an active zone.

Figure 7
Altering Brp expression levels modifies the rab3rup phenotype

If Brp levels influence the probability of Brp puncta formation, can increased Brp expression rescue the rab3rup phenotype? There are two possible outcomes. First, there may be no rescue, indicating that without Rab3, unapposed sites are fundamentally disrupted such that they cannot cluster Brp. Alternatively, increased Brp expression may increase the percentage of DGluRIII clusters apposed to Brp, demonstrating that the unapposed sites in the rab3rup mutant are less likely to cluster Brp but are not prohibited from doing so. To determine how increased Brp expression affects the rab3rup mutant phenotype, we drove a UAS-brp transgene (Wagh et al., 2006) in rab3rup mutant neurons. Surprisingly, we find strong rescue of the apposition phenotype, increasing the percentage of DGluRIII clusters apposed to Brp from 30% to 68% (Figures 7B and 7E). Furthermore, enhanced Brp expression in rab3rup mutant neurons also increases the percentage of DGluRIII clusters apposed to Cacophony-GFP from 31% to 52% (Figure S4). These results indicate that Brp-negative sites in the rab3rup mutant have the ability to cluster Brp and calcium channels, but that they require the stronger driving force provided by higher Brp levels. Interestingly, the average size of a Brp puncta in these larvae is decreased even though Brp levels are increased (Figure 7F), suggesting that that the size of the Brp puncta may be secondary to the number of available puncta. With an increased number of Brp puncta, there may be more competition for unbound Brp in the cytosol, reducing the ability of each puncta to grow. Conversely, these results suggest that the increase in Brp puncta size in the rab3rup mutant may be secondary to the decrease in the number of puncta. Taken together, these data suggest that the essential function of Rab3 is to increase the probability that a Brp punctum will form at an active zone.


We show here that the small GTPase Rab3 controls the protein composition and release probability of individual active zones at the Drosophila neuromuscular junction. In a rab3 mutant, key constituents of the presynaptic release machinery are enriched at a subset of active zones while the remaining release sites are apparently devoid of these proteins. Expression of Rab3 rapidly and reversibly rescues this altered protein distribution. Physiological studies are consistent with these morphological findings, demonstrating an increase in release probability from an apparently decreased number of release sites. Mechanistic studies indicate that Rab3 functions to increase the probability that the essential synaptic organizing molecule Bruchpilot will cluster at an active zone. This Rab3-dependent regulation of active zone protein composition and release probability provides a potential mechanism for the synapse-specific control of synaptic efficacy.

Rab3 regulates the distribution of the presynaptic release apparatus

The Drosophila NMJ consists of a motoneuron axon terminal arranged as a chain of synaptic boutons closely associated with the postsynaptic muscle membrane. Within each string of boutons are hundreds of individual synapses (Atwood et al., 1993), discrete sites of neurotransmitter release where a presynaptic active zone is directly apposed to a postsynaptic glutamate receptor cluster (Petersen et al., 1997). Such a synapse comprises: 1) the site where the axon and muscle membranes are in closest proximity, likely tethered by trans-synaptic cell adhesion molecules; 2) the presynaptic release apparatus that influences the Ca2+-mediated release of the neurotransmitter-filled vesicles; and 3) the neurotransmitter receptors and scaffolding and signaling proteins of the postsynaptic density. Here we demonstrate that disrupting Rab3 alters the distribution of proteins that make up the presynaptic release machinery without grossly disturbing the other two components of the synapse.

In the absence of rab3, a subset of synapses contain increased amounts of the active zone protein Bruchpilot, higher levels of the calcium channel Cacophony, and more electron dense T-bars at the active zone. Since Brp is a component of T-bars and influences Ca2+-channel accumulation (Fouquet et al., 2009; Kittel et al., 2006), the altered distribution of these components is likely a direct consequence of changes in Brp distribution. The creation of additional active zone markers will be necessary to determine the full extent of this altered distribution. However, since all three components examined influence the probability of evoked vesicle release, the active zones where they accumulate likely are sites of enhanced vesicle release. Conversely, the remaining sites that are devoid of these components likely exhibit impaired evoked release. Two lines of evidence support this conclusion. First, glutamate receptors preferentially cluster opposite sites with the highest release probability (Marrus et al., 2004). In the rab3rup mutant, GluR clusters are larger at Brp-positive than Brp-negative sites, suggesting that those active zones containing Brp have a higher release probability. Second, facilitation resulting from short stimulus trains is reduced in the mutant, consistent with an increased release probability (p). However, since quantal content and quantal size are unchanged, the increase in p must be balanced by a decrease in the number of sites that are firing. Hence, both the morphological and electrophysiological data are consistent with the model that Rab3 controls the distribution of active zone proteins to influence the efficacy of individual release sites.

