Drosophila studies support a role for a presynaptic synaptotagmin mutation in a human congenital myasthenic syndrome

During chemical transmission, the function of synaptic proteins must be coordinated to efficiently release neurotransmitter. Synaptotagmin 2, the Ca2+ sensor for fast, synchronized neurotransmitter release at the human neuromuscular junction, has recently been implicated in a dominantly inherited congenital myasthenic syndrome associated with a non-progressive motor neuropathy. In one family, a proline residue within the C2B Ca2+-binding pocket of synaptotagmin is replaced by a leucine. The functional significance of this residue has not been investigated previously. Here we show that in silico modeling predicts disruption of the C2B Ca2+-binding pocket, and we examine the in vivo effects of the homologous mutation in Drosophila. When expressed in the absence of native synaptotagmin, this mutation is lethal, demonstrating for the first time that this residue plays a critical role in synaptotagmin function. To achieve expression similar to human patients, the mutation is expressed in flies carrying one copy of the wild type synaptotagmin gene. We now show that Drosophila carrying this mutation developed neurological and behavioral manifestations similar to those of human patients and provide insight into the mechanisms underlying these deficits. Our Drosophila studies support a role for this synaptotagmin point mutation in disease etiology.


Introduction
Understanding the mechanisms mediating information transmission across a chemical synapse is essential to understanding brain function. During  vesicles filled with neurotransmitter must fuse with the presynaptic membrane in response to Ca 2+ influx, thereby releasing the transmitter into the synaptic cleft. The synaptic vesicle protein, synaptotagmin, is essential for this release [1][2][3] as it binds the Ca 2+ that triggers vesicle fusion [4][5][6]. Synaptotagmin has two Ca 2+ -binding pockets, C 2 A and C 2 B [7][8][9], which have both been extensively studied for their role in synchronizing vesicle fusion in vivo [4,6,[10][11][12] and in culture [13,14]. Synaptotagmin 2 is the predominant isoform found at mammalian neuromuscular junctions [15,16], and until recently, has never been linked to disease. However, whole-exome sequencing techniques have now identified point mutations in the human synaptotagmin 2 (syt2) gene in patients with a disease combining congenital myasthenic syndrome and distal motor weakness, suggesting a role for synaptotagmin in neuromuscular disease [17,18]. Two independent mutations in the C 2 B Ca 2+ -binding pocket of syt2 have been identified: one in a United States (US) family and one in a United Kingdom (UK) family. Patients in each of the two families exhibit a novel neuromuscular syndrome characterized by foot deformities, fatigable ocular movements, lower limb weakness, and potentiation of tendon reflexes following exercise [17,18]. In nerve conduction studies, all patients exhibited normal sensory responses. However, motor responses revealed low amplitude compound muscle action potentials (CMAPs) that exhibited significantly less amplitude depression following brief maximum voluntary contraction [17,18]. These findings are indicative of a presynaptic deficit of neuromuscular junction function.
Interestingly, both familial mutations are located in C 2 B Ca 2+ -binding pocket of syt2 within the highly conserved "SDPYVK" residue motif [Fig 1, [17,18]]. The US family possesses an aspartate 307 (referred to as D2) to alanine substitution within this motif (Fig 1D, first Ã ). As D2 is one of the well-ordered C 2 B aspartate residues known to coordinate the binding of Ca 2+ [9,19,20] required for triggering neurotransmitter release [4,11,14], the function of this residue has been extensively studied. The UK family possesses a proline 308 to leucine substitution within this motif (Fig 1D, second Ã ). We investigated the anticipated conformational changes of the proline to leucine (P-L) substitution by modeling this P-L mutation in silico. Simulation of the P-L mutation predicts that replacement of the highly conserved proline residue by a leucine would extend β-strand 3 into the Ca 2+ -binding pocket (Fig 1A and 1B small arrow). The extension of β-strand 3 is accompanied by a significant distortion of the backbone angles of residues on the amino terminal side of the proline (S1 Fig). The distortion of upstream residues places this critical D2 Ca 2+ -binding residue in a non-ideal conformation for Ca 2+ binding ( Fig 1C curved arrow, S1 Fig). Further, the additional amide proton from the mutant leucine residue forms a new H-bond with neighboring residues in loop 2 (not shown), thereby altering the shape, and likely the flexibility, of the Ca 2+ -binding pocket. A more rigid Ca 2+ -binding pocket, as well as the sub-optimally positioned D2 Ca 2+ -binding residue, predict deficiencies in Ca 2+ binding in the P-L mutation. Together, our modeling predicts that this mutation would result in conformational changes in synaptotagmin's C 2 B domain that would impair Ca 2+ binding. Although this proline residue is highly conserved (Fig 1D), its importance for synaptotagmin function has not been investigated previously.
To directly assess whether the syt P-L point mutation impacts Ca 2+ -binding resulting in synaptic deficits in vivo, we generated a homologous P-L mutation in the synaptotagmin isoform expressed at the Drosophila neuromuscular junction (syt P-L ). In mammals, the synaptotagmin 1 gene (syt1) is predominantly expressed throughout the cerebral hemispheres while syt2 is predominantly expressed in the brainstem and spinal cord, and hence at the neuromuscular junction [15,16]. There is no syt2 gene in Drosophila [21,22]; the syt1 gene is expressed panneuronally [23,24] and codes for the Ca 2+ sensor for fast, synchronous neurotransmitter release [4,5] throughout the nervous system. When the syt P-L mutation is expressed in Drosophila in the absence of wild type synaptotagmin, it results in lethality, demonstrating that this proline residue is critical for synaptotagmin function. As the human condition presents as a syt heterozygous dominant disorder, with patients carrying one copy of the wild type syt gene and one copy of the mutated syt gene (syt WT /syt P-L ), we simulated this expression pattern by driving pan-neuronal expression from one mutant transgene using elavGal4 in syt heterozygotes (+/-;P[syt P-L ]/+). Physiologically, mutant larvae exhibit decreased evoked transmitter release and less synaptic depression during high-frequency stimulus trains, consistent with the human electrophysiological deficits. In addition, the decreased Ca 2+ affinity of release and the associated decreased release probability documented here are consistent with the predicted disruption to the C 2 B Ca 2+ -binding pocket (Fig 1A-1C). Behaviorally, mutant adult flies exhibit less overall activity and a marked increase in fatigability, also consistent with the human phenotype. Together, these results indicate that this proline to leucine substitution impairs Ca 2+ binding and support a role for this substitution in the etiology of this human neuromuscular disease. The proline residue under investigation (P) and the adjacent, Ca 2+ -coordinating aspartate (D2) are shown as stick models. Note that the proline is at one end of β-strand 3 (small arrow) and the aspartate is in loop 1. (B) Predicted structural changes induced by the syt P-L substitution. The leucine (L) and the adjacent aspartate (D2) residues are shown as stick models and both are now within β-strand 3 (small arrow indicates extension of β-strand 3). The large arrow indicates a newly-formed β-strand. (C) Overlay of wild type (yellow) and syt P-L mutant (turquoise). The curved arrow indicates the displacement of the D2 Ca 2+ -binding residue. (D) Sequence alignment of the highly conserved SDPYVK amino acid motif of the synaptotagmin C 2 B domain from humans, rat, zebra fish, squid, Drosophila, and C. elegans. * locations of the point mutations in the US and UK families, respectively. Note: syt1 and syt2 in mammals are identical in this region. https://doi.org/10.1371/journal.pone.0184817.g001

