SYNAPSE 00:00–00 (2014)

A Nanoscale Resolution View on Synaptic Vesicle Dynamics DARIO MASCHI,1,2 AND VITALY A. KLYACHKO1,2* Department of Cell Biology and Physiology, Center for Investigations of Membrane Excitability Diseases, Washington University School of Medicine, St. Louis, Missouri 2 Department of Biomedical Engineering, Washington University, St. Louis, Missouri

1

KEY WORDS

vesicle recycling; single particle tracking

ABSTRACT The ability of synapses to sustain neurotransmitter release during continuous activity critically relies on an efficient vesicle recycling program. Despite extensive research on synaptic function, the basic mechanisms of vesicle recycling remain poorly understood due to the relative inaccessibility of central synapses to conventional recording techniques. The extremely small size of synaptic vesicles, nearly five times below the diffraction-limited resolution of conventional light microscopy, has hampered efforts to define the mechanisms controlling their cycling. The complex sequence of dynamic processes that occur within the nerve terminals and link vesicle endocytosis and the subsequent round of release has been particularly difficult to study. The recent development of nanoscale-resolution imaging techniques has provided an opportunity to overcome these limitations and begin to reveal the mechanisms controlling vesicle recycling within individual nerve terminals. Here we summarize the recent advances in the implementation of super-resolution imaging and single-particle tracking approaches to study the dynamic steps of the vesicle recycling process within presynaptic terminals. Synapse 00:000–000, 2014. VC 2014 Wiley Periodicals, Inc.

INTRODUCTION Neural activity places stringent demands on the continuous supply of synaptic vesicles to sustain neurotransmitter release. The majority of central synapses contains a very small pool of releasable vesicles (Harata et al., 2001), and therefore orchestrate a precisely controlled vesicle recycling program to sustain and control release. Several distinct steps in the vesicle recycling process have been proposed. These include, following a fusion event, (but not necessarily in this order): endocytosis, refilling with neurotransmitter, mobilization and repositioning to the active zone, docking, priming, and subsequent release (Denker and Rizzoli, 2010; Fernandez-Alfonso and Ryan, 2006). This complex process has been estimated to take at least 30 s based on the measurements of vesicle turnover in central nerve terminals (Fernandez-Alfonso and Ryan, 2006). Yet the kinetics of different stages in the vesicle cycle have thus far been difficult to determine and the rate-limiting steps in the recycling process remain poorly understood. Because demand for vesicles increases greatly during periods of high-frequency firing, it has been long hypothesized that some of the steps in the recycling process are facilitated by neural activity, presumably via changes in intraterminal calcium levels, to Ó 2014 WILEY PERIODICALS, INC.

promote vesicle availability for release. Despite extensive research, however, it remains largely unknown how neural activity regulates the vesicle cycle and availability for reuse. Development of pH-sensitive genetically encoded indicators known as synaptopHluorins have permitted analysis of fusion events at single-vesicle resolution and provided a wealth of information on a phase of the vesicle cycle that occurs at the plasma membrane, particularly the processes of exo- and endocytosis (Kavalali and Jorgensen, 2013). Yet we know much less about the events that occur in the interior of the nerve terminals, which are particularly difficult to access. Vesicle mobility during the recycling process has attracted particularly extensive research efforts because of its hypothesized role as a ratelimiting step in the recycling process (Shtrahman et al., 2005; Yeung et al., 2007). The conventional *Correspondence to: Vitaly A. Klyachko, Departments of Cell Biology and Physiology, Biomedical Engineering, Center for Investigations of Membrane Excitability Diseases, Washington University School of Medicine, St. Louis, MO, USA. E-mail: [email protected] Received 7 October 2014; Revised 20 November 2014; Accepted 27 November 2014 DOI: 10.1002/syn.21795 Published online 00 Month 2014 in Wiley Online Library (wileyonlinelibrary.com).