Other Drosophila mutants have active zone phenotypes, but to our knowledge none have the combination of phenotypes described here. Mutations in synaptojanin (Dickman et al., 2006), liprin (Kaufmann et al., 2002), neurexin (Li et al., 2007), and spectrin (Pielage et al., 2006) affect the size and spacing of the entire array of active zones. Mutations in the Unc-51 kinase and the protein phosphatase PP2A have differential affects on active zones, resulting in a subset of glutamate receptors unapposed to Bruchpilot puncta as in the rab3rup mutant (Viquez et al., 2009; Wairkar et al., 2009). However, in the unc-51 and PP2A mutants the remaining Brp puncta are not enlarged and there is no increase in the proportion of active zones with multiple T-bars. Such phenotypes are consistent with defects in active zone formation, rather than in the distribution of proteins across active zones. Finally, GluR clusters unapposed to Brp puncta occurs following synaptic retraction, but in such mutants the active zone defects are secondary to the loss of the entire presynaptic terminal (Eaton et al., 2002). Hence, Rab3 participates in a previously undescribed mechanism that differentially regulates active zones within an NMJ.

Rab3 increases the probability that Bruchpilot will cluster at an active zone

Our findings demonstrate that Rab3 plays a central role in the localization of Bruchpilot to individual active zones. In the absence of Rab3, approximately 70% of active zones are devoid of Brp while the other 30% contain an excess of Brp. What is the function of Rab3 such that its loss leads to this altered Brp distribution? We suggest that Brp is present in two pools: one fraction bound in complexes at active zones and a second mobile fraction in the cytosol. We further suggest that Brp is dynamic and may alternate between these two pools by associating with or dissociating from the active zone complex. As such, unbound Brp in the cytosol may either nucleate a cluster at an active zone, creating a new Brp punctum, or add to a pre-existing Brp punctum making it larger. Given this scenario, the rab3 phenotype may be explained by two alternative models of Rab3 function: 1) Rab3 limits Brp puncta size, or 2) Rab3 increases the ability of Brp to nucleate new Brp clusters at active zones. If Rab3 functions to limit the addition of Brp to already existing sites, disruption of Rab3 would allow Brp clusters to grow to a maximal size, reducing the availability of cytosolic Brp to create new puncta and consequently constraining the number of puncta formed. In such a model, we would predict that Brp puncta size would be large at rab3rup mutant NMJs regardless of Brp expression levels. Instead, decreasing Brp levels in the rab3rup mutant decreases the size of Brp puncta. Even more telling, increasing Brp levels in the rab3rup mutant also reduces the size of Brp puncta. These results are inconsistent with the model that primary function of Rab3 is to limit the size of Brp puncta.

Instead, we suggest that Rab3 functions to increase the probability that Brp will nucleate a new cluster at an active zone. The presence of Brp at some active zones demonstrates that Rab3 is not absolutely required for Brp localization. Why then is Rab3 required for Brp localization to the 70% of active zones that are bereft of Brp? Rather than posit that these two classes of active zones are fundamentally different in the rab3rup mutant, we suggest that in the absence of Rab3, Brp is much less likely to nucleate a cluster at an active zone (Figure 8). In this view, whether or not an active zone contains a Brp puncta is determined stochastically. This reduction in Brp puncta number would leave a larger pool of cytosolic Brp and fewer puncta with which Brp can associate, secondarily leading to an increase in puncta size due to excess unbound Brp. We estimate that Rab3 increases the odds of Brp clustering at an active zone by about three-fold, as the absence of Rab3 leads to approximately 70% fewer Brp puncta with wild type levels of Brp (30% apposition vs. 100% apposition) as well as reduced levels of Brp (13% apposition vs. 50% apposition). This model highlights the importance of both Rab3 and the levels of Brp in controlling the number and size of Brp clusters at active zones.

Figure 8
A model for the role of Rab3 in influencing Brp distribution at active zones. Unbound Brp (red triangle) may either nucleate a new Brp cluster at an active zone (black rectangle) or add to a pre-existing Brp cluster making it larger. In the rab3rup mutant, ...