Results
In the absence of wild type synaptotagmin, the P[syt P-L ] mutation increases the lethality rate To assess the effects of the syt P-L mutation in the absence of the wild type protein, we drove pan-neuronal expression of syt transgenes in the syt null background using the elavGal4 promoter (referred to as -/-;P[syt WT ]/+ or -/-;P[syt P-L ]/+, see methods for full genotype). Rates of survival to the first instar larval stage were also compared to larvae lacking any synaptotagmin 1 expression (-/-). Balancer chromosomes used in our crossing scheme result in embryonic lethality for any wild type synaptotagmin homozygotes (+/+ for syt, included in Table 1 under unhatched) due to the presence of a homozygous lethal Cy mutation (see methods and S2 Fig) with an expected frequency of~25%. Table 1 depicts percentages and raw counts (in parentheses) of first instar larvae in the syt null (-/-, top row) and syt heterozygous (+/-, second row) backgrounds and embryonic lethality (unhatched) of progeny containing one copy of synaptotagmin transgenes (P[syt WT ] or P[syt P-L ]) or of progeny containing no transgenic syt (last column). All crosses were carried out using the CyO second chromosome balancer such that syt homozygotes are lethal (included in unhatched) due to the presence of the homozygous lethal Cy mutation (syt WT Cy/ syt WT Cy). There was a significant decrease in the percentage of first instars expressing one copy of P[syt P-L ] in the absence of any wild type synaptotagmin (-/-;P[syt P-L ]/+, 12.0%) compared to control (-/-;P[syt WT ]/+, 25.7%, p < 0.0001, Chi-squared test). There was also a significant decrease in the percentage of first instars expressing one copy of the P[syt P-L ] transgene (-/-;P[syt P-L ]/+, 12.0%) compared to no synaptotagmin expression at all (-/-; no transgenic syt, 18.0%, p < 0.0001, Chi-squared test).
As expected for a fully functional protein, the wild type transgenic cross resulted in progeny where~25% were syt null mutants expressing synaptotagmin from the wild type transgene (-/-; P[syt WT ]/+; Table 1, 25.7%). This is significantly larger than in the P[syt P-L ] cross, in which only 12.0% of progeny were syt null mutants expressing the mutated transgene (-/-;P[syt P-L ]/+, Table 1, 12.0%, p < 0.0001). As previously shown, syt null mutants are viable [25], and the cross lacking any source of synaptotagmin resulted in 18.0% of progeny that were syt null mutants (Table 1, no transgenic syt, 18.0%). Yet the expression of the mutant transgene resulted in a significant decrease in the rate of survival to the 1 st instar larval stage (Table 1, 18.0% down to 12.0%, p < 0.0001). Thus, in the absence of wild type synaptotagmin, the expression of P [syt P-L ] not only fails to rescue synaptotagmin function, it results in a negative impact on survival. This is the first study to show the critical nature of this proline residue for synaptotagmin function.

Transgenic synaptotagmin is expressed appropriately
To verify appropriate transgenic synaptotagmin expression levels for modeling this disease in Drosophila, western blot analysis was performed on the central nervous system (CNS) of larvae. Comparison of CNS samples from syt heterozygotes (+/-) and syt heterozygotes expressing  Drosophila studies implicating a presynaptic synaptotagmin mutation in a congenital myasthenic syndrome the transgenic wild type and transgenic mutant protein are appropriately targeted to the synapse (Fig 2D-2G). As seen in wild type Drosophila [23,24], synaptotagmin staining in muscles is restricted to the neuromuscular junction in third instar larvae of both the P[syt WT ] heterozygote controls and the P[syt P-L ] heterozygotes ( Fig 2D and 2E, respectively). Staining of first instar larvae expressing the syt transgenes in the absence of native synaptotagmin also revealed that the transgene is appropriately expressed at the neuromuscular junction (

