Abstract
Sustained neuronal communication relies on the coordinated activity of multiple proteins that regulate synaptic vesicle biogenesis and cycling within the presynaptic terminal. Synaptogyrin and synaptophysin are conserved MARVEL domain-containing transmembrane proteins that are among the most abundant synaptic vesicle constituents, although their role in the synaptic vesicle cycle has remained elusive. To further investigate the function of these proteins, we generated and characterized a synaptogyrin (gyr)-null mutant in Drosophila, whose genome encodes a single synaptogyrin isoform and lacks a synaptophysin homolog. We demonstrate that Drosophila synaptogyrin plays a modulatory role in synaptic vesicle biogenesis at larval neuromuscular junctions. Drosophila lacking synaptogyrin are viable and fertile and have no overt deficits in motor function. However, ultrastructural analysis of gyr larvae revealed increased synaptic vesicle diameter and enhanced variability in the size of synaptic vesicles. In addition, the resolution of endocytic cisternae into synaptic vesicles in response to strong stimulation is defective in gyr mutants. Electrophysiological analysis demonstrated an increase in quantal size and a concomitant decrease in quantal content, suggesting functional consequences for transmission caused by the loss of synaptogyrin. Furthermore, high-frequency stimulation resulted in increased facilitation and a delay in recovery from synaptic depression, indicating that synaptic vesicle exo-endocytosis is abnormally regulated during intense stimulation conditions. These results suggest that synaptogyrin modulates the synaptic vesicle exo-endocytic cycle and is required for the proper biogenesis of synaptic vesicles at nerve terminals.
Introduction
Synaptogyrin and synaptophysin are evolutionarily related members of the MARVEL domain family, which defines a group of tetraspanning membrane proteins involved in vesicle trafficking and membrane apposition events (Hübner et al., 2002; Sánchez-Pulido et al., 2002). The two proteins are abundantly expressed on synaptic vesicles, with synaptophysin alone constituting ∼10% of synaptic vesicle protein by mass in rats (Takamori et al., 2006). Synaptophysin can oligomerize into homomultimers and has been proposed to form an ion channel or proteinaceous fusion pore (Rehm et al., 1986; Thomas et al., 1988; Gincel and Shoshan-Barmatz, 2002; Arthur and Stowell, 2007). Synaptophysin also has been implicated in regulating synaptic vesicle exocytosis, perhaps via its interaction with synaptobrevin/VAMP2, which precludes the association of synaptobrevin with other members of the SNARE complex (Edelmann et al., 1995).
Synaptophysin and synaptogyrin family members may also play a role in shaping lipid membranes and promoting vesicle biogenesis. Synaptophysin binds cholesterol and promotes the formation of highly curved membranes (Thiele et al., 2000). Overexpression of cellugyrin, a non-neuronal paralog of synaptogyrin, induces the formation of synaptic-like microvesicles in neuroendocrine cells (Belfort et al., 2005). Synaptophysin may also participate in synaptic vesicle recycling through a Ca2+-dependent interaction with dynamin, as the inhibition of this interaction results in a smaller synaptic vesicle pool following high-frequency stimulation (Daly et al., 2000; Daly and Ziff, 2002). Similarly, a recent study demonstrated the kinetics of endocytosis is slowed in cultured neurons during periods of sustained stimulation in the absence of synaptophysin (Kwon and Chapman, 2011).
Although synaptophysin and synaptogyrin are evolutionarily conserved and abundant in neurons, the loss of these proteins results in relatively mild phenotypes. Initial characterization of a Caenorhabditis elegans triple knock-out of synaptophysin, synaptogyrin, and the tetraspanning membrane protein SCAMP revealed no detectable phenotypes in addition to a slight increase in clathrin-coated synaptic vesicles (Abraham et al., 2006). Closer examination of the C. elegans synaptogyrin mutant identified slightly altered motility and changes in sensitivity to drugs affecting GABAergic and cholinergic signaling (Abraham et al., 2011). Similarly, synaptophysin knock-out mice, while viable and fertile, have mild defects, including altered synaptic vesicle density and morphology in photoreceptors that lack the synaptophysin paralog synaptoporin (McMahon et al., 1996; Spiwoks-Becker et al., 2001). Synaptophysin/synaptogyrin double knock-out mice also display defects in paired-pulse facilitation, post-tetanic potentiation, and long-term potentiation (Janz et al., 1999).
The genetic analysis of synaptophysin and synaptogyrin in mice is complicated by the presence of multiple homologs with some degree of functional redundancy (Janz et al., 1999; Spiwoks-Becker et al., 2001). We therefore sought to investigate synaptogyrin function in Drosophila, which lacks a synaptophysin homolog and has only one synaptogyrin family member. The loss of synaptogyrin in Drosophila leads to changes in synaptic vesicle size and delays in resolving endocytic cisternae, as well as alterations in quantal size and evoked release during high-frequency stimulation.
Materials and Methods
Homology searches and protein alignment.
NCBI BLAST (blastp) was used to identify potential homologs, which were confirmed using a reciprocal blastp search (http://blast.ncbi.nlm.nih.gov). TMHMM Server v. 2.0 (http://www.cbs.dtu.dk/services/TMHMM) was used to verify prospective homologs had four transmembrane domains in the correct orientation (cytoplasmic N and C termini). The synaptogyrin protein alignment was generated using the T-Coffee algorithm (http://tcoffee.org) with default settings (Notredame et al., 2000; Di Tommaso et al., 2011). CLC DNA Workbench 4.0 was used to visualize the alignment. Synaptogyrin GenBank accession numbers used for Figure 1: Drosophila melanogaster, AAF58329; Homo sapiens, EAW60323 (isoform 1a); Monosiga brevicollis, EDQ85723.
Fly stocks and transgenics.
Flies were cultured using standard media and techniques at room temperature (∼22°C) unless otherwise noted. The GAL4/UAS system was used to drive neuronal expression of selected transgenes using the driver elavc155-GAL4 (Brand and Perrimon, 1993). The UAS-synaptogyrin-GFP (gyrin-GFP) line was generated by subcloning full-length Drosophila synaptogyrin cDNA (minus the stop codon) into a pUAST vector containing a C-terminal GFP tag. This construct was then injected into w1118 embryos at Duke University Model Systems Genomics (Durham, NC). The UAS-myc-synaptogyrin construct used for rescue experiments was constructed by inserting full-length synaptogyrin cDNA into a modified pValum vector containing an N-terminal myc tag (Cho et al., 2010). This construct was injected into a strain carrying the attP2 site (yv;;attP2) for targeted transgene insertion on the third chromosome (Markstein et al., 2008; Ni et al., 2008) at Genetic Services.
Synaptogyrin antibody generation and Western blotting.
The C-terminal fragment of Drosophila synaptogyrin (gyrin-cterm) encoding amino acids 181–241 was amplified from a Drosophila cDNA library and subcloned into the pGEX-4T-1 vector (GE Healthcare). Recombinant GST-gyrin-cterm was expressed and processed in Escherichia coli (BL21) using standard protocols. The fusion protein was purified from cell lysates using Glutathione Sepharose 4B (GE Healthcare) and used to immunize rabbits for polyclonal antibody production (Invitrogen). Western analysis was performed using standard protocols. Primary antibody dilutions were as follows: synaptogyrin, 1:20,000 and complexin, 1:5000 (Huntwork and Littleton, 2007). IRDye 800-conjugated goat anti-rabbit secondary antibodies were used at a dilution of 1:10,000 (LI-COR Biosciences). Western blots were visualized with an Odyssey infrared imaging system (LI-COR Biosciences).
