Soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complexes execute synaptic vesicle (SV) fusion. Vesicle fusion is preceded by an obligatory Munc13-dependent priming process that conveys fusion competence to SVs by facilitating SNARE complex assembly. Ultrastructural studies after chemical fixation indicated that vesicle docking to the plasma membrane is independent of Munc13s but these results may be misleading because aldehyde fixatives modify the localization of SVs with respect to the plasma membrane. To reinvestigate the role of Munc13s in vesicle docking, cultured hippocampal slices were immobilized using high-pressure freezing, which circumvents aldehyde artifacts. High-pressure freezing was combined with electron tomography to reach a resolution that allows the characterization of details of SV docking in a close-to-native state. In control slices, docked vesicles are not hemifused with the plasma membrane but linked to it and to dense material at the active zone by small strands. In slice cultures from Munc13-deficient mice, vesicles are not docked to the active zone plasma membrane. These results indicate that SV docking at the plasma membrane and functional priming are respective morphological and physiological manifestations of the same molecular process mediated by SNARE complexes and Munc13s.
Action potentials trigger neurotransmitter release on a sub-millisecond time scale by inducing the calcium-dependent fusion of synaptic vesicles (SVs) at the active zone (AZ) plasma membrane (Sabatini & Regehr, 1996; Lisman et al., 2007). To allow such rapid excitation/secretion coupling, SVs are docked to the plasma membrane at the site of release and primed to fusion competence. Fusion of SVs with the plasma membrane requires the formation of the soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex (Sudhof, 2004; Jahn & Scheller, 2006). Munc13 proteins regulate this SNARE complex assembly process and thereby prime SVs to fusion competence (Richmond et al., 2001; Wojcik & Brose, 2007; Rizo & Rosenmund, 2008). Indeed, Munc13-mediated SV priming is absolutely essential for SV fusion. Synapses lacking Munc13s, e.g. in hippocampal neurons of Munc13-1/2 double knock-out (DKO) mice, are entirely devoid of fusion-competent SVs and are, as a consequence, completely unable to execute spontaneous or evoked SV fusion (Varoqueaux et al., 2002). Similar observations were made in Caenorhabditis elegans (C. elegans) and Drosophila mutants lacking the only nematode and fly Munc13 homologues Unc-13 and Dunc-13, respectively (Aravamudan et al., 1999; Richmond et al., 1999). Interestingly, initial electron microscopic studies of these mutants after aldehyde fixation showed no loss of SVs docked at the AZ (Aravamudan et al., 1999; Richmond et al., 1999; Varoqueaux et al., 2002). These findings led to the conclusion that Munc13 proteins act at a post-docking priming step, which renders SVs fusion-competent. However, recent electron microscopic studies on Unc-13- and Syntaxin-deficient synapses in C. elegans mutants that had been physically immobilized by high-pressure freezing (HPF) uncovered a loss of docked vesicles in both types of mutant synapses (Weimer et al., 2006; Hammarlund et al., 2007). The contradiction between these recent findings and earlier reports can be explained by the difference in the fixation methods used, as aldehydic fixatives modify SV localization with respect to the plasma membrane (Siksou et al., 2007).
In view of the conflicting datasets on SV docking in Munc13-, Unc-13- and Dunc-13-deficient synapses, we reinvestigated the role of Munc13 proteins in SV docking by examining the ultrastructure of wild-type and Munc13-deficient (Munc13-1/2 DKO) mouse brain synapses using HPF immobilization. HPF permits a very rapid (∼10 ms) immobilization of the tissue. However, it allows proper freezing only at the border of nerve tissue slices (Rostaing et al., 2006). As acute slices present dissection artifacts at their borders, HPF was performed on cultured brain slices. Indeed, during time in culture, disrupted cellular elements regress and cells rearrange so that the slice reforms a continuous border without dissection artifacts (Buchs et al., 1993). In this preparation, the maturation of synapses is very similar to that in the in-vivo situation (Buchs et al., 1993; Muller et al., 1993). Thus, HPF can instantly immobilize physiologically active synapses. Moreover, as reported here, synapses can be obtained in slice cultures from embryos of synaptic protein mutants that die at birth. Classical transmission electron microscopy does not permit accurate localization of SVs with respect to the plasma membrane because of the thickness of the sections. Therefore, HPF of slice cultures was combined with electron tomography to increase resolution in the z-axis. This allows the study of SV docking and its alteration in Munc13-1/2 DKO.
