Intersectin is a multidomain dynamin-binding protein implicated in numerous functions in the nervous system, including synapse formation and endocytosis. Here, we demonstrate that during neurotransmitter release in the central synapse, intersectin, like its binding partner dynamin, is redistributed from the synaptic vesicle pool to the periactive zone. Acute perturbation of the intersectin–dynamin interaction by microinjection of either intersectin antibodies or Src homology 3 (SH3) domains inhibited endocytosis at the fission step. Although the morphological effects induced by the different reagents were similar, antibody injections resulted in a dramatic increase in dynamin immunoreactivity around coated pits and at constricted necks, whereas synapses microinjected with the GST (glutathione S-transferase)–SH3C domain displayed reduced amounts of dynamin in the neck region. Our data suggest that intersectin controls the amount of dynamin released from the synaptic vesicle cluster to the periactive zone and that it may regulate fission of clathrin-coated intermediates.
- presynaptic mechanisms
- synaptic vesicle recycling
- periactive zone
- glutamatergic neurons
- spinal cord
Clathrin-mediated endocytosis is one of the major pathways for synaptic vesicle recycling (Murthy and De Camilli, 2003; Granseth et al., 2006). Along with numerous other proteins implicated in this process, the GTPase dynamin is a key element that is required for the fission of newly formed vesicles from the plasma membrane. Recent in vitro studies have provided clues about how dynamin functions in fission (Chen et al., 2004; Roux et al., 2006). However, it is still unclear how the protein is delivered to coated intermediates to perform its function. It is believed that the C-terminal PRD (proline-rich domain) plays an important role. This domain interacts with Src homology 3 (SH3) domains of several endocytic proteins enriched in synapses such as amphiphysin, endophilin, syndapin, and intersectin (Praefcke and McMahon, 2004; Anggono et al., 2006). All of these proteins have been implicated in specific protein–protein and protein–lipid interactions in the synapse, suggesting that they are interacting with dynamin at particular steps of the vesicle cycle. In addition, some of these interactions are modulated by phosphorylation, which involves proteins such as calcineurin and Cdk5 (cyclin-dependent kinase 5), further indicating that specific interactions dominate in resting synapses, whereas others occur during activity (Cousin et al., 2001; Anggono et al., 2006).
The role of the scaffolding protein intersectin in vertebrate synapses remains obscure. Two intersectin genes (1 and 2) have been identified, each producing long (L) and short isoforms with many splice variants (Guipponi et al., 1998; Pucharcos et al., 2000). Only the long isoform of intersectin 1 is selectively expressed in neurons. Intersectin 1L consists of two epidermal growth factor receptor pathway substrate clone 15 (Eps15) homology (EH) domains along with a coiled-coil domain, which bind epsin 1/2, Eps15, stoned B/stonin 2, SNAP-25 (synaptosome-associated protein of 25 kDa) (Okamoto et al., 1999; Sengar et al., 1999; Martina et al., 2001), and five SH3 domains located in the middle part; three of these interact with dynamin. In addition, synapsin, synaptojanin, and stoned B/stonin 2, the Arp2/3 (actin-related protein 2/3)-interacting protein, neuronal Wiskott–Aldrich syndrome protein (N-WASP), and a signaling protein, Sos (son of sevenless), have also been identified as binding partners for this region (Tong et al., 2000; Hussain et al., 2001; Zamanian et al., 2003). Intersectin 1L also possesses a PH (pleckstrin homology) domain, which interacts with phospholipids, and a DH (Dbl homology) domain, which functions as a GEF (guanine nucleotide exchange factor) for Cdc42 (Zamanian et al., 2003), indicating a potential role for intersectin in the organization of the actin cytoskeleton (Hussain et al., 2001; McPherson, 2002). Loss-of-function mutations of the Drosophila ortholog of intersectin, dynamin-associated protein of 160 kDa (Dap160), are lethal, demonstrating that the gene is essential (Koh et al., 2004; Marie et al., 2004). Null mutant flies show severe defects in the formation of neuromuscular junctions (NMJs) and in synaptic vesicle recycling (Koh et al., 2004; Marie et al., 2004). These data suggest that intersectin may play an important role in sustaining neurotransmitter release in vertebrate synapses.
In this study, we investigated the role of intersectin–dynamin interactions in synaptic vesicle recycling by using the lamprey giant reticulospinal synapse as a model system. We show that intersectin redistributes to sites of clathrin-mediated endocytosis from the synaptic vesicle cluster during activity and controls the amount of dynamin destined for migration to the periactive zone. In addition, our data suggest that intersectin is involved in the regulation of fission.
Materials and Methods
Cloning of lamprey intersectin.
