Uncoupling the roles of synaptotagmin I as a dual Ca²⁺ sensor during endo- and exocytosis of synaptic vesicles

Uncoupling syt1 in exo- and endocytosis via mutations

During exocytosis, the C2-domains of syt1 interact with Ca²⁺, SNAREs, and lipids in the plasma membrane². Recent studies indicate that Ca²⁺•syt1 must penetrate, and bend, the target membrane to facilitate fusion^19,28,29. Because Ca²⁺ and membrane-bending are also involved in endocytosis^13,14,30,31, we sought to determine whether interactions between syt1 and Ca²⁺/membranes are important for syt1-mediated acceleration of SV endocytosis.

We made use of previously reported mutations that alter the membrane penetration (and thereby membrane bending) and/or Ca²⁺ binding activity of syt1^19,28. In the first Ca²⁺ ligand mutant form of syt1 that we examined, designated as syt1^4DN, two Ca²⁺ ligands in C2A and two ligands in C2B were neutralized by substituting the native aspartatic acid residues with asparagines, such that syt1 cannot bind to Ca²⁺ and is therefore unable to penetrate membranes. In the alanine mutant (syt1^4A), four non-polar residues within the membrane-penetration loops of C2A and C2B were replaced with alanines, such that the volume of the membrane penetration loops was reduced; these mutations virtually abolish the membrane penetration and bending activity of Ca²⁺•syt1¹⁹. Although the Ca²⁺ binding activity of syt1 was attenuated by these mutations, vesicle co-flotation assays indicate that syt1^4A can still bind to SNARE proteins to some extent in a Ca²⁺-promoted manner¹⁹, whereas Ca²⁺-promoted SNARE binding activity was completely eliminated in syt1^{4DN 32}. When expressed in syt1 KO neurons, both of these syt1 mutants were present in presynaptic boutons (Supplementary Fig. 4). Nevertheless, syt1^4A and syt1^4DN both failed to rescue fast vesicle exocytosis (Fig. 3a, b; 793 ± 60 pA for syt1^FL; 107 ± 27 pA for syt1^4A, 83 ± 14 pA for syt1^4DN, p<0.001).

Open in a separate window

Figure 3

Ca²⁺ binding activity is required for syt1 to accelerate SV endocytosis. (a) Representative traces of EPSCs triggered by single action potentials. (b) Bar graph summarizing the peak amplitude of EPSCs recorded from syt1 KO neurons with lentiviral expression of syt1^FL (n = 29), syt1^4A (n = 15), syt1^4W (n = 24) or syt1^4DN (n = 17). (c) Representative traces showing EPSCs evoked by a 2 s 10 Hz stimulus train in syt1 KO neurons expressing syt1^FL or syt1^4W. (d) Normalized peak amplitudes of EPSCs during 2 s 10 Hz train stimulation (syt1^FL, n = 14; syt1^4W, n = 13). (e) Sample images showing sypHy fluorescence in presynaptic boutons before, 1 s after and 40 s after 100 action potentials at 20 Hz. Images were acquired from syt1 KO neurons and KO neurons expressing syt1^4A, syt1^4W, or syt1^4DN. Scale bar, 2 μm. (f) Normalized fluorescence changes of sypHy in syt1 KO neurons (n = 95), and KO neurons expressing syt1^FL (n = 102), syt1^4A (n = 100), syt1^4W (n = 110), or syt1^4DN (n = 84). ‘n’ is total number of boutons pooled from 4–5 coverslips. (g) Bar graph summarizing the endocytic time constants of syt1^FL, syt1^4A, syt1^4W, or syt1^4DN. ** p<0.001. Error bars represent SEM.

In contrast to the alanine mutant, we also increased the volume of the membrane penetration loops by substituting the same non-polar residues with tryptophans (syt1^4W); these mutations enhance the Ca²⁺ dependent phospholipid binding - and membrane bending - activity of syt1^19,28,33. When expressed in syt1 KO neurons, syt1^4W was targeted to presynaptic boutons (Supplementary Fig. 4) where it mediated larger EPSCs as compared to syt1^FL (1460 ± 130 pA, p<0.001). This increase was caused by higher SV release probability, since syt1^4W resulted in a faster depletion of the SV pool during a 2 s 10 Hz stimulus train (Fig. 3c, d) ³³.