Materials and methods
Hippocampal slice cultures
All relevant experimental procedures followed the guidelines of the European Commission and were approved by the Regional Ethics Committee no 3 of Ile-de-France region on Animal Experiments. Hippocampal slice cultures from two 7-day-old C57/Bl6 male mice (Janvier, Le Genest-St-Isle, France) were prepared according to the protocol developed by Stoppini et al. (1991). Briefly, the pups were decapitated and their brains rapidly transferred to ice-cold dissection medium. The hippocampus was dissected and 400 μm coronal slices were sectioned using a McIllwain tissue chopper (Mickle Laboratory, Surrey, UK). Slices were placed on Millicell-CM inserts (Millipore, Billerica, MA, USA) and maintained in slice culture medium for 3 weeks. Slices were cultured with appropriate culture medium (Poncer et al., 2002). The slice culture medium was made of 10.63 g/L of minimum essential medium (MEM) Eagle medium (50-019-PC, Cellgro) reconstituted in milliQ water, 20% heat-inactivated horse serum, 1 mm l-glutamine, 1 mm CaCl2, 2 mm MgSO4, 1 mg/L insulin (I5500; Sigma-Aldrich), 61 μm ascorbic acid (A4034; Sigma-Aldrich), 11 mm d-glucose, 5 mm NaHCO3 and 30 mm Hepes (223–778; Boehringer). The pH was adjusted to 7.27. As Munc13-1/2 DKOs die immediately after birth, organotypic hippocampal slice cultures were prepared from embryonic day 18 mice. Pregnant mice were killed by cervical dislocation, embryos were removed and the slices prepared as described above. These slices were cultured for 5 weeks. Three Munc13-1/2 DKO mice and three Munc13-1/2 double heterozygous control littermates, which are phenotypically identical to wild-type mice (Varoqueaux et al., 2002), were processed from two different litters in two different sessions, so that two independent experiments were performed.
Hippocampal slices from post-natal day 7 mice cultured for 3 weeks were immersed in slice medium and detached from the Millicell-CM membrane with a razor blade. Slices were then transferred to an electrophysiology setup. Neurons were recorded in voltage clamp (−70 mV) in the whole-cell configuration using a Multiclamp 700B amplifier (Axon, Sunnyvale, CA, USA). Recordings were filtered at 2–5 kHz and acquired at 5–10 kHz. Currents were recorded in voltage clamp. Throughout the experiment, the access resistance was periodically tested and, if the access resistance changed by more than 10% or was higher than 20 MΩ, the cells were discarded. Data were acquired using clampex 10 software and analysed using the clampfit 10 program (Axon). Six neurons from six different slices and three different experiments were patch clamped. All neurons exhibited spontaneous activity. In three of the cells, d(−)-2-amino-5-phosphonopentanoic acid (D-APV) (50 μm) and 2-3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[g]quinoxaline-7-sulfonamide disodium salt (NBQX) (50 μm) were applied, which blocked spontaneous activity.
Slices were immersed in slice medium and detached from the Millicell-CM membrane with a razor blade. Whole slices were immediately transferred to a high-pressure chamber and frozen in an HPF machine (HPM 010; Bal-Tec, Leichtenstein). After freezing, the samples were rapidly transferred to liquid nitrogen for storage.
Cryosubstitution and embedding
Cryosubstitution and embedding were performed in a Reichert AFS apparatus (Leica, Vienna, Austria), as described previously (Rostaing et al., 2006). Briefly, cryosubstitution was performed in acetone with 0.1% tannic acid at −90°C for 4 days, followed by acetone with 2% osmium during the last 7 h. The slices were warmed (5°C/h) to −20°C and incubated for an additional 16 h, before being warmed (10°C/h) to 4°C. At 4°C, the slices were washed in acetone and then warmed to room temperature (20°C). They were then embedded in araldite.