The data presented here are consistent with this model. First, late rescue with rab3 leads to the rapid addition of new, small Brp puncta and, on a slower time scale, a decrease in the size of the large Brp puncta. This demonstrates that Brp is dynamic and can move into and out of active zones. Second, reducing the levels of Brp at wild type synapses leads to a decrease in both the number of Brp puncta formed as well as their size. An increase in Brp expression at a wild type synapse cannot increase the number of puncta since essentially 100% of active zones already contain a Brp puncta, but it does lead to an increase in the size of the puncta. Hence, Brp levels affect both the likelihood of forming a Brp puncta at an active zone as well as the ultimate size of the Brp puncta. Third, increased Brp expression enhances the ability of Brp to cluster at active zones, overcoming the absence of Rab3 and leading to the formation of more Brp puncta in the rab3rup mutant. This demonstrates that these mutant active zones do have the capacity to cluster Brp, but that it requires the stronger driving force provided by the additional Brp to overcome the absence of Rab3. Finally, when Brp is overexpressed in the rab3rup mutant the Brp puncta are smaller than when Brp is expressed at wild type levels. This apparent paradox suggests that Brp puncta compete for unbound Brp, and that the large increase in the number of Brp puncta provides more sites for unbound Brp and so ultimately results in smaller puncta. Hence, this model explains the variation in the number and size of Brp puncta present in the various genetic backgrounds tested above, and highlights a novel role for Rab3 in controlling the protein composition of active zones.

Effectors of Rab3 function

Brp appears to play a prominent role in the mechanism by which Rab3 regulates the distribution of active zone components to release sites. However, we have no evidence that Rab3 interacts directly with Brp, and such a direct interaction between Rab3 and members of the CAST/ERC family, of which Brp is an ortholog, has not previously been reported. Other proteins could mediate the interaction between Rab3 and Brp. In other species, Rab3 is known to interact with proteins involved in the Rab3 GTPase cycle such as Rab3-GEF (Wada et al., 1997), Rab3-GAP (Fukui et al., 1997), and GDI (Araki et al., 1990), as well as the putative Rab3 effectors Sec15 (Wu et al., 2005), Rabphilin (Shirataki et al., 1993), and Rim (Wang et al., 1997). Among the Rab3 effectors, Rabphilin is an unlikely candidate because Rabphilin knock-out mice and worms have no observable morphological or physiological synaptic defects (Schluter et al., 1999). Rim is a more plausible candidate because it is a constituent of the presynaptic release apparatus and binds to many other presynaptic active zone proteins including orthologs of Brp (Schoch et al., 2002). Alternatively, Rab3 may act on a yet unidentified target to regulate the molecular properties of Brp. Understanding Rab3 function at the fly NMJ will require the identification of the protein(s) Rab3 interacts with to distribute active zone components among release sites.

Comparison of Rab3 function in Drosophila, C. elegans, and mice

Previous studies of rab3 knockouts in other organisms suggest that Rab3 is involved in regulating vesicle cycling, docking, and exocytosis (Sudhof, 2004). While Rab3 may play a direct role in vesicle dynamics and release at the Drosophila NMJ, we suggest that it also plays a second, separate role in influencing the distribution of the presynaptic release apparatus. Defects at the active zone in the rab3rup mutant are unlikely to be secondary to altered synaptic vesicle release because 1) other mutants affecting release do not disrupt the composition of the active zone and 2) neither increasing nor decreasing activity in the rab3rup mutant exacerbates or ameliorates the active zone phenotype.

If Rab3 controls the protein composition of the active zone, then why have genetic analyses of rab3 in mice and C. elegans not identified structural abnormalities at the synapse? In Drosophila, loss of rab3 results in a very specific ultrastructural phenotype. The active zone, visualized as an electron dense thickening of tightly apposed pre- and postsynaptic membranes, is normal in Drosophila rab3rup mutants as it is in worms and mice. However, some synapses including Drosophila NMJ synapses contain prominent electron dense specializations such as T-bars that are thought to promote transmitter release. It is the distribution of these T-bars that is altered in the Drosophila rab3rup mutant, which would not be apparent at, for example, hippocampal synapses, where such dense bodies are not readily observed. While structural defects have not been detected in other organisms, the electrophysiological findings in mice show interesting parallels to the fly phenotype. The quadruple knockout of Rab3A, Rab3B, Rab3C, and Rab3D in mice demonstrates that Rab3 increases the release probability of a subset of vesicles in the readily releasable pool (Schluter et al., 2006). The authors of this study propose two hypotheses to explain these findings. The first stays within the traditional vesicle-centric framework for Rab3, suggesting that Rab3 docks specific vesicles to sites of high release probability. The second hypothesis posits that Rab3 recruits additional proteins to the release machinery at certain synapses, thereby making Ca2+-mediated release more efficient (Schluter et al., 2006). This second possibility is consistent with our findings in Drosophila that Rab3 regulates the distribution of release apparatus proteins to control the efficacy of individual sites.