Lifespan is decreased in P[syt P-L ] heterozygotes
While P[syt P-L ] heterozygotes are easily generated and maintained throughout all larval and adult stages, we wanted to ascertain whether this mutation had any impact on lifespan in heterozygotes. Therefore, we performed a longevity assay. Lifespans between P[syt WT ] heterozygotes (51.4 ± 1.50 days, mean lifespan ± SEM, n = 151) and P[syt P-L ] heterozygotes (39.2 ± 1.25 days, mean lifespan ± SEM, n = 149) are significantly different (Fig 3, p < 0.0001, Wilcoxon Rank-Sum test). However, this does not relate directly to the human condition, who show no indication of this P-L mutation impacting longevity, as these flies do not receive any medical intervention. These results support a detrimental role of the proline to leucine mutation in survival.
The syt P-L mutation disrupts evoked transmitter release In human motor nerve conduction studies, the summated response of all muscle fibers supplied by a particular motor nerve can be measured as a CMAP. The syt P-L -affected family members present with decreased CMAP amplitudes [17,18], consistent with a decrease in evoked transmitter release. To test evoked release in Drosophila P[syt P-L ] heterozygotes, we measured excitatory junction potentials (EJPs) at the larval neuromuscular junction. Upon nerve stimulation, P[syt P-L ] heterozygotes exhibited a significant decrease in the amount of neurotransmitter released (Fig 4A and 4B). Average EJP ± SEM for +/-;P[syt WT ]/+ was 34.8 ± 0.83 mV, and for +/-;P[syt P-L ]/+ was 29.6 ± 1.5 mV (p = 0.002, student t-test). This decrease in release is consistent with the decrease in baseline amplitude in the human motor nerve conduction studies. One potential explanation for this observed 20% decrease in evoked response could be a decrease in quantal content. To determine quantal size, we calculated the mean amplitudes of Ca 2+ -independent spontaneous miniature EJPs (mEJPs) in both P[syt P-L ] heterozygotes and their controls. Each mEJP results from one vesicle fusing with the presynaptic membrane. Thus, the amplitude of the response is an indication of the amount of neurotransmitter loaded into each vesicle. Neither the mean amplitude nor frequency of mEJPs in controls and P[syt P-L ] heterozygotes were significantly different (Fig 4C-4E). Mean mEJP amplitude ± SEM for P[syt WT ] heterozygotes was 0.96 ± 0.03 mV (n = 50 mEJP events from 19 fibers), and for P[syt P-L ] heterozygotes was 1.06 ± 0.05 mV (n = 50 mEJP events from 12 fibers, p = 0.09, Wilcoxon Rank-Sum Test). Mean mEJP frequency ± SEM for P[syt WT ] heterozygotes was 1.55 ± 0.15 Hz (n = 19 fibers), and for P[syt P-L ] heterozygotes was 1.94 ± 0.26 Hz (n = 12 fibers, p = 0.18, student ttest). These results demonstrate that the decrease in evoked release is not due to a change in quantal content.
Transmitter release from an activated nerve terminal is a Ca 2+ -dependent, cooperative process [26]. Evoked release was measured at a variety of extracellular Ca 2+ concentrations, ranging from 0.05 mM to 5 mM, to assess whether the Ca 2+ dependence of release was altered in P [syt P-L ] heterozygotes. At all Ca 2+ concentrations above 0.1 mM, P[syt P-L ] heterozygotes exhibit a decrease in evoked release compared to controls (Fig 5A). A nonlinear regression analysis determined a change in the Ca 2+ dependence of P[syt P-L ] heterozygotes. The extracellular Ca 2+ concentration at which a 50% maximum response is reached (EC 50 ) is statistically shifted in P[syt P-L ] heterozygotes compared to controls. In +/-;P[syt WT ]/+ EC 50 = 0.33 mM Ca 2+ (95% confidence intervals from 0.30-0.36 mM Ca 2+ ) and in +/-;P[syt P-L ]/+ EC 50 = 0.44 mM Ca 2+ (95% confidence intervals from 0.41-0.48 mM Ca 2+ , Fig 5B). The non-overlapping confidence intervals demonstrate that the proline to leucine substitution in synaptotagmin results in a significant decrease in the Ca 2+ affinity of release.

Release probability is decreased in P[syt P-L ] heterozygotes
A decrease in the size of the readily releasable pool and/or a decrease in release probability for individual vesicles could result in decreased evoked transmitter release. To determine whether P[syt P-L ] heterozygotes exhibit alterations in the size of their readily releasable pool of synaptic vesicles, we triggered Ca 2+ -independent vesicle fusion events using a hypertonic solution (0.5 M sucrose). Before sucrose stimulation, mEJP frequency is similar between both P[syt WT ] and P[syt P-L ] heterozygotes [ Fig 5C filled symbols, +/-;P[syt WT ]/+ (black), 1.95 ± 0.07 Hz and +/-;P[syt P-L ]/+ (red), 1.81 ± 0.10 Hz, mean mEJP frequency ± SEM, n = 11 fibers for both genotypes, p = 0.28, student t-test). Increases in mEJP frequencies are observed in both controls (p < 0.0001, student t-test) and P[syt P-L ] heterozygotes (p < 0.0001, student t-test) upon sucrose stimulation. These mEJP frequencies are also similar between genotypes during the maximum sucrose response period ( To address release probability, we used a paired-pulse analysis examining the ratio of the second response divided by the first response using a 20 msec interpulse interval. 1 ± 1.02 mV, second pulse = 30.6 ± 0.84 mV, EJP amplitude ± SEM), p = 0.0008, student t-test]. The significantly larger paired pulse ratio indicates a decrease in release probability in P[syt P-L ] heterozygotes. Thus, our electrophysiological analysis of P[syt P-L ] heterozygotes indicates the decrease in evoked transmitter release is not due to changes in the readily releasable pool, but rather results from a decrease in vesicular release probability. A distinctive feature of the human condition is that following brief maximal voluntary muscle contraction, the CMAP amplitude is less depressed, typically for several minutes [17,18]. Since a decrease in release probability could result in less depression, we assessed this unique phenotype in Drosophila. We measured the amplitude of evoked release at the larval neuromuscular junction following a 10 Hz stimulation for 2 seconds. Responses were normalized to the initial EJP response to account for the decrease in EJP amplitude seen in the P[syt P-L ] heterozygotes. Beginning at the fourth pulse of this stimulation period, P[syt P-L ] heterozygotes exhibited a significant 5-10% decrease in synaptic depression relative to control (Fig 6, S2  Table). However, when a test stimulation was applied 1 minute after the end of the stimulus train, this effect was no longer observed; the amplitude of the evoked response in both the P [syt P-L ] heterozygotes and the controls had returned to resting levels (Fig 6, S2 Table).
Since maximum voluntary contraction can produce motor nerve firing rates in excess of 50 Hz [27], larval preparations were stimulated at 50 Hz for 2 seconds in an attempt to more closely approximate the stimulus that results in the prolonged increase in release seen in the affected family members. Normalized post-stimulus train responses were tested at 1 second and then at 30-second intervals out to 4 minutes. This protocol displayed similar results to the 10 Hz stimulation experiment. Both at the end of the stimulus train and at 1 second post-stimulation, there is no synaptic depression in the P[syt P-L ] heterozygotes, unlike in the control (Fig 7, S3 Table). However, normalized responses to test stimulations at 30 seconds or longer after cessation of the stimulation train showed no significant differences in release between the  Table). Thus, our transgenic model system cannot provide insight into the prolonged increase of the CMAP amplitude following maximal voluntary contraction, which can last 10 minutes or longer in humans [17,18].