Sectioning, counterstaining and examination of the sections
Ultrathin (40–70 nm) sections were cut using an EM UC6 (Leica Microsystems, Wetzlar, Germany) and collected on 400 mesh copper grids. For tomographic analysis, 100-nm-thick sections were collected and incubated for 2 h with immunogold-conjugated Protein A 10 nm (EMPAG10; BritishBiocell International, Cardiff, UK) (1 : 2 in filtered phosphate-buffered saline). All sections were stained by incubation with 5% uranyl acetate in 70% methanol for 5 min and then with lead citrate (0.08 m lead nitrate, 0.12 m sodium citrate in CO2-free dH2O) for 5 min. The sections were observed with a Philips TECNAI 12 (FEI, Eindhoven, the Netherlands).
Sampling of the synapses
All experimental data sets were focused on spine synapses. Spine synapses were discriminated from shaft synapses on the basis of the small size of the post-synaptic profile and the absence of mitochondria and microtubules in the post-synaptic compartment (Peters et al., 1991). The 40–100-nm-thick sections used for electron microscopy did not include synapses in their entire width. Thus, synapses analysed in this study are referred to as synaptic profile. Segregation patterns of nuclei that are probably due to the nucleation of small ice crystals (Hohenberg et al., 1996) start at about 5 μm from the border of the slices and increase towards the center. Thus, analyses were performed within 5 μm from the border of the slices. This thin depth of tissue frozen without segregation patterns may be due to the fact that cultured brain slices contain more water and are thus more difficult to freeze than acute brain slices (Moor, 1987; Siksou et al., 2009).
The electron microscope was operated at 120 kV. Thirty-five synapses were recorded with a 1300 × 1030 Dualvision 300W CCD camera (Gatan, Evry, France) at 87 000-fold magnifications (corresponding to a pixel size of 0.60 nm). Tilted series (from about −60° to +60°, with 1° increments) were acquired using the DigitalMicrograph program (Gatan). The sections were rotated by 90° and another series was acquired, allowing better imaging of the synapse as the missing information was reduced. Images of each tomographic tilt series were aligned, the tilt angle was calculated by tracking of the gold particles and the final volume was reconstructed with a weighted back-projection algorithm, using the imod software for dual axis reconstruction (Kremer et al., 1996). Synaptic profiles from eight control and 10 Munc13-1/2 DKO synapses were analysed.
The amira software (Version 4.1.2; Mercury Computer Systems, San Diego, CA, USA) was used for the three-dimensional reconstruction of the synapses. The docked SVs, membranes and cytoskeletal elements at the AZ were reconstructed by semi-automatic rendering. Alternatively, SVs were spotted using spheres of the corresponding diameter. The AZ was defined as the pre-synaptic plasma membrane next to the electron-dense material filling the synaptic cleft and/or of the postsynaptic density. This is in accord with the definition of the synaptic junction at interneuronal chemical synapses (Peters et al., 1991). Perforated synapses were not observed in our samples, probably due to their low frequency of occurrence in young animals (Harris et al., 1992).
All quantifications were performed blindly. Three Munc13-1/2 DKO mice and three Munc13-1/2 double heterozygous control littermates from two independent experiments were used for quantification. To quantify the total number of SVs per synaptic profile, photomicrographs of 139 synaptic profiles from three Munc13-1/2 DKO mice and 140 synaptic profiles from three control littermates were taken from ultrathin sections at a 60 000-fold magnification. The statistical significance of differences in the total number of SVs was calculated using a two-tailed, unpaired t-test on the means obtained from the three animals per genotype. To quantify the number of docked vesicles, tomographic reconstructions of synaptic profiles from eight control and 10 Munc13-1/2 DKO synapses were analysed. SVs were scored as ‘docked’ when the outer layer of their membrane was contacting the inner layer of the plasma membrane. The distance of the SV with respect to the plasma membrane was measured with ImageJ software (Rasband, 1997).
Ultrastructure of pre-synaptic terminals after high-pressure freezing of cultured slices
Hippocampal slices from 7-day-old wild-type mice were cultured for 3 weeks. Neurons from these cultures exhibited spontaneous glutamatergic activity (Fig. 1A). HPF was then performed on these organotypic hippocampal slices. The appearance of nuclei was analysed to evaluate the quality of freezing. Nuclei next to the slice border were well preserved (Fig. 1B) but appeared segregated beyond 5 μm from the slice border, indicating the nucleation of ice crystals (Fig. 1C) (Hohenberg et al., 1996). Spine synapses were therefore analysed within the first 5 μm from the slice border, which contained numerous synapses (Fig. 1D).