2

D. MASCHI AND V.A. KLYACHKO

view on vesicle recycling suggests that following endocytosis, newly formed vesicles must undergo physical relocation within the terminals to reach the subsequent fusion sites (Denker and Rizzoli, 2010). This view is based on the observations that newly endocytosed vesicles are located throughout the nerve terminal, often hundreds of nanometers away from the active zone (Schikorski and Stevens, 2001; Schikorski, 2014), suggesting that some form of motion or a transport mechanism is employed by the synapses to deliver vesicles to the fusion sites. This model is further supported by recent evidence using ultrafast optogenetics-based flash-and-freeze electron microscopy showing that sites of exo- and endocytosis are spatially separated in both peripheral and central synapses (Watanabe et al., 2013a,b). Several recent models extend this view by incorporating an intermediate immobile state such that following endocytosis, vesicles incorporate into an immobile cluster and then are subsequently mobilized for fusion when neural activity occurs. Vesicles are thus thought to assume multiple dynamic states within the nerve terminals that represent different stages in the vesicle cycle (Denker and Rizzoli, 2010). Until recently, however, defining such vesicle dynamic states and mobilization mechanisms in central synapses has been limited by the inability of existing techniques to resolve these processes at the level of individual vesicles. In neuroendocrine cells and in ribbon synapses, some of these questions could be approached with total internal reflection fluorescence (TIRF) microscopy. Vesicles exhibited high mobility in their approach to the active zone, which was directed and targeted toward release sites (Holt et al., 2004; Zenisek et al., 2000). However, TIRF has not thus far been successfully applied to study vesicle dynamics in mammalian small central synapses. Understanding the dynamic components of vesicle recycling has been further complicated by earlier findings indicating that the majority of vesicles are largely immobile within synaptic terminals both at rest and during neural activity (Gaffield et al., 2006; Henkel et al., 1996; Kraszewski et al., 1996; Lemke and Klingauf, 2005; Shtrahman et al., 2005). These and other studies led to the development of a stickand-diffuse model and the hypothesis that vesicle mobility toward the active zone is the rate-limiting step in the recycling process (Yeung et al., 2007). Indeed, in both the neuromuscular junction (NMJ) and central neurons, the very limited vesicle mobility could be increased several fold by a phosphatase inhibitor okadaic acid (Henkel et al., 1996; Shtrahman et al., 2005), suggesting that vesicle mobility is reduced greatly by binding to the synaptic structural elements in a phosphorylation-dependent manner, thus keeping the majority of vesicles immobile. These earlier studies, however, were performed at room Synapse

temperature and using bulk measurements of vesicle dynamics. More recent analyses have demonstrated that vesicle mobility is strongly temperature dependent and is apparent only at physiological temperatures both in the NMJ and in central neurons (Gaffield and Betz, 2007; Peng et al., 2012). This notion is further supported by the findings that synaptic processes that rely on vesicle recycling, such as recovery from vesicle depletion, are markedly temperature dependent (Klyachko and Stevens, 2006; Kushmerick et al., 2006). Elegant recent studies also highlighted the large heterogeneity in vesicle populations within individual synapses, both in terms of their composition and release properties (Bal et al., 2013; Chung et al., 2010; Fredj and Burrone, 2009; Hua et al., 2011b; Raingo et al., 2012; Sara et al., 2005). Such intrinsic vesicle heterogeneity further underscores the limitations of using bulk measurements to understand vesicle dynamics during recycling. Development of nanoscale-resolution single-particle tracking (SPT) approaches and super-resolution imaging techniques over the last decade have for the first time allowed investigators to approach these fundamental questions in central nerve terminals at the level of individual synaptic vesicles. In this review we summarize the application of nanoscaleresolution imaging approaches to studies of synaptic vesicle recycling with a particular focus on hightemporal resolution techniques that allow rapid tracking of individual synaptic vesicle behavior during recycling. Overcoming the diffraction limit of resolution The resolution of the optical microscope is limited by the wave nature of light. Consequently, light emitted by even the smallest fluorescently-labeled object always results in a diffracted spot represented by the point spread function (PSF) of the optical system. The ability of the optical system to separate two such point light sources defines the resolution limit, which for light microscopy applications is determined by the formula originally developed by Ernst Abbe in 1873, to be 200–250 nm (Hell, 2007). The imaging techniques that achieve resolution below the diffraction limit are collectively known as super-resolution microscopy. These techniques work around the diffraction barrier, rather than breaking it. The main idea behind many of these methods is to detect single light sources (or few light sources in a subdiffraction-sized region) and rely on the principle that a single fluorescent light source can be localized with very high accuracy if enough numbers of photons are collected (Thompson et al., 2002). Many nanoscale resolution imaging techniques2including all SPT approaches (Alcor et al., 2009), Fluorescence Imaging with One Nanometer Accuracy (FIONA) (Park et al., 2007), PhotoActivated