Generation of a synaptogyrin-null mutant.
Two independent deletions of the synaptogyrin locus were generated by imprecise excision of a P-element (P{lacW}l(2)SH0644SH0644) located ∼500 bp upstream of the synaptogyrin translation start site in the first exon of the gene. Two hundred white− excision events were screened by PCR and two resulted in deletions that extended into the synaptogyrin locus beyond the translation start site. A precise excision line isolated from the screen, gyrPE, was confirmed by sequencing and used as a control for genetic background in all experiments. To confirm that the deletions in gyr1 and gyr2 did not generate truncated versions of synaptogyrin, Western blots were performed on protein extracts from these lines. The polyclonal antibody raised against the synaptogyrin protein targets the C terminus, and this coding region was left intact in both alleles. Synaptogyrin immunoreactivity was completely absent in gyr1 and gyr2 animals by Western blot, suggesting these represent null mutations. gyr1 and gyr2 are in the white (w1118) background unless otherwise indicated.
Immunohistochemistry.
Wandering third instar larvae were dissected in Ca2+-free HL3.1 saline (70 mm NaCl, 5 mm KCl, 10 mm NaHCO3, 4 mm MgCl2, 5 mm trehalose, 115 mm sucrose, and 5 mm HEPES, pH 7.2) and fixed for 45 min in HL3.1 containing 4% formaldehyde. Fixations using the synaptogyrin antibody (α-Gyr) were performed for 5 min in ice-cold 100% methanol. Primary antibody dilutions were as follows: synaptogyrin, 1:500; Discs large (Dlg), 1:250; nc82/bruchpilot, 1:100; synaptotagmin 1, 1:200 (3H2 2D7, from K. Zinn). Secondary antibodies were used at a dilution of 1:250 and include Cy2-conjugated goat anti-rabbit, Cy3-conjugated goat anti-rabbit, Rhodamine Red-conjugated donkey anti-mouse (Jackson ImmunoResearch), and Alexa Fluor 488-conjugated goat anti-mouse (Invitrogen). Goat α-horseradish peroxidase (HRP) antibodies conjugated to either fluorescein isothiocyanate or Rhodamine Red were added as secondary antibodies and were used at a concentration of 1:10,000 or 1:500, respectively. Images were acquired with an Axoplan 2 confocal microscope using PASCAL software (Carl Zeiss MicroImaging).
Bouton and active zone counting.
Age-matched wandering third instar larvae were grown at low density at 25°C and immunohistochemistry was performed as described above. Bouton quantification was performed using antibodies against HRP and Dlg. Active zone number was quantified using monoclonal nc82 antibodies, which recognize the active zone component bruchpilot (Wagh et al., 2006). To control for staining variability, larvae of different genotypes were processed together. All bouton and active zone measurements were quantified at muscle 6/7 of segment A3. Image analysis was performed using ImageJ software (National Institutes of Health; NIH). Bouton number is known to be proportional to muscle size (Lnenicka and Keshishian, 2000); however, since muscle area did not differ significantly between gyr and control larvae, no corrections were necessary. Bouton and active zone quantification were performed blind to genotype.
Electron microscopy.
Dissected Drosophila third instar larvae were fixed for 1 h in 4% paraformaldehyde/1% glutaraldehyde in 0.1 m cacodylate buffer, pH 7.2. For the indicated experiments, preparations were stimulated before fixation with 90 mm high K+ solution or with 10 consecutive stimulation trains at a frequency of 100 Hz. Each pulse was 1 s in duration and was followed by a 1 s rest period. Preparations were fixed immediately after stimulation. Fixed larvae were washed in standard HL3 solution for 1 h and postfixed for 40 min in 1% osmium tetroxide. Specimens were dehydrated in a graded series of ethanol and water mixtures up to 100% ethanol followed by acetone. Samples were pre-infiltrated for 1 h in a 1:1 mixture of acetone and Epon and then embedded in fresh Epon overnight at 60°C. Thin sections (60–70 nm) were contrasted with 0.2% lead citrate and imaged at 80 kV on an FEI Tecnai G2 Spirit transmission electron microscope (TEM) equipped with an AMT CCD camera. Imaging was performed at the W.M. Keck Microscopy Facility at the Whitehead Institute. Quantification of synaptic vesicle diameter and density was performed using randomly sampled images from type Ib boutons with a diameter >2 μm (to ensure the exclusion of type Is boutons). The diameter of synaptic vesicles and cisternae were measured from the outside of the vesicle membrane along the long axis. A vesicle was measured only if its borders were well defined and it had a clear center (i.e., it was not a dense-core vesicle). Structures larger than 60 nm in diameter were excluded from calculations of mean synaptic vesicle diameter. High K+ stimulation was performed using a modified Jan and Jan solution (Jan and Jan, 1976; 45 mm NaCl, 90 mm KCl, 36 mm sucrose, 2 mm CaCl2, 2 mm MgCl2, and 5 mm HEPES, pH 7.3). Cisternae were defined as intracellular single-membrane structures with a diameter greater than or equal to 80 nm. In certain cases, the exact size of a cisterna could not be measured; however, if a cisterna clearly was >80 nm in diameter, it was included in the analysis of the number of cisternae per unit area (see Figs. 6, 8). The analysis was done blind to genotype whenever possible. Measurements were performed using ImageJ (NIH) and Adobe Photoshop (Adobe Systems) software.
FM1–43 staining.
Wandering third instar larvae were dissected in HL3.1 solution with 1 mm CaCl2 and the motor nerves were severed to prevent spontaneous muscle contractions. Larvae were stimulated for 5 min with high K+ (90 mm) Jan and Jan solution containing 4 μm FM1–43 (Invitrogen). Preparations were washed in standard HL3.1 and briefly incubated in HL3.1 saline containing 100 μm Advasep-7 (Biotium), which has been shown to reduce background fluorescence (Kay et al., 1999). Larvae were washed again in HL3.1 and then imaged to quantify the level of FM1–43 uptake. After 10 min, the neuromuscular junction (NMJ) was imaged a second time, then immediately incubated for 1 min with high K+ (90 mm) Jan and Jan solution, washed, and imaged a final time to determine the extent of unloading. For frequency-induced FM1–43 unloading, preparations were stained with 4 μm FM1–43 as described above and stimulated for 10 min at 1 Hz. Image acquisition was performed using an UltraVIEW VoX confocal imaging system (PerkinElmer) equipped with an ImagEM camera (C9100–13; Hamamatsu) and a Yokogawa CSU-X1 spinning disk head. Images were acquired as a Z series taken at 0.3 μm steps using a 40× water-immersion objective. Mean bouton intensity was determined using Volocity 3D Image Analysis Software (PerkinElmer). Background intensity of the surrounding region was subtracted from bouton labeling.
Electrophysiology.
Postsynaptic currents were recorded at segment A3 of ventral longitudinal muscle 6 in third instar larvae using two-electrode voltage-clamp with a −80 mV holding potential in HL3.1 solution. The final Ca2+ concentration was adjusted to the desired level indicated in the text. Data acquisition and analysis were performed using Axoscope 9.0 and Clampfit 9.0 software (Molecular Devices). Quantal content was estimated by dividing the current integral of nerve-evoked currents by the current integral from individual quanta as previously described (Barber et al., 2009). Motor nerves innervating the musculature were severed and placed into a suction electrode so action potential stimulation could be applied at the indicated frequencies using a programmable stimulator. Statistical analysis was performed using GraphPad Prism for Mac OS X version 5.0a and Origin Software (OriginLab). Statistical significance was determined using two-tailed Student's t tests or one-way ANOVA and noted as follows: *p < 0.05; **p < 0.01; ***p < 0.001. For all data, error bars indicate SEM.
