Main

Several mechanisms for facilitation have been proposed (Extended Data Fig. 1). In the ‘buffer saturation’ model, high concentrations of presynaptic Ca2+ buffer capture incoming Ca2+ before it binds to the rapid synaptotagmin isoforms (1, 2 and 9) that trigger vesicle fusion at most synapses7. If the Ca2+ buffer saturates during the first action potential, more Ca2+ reaches release sites during subsequent action potentials, producing facilitation6,8. Yet many facilitating synapses lack sufficient presynaptic Ca2+ buffer to account for this form of facilitation9. Another theory suggests that a specialized Ca2+ sensor responds to the smaller, longer-lasting Ca2+ signals between action potentials4. In one scenario, this sensor modulates Ca2+ channels to produce use-dependent increases in Ca2+ influx10. Several candidate proteins have been proposed to act in this manner11,12, but increased Ca2+ influx cannot account for facilitation at most synapses13. Alternatively, an unidentified Ca2+ sensor could mediate facilitation by directly increasing the probability of release (p).

Syt7 is located presynaptically, and binds Ca2+ with high affinity and slow kinetics14,15,16, making it a promising candidate sensor for the modest increases in residual Ca2+ that mediate facilitation. Previous studies suggest that Syt7 contributes to a slow phase of transmission known as asynchronous release17,18, and to Ca2+-dependent recovery from depression19, but the role of Syt7 in facilitation was not examined because these studies used synapses with prominent depression that obscures facilitation. We therefore examined synaptic transmission at four facilitating synapses: Schaffer collateral synapses between hippocampal CA3 and CA1 pyramidal cells9 (Fig. 1a), thalamocortical synapses between layer 6 cortical pyramidal cells and thalamic relay cells20 (Fig. 1b), mossy fibre synapses between dentate granule and CA3 cells9 (Fig. 1c), and perforant path synapses between layer II and III cells of the entorhinal cortex and dentate granule cells21 (Fig. 1d). Immunohistochemistry shows that Syt7 is present in regions where these synapses are located (Extended Data Figs 2 and 3). Facilitation is often assessed using pairs of closely spaced stimuli. In slices from wild-type mice, paired-pulse facilitation resulted in ~2-fold enhancement of neurotransmitter release lasting several hundred milliseconds (Fig. 1a–d, black traces). In Syt7-knockout mice, paired-pulse facilitation was eliminated (Fig. 1a–d, red traces). Sustained high frequency activation produces up to tenfold enhancement in wild-type animals, but facilitation is eliminated in knockouts at all synapses except for mossy fibre synapses, where the remaining enhancement is consistent with use-dependent spike broadening that occurs at this synapse22 (Fig. 1e–h and Extended Data Fig. 4).

Figure 1: Facilitation is absent in Syt7 knockout mice.
figure 1

a–d, Representative traces (top) and average paired-pulse ratio (PPR) at different interstimulus intervals (Δt) (bottom) recorded in slices prepared from wild-type (WT; black) and Syt7-knockout (KO; red) animals. Postsynaptic responses were recorded using whole-cell voltage clamp from hippocampal CA1 pyramidal cells (a), and thalamic relay cells (b). fEPSPs were recorded from hippocampal-mossy-fibre to CA3 synapses (c), and lateral-performant-path synapses in the dentate gyrus (d). Vertical scale bars, 100 pA (a, b) and 100 μV (c, d). e–h, Synaptic responses to 20-Hz trains from the same preparations as a–d (top), normalized amplitudes during 20-Hz trains (middle), and normalized responses to the tenth stimulus as a function of stimulus frequency (bottom). Peak PPR was significantly different for wild-type and Syt7-knockout mice at all synapses, as was response10/response1 for 5–50-Hz trains (P < 0.01, Student’s t-test). Data represent mean ± s.e.m. Number of experiments is shown in Extended Data Table 1.