Virtual sections from tomographic reconstructions allowed a detailed visualization of the cytoskeleton. One to five microtubules were observed per synaptic profile. They were located at the center or at the periphery of the bouton. In a few instances, they were linked to SVs by small filaments (data not shown). In addition, long and straight filaments were observed in some reconstructions, oriented in every direction and sometimes towards the AZ (Fig. 2A). On occasion, SVs were connected to these filaments (Fig. 2B). In addition to links between SVs and cytoskeletal elements, short and occasionally V-shaped strands linking SVs together were present in all of the tomographic reconstructions analysed (Fig. 2A and B). On average, a single SV was directly linked to 1.7 other SVs by these small tethers (n = 35 SVs analysed). Longer links between SVs and the plasma membrane at the AZ were also observed (Fig. 2C).
Synaptic vesicle docking
Electron tomography allowed a clear identification of docked vesicles (Fig. 3A–C), as virtual sections from tomographic reconstructions have a 5–10 nm resolution in the z-axis. It has been proposed that primed vesicles might be stabilized as hemifusion intermediates, with their outer membrane leaflet continuous with the inner leaflet of the plasma membrane (Kesavan et al., 2007; Wong et al., 2007; Chernomordik & Kozlov, 2008). To evaluate SV hemifusion, 13 synaptic profiles were sampled in 100-nm-thick sections of wild-type synapses. Membrane hemifusion was never observed in the 37 docked vesicles analysed (Fig. 3A and B); rather docked vesicles were always apposed to the plasma membrane.
Also, small filaments linked some of the docked SVs to the pre-synaptic membrane (Fig. 3C) and multiple fine strands contacting the plasma membrane were found in the vicinity of docked SVs (Fig. 3D and E). In three-dimensional reconstructions, this strand-like material did not cover the plasma membrane homogeneously and did not seem to have a regular organization. However, it always surrounded docked SVs. Filamentous links were observed between docked SVs and the pre-synaptic dense material (Fig. 3C).
Munc13s are required to dock synaptic vesicles at the plasma membrane
To investigate the role of Munc13s in SV docking, hippocampal synapses from Munc13-1/2 DKOs were examined. Because these mutants, which are defective for synaptic transmission, die at birth, hippocampal slices were obtained from embryonic day 18 Munc13-1/2 DKOs and control littermates, and cultured for 5 weeks. Gross examination of ultrathin sections indicated that synapses developed normally in this preparation (Fig. 4A). The overall number of SVs per synaptic profile in Munc13-1/2 DKO samples was similar to the controls (33 ± 4 vesicles per profile in Munc13-1/2 DKOs and 26 ± 2 in controls, mean ± SEM, P = 0.2, Fig. 4B). In tomographic reconstructions, control spine synapses (n = 8) had a similar morphology as synapses in slices obtained from post-natal day 7 mice and cultured for 3 weeks, with some of the SVs directly apposed to the membrane (Fig. 5A and C). Furthermore, the mean number of SVs within the first 40 nm from the AZ was similar in tomographic reconstructions of Munc13-1/2 DKO (5.2 ± 0.6, mean ± SEM; 10 synapses from n = 3 animals) and control synaptic profiles (6.3 ± 1.4, mean ± SEM; eight synapses from n = 3 animals). However, Munc13-1/2 DKO spine synapses displayed an almost complete loss of SVs directly contacting the plasma membrane (Fig. 5B, D and F). Only one out of 51 SVs examined within the first 40 nm from the AZ was docked at the AZ, whereas 26 out of 48 SVs were docked in control synapses. In the Munc13-1/2 DKO, SVs remained at a short distance from the membrane (Fig. 5B and F) and sometimes a filament linking these SVs to the plasma membrane was observed (Fig. 5B).
Electron tomography of synapses after HPF of organotypic slice cultures has opened new perspectives to the study of synapse ultrastructure. It permits the discrimination of detailed morphological features such as the cytoskeleton or the tethering and docking of SVs. In addition, synapses from mutant embryos that die at birth can be analysed. Using this method, we observed that docked SVs are not hemifused with the plasma membrane but linked to it and to the dense material at the AZ by small strands. In synapses from Munc13-deficient mice, SVs are not docked at the plasma membrane but remain at a short distance from the membrane.
