Ca²⁺-Dependent, Phospholipid-Binding Residues of Synaptotagmin Are Critical for Excitation–Secretion Coupling In Vivo

Abstract

Synaptotagmin I is the Ca²⁺ sensor for fast, synchronous release of neurotransmitter; however, the molecular interactions that couple Ca²⁺ binding to membrane fusion remain unclear. The structure of synaptotagmin is dominated by two C₂ domains that interact with negatively charged membranes after binding Ca²⁺. In vitro work has implicated a conserved basic residue at the tip of loop 3 of the Ca²⁺-binding pocket in both C₂ domains in coordinating this electrostatic interaction with anionic membranes. Although results from cultured cells suggest that the basic residue of the C₂A domain is functionally significant, such studies provide contradictory results regarding the importance of the C₂B basic residue during vesicle fusion. To directly test the functional significance of each of these residues at an intact synapse in vivo, we neutralized either the C₂A or the C₂B basic residue and assessed synaptic transmission at the Drosophila neuromuscular junction. The conserved basic residues at the tip of the Ca²⁺-binding pocket of both the C₂A and C₂B domains mediate Ca²⁺-dependent interactions with anionic membranes and are required for efficient evoked transmitter release. Our results directly support the hypothesis that the interactions between synaptotagmin and the presynaptic membrane, which are mediated by the basic residues at the tip of both the C₂A and C₂B Ca²⁺-binding pockets, are critical for coupling Ca²⁺ influx with vesicle fusion during synaptic transmission in vivo. Our model for synaptotagmin's direct role in coupling Ca²⁺ binding to vesicle fusion incorporates this finding with results from multiple in vitro and in vivo studies.

Introduction

The synaptic vesicle protein, synaptotagmin, functions as the Ca²⁺ sensor for synchronous neurotransmitter release. Synaptotagmin contains two C₂ domains, C₂A and C₂B, that coordinate Ca²⁺ ions. Whereas Ca²⁺ binding by the C₂B domain is essential for synchronous transmitter release, Ca²⁺ binding by C₂A plays only a modest role (Fernández-Chacón et al., 2002; Mackler et al., 2002; Robinson et al., 2002). Synaptotagmin's C₂ domains interact with anionic phospholipids in a Ca²⁺-dependent manner in vitro (Perin et al., 1990; Earles et al., 2001; Fernández-Chacón et al., 2001; Bai et al., 2002), and deficits in Ca²⁺-triggered fusion in several synaptotagmin mutants parallel the decrease in Ca²⁺-dependent phospholipid binding (Fernández-Chacón et al., 2001; Mackler et al., 2002; Sørensen et al., 2003; Wang et al., 2003; Nishiki and Augustine, 2004; Li et al., 2006). Thus, a Ca²⁺-dependent interaction between synaptotagmin and phospholipids is postulated to be critical in mediating Ca²⁺-triggered vesicle fusion.

Specific residues within synaptotagmin are required for Ca²⁺-dependent phospholipid binding in vitro. At the tip of the Ca²⁺-binding loops of each C₂ domain, there is a conserved basic residue (Fig. 1, ⊕) that is thought to mediate an electrostatic interaction between synaptotagmin and the anionic presynaptic membrane (Chae et al., 1998; Fernández-Chacón et al., 2001; Wang et al., 2003). This interaction may contribute to the electrostatic switch that drives vesicle fusion after Ca²⁺ influx (Davletov et al., 1998; Ubach et al., 1998). In C₂A, this basic residue is likely functionally significant, because it is necessary for efficient Ca²⁺-triggered vesicle fusion in a variety of cultured cell types (Fernández-Chacón et al., 2001; Sørensen et al., 2003; Wang et al., 2003). The functional significance of this basic residue in C₂B, however, is controversial, because studies from cultured cells provide contradictory results (Wang et al., 2003; Li et al., 2006). In addition, cultured cells do not necessarily reproduce the exact behavior of intact synapses (Banker and Goslin, 1998), so it is critical to test each of these residues for function in vivo.

Figure 1.

Both the C₂A and C₂B domains of synaptotagmin I have a conserved basic residue at the tip of the Ca²⁺-binding pocket. A, Alignment of synaptotagmin I from human, rat, mouse, and Drosophila. Bars indicate β-sheets, ⊕ indicates the conserved basic residues, dots indicate Ca²⁺-binding residues, and open boxes indicate conserved hydrophobic residues. Within the alignment, conserved residues are shown in gray, and identical residues are boxed. B, Schematic representation of loops 1 and 3, which form the Ca²⁺-binding pockets of both C₂ domains [adapted with permission from Fernandez et al. (2001), their Fig. 6, using the Drosophila sequence to show the conserved basic residues (⊕) examined in this study].

Because (1) multiple residues located at the tip of the Ca²⁺-binding pockets, including these basic residues, in both the C₂A and C₂B domains are critical for Ca²⁺-dependent phospholipid interactions (Chae et al., 1998; Chapman and Davis, 1998; Davis et al., 1999; Fernández-Chacón et al., 2001; Bai et al., 2002; Frazier et al., 2003; Wang et al., 2003; Araç et al., 2006) (but see Li et al., 2006) and (2) multiple fusion assays implicate these tip residues in Ca²⁺-triggered fusion (Fernández-Chacón et al., 2001; Sørensen et al., 2003; Wang et al., 2003; Rhee et al., 2005; Martens et al., 2007), we directly tested the functional significance of each of these basic residues at an intact synapse by individually neutralizing them and measuring evoked release at the Drosophila neuromuscular junction. Here, we demonstrate that the conserved basic residues at the tip of the C₂A and C₂B Ca²⁺-binding pockets each mediate interactions with anionic phospholipids in vitro and are each critical for synaptotagmin function in vivo.