Implications for synaptic plasticity

Many neurons differentially regulate the release properties of individual release sites along their axonal lengths through presynaptic, synapse-specific mechanisms. These include the regulation of Ca2+-channel localization and function and the selective accumulation of group III metabotropic glutamate receptors to specific presynaptic active zones (Pelkey and McBain, 2007). We suggest that Rab3 is well positioned to participate in such synapse-specific plasticity mechanisms. Our finding that late expression of Rab3 can rapidly reverse the apposition phenotype of the mutant and redistribute Brp to active zones that previously lacked the protein indicates that 1) Brp is highly mobile and 2) Rab3 can rapidly modulate its distribution among individual sites. Multiple proteins control Rab3 function via its GTPase cycle (Sudhof, 2004), so mechanisms that locally activate or inhibit Rab3 could lead to rapid and local changes in active zone structure and function. Thus, Rab3 is a candidate to participate in plasticity mechanisms that regulate the protein composition and efficacy of individual release sites.

Materials and Methods

Fly Stocks

Flies were maintained at 25°C on standard fly food. Wild type (WT) flies were Canton S (CS) or CS outcrossed to elav-Gal4 (Yao and White, 1994), ELAV-GeneSwitch (Osterwalder et al., 2001), or dvglutNMJX-Gal4 (Daniels et al., 2008) based on the experiment. The following fly lines were obtained from the Bloomington Stock Center: P-element collection on the second chromosome (Bellen et al., 2004), the P-element line P{SUPor-P}KG07292, the deficiency lines Df(2R)ED2076 and Df(2R)BSC29, the UAS-Cacophony-GFP line P{UAS-cac1-EGFP}422A (Kawasaki et al., 2004), and the line containing the ropG27 allele (Wu et al., 1998). The brp69 mutant and the lines containing the UAS-brp and UAS-brp RNAi transgenes were obtained from Stephan Sigrist (University of Berlin, Berlin, Germany) (Kittel et al., 2006; Wagh et al., 2006). The UAS-DGluRIII RNAi transgene line was driven by the muscle driver 24B-Gal4 (Brand and Perrimon, 1993). napts was obtained from Barry Ganetzky (University of Wisconsin, Madison, WI) (Wu et al., 1978) and UAS-Shaker dominant-negative (Sh DN) was obtained from Haig Keshishian (Yale University, New Haven, CT) (Mosca et al., 2005). The UAS-rab3 transgene was generated by cloning the rab3 cDNA (LP05860, obtained from the Drosophila Genomics Resource Center, Bloomington, IN) into a pUAST vector (Brand and Perrimon, 1993). UAS-rab3N134I was created using site-directed mutagenesis.

Antibody Generation

Polyclonal anti sera were generated against a synthetic drab3 peptide (CDADPTLVGGGQKGQRLTDQ). The peptide was conjugated to KLH and injected into rabbits for the generation of antiserum. (BabCo,Berkeley, CA). The antiserum was affinity purified using the peptide conjugated to Affi-Gel 102 (BioRad) using sulfo-SMCC (Pierce) as the cross-linking agent. The antibodies were eluted from the affinity column using 0.1 M glycine, pH 2.8. The purified α-Rab3 antibody was used at 1:1000.


Third-instar larvae were dissected in PBS and fixed in either Bouin’s fixative for 5 min or 4% formaldehyde for 20 min. Larvae were washed with PBS containing 0.1% Triton X-100 (PBT) and blocked in 5% NGS in PBT for 30 min, followed by overnight incubation in primary antibodies in 5% NGS in PBT, three washes in PBT, incubation in secondary antibodies in 5% NGS in PBT for 45 min, three final washes in PBT, and equilibration in 70% glycerol in PBS. Samples were mounted in VectaShield (Vector, Burlingame, CA). The following primary antibodies were used: mouse α-Brp, 1:250 (Developmental Studies Hybridoma Bank), mouse α-Dlg monoclonal antibody (mAb) 4f3, 1:2000 (Developmental Studies Hybridoma Bank), mouse α-synapsin, 1:10 (Developmental Studies Hybridoma Bank), rabbit α-DGluRIII, 1:2500 (Marrus et al, 2004), rabbit α-DVGLUT, 1:10,000 (Daniels et al., 2004), and rabbit α-dPak, 1:2000 (gift from Nicholas Harden, Simon Fraser University, Burnaby, BC, Canada) (Harden et al., 1996). Goat Cy5-, Cy3-, and FITC-conjugated secondary antibodies against mouse and rabbit IgG were used at 1:1000 and were obtained from Jackson ImmunoResearch. Antibodies obtained from the Developmental Studies Hybridoma Bank were developed under the auspices of the National Institute of Child Health and Human Development and maintained by the Department of Biological Sciences of the University of Iowa, Iowa City, IA.