Basal levels of locomotor activity are decreased in P[syt P-L ] heterozygotes
The P-L familial condition presents with distal limb deformities, muscle wasting, and difficulty walking [17,18]. To assess the impact of this mutation on overall motor function in our model system, the Drosophila activity-monitoring (DAM) assay was performed to quantify basal locomotor activity levels. This assay determines total activity over time. However, as activity levels in Drosophila males and females are known to have different circadian cycles [28,29], males and females were tested separately. Activity level was determined by monitoring individual flies for 6 days in a 12hr light: 12hr dark cycle (Fig 8A and 8B) in age-and sex-matched adult Drosophila P[syt WT ] or P[syt P-L ] heterozygotes.
Drosophila are typically active during periods of light and inactive during periods of dark. Furthermore, flies have an increase in locomotor activity in anticipation of the transition between dark to light and light to dark. We analyzed the locomotor activity at the two anticipatory peaks of activity (Active 1 and 2, respectively), the non-peak period during lights on (Active 3) and the inactive period approximately corresponding to lights off (Inactive). We find distinct periods of activity between males and females, consistent with published data [28,29]. During all activity periods, both male and female P[syt P-L ] heterozygotes displayed significantly less movement (Fig 8C and 8D, S4 Table).

The rate of muscle fatigue is faster in P[syt P-L ] heterozygotes
Affected individuals with the human disease also present with cramping and pain upon physical exertion, difficulty with sports, and some fatigable eye movements [18]. Since these can all be associated with fatigability in muscle, we analyzed fatigue rate in adult Drosophila using a modified climbing assay [30]. Flies were repetitively knocked to the bottom of a vial. This stimulates negative geotaxis, an innate escape response in which flies climb up the vial wall. Vials were dropped (see methods) every 10 seconds for 30 minutes. After each drop, the distance Drosophila studies implicating a presynaptic synaptotagmin mutation in a congenital myasthenic syndrome each fly climbed upward within 5 seconds was measured and averaged. As with the DAM assay, we tested age-matched female and male adult flies separately. Differences were observed as early as the first drop for both males and females of P[syt P-L ] heterozygotes and controls (female +/-;P[syt WT ]/+ = 6.49 ± 0.60 cm, female +/-;P[syt P-L ]/+ = 3.67 ± 0.91 cm, p = 0.017, student t-test, male +/-;P[syt WT ]/+ = 6.99 ± 0.45 cm, and male +/-;P[syt P-L ]/+ = 4.68 ± 0.64 cm, p = 0.009, student t-test, average distance traveled ± SEM after the first drop). To specifically measure fatigue, we controlled for the decreased overall activity in the mutants by normalizing all measurements to the distance traveled after the initial drop. The normalized data was then graphed over time for both females and males (Fig 9A and 9B, respectively). As a measure of the rate of fatigue, we calculated the log of the raw data and used linear regressions to provide predicted lines of best fit (not shown). The slopes of these lines were statistically significantly different between mutants and controls in both females and males (slopes for female +/-;P