Materials and Methods

Site-directed mutagenesis.

To neutralize the positive charge without disrupting the structure of the C₂ domain (Fernández-Chacón et al., 2001), arginine residues 285 and 419 (Fig. 1, ⊕) of Drosophila synaptotagmin I (syt) were mutated to glutamines using PCR. To mutate arginine 285, a specifically mutated oligonucleotide (CGAGAACTGATCGAAGTCGAAAATGGC) was paired with a wild-type (WT) oligonucleotide that flanked a unique StyI site. The PCR product was gel purified and used as a macroprimer in a second round of PCR with a wild-type oligonucleotide that flanked a unique EcoRV site. This second-round, mutant PCR product was then subcloned into an otherwise wild-type Drosophila syt cDNA construct in pBluescript II (Mackler and Reist, 2001). To mutate arginine 419, a specifically mutated oligonucleotide (TGCAGCGGCCGATGGGTTCGGAGGTGCCAATCTGATCGTAGTCCACGACGGTCACAACG) containing a unique EagI site was paired with a wild-type oligonucleotide that flanked a unique EcoRV site. That mutant PCR product was then gel purified and subcloned into the Drosophila syt cDNA construct in pBluescript mentioned above. DNA sequencing confirmed that either R285Q or R419Q was the only mutation harbored in the entire region generated by PCR. Each mutant syt cDNA was subcloned into a pUAST vector to place the mutant syt gene under the control of the UAS promoter (Brand and Perrimon, 1993).

Generation of mutant transgenic lines.

Drosophila embryos were transfected with the mutant pUAST plasmids as described previously (Mackler and Reist, 2001). At least two lines carrying separate insertions of the mutant syt transgenes were isolated for each genotype. Expression of each transgene was localized to the nervous system using the elav promoter to drive Gal4, and the Gal4/UAS system was used to amplify expression levels (Brand and Perrimon, 1993; Yao and White, 1994). Standard genetic techniques were used to cross the transgenes into the syt^null background to express the transgene in the absence of endogenous synaptotagmin I for all experiments. The genotypes of the mutant lines were yw; syt^AD4 elav GAL4/syt^AD4; P[UAS syt^A-R285Q]/+, and yw; syt^AD4 elav GAL4/syt^AD4; P[UAS syt^B-R419Q]/+, which are written as P[syt^A-RQ] and P[syt^B-RQ], respectively, in the text. The genotype of the control was yw; syt^AD4 elav GAL4/syt^AD4; P[UAS syt^wild-type]/+, which is written as P[syt^WT] in the text.

Electrophysiology.

Evoked and spontaneous excitatory junction potentials (EJPs) were recorded from muscle 6 of segments 3 and 4 of third instars in HL3 saline [5 mm KCl, 1.5 mm CaCl₂, 70 mm NaCl, 20 mm MgCl₂, 10 mm NaCHO₃, 5 mm HEPES, 115 mm sucrose, and 5 mm trehalose (Stewart et al., 1994)] as described previously (Loewen et al., 2001). Briefly, third-instar larvae were dissected in Ca²⁺-free HL3 to expose the body wall musculature. After changing to HL3 saline containing 1.5 mm Ca²⁺, muscle 6 was impaled with a recording electrode having a resistance between 10 and 40 MΩ. Evoked EJPs were generated by stimulating segmental nerves with a suction electrode filled with HL3. The Ca²⁺ dependence curve was generated by evoking EJPs in external Ca²⁺ concentrations ranging from 0.6 to 5.0 mm. Muscles were impaled in 1.5 mm Ca²⁺ HL3, and recordings in several different Ca²⁺ concentrations were obtained from each muscle fiber. The trehalose was varied between 0.5 and 5.0 mm, although this had no effect on evoked release (data not shown). The predicted maximal response was calculated by fitting the Hill equation to the mean response at each extracellular Ca²⁺ concentration (Kalediagraph). The Ca²⁺ cooperativity coefficient was estimated from the slope of a double-log plot of EJP amplitude versus Ca²⁺ concentration (Kalediagraph). All events were collected using an AxoClamp 2B (Molecular Devices) and digitized using a MacLab4s analog-to-digital converter (ADInstruments). Spontaneous events were recorded in Chart Software, and evoked events were recorded in Scope software (ADInstruments). Spontaneous fusion events were identified manually, blind to genotype.

Immunolabeling.

For immunolabeling of the neuromuscular junction, third instars of the indicated genotypes were dissected in Ca²⁺-free HL3 saline to expose their body wall muscles and fixed in 4% paraformaldehyde in PBS. This whole-mount preparation was incubated overnight in anti-synaptotagmin antibody [Dsyt-CL1 (Mackler et al., 2002)], diluted 1:1000 in PBST-NGS [PBS with 0.1% Triton, 1% BSA, and 1% normal goat serum (NGS from Jackson ImmunoResearch)], washed in PBST for 30–60 min, incubated in Alexa Fluor 488 goat-anti-rabbit antibody (Invitrogen) diluted 1:5000 in PBST-NGS for 1 h, washed in PBST for 1–2 h, mounted in Citifluor (Ted Pella), and then visualized on a Zeiss LSM 510Meta confocal microscope equipped with an argon laser (Zeiss Microimaging). Emissions were collected using a bandpass 505–530 emission filter at 40× with a pinhole set for 1 Airy unit.

Immunoblotting.