PowerPoint slide

Full size image

The loss of facilitation in Syt7 knockouts cannot be accounted for by slowed recovery from depression reported with Syt7 deletion19, because recovery from depression is too slow to influence rapid facilitation strongly, nor can it produce the large increase in release associated with facilitation. There are several possible explanations for the loss of facilitation in knockouts: (1) the presynaptic Ca2+ signal that induces facilitation could be altered, (2) the probability of release (p) for synaptic vesicles could be increased, which by promoting vesicle depletion would indirectly reduce facilitation, or (3) the mechanism for facilitation could be disrupted directly. We assessed these possibilities at the CA3–CA1 synapse.

Action-potential-evoked increases in presynaptic Ca2+ consist of a large, brief localized Ca2+ signal that activates the low-affinity Ca2+ sensor synaptotagmin 1 to trigger neurotransmitter release23, and a small residual Ca2+ signal (Cares) that persists for tens of milliseconds and has been implicated in facilitation2. It is difficult to measure local Ca2+ signals that trigger release, but Cares is readily measured. We used a low-affinity Ca2+ indicator to measure the time course of Cares in CA3 presynaptic terminals, because facilitation can be attenuated by the accelerated decay of Cares (ref. 4). Cares decayed similarly in wild-type and Syt7-knockout animals (Fig. 2a), indicating that the loss of facilitation in knockout mice is not a consequence of accelerated Cares decay. We also used Cares as a measure of Cainflux to determine whether there are use-dependent changes in Ca2+ entry. However, each of two closely spaced stimuli evoked the same incremental increase in Cares in both wild types and knockouts (Fig. 2b), indicating that use-dependent changes in total Cainflux cannot account for facilitation. This suggests that if changes in Cainflux contribute to facilitation at this synapse, they must be restricted to the small subset of presynaptic calcium channels that evoke neurotransmitter release. We repeated the experiment using a high-affinity Ca2+ indicator, in which the degree of saturation during paired stimuli can be used to measure the magnitude of Cares evoked by the first stimulus (see Methods). We conclude that Cainflux evoked by the first stimulus is the same in wild-type and knockout animals (Fig. 2c).

Figure 2: Facilitation is altered in Syt7-knockout animals despite similar presynaptic Ca2+ signals.
figure 2

a, Presynaptic Cares evoked by a single stimulus recorded from Schaffer collateral fibres loaded with a low-affinity Ca2+ indicator (left), and Cares half-decay times (right). NS, not significant. b, Cares signals recorded with low-affinity indicator evoked by one or two stimuli (left). The ratio of the increase in Cares evoked by the first (ΔF1) and second (ΔF2) stimuli (right). c, Cares signals recorded with high-affinity indicator evoked by one or two stimuli. d, Average EPSC amplitudes for CA3–CA1 synapses recorded in different external Ca2+ (Cae) concentrations, normalized to the amplitude in 2 mM Cae. e, EPSCs recorded in different Cae. Vertical scale bars, 50, 100, 200 and 300 pA in 0.5, 1, 2 and 3 mM Cae, respectively. f, PPR for interstimulus interval of 20 ms recorded in different Cae. In 0.5 mM Ca2+, the PPR in knockout (1.24 ± 0.12) was not significantly different from 1 (P = 0.084, Wilcoxon signed rank test). Data represent mean ± s.e.m. Number of experiments is shown in Extended Data Table 2.

PowerPoint slide

Full size image

We further explored the role of Ca2+ in facilitation by examining the Ca2+-dependence of excitatory postsynaptic currents (EPSCs) and facilitation. Raising extracellular Ca2+ leads to a steep increase in EPSC amplitude (Fig. 2d) but a decrease in facilitation (Fig. 2e, black traces), even though high extracellular Ca2+ should increase the Cares available to evoke facilitation. This paradox is resolved by realizing that increased Ca2+ influx increases p, which depletes presynaptic vesicles, saturates release and limits the extent of facilitation. The Ca2+-dependence of EPSC amplitudes was unaffected in knockout animals (Fig. 2d), but facilitation was absent for all values of external Ca2+ (Fig. 2e, f). Meanwhile, there was no difference in basal release properties measured by the rate of spontaneous EPSCs (Extended Data Fig. 5). These findings suggest that the loss of facilitation in knockouts is not a consequence of higher initial p, because facilitation was absent even when the initial p was strongly attenuated by reducing external Ca2+.