Discussion
In this study we have investigated a synaptotagmin point mutation that was implicated in the etiology of a congenital myasthenic syndrome found in a single family. By modeling this point mutation in silico, we predict significant conformation changes to a key, well-ordered Ca 2+binding residue essential for fast, synchronous neurotransmitter release. The role of this conserved proline residue in these C 2 domains could be to punctuate the end of the β-strand leading into the Ca 2+ -binding pocket thereby providing maximum Ca 2+ binding volume optimizing Ca 2+ sensitivity.
We expressed the homologous mutation in Drosophila and documented several similar deficits to those seen in the affected patients. In addition, we found a decreased Ca 2+ affinity for release and decreased release probability that are consistent with the conformational changes predicted by our in silico analysis. Notably, this syt P-L mutation not only fails to rescue synaptotagmin function, it has a negative impact on survival when expressed as the sole source of synaptotagmin (-/-;P[syt P-L ]/+); the survival rate is significantly lower than that of larvae with no synaptotagmin expression (-/-). This finding demonstrates for the first time that this proline residue in the C 2 B Ca 2+ -binding pocket is essential for synaptotagmin's Ca 2+ -sensing function. By driving expression of the mutant transgene in syt heterozygotes (+/-;P[syt P-L ]/+), we were able to achieve expression levels of syt WT and syt P-L similar to that in the human disease.
Electrophysiological analyses demonstrated deficits that may help inform the human condition. Consistent with the human nerve conduction tests, the EJP amplitude at the larval neuromuscular junction in P[syt P-L ] heterozygotes is significantly decreased. This decrease in the initial evoked response is not due to a decrease in quantal size or the readily releasable pool. P [syt P-L ] heterozygotes exhibit a decreased Ca 2+ affinity, as documented by the increased EC 50 for Ca 2+ , which resulted in a decreased release probability, as documented by the facilitated responses in paired pulse experiments, and either no, or less, depression during high frequency stimulations. This decreased probability of release could account for an increased failure rate at the neuromuscular junction, consistent with the decreased CMAP seen in humans.
The P[syt P-L ] mutation resulted in significant electrophysiological deficits when expressed in a heterozygous background. This is notable, as synaptotagmin heterozygotes previously have been shown to exhibit no significant changes in electrophysiological responses compared to synaptotagmin homozygotes [31]. Although we did not express this mutation in an otherwise wild type background, this would be consistent with a dominant negative phenotype.
This proline to leucine mutation presents with behavioral deficits similar to those in patients, most notably decreased movement and increased fatigability. This similarity of multiple deficits found in the familial disease strongly supports a role for this residue in the etiology of this form of congenital myasthenic syndrome. It is only recently that mutations in the syt1 and syt2 genes have been implicated in human disease [17,18,32]. A patient with a de novo mutation in the human syt1 gene presented with an early onset dyskinetic movement disorder, severe motor delay, and profound cognitive deficits attributed to a single amino acid substitution in the C 2 B Ca 2+ -binding pocket: an isoleucine to threonine (I-T) substitution [32]. This hydrophobic isoleucine residue had been shown previously to mediate Ca 2+ -dependent membrane penetration by synaptotagmin [33][34][35][36]: an effector interaction critical for coupling Ca 2+ influx with neurotransmitter release from neurons [37,38]. Since this syt mutation results in the most severe deficits in an animal system [37] and is located in the syt1 gene, which is preferentially expressed in the cerebral hemispheres [15,16], it is not surprising that this patient experienced extreme cognitive deficits as well as the most severe motor deficits observed in a synaptotagmin human disease to date [32].
Herrmann et al, 2014 reported two mutations in the human syt2 gene implicated in the etiology of congenital myasthenic syndrome patients: the proline to leucine (syt P-L ) mutation analyzed in this paper and an adjacent aspartate to alanine (syt D-A ) mutation. Both of these highly conserved residues are located in synaptotagmin's C 2 B Ca 2+ -binding pocket [9,19,39]. While the function of this proline residue had not been previously studied in animal models, previous studies have demonstrated that the Ca 2+ -binding aspartate residues of the C 2 B domain are essential for synaptotagmin function [4,14,17,35], at times resulting in lethality in animal models [4,17]. The single viable Drosophila line from our lab investigating C 2 B aspartate mutations had somewhat reduced expression of the mutant transgene [4]. Since these mutations result in a dominant negative decrease in evoked transmitter release (worse than no synaptotagmin at all) [4], the decreased expression level in our line may have contributed to its viability. When driven in the presence of native synaptotagmin, our Drosophila lines that disrupted Ca 2+ binding by C 2 B were viable, but they did result in a significant dominant negative knockdown in function [4]. For the syt D-A familial mutation, expression of the mutant syt from a transgene in syt heterozygotes resulted in lethality in 4 of 6 independent transgenic lines. There was a dramatic decrease in the evoked response in the remaining two lines [17]. The syt P-L mutation investigated here, however, resulted in no viability issues in the heterozygous background, and only a 20% decrease in the evoked response. Thus, it is surprising that both the syt P-L and the syt D-A human families present with such similar symptoms [17,18]. Given the severe consequences of mutating the C 2 B Ca 2+ -binding aspartate residues in animal models, it is noteworthy that the deficits seen in the affected human family are comparatively mild [17,18].
It has been shown that Ca 2+ -binding by the aspartate residues [9,19,35] as well as membrane penetration by the hydrophobic residue [33][34][35][36], both located in the C 2 B domain of synaptotagmin, are of critical importance for proper synchronous transmitter release [4,14,17,18,32,37]. Interestingly, the proline residue investigated in this study is located directly adjacent to one of these aspartates residues. We speculate that this proline residue may provide conformational rigidity important for stabilizing the C 2 B Ca 2+ -binding pocket. By mutating the proline to a leucine, the rigid R group found in the proline residue would be lost, potentially affecting the precise conformation of the pocket and impacting the ability of the adjacent aspartate to bind Ca 2+ . Such a mechanism could result in a decreased, albeit not demolished, ability of the C 2 B domain to bind Ca 2+ , resulting in the less severe phenotype documented here compared to both the aspartate and isoleucine mutations in vivo [4,11,17,18,32,37].
In our Drosophila P[syt P-L ] heterozygotes, the impact of high frequency stimulation on EJP amplitude was short-lived. Thus our model system cannot provide insight into the long-lasting decrease in depression of the CMAP seen in the human patients following maximal voluntary contraction. The initial facilitation and following lack of depression observed in our P[syt P-L ] heterozygotes during and immediately after high frequency stimulation is predicted by the residual Ca 2+ hypothesis for short-term plasticity [40][41][42]. The increase in paired pulse facilitation ( Fig 5D) and lack of depression observed during high frequency stimulation (Figs 6 and  7) demonstrate that the syt P-L mutation decreased release probability compared to wild type. Accordingly, the build-up of Ca 2+ that occurs during high-frequency stimulation is predicted to contribute to a relatively increased release probability during and shortly after the stimulus train. As this syt P-L mutation is predicted to disrupt the critical C 2 B Ca 2+ -binding pocket of synaptotagmin (Fig 1A-1C), which is the Ca 2+ sensor for fast, synchronous neurotransmitter release, a decrease in release probability was not unexpected. The note-worthy long-lasting potentiation of the CMAP, which can last up to 60 minutes in human syt P-L patients [18] remains unexplained; mutation of the Ca 2+ sensor would not be predicted to cause such a long-lasting effect.
The decreasing cost and increasing availability of DNA sequencing has led to an increased incidence of genomic sequencing during patient diagnoses. This ever-increasing process will undoubtedly lead to the discovery of additional synaptic mutations and associated genetic disorders. The need for a relatively quick and cost effective method to begin elucidating the molecular mechanisms underlying these synaptic disorders is vital. Using Drosophila to model these newly discovered conditions in highly conserved pathways like synaptic transmission is ideal for elucidating their underlying mechanisms, as Drosophila provide a cost-effective and rapid mechanism for analysis.
In summary, we have identified the functional role of a predicted leucine-induced conformational change in the C 2 B Ca 2+ -binding pocket of synaptotagmin. Expression of the homologous human mutation in Drosophila resulted in deficits similar to many of the human symptoms. Thus, this work supports a role for this synaptotagmin point mutation in disease etiology and elucidates the molecular mechanisms responsible.