Similar levels of transgene expression were verified by Western blot analysis. The nervous system of a single third instar of the indicated genotype was homogenized in protein loading buffer (Bio-Rad). Proteins were separated via SDS-PAGE and transferred to a polyvinylidene difluoride membrane as described previously (Mackler and Reist, 2001). All antibodies were diluted in PBS containing 0.05% Tween and 10% NGS. Blots were probed with Dsyt-CL1 diluted between 1:1250 and 1:5000, and an anti-actin antibody (MAB 1501; Millipore Bioscience Research Reagents), diluted between 1:20,000 and 1:80,000. Actin levels were used to normalize for equal protein loading. These antibodies were visualized with HRP-tagged donkey anti-rabbit IgG, diluted 1:10,000 to 1:20,000, and HRP-tagged donkey anti-mouse IgG, diluted 1:2500 to 1:40,000 (Jackson ImmunoResearch). The HRP-tagged antibodies were detected using a Supersignal West Dura Extended Duration Substrate kit (Pierce) in an Epichemi3 Darkroom (UVP). The synaptotagmin:actin signal ratio was determined for each CNS, then normalized to the mean synaptotagmin:actin ratio of the P[syt^WT] lanes on each blot to allow comparison of signal between multiple blots.

C₂AB–phospholipid binding.

cDNA encoding the cytoplasmic domain of Drosophila synaptotagmin (C₂AB, residues 191–474) was generated by PCR using primers AGCAGAGAATTCAGAAGCTGGGGCGCC and CCGCCGAAGCTTTTACTTCATGTTCTT. WT, C₂A mutant (A-R285Q), and C₂B mutant (B-R419Q) C₂AB constructs were subcloned into the expression vector, pGEX-KG (kindly provided by Dr. Sandra Bajjalieh, University of Washington, Seattle, WA). Mammalian cDNA encoding WT, C₂A mutant (A-R233Q), and C₂B mutant (B-K366Q) C₂AB in pGEX-KG (Li et al., 2006) was kindly provided by Dr. Thomas C. Südhof (UT Southwestern Medical Center, Dallas, TX). The C₂AB domains were expressed as GST fusion proteins and purified using glutathione-Sepharose beads [GE Healthcare (Chapman et al., 1995)]. Recombinant synaptotagmin harbors tightly bound nucleic acid contaminants that may affect its properties (Ubach et al., 2001). These contaminants were removed by DNase/RNase and high-salt washes as described previously (Bai et al., 2004). Synthetic 1,2-dioleoyl-sn-glycero-3-phospho-l-serine [phosphatidylserine (PS)], 1,2-dioleoyl-sn-glycero-3-phosphocholine [phosphatidylcholine (PC)], and N-(lissamine rhodamine B sulfonyl)-1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine were purchased from Avanti Polar Lipids. A phospholipid mixture containing 15% negatively charged phospholipid was chosen to approximate the amount of negatively charged phospholipids found in neuronal membranes (Takamori et al., 2006). This mixture [15% PS + 84% PC + 1% N-(lissamine rhodamine B sulfonyl)-1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine] was dried under a stream of nitrogen and suspended in HEPES-buffered saline (50 mm HEPES/150 mm NaCl, pH 7.4). Large (100 nm) unilamellar liposomes were prepared using an extruder from Avanti Polar Lipids, as described previously (Davis et al., 1999). Rhodamine-labeled liposome-binding assays were performed as described previously (Hui et al., 2005) in 150 μl of HEPES-buffered saline (50 mm HEPES/150 mm NaCl, pH 7.4) using 6 μg of immobilized protein and 22 nm liposomes per data point. Bound liposomes were eluted with HEPES-buffered saline containing 1% Triton X-100. The solubilized lipids were collected, and binding was quantified by measuring the emission fluorescence intensity of rhodamine at 580 nm. To determine the apparent Ca²⁺ affinity for WT and mutant C₂AB, we assayed for PS/PC binding as a function of the indicated Ca²⁺ concentrations. Free Ca²⁺ concentration was calculated by using WEBMAXC STANDARD software developed by Stanford University (http://www.stanford.edu/∼cpatton/webmaxc/webmaxcS.htm). Data were plotted and fitted with sigmoidal dose–response curves (variable slope) using PRISM 5.0 software (GraphPad). In all experiments, error bars represent the SD from at least three independent determinations.

Results

The conserved basic residues at the tips of synaptotagmin's C₂A and C₂B Ca²⁺-binding pockets are both required for efficient synaptic transmission

A highly conserved basic residue is present in loop 3 of the Ca²⁺-binding pocket in both the C₂A and C₂B domains of synaptotagmin I (Fig. 1, ⊕). To examine the role of these residues during synaptic transmission at an intact synapse, we separately mutated the conserved basic residue in each C₂ domain to a glutamine and assessed evoked release at the Drosophila neuromuscular junction. We will denote the mutation of these residues as syt^A-RQ and syt^B-RQ. All experiments were performed on third instars expressing the indicated form of synaptotagmin from a transgene in the absence of endogenous synaptotagmin I. To indicate their transgenic origin, we will refer to the C₂A mutants as P[syt^A-RQ], the C₂B mutants as P[syt^B-RQ], and the transgenic controls as P[syt^WT]. Finally, because the random insertion of a transgene could potentially disrupt another functionally important gene, two independent insertions of each mutant transgene were examined to ensure that any deficits found resulted from the mutation rather than the insertion site.