To test further whether initial p is increased in Syt7 knockouts, we measured how field excitatory postsynaptic potentials (fEPSPs) scaled with stimulus intensity24 (Fig. 3a). The slope of the fEPSP versus presynaptic volley gives a relative measure of p (see Methods), which was unchanged in knockouts (Fig. 3b). Moreover, the fEPSP to presynaptic volley ratio changed steeply with extracellular Ca2+, showing that this method is sensitive to p (Fig. 3c, d). We also assessed p using pharmacological blockade of synaptically activated NMDARs (N-methyl-d-aspartate receptors) by the use-dependent blocker MK801 (ref. 25; Fig. 3e–g). This approach is widely used to detect changes in p: an increase in p leads to more glutamate release, and more activation and rapid blockade of NMDARs, while a decrease in p leads to a slower blockade (Extended Data Fig. 6). The rate of blockade of NMDAR-mediated fEPSPs (NMDAR-fEPSPs) was unaffected by Syt7 deletion (Fig. 3e), indicating similar initial p. However, when we evoked NMDAR-fEPSPs with trains of three stimuli25, amplitudes decayed more rapidly in wild types (Fig. 3f, g), suggesting that Syt7 is required to increase p for the second and third stimuli. Thus, initial p and presynaptic Ca2+ signalling are unaffected by Syt7 deletion, but knockouts lack the use-dependent increase in p that underlies facilitation. This suggests that the mechanism underlying facilitation is directly impaired by Syt7 deletion.

Figure 3: Change in the initial probability of release does not underlie the absence of facilitation in Syt7-knockout mice.
figure 3

a, Extracellular recordings of presynaptic fibre volley and fEPSP evoked by the indicated stimulus intensities. Scale bar, 200 μV. b, fEPSP slope plotted against fibre volley amplitude, for 20–100 μA stimulation. c, fEPSPs recorded in 1 and 3 mM Cae. Scale bar, 100 μV. d, Average ratio of the fEPSP to the fibre volley in different Cae. e, Top, initial release probability was measured by stimulating Schaffer collaterals every 10 s while recording NMDAR-fEPSPs before and after MK801 bath application. Middle, traces averaged from 10 trials before (dark traces), and trials 10–15 after (light traces) MK801 application. Bottom, average NMDAR-fEPSP amplitudes evoked in the presence of MK801. f, Same as in e but with three stimuli at 50 Hz every 30 s. First response to trains is shown. g, Half-decay times of NMDAR-fEPSP amplitudes in the presence of MK801. *P < 0.05, one-way analysis of variance (ANOVA) with Tukey’s post-hoc test. Data represent mean ± s.e.m. Number of experiments shown in Extended Data Table 2.

PowerPoint slide

Full size image

Syt7 is implicated in neuroendocrine release16, insulin secretion26 and exocytosis of lysosomes27, which could all indirectly influence synaptic transmission in global Syt7 knockouts. Therefore, to determine whether Syt7 controls facilitation by acting in presynaptic neurons in a cell-autonomous manner, we tested whether viral expression of Syt7 in CA3 pyramidal cells of Syt7 knockouts rescued facilitation. This approach is complicated by our inability to virally transduce all CA3 pyramidal cells, which prohibits the use of extracellular stimulation that would activate some presynaptic cells that express Syt7 and others that do not. We overcame this problem with an adeno-associated virus (AAV) that drove bicistronic expression of both channelrhodopsin-2 (ChR2) and Syt7, allowing optical stimulation of only those fibres expressing Syt7.