Results
Evolutionary analysis of synaptogyrin and synaptophysin
Synaptogyrin and synaptophysin have previously been established as evolutionarily conserved proteins (Hübner et al., 2002; Sánchez-Pulido et al., 2002; Abraham et al., 2006). We were interested in examining the emergence of these and several additional MARVEL domain-containing proteins in relation to other synaptic vesicle components. While many synaptic vesicle proteins such as v- and t-SNAREs, the vacuolar ATPase, and Rab 3 are ancient eukaryotic proteins, several key proteins involved in neurotransmitter exocytosis emerged more recently in evolution (Srivastava et al., 2010). For example, synaptotagmin 1 and complexin homologs are absent in the yeast Saccharomyces cerevisiae and the amoeba Dictyostelium discoideum, but are present in the placozoan Trichoplax adhaerens, a simple multicellular organism that lacks a nervous system, but has several organized cell layers (Srivastava et al., 2008; Barber et al., 2009). A reciprocal protein-protein BLAST search revealed synaptogyrin and synaptophysin orthologs in the T. adhaerens genome, indicating that they too evolved before the emergence of defined synapses (Table 1). Interestingly, while the S. cerevisiae genome lacks synaptogyrin or synaptophysin orthologs, we were able to identify a putative synaptogyrin family member in the unicellular choanoflagellate M. brevicollis with 33% identity to human synaptogyrin 1 and four predicted transmembrane domains (King et al., 2008). Our BLAST search did not reveal a M. brevicollis synaptophysin homolog, raising the possibility that the synaptogyrin and synaptophysin families (gyrins and physins, respectively) are descended from an ancestral synaptogyrin-like protein. Since choanoflagellates are thought to be the closest extant unicellular relative of metazoans (King et al., 2008), the presence of a synaptogyrin-like ortholog in M. brevicollis but not in yeast suggests a unique role for synaptogyrin in early metazoan evolution. This putative M. brevicollis synaptogyrin ortholog is almost entirely comprised of the MARVEL domain, with its short N and C termini each predicted to be <15 aa in length. Therefore, it is likely that the MARVEL domain itself has an important cellular function and that the N and C termini of gyrins and physins may have elaborated and adapted over the course of evolution to perform a variety of specialized tasks in other organisms. Interestingly, we were unable to identify a gyrin ortholog in the sea anemone Nematostella vectensis, but did uncover at least one physin ortholog, suggesting that some metazoans do not require both proteins or that the functions of physins and gyrins are interchangeable (Putnam et al., 2007).
Evolutionary conservation of selected MARVEL domain and SCAMP proteins
We also searched for homologs of several other MARVEL domain-containing proteins and found that some, including MAL (myelin and lymphocyte protein), MYADM (myeloid-associated differentiation marker gene), and occludin (a component of tight junctions), are absent from invertebrate genomes. This lack of conservation is unsurprising given that vertebrate myelin and the myelin-like sheaths found in some invertebrates are hypothesized to have arisen through convergent evolution (Hartline and Colman, 2007). Similarly, invertebrates do not have tight junctions, but rather have analogous structures known as septate junctions (Furuse and Tsukita, 2006). We also examined the CMTM family (CKLF-like MARVEL transmembrane domain-containing family, where CKLF stands for chemokine-like factor), a novel group of proteins with a highly conserved member named CMTM4 that has been implicated in regulating the cell cycle and cellular growth (Plate et al., 2010). CMTM4 homologs are found in Drosophila, C. elegans, N. vectensis, and potentially in T. adhaerens, although this putative homolog has only three predicted transmembrane domains. Finally, we investigated the evolutionary conservation of the secretory carrier-associated membrane proteins (SCAMPs), which lack the MARVEL domain, but have the same tetraspanning membrane topology (Hubbard et al., 2000). Interestingly, SCAMP homologs are also found in the plant kingdom, suggesting they perform a more ubiquitous role in multicellular organisms (Fernández-Chacón and Südhof, 2000).
The D. melanogaster genome encodes a single synaptogyrin homolog (CG10808) that is 42% identical to human synaptogyrin 1 (Fig. 1). Unlike many other invertebrates, including C. elegans and the red flour beetle Tribolium castaneum, all of the sequenced Drosophila species lack a synaptophysin ortholog. However, the D. melanogaster genome does encode other MARVEL domain-containing proteins, including a CMTM4 ortholog (CG15211) and Singles Bar (CG13011), a protein involved in myoblast fusion (Estrada et al., 2007). In C. elegans, synaptophysin expression is largely restricted to muscle cells in the pharynx and anal sphincter (Abraham et al., 2006), while synaptogyrin is expressed in most neurons (Nonet, 1999; Abraham et al., 2011), indicating that synaptogyrin is likely to be the predominant synaptic vesicle MARVEL protein in nematodes.
Protein sequence alignment of synaptogyrin homologs. Transmembrane (TM) domains are indicated for Drosophila synaptogyrin (bars above the alignment) and M. brevicollis synaptogyrin (bars below the alignment). Conserved residues are indicated in black. Sequence abbreviations: D.m., Drosophila melanogaster; H.s., Homo sapiens; M.b., Monosiga brevicollis.
Synaptogyrin is not an essential protein in Drosophila
Synaptogyrin has not previously been characterized in Drosophila. Therefore, we examined the localization of endogenous synaptogyrin by generating antisera against a recombinant C-terminal fusion protein of Drosophila synaptogyrin. As expected for a presumptive synaptic vesicle protein, immunohistochemistry of third instar Drosophila larvae reveals abundant synaptogyrin expression throughout the brain and ventral nerve cord (Fig. 2A). Synaptogyrin is also found at the larval NMJ, where it strongly colocalizes with the synaptic vesicle protein synaptotagmin 1 (Syt1) and partially overlaps with the active zone protein bruchpilot (Fig. 2B,C). Similarly, expression of a GFP-tagged synaptogyrin transgene in neurons using the elavc155-GAL4 driver resulted in presynaptic targeting at the NMJ that overlapped with Syt1 labeling (Fig. 2D). We conclude that, like the murine and worm synaptogyrin homologs, Drosophila synaptogyrin is a presynaptic neuronal protein.
Synaptogyrin localization at synapses. A, Immunohistochemistry with antisera against the neuronal membrane marker HRP (magenta) and synaptogyrin (Gyr, green) indicates that synaptogyrin is broadly expressed in the synaptic neuropil of the larval CNS. B, Synaptogyrin (green) is expressed presynaptically at the larval NMJ where it colocalizes with synaptotagmin 1 (Syt1, magenta). C, Synaptogyrin is not confined to active zones, which are indicated by nc82 staining (magenta). D, Synaptogyrin-GFP (gyrin-GFP, green) driven by the pan-neuronal driver elavc155-GAL4 localizes to synapses. Scale bars: A, 100 μm; B, 25 μm; C, 2 μm; D, 20 μm.