Evoked EJPs and spontaneous miniature excitatory junction potentials (mEJPs) were recorded from larval neuromuscular junctions in HL3 saline containing 1.5 mm Ca²⁺. Mutation of either of the conserved basic residues that mediate synaptotagmin's electrostatic interaction with anionic phospholipids decreased the evoked response (Fig. 2). The mutation within the C₂A domain reduced evoked release by ∼50% compared with the transgenic wild-type control. Evoked release in P[syt^A-RQ] was 13.6 ± 2.1 mV (line 7) or 11.5 ± 0.6 mV (line 8) compared with 27.4 ± 1.5 mV in P[syt^WT] (Fig. 2B, asterisks) (p < 0.0001). This decrease is similar to the result observed in cultured cells from mice and rats harboring a homologous mutation (Fernández-Chacón et al., 2001; Sørensen et al., 2003; Wang et al., 2003; Han et al., 2004). The mutation within the C₂B domain reduced evoked release by ∼80% compared with the transgenic control. Evoked release in the P[syt^B-RQ] mutants was 5.4 ± 0.7 mV (line 3) or 4.4 ± 0.7 mV (line 4), compared with 27.4 ± 1.5 mV in P[syt^WT] (Fig. 2B, double asterisks) (p < 0.0001). This decrease is consistent with results from cultured rat neuroendocrine cells, PC12 cells (Wang et al., 2003), but in direct contrast to the results from cultured mouse neurons, hippocampal autapses (Li et al., 2006). The level of evoked release remaining in the P[syt^B-RQ] mutants is significantly less than that in the P[syt^A-RQ] mutants (Fig. 2B, asterisks vs double asterisks) (p < 0.01). No difference in mean EJP amplitude was found between the insertions of a given genotype for either P[syt^A-RQ] or P[syt^B-RQ] (p > 0.2). Thus, the reduction in evoked release results from the specific synaptotagmin mutations and not from insertion of the transgene disrupting an unspecified gene.

Figure 2.

Evoked release is reduced in phospholipid-binding mutants of both C₂ domains. A, Representative traces recorded from larval muscle fiber 6. Each trace represents the mean of 30 consecutive sweeps from the same muscle fiber. B, Compared with P[syt^WT], the mean EJP amplitude of all P[syt^RQ] lines was significantly decreased (*, **p < 0.0001, 1-way ANOVA; P[syt^WT], n = 16; P[syt^A-RQ] line 7, n = 12; P[syt^A-RQ] line 8, n = 14; P[syt^B-RQ] line 3, n = 16; P[syt^B-RQ] line 4, n = 13). Additionally, the evoked responses of the P[syt^B-RQ] lines were significantly lower than those of the P[syt^A-RQ] lines (*, **p < 0.01), but no differences were found between P[syt^A-RQ] lines 7 and 8, or between P[syt^B-RQ] lines 3 and 4 (p > 0.2).

Mutation of either of the conserved basic residues increases the rate of spontaneous release at third-instar neuromuscular junctions (Fig. 3A). The mutation within the C₂A domain at least doubled the rate of mEJPs, with a frequency of 3.5 ± 0.4 Hz (line 7) or 2.8 ± 0.3 Hz (line 8) in P[syt^A-RQ] compared with 1.4 ± 0.1 Hz in P[syt^WT] (Fig. 3A, asterisks) (p < 0.0001). The mutation within the C₂B domain increased the rate of mEJPs by ∼40%, to 1.9 ± 0.2 Hz (line 3) or 2.0 ± 0.3 Hz (line 4) compared with 1.4 ± 0.1 Hz for P[syt^WT] (Fig. 3A, double asterisks) (p < 0.05). Although the trend toward a less severe increase in mEJP frequency in the C₂B mutants suggests that these mutants may maintain more of a vesicle-clamping function, the frequency of mEJPs in the P[syt^A-RQ] and P[syt^B-RQ] mutants was not significantly different (p > 0.05, single asterisks vs double asterisks, one-way ANOVA). No difference in mEJP frequency was found between independent insertions of the synaptotagmin gene for P[syt^A-RQ] or P[syt^B-RQ] (Fig. 3A) (p > 0.2). The amplitude of mEJPs was unchanged in the mutants (P[syt^WT], 1.10 ± 0.03 mV; P[syt^A-RQ] line 7, 1.00 ± 0.08 mV; P[syt^A-RQ] line 8, 1.24 ± 0.06 mV; P[syt^B-RQ] line 3, 1.09 ± 0.06 mV; P[syt^B-RQ] line 4, 1.10 ± 0.07 mV; p > 0.1, one-way ANOVA). In addition, we compared the frequency of quantal amplitudes for each mutant line (Fig. 3B) (P[syt^WT], P[syt^A-RQ] line 8, and P[syt^B-RQ] line 3 shown). The constant mEJP amplitude indicates that neither the C₂A nor the C₂B mutation perturbs synaptic vesicle filling or the postsynaptic machinery and is consistent with previous studies of spontaneous release characteristics in syt^A-RQ mutants (Sørensen et al., 2003).

Figure 3.

mEJP frequency is increased in phospholipid-binding mutants of both C₂ domains. A, Mean mEJP frequencies of each mutant and the transgenic control. P[syt^WT], n = 55 fibers; P[syt^A-RQ] line 7, n = 7 fibers; P[syt^A-RQ] line 8, n = 17 fibers; P[syt^B-RQ] line 3, n = 27 fibers; P[syt^B-RQ] line 4, n = 12 fibers. Compared with P[syt^WT], all genotypes had an increased mEJP frequency (p < 0.05). No significant difference was detected between P[syt^A-RQ] and P[syt^B-RQ] (p > 0.05), between P[syt^A-RQ] lines 7 and 8 (p > 0.1), or between P[syt^B-RQ] lines 3 and 4 (p > 0.1). *p < 0.0001, **p < 0.05, compared with P[syt^WT]. B, Frequency distribution curves of mEJP amplitudes calculated from 1000 individual events per transgenic line in 0.2 mV bins. P[syt^WT], P[syt^A-RQ] line 8, and P[syt^B-RQ] line 3 are shown.