RU486-GeneSwitch experiments

The RU486-GeneSwitch system was administered according to Osterwalder et al. (2001). For early expression of UAS-tagged transgenes driven by ELAV-GeneSwitch, females were maintained on fly food containing 25 μg/ml RU486 (Mifepristone; Sigma, St. Louis, MO) for 2 days before mating and allowed to lay eggs on RU486-containing fly food. For late expression of UAS-rab3, adult females and their larval offspring were both raised upon non-RU486 food. 72 hours after egg-laying, early third-instar larvae were placed on RU486-containing food (Time 0) and removed and dissected at the appropriate time-points following RU486 exposure.

Western Blots

Third-instar larval brains were homogenized in ice-cold homogenization buffer (67mM Tris-HCl, pH 8.0, 67mM NaCl, 2 M urea, 1 mM EDTA, and 1.3% SDS), and samples were run on 15% SDS-PAGE gels according to standard procedures. Rabbit α-Rab3 was used at 1:1000. Mouse α-β-tubulin (E7) (Developmental Studies Hybridoma Bank) was used at 1:100. HRP-conjugated goat α-rabbit and α-mouse (Jackson ImmunoResearch) were used at 1:10,000.

Imaging and analysis

Samples were imaged using a Nikon (Tokyo, Japan) C1 confocal microscope. All genotypes for an individual experiment were imaged at the same gain and set such that signals from the brightest genotype for a given experiment were not saturating. Images were randomized and analyzed using MetaMorph software (Universal Imaging Corporation, West Chester, PA). Statistical analysis was performed using ANOVA for comparison of samples within an experimental group. All histograms and measurements are shown as mean±SEM.

Brp and DGluRIII puncta were manually counted at MN4b NMJs on muscle 4. DGluRIII clusters that were not opposite to a detectable Brp puncta were counted as unapposed DGluRIII clusters. MetaMorph software was used for the quantification of Brp and DGluRIII puncta size and Cac-GFP average intensity. For measurement of Brp or DGluRIII puncta areas, thresholds were kept constant across all genotypes for a given experiment. Although most Brp puncta were distinct, occasional overlapping puncta were separated with the cut drawing tool. DGluRIII clusters were less distinct and often had to be separated with the cut drawing tool. Only DGluRIII clusters with obvious boundaries were analyzed. For measurements of Cac-GFP intensity, DGluRIII puncta were first defined. The average intensity of Cac-GFP signal within each defined DGluRIII cluster was then calculated, and the average muscle background intensity was subtracted.


Electrophysiological experiments were performed as previously described (Daniels et al., 2006). Female third instar larvae were dissected in HL-3 saline containing 0.2 mM Ca2+ and then washed with and recorded in HL-3 saline containing the indicated calcium concentration. Muscle 6 in segments A3 and A4 were clamped at −70 mV and rejected if more than 1nA of holding current was required. For evoked recordings, the end of the cut segmental nerve was sucked into an electrode and stimulated with a 1 msec pulse. The amplitude of the stimulus was set to ensure recruitment of both nerves innervating muscle 6. EJC amplitude was determined by averaging 75 evoked events at 2 Hz. For stimulus trains, the nerve was stimulated with 10 trains of 5 pulses at either 10 or 20 Hz with 5 seconds of rest between each train.

Electron Microscopy

Tissue for electron microscopy experiments was prepared as previously described (Viquez et al., 2009). Pictures were taken on a Hitashi H-7500 TEM using 80 kV accelerating voltage. Image magnifications ranged from 20000 to 80000 to allow visualization of structures ranging in size from an entire bouton to a single active zone. Electron micrographs of NMJs were taken from muscles 6–7 and segments A2-A3 in two larvae of each genotype. Images from a total of 78 boutons from wild type and 70 boutons from rab3/Df mutants were used for analysis. Image files were randomized during analysis. Metamorph imaging software was used to measure bouton circumference and active zone length. Active zones were identified as linear electron densities found between pre- and post-synaptic membranes. High magnification images (>50000×) were used to determine the presence of a t-bar, which was defined as an electron dense rod surrounded by vesicles and apposed to the presynaptic membrane.