Using the QuickPrep mRNA Extraction kit (GE Healthcare, Little Chalfont, UK), mRNA was isolated from lamprey (Lampetra fluviatilis) brain. cDNA was synthesized with the ThermoScript reverse transcriptase (Invitrogen, Carlsbad, CA) and random or oligo-dT primers. An intersectin ortholog was partially cloned using degenerate primers against the conserved SH3 domain region and a random-primed lamprey brain cDNA. The resulting bands were subcloned into the pCR2.1–TOPO vector (Invitrogen) and sequenced. Sequencing was performed at KISeq, Karolinska Institutet (Stockholm, Sweden). Using oligo-dT- and random-primed cDNA, additional independent clones were generated confirming the found sequence. 3′ RACE (rapid amplification of cDNA ends) was performed using an oligo-dT-primed lamprey brain cDNA, a specific primer from the 3′ end of the obtained sequence, and an oligo-dT primer. This clone contained an in-frame stop codon followed by an ∼1300 bp untranslated region and ended with a poly-A tail. The 5′ end of the SH3 domain region was cloned using 5′-poly-A-tailed random-primed cDNA, an oligo-dT primer, and a specific reverse primer in the 5′ end of the known sequence.
In addition, a random-primed λ-ZAP cDNA library from lamprey brain mRNA was screened using standard methods (Sambrook et al., 1989). A probe was generated by restriction digestion of the subcloned band from the initial PCR, labeled with [32P]dCTP using the Ready-To-Go DNA Labeling Beads (GE Healthcare) and purified with a ProbeQuant G-50 Micro Column (GE Healthcare). Positive clones were sequenced.
Glutathione S-transferase (GST)-fusion protein constructs of the lamprey SH3 domains [A, amino acids 740–806; B, 913–972; C, 1002–1060; D, 1073–1138; E, 1154–1214; A–C, 762–1064; amino acid numbering of lamprey intersectin here corresponds to that of human intersectin 1L; GenBank accession number NP003015 (see also Fig. 1A)] were generated, using the pGEX-6P-2 vector (GE Healthcare). The numbering of the Drosophila Dap160 aa sequence in Figure 1A is shown according to accession number AAC39138. The accession number of lamprey intersectin is EF134956.
Polyclonal antibodies were raised in rabbit against GST-fusion proteins of lamprey SH3 domains A–C (LIS-AC; corresponding to amino acids 762–1064 in human intersectin 1L) (see Fig. 1A) and SH3 domains C–E (LIS-CE; corresponding to amino acids 1034–1173 in human intersectin 1L) (see Fig. 1A). The antibodies were purified on protein A Sepharose (GE Healthcare) and affinity purified on a Hi-Trap column (GE Healthcare) with the antigen covalently coupled. Nonspecific rabbit IgGs were used in control experiments. Antibodies raised against Xenopus and rat intersectin EH domains (Hussain et al., 1999; Okamoto et al., 1999) were used for detection of intersectin. Dynamin was detected using the DG-1 (Grabs et al., 1997) and LD-1 (Evergren et al., 2004a) antibodies. D domain antibodies were used to detect lamprey synapsin (Bloom et al., 2003), and amphiphysin was detected using lamprey SH3 domain antibodies (Evergren et al., 2004b). The synaptojanin and Eps15 antibodies have been described previously (Gad et al., 2000; Kent et al., 2002). Synapsin E-domain antibodies (Pieribone et al., 1995) were used in microinjection experiments.
Affinity chromatography of tissue extracts and immunoprecipitation experiments.
Intersectin antibodies were coupled to protein A Sepharose (GE Healthcare), and GST-fusion proteins were coupled to glutathione Sepharose (GE Healthcare) and incubated with a 10% lamprey brain detergent extract (Gad et al., 2000) in buffer A (10 mm HEPES buffer, pH 7.4, containing 100 mm KCl, 2 mm MgCl2, and 1% Triton X-100) plus protease inhibitors or only with buffer A for 2 h at 4°C on a rotating wheel. Samples were washed three times with buffer A with 1% Triton X-100 and three times with buffer A without Triton X-100, eluted with sample buffer, and analyzed by SDS-PAGE.
Dissection of the trunk region of the lamprey spinal cord, microinjection procedures, stimulation, and fixation were performed as described previously (Gad et al., 2000). Animals were treated according to the Swedish Animal Welfare Act (SFS 1988:534), as approved by the Local Animal Research Committee of Stockholm. All efforts were made to minimize animal suffering and to reduce the number of animals used. Giant axons were stimulated at 5 Hz using extracellular electrodes (30 min; ∼9000 action potentials). The effects described in this study were detectable even for stimulation at 0.2 Hz (30 min; 360 action potentials; data not shown). Naturally, fewer intermediates were observed around synapses stimulated at 0.2 Hz, and the statistical analysis became very complex, because it required a quantitative analysis of hundreds of sections. Stimulation at 20 Hz was accomplished with intracellular electrodes. Action potentials in giant axons were monitored with an extracellular suction electrode placed at the caudal part of the preparation. Axons referred to as nonstimulated controls were either adjacent to axons stimulated intracellularly or from a specimen left at rest in a low Ca2+ Ringer's solution for 45–90 min. Microinjections of fluorescent reagents were monitored with a CCD camera (Princeton Instruments, Trenton, NJ) connected to a fluorescence microscope and a Nikon (Tokyo, Japan) D-eclipse C1 confocal microscope using a 40× water-immersion objective (numerical aperture, 0.80). Confocal images were analyzed using NIH ImageJ software (http://rsb.info.nih.gov/ij/).