The decreased evoked release observed in P[syt^A-RQ] and P[syt^B-RQ] mutants is not the result of protein misexpression

It is conceivable that the decreased evoked release demonstrated in both the P[syt^A-RQ] and P[syt^B-RQ] mutants results from protein misexpression. To assess the expression level of each transgenic line, we probed Western blots of larval CNSs with an anti-synaptotagmin antibody (Fig. 4A) (P[syt^WT], P[syt^A-RQ] line 8, and P[syt^B-RQ] line 3 shown). The two independent lines of both P[syt^A-RQ] and P[syt^B-RQ] used for the electrophysiological experiments expressed approximately the same amount of transgenic synaptotagmin as the transgenic control line (Fig. 4B). Thus, the deficits in evoked release seen in both the P[syt^A-RQ] and P[syt^B-RQ] mutants are not the result of insufficient expression of the transgene. To determine whether the mutant proteins were appropriately localized to synaptic sites, the neuromuscular junctions of mutant and control transgenic larvae were immunolabeled with an anti-synaptotagmin antibody. In all lines, transgenic synaptotagmin was properly localized to the neuromuscular junction (Fig. 4C) (P[syt^WT], P[syt^A-RQ] line 8, and P[syt^B-RQ] line 4 shown). Thus, the decrease in evoked release did not result from either a deficiency in gene dosage or improper localization.

Figure 4.

Synaptotagmin expression is unaltered in phospholipid-binding mutants. A, Synaptotagmin is expressed at similar levels in each of the transgenic synaptotagmin lines. Representative Western blots of homogenized CNSs of third instars from the indicated lines were probed with an anti-synaptotagmin antibody. To confirm equal loading, they were also probed with an anti-actin antibody. B, Comparison of synaptotagmin/actin ratio normalized to the mean ratio of the transgenic control for P[syt^WT] (n = 45), P[syt^A-RQ] line 7 (n = 6), P[syt^A-RQ] line 8 (n = 14), P[syt^B-RQ] line 3 (n = 18), and P[syt^B-RQ] line 4 (n = 10). An ANOVA showed no significant difference between any of the genotypes (p > 0.1). C, Synaptotagmin is localized to the larval neuromuscular junction in each of the transgenic synaptotagmin lines. Representative confocal Z-stack projections are shown for P[syt^WT], P[syt^A-RQ] line 8, and P[syt^B-RQ] line 4. Scale bar, 10 μm.

The syt^A-RQ and syt^B-RQ mutations decrease the Ca²⁺ affinity of neurotransmitter release

Release of neurotransmitter has long been known to be a Ca²⁺-dependent, cooperative process (Dodge and Rahamimoff, 1967). The Ca²⁺ cooperativity (“n”) of release may represent the mean number of Ca²⁺ ions used to trigger a vesicle fusion event (Dodge and Rahamimoff, 1967; Stevens and Sullivan, 2003; Tamura et al., 2007). To assess the Ca²⁺ dependence of the release properties in the syt^A-RQ and syt^B-RQ mutants, evoked release was measured at a variety of extracellular Ca²⁺ concentrations, ranging from 0.6 to 5.0 mm. At all Ca²⁺ concentrations, these mutants exhibit a decrease in evoked transmitter release compared with control (Fig. 5A). To determine whether mutation of either of these basic residues changes the Ca²⁺ cooperativity of release, we plotted the mean EJP amplitude versus extracellular Ca²⁺ concentration on a double-log plot in nonsaturating Ca²⁺ ranges (Fig. 5B). As estimated from the slope of these double-log plots, n = 3.2 for P[syt^WT], n = 3.0 for P[syt^A-RQ], and n = 2.9 for P[syt^B-RQ], similar to previously recorded values (n = 3.0–3.6) at wild-type neuromuscular junctions in Drosophila (Littleton et al., 1994; Stewart et al., 2000; Yoshihara and Littleton, 2002; Okamoto et al., 2005). Thus, neither of the mutations changes the Ca²⁺ cooperativity of release. This finding is consistent with the hypothesis that synaptotagmin's interaction with phospholipids functions downstream of Ca²⁺ binding and does not affect the number of Ca²⁺ ions needed to trigger vesicle fusion.

Figure 5.

Phospholipid-binding mutants decrease the apparent Ca²⁺ affinity but do not affect the Ca²⁺ cooperativity of release. A, EJPs were evoked in P[syt^WT], P[syt^A-RQ] line 8, and P[syt^B-RQ] lines 3 and 4 by 0.05 Hz stimulation, and 10 sweeps were averaged for each fiber at each [Ca²⁺]. The responses from lines 3 and 4 of P[syt^B-RQ] were not significantly different, so the results were pooled. For all genotypes at all [Ca²⁺], n = 10–18 muscle fibers, except for 1.5 mm Ca²⁺, at which n = 40–49 muscle fibers. The Hill equation was fit to the data. Error bars are SEM. B, The EJP amplitudes within the nonsaturating range of Ca²⁺ were plotted on a double-log plot, and a linear regression line was used to determine the slope (n) (P[syt^WT]: n = 3.2, r = 0.999; P[syt^A-RQ]: n = 3.0, r = 0.999; P[syt^B-RQ]: n = 2.9, r = 0.933). C, The EJP amplitudes at each Ca²⁺ concentration were normalized to the maximum predicted by the Hill equation for each genotype and replotted to illustrate the shift in EC₅₀. P[syt^wt]: EC₅₀ = 1.4 ± 0.1 mm; P[syt^A-RQ]: EC₅₀ = 2.0 ± 0.1 mm; P[syt^B-RQ]: EC₅₀ = 2.3 ± 0.2 mm. For all panels, black filled circles indicate P[syt^WT], gray open squares indicate P[syt^A-RQ], and gray open diamonds indicate P[syt^B-RQ].

