Synaptotagmin 7 switches short-term synaptic plasticity from depression to facilitation by suppressing synaptic transmission

Short-term synaptic plasticity is a fast and robust modification in neuronal presynaptic output that can enhance release strength to drive facilitation or diminish it to promote depression. The mechanisms that determine whether neurons display short-term facilitation or depression are still unclear. Here we show that the Ca2+-binding protein Synaptotagmin 7 (Syt7) determines the sign of short-term synaptic plasticity by controlling the initial probability of synaptic vesicle (SV) fusion. Electrophysiological analysis of Syt7 null mutants at Drosophila embryonic neuromuscular junctions demonstrate loss of the protein converts the normally observed synaptic facilitation response during repetitive stimulation into synaptic depression. In contrast, overexpression of Syt7 dramatically enhanced the magnitude of short-term facilitation. These changes in short-term plasticity were mirrored by corresponding alterations in the initial evoked response, with SV release probability enhanced in Syt7 mutants and suppressed following Syt7 overexpression. Indeed, Syt7 mutants were able to display facilitation in lower [Ca2+] where release was reduced. These data suggest Syt7 does not act by directly sensing residual Ca2+ and argues for the existence of a distinct Ca2+ sensor beyond Syt7 that mediates facilitation. Instead, Syt7 normally suppresses synaptic transmission to maintain an output range where facilitation is available to the neuron.

www.nature.com/scientificreports/ mirrored by changes in short-term presynaptic plasticity. High levels of Syt7 enabled robust facilitative responses while loss of Syt7 switched the normally facilitating synapse into one that displayed short-term depression. This work reveals that Syt7 normally reduces synaptic transmission to scale it to an appropriate range where facilitation is allowed, providing a bi-directional switch for short-term synaptic plasticity.

Results
Syt7 switches short-term synaptic plasticity from depression to facilitation. Whole-cell voltage clamp recordings were performed at muscle fiber 6 in Drosophila embryos at hatching stage (21-24 h after fertilization) to record synaptic currents (EPSCs) elicited by stimulation of the glutamatergic motoneurons innervating the muscle. All recordings were done in the background of a null mutation in Myosin heavy chain (Mhc 1 ) to prevent muscle contraction 7 . We stimulated the nerve for 10 pulses at 10 Hz, commonly used in studies of both mammalian central synapses and Drosophila NMJs for mimicking natural communication 11 , at the reported physiological concentration of Ca 2+ (1.5 mM) 12 . As shown in Fig. 1a,b, wild type (WT) embryos display a cumulating increase in response to three consecutive stimuli at the beginning of the 10 Hz stimulation, while Syt7 null homozygous mutant (Syt7 −/− ) 13 embryos show a decreasing response from the larger 1st EPSC than that of WT. The response then reaches a plateau until presumed depletion of the immediate releasable SV pool (IRP) occurs 14 . Strikingly, heterozygotes (Syt7 +/− ) containing only a single copy of Syt7 showed an intermediate phenotype with almost no facilitation or depression, consistent with the intermediate size of the 1st EPSC compared to WT or Syt7 −/− (Fig. 1a,b). The ratio of the 3rd EPSC to 1st EPSC in Fig. 1c demonstrates switching from depression to facilitation with increasing amounts of Syt7. These results suggest Syt7 normally suppresses synaptic transmission to an appropriate range to prevent depletion of the IRP and ensure synaptic facilitation.
To test this hypothesis directly, we overexpressed Syt7 with the elav-GAL4 pan-neuronal driver. Syt7 overexpression suppressed the amplitude of the 1st EPSC compared to WT (Fig. 1a,b) and increased the facilitation ratio by more than sixfold compared with elav-GAL4 controls (Fig. 1d). These results indicate the levels of Syt7 set the initial magnitude of presynaptic output, with normal levels of Syt7 suppressing synaptic transmission to a range where facilitation can occur.
Syt7 null mutants show enhanced nerve-evoked transmission with higher release probability than WT. We next analyzed individual EPSCs from WT and Syt7 −/− . As shown in Fig. 2a,b, at all Ca 2+ concentrations tested Syt7 −/− mutants show dramatically enhanced synaptic currents that are three-fold larger than WT at 0.5 mM Ca 2+ . Presynaptic changes that can drive larger synaptic responses can be secondary to increased release probability (P) of single SV fusion events or an increased number (N) of readily-releasable SVs.