The antibodies and GST-fusion proteins used in microinjection experiments were labeled with Alexa-602, Alexa-594 (Invitrogen), or Cy5 (GE Healthcare) according to the manufacturer's instructions. Phalloidin–Alexa-488 was purchased from Invitrogen. GTPγS (Roche Diagnostics, Indianapolis, IN) was diluted in Texas Red dextran (molecular weight, 3000; Invitrogen) to a final concentration of 10 mm.
Spinal cords were fixed during stimulation in 3% glutaraldehyde/0.5% paraformaldehyde/4% tannic acid in 0.1 m cacodylate buffer, pH 7.4, for 1 h, followed by the same fixative without tannic acid for an additional 3 h. After postfixation in 1% osmium tetroxide for 1 h and dehydration in alcohol, the specimens were embedded in Durcupan ACM (Fluka, Saint Quentin Fallavier, France). Serial ultrathin sections were cut with a diamond knife and viewed at 80 kV in a Tecnai 12 or a Philips CM12 electron microscope (FEI, Ekerö, Sweden).
Effects of microinjections were analyzed in 20–60 synapses cut in serial ultrathin sections at various distances from microinjection sites (i.e., different concentrations). All described effects were reproduced in at least two independent preparations in which three to six axons were microinjected. The number of synaptic vesicles and coated pits and the membrane length of invaginations in the endocytic zone in antibody injection experiments were determined from middle sections of at least five serially cut synapses. Because the number of synaptic vesicles in the cluster is correlated to the length of the active zone in the reticulospinal synapse, the values for number of synaptic vesicles and coated pits were normalized to the length of the active zone (Gustafsson et al., 2002). The length of endocytic zone membrane invaginations in the axoplasm was measured in a 5 × 3.5 μm area above the active zone using NIH ImageJ software. In addition, the area of membrane invaginations was calculated in five injected synapses stimulated at 5 Hz. The area of membrane folds was correlated to the expected membrane area of fused vesicles. The number of fused vesicles was estimated from the reduction of vesicles in a middle section, which correlates with the total number of vesicles in the synapse (Gustafsson et al., 2002).
Postembedding immunogold labeling of giant reticulospinal synapses.
An unstimulated lamprey spinal cord was dissected out and fixed in 4% paraformaldehyde/0.5% glutaraldehyde/4% tannic acid in 0.1 m cacodylate buffer, pH 7.4, for 1 h at 4°C and then for 3 h in fixative without tannic acid. Specimens were stained en bloc with 2% uranyl acetate, dehydrated in graded ethanol series, and embedded at −25°C in LR Gold resin (Fluka). Serial ultrathin sections were collected on nickel mesh grids and incubated overnight at 4°C with primary antibodies (LIS-AC antibodies, 0.03 mg/ml; DG-1 antibodies, 0.01 mg/ml) diluted in 1% human serum albumin/5% bovine serum albumin in Tris-PBS (TPBS), pH 7.4. Secondary antibodies conjugated to 5 nm colloidal gold (GE Healthcare) were used at a dilution of 1:25. Gold particles were enhanced using the IntenSE Silver enhancement kit (GE Healthcare). Sections were counterstained with uranyl acetate and lead citrate and examined in a Tecnai 12 electron microscope. Labeling experiments, in which primary antibodies were omitted, did not produce any specific labeling. The number of particles was quantified in selected regions, and the density of gold particles per square micrometer was calculated. The synaptic vesicle cluster was defined as a region containing synaptic vesicles clustered at an active zone. The periactive zone was defined as a presynaptic area (0.5 × 0.5 μm) lateral to the active zone, and the axoplasmic matrix was defined as an area (0.5 × 0.5 μm) >0.5 μm from the synaptic vesicle cluster.
Preembedding immunocytochemistry for lamprey preparations.
Lamprey spinal cords were incubated in either low Ca2+ or high K+ Ringer's solution at 4.0°C for 20 min or stimulated electrically at 5 Hz for 30 min, fixed in 3% paraformaldehyde/0.5% glutaraldehyde in 0.1 m phosphate buffer, pH 7.4, for 4 h. Longitudinal sections were cut in a vibratome to open up giant reticulospinal axons and thereby provide access for antibodies. Microinjected specimens were also cut transversally at the site of injection (supplemental Fig. 6, available at www.jneurosci.org as supplemental material). The labeling procedure was performed as described previously (Evergren et al., 2004a). Briefly, specimens were incubated with primary antibodies (LIS-AC and LD-1 antibodies, 0.03 mg/ml; amphiphysin–SH3 antibodies, 0.13 mg/ml) diluted in 1% human serum albumin in TPBS overnight at 4°C on a shaker. After incubation with secondary antibodies (5-nm-gold-conjugated diluted 1:50; GE Healthcare), the specimens were postfixed in 3% glutaraldehyde in 0.1 m phosphate buffer, pH 7.4, for 1 h, and in 1% osmium tetroxide for 1 h. Sections were stained en bloc with uranyl acetate and embedded in Durcupan ACM (Fluka). Ultrathin sections were counterstained with uranyl acetate and lead citrate and examined in a Tecnai 12 microscope. Quantification of immunogold labeling at constricted collared clathrin-coated intermediates was performed as described in the legend to supplemental Figure 7 (available at www.jneurosci.org as supplemental material). Omission of primary antibodies did not produce any labeling. The periactive zone was defined as above.