To assess the apparent Ca²⁺ affinity of release, we fit the Hill equation to the data and normalized to the predicted maximal response within each line. Figure 5C shows that the apparent Ca²⁺ affinity of release in vivo was decreased in both the P[syt^A-RQ] (EC₅₀ = 2.0 ± 0.1 mm) and P[syt^B-RQ] (EC₅₀ = 2.3 ± 0.2 mm) mutants compared with the transgenic wild-type control (EC₅₀ = 1.4 ± 0.1 mm). A rightward shift of the EC₅₀ for Ca²⁺-evoked transmitter release was previously seen for syt^A-RQ mutants in several cell culture systems, including hippocampal autapses, chromaffin cells, and PC12 cells (Fernández-Chacón et al., 2001; Sørensen et al., 2003; Wang et al., 2003), and our studies now show that the syt^A-RQ mutation decreases the Ca²⁺ affinity of release at an intact synapse. In addition, our experiments demonstrate that the syt^B-RQ mutation also results in a severe disruption of Ca²⁺-evoked transmitter release at an intact synapse, indicating that the function of the basic residue at the tip of loop 3 is conserved.

The syt^A-RQ and syt^B-RQ mutations decrease the Ca²⁺ affinity of phospholipid binding

To determine whether the conserved basic residue at the tip of each C₂ domain participates in the electrostatic interaction between synaptotagmin and anionic phospholipids, we measured the Ca²⁺ affinity of C₂AB domain binding to PS/PC liposomes in vitro. Similar to previous reports from mammalian systems (Chae et al., 1998; Davletov et al., 1998; Fernandez et al., 2001; Fernández-Chacón et al., 2001; Wang et al., 2003), the Drosophila syt^A-RQ mutation decreased the Ca²⁺ affinity of C₂AB binding to negatively charged phospholipids (Fig. 6A) (WT EC₅₀ = 238 ± 20 μm Ca²⁺; syt^A-RQ EC₅₀ = 368 ± 35 μm Ca²⁺). Using Drosophila C₂AB domains, the syt^B-RQ mutation decreased the Ca²⁺ affinity of binding to negatively charged phospholipids to a similar extent as the syt^A-RQ mutation (Fig. 6A) (WT EC₅₀ = 238 ± 20 μm Ca²⁺; syt^B-RQ EC₅₀ = 344 ± 32 μm Ca²⁺). Compared with mammalian systems, this result is consistent with the findings of Wang et al. (2003) but in conflict with the findings of Li et al. (2006). To directly examine whether or not the Drosophila and mammalian systems differ, we also measured the Ca²⁺ affinity of mammalian C₂AB domain binding to PS/PC liposomes. As shown in Figure 6B, both the syt^A-RQ and the syt^B-KQ mutations decreased the Ca²⁺ affinity of C₂AB binding to negatively charged phospholipids to a similar extent (WT EC₅₀ = 98 ± 6 μm Ca²⁺; syt^A-RQ EC₅₀ = 154 ± 10 μm Ca²⁺; syt^B-KQ EC₅₀ = 147 ± 7 μm Ca²⁺). The EC₅₀ values measured in this study are higher than those previously reported (Fernández-Chacón et al., 2001; Wang et al., 2003) because of a lower, and more physiological (Takamori et al., 2006), percentage of negatively charged phospholipids in our liposomes. Thus, in Drosophila and mammals, mutation of either the C₂A or the C₂B conserved basic residue decreases the Ca²⁺ affinity of interactions with anionic phospholipids.

Figure 6.

Mutation of the conserved basic residue in either C₂A or C₂B decreases Ca²⁺-dependent interactions between anionic phospholipids and synaptotagmin in both Drosophila and mammals. Phospholipid binding for wild-type (black filled circles), C₂A mutant (gray open squares; A-RQ in both Drosophila and mammals), and C₂B mutant (gray open diamonds; B-RQ in Drosophila and B-KQ in mammals) C₂AB domains are graphed versus Ca²⁺ concentration. A, Immobilized WT or A-RQ or B-RQ mutant versions of Drosophila C₂AB were assayed for binding PS/PC liposomes. The A-RQ and the B-RQ mutations each decreased the apparent Ca²⁺ affinity of binding by ∼1.5-fold compared with WT without altering the Hill coefficient. The EC₅₀ (mean ± SD in micromolar concentration) was 368 ± 35 for A-RQ, 344 ± 32 for B-RQ, and 238 ± 20 for WT. B, Immobilized WT, A-RQ, or B-KQ mutant versions of mammalian C₂AB were assayed for binding PS/PC liposomes. The A-RQ and the B-KQ mutations each decreased the apparent Ca²⁺ affinity of binding by ∼1.5-fold compared with WT without altering the Hill coefficient. The EC₅₀ (mean ± SD in micromolar concentration) was 154 ± 10 for A-RQ, 147 ± 7 for B-KQ, and 98 ± 6 for WT.

Discussion

Several reports indicate that residues at the tip of the Ca²⁺-binding pockets, including the conserved basic residues, in both C₂ domains of synaptotagmin are critical for Ca²⁺-dependent phospholipid interactions in vitro and Ca²⁺-triggered fusion in cultured cells (Chapman and Davis, 1998; Fernández-Chacón et al., 2001; Bai et al., 2002; Frazier et al., 2003; Sørensen et al., 2003; Wang et al., 2003; Rhee et al., 2005; Araç et al., 2006). Studying cultured cells provides valuable insight into possible protein functions, yet some of the properties of vesicle fusion in PC12 cells and chromaffin cells are quite different from those mediating fast transmitter release at a synapse. Even cultured neurons do not necessarily faithfully reproduce all aspects of synaptic behavior at an intact synapse. For example, at cultured hippocampal autapses, the amplitude of mEJPs is larger (Bekkers et al., 1990), the paired-pulse ratio is decreased (Sippy et al., 2003), and long-term potentiation cannot be reliably elicited (Bekkers and Stevens, 1990) compared with recordings from hippocampal slices. Additionally, synaptotagmin mutant autapses exhibit no change in mEJP frequency (Geppert et al., 1994), whereas synaptotagmin mutations increased the mEJP frequency at several other synapses, including excitatory and inhibitory synapses between cultured cortical neurons, the calyx of Held, and mature neuromuscular junctions in Drosophila and mice (DiAntonio and Schwarz, 1994; Pang et al., 2006a,c) (but see Marek and Davis, 2002). Therefore, it is critical to test the function of the basic residues at the tips of the C₂ domains at an intact synapse to determine their role in synaptic transmission in vivo.