Using conditions we have previously shown allow facilitation to be studied with optogenetic stimulation (see Methods), we confirmed that when ChR2 alone was expressed, optical and electrical stimulation produced similar facilitation in wild types (Fig. 4a, e, f), and similar depression in knockouts (Fig. 4b, e, f). We next used a bicistronic vector to express both ChR2 and wild-type Syt7 in knockout animals. Light-evoked responses exhibited facilitation, whereas electrically evoked responses did not (Fig. 4c, e, f). This suggests that bicistronic expression of ChR2 along with a presynaptic protein of interest offers a powerful new approach to characterize the effect of gene manipulation on presynaptic function within intact neural circuits. When Syt7 was expressed in wild-type animals, the peak facilitation was unaffected (Fig. 4e, f and Extended Data Fig. 7a). Thus, expressing Syt7 in CA3 pyramidal cells rescued facilitation in a cell-autonomous manner, with facilitation restored only at synapses expressing Syt7 and ChR2.

Figure 4: Viral expression of Syt7 restores facilitation at Schaffer collateral synapses.
figure 4

a–d, Top, fluorescence images of yellow fluorescent protein (YFP)-tagged ChR2 and Syt7 immunostaining in the CA1 region after AAV injection into CA3 to express the indicated proteins in wild-type animals (a) or Syt7-KO animals (b–d). PY, stratum pyramidale; SR, stratum radiatum. Scale bar, 100 μm. Bottom, EPSCs and PPRs for responses evoked electrically (open symbols) and optically (blue symbols). In a and b, only ChR2–YFP was expressed; in c, both ChR2–YFP and wild-type Syt7 were expressed (separated by a porcine teschovirus-1 2A (P2A) cleavage peptide); and in d, ChR2–YFP and Ca2+-insensitive Syt7(C2A*) were expressed. e, f, Summary of PPRs for 50-ms interstimulus interval. Asterisks denote significant difference from responses evoked electrically in uninjected wild-type animals (e), or optically in wild-type animals expressing ChR2 alone (f). *P < 0.05, one-way ANOVA with Tukey’s post-hoc test. Data represent mean ± s.e.m. Number of experiments is shown on bar graphs.

PowerPoint slide

Full size image

To determine whether Ca2+ binding by Syt7 is important for facilitation, we assessed whether facilitation is rescued by Syt7 with a mutated Ca2+-insensitive C2A domain (Syt7(C2A*)). Previous studies established that Ca2+ binding to the C2A domain of Syt7 is required for Syt7 to mediate asynchronous release18. We found that Syt7(C2A*) did not rescue facilitation in knockouts (Fig. 4d–f). Moreover, in wild-type animals, Syt7(C2A*) expression strongly attenuated facilitation (Fig. 4e, f and Extended Data Fig. 7b), suggesting that Syt7(C2A*) competes with native Syt7 to suppress facilitation.

Our results indicate that facilitation requires Ca2+ binding to the C2A domain of Syt7, and also provide insight into the role of Syt7 in facilitation. We conclude that Syt7 does not produce facilitation by altering the amplitude and time course of Cares (Fig. 2), by increasing initial p (Fig. 3), by acting as a Ca2+ buffer (Extended Data Fig. 8), or through use-dependent increases in the total Cainflux (Extended Data Fig. 1b and Fig. 2). The observation that initial p is unaltered in Syt7 knockouts indicates that local Cainflux is unaffected for the first stimulus, but it is difficult to rule out the possibility that Syt7 mediates a use-dependent increase in Cainflux through the subset of channels that trigger vesicle fusion. There is, however, no evidence for Syt7 associating with or regulating calcium channels. By contrast, Syt7 is known to interact with Syt1 and can mediate vesicle fusion16,17,18. The most parsimonious explanation is that Syt7 acts as the proposed specialized Ca2+ sensor to increase p during facilitation. Facilitated release exhibits rapid kinetics, suggesting that Syt7 somehow increases the probability of Syt1-dependent vesicle fusion. Whether this is through a direct interaction of Syt7 with a fast synaptotagmin isoform such as Syt1 remains an open question. It is also unclear whether the recently described interaction between Syt7 and calmodulin that promotes vesicle replenishment19 is similarly required for facilitation. Finally, it is possible that at other synapses facilitation is mediated by additional specialized Ca2+ sensors, or involves other mechanisms. Further studies are needed to clarify these issues.