We next sought to investigate the function of Drosophila synaptogyrin by generating a synaptogyrin (gyr)-null mutant. We isolated two independent partial deletions of the synaptogyrin genomic locus via imprecise excision of a P-element (P{lacW}l(2)SH0644SH0644) located just upstream of the synaptogyrin translation start site. The first deletion, gyr1, extends 2.5 kb into the synaptogyrin locus and removes the first two exons and a portion of the third exon, while the second deletion, gyr2, is a smaller 1.7 kb deletion that removes the first exon (Fig. 3A). A precise excision line, gyrPE, was chosen to serve as a control for genetic background in all experiments. The extent of the deletions and the precise excision event were confirmed by PCR and sequencing. Western analysis revealed the complete absence of synaptogyrin immunoreactivity in both gyr1 and gyr2 animals (Fig. 3B). Similarly, synaptogyrin staining is absent at the larval NMJ in mutant animals (Fig. 3C). The antibody raised against synaptogyrin targets the C terminus of the protein that is left intact in both deletions, indicating that a truncated version of the protein is not produced. We conclude that gyr1 and gyr2 represent null mutations in the synaptogyrin locus. Unless otherwise noted, gyr1 animals were used in all experiments. In agreement with mouse and C. elegans knock-outs, Drosophila gyr mutant animals are viable, fertile, and appear behaviorally normal. They develop at a similar rate and have a normal life span compared with control animals (data not shown). Motor function and coordination appear to be unaffected at both the larval and adult stages. Additionally, gyr males are capable of performing all aspects of the Drosophila courtship ritual and display a normal courtship index (data not shown). Therefore, we conclude that synaptogyrin is not essential for viability, fertility, or basic motor functions in Drosophila.
Generation of a Drosophila synaptogyrin (gyr) mutant. A, The synaptogyrin locus is diagrammed with the two neighboring genes (CG30484 and CG6357). The location of the P-element used for the excision screen is indicated by the pink triangle. Two separate deletions were isolated and are indicated by black lines. The green arrows mark the locations for the PCR primers used to determine the extent of each deletion. B, Western blot analysis of synaptogyrin protein expression levels from homogenates made from white, gyrPE, gyr1, and gyr2 adult heads. Complexin (Cpx) immunoreactivity was used as a loading control. Synaptogyrin immunoreactivity is completely absent in gyr1 and gyr2. C, Immunohistochemistry at third instar larval NMJs confirms that synaptogyrin (red) is absent in the gyr1 mutant but not in the control (gyrPE). Synaptic varicosities were identified using α-HRP antibodies (green).
gyr mutants have normal synaptic growth and synaptic architecture
As Drosophila proceed through the larval stages, the surface area of the body wall muscles increases ∼100-fold, and synaptic innervation at the NMJ increases in parallel through the addition of new boutons to maintain proper muscle depolarization. Disruptions in trans-synaptic growth signaling pathways (e.g., Wnt or TGFβ) or alterations in synaptic activity levels can lead to synaptic undergrowth or overgrowth (Budnik et al., 1990; Guan et al., 2005; Collins and DiAntonio, 2007). Furthermore, many Drosophila endocytic and membrane recycling mutants, including rab11, endophilin, synaptojanin, and dap160, display an increase in supernumerary or “satellite” boutons (Koh et al., 2004; Dickman et al., 2006; Khodosh et al., 2006). At gyr NMJs, overall synaptic morphology is normal with no obvious overgrowth, undergrowth, or satellite bouton phenotypes. The total number of synaptic boutons at muscle 6/7 of segment A3 is unchanged in both gyr1 and gyr2 mutants compared with controls (Fig. 4A,B; p = 0.48, one-way ANOVA). There also is no significant difference in muscle size, consistent with our previous observation that gyr mutants do not have developmental delays (control = 56,000 ± 1800 μm2, n = 22; gyr1 = 61,000 ± 2200 μm2, n = 23; gyr2 = 59,000 ± 2000 μm2, n = 20; p = 0.20, one-way ANOVA). Similarly, the number of active zones as defined by nc82 staining is unchanged in gyr1, with no noticeable alterations in active zone size or spacing (Fig. 4C,D; control = 529 ± 24.4, n = 14; gyr = 510 ± 28.7, n = 14, p = 0.61). We conclude that the loss of synaptogyrin in Drosophila does not significantly alter synaptic growth or influence the number of synaptic vesicle release sites.
The number of synaptic varicosities and active zones is unchanged in gyr mutants. A, Synaptic bouton number and morphology at muscle 6/7 in body segment A3 are normal in gyr1 and gyr2 third instar larvae relative to controls (gyrPE), as determined by immunostaining against HRP (magenta) and Discs large (Dlg), a postsynaptic scaffolding protein (green). B, Quantification of total bouton number at muscle 6/7 revealed no significant difference between control, gyr1, and gyr2 (p = 0.48, one-way ANOVA). Average total bouton number ± SEM: control = 70.8 ± 2.6, n = 23; gyr1 = 66.9 ± 2.8, n = 22; gyr2 = 65.2 ± 4.5, n = 21. C, Active zone number is not significantly different in gyr compared with controls (p = 0.61). Average active zone number ± SEM: control = 529 ± 24.4, n = 14; gyr = 510 ± 28.7, n = 14. D, Representative images of control and gyr NMJs stained with the active zone marker nc82. Boxed areas are shown at higher magnification. Scale bars: A, 40 μm; D, 20 μm; D (inset), 2.5 μm.
Ultrastructural analysis reveals alterations in synaptic vesicle diameter and density
Although bouton morphology appears normal at the light microscopy level in gyr mutants, ultrastructural analysis by TEM of type Ib boutons reveals several abnormalities in gyr animals. In particular, boutons at the larval NMJ in gyr animals display variable changes in synaptic vesicle diameter and density. In control boutons, synaptic vesicles are uniform in size and cluster around the periphery of the bouton (Fig. 5A,B). Synaptic vesicle diameter and density in some gyr boutons appear similar to controls (Fig. 5E). However, other gyr boutons (∼30–40%) have an increase in the number of large synaptic vesicles, the largest of which may represent endosomes or endocytic cisternae (Fig. 5C,D). This dramatic increase in the variability of synaptic vesicle diameter in certain gyr boutons is illustrated in Figure 5D (compare Fig. 5B). A few (∼10%) gyr boutons have a marked decrease in synaptic vesicle density (Fig. 5F), and while many gyr boutons appear normal, overall there is a slight but statistically significant decrease in the density of synaptic vesicles (Fig. 5G; control = 131.7 ± 9.4 vesicles/μm2, n = 21; gyr = 101.5 ± 9.5 vesicles/μm2, n = 21; p = 0.03). In contrast, there is a statistically significant increase in mean synaptic vesicle diameter in gyr boutons compared with controls (Fig. 5H; control = 42.7 ± 0.50 nm, gyr = 45.2 ± 0.93 nm; p = 0.02). There is also an overall shift in the distribution of synaptic vesicle diameter toward larger values, with an increased variability in diameter, suggesting synaptogyrin directly or indirectly regulates synaptic vesicle size (Fig. 5I).