Characterization of lamprey intersectin–dynamin interactions
By screening a randomly primed cDNA library from lamprey brain using degenerate and specific primers, we acquired a sequence corresponding to the lamprey intersectin SH3 domains A–E. The cloned sequence had 79% similarity (58% identity) to the SH3 domain region of human intersectin 1L and 59% similarity to Drosophila Dap160 (Fig. 1A). The similarity of the lamprey sequence to human intersectin 2L was 72%.
GST-fusion proteins of SH3 domains A–C and C–E were expressed and used to generate antibodies (Fig. 1A). Long and short isoforms of mammalian and Xenopus intersectin migrate at 200 and 170 kDa by SDS-PAGE, respectively (Hussain et al., 1999; Okamoto et al., 1999). The affinity-purified antibodies LIS-AC and LIS-CE recognized two bands of the expected molecular weight in Western blots of lamprey brain detergent extract (Fig. 1B). Bands of the same molecular weight were also detected by polyclonal antibodies raised against the EH domains of rat (S750) (Fig. 1B) and Xenopus intersectin, as well as by an antibody raised against the SH3 domains of rat intersectin (data not shown). To provide additional evidence that the antibodies recognize lamprey intersectin, we immunoprecipitated the protein from lamprey brain extract using LIS-AC antibodies and detected the precipitate using the antibodies LIS-CE and S750. Both antibodies recognized bands at 200 and 170 kDa (Fig. 1C). Intersectin could also be immunoprecipitated using the LIS-CE antibody (Fig. 1E). We thus conclude that the proteins identified by the LIS-AC and LIS-CE antibodies are the lamprey intersectin orthologs.
We next investigated whether lamprey intersectin interacts with dynamin as well as with other endocytic proteins via its SH3 domains. Interactions with synaptojanin and synapsin in addition to dynamin have been reported previously for mammalian and Drosophila intersectin/Dap160 (Roos and Kelly, 1998; Okamoto et al., 1999; Zamanian et al., 2003). GST-fusion proteins of single SH3 domains of lamprey intersectin were used to affinity purify binding partners from lamprey brain extract. Bound material was analyzed by SDS-PAGE and Western blot (Fig. 1D). Dynamin was found to interact with the SH3 domains A, C, and E. Synaptojanin and synapsin bound mainly to SH3A. Thus, our data indicate that the interactions with dynamin, synaptojanin, and synapsin are conserved in lamprey.
The goal of our study was to clarify the role of intersectin–dynamin interactions in the synapse by using antibodies to acutely perturb these interactions. We therefore investigated whether the antibodies were able to inhibit the binding of dynamin to intersectin. Intersectin was immunoprecipitated using either LIS-AC or LIS-CE antibodies. “Bound” and “unbound” fractions were analyzed by Western blot (Fig. 1E). Both antibodies efficiently precipitated intersectin, depleting it from the extract. However, only a small amount of dynamin was detected in the bound fractions, indicating that the antibodies were blocking the binding sites for dynamin efficiently (Fig. 1E). Eps15, an endocytic protein that binds to the intersectin coiled-coil region (Okamoto et al., 1999; Sengar et al., 1999), was, however, efficiently coprecipitated (data not shown). Immunoprecipitation of intersectin using the EH domain antibody S750 efficiently coimmunoprecipitated dynamin. We thus conclude that the lamprey intersectin antibodies occupy the binding sites for dynamin.
Localization of intersectin in living synapses in situ
Previous studies in Drosophila NMJs have demonstrated that dynamin redistributes to endocytic sites during synaptic activity (Estes et al., 1996), whereas intersectin is stably localized to the endocytic zone (Roos and Kelly, 1998, 1999). To investigate whether this is also true for central vertebrate synapses, Alexa-tagged intersectin antibodies were microinjected into living reticulospinal axons. This resulted in an accumulation of fluorescence at spots on the axonal surface (Fig. 2A). Concentration of the labeling was facilitated by the induction of action potentials in the axons. An accumulation of fluorescence in spots was also observed after microinjections of antibodies against the synaptic vesicle proteins synapsin (Fig. 2A) and synaptotagmin (Pieribone et al., 1995), as well as phalloidin (Fig. 2A), which labels filamentous actin in the periactive zone of synapses (Shupliakov et al., 2002). Nonspecific IgGs failed to accumulate in spots (Fig. 2A). This strongly indicated that the intersectin antibodies accumulated at synaptic regions.