Based primarily on theoretical considerations, facilitation is thought to influence both information transfer and network dynamics profoundly. In the hippocampus, the high-pass filtering imposed by facilitating synapses may account for the burst firing in place cells that encode spatial information28. In the auditory pathway, facilitation is proposed to counteract short-term depression to maintain linear transmission of rate-coded sound intensity29. It has even been suggested that facilitation forms the basis of short-term memory, as facilitating recurrent connections within cortical networks could support the persistent activity states associated with working memory30. In future studies, the selective elimination of Syt7 from specific cell types could allow the first direct tests of the effect of facilitation on neural circuits and behaviour.

Methods

Animals and viruses

All mice were handled in accordance with NIH guidelines and protocols approved by Harvard Medical School. Syt7 knockout mice31 (Jackson Laboratory) and wild-type littermates of either sex were used. Statistical tests were not used to predetermine sample size. Blinding and randomization were not performed. AAV2/9-hSyn-hChR2(H134R)-EYFP and its pAAV backbone (Addgene 26973) were obtained from the University of Pennsylvania Vector Core. Complementary DNA encoding the rat Syt7 wild-type α isoform and C2A* mutant (D225A, D227A and D233A)18 were provided by T. Bacaj and T. Sudhof. For rescue experiments involving Syt7 with mutated Ca2+ binding domains, we used the mutated C2A* version instead of the C2A*C2B* double mutant, as mutation of both C2 domains leads to lower levels of expression. The P2A cleavage sequence32 and Syt7 were inserted after the ChR2 carboxy terminus in the pAAV backbone (Genscript). Plasmid-driven expression of ChR2–YFP and Syt7 was confirmed in HEK cells by Syt7 immunostaining and patch-clamp recording of ChR2 photocurrents. AAVs were produced and purified from HEK cells as previously described33.

Stereotaxic surgeries were performed as described34. Postnatal day (P) 18–30 mice were anaesthetized with ketamine/xylazine/acepromazine (100/10/3 mg kg−1) supplemented with 1–4% isoflurane. Viruses were injected through glass capillary needles using a syringe (Hamilton) mounted on a stereotaxic instrument (Kopf). Injection coordinates from lambda were 2.69 mm (rostral), 3 mm (lateral) and 2.8 mm (ventral). One microlitre of virus suspension was delivered at a rate of 0.1 μl min−1 using a microsyringe pump (WPI; UMP3) and microsyringe pump controller (WPI; Micro4). The needle was slowly retracted 5–10 min after injection, and the scalp incision was closed with gluture. Post-injection analgesic (buprenophrine, 0.05 mg kg−1) was administered subcutaneously for 48 h.

Acute slice preparation

P30–P60 animals were euthanized under isoflurane anaesthesia, 14–30 days after AAV injection. Brains were removed and placed in ice-cold solution containing (in mM): 234 sucrose, 25 NaHCO3, 11 glucose, 7 MgCl2, 2.5 KCl, 1.25 NaH2PO4 and 0.5 CaCl2. Then, 270-μm-thick transverse slices (hippocampal recordings) or 250-μm-thick sagittal slices (thalamic recordings) were prepared on a vibrotome (Leica, VT1000s), and a cut was made between CA3 and CA1 to prevent recurrent excitation. Slices were transferred for 30 min to 32 °C artificial cerebrospinal solution (ACSF) containing (in mM): 125 NaCl, 26 NaHCO3, 25 glucose, 2.5 KCl, 2 CaCl2, 1.25 NaH2PO4 and 1 MgCl2, adjusted to 315 mOsm, and allowed to equilibrate to room temperature for >30 min. Experiments were performed at 33 ± 1 °C with flow rates of 2 ml min−1.