Supplementary Material



We wish to thank members of the DiAntonio lab for helpful discussions and Xiaolu Sun and Howard Wynder for technical assistance. We also thank Stephan Sigrist, Barry Ganetzky, Haig Keshishian, and Nicholas Harden for gifts of fly strains and antibodies. E.R.G. was supported by NIH training grant F32 NS056765. This work was supported by a grant from the NIH (NS043171) to A.D.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Albin SD, Davis GW. Coordinating structural and functional synapse development: postsynaptic p21-activated kinase independently specifies glutamate receptor abundance and postsynaptic morphology. J Neurosci. 2004;24:6871–6879. [PubMed]
  • Araki S, Kikuchi A, Hata Y, Isomura M, Takai Y. Regulation of reversible binding of smg p25A, a ras p21-like GTP-binding protein, to synaptic plasma membranes and vesicles by its specific regulatory protein, GDP dissociation inhibitor. J Biol Chem. 1990;265:13007–13015. [PubMed]
  • Atwood HL, Govind CK, Wu CF. Differential ultrastructure of synaptic terminals on ventral longitudinal abdominal muscles in Drosophila larvae. J Neurobiol. 1993;24:1008–1024. [PubMed]
  • Bellen HJ, Levis RW, Liao G, He Y, Carlson JW, Tsang G, Evans-Holm M, Hiesinger PR, Schulze KL, Rubin GM, et al. The BDGP gene disruption project: single transposon insertions associated with 40% of Drosophila genes. Genetics. 2004;167:761–781. [PubMed]
  • Brand AH, Perrimon N. Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development. 1993;118:401–415. [PubMed]
  • Dai Y, Taru H, Deken SL, Grill B, Ackley B, Nonet ML, Jin Y. SYD-2 Liprin-alpha organizes presynaptic active zone formation through ELKS. Nat Neurosci. 2006;9:1479–1487. [PubMed]
  • Daniels RW, Collins CA, Chen K, Gelfand MV, Featherstone DE, DiAntonio A. A single vesicular glutamate transporter is sufficient to fill a synaptic vesicle. Neuron. 2006;49:11–16. [PMC free article] [PubMed]
  • Daniels RW, Collins CA, Gelfand MV, Dant J, Brooks ES, Krantz DE, DiAntonio A. Increased expression of the Drosophila vesicular glutamate transporter leads to excess glutamate release and a compensatory decrease in quantal content. J Neurosci. 2004;24:10466–10474. [PubMed]
  • Daniels RW, Gelfand MV, Collins CA, DiAntonio A. Visualizing glutamatergic cell bodies and synapses in Drosophila larval and adult CNS. J Comp Neurol. 2008;508:131–152. [PubMed]
  • DiAntonio A, Burgess RW, Chin AC, Deitcher DL, Scheller RH, Schwarz TL. Identification and characterization of Drosophila genes for synaptic vesicle proteins. J Neurosci. 1993;13:4924–4935. [PubMed]
  • Dickman DK, Lu Z, Meinertzhagen IA, Schwarz TL. Altered synaptic development and active zone spacing in endocytosis mutants. Curr Biol. 2006;16:591–598. [PubMed]
  • Eaton BA, Fetter RD, Davis GW. Dynactin is necessary for synapse stabilization. Neuron. 2002;34:729–741. [PubMed]
  • Fejtova A, Gundelfinger ED. Molecular organization and assembly of the presynaptic active zone of neurotransmitter release. Results Probl Cell Differ. 2006;43:49–68. [PubMed]
  • Fouquet W, Owald D, Wichmann C, Mertel S, Depner H, Dyba M, Hallermann S, Kittel RJ, Eimer S, Sigrist SJ. Maturation of active zone assembly by Drosophila Bruchpilot. J Cell Biol. 2009;186:129–145. [PMC free article] [PubMed]
  • Fukui K, Sasaki T, Imazumi K, Matsuura Y, Nakanishi H, Takai Y. Isolation and characterization of a GTPase activating protein specific for the Rab3 subfamily of small G proteins. J Biol Chem. 1997;272:4655–4658. [PubMed]
  • Harden N, Lee J, Loh HY, Ong YM, Tan I, Leung T, Manser E, Lim L. A Drosophila homolog of the Rac- and Cdc42-activated serine/threonine kinase PAK is a potential focal adhesion and focal complex protein that colocalizes with dynamic actin structures. Mol Cell Biol. 1996;16:1896–1908. [PMC free article] [PubMed]