Ultrastructural analysis reveals alterations in synaptic vesicle diameter in gyr mutants. A, Representative image of a type Ib synaptic bouton from a control larva. B, The boxed region in A is shown at higher magnification to illustrate homogeneity of synaptic vesicle diameter in control animals. C, Example of a gyr bouton with abnormal synaptic vesicle diameter and large cisternal-like structures (indicated by asterisk). D, The boxed region in C is shown at higher magnification to illustrate the variability in synaptic vesicle diameter (see asterisk). E, Representative gyr bouton with similar synaptic vesicle diameter and density relative to controls. F, Example of a gyr bouton with decreased synaptic vesicle density. G, Synaptic vesicle density is significantly decreased in gyr mutants compared with controls (p = 0.03). Each point on the graph represents the synaptic vesicle density of a single bouton. Average synaptic vesicle density per bouton area (μm2) ± SEM: control = 131.7 ± 9.4, n = 21; gyr = 101.5 ± 9.5, n = 21; p = 0.03. H, The average synaptic vesicle diameter (vesicles < 60 nm in diameter) is increased in gyr mutants relative to controls (p = 0.02). Each data point represents the average synaptic vesicle diameter for a single bouton. At least 20 synaptic vesicles were measured in each bouton to obtain the mean synaptic vesicle diameter. Average synaptic vesicle diameter (in nanometers) ± SEM: control = 42.7 ± 0.50, n = 21 boutons; gyr = 45.2 ± 0.93, n = 21 boutons. I, Synaptic vesicle diameter is shifted to higher values relative to the control in gyr mutant larvae. Since varying numbers of synaptic vesicles were measured in each bouton, the frequency distribution of each bouton was calculated and normalized to obtain the overall frequency distribution. Scale bars: A, C, E, F, 1 μm; B, D, 100 nm.
gyr mutants retain endocytic cisternae following intense stimulation
Synaptic vesicles are known to form through several pathways, including traditional clathrin-mediated endocytosis from the plasma membrane (De Camilli and Takei, 1996) and via endosomal intermediates (Heuser and Reese, 1973; Takei et al., 1996; de Lange et al., 2003). During periods of relatively low activity, clathrin-mediated endocytosis appears to be the predominant form of synaptic vesicle retrieval at central synapses (Granseth et al., 2006). However, tetanic stimulation or intense nonphysiological stimuli can induce bulk endocytosis, a process whereby large plasma membrane invaginations are internalized to form endocytic cisternae from which synaptic vesicles then bud (Miller and Heuser, 1984; Richards et al., 2000; Richards et al., 2003; Evans and Cousin, 2007). The presence of many large, endosomal-like structures in a subset of gyr mutant boutons prompted us to test if synaptic vesicle recycling via endosomal intermediates was disrupted in these boutons, resulting in a buildup of endocytic cisternae.
To test this hypothesis, we incubated larvae with a high K+ solution known to induce the formation of endocytic cisternae in response to strong synaptic vesicle exocytosis caused by continuous membrane depolarization (Marxen et al., 1999; de Lange et al., 2003; Akbergenova and Bykhovskaia, 2009). We then investigated whether gyr mutants were impaired in recovering vesicle membrane via bulk endocytosis or in resolving endocytic cisternae into synaptic vesicles. Under normal resting conditions, gyr mutants have a statistically significant increase in the number of cisternae per μm2 (cisternae are defined as structures with a diameter >80 nm; Fig. 6 [left]; control = 0.53 ± 0.10 cisternae/μm2, n = 21; gyr = 1.57 ± 0.33 cisternae/μm2, n = 23; p < 0.01). Immediately after a 5 min incubation in high K+ (90 mm) Jan and Jan solution (Jan and Jan, 1976; Akbergenova and Bykhovskaia, 2009), both control and gyr animals show a similar increase in the number of cisternae, suggesting that gyr mutants are not impaired in this step of bulk endocytosis (Fig. 6 [center]; control = 15.0 ± 0.62 cisternae/μm2, n = 8; gyr = 12.7 ± 2.0 cisternae/μm2, n = 5; p = 0.21). However, when larvae are subsequently allowed to recover for 10 min in normal (low K+) HL3.1 saline before fixation, gyr animals have a significantly higher amount (∼50% by area) of endocytic cisternae remaining compared with controls (Fig. 6 [right]; control = 5.35 ± 0.81 cisternae/μm2, n = 21; gyr = 8.26 ± 0.77 cisternae/μm2, n = 22; p = 0.01). We confirmed by serial EM reconstruction of several cisternae that these membrane compartments are not connected to the plasma membrane, and thus do not represent membrane infoldings (data not shown). These observations suggest that the process of resolving synaptic vesicles from endocytic cisternae is delayed in gyr mutants. To examine cisternae following intense nerve stimulation, we treated larvae with a protocol that included 10 trains at 100 Hz (each lasting 1 s with a 1 s gap between each pulse for a total of 1000 stimuli) and fixed the preparations immediately afterward for EM analysis. gyr boutons showed a significantly higher number of cisternae compared with control boutons following stimulation (control = 0.29 ± 0.17 cisternae/μm2, n = 9; gyr = 1.24 ± 0.29 cisternae/μm2, n = 15; p = 0.03).
gyr mutants are delayed in resolving endocytic cisternae following a high K+ saline shock. Left, At rest, gyr mutants have a statistically significant increase in the number of cisternae (endosomal-like structures larger than 80 nm in diameter) per μm2 compared with controls (p < 0.01). Middle, Immediately following a 5 min incubation with high K+ (90 mm) Jan and Jan solution, control and gyr boutons have a similar dramatic increase in the number of endocytic cisternae due to bulk endocytosis (p = 0.21). Right, Ten minutes after high K+ stimulation, gyr boutons have ∼50% more endocytic cisternae than controls, suggesting that they cannot resolve endocytic cisternae into synaptic vesicles as quickly as controls (p = 0.01). Arrows highlight cisternal structures. Scale bar, 500 nm.
To further examine synaptic vesicle recycling, we turned to the styryl dye FM1–43, which reversibly binds membranes and increases in fluorescence intensity upon membrane integration, allowing one to track compartments as they move through the exo-endocytic cycle (Cochilla et al., 1999). Mutations in a variety of endocytic proteins, such as endophilin, eps15, synaptojanin, and dap160, result in decreased loading of FM1–43 dye (Verstreken et al., 2002, 2003; Koh et al., 2004, 2007). When we incubated gyr and control larvae for 5 min with 90 mm K+ Jan and Jan solution containing 4 μm FM1–43, we found no significant difference in the amount of dye uptake (Fig. 7A–C; control = 13,600 ± 400 A.U.; gyr = 13,700 ± 600 A.U.; p = 0.83). This result is consistent with our EM results that demonstrate gyr boutons are capable of taking up large quantities of membrane in the form of endocytic cisternae immediately following a K+ shock (Fig. 6). We also examined the capacity of gyr and control boutons to unload dye after 10 min of rest in normal saline to determine whether exocytosis is significantly impaired. A 1 min incubation with high K+ saline resulted in a similar level of destaining relative to loading fluorescence levels (control = 31.1 ± 2.40% of loading fluorescence, n = 13 NMJs; gyr = 37.9 ± 2.50%, n = 11 NMJs, p = 0.06), indicating that gyr boutons are not defective in releasing FM1–43. Since unloading was performed with an intense, nonphysiological stimulus, this assay is poorly suited to determine whether gyr mutants sequester FM1–43 in endocytic cisternae to a greater extent than controls. The second high K+ stimulus used to unload the dye may have caused cisternae to fuse in addition to synaptic vesicles, which would mask any defect in synaptic vesicle budding from cisternae. To bypass this problem, we quantified bouton fluorescence intensity 10 min after loading during a rest period where only spontaneous vesicle release occurs, allowing a more sensitive measure for vesicle availability following intense stimulation. Under these conditions, we found that gyr boutons retain a higher percentage of dye relative to controls (Fig. 7C; control = 90.8 ± 1.83% of loading fluorescence, n = 13 NMJs; gyr = 98.9 ± 1.48%, n = 11 NMJs; p < 0.01), consistent with the observation that gyr mutants are delayed in resolving synaptic vesicles from cisternae. To directly assay whether gyr boutons are defective in vesicle availability following a milder stimulation protocol, we examined FM1–43 unloading after 10 min of stimulation at 1 Hz (Fig. 7D–F). Using this stimulation protocol for unloading, we observed a significantly higher level of fluorescence intensity retained in gyr boutons compared with control preparations (control = 34.4 ± 3.28%, n = 19 NMJs; gyr = 45.6 ± 4.47%, n = 18 NMJs; p < 0.05).