Next, we evaluated the performance of all three syt1 point mutants during SV endocytosis using sypHy (Fig. 3e). The slow endocytosis observed in syt1 KO neurons was completely rescued by syt1^4A or syt1^4W (Fig. 3f, g; τ: 21.9 ± 0.6 s for syt1^FL, 23.2 ± 0.5 s for syt1^4A, 23.5 ± 0.6 s for syt1^4W, p>0.05, compared to syt1^FL). The similar time constants for the alanine and tryptophan mutants indicate that membrane penetration does not play a role in syt1 promoted endocytosis. Furthermore, enhancement of the Ca²⁺/membrane binding affinity in syt1^4W did not further accelerate endocytosis, consistent with our observations that increasing the extracellular [Ca²⁺] from 1 to 3.5 mM does not accelerate the kinetics of endocytosis (Supplementary Fig. 5); physiological extracellular [Ca²⁺] appears to be saturating under these conditions. In marked contrast to the syt1^4A and syt1^4W mutants, syt1^4DN did not rescue slow endocytosis in syt1 KO neurons (Fig 3f, g; τ: 33.1 ± 1.1 s; p>0.05, compared to syt1 KO). This observation was confirmed via FM4-64 uptake experiments (Supplementary Fig. 6).

We reiterate that membrane penetration loop mutations in syt1 either reduced (syt1^4A) or enhanced (syt1^4W) the function of syt1 during exocytosis without any secondary effects on the function of syt1 during endocytosis. Analogous to the retargeting experiments described above, these mutations clearly uncouple the role of syt1 in endo- and exocytosis. More importantly, both syt1^4A and syt1^4DN failed to rescue rapid exocytosis, but syt1^4A did rescue normal rates of endocytosis in syt1 KO neurons. This difference between syt1^4A and syt1^4DN during vesicle retrieval can be attributed to the lack of Ca²⁺ binding activity in syt1^4DN, but not to reductions in exocytosis. Together, these results strongly suggest that syt1 functions as a Ca²⁺ sensor during SV endocytosis.

Ca²⁺ independent SV endocytosis does not rely upon syt1

The experiments reported thus far indicate that syt1 functions to accelerate Ca²⁺-promoted endocytosis, but does syt1 also regulate Ca²⁺-independent vesicle retrieval? To address this question, 20 Hz field stimulation was replaced with a 5 s superfusion of a hypertonic sucrose solution (0.5 M), to induce Ca²⁺ independent exo-and endocytosis (Fig. 4a, Supplementary Fig. 8). Interestingly, endocytosis following sucrose treatment occurred with similar kinetics in WT and syt1 KO neurons (Fig. 4b, c) (τ of WT neurons: 17.0 ± 0.5 s for train stimulation, 15.5 ± 0.4 s for sucrose, p>0.05) (τ of syt1 KO neurons: 27.8 ± 0.9 s for train stimulation, 16.0 ± 0.4 s for sucrose, p<0.001), indicating that syt1 specifically functions during Ca²⁺-regulated SV retrieval. We confirmed that the sucrose-induced changes in sypHy fluorescence intensity reflect changes in SV retrieval and are not due to mechanical effects (Supplementary Fig. 7).

Open in a separate window

Figure 4

Ca²⁺ independent SV endocytosis does not rely upon syt1. (a) Experimental protocol and typical images showing exo- and endocytosis in syt1 KO neurons. Neurons were subjected to 5 s field-stimulation (20 Hz) or, for induction of Ca²⁺ independent exocytosis and subsequent endocytosis, were superfused with a 0.5 M hypertonic sucrose solution for 5 s. Scale bar, 2 μm. (b) Normalized fluorescence changes of sypHy in WT neurons and syt1 KO neurons in response to train stimulation or sucrose superfusion. WT/sucrose, n = 79; WT/train, n = 137; KO/sucrose, n = 77; KO/train, n = 103. ‘n’ is total number of boutons pooled from 4–5 coverslips. (c) Bar graph summarizing the τ values for endocytosis in WT and syt1 KO neurons in response to train stimulation or sucrose superfusion. * p<0.05, ** p<0.001. Error bars represent SEM.

Both C2 domains can act as Ca²⁺ sensors for SV endocytosis

The C2B domain of syt1 was originally proposed to be critical for endocytosis, because only this domain was found to bind AP-2^3,34. However, recent studies suggested that syt1 interacts with stonin-2 through either C2 domain^7,35. Here, we sought to determine which C2-domain of syt1 must sense Ca²⁺ in order to accelerate endocytosis, by mutating two Ca²⁺-ligands in either C2A (C2A^2DN-C2B) or C2B (C2A-C2B^2DN) (Fig. 5a), expressing these constructs in syt1 KO neurons, and conducting pHluorin imaging experiments. To avoid possible contributions of syt1 “null SVs” - which would, in effect, decelerate endocytosis - we tagged pHluorin directly to the luminal tail of the syt1 constructs used for these experiments. We note that the rate of endocytosis was identical in syt1 KO neurons expressing pH-syt1, as compared to WT neurons expressing sypHy (Supplementary Fig. 9).