To differentiate between these two possibilities, we measured the readily-releasable pool size using hypertonic stimulation 15 . As shown in Fig. 3a,b, hypertonic stimulation in Syt7 −/− embryos with 500 mM sucrose solution induced SV release levels similar to WT, indicating mutants do not have a larger releasable SV pool at embryonic synapses. Although the SV pool size is different in Syt7 3rd instar mutants 13 , SV pools increase dramatically from embryonic development through the 3rd instar larval stage as synaptic maturation and profound synaptic growth occur. Given the dramatic increase in presynaptic release in embryonic Syt7 terminals can occur without a change in the releasable SV pool size at this stage of development, release probability for individual SVs is also enhanced following loss of Syt7. We previously found that Syt1 −/− null mutants have a smaller releasable SV pool than WT 8,9 . We assayed for genetic interactions in SV pool size between Syt1 −/− and Syt7 −/− and found that loss of Syt7 did not change the smaller response in Syt1 −/− mutants (Fig. 3a,b), consistent with enhanced synaptic transmission occurring without an elevated SV pool size in Syt7 −/− . We next measured quantal size generated by release of single SVs during hypertonic stimulation, where release is Ca 2+ -independent and does not show multi-quantal release 16 . Quantal sizes of WT and Syt1 −/− are similar to those previously found 8 , and both Syt7 −/− and Syt1 −/− ; Syt7 −/− double mutants did not alter quantal size (Fig. 3c). These findings indicate there is no defect in the postsynaptic response to SV release in the absence of Syt7.
These data indicate the stronger synaptic transmission in Syt7 −/− mutants is likely due to increased P rather than changes in N. It was previously proposed that Syt1 triggers SV fusion upon action potential-induced Ca 2+ influx and also clamps fusion at lower [Ca 2+ ] to generate a better signal-to-noise ratio 17 . Experimental evidence supporting the clamping model were recorded at Drosophila 3rd instar NMJs where miniature frequency in Syt1 −/− is increased compared to WT 18,19 . In the case of embryonic NMJs, we did not detect an elevated frequency of miniature events in Syt1 −/− 8,9 . However, Syt7 −/− mutants displayed more frequent miniature release than WT, and Syt1 −/− ; Syt7 −/− double mutants showed even more frequent miniature release than Syt7 −/− single mutants (Fig. 4a,b). These results support a model where both Syt7 and Syt1 may clamp SV fusion in a synergistic manner as observed at mammalian synapses 20 . We hypothesize that elevated miniature release rate in Syt1 −/− single mutant embryos is cancelled out by the reduced SV pool size at these synapses, as well as its reported role in SV docking and endocytosis 8,9 . Syt7 suppresses release probability and enables paired-pulse facilitation. The higher release probability in Syt7 −/− mutants predicts decreased facilitation would occur, given synapses with stronger release probability have a lower paired pulse facilitation (PPF) ratio due to depletion of SVs during the 1st response 3,11 . Indeed, the PPF ratio in Syt7 −/− was much smaller than WT at 0.5 mM and 1.5 mM extracellular [Ca 2+ ] (Fig. 5a,b). However, at a lower [Ca 2+ ] of 0.3 mM, a positive PPF ratio was observed in Syt7 −/− in accordance with the smaller amplitude of the 1st EPSC (Fig. 5a,b). As shown in Fig. 5c, Syt7 −/− shows a relatively high PPF ratio (arrowhead in Fig. 5c) when synaptic currents are small, similar to that observed in WT. Together with the enhanced facili- www.nature.com/scientificreports/ tation observed following Syt7 overexpression (Fig. 1d), these data indicate Syt7 levels bi-directionally gate the sign of short-term plasticity (facilitation versus depression) by controlling the levels of presynaptic output.