Mutagenesis
In Drosophila, the syt1 gene is expressed in all neurons [23,24] and is homologous to both syt1 and syt2 in mammals [21]. Proline residue 363 of Drosophila syt1 is homologous to proline 308 in human syt2, implicated in the human disease, and proline 310 in rat syt1, modeled in Fig 1. Using the Drosophila syt1 coding sequence [43,44], both a wild type control and a P363L mutant cDNA flanked by EcoRI and BglII restriction sites were synthesized by GeneWiz (La Jolla, CA). The synthesized cDNAs were then subcloned into the PUAST-attB vector to place the transgenes under the control of the UAS promoter and the PhiC31 targeted insertion system was used to target the transgene containing vectors to the attP2 landing site on the Drosophila third chromosome [BestGene, Chino Hills, CA; [45,46]]. We refer to the mutant transgene as P[syt P-L ] and the control transgene as P[syt WT ].

Fly lines
Expression of the UAS-driven syt transgenes was localized to the nervous system using a panneuronal source of GAL4 [elavGAL4 [47,48]]. Syt AD4 was the syt null mutation used [44]. Synaptotagmin is found natively on the second chromosome, and the syt transgene used was incorporated into a third chromosome landing site. To directly assess the function of the syt P-L mutation, the impact of transgene expression on survival was assessed in the syt null background. For these experiments, the following genotype yw;syt null P[elavGal4w + ]/CyO,GFPw + was crossed to yw;syt null /CyO,GFP;P[UASsyt WT ,y + w + ] or yw;syt null /CyO,GFP;P[UASsyt P-L y + w + ] or yw;syt null /CyO,GFP to generate: 1) the transgenic control in the syt null background yw; syt null P[elavGAL4w + ]/syt null ;P[UASsyt WT y + w + ]/+ (referred to as -/-; P[syt WT ]/+), 2) the transgenic mutant in the syt null background, yw;syt null P[elavGAL4w + ]/syt null ;P[UASsyt P-L y + w + ]/+ (referred to as -/-;P[syt P-L ]/+), and 3) syt null mutants, yw; syt null P[elavGAL4w + ]/syt null (referred to as -/-), respectively. Expression of the mutant transgene resulted in embryonic or first instar lethality. Therefore, for most experiments, the P[syt P-L ] transgene was expressed in a syt heterozygous background in order to mimic the heterozygous nature of the affected human population (syt P-L /syt WT ). This results in synaptotagmin expression from a single copy of the endogenous syt gene and from a single copy of the mutant syt transgene. For these experiments, the following genotype yw;syt null P[elavGal4w + ]/CyO,GFPw + was crossed to yw;+;P[UASsyt P-L y + w + ] or yw;+;P[UASsyt WT y + w + ] (note: both of these transgenic lines are homozygous for syt WT on the second chromosome) to generate: 4) the transgenic mutant in the syt heterozygous background, yw; syt null P[elavGAL4w + ]/syt WT ;P[UASsyt P-L y + w + ]/+ (labeled as +/-;P [syt P-L ]/+ in the figures and referred to as P[syt P-L ] heterozygotes in the text), and 5) the transgenic control in the syt heterozygous background, yw;syt null P[elavGAL4w + ]/syt WT ;P[UASsyt WT y + w + ]/+ (labeled as +/-;P[syt WT ]/+ in the figure legends and referred to as P[syt WT ] heterozygotes in the text). For western analyses, the additional genotypes used for comparisons were: 6) syt heterozygotes, yw; syt null /syt WT ; P[UASsyt P-L y + w + ]/+ (labeled as +/-in Fig 2A), and 7) syt homozygotes, yw; syt WT /syt WT ; P[UASsyt P-L y + w + ]/+ (labeled as +/+ in Fig 2B). The experimental syt heterozygotes (+/-in Fig 2A) and homozygotes (+/+ in Fig 2B) both contained the P[syt P-L ] transgene but lack the elavGal4w + driver, so the transgene was not expressed.

Molecular dynamics
As the crystal structures of mammalian synaptotagmin 2 proteins have not been solved, we obtained the crystal structures of rat Syt1 C 2 B (PDB file1TJX) [20] from the Protein Data Bank. The C 2 B domains of mammalian synaptotagmin 1 and synaptotagmin 2 are >90% identical (8 substitutions in the core of 93 C 2 B residues) and most of these substitutions are conservative. We generated models of the P-L mutation using the mutagenesis wizard in Pymol Molecular Graphics System, Version 1.8, Schrödinger, LLC. The QuikMD utility in NAMD [49] was used for the simulation. The structures were solvated using the solvate TCL utility in a cube of approximately 7,000 water molecules (TIP3) with a positive and negative padding of 10 Å on each axis. Residue 359 in C 2 B was restrained with harmonic constraints to dampen coordinate drift within the simulation box. Both wild type and mutant C 2 B models were simulated using NAMD [49] with different random seeds for 50,000,000 frames with a time step of two femtoseconds (total of 100 ns) and a write frequency of 25,000. Periodic boundary conditions were used with PME and a grid spacing of 1.0. We used rigid bonds and constant group pressure control. The pressure was controlled with a Langevin piston set at 1.01325 bar (atmospheric pressure). The temperature was controlled at 310 K with Langevin dynamics and a dampening coefficient of 5 ps -1 was applied to each trajectory. We validated that the trajectories reached equilibrium by calculating the root mean squared displacements of all atoms in the protein after alignment over the course of the run. Each unique simulation setup was run three times.

Lethality assay
The only source of synaptotagmin from the native locus (syt WT ) in the crosses used to assess the effect of transgene (P[syt]) expression was located on the CyO,GFP balancer chromosome (see fly lines section above). Therefore, any progeny that were homozygous for native synaptotagmin are categorized as unhatched since they are also homozygous for Cy, which is embryonic lethal (Table 1, S2 Fig). Based on Mendelian genetics, the expected frequency is~25% (S2 Fig). At least three bottle crosses using at least 50 virgin females each were fed on molasses plates supplemented with a dab of yeast/baby food mixture and were left to mate for two days. A fresh molasses plate dabbed with yeast/baby food mixture was then provided and the flies were allowed to lay eggs for 6 hours to provide age-matched progeny. A total of 4476 eggs were collected and observed. First instar progeny were then scored by phenotype, and the number of unhatched eggs was counted. The plates were observed daily for three consecutive days to control for any developmental delays. A chi-squared test between the three genotypes was applied to reject the test of independence. Additional chi-square tests were applied to test for differences between -/-;P[syt WT ]/+ and -/-;P[syt P-L ]/+, as well as between -/-and -/-;P[syt P-L ]/+.