The C₂A and C₂B domains are structurally highly homologous and exhibit many similar biochemical interactions in vitro (Geppert et al., 1991; Sutton et al., 1995; Chae et al., 1998; Shao et al., 1998; Ubach et al., 1998; Fernandez et al., 2001; Cheng et al., 2004). Analyses of these interactions have provided critical insights into the molecular mechanisms mediating synaptic vesicle fusion. Conditions in vitro, however, can affect biochemical interactions. Interactions between the C₂A basic residue and negatively charged phospholipids can differ when isolated C₂A domains are used: Fernández-Chacón et al. (2001) found that the Ca²⁺ affinity of the interaction was decreased in syt^A-RQ mutants, whereas Zhang et al. (1998) found no change. But studies using tandem C₂AB domains all show a decrease in Ca²⁺-dependent phospholipid binding (Fernández-Chacón et al., 2001; Wang et al., 2003; Li et al., 2006). The homologous C₂B residue shows variable results using tandem C₂AB domains: similar to our results in Drosophila (Fig. 6A), Wang et al. (2003) found that this mutation decreased Ca²⁺-dependent interactions with negatively charged phospholipids, whereas Li et al. (2006) found no change. We therefore repeated these experiments using the C₂AB constructs from Li et al. (2006) and found that the Ca²⁺ affinity of this interaction decreased (Fig. 6B). Thus, Ca²⁺-dependent interactions between these positively charged residues and negatively charged phospholipids appear to be quite sensitive to experimental conditions. Still, most studies find a decrease in Ca²⁺-dependent phospholipid interactions when either the C₂A or C₂B basic residue is neutralized.

Additional studies indicate that Ca²⁺-dependent interactions between each C₂ domain and anionic membranes are conserved and mediated by residues at the tip of the Ca²⁺-binding pockets. Immediately adjacent to the conserved basic residues (Fig. 1, ⊕), there are hydrophobic residues (Fig. 1A, □). In both C₂ domains, these hydrophobic residues interact with phospholipids in a Ca²⁺-dependent manner by inserting into the hydrophobic core of the membrane (Bai et al., 2002). Increasing the hydrophobicity of three residues located around the rim of each Ca²⁺-binding pocket substantially increased the Ca²⁺ affinity of these interactions (Rhee et al., 2005). Together, these results provide strong support for the hypothesis that Ca²⁺-dependent interactions between phospholipids and both the C₂A and C₂B domains are mediated by residues located at the tip of the Ca²⁺-binding pockets.

But are these interactions between synaptotagmin and anionic membranes relevant for synaptic transmission? Results from cultured cells suggest that residues at the tip of the C₂A Ca²⁺-binding pocket maybe functionally significant; the syt^A-RQ mutation decreased evoked transmitter release by decreasing the apparent Ca²⁺ affinity of release (Fernández-Chacón et al., 2001; Sørensen et al., 2003; Wang et al., 2003). The homologous mutation in C₂B has been examined twice in culture with directly contradictory results. At cultured hippocampal autapses, this mutation showed no decrease in Ca²⁺-evoked release (Li et al., 2006). However, in cultured PC12 cells, the C₂B mutation decreased Ca²⁺-evoked release (Wang et al., 2003). Thus, the functional relevance of the C₂B interaction remained inconclusive.

To determine whether the results from cultured cells are relevant for synaptic transmission in vivo, we tested the function of the basic residue at the tip of each Ca²⁺-binding pocket at intact neuromuscular junctions. We found that the syt^A-RQ mutation decreases the Ca²⁺ affinity of release at the neuromuscular junction. Importantly, we found that the syt^B-RQ mutation also decreases Ca²⁺-evoked transmitter release by decreasing Ca²⁺ affinity. Western analysis and immunohistochemical localization studies demonstrate approximately equal levels of transgene expression and appropriate synaptic localization in multiple, independent mutant and control lines. Therefore the decrease in evoked release is a direct result of the mutation. Thus, both the C₂A and C₂B positively charged residues mediate Ca²⁺-dependent interactions with anionic phospholipids and are required for efficient evoked transmitter release at intact synapses.

Our results are consistent with those from PC12 cells expressing the mutant C₂B protein (Wang et al., 2003). When using high K⁺ to trigger release, these authors found that the rate of evoked release was decreased by ∼50% in both the C₂A and C₂B basic residue mutations. Interestingly, the cumulative amount of release was lower in C₂B mutants than in C₂A mutants, although this effect was not quantified. Yet our results at intact synapses, like the results from PC12 cells, are in direct contrast with the lack of effect seen at hippocampal autapses (Li et al., 2006).

Importantly, the replacement of three hydrophobic residues around the rim of either the C₂A or the C₂B Ca²⁺-binding pocket with residues of increased hydrophobicity increased both the Ca²⁺ affinity of phospholipid binding in vitro and the Ca²⁺ affinity of evoked release at autapses (Rhee et al., 2005). Thus, even at hippocampal autapses, the tip of the C₂B Ca²⁺-binding pocket can interact with phospholipid membranes during synaptic transmission.