Discussion
The current study indicates Syt7 is indispensable for facilitation across the physiological range of Ca 2+ concentrations at Drosophila embryonic NMJs as previously shown for mammalian preparations 2 . In the absence of Syt7, the normally facilitating embryonic NMJ now displays depression. Following Syt7 overexpression, facilitation is . WT, Syt7 +/− and Syt7 −/− mutants were analyzed with the Kruskal-Wallis test using a one-way ANOVA by ranks and significant difference between the groups was found (P < 0.0001). *P < 0.05 by Dunn's post-hoc multiple comparison test between groups. The red line indicates when no facilitation or depression is observed (ratio = 1). (d) Facilitation ratio shown as 3rd pulse-induced EPSC (EPSC 3 )/1st pulse-induced EPSC (EPSC 1 ) for elav-GAL4 controls and Syt7 OE (elav-GAL4 > UAS-Syt7). The results were analyzed with the Mann-Whitney U test and significant difference between the groups was found (***P < 0.001). The numbers of recorded cells analyzed for each genotype: elav-GAL4 , 13; elav-GAL4 > UAS-Syt7, 9. Error bars are SEM.   S y t7 -/- www.nature.com/scientificreports/ greatly enhanced. Our data indicate the major reason for defective facilitation in Syt7 −/− mutants is due to loss of Syt7's ability to suppress release, which likely causes rapid SV depletion that is non-compatible with short-term synaptic facilitation. Likewise, overexpression of Syt7 reduces SV release and allows for enhanced facilitation. This role for Syt7 contrasts with current models proposed in mammals where Syt7 is hypothesized to bind residual Ca 2+ to directly act as a facilitation Ca 2+ sensor. Although our data indicate Syt7 is not the primary Ca 2+ sensor for facilitation, we cannot rule out the possibility that Syt7 has dual roles in both suppressing and facilitating SV fusion as observed for Syt1 9,21 . If Syt7 has a dual role with C2A functioning for clamping and C2B for facilitation, the null mutant would lack both properties. It is possible the lack of clamping is the dominant phenotype, with any facilitative function being masked by SV depletion at higher Ca 2+ concentrations. Thus, we cannot rule out a Syt7-dependent component of facilitation. " However, the presence of facilitation in Syt7 −/− mutants at lower [Ca 2+ ] (Fig. 5) indicate there is a facilitation sensor besides Syt7 that monitors residual Ca 2+ to directly activate this form of short-term plasticity. Although we cannot completely rule out a role for residual maternally supplied Syt7 at the embryonic stage in Syt7 −/− mutants, there is no evidence from RNA profiling studies (Flybase) that indicate Syt7 is present at earlier stages of embryonic development prior to nervous system formation. Thus, it is unlikely residual Syt7 could sustain normal levels of facilitation as observed in low [Ca 2+ ], consistent with the enhanced synaptic transmission observed across a broad [Ca 2+ ] range in Syt7 −/− mutants. Given facilitation is also present in low [Ca 2+ ] in Syt7 −/− mutants at the 3rd instar stage when any maternal contribution would be depleted 13 , we conclude that facilitation can occur in the complete absence of Syt7 under conditions where the initial response is reduced.
A key advantage of the Drosophila embryonic NMJ preparation is the ability to unambiguously monitor the absolute baseline values of synaptic strength even in high [Ca 2+ ] using the non-contracting Mhc mutant. In this regard, it is clear that synaptic transmission at Drosophila embryonic NMJs is much stronger in Syt7 −/− mutants at all Ca 2+ concentrations tested. Moreover, the stronger transmission is due to higher release probability rather than an increased number of releasable SVs. Thus, our data predict that higher release probability leads to a lower facilitation ratio secondary to vesicle depletion 3,11 . The precise mechanisms by which Syt7 suppresses SV release to enable facilitation will require further study. Beyond a potential clamping function for Syt7, the protein could alter local Ca 2+ buffering or cause increased Ca 2+ influx that could contribute to elevated SV release. Syt7 does not localize to SVs and may instead act from the plasma membrane 22 or internal membrane compartments 13 , allowing for several potential mechanisms for Syt7 to suppress release. Ca 2+ binding to the C2A and C2B domains of Syt1 have been shown to have distinct functions in SV release, with C2B playing a dominant role in triggering SV fusion and C2A acting to clamp release 9,23 . It is unclear if the C2A and C2B domains of Syt7 act similarly in Drosophila or have independent functions compared to Syt1. One possibility is that the C2A domain of Syt7 suppresses SV fusion and the C2B domain facilitates release, similar to Syt1. Structure function studies of Syt7 should help elucidate this biology in Drosophila, similar to our prior studies of Syt1 function.