Open in a separate window

Figure 5

C2A and C2B are redundant Ca²⁺ sensors for endocytosis. (a) Diagram showing the Ca²⁺ ligand mutant forms of pHluorin tagged syt1 employed for fluorescence imaging experiments. (b) Normalized fluorescence changes in syt1 KO neurons expressing sypHy (n = 52), pH-syt1^FL (n = 64) or Ca²⁺ ligand mutants of pH-syt1^FL (pH-syt1^4DN, n = 48; C2A^2DN-C2B, n = 72; C2A-C2B^2DN, n = 61; C2B^2DN-C2A, n = 50; C2B-C2A^2DN, n = 60). ‘n’ is total number of boutons pooled from 4 coverslips. (c) Bar graph summarizing the τ values for endocytosis in (B). ** p<0.001. Error bars represent SEM.

Similar to syt1^4DN, which was characterized above, pH-syt1^4DN was also unable to rescue rapid endocytosis (Fig. 5b, c; τ: 39.3 ± 1.1 s for pH-syt1^4A, 17.5 ± 0.6 s for pH-syt1^FL, p<0.001). Surprisingly, both C2A^2DN-C2B (16.1 ± 1.0 s) and C2A-C2B^2DN (17.5 ± 0.5 s) exhibited normal rates of endocytosis (p>0.05), and the C2-domains appear to be completely redundant at this level of analysis. To further verify these findings, we investigated the performance of the Ca²⁺-ligand mutants based on a ‘scrambled’ form of pH-syt1^FL, in which the order of C2A and C2B was exchanged in the primary sequence of the protein (C2B-C2A; Fig. 5a). Both C2B^2DN-C2A (17.5 ± 0.6 s), as well as C2B-C2A^2DN (19.5 ± 0.6 s), restored normal rates of endocytosis in syt1 KO neurons (Fig. 5b, c; p>0.05, compared to pH-syt1^FL). We reiterate that Ca²⁺-ligand mutations in the C2B domain disrupt the function of syt1 during exocytosis while analogous mutations in C2A do not^1,12,36,37. Here, in sharp contrast to the function of syt1 in exocytosis, our data suggest that endocytosis can employ either C2-domain of syt1 to sense Ca²⁺.

Uncoupling the function of syt1 in SV exo- and endocytosis

In previous studies, defects in endocytosis induced by mutations or complete loss of syt1 were always associated with defects in exocytosis^8–10. In a seminal study using acute photoinactivation of syt1^9,39, exocytosis was allowed to proceed in shibire^ts mutants at the restrictive temperature, then photoinactivation was carried out, and endocytosis was subsequently monitored at the permissive temperature and found to be absent. Recovery of endocytosis was not observed over the time frame of this experiment, and since syt1 is not required for endocytosis per se, these findings raise the possibility that the photoinactivated syt1 adduct actively inhibits endocytosis that would normally occur.

Uncoupling the roles of syt1 in endo- and exocytosis would greatly facilitate our understanding of the contribution of this molecule to the vesicle retrieval pathway. Here, we first addressed this problem by altering the local distribution of syt1 in neurons. Retargeting the cytoplasmic domain of syt1 to the plasma membrane uncoupled its roles in of exo- and endoctosis. In a second approach, we utilized point mutations to selectively alter exocytosis. For example, abolishing the membrane penetration activity of syt1 (syt1^4A) largely abolished exocytosis while leaving the ability of syt1 to facilitate endocytosis intact. This experiment revealed that changes in exocytosis do not always result in changes in endocytosis, at least in mouse hippocampal neurons; this result significantly impacts interpretation of the Ca²⁺-ligand mutants, as detailed further below. Third, analysis of deletion mutants demonstrated that the C2A domain alone cannot trigger exocytosis but can sense Ca²⁺ to facilitate rapid endocytosis.

Together, all three sets of experiments demonstrate that syt1 has a direct function in endocytosis, independent of its function in exocytosis.