Electrophysiology

For ChR2 stimulation, 160 mW mm−2 laser pulses (0.2–0.5 ms) from a 100-mW 473 nm laser (OptoEngine, MBL-III) were focused through the ×60 objective of the microscope (Olympus, BX51WI) to produce a 80-μm diameter spot over the stratum radiatum, >500 μm from the recorded cell to avoid activating ChR2 in presynaptic boutons, which can artificially raise the probability of release and obscure facilitation34. Extracellular stimulation was performed with a stimulus isolation unit (WPI, A360) using glass monopolar electrodes (0.5–1 MΩ) filled with ACSF. Stimulus electrodes were positioned ~500 μm from the recording electrode in the stratum radiatum (Schaffer collaterals), the internal capsule (corticothalamic), the hillus adjacent to the dentate granule cell layer (mossy fibres), and the outer molecular layer (lateral performant path). To ensure that mossy fibre responses were not contaminated by associational/commissural inputs the metabotropic glutamate receptor agonist DCG-IV (1 μM) was applied at the end of experiments to block mossy fibre responses selectively35. Data were included only if responses were reduced by more than 80% (average reduction was 88 ± 1% in wild-type and 90 ± 2% in Syt7-knockout mice), and the amplitude of mossy fibre responses was measured after subtracting the response remaining in the presence of DCG-IV. Stimulus trials were repeated at 0.1 Hz (0.033 Hz at mossy fibres to avoid potentiation), and artefacts were deleted for display. Recordings were acquired using an amplifier (Axon Instruments, Multiclamp 700B) controlled by custom software written in IgorPro (provided by Matthew Xu-Friedman, SUNY Buffalo), and low-pass filtered at 2 kHz. Whole-cell recordings were obtained using borosilicate patch pipettes (2–5 MΩ) pulled with a horizontal puller (Sutter P-97). The internal recording solution contained (in mM): 150 Cs-gluconate, 3 KCl, 10 HEPES, 0.5 EGTA, 3 MgATP, 0.5 NaGTP, 5 phosphocreatine-Tris and 5 phosphocreatine-Na; pH 7.2. Cells were held at −70 mV, and series resistance was monitored during recordings. fEPSPs were recorded in current-clamp mode with ACSF-filled patch pipettes (0.5–1 MΩ). Inhibition was blocked with picrotoxin (50 μM), and during fEPSP recordings, CPP (2 μM) and CGP (3 μM) was added to the bath. Approximately 4–10 trials were conducted for each stimulus frequency, and recordings were averaged over trials. Data in all figures represent the mean ± s.e.m. Average responses are displayed with double exponential or polynomial curves fit in IgorPro. Unless stated otherwise, statistical significance was assessed by unpaired two-tailed Student’s t-test, or one-way ANOVA followed by Tukey’s post-hoc test.

Probability of release

To record NMDAR-EPSCs, cells were voltage clamped at +40 mV, and the internal solution contained (in mM): 85 Cs-methanesulfonate, 4 NaCl, 10 HEPES, 0.2 EGTA, 30 BAPTA, 2 MgATP, 0.4 NaGTP, 10 phosphocreatine-Na, 25 TEA, 5 QX-314; pH 7.3. For recording NMDAR-fEPSPs, Mg2+ was excluded from ACSF to relieve Mg2+ block of NMDA receptors. Picrotoxin (100 μM) and NBQX (5 μM) were added to the bath, and stimulation was conducted at 0.1 Hz (unless otherwise indicated) for 5 min to obtain a baseline response. Stimulation was halted for 10 min while (+)-MK801 (40 μM) was added and allowed to equilibrate. For experiments involving fEPSPs versus presynaptic volley, the postsynaptic response was measured by the slope of the fEPSP, while the amplitude of the presynaptic volley was used to determine the number of activated fibres. If p increases, the same number of activated presynaptic fibres will produce a larger fEPSP. The ratio between fEPSP and volley was determined by line fits to the linear regime of the input–output curve of individual experiments (20–80 μA stimuli).