  • Iwasaki K, Staunton J, Saifee O, Nonet M, Thomas JH. aex-3 encodes a novel regulator of presynaptic activity in C. elegans. Neuron. 1997;18:613–622. [PubMed]
  • Johnston PA, Archer BT, 3rd, Robinson K, Mignery GA, Jahn R, Sudhof TC. rab3A attachment to the synaptic vesicle membrane mediated by a conserved polyisoprenylated carboxy-terminal sequence. Neuron. 1991;7:101–109. [PubMed]
  • Kaufmann N, DeProto J, Ranjan R, Wan H, Van Vactor D. Drosophila liprin-alpha and the receptor phosphatase Dlar control synapse morphogenesis. Neuron. 2002;34:27–38. [PubMed]
  • Kawasaki F, Zou B, Xu X, Ordway RW. Active zone localization of presynaptic calcium channels encoded by the cacophony locus of Drosophila. J Neurosci. 2004;24:282–285. [PubMed]
  • Kittel RJ, Wichmann C, Rasse TM, Fouquet W, Schmidt M, Schmid A, Wagh DA, Pawlu C, Kellner RR, Willig KI, et al. Bruchpilot promotes active zone assembly, Ca2+ channel clustering, and vesicle release. Science. 2006;312:1051–1054. [PubMed]
  • Li J, Ashley J, Budnik V, Bhat MA. Crucial role of Drosophila neurexin in proper active zone apposition to postsynaptic densities, synaptic growth, and synaptic transmission. Neuron. 2007;55:741–755. [PMC free article] [PubMed]
  • Mahoney TR, Liu Q, Itoh T, Luo S, Hadwiger G, Vincent R, Wang ZW, Fukuda M, Nonet ML. Regulation of synaptic transmission by RAB-3 and RAB-27 in Caenorhabditis elegans. Mol Biol Cell. 2006;17:2617–2625. [PMC free article] [PubMed]
  • Marrus SB, Portman SL, Allen MJ, Moffat KG, DiAntonio A. Differential localization of glutamate receptor subunits at the Drosophila neuromuscular junction. J Neurosci. 2004;24:1406–1415. [PubMed]
  • Mosca TJ, Carrillo RA, White BH, Keshishian H. Dissection of synaptic excitability phenotypes by using a dominant-negative Shaker K+ channel subunit. Proc Natl Acad Sci U S A. 2005;102:3477–3482. [PubMed]
  • Nonet ML, Staunton JE, Kilgard MP, Fergestad T, Hartwieg E, Horvitz HR, Jorgensen EM, Meyer BJ. Caenorhabditis elegans rab-3 mutant synapses exhibit impaired function and are partially depleted of vesicles. J Neurosci. 1997;17:8061–8073. [PubMed]
  • Ohtsuka T, Takao-Rikitsu E, Inoue E, Inoue M, Takeuchi M, Matsubara K, Deguchi-Tawarada M, Satoh K, Morimoto K, Nakanishi H, Takai Y. Cast: a novel protein of the cytomatrix at the active zone of synapses that forms a ternary complex with RIM1 and munc13-1. J Cell Biol. 2002;158:577–590. [PMC free article] [PubMed]
  • Osterwalder T, Yoon KS, White BH, Keshishian H. A conditional tissue-specific transgene expression system using inducible GAL4. Proc Natl Acad Sci U S A. 2001;98:12596–12601. [PubMed]
  • Owald D, Sigrist SJ. Assembling the presynaptic active zone. Curr Opin Neurobiol 2009 [PubMed]
  • Patel MR, Shen K. RSY-1 is a local inhibitor of presynaptic assembly in C. elegans. Science. 2009;323:1500–1503. [PMC free article] [PubMed]
  • Pelkey KA, McBain CJ. Differential regulation at functionally divergent release sites along a common axon. Curr Opin Neurobiol. 2007;17:366–373. [PubMed]
  • Petersen SA, Fetter RD, Noordermeer JN, Goodman CS, DiAntonio A. Genetic analysis of glutamate receptors in Drosophila reveals a retrograde signal regulating presynaptic transmitter release. Neuron. 1997;19:1237–1248. [PubMed]
  • Pielage J, Fetter RD, Davis GW. A postsynaptic spectrin scaffold defines active zone size, spacing, and efficacy at the Drosophila neuromuscular junction. J Cell Biol. 2006;175:491–503. [PMC free article] [PubMed]
  • Rasse TM, Fouquet W, Schmid A, Kittel RJ, Mertel S, Sigrist CB, Schmidt M, Guzman A, Merino C, Qin G, et al. Glutamate receptor dynamics organizing synapse formation in vivo. Nat Neurosci. 2005;8:898–905. [PubMed]