3

SYNAPTIC VESICLE DYNAMICS

Fig. 1. Nanoscale resolution imaging techniques to visualize dynamic synaptic processes. A–D. Single particle tracking. Movies of sparsely labeled synaptic vesicles in cultured neurons are acquired with high temporal resolution (A); each vesicle’s image is fitted by a Gaussian representation of the PSF to determine the peak location with sub-pixel accuracy (B) trajectories of individual vesicles are assembled from their localizations in each frame (C). Sample track of a synaptic vesicle moving within a hippocampal

synapse at 37 C determined at 35 frames/s using an SPT approach described in Peng et al. (2012) (D). E, F. STED. Comparison of confocal and STED images of synaptic vesicles in hippocampal cultures labeled with Syt1 antibody (E). By dramatically reducing the effective PSF size, STED improves resolution 10 fold compared with confocal imaging (F). (E) and (F) are modified from Westphal et al. (2008).

Localization Microscopy (PALM) (Betzig et al., 2006), Super Resolution Fluorescence Localization Microscopy (SRFLM) (Nelson and Hess, 2014), and Stochastic Optical Reconstruction Microscopy (STORM) (Rust et al., 2006)2employ this basic concept. An alternative approach based on application of two laser beams to create stimulated emission depletion (STED) process is another widely used technique to achieve nanoscale resolution (Hell, 2007). These techniques have been described in detail in several excellent recent reviews (Alcor et al., 2009; Bates et al., 2008; Hell, 2007; Huang et al., 2009; Maglione and Sigrist, 2013; Nelson and Hess, 2014; Sigrist and Sabatini, 2011). In recent years, application of these techniques to neurobiology has revealed a wealth of structural information about synaptic organization. Most of these approaches have intrinsic temporal resolution limits that make them particularly useful to study synaptic organization, but limits their access to millisecond-timescale dynamic synaptic processes. Among nanoscale-resolution approaches, SPT and STED are the two major approaches with temporal resolution on the millisecond scale (Alcor et al., 2009; Hell, 2007). Consequently

these are the two main nanoscale-resolution approaches that have been predominately used in studies of dynamic synaptic processes. In this review we describe these relevant techniques and the recent advances in our understanding of rapid synaptic processes revealed by application of these nanoscopy approaches. Nanoscale-resolution approaches to study dynamic synaptic processes Single-particle tracking (SPT) This family of nanoscale-resolution approaches is based on a principle that the spatial localization of an isolated fluorescent particle is not subjected to diffraction limit of resolution. The image of a single particle is represented by the PSF of the optical system (Figs. 1A–1C), and its peak position can be localized with a precision many times greater than the diffraction limit. In some cases, it has been possible to achieve fluorophore localization with one-nanometer accuracy giving rise to FIONA (Park et al., 2007). Typically the localization of the fluorophore is Synapse