Even when using a CCD camera (Fig. 2A), it was evident that the spots of fluorescence induced by the reagents had different appearances. To obtain more detailed information on the localization of intersectin, we used confocal microscopy. Axons were microinjected with Alexa-594-tagged intersectin antibodies, which was followed by microinjection of Alexa-488–phalloidin. This resulted in an accumulation of both reagents at synaptic regions (Fig. 2B–D). Figure 2B shows a low-power image of a reticulospinal axon microinjected under resting conditions. Analysis of the labeling pattern revealed that intersectin antibodies accumulated in a spot, whereas the majority of the periactive zone marker phalloidin surrounded this site (Fig. 2C). A merge of the two images placed the intersectin labeling in the middle of the “phalloidin ring,” corresponding to the localization of the synaptic vesicle cluster (Shupliakov et al., 2002). Figure 2D shows an image of a synapse acquired during stimulation at 5 Hz. Under these conditions, the intersectin antibodies were dispersed over a larger area of the synapse, and colocalized with phalloidin rings (Fig. 2D). These results indicate that intersectin relocates to the periactive zone during synaptic activity in a living vertebrate synapse. To investigate this observation further, we studied the localization of intersectin in the periactive zone at the ultrastructural level.
Subcellular localization of intersectin and dynamin in the giant reticulospinal synapse
To localize intersectin at the ultrastructural level, we first used the postembedding immunogold technique, which provides antibodies equal access to epitopes exposed on the surface of a section (Ottersen, 1989). In agreement with light microscopic observations (Fig. 2C), periactive zones in synapses fixed at rest contained only a few gold particles when labeled with intersectin antibodies (Fig. 3A). Immunogold particles were predominantly associated with synaptic vesicle clusters (Fig. 3A) (Evergren et al., 2006). The density of gold particles over this synaptic region (21.6 ± 5.6 particles/μm2; n = 14; mean ± SEM) was 20 times higher than in the axoplasmic matrix (1.4 ± 0.5 particles/μm2; n = 13; mean ± SEM; p < 0.0001; two-tailed Student's t test). The level of labeling in the periactive zone area (1.8 ± 1.0 particles/μm2; n = 21; mean ± SEM) was not significantly different from the labeling over the axoplasmic matrix (p > 0.05; two-tailed Student's t test). Labeling for dynamin showed a distribution pattern similar to that of intersectin (Fig. 3B). Gold particles were predominantly accumulated over the synaptic vesicle cluster (20.2 ± 4.1 particles/μm2; n = 16; mean ± SEM), compared with the axoplasmic matrix (1.5 ± 0.6 particles/μm2; n = 16; mean ± SEM; p < 0.0001; two-tailed Student's t test) or the periactive zone (1.0 ± 0.3 particles/μm2; n = 16; mean ± SEM; p < 0.0001; two-tailed Student's t test).
To achieve a higher labeling efficiency at periactive zones, we used a preembedding immunogold technique (Evergren et al., 2004a). No clear association of gold particles with this region in synapses at rest was detected for intersectin (Fig. 3C) (6.7 particles/μm2; mean ± SEM; n = 14). However, strong labeling of periactive zones was observed in synapses stimulated with high K+ for 20 min (Fig. 3D) (54.7 ± 7.0 particles/μm2; mean ± SEM, n = 14). The density of gold particles was increased more than eightfold compared with resting synapses (p < 0.0001; two-tailed Student's t test). Gold particles were found predominantly on endocytic intermediates from early to late stages and were associated with the coat region (Fig. 3D,E). A similar labeling pattern was obtained with LIS-AC and LIS-CE intersectin antibodies. Similar to intersectin, the dynamin immunolabeling was also associated with clathrin-coated intermediates in the periactive zone in stimulated synapses at different stages (Fig. 3G,H). The labeling in the periactive zone was significantly increased in stimulated synapses (34.6 ± 3.2 particles/μm2; mean ± SEM; n = 8) compared with resting terminals (Fig. 3F,G) (0.6 ± 0.4 particles/μm2; mean ± SEM; n = 8; p < 0.0001; two-tailed Student's t test). Together, these results show that intersectin and its binding partner dynamin are redistributed after stimulation from the synaptic vesicle cluster to the periactive zone and that they are associated with endocytic intermediates in synapses from early stages on.