  • Schluter OM, Basu J, Sudhof TC, Rosenmund C. Rab3 superprimes synaptic vesicles for release: implications for short-term synaptic plasticity. J Neurosci. 2006;26:1239–1246. [PubMed]
  • Schluter OM, Schmitz F, Jahn R, Rosenmund C, Sudhof TC. A complete genetic analysis of neuronal Rab3 function. J Neurosci. 2004;24:6629–6637. [PubMed]
  • Schluter OM, Schnell E, Verhage M, Tzonopoulos T, Nicoll RA, Janz R, Malenka RC, Geppert M, Sudhof TC. Rabphilin knock-out mice reveal that rabphilin is not required for rab3 function in regulating neurotransmitter release. J Neurosci. 1999;19:5834–5846. [PubMed]
  • Schmid A, Hallermann S, Kittel RJ, Khorramshahi O, Frolich AM, Quentin C, Rasse TM, Mertel S, Heckmann M, Sigrist SJ. Activity-dependent site-specific changes of glutamate receptor composition in vivo. Nat Neurosci. 2008;11:659–666. [PubMed]
  • Schoch S, Castillo PE, Jo T, Mukherjee K, Geppert M, Wang Y, Schmitz F, Malenka RC, Sudhof TC. RIM1alpha forms a protein scaffold for regulating neurotransmitter release at the active zone. Nature. 2002;415:321–326. [PubMed]
  • Shirataki H, Kaibuchi K, Sakoda T, Kishida S, Yamaguchi T, Wada K, Miyazaki M, Takai Y. Rabphilin-3A, a putative target protein for smg p25A/rab3A p25 small GTP-binding protein related to synaptotagmin. Mol Cell Biol. 1993;13:2061–2068. [PMC free article] [PubMed]
  • Sudhof TC. The synaptic vesicle cycle. Annu Rev Neurosci. 2004;27:509–547. [PubMed]
  • Takao-Rikitsu E, Mochida S, Inoue E, Deguchi-Tawarada M, Inoue M, Ohtsuka T, Takai Y. Physical and functional interaction of the active zone proteins, CAST, RIM1, and Bassoon, in neurotransmitter release. J Cell Biol. 2004;164:301–311. [PMC free article] [PubMed]
  • Viquez NM, Fuger P, Valakh V, Daniels RW, Rasse TM, DiAntonio A. PP2A and GSK-3beta act antagonistically to regulate active zone development. J Neurosci. 2009;29:11484–11494. [PMC free article] [PubMed]
  • Wada M, Nakanishi H, Satoh A, Hirano H, Obaishi H, Matsuura Y, Takai Y. Isolation and characterization of a GDP/GTP exchange protein specific for the Rab3 subfamily small G proteins. J Biol Chem. 1997;272:3875–3878. [PubMed]
  • Wagh DA, Rasse TM, Asan E, Hofbauer A, Schwenkert I, Durrbeck H, Buchner S, Dabauvalle MC, Schmidt M, Qin G, et al. Bruchpilot, a protein with homology to ELKS/CAST, is required for structural integrity and function of synaptic active zones in Drosophila. Neuron. 2006;49:833–844. [PubMed]
  • Wairkar YP, Toda H, Mochizuki H, Furukubo-Tokunaga K, Tomoda T, Diantonio A. Unc-51 controls active zone density and protein composition by downregulating ERK signaling. J Neurosci. 2009;29:517–528. [PMC free article] [PubMed]
  • Wang Y, Okamoto M, Schmitz F, Hofmann K, Sudhof TC. Rim is a putative Rab3 effector in regulating synaptic-vesicle fusion. Nature. 1997;388:593–598. [PubMed]
  • Weber E, Jilling T, Kirk KL. Distinct functional properties of Rab3A and Rab3B in PC12 neuroendocrine cells. J Biol Chem. 1996;271:6963–6971. [PubMed]
  • Wu CF, Ganetzky B, Jan LY, Jan YN, Benzer S. A Drosophila mutant with a temperature-sensitive block in nerve conduction. Proc Natl Acad Sci U S A. 1978;75:4047–4051. [PubMed]
  • Wu MN, Littleton JT, Bhat MA, Prokop A, Bellen HJ. ROP, the Drosophila Sec1 homolog, interacts with syntaxin and regulates neurotransmitter release in a dosage-dependent manner. Embo J. 1998;17:127–139. [PubMed]
  • Wu S, Mehta SQ, Pichaud F, Bellen HJ, Quiocho FA. Sec15 interacts with Rab11 via a novel domain and affects Rab11 localization in vivo. Nat Struct Mol Biol. 2005;12:879–885. [PubMed]
  • Yao KM, White K. Neural specificity of elav expression: defining a Drosophila promoter for directing expression to the nervous system. J Neurochem. 1994;63:41–51. [PubMed]
  • Zucker RS, Regehr WG. Short-term synaptic plasticity. Annu Rev Physiol. 2002;64:355–405. [PubMed]