FM1–43 loading and unloading in gyr mutants. A, Schematic of the FM1–43 loading/unloading protocol. Arrows indicate points at which images were acquired. B, Representative images of control and gyr boutons from FM1–43 uptake experiments following 5 min of dye loading, 10 minutes of rest, and 1 min of dye unloading. Mean loading fluorescence intensity was not significantly different in gyr or control boutons. C, Unloading fluorescence intensity relative to loading intensity was also unchanged in gyr mutants. Following a 10 min incubation in low K+ saline, gyr boutons retained a statistically significant higher in fluorescence intensity relative to controls (control = 90.1 ± 1.83% relative to loading fluorescence, n = 13 NMJs; gyr = 98.9 ± 1.48%, n = 11 NMJs, p < 0.01). D, Schematic of the FM1–43 loading/unloading protocol. E, Representative images of control and gyr boutons after FM1–43 loading and 1 Hz frequency-induced unloading. F, A significantly higher level of fluorescence intensity is retained in gyr boutons compared with control preparations following 1 Hz frequency-induced unloading (control = 34.5 ± 3.28% relative to loading fluorescence, n = 19 NMJs; gyr = 45.6 ± 4.47% n = 18 NMJs, p < 0.05). Scale bars: 5 μm.
If synaptogyrin regulates synaptic vesicle budding from endocytic cisternae, synaptic vesicle recycling induced by high K+ stimulation might exacerbate the misregulation of synaptic vesicle size seen in gyr mutants under nonstimulated conditions. However, when we measured the diameter of synaptic vesicles 10 min after high K+ stimulation, we found that the distribution of synaptic vesicle diameter was more similar to controls compared with pre-K+ stimulation (Fig. 8A,B). The average synaptic vesicle diameter following the K+ shock was not significantly different between gyr and controls (Fig. 8C,D; control = 43.33 ± 0.58 nm, n = 18 boutons; gyr = 43.28 ± 0.49 nm, n = 20 boutons; p = 0.95), and individual gyr boutons had much less variation in their mean synaptic vesicle diameter (Fig. 8D). So while the number of large cisternae is increased in gyr mutants after intense stimulation, synaptic vesicles (<60 nm in diameter) that form soon after stimulation are more similar in size to controls (Fig. 8D). These findings suggest that intense stimulation results in the restoration of normal synaptic vesicle diameter immediately after the stimulus, likely by forcing abnormally large synaptic vesicles to fuse.
Effects of high K+ stimulation on synaptic vesicle diameter and synaptic vesicle density. A, Same as Figure 5I. At rest, gyr synaptic vesicles are larger and more variable in size. B, Following a 5 min high K+ shock and 10 min rest period, the distribution of synaptic vesicle diameter in gyr mutants is much more similar to controls. The mean synaptic vesicle diameter following the K+ shock is not significantly different in gyr mutants relative to controls (control = 43.3 ± 0.58 nm, n = 18; gyr = 43.3 ± 0.50 nm, n = 20; p = 0.95). C, A normalized cumulative frequency diagram illustrates the differences in vesicle diameter before (Pre-K+) and after (Post-K+) stimulation. Before the K+ shock, the gyr distribution is slightly shifted to the right compared with the control. Following the K+ shock, gyr and control synaptic vesicles have much more similar distributions. D, The mean synaptic vesicle diameter of gyr boutons returns to control values following high K+ stimulation. However, synaptic vesicle diameter in shits1; gyr1 larvae is significantly greater than shits1; gyrPE (shits1; gyrPE = 42.3 ± 0.58 nm, n = 14; shits1; gyr1 = 46.2 ± 0.52 nm, n = 16; p < 0.001). E, Representative images of shits1; gyrPE and shits1; gyr1 boutons 20 min after a 5 min heat shock and high K+ stimulation. F, shits1; gyrPE and shits1; gyr1 have an increase in endocytic cisternae >80 nm in diameter after 20 min of recovery. G, shits1; gyr1 boutons have a dramatic increase in the number of cisternae >200 nm in diameter (p < 0.05). Average number of cisternae >200 nm/μm2 ± SEM: control (pre-K+) = 0.042 ± 0.03, n = 21; gyr1 (pre-K+) = 0.20 ± 0.08, n = 23; control (post-K+) = 0.28 ± 0.10, n = 21; gyr1 (post-K+) = 0.43 ± 0.11, n = 22; shits1; gyrPE (post-K+) = 0.66 ± 0.22, n = 14; shits1; gyr1 (post-K+) = 1.58 ± 0.31, n = 17). H, Control and gyr boutons have similar synaptic vesicle densities after a high K+ shock, while the presence of the shibire mutation results in a significant decrease in synaptic vesicle density following synaptic vesicle depletion, with shits1; gyr1 boutons displaying significantly lower levels than shits1; gyrPE (p < 0.01).
To further investigate recycling through endosomal intermediates, we used the temperature-sensitive dynamin mutant shibire (shits1) in combination with gyr to examine the recovery of synaptic vesicles following a complete endocytic block. In shits1 larvae, synaptic vesicles are depleted and synapses accumulate numerous collared pits and membrane invaginations at restrictive temperatures. Upon subsequent recovery at the permissive temperature, cisternae are generated from which synaptic vesicles then bud (Koenig and Ikeda, 1996). We repeated the previously described high K+ incubation with shits1; gyr1 and shits1; gyrPE larvae at 32°C and allowed larvae to recover at room temperature (∼22°C) in low K+ saline before fixation for EM. Ten minutes after a 5 min simultaneous heat shock and K+ stimulation, both shits1; gyr1 and shits1; gyrPE boutons contained numerous overlapping and misshapen cisternal-like structures that were difficult to measure, therefore we extended the recovery period to 20 min. We found that shits1; gyrPE and shits1; gyr1 boutons contained a significantly greater number of cisternae compared with poststimulation non-shibire larvae (p < 0.05), suggesting that the inhibition of dynamin activity significantly impacts endocytic recovery through bulk endocytosis following K+ stimulation. Furthermore, as was the case in the absence of dynamin inhibition, shits1 larvae carrying the gyr mutation tended to have more cisternae than shits1 controls, although this difference was not statistically significant (Fig. 8E,F; shits1; gyrPE = 8.63 ± 1.2 cisternae/μm2, n = 14; shits1; gyr1 = 10.92 ± 0.87 cisternae/μm2, n = 17; p = 0.13). Interestingly, shits1; gyr1 boutons had a dramatic increase (>2-fold) in very large (>200 nm) endocytic structures, while shits1; gyrPE boutons at 20 min poststimulation had values more similar to those seen in gyrPE and gyr1 larvae 10 min after stimulation (Fig. 8G; shits1; gyrPE = 0.66 ± 0.22 cisternae/μm2, n = 14; shits1; gyr1 = 1.58 ± 0.31 cisternae/μm2, n = 17; p = 0.03). This is likely a consequence of dynamin inhibition forcing a greater fraction of synaptic vesicle membrane to recycle through cisternae via the bulk endocytosis pathway, thereby exacerbating the gyr phenotypes. Indeed, we observed that synaptic vesicle diameter in shits1; gyr1 larvae poststimulation reverted to values similar to those observed in gyr larvae prestimulation, suggesting that these boutons again become defective in regulating synaptic vesicle size when bulk endocytosis contributes prominently to the recovery of synaptic vesicles (Fig. 8D). Intriguingly, when we quantified synaptic vesicle density poststimulation, we discovered that both gyr and control non-shibire larvae recovered to levels similar to those seen before stimulation within 10 min (Fig. 8H; control = 125.8 ± 10.6 vesicles/μm2, n = 18; gyr = 131.1 ± 8.1 vesicles/μm2, n = 21; p = 0.69). However, at 20 min poststimulation, shits1; gyrPE and shits1; gyr1 boutons still had a large decrease in synaptic vesicle density compared with control and gyr, and shits1; gyr1 larvae had significantly lower density than shits1; gyrPE (Fig. 8H; shits1; gyrPE = 80.6 ± 7.7 vesicles/μm2, n = 14; shits1; gyr1 = 51.8 ± 4.0 vesicles/μm2, n = 17; p < 0.01). While the shibire mutation alone inhibits endocytic recovery following synaptic vesicle depletion, the combination of shibire and gyr dramatically enhances this phenotype, indicating synaptogyrin plays an important role in modulating synaptic vesicle regeneration via endocytic cisternae following intense stimulation.