Acute perturbation of intersectin–dynamin interactions disrupts synaptic vesicle membrane recycling
Previous studies in non-neuronal cells have demonstrated that the overexpression of SH3 domains efficiently inhibits clathrin-mediated endocytosis (Simpson et al., 1999). To test whether the blockade of intersectin-binding sites on dynamin may negatively affect endocytosis in a living reticulospinal synapse, GST–SH3C domain was microinjected into reticulospinal axons and stimulated to induce synaptic vesicle recycling at a physiological rate (5 Hz) (Deliagina and Fagerstedt, 2000) and then analyzed in serial ultrathin sections (supplemental Fig. 1B, available at www.jneurosci.org as supplemental material). The domain induced a significant reduction in the size of the synaptic vesicle pool compared with uninjected control synapses (Fig. 4A,C,E) (p < 0.0001; two-tailed Student's t test). An increase in the number of constricted clathrin-coated pits with electron-dense collars (Fig. 4A–C,F) (p < 0.0001; two-tailed Student's t test) and an accumulation of membrane infolds at periactive zones occurred (Fig. 4A,C,G) (p < 0.05; two-tailed Student's t test), indicating that endocytosis was inhibited. The morphology of synapses injected with GST did not differ significantly from uninjected control synapses (Fig. 4C–G) (p > 0.05; two-tailed Student's t test). GST–SH3B domain was used as a control, because it does not bind dynamin (Fig. 1D). It did not produce an accumulation of coated pits (supplemental Fig. 2A,D, available at www.jneurosci.org as supplemental material) (p > 0.05; two-tailed Student's t test) or membrane invaginations (supplemental Fig. 2A,E, available at www.jneurosci.org as supplemental material) (p > 0.05; two-tailed Student's t test). Together, these data show that acute perturbation of dynamin function by intersectin SH3C domain inhibits synaptic vesicle recycling and interferes with the fission of vesicles from the plasma membrane (Fig. 4H, schematics).
To block intersectin SH3 domain interactions, LIS-AC antibodies were microinjected into living reticulospinal synapses. As in the case of the SH3C domain injection, a significant reduction in number of synaptic vesicles (Fig. 5G) (p < 0.0001; two-tailed Student's t test), an accumulation of clathrin-coated pits (Fig. 5H) (p < 0.0001; two-tailed Student's t test), and an increase in membrane folds occurred (Fig. 5I) (p < 0.05; two-tailed Student's t test) compared with uninjected synapses stimulated at 5 Hz (Figs. 4C, 5A–C, schematics in 5J, supplemental Figs. 1, 5, available at www.jneurosci.org as supplemental material). Several control experiments confirmed that the effects of the antibodies on synaptic vesicle recycling were specific. First, they were not observed in synapses of axons microinjected with intersectin antibodies at rest (Fig. 5F–I) (p > 0.05; two-tailed Student's t test). Second, the induced defects were different compared with those produced by synapsin E-domain antibodies, which cause disruption of the distal pool of vesicles in the synaptic vesicle cluster (supplemental Fig. 5D, available at www.jneurosci.org as supplemental material) (Pieribone et al., 1995). Third, injection of nonspecific rabbit IgGs did not induce changes in the presynaptic membrane organization in the periactive zone (Fig. 5E,G–I) (p > 0.05; two-tailed Student's t test). In addition, antibodies raised against the SH3 domain of another endocytic protein, endophilin, induced numerous shallow coated pits, as we reported previously (Ringstad et al., 1999).
An electron-dense cytomatrix occurred within the pool of remaining vesicles exposed to the antibodies (Fig. 5D), suggesting that they interacted with intersectin in this region and formed a complex that was detectable in the electron microscope. The dense matrix did not block fusion of vesicles, because a full depletion of vesicles and large membrane folds occurred in synapses exposed to high-frequency stimulation (supplemental Fig. 5A,B,E, available at www.jneurosci.org as supplemental material).
Previous studies have demonstrated that an actin-rich cytomatrix is present at the periactive zone (Shupliakov et al., 2002; Bloom et al., 2003). This cytomatrix is poorly preserved in osmicated material used for conventional electron microscopy. To test whether a disruption of the actin cytoskeleton might have occurred, microinjections of GST–SH3C domain and LIS-AC antibodies were followed by fluorescently labeled phalloidin. Contrary to the control synapses, in which phalloidin accumulated in ring-like structures, many of the phalloidin rings were fragmented in the axons injected with the reagents and had a larger diameter (supplemental Fig. 3, available at www.jneurosci.org as supplemental material) (p < 0.01; two-tailed Student's t test; n = 21).
Together, the results from domain and antibody injections allow us to conclude that acute perturbation of intersectin–dynamin interactions in a living vertebrate synapse perturbs the organization of the periactive zone and interferes with the recycling of synaptic vesicle membrane (Figs. 4H, 5J, schematics).