Syt1 as a dual Ca²⁺ sensor for SV endo- and exocytosis

While both syt1^4A and syt1^4W were able to mediate rapid efficient endocytosis, syt1^4DN failed to rescue normal endocytosis in syt1 KO neurons. Since syt1^4A retains Ca²⁺-sensing activity while syt1^4DN is unable to bind Ca²⁺, these results strongly suggest that syt1 must bind Ca²⁺ in order to accelerate endocytosis. We further demonstrate that in hippocampal neurons, both the C2A and C2B domains of syt1 are capable of rescuing defects in endocytosis through binding to Ca²⁺. Our results differ from a previous study carried out using Drosophila, in which a truncation mutant lacking C2B - but containing an intact C2A domain - could not rescue defects in endocytosis measured using a synaptobrevin-based pHluorin¹¹. This apparent discrepancy might be due to inefficient targeting of truncated syt1 (Fig. 6b, c); a large number of synapses, devoid of mutant syt1, are likely to express the synaptobrevin-based optical reporter. To mitigate this concern, we fused a pHluorin directly to syt1. By fusing the pH sensor directly to WT or truncated syt1 proteins, changes in fluorescence exclusively represent the recycling of SVs that harbor the mutant protein. Using this approach, we report for the first time that either C2A or C2B is able to accelerate endocytosis in neurons. However, in the Drosophila neuromuscular junction (NMJ), Ca²⁺-ligand mutations in C2B alone disrupted endocytosis¹¹ and thus it appears that fly and mouse syt1 operate in distinct ways during vesicle retrieval.

We note that the study by Poskanzer et al. (2006) has been used as evidence for the idea that the decelerated endocytosis observed in syt1 deficient neurons was secondary to reduced rates of exocytosis¹⁴. However, this idea does not appear to apply to the system used in our studies, since we have shown that mutations in syt1 that reduce exocytosis do not always give rise to changes in the rate of endocytosis. If this is also true at the Drosophila NMJ, the C2B Ca²⁺-ligand studies reported by Poskanzer et al. (2006) also support the notion that syt1 acts as a Ca²⁺ sensor for SV endocytosis.

Our results regarding syt1 deletion mutants, that lacked either C2A or C2B, indicate that either C2-domain can accelerate endocytosis. The notion that the C2A domain participates in vesicle retrieval is novel and might be viewed as surprising since this domain cannot directly bind to AP-2μ^3,35. However, as alluded to above, stonin-2, a sorting protein adaptor specific for syt1, is now believed to bind to C2A even more strongly than to C2B³⁵. Thus the internalization of C2A is likely to involve the recruitment of stonin-2, as suggested by recent studies using non-neuronal cells³⁵.

Cell culture and lentiviral infection

Cultured hippocampal neurons and chromaffin cells were prepared from postnatal day 0–1 mice as described previously²². Lentiviral particles were generated as described previously¹⁹. Neurons were infected with virus at 5 days in vitro (DIV) and analyzed at 16–21 DIV. Chromaffin cells were infected at 0 DIV and analyzed at 4 DIV.

Electrophysiology

Whole-cell recordings were performed in voltage-clamp mode using a MultiClamp 700B amplifier (Molecular Devices). The recording chamber was continuously perfused with a bath solution (128 mM NaCl, 30 mM glucose, 5 mM KCl, 5 mM CaCl₂, 1 mM MgCl₂, 25 mM HEPES; pH 7.3) at room temperature. 50 μM D-AP5 (NMDA receptor antagonist; Tocris) and 20 μM bicuculline (GABA_A receptor antagonist; Tocris) were applied to isolate AMPA receptor-mediated EPSCs. Patch pipettes were pulled from borosilicate glass and had resistances of 3–5 MΩ when filled with internal pipette solution (130 mM K-gluconate,1 mM EGTA, 5 mM Na-phosphocreatine, 2 mM Mg-ATP, 0.3 mM Na-GTP, 10 mM HEPES; pH 7.3). The series resistance was typically <15 MΩ and was partially compensated to 60–80%. The membrane potential was held at −70 mV. To trigger an action potential, presynaptic neurons were depolarized with a theta stimulating electrode, with a voltage step from 0 V to 20–30 V for 1 millisecond; EPSCs were recorded from connected post-synaptic neurons. For mEPSC recording in Supplementary Fig. 2, the bath solution was supplemented with 0.5 μM tetrodotoxin (TTX) to block Na⁺ channels. Data were acquired using pClamp software (Molecular Devices), sampled at 10 kHz, and filtered at 2 kHz. Off-line data analysis of EPSCs was performed using Clampfit and Igor (WaveMetrics) software.