Our results indicate that the function of this basic residue is conserved in each C₂ domain. Yet other studies have discovered dramatic functional differences between the C₂ domains during synaptic transmission. The most striking difference is the relative importance of Ca²⁺ binding in triggering vesicle fusion. Mutations within the C₂A Ca²⁺-binding motif resulted in either subtle or no disruptions in evoked release (Fernández-Chacón et al., 2002; Robinson et al., 2002; Stevens and Sullivan, 2003; Pang et al., 2006b). Yet mutations within the C₂B Ca²⁺-binding motif inhibited evoked release by up to 99% (Mackler and Reist, 2001; Nishiki and Augustine, 2004; Tamura et al., 2007). Thus, Ca²⁺ binding by the C₂B domain plays the crucial role in triggering fast, synchronous vesicle fusion.

The key difference between the C₂ domains likely resides within their polylysine motifs. The C₂B polylysine motif mediates a unique set of interactions that are not shared by the C₂A domain, including Ca²⁺-independent interactions with phosphatidylinositol 4,5-bisphosphate (PIP₂) and soluble N-ethylmaleimide-sensitive factor attachment receptor (SNARE) proteins (Zhang et al., 2002; Bai et al., 2004; Rickman et al., 2004). Because the polylysine motif is on the side of the C₂B domain, interactions with the SNARE proteins could hold the C₂B Ca²⁺-binding pocket in close proximity to the presynaptic membrane, priming vesicles for immediate fusion after Ca²⁺ influx (Fig. 7A) (Loewen et al., 2006). Then, after Ca²⁺ influx, the conserved basic residues of each C₂ domain mediate similar Ca²⁺-dependent interactions with anionic phospholipids (Fig. 6) (Wang et al., 2003), which pulls the synaptic vesicle toward the membrane (Fig. 7B). The insertion of hydrophobic residues on the tips of the C₂ domains may destabilize the presynaptic membrane, aiding the fusion reaction (Bai et al., 2002; Martens et al., 2007). The relative positioning of the C₂A versus C₂B domains with respect to the SNARE complex may determine the relative importance of the Ca²⁺-binding sites, and the basic residues reported in this study, for triggering vesicle fusion in vivo. This model is consistent with the data from multiple in vitro and in vivo systems, and a remarkably similar model, postulating a more distally positioned C₂A domain, has recently been proposed (Dai et al., 2007).

Figure 7.

Model of the role played by the conserved basic residues in Ca²⁺-dependent interactions with the anionic presynaptic membrane. The crystal structure of the core complex [PDB file 1SFC, containing syntaxin (red), SNAP-25 (green), and VAMP (vesicle-associated membrane protein)/synaptobrevin (blue)], the nuclear magnetic resonance structures of the C₂A (PDB file 1BYN) and C₂B (PDB file 1K5W) domains of synaptotagmin (yellow), and Ca²⁺ (pink) are shown to scale using PyMOL Molecular Graphics System (DeLano Scientific). The membranes, the transmembrane domains, and the link between C₂A and C₂B were added in Adobe Photoshop. Left, A Ca²⁺-independent priming interaction between the C₂B polylysine motif (yellow, space-filled residues) and SNAP-25 (green, space-filled residues) holds the C₂A and C₂B Ca²⁺ binding sites in close proximity to the presynaptic membrane. We diagrammed the interaction between the C₂B polylysine motif and SNAP-25 within the SNARE complex (Zhang et al., 2002; Rickman et al., 2004, 2006; Dai et al., 2007), because mutation of the C₂B polylysine motif impairs this interaction (Bai et al., 2004) and disrupts synaptotagmin's ability to increase the speed of SNARE-mediated liposome fusion in the absence of any PIP₂ (Loewen et al., 2006). However, an interaction with PIP₂, which is located specifically in the presynaptic membrane, could also serve this purpose (Bai et al., 2004; Araç et al., 2006). In the absence of Ca²⁺, the high concentration of negative charge in the Ca²⁺-binding pockets repulses the negatively charged presynaptic membrane, preventing synaptotagmin's conserved basic residues (blue, space-filled residues) from interacting with the membrane. Right, After Ca²⁺ entry, the negative charge of the Ca²⁺-binding pockets is neutralized by the bound Ca²⁺, which initiates the electrostatic switch: a strong attraction of the negatively charged, phospholipid head groups by the bound Ca²⁺ and the basic residues at the tips of Ca²⁺-binding pockets that draws the synaptic vesicle (SV) toward the presynaptic membrane (PM). Insertion of the hydrophobic residues (gray, stick residues) at the tips of the C₂ domains into the core of the presynaptic membrane (Chapman and Davis, 1998) may help trigger fusion by promoting a local Ca²⁺-dependent buckling of the plasma membrane (Chapman and Davis, 1998; Martens et al., 2007). The Ca²⁺-induced increase in positive charge at the end of the C₂B domain also likely increases the strength of the electrostatic interaction between the C₂B polylysine motif and the SNARE complex, resulting in simultaneous binding of the SNARE complex and the presynaptic membrane (Davis et al., 1999; Bhalla et al., 2006; Loewen et al., 2006; Dai et al., 2007).

In summary, synchronous vesicle fusion triggered by Ca²⁺ requires the coordinated interactions of many presynaptic molecules. Examination of isolated interactions in vitro provides insight into possible molecular mechanisms for fusion; however, only the analysis of synaptic transmission at intact synapses can determine which interactions are likely to function in vivo. Our findings demonstrate that the positively charged residue located at the tip of each Ca²⁺-binding pocket mediates Ca²⁺-dependent interactions with negatively charged membranes and is required for efficient synaptic transmission. These findings indicate that the function of this region of C₂A and C₂B is likely conserved and support the hypothesis that Ca²⁺-dependent interactions between the tips of each C₂ domain with anionic phospholipids are requisite for efficient excitation–secretion coupling during synaptic transmission.