Morphology of clathrin-coated pits and immunolocalization of dynamin after perturbation of intersectin–dynamin interactions
The most striking effect of microinjections of intersectin domains and antibodies was a 13-fold increase in clathrin-coated pits compared with uninjected control synapses (Figs. 4F, 5H). The majority (90%) of clathrin-coated pits were trapped at a stage of constriction and contained electron-dense collars (Fig. 6A,B), compared with 10% in uninjected control synapses (Fig. 6A,B) (IgG, 2%; GST, 3%; SH3B, 3%) (supplemental Figs. 2F, 4, available at www.jneurosci.org as supplemental material). To investigate whether this was caused by a perturbation of the dynamin-based fission mechanism, we first studied the endocytic intermediates induced by SH3C and LIS-AC antibody injections at high resolution. Interestingly, the collars observed in synapses exposed to GST–SH3C domain or LIS-AC antibodies (Fig. 6C,D) were thinner than those found on constricted coated pits in noninjected control axons or on intermediates trapped by the microinjection of GTPγS (Fig. 6E,F), suggesting that some components in the fission complex might be missing. To investigate this further, dynamin and amphiphysin, two key components of the fission complex, were immunolocalized at these endocytic intermediates using preembedding immunocytochemistry in microinjected axons cut in two halves after fixation (Fig. 7, supplemental Fig. 6, available at www.jneurosci.org as supplemental material). A twofold reduction of dynamin labeling at the fission complex region of endocytic intermediates trapped by GST–SH3C was observed, compared with that of control (Fig. 7D) (p < 0.05; two-tailed Student's t test), whereas amphiphysin labeling was not significantly changed (Fig. 7D) (p > 0.05; two-tailed Student's t test). This was not attributable to an overall reduction of dynamin recruitment to the endocytic zone, because the total dynamin labeling per coated pit was similar to the control (Fig. 7D) (p > 0.05; two-tailed Student's t test). Surprisingly, a fourfold increase in labeling at the fission complex region of intermediates trapped by LIS-AC microinjection occurred (Fig. 7D) (p < 0.0001; two-tailed Student's t test), along with a small increase in amphiphysin labeling (Fig. 7D) (p < 0.01; two-tailed Student's t test). This was accompanied by an overall increase in dynamin labeling around coated pits (Fig. 7D) (p < 0.0001; two-tailed Student's t test), demonstrating that more dynamin was detectable in the endocytic zone after intersectin antibody microinjection. There was no significant change in amphiphysin labeling per coated pit in these experiments compared with uninjected control (Fig. 7D) (p > 0.05; two-tailed Student's t test). A dramatic increase in dynamin immunoreactivity around coated intermediates after perturbation of intersectin SH3 domain interactions indicates that intersectin negatively controls dynamin recruitment to the periactive zone (Fig. 8). This increase in dynamin level does not, however, lead to efficient fission. This further implies that a mechanism that actively directs dynamin to the neck region of the clathrin-coated pit may exist, which becomes inhibited when the intersectin–dynamin interaction is acutely perturbed.
Previous studies in Drosophila have revealed an essential role for intersectin in the regulation of synaptic vesicle recycling. These studies suggested that it is an integral component of the periactive zone and that it functions as a molecular scaffold to concentrate endocytic proteins (Roos and Kelly, 1999; Koh et al., 2004; Marie et al., 2004). In agreement with this function, intersectin deletion mutants displayed a dramatic reduction in several proteins involved in clathrin-mediated endocytosis, including dynamin, endophilin, synaptojanin, and AP180 (adaptor protein 180) (Koh et al., 2004). Our present study shows that intersectin has a dynamic localization in the lamprey giant synapse and that its function is not restricted to the periactive zone. It relocates to endocytic sites after stimulation and is localized to the synaptic vesicle cluster at rest. Our data indicate that during its cycle, intersectin controls the function of its major binding partner, dynamin, at two different steps. First, it functions as a negative regulator of dynamin redistribution from the vesicle cluster to the periactive zone, and second, it may facilitate the function of dynamin during fission of clathrin-coated vesicles.
Intersectin and dynamin interact with each other in vitro and colocalize in the vesicle cluster at rest. This strongly suggests that they form a complex in this synaptic compartment (Fig. 8A). Imaging experiments show that intersectin antibodies specifically accumulate at vesicle clusters in living synapses at rest. We suggest that they efficiently compete with the binding to dynamin, in agreement with our in vitro experiments. This competition would cause a disruption of dynamin–intersectin interactions in the cluster, creating a pool of dynamin not bound to intersectin. A dramatic increase in dynamin immunoreactivity at periactive zones occurred after antibody microinjection, suggesting that this pool of dynamin is relocated to the periactive zone in a nonregulated manner. Together, these data indicate that intersectin negatively regulates dynamin redistribution to the periactive zone (Fig. 8B). This function was not affected by microinjection of the SH3C domain alone (Fig. 8C). In agreement with these results, synapses in Drosophila intersectin-null mutants displayed significantly lower levels of dynamin than wild type, supporting the idea that intersectin is involved in control of the amount of dynamin circulating at synapses (Koh et al., 2004).
During stimulation, both proteins reach the periactive zone (Fig. 8B). This migration was not affected by the antibody, thus indicating that dynamin and intersectin are recruited via independent, parallel pathways. It has been shown that the targeting of intersectin to the coat occurs via its Eps15 interaction (Hussain et al., 2001). Several SH3 domain-containing proteins have been proposed to perform this function for dynamin (e.g., intersectin, amphiphysin, syndapin, and cortactin) (Shupliakov et al., 1997; Merrifield et al., 2005; Anggono et al., 2006). Although it remains to be clarified which of these SH3 domain proteins delivers dynamin to the periactive zone, it is evident that the delivery can be efficiently performed even when the intersectin–dynamin interaction is perturbed.