Advantages of high-pressure freezing of slice cultures to study the pre-synaptic cytoskeleton
The organization of the pre-synaptic cytoskeleton in cultured slices is similar to that previously observed in acute slices (Siksou et al., 2007). In both preparations, small filaments were observed linking SVs together. The average number of filamentous connections between SVs in the cultured slices reported here (1.7) is very similar to that described in acute slices (1.5). In accord with a previous study, longer links between SVs and the AZ plasma membrane were observed. In addition, long and straight actin-like filaments were detected. The filaments emerging from the AZ with vesicles along their length previously observed in acute slices could be similar microfilaments (Siksou et al., 2007). Also, in both preparations docked SVs are linked to the plasma membrane by small strands. Thus, electron tomography after HPF of cultured slices confirms the results obtained previously with acute slices, indicating that this in-vitro preparation is very well suited to the study of synapses with a structure close to the in-vivo situation.
In addition, finer details of pre-synaptic structure were observed following HPF of slice cultures. The microfilaments, as well as the small links between SVs, appeared thinner than in acute slices (Siksou et al., 2007). Further, pre-synaptic densities comprised many small filamentous structures rather than appearing as compact structures, which corroborates a previous quick-freeze study (Landis et al., 1988). Filaments linking docked vesicles to this pre-synaptic dense material were observed, which is reminiscent of the dense material that anchors SVs at the frog neuromuscular junction (Harlow et al., 2001). This better visualization of the cytomatrix is probably due to the better freezing of synapses in the sampled area (Siksou et al., 2009). Contrary to acute slices, the border of slice cultures is devoid of dissection artifacts. At this border, the freezing is optimal so that the synaptic ultrastructure can be analysed without ice crystal artifacts.
Mode of synaptic vesicle docking
Hemifusion of docked SVs was never observed in the slices studied here, despite the fact that synapses in the preparation are spontaneously active. A recent electron tomography study of mammalian central synapses after chemical fixation indicated that 74% of docked SVs are hemifused with the plasma membrane (Zampighi et al., 2006). In view of the present findings, these numerous hemifusion structures are likely to have been induced by aldehydes, as these fixatives stimulate SV exocytosis (Smith & Reese, 1980). Indeed, our results indicate that docked SVs are not hemifused. Rather, hemifusion probably represents a rapidly induced and very unstable transition state between docking and fusion.
In Munc13-1/2 DKO synapses, SVs were not apposed to the plasma membrane but were sometimes linked to it by small filaments. This indicates that SVs can be attached to the AZ by tethers before being docked. This scenario is similar to the situation in endocrine cells, where tethered granules perform a 20 nm step toward the plasma membrane and become docked upon Ca2+ stimulation (Karatekin et al., 2008).
Role of Munc13s in synaptic vesicle docking
The observation that Munc13s are required for SV docking complements and extends the results of recent studies on the ultrastructure of C. elegans synapse, which showed that Unc-13 activity and the presence of the target-SNARE. Syntaxin are required for the contact formation between SVs and the pre-synaptic plasma membrane (Weimer et al., 2006; Hammarlund et al., 2007). Munc13s are thought to facilitate SNARE complex formation by inducing a conformational switch in Syntaxin from a ‘closed’ to an ‘open’ conformation or by providing a template for the assembly of Syntaxin-1/synaptosomal-associated protein of 25 kDa (SNAP-25) heterodimers (Wojcik & Brose, 2007; Rizo & Rosenmund, 2008). In view of this mechanistic model of Munc13 action, the fact that Munc13/Unc-13 and Syntaxin deficiencies lead to the same SV docking defect indicates that SNARE complexes, assembled or stabilized by accessory proteins such as Munc13s, maintain SVs docked to the membrane in a fusion-competent state. Accordingly, morphological docking and physiological priming are not sequential events but rather two different manifestations of the same molecular process that is mediated by Munc13s and SNARE complexes.
We thank O. Medalia and T. Boudier for helpful discussion, and M. Zhen and J.L. Bessereau for critical reading of this manuscript. This work was supported by grants from Fondation pour la Recherche Médicale, Fédération pour la Recherche sur le Cerveau, Rotary International, Agence Nationale de la Recherche, Associaton Française contre les Myopathies and the Max Planck Society. S.M. is a CNRS investigator.
soluble N-ethylmaleimide-sensitive factor attachment protein receptor