As suppression of SV release by Syt7 is dose-dependent 13 (Fig. 1b), increasing levels of Syt7 would elevate the ratio of facilitation as shown in Fig. 1a-d. These results suggest the degree of facilitation across distinct neuronal populations may be set by Syt7 levels similar to a potentiometer. Our analysis of Syt7 −/− nulls, heterozygotes and overexpression lines support such a model that changes release and short-term plasticity in a graded fashion. Depending on whether a synapse is facilitative or depressive 11 , Syt7 expression could be modulated to gate plasticity to the level that most benefits the local circuit, similar to how Syt1 and Syt2 levels variably control release synchronicity across neuronal populations 20 . Indeed, the squid giant synapse is facilitative only when Ca 2+ is lowered from normal saline (artificial sea water) 24,25 , similar to Syt7 −/− mutants, suggesting synapses that normally depress may have reduced levels of Syt7. Indeed, recent evidence suggests that species-specific differences in presynaptic plasticity in rodents is linked to the levels of Syt7 26 . In shrews, the levels of Syt7 are lower in www.nature.com/scientificreports/ hippocampal CA3 synapses and they show reduced presynaptic plasticity. In contrast, Syt7 levels are much higher in mice, with their CA3 output synapses displaying far greater forms of presynaptic plasticity. Drosophila adults 27 and 3rd instar larvae 28 also have less facilitative NMJs than embryonic NMJs. This difference may contribute to the distinct effects of Syt7 on clamping spontaneous SV release that is observed between embryonic and 3rd instar NMJs 13 . Mammalian studies identified redundant functions for Syt1 and Syt7 in clamping spontaneous fusion at inhibitory synapses 29 . While reductions in Syt7 levels alone did not increase spontaneous SV release, removal of both Syt1 and Syt7 enhanced mini frequency to a far greater level that loss of Syt1 alone. In addition, a Syt7 transgene was able to rescue the elevated miniature frequency in Syt1 mutants. These differences in clamping properties were attributed to an insufficient level of Syt7 expression compared to Syt1. Differences in the Syt1/Syt7 ratio between Drosophila 3rd instar and embryonic NMJs may also contribute to distinct effects on spontaneous SV clamping observed in Syt1 and Syt7 mutants at these distinct developmental stages. In conclusion, controlling expression level of Syt7 provides an attractive mechanism for activity-dependent presynaptic scaling of release probability as a homeostat for both presynaptic output and short-term facilitation, similar to postsynaptic scaling mechanisms previously described for chronic forms of synaptic plasticity 30 .

Methods
Drosophila strains. Drosophila melanogaster were cultured on standard medium at 25 °C. The Syt7 M2 null mutant was used in this study 13 . Syt7 M2 has an insertion of a Minos transposon into the second exon of the Syt7 that generates a premature stop codon before the C2A domain, resulting in loss of Syt7 protein by Western analysis 13 . We observed similar results for Syt7 M1 , an independent null mutant generated via CRISPR-Cas9 (data not shown). All electrophysiological recordings were carried out in the background of a null mutant of muscle-specific myosin heavy chain (Mhc 1 ) as previously described 7  Electrophysiological analysis. All electrophysiology experiments were carried out as described previously 8,9 . Briefly, synaptic currents were recorded with the patch-clamp technique in whole-cell configuration from embryonic muscle fiber 6 at segments A2-A4 that were maintained at a holding potential of − 60 mV. Embryos were aged 21-24 h after fertilization and recorded in HL3.1 saline solution 33  For nerve stimulation, a small part of an intact (uncut) motor nerve was sucked into a suction electrode containing bath solution at its site of emergence from the CNS, and 1.5 μA of positive current was delivered for 1 ms. For extensive electrical isolation, a negative glass electrode with a blunt end was loosely attached to the nerve at a more distal site than that of the suction electrode. Clampex 10.7 in the pCLAMP10.7 package (Molecular Devices) was used for data acquisition and data were analyzed with Clampfit 10.7 in the pCLAMP10.7 (Molecular Devices). For hypertonic-stimulated release, 500 mM sucrose dissolved in Ca 2+ -free HL-3.1 saline solution was included in a puff pipette with a 1-μm tip. The pipette was placed in close vicinity of the boundary between muscle fibers 6 and 7 where the NMJ forms. Hypertonic solution was puffed using positive pressure for 3 s. Slow responses originating from electrically coupled muscle fibers were excluded in subsequent analysis in all genotypes. These experiments were performed in Ca 2+ -free saline to avoid enhancements of presynaptic release mediated by retrograde signaling downstream of postsynaptic Ca 2+ influx 6,34 . Miniature analysis was performed with 1.5 mM Ca 2+ HL3.1 containing 3 μM tetrodotoxin (TTX). Through the "Gap free" recording of Clampex, miniature release was counted for 10 min for each cell and reported as a frequency per minute.
Statistical analysis was performed using Prism6 software (GraphPad). Since the variation of data between animals or cells was small compared with the larger variation in data from one cell or one animal, results were combined and the number of events recorded are shown as N. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.