Immunoblotting
Western analysis of larval CNSs was used to determine relative levels of synaptotagmin expression. The CNS was isolated from third instars dissected in Ca 2+ -free HL3.1 containing 70 mM NaCl, 5 mM KCl, 4 mM MgCl 2 , 10mM NaHCO 3 , 5 mM Trehalose, 115 mM sucrose, 5 mM HEPES, pH 7.2 [50]. Samples were sonified for 10 pulses using a Branson Sonifier 450 (VWR Scientific, Westchester, PA) in Laemmli buffer (Bio-Rad, Hercules, CA) containing 5% 2-mercaptoethanol. The indicated genotype was loaded one CNS per lane. Samples were electrophoresed, transferred to Immobilon membranes (Millipore, Bedford, MA), and washed in blocking solution according as described [31]. Blots were probed overnight at 4˚C with an anti-synaptotagmin antibody, Dsyt-CL1 [4], diluted 1:2500 and an anti-actin antibody, MAB 1501 (Millipore Bioscience Research Reagents, Billerica, MA), diluted 1:10,000. Actin levels were used to normalize for equal protein loading. Blots were visualized using an EpiChemi 3 Darkroom and Labworks Imaging software (UVP BioImaging, Upland, CA). The synaptotagmin:actin ratio was determined for each lane, then normalized to the mean synaptotagmin: actin ratio of all control lanes on an individual blot to allow comparison of signal between multiple blots. Outliers in total protein loaded (as indicated by actin levels) were not included in the analysis. Statistical significance for each comparison was determined using a student's ttest.

Immunolabeling
For immunolabeling of the neuromuscular junction, third instars of P[syt WT ] heterozygotes and P[syt P-L ] heterozygotes were dissected in Ca 2+ -free HL3.1 saline to expose their body wall muscles and fixed in phosphate-buffered saline (PBS, 137 mM NaCl, 1.5 mM KH 2 PO 4 , 2.7 mM KCl, 8.1 mM Na 2 HPO 4 ) containing 2% formaldehyde for 1 hour. Whole-mounts were incubated overnight in Dsyt-CL1 diluted 1:400 in dilution media [PBS with 0.1% Triton, 1% bovine serum albumin, and 1% normal goat serum (NGS from Jackson ImmunoResearch, West Grove, PA)], washed in PBST (PBS with 0.1% Triton) for 3 hours, incubated in Alexa Fluor 488 goat-anti-rabbit antibody (Invitrogen, Carlsbad, CA) diluted 1:400 and Texas Red anti-HRP (Jackson ImmunoResearch, West Grove, PA) diluted 1:50 in dilution media for 1 hour, washed in PBST for 1 hour, and mounted in Citifluor (Ted Pella, Redding, CA). To label transgenic synaptotagmin expressed in the syt null background, the above protocol was applied to first instars rather than third instars with the following changes: first instars were fixed in PBS containing 4% formaldehyde for 2 hours and Texas Red anti-HRP was omitted. Neuromuscular junctions at the muscle 6/7 junction were imaged for both first and third instars with a Zeiss 880 light scanning microscope using a 40X objective, and acquired using Zeiss Zen 2.1 acquisition software, version 11,0,3,190.

Longevity assay
Lifespans between +/-;P[syt WT ]/+ and +/-;P[syt P-L ]/+ were recorded from at least three separate crosses using at least 50 virgins in each cross. Lenth's analysis was used to determine adequate sample size. N = 151 for +/-; P[syt WT ]/+ and n = 149 for +/-; P[syt P-L ]/+. Ten newly hatched adult flies/vial were observed/flipped into fresh food every Monday/Wednesday/Friday and number of living or dead flies were recorded. Flies that were found alive but stuck in the food were omitted from analysis. Any flies that escaped during flipping were also omitted from analysis, as there was no way to determine the length of natural life. The Wilcoxon (two-sample) Rank-Sum test was used to determine significance between genotypes.

Excitatory junction potentials
Third instars were dissected in Ca 2+ -free HL3.1 saline. EJPs were evoked from muscle fiber 6 of abdominal segments 3 and 4 in HL3.1 containing 1.0 mM Ca 2+ using standard techniques [25,31,37]. Intracellular electrodes of 10-25 MO contained 3 parts 2M potassium citrate to 1 part 3M potassium chloride. The resting membrane potential of each fiber was maintained at -55 mV by passing no more than 1 nA of current. Nerves were stimulated with a suction electrode filled with HL3.1 containing 1.0 mM Ca 2+ . All events were collected using an AxoClamp 2B (Molecular Devices, Sunnyvale, CA) amplifier and digitized using a Powerlab 4/35 A/D converter (ADInstruments, Sydney, Australia). EJPs were recorded in LabChart software (ADInstruments, Sydney, Australia). For single evoked responses, averages of 10 EJPs collected at 0.04 Hz were calculated for each individual fiber. Statistical significance was determined using a student's t-test. Spontaneous events were identified manually after recordings had been randomized and blinded to the researcher. Average mEJP amplitudes were determined from 50 consecutive mEJP events/fiber, taken from consistent time periods across fibers. Statistical significance for mEJP amplitude was determined by the Wilcoxon Rank-Sum test. Frequency of spontaneous events was determined by manually counting number of mEJP events during one minute of baseline activity/fiber. Each fiber was analyzed for the same time period during the recording to eliminate bias.

Ca 2+ dependence of release
Individual fibers were bathed in HL3.1 saline containing Ca 2+ levels varying from 0.05 mM-5.0 mM. Each fiber was recorded in at least three different Ca 2+ concentrations to be included in analysis. At each Ca 2+ concentration, the mean response of 5 EJPs collected at 0.5 Hz were calculated. Lines of best fit were calculated using a nonlinear regression analysis, which provided EC 50 values and their confidence intervals for each genotype.

Paired pulse
Experiments were conducted using a 0.02 second delay between stimulations, and averages of EJP responses were normalized to the first stimulation. Statistical significance was determined using a student t-test.