Fluorescence imaging

Neurons were continuously perfused with bath solution (in mM: 140 NaCl, 5 KCl, 2 CaCl₂, 2 MgCl₂, 10 HEPES, 10 glucose; in μM: 10 CNQX, 50 AP-5; pH 7.4). When Ca²⁺ free bath solution was used, 2 mM CaCl₂ was replaced with 2 mM MgCl₂. Neurons were depolarized using 1 ms, 70 V, bipolar pulses via two parallel platinum wires embedded in the imaging chamber, or were superfused with hypertonic sucrose solution. For the AM-BAPTA experiments in Supplementary Fig. 8, neurons were perfused with Ca²⁺ free bath solution containing 30 μM AM-BAPTA, and were pre-incubated with AM-BAPTA for 15 min prior to imaging experiments. A Xenon arc lamp was used as the light source for excitation, and images were acquired at 1 s intervals with a typical exposure time of 400 ms for pHluorin or 100 ms for FM4-64 on a Nikon TE300 inverted microscope. The following band pass optical filters (± 15~20 nm) were used: pHluorin (Excitation 484 nm, Emission 517 nm), FM4-64 (Ex 484 nm, Em 645 nm). A neutral density filter (ND 0.3, Chroma) was used for FM4-64 to minimize photobleaching. Fluorescence changes at individual boutons were monitored using an EMCCD camera (Cascade 512IIB, PhotoMetrics) with 2 × 2 binning, collected and analyzed off-line using MetaMorph software (Universal Imaging). In all experiments, images were acquired for 10 s prior to stimulation to establish a stable baseline. The regions that showed ΔF larger than 5 A.U. during stimulation were initially selected and marked with a 7-pixel diameter circle (corresponding to ~1.5 μm). This cut-off for selection (5 A.U.), which corresponds to 1.5 times the full-width half maximum of the Gaussian distribution of fluorescence noise, was determined as described previously ⁵⁰. Those regions that showed abrupt ΔF larger than 5 A.U. over any 3 consecutive time-points during the baseline acquisition were discarded. Further selection was based on morphology (oval shape) and immobility during acquisition. Bleaching was less than 1% for 120 s ~ 180 s time-lapse experiments and was not corrected.

Antibodies and Immunocytochemistry

Mouse monoclonal antibodies specific for syt1 (41.1), NMDAR (54.1) or synaptophysin (7.2) used for immunoblot analysis were kindly provided by R. Jahn (Gottingen, Germany). The same antibody for syt1 (1:400), a chick polyclonal antibody specific for GFP (1:1000, Abcam), and a rabbit polyclonal antibody specific for synapsin I (1:200, Chemicon) were used to localize GAP43-C2AB and syt1 mutants. Cy3-conjugated anti-mouse (Jackson ImmunoResearch Laboratories) and Alexa647-conjugated anti-rabbit (Invitrogen) secondary antibodies (1:500) were used to visualize immunofluorescence. Immunocytochemistry was carried out and analyzed as described previously²². All images were collected using a Confocal microscope (FV1000; Olympus) with a 60X NA 1.10 water immersion objective (Olympus). Images were acquired using Fluoview software version 1.6 (Olympus), and analyzed using ImageJ (NIH) or Adobe Photoshop software (Adobe).

Cell fractionation and Immunoblot analysis

Neurons were homogenized with a Dounce tissue grinder and centrifuged at 1k × g for 10 min to remove unbroken cells and large fragments. A crude synaptosomal pellet was collected after further high speed centrifugation at 12k × g for 15 min. The synaptosomal pellet was resuspended in HEPES buffer supplemented with 1 volume of 320 mM sucrose and lysed with 9 volumes of ice-cold water, followed by centrifugation at 66K × g for 25 minutes. The supernatant was collected and mixed with equal volumes of Optiprep solution and layered with 30% and 0% Optiprep. The step gradient was centrifuged at 20K × g for 5 hours and SVs were collected from the 0 ~ 30% interface. The protein concentration was determined using the BCA assay and 1 μg of each sample was resolved by SDS-PAGE. The samples were transferred to nitrocellulose and immunoblotted for SV and plasma membrane markers.

We thank E. Hui for providing constructs used for the electrophysiology experiments in Fig. 3. We also thank M.B. Jackson, X. Lou, and the Chapman lab for their comments on this manuscript. This study was supported by a grant from the National Institutes of Health (MH 61876). J.Y. is supported by an American Heart Association postdoctoral fellowship (11POST5720016). S.E.K was supported by an Epilepsy Foundation predoctoral fellowship. E.R.C. is an Investigator of the Howard Hughes Medical Institute.