4

D. MASCHI AND V.A. KLYACHKO

determined by approximating PSF with a Gaussian fit to the image intensity data (Figs. 1B and 1C). However, Gaussian fitting is not the only way to determine the localization; other mathematical algorithms have been shown to be more efficient in terms of computational cost and speed, enabling real-time super-resolution imaging analysis (Small and Stahlheber, 2014; Smith et al., 2010; Wolter et al., 2010). Individual synaptic vesicles can be fluorescently labeled in nerve terminals via a number of approaches, most commonly via compensatory endocytosis and uptake from the extracellular fluid of fluorescent lipophilic dyes (such as FM1-43 or SCG5) (Hoopmann et al., 2012; Peng et al., 2012; Wu et al., 2009), Quantum dots (Q-dots) (Lee et al., 2012; Park et al., 2012; Zhang et al., 2009), or fluorescently labeled antibodies (Abs) against specific vesicular proteins (such as synaptotagmin I (syt I)) (Hell, 2007). Since synaptic vesicles are subresolution organelles with a size 40 to 50 nm, they can be viewed to represent a point light source and therefore their localization can be determined by PSF fitting with very high accuracy, reaching 20 nm in central neurons (Fig. 1D) (Park et al., 2012; Peng et al., 2012) and temporal resolution of over 125 frames/s (D.M. and V.A.K., personal communication). A necessary condition for the application of SPT technique is sparse vesicle labeling with only one or a few synaptic vesicles labeled per synapse. This is typically achieved by controlling the number of stimuli used to evoke release and compensatory vesicle uptake. At physiological temperatures, single AP stimulation in cultured hippocampal neurons leads to an uptake of one or two vesicles at most per synapse in the majority of cases (Peng et al., 2012). Adjustments of the strength of the stimulus and the timing of dye application further allow specific labeling of early, late and spontaneously endocytosed vesicles (Aravanis et al., 2003; Lemke and Klingauf, 2005; Murthy and Stevens, 1998; Peng et al., 2012; Ryan et al., 1997). In combination with additional postprocessing computational tools, such as multiple-hypothesis tracking, SPT can be used to achieve super-resolution tracking of multiple simultaneously imaged particles (Jaqaman et al., 2008). The key advantage of SPT over other super-resolution approaches, particularly in applications to studies of dynamics processes, is very high temporal resolution. SPT temporal resolution is only limited by the signal-to-noise ratio and has already achieved 125 frames/s for single synaptic vesicles (D.M. and V.A.K., personal communication). This approach has also been applied to simultaneously track multiple particles in large and densely populated fields of view, allowing simultaneous sampling of tens and, in some cases, hundreds of single fluorescent particles (Jaqaman et al., 2008). The main limitation of this approach is the requirement Synapse

for sparse particle labeling that has thus far prevented the use of genetically encoded fluorescent labels. Stimulated emission depletion (STED) STED is a breakthrough super-resolution imaging technique pioneered by Stefan Hell and colleagues, with exceptionally high temporal and spatial resolution. Many excellent reviews of STED are available (Hell, 2003, 2007; Hell et al., 2004). In brief, in STED the sample is illuminated by two laser beams: first the dye is excited using a sub-picosecond laser pulse producing the typical diffraction-limited spot of excited fluorophores (excitation pulse). A second, redshifted pulse laser (depletion pulse) that immediately follows the first pulse is used to return the fluorophores to their ground state. The depletion pulse or STED pulse is focused in a toroid-like shape, which effectively allows only the fluorophores in the center of the toroid to emit light. This stimulated depletion reduces the effective size of the PSF to as little as 20 nm in fixed tissue and 50 nm in live tissue in the XY dimensions (Figs. 1E and 1F). Recent studies have demonstrated the ability of STED to reach video rates and have extended its applications to threedimensional imaging with axial resolution of 100 nm (Osseforth et al., 2014; Westphal et al., 2008; Wildanger et al., 2009). More recently STED microscopy with continuous wave beams (CW STED) in video-rate STED recording of living neurons has been demonstrated. Unlike pulse excitation, with continuous-wave excitation, the emission occurs continuously, allowing capture of more photons per unit of time (Lauterbach et al., 2010; Willig et al., 2007). In applications to dynamic synaptic processes, STED offers exceptional spatial and temporal resolution, with its only downside being a relatively limited field of view. This limitation arises from the necessity to scan the field of view, limiting the effective imaging area to 1.8 3 2.5 mm (Westphal et al., 2008; Willig et al., 2006). Synaptic vesicle dynamics during recycling The initial application of nanoscale resolution imaging in central synapses to study synaptic vesicle dynamics used the lipophilic styryl dye FM1–43 to label individual synaptic vesicles via compensatory endocytosis and analyzed by laser-scanning microscopy (Lemke and Klingauf, 2005). In line with the earlier bulk measurements of vesicle dynamics, this work also showed that synaptic vesicle mobility in cultured hippocampal neurons is very low at rest and remains very low during stimulation. The vesicle motion was also highly restricted and consistent with obstructed or caged diffusion (also known as anomalous subdiffusion), with a maximal cage radius of