In the present study, we show that intersectin–dynamin interactions are important during fission in a vertebrate synapse. In agreement with previous studies in which dynamin-interacting SH3 domains were overexpressed in fibroblasts (Simpson et al., 1999), microinjection of the SH3C domain of intersectin in living reticulospinal synapses resulted in the inhibition of clathrin-mediated endocytosis. This inhibition was accompanied by a reduction in dynamin immunoreactivity in the neck region of constricted coated pits. Surprisingly, an increase in dynamin levels around the necks of clathrin-coated pits after microinjection of intersectin antibodies was observed. Despite the presence of dynamin around the necks of coated pits, the fission reaction was inhibited. In vitro experiments with purified dynamin have demonstrated that the GTPase activity during oligomerization is strongly enhanced by an increase in dynamin concentration (Hinshaw, 2000). It cannot be excluded that intersectin antibodies accumulated at necks of constricted coated intermediates might create steric hindrance for dynamin assembly. Previous experiments, however, show that antibodies that interact with endocytic proteins localized to clathrin-coated intermediates, synaptojanin, and amphiphysin do not block fission (Haffner et al., 1997; Gad et al., 2000; Evergren et al., 2004b). We hypothesize that an active, defined positioning of dynamin, as well as of other components of the fission complex, is required for efficient membrane severing in vivo. Several SH3 domain-containing proteins that are present in the fission complex have been implicated in this function (e.g., amphiphysin and endophilin) (Shupliakov et al., 1997; Gad et al., 2000). Our data indicate that intersectin is an important regulatory protein in this process. Spirals, observed at the necks of coated pits, were thinner after perturbation of intersectin SH3 domain interactions than those observed after microinjection of GTPγS, suggesting that components of the fission complex were missing. In agreement with this finding, thin collars at the necks of coated pits were also observed in Drosophila NMJs in intersectin-null mutants (Koh et al., 2004). Perturbation of other SH3 domain-containing proteins (e.g., amphiphysin and endophilin) (Shupliakov et al., 1997; Gad et al., 2000) also resulted in an inhibition of fission. These intermediates had different morphological appearances from those observed after the perturbation of intersectin SH3 domain interactions, thus indicating that intersectin is involved at a step of the fission reaction distinct from those including amphiphysin and endophilin.
Intersectin interacts with several important proteins regulating actin polymerization (e.g., N-WASP and Cdc42) (Roos and Kelly, 1998; Hussain et al., 2001). Recent studies have shown that dynamin also interacts with proteins that function in actin regulation, such as cortactin and syndapin (Orth and McNiven, 2003; Merrifield et al., 2005; Anggono et al., 2006). In synapses at which intersectin–dynamin interactions were perturbed, we observe a reorganization of the actin matrix in the periactive zone. Thus, we suggest that the membrane infolds observed are a consequence of a loss of membrane rigidity normally provided by the cortical actin cytoskeleton. This notion is in agreement with data obtained from other model systems (Trifaro and Vitale, 1993; Qualmann et al., 2000; Itoh et al., 2005). It cannot be excluded that the actin reorganization observed after perturbation of intersectin interactions also reduces the efficiency of membrane scission. It has been suggested that polymerization of actin is important for creating a tension required for the dynamin-mediated scission of the membrane (Itoh et al., 2005; Roux et al., 2006).
Together, our data show that the release of dynamin to the periactive zone is regulated by intersectin interactions and that both proteins redistribute to the periactive zone, in which intersectin enhances dynamin-mediated fission. Our results demonstrate that the synaptic vesicle pool serves as an important compartment concentrating key components of the endocytic machinery, which relocate to the periactive zone during synaptic activity.
This work was supported by grants from the Swedish Research Council (13473), Svenska Läkaresällskapet, the Swedish Royal Academy, the Wallenberg Foundation, and Erik Fernström's Foundation (O.S.). H.G. was supported by the Swedish Brain Foundation, N.T. received support from the Russian Foundation for Basic Research (05-04-08973), and K.W. received stipends from the Wenner-Gren Foundation and the Deutsche Forschungsgemeinschaft. We thank Drs. H. T. McMahon, P. McPherson, and T. Südhof for the generous gift of intersectin and Eps15 antibodies, Drs. L. Brodin and P. DeCamilli for the generous gift of dynamin and amphiphysin antibodies, and Dr. P. Greengard for the random-primed cDNA library and synapsin antibodies. We thank Dr. P. Löw for expert advice and Drs. H. Bellen, H. T. McMahon, P. DeCamilli, S. Schmid, and D. Parker for discussions and critical comments.
- Correspondence should be addressed to Dr. Oleg Shupliakov, Department of Neuroscience, Karolinska Institutet, 171 77 Stockholm, Sweden.