The study of probability of release is complicated because many people use p to refer to the probability of release of a vesicle (pv) and others refer to probability of release from an active zone (psynapse) that contains N vesicles in its readily releasable pool. Thus, an increase in the size of the readily releasable pool for an active zone can increase psynapse even if pv is unaltered. Although MK801 blockade25 and fEPSPs versus presynaptic volley24 are both widely used methods to detect changes in the probability of release, for both approaches it is conceivable (although unlikely) that increases in pv could be obscured by a perfectly balanced decrease in the readily releasable pool size. However, the relationship between EPSC amplitude and extracellular Ca2+ is similar in wild-type and Syt7-knockout animals. This suggests there is no increase in pv, which would cause this curve to saturate at lower values of Cae for Syt7-knockout animals. Moreover, the large differences in facilitation in wild-type and Syt7-knockout animals were even more pronounced when the probability of release was reduced tenfold by lowering Cae from 2 mM to 0.5 mM, which is incompatible with an increase in pv obscuring facilitation by depleting vesicles.

Ca2+ measurements

Ca2+ was measured as described previously4. In brief, CA3 fibres were labelled for 3 min using an ACSF-filled pipette containing either magnesium green AM or fura-2 AM (240 μM) and 1% fast green, placed into the border of the CA3–CA1 field. A vacuum pipette placed above the loading site removed excess indicator. Slices were incubated for at least 1 h and imaging was performed in stratum radiatum of CA1 at least 500 μm from the injection site using a 60× objective and custom-built photodiode. Excitation was achieved using a tungsten (magnesium green) or xenon (fura-2) lamp. Schaffer collaterals were stimulated using a glass electrode placed at least 300 μm from the imaging site. To prevent recurrent excitation, experiments were performed in the presence of NBQX (10 μM), CPP (2 μM) and picrotoxin (50 μM).

Magnesium green is a low-affinity calcium indicator36 (KD = 7 μM) that provides an approximately linear measure of Cares (ref. 37). As such it is well suited to measuring the time course of presynaptic Cares (Fig. 2a) and detecting changes in Cainflux during successive stimulations (Fig. 2b). However, with the bulk loading approach the size of the fluorescence change is proportional to the number of stimulated fibres, so the absolute Cares signal is not readily quantified with magnesium green. By contrast, fura-2 has a high affinity for calcium38,39 (KD = 131 nM) so it provides a saturating sublinear response to increases in Cares (refs 40, 41, 42). This can be used to test for changes in the absolute size of Cainflux because a change in the Cainflux per stimulus would change the ratio between the fluorescence change produced by the first and second stimuli.

Immunohistochemistry

Two to four weeks after AAV injection, mice were anaesthetized with ketamine and transcardially perfused with 4% paraformaldehyde (PFA) in PBS. The brain was removed and post-fixed for 24 h. Slices (50 μm thick) were permeabilized (PBS plus 0.4% Triton X-100) for 30 min and then prepared in blocking solution (PBS plus 0.2% Triton X-100 and 2% normal goat serum; PBST) for 30 min at room temperature. Slices were incubated overnight at 4 °C in PBST with primary antibodies (anti-Syt7 (Synaptic Systems, 105173), 1 μg ml−1; 1:200, targeting amino acids 46–133 of Syt7α, anti-vGlut1 (Synaptic Systems, 135304), 1 μg ml−1; 1:500, and anti-calbindin-D28k (Sigma Aldrich, C9848), 1 μg ml−1; 1:500), followed by incubation with secondary antibodies in PBST for 2 h at room temperature. For both wild-type and Syt7-knockout mice, images from each brain region were acquired on a laser scanning confocal (Olympus, FluoView1200) using the same laser/microscope settings and processed in ImageJ identically.