gyr mutants exhibit alterations in presynaptic function
The increase in synaptic vesicle diameter seen in gyr mutants under resting conditions raises the possibility that there may be a concomitant change in quantal size. Indeed, postsynaptic recordings of spontaneous miniature excitatory junctional currents (mEJCs or minis) in third instar larvae using two-electrode voltage-clamp revealed a significant increase in quantal size in the gyr mutant (Fig. 9A; control = 0.83 ± 0.09 nA, n = 5; gyr = 1.35 ± 0.11 nA, n = 6; p < 0.01). gyr larvae also display a slight but significant decrease in spontaneous mEJC frequency (Fig. 9B; control = 2.4 ± 0.2 Hz, n = 5; gyr = 1.8 ± 0.1 Hz, n = 6; p < 0.05). Presynaptic expression of synaptogyrin using the elavc155-GAL4 driver restored both mEJC amplitude and frequency to wild-type values (Fig. 9A,B; mini amplitude = 0.75 ± 0.18 nA, n = 4; mini frequency = 2.6 ± 0.3 Hz, n = 4). While mEJC amplitude is increased in gyr, there is no significant change in evoked EJC amplitude at 0.2 mm Ca2+ (control = 4.1 ± 0.1 nA, n = 5; gyr = 3.8 ± 0.1 nA, n = 6; p > 0.05), indicating that quantal content is reduced in gyr mutants (Fig. 9C; control = 5.1 ± 0.4 quanta, n = 5; gyr = 3.2 ± 0.3 quanta, n = 6; p < 0.001). We next examined short-term regulation of neurotransmitter release by conducting paired-pulse recordings in 0.2 mm Ca2+ and discovered a significant increase in the paired-pulse ratio in gyr mutants (Fig. 9D; control = 1.5 ± 0.1, n = 5; gyr = 2.2 ± 0.2, n = 6; p < 0.001), consistent with a reduction in release probability in gyr mutants. Although gyr larvae exhibit an increase in paired-pulse facilitation, this enhanced facilitation is transient when examined using a longer stimulation protocol of 1500 stimuli at 20 Hz. After ∼500 stimuli, the elevated facilitation in gyr larvae peaks and subsequently declines (Fig. 9E,F), indicating gyr synapses cannot maintain elevated levels of release and become depleted. We then sought to examine whether the delay in synaptic vesicle biogenesis from endocytic cisternae at gyr synapses translated into a defect in more long-term forms of synaptic plasticity induced by a brief 20 Hz stimulation in 2 mm Ca2+. While gyr and control synapses exhibit a similar level of depression, the kinetics of recovery is slowed in gyr larvae as they have a more persistent post-tetanic depression (Fig. 9G), suggesting a defect in the replenishment of releasable vesicles. Although it is currently unclear how the loss of synaptogyrin precisely elicits changes in release properties, the defect in synaptic vesicle recycling and vesicle biogenesis may extend beyond regulating synaptic vesicle size to influencing synaptic vesicle protein content. An alteration in the normal distribution of synaptic vesicle proteins per vesicle could contribute to the abnormal release properties and synaptic plasticity in gyr mutants.
Electrophysiological analysis of gyr mutants. A, Postsynaptic current recordings of spontaneous release at muscle 6 synapses in control (black), gyr (red), and rescue (gray) larvae. This genotypic color code is maintained throughout the figure. Histograms of mEJC peak amplitudes reveal a shift in the distribution toward higher values in gyr larvae, resulting in a significant increase in average mEJC amplitude (p < 0.01). Presynaptic expression of synaptogyrin driven by elavc155-GAL4 rescues the phenotypes. B, Mini frequency is significantly decreased in gyr larvae (p < 0.05). C, There is no significant change in evoked EJC amplitude (p > 0.05), while quantal content is reduced from 5.1 ± 0.4 quanta in controls to 3.2 ± 0.3 quanta in gyr (p < 0.01). D, At 0.2 mm Ca2+, gyr mutants have an increase in the paired-pulse ratio (PPR) compared with controls (p < 0.001). The PPR is defined as the amplitude of the second peak divided by the first peak. E, Representative EJCs in low external Ca2+ (0.2 mm) in control and gyr synapses during a long 20 Hz tetanic stimulation. After an initial increase in facilitation, gyr larvae fail to maintain the increased EJC amplitude. F, EJC amplitude change during a 20 Hz tetanic stimulation in control (black) and gyr (red) synapses. G, Control and gyr synapses were stimulated at 1 Hz for 3 min to induce a low-frequency synaptic depression. This was followed by a 20 Hz stimulation train for 50 s in 2 mm Ca2+ to induce post-tetanic depression. Post-tetanic responses were then monitored at 1 Hz to examine synaptic recovery. gyr larvae require a significantly longer time to reach stable pre-tetanus EJC values following post-tetanic depression (5.1 ± 0.7 min in controls; 10.0 ± 1.3 min in gyr; p < 0.01). Normalized responses are plotted in the graph.
Discussion
Synaptogyrin is not essential for synaptic vesicle trafficking
Our results indicate that synaptogyrin, while not essential for basal synaptic transmission in Drosophila, modulates synaptic vesicle cycling and impacts neurotransmitter release properties. Our findings are consistent with knock-out studies in other species that have identified relatively mild alterations in synaptic activity and vesicle morphology due to the loss of synaptogyrin and/or synaptophysin (McMahon et al., 1996; Janz et al., 1999; Abraham et al., 2006, 2011). While synaptogyrin and synaptophysin are not required for the generation of synaptic vesicles per se, increasing evidence suggests that these proteins play a facilitatory role in vesicle biogenesis and/or endocytosis. The C. elegans synaptogyrin/synaptophysin/SCAMP triple knock-out displayed an increase in clathrin-coated vesicles, which may reflect a compensatory mechanism for defects in other endocytic pathways (Abraham et al., 2006). Furthermore, the phenotypic severity of the endocytic mutants endophilin, synaptotagmin, and synaptojanin was enhanced when combined with mutations in synaptogyrin (Abraham et al., 2006, 2011). Murine retinal cells lacking both synaptophysin and its paralog synaptoporin had a decrease in synaptic vesicle density, and this reduction became more pronounced during periods of high activity (i.e., during dark adaptation). Moreover, during these periods of elevated activity there was an increase in the number of large cisternal-like structures, as well as an increase in synaptic vesicle diameter in synaptophysin knock-outs (Spiwoks-Becker et al., 2001). These results indicate that synaptophysin promotes the efficient formation of synaptic vesicles and that increased activity exacerbates this defect. Similarly, cultures derived from synaptophysin knock-out mice were slower to internalize tagged synaptic vesicle proteins during sustained stimulation, but not under mild stimulation conditions (Kwon and Chapman, 2011). These neurons also exhibited increased synaptic depression and delayed recovery of the synaptic vesicle pool, indicating that synaptophysin facilitates vesicle recovery. Our work suggests that synaptogyrin also modulates synaptic vesicle biogenesis in Drosophila, a species that lacks a synaptophysin homolog. The phenotypic variation we see at gyr boutons under resting conditions (Fig. 5) may reflect a difference in the levels of prior activity between boutons, which has previously been documented at Drosophila NMJs (Guerrero et al., 2005; Peled and Isacoff, 2011). Similar variability in phenotypic severity was observed in mammalian synaptojanin and dynamin 1 knock-out cultures, in which only a minority of synapses (predominantly GABAergic interneurons) displayed a dramatic accumulation of clathrin-coated pits, likely due to increased activity in these neurons (Hayashi et al., 2008).