High frequency stimulation
Assays to observe depression or facilitation were first assessed by stimulating the nerve at 10 Hz for 2 seconds followed by a test pulse 60 sec post-stimulus train. EJP amplitudes were normalized to that of the first EJP prior to calculating the average response within a genotype. A repeated measures analysis of variance was used to compare the two genotypes at each stimulus. Post-hoc t-tests were then used to determine significance. To more closely mimic maximal voluntary contraction at the human neuromuscular junction, another protocol was designed using 50 Hz stimulation for 2 seconds, followed by test pulses at 1 second post-stimulus train, and then at 30 second intervals to 4 minutes. Similar statistical methods as above were applied to determine significant differences between amplitudes at each individual time point.

Hypertonic solution stimulations
A PicoSpritzer III was used to administer a puff application of either 0.5M sucrose or HL3.1 bath solution to the junctional region of muscle fibers 6 at 5 pounds per square inch (psi) for 10 seconds. Hypertonic stimulations were conducted in Ca 2+ -free HL3.1. mEJP events were counted manually for all time points analyzed. To determine baseline mEJP frequency, 10 seconds of mEJP events immediately prior to sucrose or HL3.1 application were analyzed. To determine average mEJP frequency during maximum sucrose stimulation, 15 seconds of mEJP events were analyzed, beginning 5 seconds into the sucrose or HL3.1 application. Statistical significance was determined by student t-tests.
For all electrophysiological experiments, post-hoc Lenth's analyses were performed to ensure adequate sample size. If fibers were unable to maintain a physiological potential when injected with a maximum of 1 nA of current, they were excluded.

Drosophila activity monitoring
The basal locomotor activity of adult Drosophila was assessed using the TriKinetics Locomotor Activity Monitoring System (Trikinetics, Waltham, MA). Age-matched 5-7 day old adult flies of the indicated genotypes were loaded into DAM2 Drosophila activity monitors. Males and females were tested separately to control for sex differences in circadian rhythms [28,29]. Animals were placed on a 12hr light: 12hr dark cycle on standard fly food media and activity was recorded for 7 days using DAM system Acquisition Software. Data from the first day were not included in the analysis to allow flies to acclimate to their new environment. Activity levels were binned into 30-minute intervals and uploaded into Microsoft Excel using DAM Filescan software. Using these 30-minute intervals, an averaged 24-hour day was created for each genotype and sex. Four distinct activity periods were determined post-hoc that correspond to Drosophila's two anticipatory peaks (Active 1 and 2), a lights-on active period (Active 3), and a lights off-inactive period (Inactive). T-tests were performed to compare genotypes during each activity period for both males and females. Flies that died during the course of the experiment were excluded from the analysis. Post-hoc Lenth's analysis was performed to ensure adequate sample size.

Fatigability assay
To assess fatigue, a modified negative geotaxis assay was used [30]. For each genotype, 10 unanesthetized, sex-and age-matched 5-7 day old adult flies were loaded by aspiration into a 9 cm tall standard Drosophila vial. Flies were not exposed to CO 2 to avoid possible side effects and were left to acclimate for 1 hour. Experimental and control vials were secured side by side in a vial rack, raised 6-8 inches above a table, and then dropped, causing all flies to fall to the bottom of the vial. Vials were dropped every 10 seconds for 30 minutes, and a picture was recorded 5 seconds after every drop. Each picture was analyzed in ImageJ for the distance each fly had traveled up the side of the vial. The average for each sex and genotype was calculated for each picture, and the data was binned into 1 minute averages. To control for any genotype-dependent decrease in mobility and thereby specifically assess fatigue, all average distances were normalized to the average distance traveled after the first drop. Linear regressions of the log of the raw data provided predicted lines of best fit for each sex and genotype. A significant interaction between time and genotype (for both sexes) in the log of the raw data indicates that there is a difference between the slopes for the two genotypes. Fitted curves, after back transformation, are shown in Fig 9. Post-hoc Lenth's analysis was performed to ensure adequate sample size.

Conclusions
In this study, we investigate the functional role of a synaptotagmin 2 residue implicated in disease. Recently, a neuromuscular disorder has been reported in which a highly conserved proline residue is mutated to a leucine within synaptotagmin. The functional importance of this residue was unknown. Here we model the impact of this mutation on the structure of synaptotagmin and examine the effects of a homologous mutation in vivo using Drosophila. We found that this mutation is predicted to produce a conformational change in the C 2 B domain of synaptotagmin that would decrease Ca 2+ binding. When expressed in flies containing no synaptotagmin, this mutation was lethal, demonstrating the critical nature of the examined residue. When expressed in combination with wild type synaptotagmin, we demonstrated a decreased Ca 2+ affinity for neurotransmitter release and a decreased release probability. These deficits can account for the defects in neuromuscular function and behavior, which are similar to those of the human patients. Thus, these results support a role for this amino acid substitution in human disease etiology. 25% of F1 progeny in control crosses (syt homozygotes at the native locus) are expected to remain unhatched, as they are also homozygous for Cy, which is embryonic lethal. The remaining progeny should present as~25% lacking GFP (sytnull at the native locus) and~50% GFP (syt heterozygotes at the native locus). (TIFF) S1  Table. P[syt P-L ] heterozygotes exhibit less synaptic depression relative to the control throughout a 10 Hz stimulation, but fail to maintain this relative increase in release upon cessation of the stimulus train. Table providing normalized mean responses, SEM, and p-values for all points tested during and after a 10 Hz stimulation train (Fig 6), where Ã p < 0.05, and ÃÃ p < 0.001. (DOCX) S3 Table. P[syt P-L ] heterozygotes do not exhibit synaptic depression during and shortly after 50 Hz stimulation, but this increase in release relative to controls is not prolonged. Table providing normalized mean responses, SEM, and p-values during and after a 50 Hz stimulation train (Fig 7), where ÃÃ p << 0.0001, and Ã p < 0.01. (DOCX) S4 Table. P[syt P-L ] heterozygotes display decreased motor output compared to controls. Table providing mean sensor crossings and SEM acquired during the Drosophila Activity Monitoring assay (Fig 8).