Functional redundancy between synaptogyrin and synaptophysin has been demonstrated in mice (Janz et al., 1999). Therefore it is possible that the lack of strong phenotypes in flies, mice, and nematodes is due to compensation by a yet unidentified protein(s). SCAMPs, although lacking a MARVEL domain, have historically been linked with physins and gyrins because they share the same transmembrane topology and some SCAMP paralogs are located on synaptic vesicles (Brand et al., 1991; Fernández-Chacón and Südhof, 2000; Hübner et al., 2002). However, the lack of severe phenotypes in the C. elegans synaptogyrin/synaptophysin/SCAMP triple knock-out demonstrates that the absence of all of these protein families does not further impact basal synaptic function (Abraham et al., 2006). Drosophila has a single SCAMP homolog, and is therefore amenable to further genetic analysis to examine issues of redundancy at fly synapses.
A model to explain synaptogyrin's role in synaptic vesicle formation
While there are still questions about synaptogyrin function, we propose a preliminary model to account for the gyr phenotypes related to abnormal synaptic vesicle diameter and delayed synaptic vesicle biogenesis from endocytic cisternae. Numerous studies have established the existence of multiple endocytic pathways at the NMJ, including clathrin-mediated and bulk endocytosis, as well as the controversial kiss-and-run pathway (Koenig and Ikeda, 1996; Verstreken et al., 2002; Dickman et al., 2005). We propose there are at least two pathways responsible for budding of synaptic vesicles from endocytic cisternae, one of which is regulated by synaptogyrin and the other that is independent of synaptogyrin function. In a wild-type bouton, both pathways are intact and produce vesicles of normal size, resulting in a low level of abnormal synaptic vesicles or residual endocytic cisternae. In the absence of synaptogyrin, the synaptogyrin-dependent pathway no longer functions properly and produces vesicles of abnormal size, while the synaptogyrin-independent pathway is still capable of generating normal synaptic vesicles. Nevertheless, the loss of synaptogyrin results in an overall increase in synaptic vesicle diameter in gyr animals.
During periods of intense activity that induce bulk endocytosis, a wild-type synapse has both pathways fully capable of regenerating synaptic vesicles from endocytic cisternae, and these structures quickly resolve into vesicles. However, in the gyr mutant, the synaptogyrin-dependent pathway is impaired, and the time it takes to resolve cisternae into synaptic vesicles is increased relative to controls. Furthermore, if the synaptogyrin-independent pathway is dominant in the gyr mutant, the synaptic vesicles that form shortly after intense stimulation would be more uniform in size because the vast majority would have been generated through the unimpaired synaptogyrin-independent pathway or via clathrin-mediated endocytosis from the plasma membrane.
An alternative model to the multiple endocytic pathways described above is that a single major endocytic pathway is responsible for vesicle biogenesis from cisternae, and the loss of synaptogyrin merely slows the process of synaptic vesicle budding. If synaptogyrin were involved in the budding of all synaptic vesicles from endocytic cisternae, one would expect to see a population of abnormally shaped synaptic vesicles shortly after the high K+ shock as the cisternae begin to resolve, which we did not observe. However, it is unknown what fraction of vesicles generated after the K+ shock originate directly from the plasma membrane as opposed to cisternae. If the majority of synaptic vesicles observed 10 min after high K+ stimulation derive from the plasma membrane and this form of endocytosis is independent of synaptogyrin, then plasma membrane-derived synaptic vesicles may mask a small population of abnormal synaptic vesicles originating from cisternae in the gyr mutant. Indeed, the observation that synaptic vesicle diameter is increased and synaptic vesicle density is decreased in shits1; gyr1 larvae following high K+ stimulation is consistent with this model. Assuming that dynamin inhibition forces a greater amount of synaptic membrane to cycle through endocytic cisternae, boutons lacking synaptogyrin would exhibit increased phenotypic severity. It should be noted, however, that the extent to which synaptogyrin influences traditional clathrin-mediated endocytosis from the plasma membrane is currently unknown.
Synaptogyrin regulates neurotransmitter release and short-term synaptic plasticity
Synaptic vesicle biogenesis, whether from an endosomal compartment or the plasma membrane, requires the recruitment of a wide variety of essential synaptic vesicle proteins. Maintaining the correct complement of synaptic vesicle proteins could theoretically be achieved through a kiss-and-run endocytic mechanism, which would allow a transiently fused vesicle to directly reform after fusion pore closure rather than collapse into the plasma membrane. While rapid endocytosis has been observed at certain synapses such as the calyx of Held (Wu et al., 2005), the contribution of kiss-and-run endocytosis at other synapses, including the Drosophila NMJ, is controversial (He and Wu, 2007). An alternative way to promote proper synaptic vesicle sorting would be through the creation of microdomains on the plasma membrane and endosomal cisternae enriched in the lipids and proteins found on individual synaptic vesicles. Synaptophysin, via its interactions with cholesterol and synaptobrevin, as well as its ability to multimerize, is an intriguing candidate to seed this sort of microdomain (Rehm et al., 1986; Thomas et al., 1988; Calakos and Scheller, 1994; Washbourne et al., 1995; Thiele et al., 2000).
If synaptogyrin influences the recruitment of synaptic vesicle proteins through the formation of membrane microdomains during endocytosis, the loss of synaptogyrin could lead to alterations in protein stoichiometry or protein sorting, which could impact synaptic vesicle release. Indeed, there is precedent for protein-sorting defects resulting in changes in exocytosis. A recent study investigating the function of a neuronal Rab35 GTPase activating protein named Skywalker (Sky) discovered that sky mutant larvae accumulate an excess of cisternal-like structures after intense stimulation and show increased neurotransmitter release relative to controls (Uytterhoeven et al., 2011). It remains to be determined whether the changes in neurotransmitter release observed in gyr mutants are the result of altered synaptic vesicle protein composition or whether synaptogyrin directly influences exocytosis through a currently unknown mechanism. Interestingly, several recent studies have associated mutations in the synaptogyrin locus with schizophrenia susceptibility in humans (Verma et al., 2004, 2005; Cheng and Chen, 2007; Iatropoulos et al., 2009). Therefore, defining how synaptogyrin controls synaptic vesicle cycling may provide insights into neuropsychiatric pathology.
Footnotes
This work was supported by National Institutes of Health Grant NS40296 (J.T.L.). R.A.J. was a recipient of the PEW Latin American Fellows Program in the Biomedical Sciences.
The authors declare declare no competing financial interests.
- Correspondence should be addressed to J. Troy Littleton, The Picower Institute for Learning and Memory, Massachusetts Institute of Technology, 43 Vassar Street, 46-3243, Cambridge, MA 02139. troy{at}mit.edu