SYNAPTIC VESICLE DYNAMICS

Fig. 2. Extensive synaptic vesicle mobility within central nerve terminals revealed by nanoscale-resolution imaging. A. Sample synaptic vesicle trajectories in hippocampal neurons at 37 C determined using an SCG5 lipophilic dye labeling and an SPT approach. Left: trajectory of a vesicle undergoing activity-evoked recycling; Right: trajectory of a vesicle undergoing spontaneous recycling. Modified from Peng et al. (2012). B. Sample trajectories of synaptic vesicles labeled with Q-dots in hippocampal neurons and determined using a threedimensional SPT approach. Modified from Park et al. (2012). C. Sample STED images and trajectories of synaptic vesicles labeled with Syt 1 antibodies. Modified from Westphal et al. (2008).

0.1 lm for vesicles during resting conditions. The average vesicle mobility was indistinguishable in live and paraformaldehyde-fixed cultures, indicating that vesicles inside the nerve terminals are effectively immobile. The authors also did not observe major differences in vesicle mobility of early, late, and spontaneously endocytosed vesicles (Lemke and Klingauf, 2005). This lack of vesicle mobility at rest and even shortly prior to fusion is difficult to reconcile with the large scatter of the recently endocytosed vesicles (Schikorski and Stevens, 2001; Schikorski, 2014) suggesting the need for large-scale motion from the endocytic sites toward the active zone. Furthermore, earlier studies using TIRF microscopy have demonstrated that extensive and directed vesicle mobility exists at least near the plasma membrane in ribbon synapses (Holt et al., 2004; Zenisek et al., 2000).

5

Indeed, several recent studies using either SPT of vesicles labeled with lipophilic dyes (Peng et al., 2012) or Q-dots (Lee et al., 2012; Park et al., 2012), as well as STED measurements (Kamin et al., 2010; Westphal et al., 2008) showed that a large proportion of recycling vesicles undergoes large-scale and complex movements within small central synapses (Fig. 2). SPT using non-scanning fluorescence microscopy (Peng et al., 2012) and taking advantage of the use of SGC5, a lipophilic dye similar in spectral properties to FM1–43 but several-fold brighter, was able to achieve much higher temporal resolution (10–125 Hz frame rate) and much increased sample size compared with the earlier SPT approach. The high frame rate, improved detection abilities and, most importantly, recordings at physiological temperatures in this study allowed to demonstrate that the majority of vesicles in hippocampal synapses exhibit complex motion patterns beyond caged diffusion, with trajectories often spanning several hundred of nanometers over the 30-sec observation window (Fig. 2A) (Peng et al., 2012). This high mobility was not due to vesicles traveling extrasynaptically from one bouton to another, a process that was observed in a subpopulation of labeled vesicles, but excluded from analysis in that study. Most interestingly, using transient motion analysis tools and pharmacology this study found that vesicle motion is, in part, directed and mediated by molecular motors (discussed in detail below). Analysis of vesicles undergoing various forms of endocytosis has also demonstrated that major differences in mobility exist between vesicles that undergo spontaneous versus activity-evoked endocytosis (Fig. 2A). Evoked vesicles moved on average twice as fast and spent twice as much time in directed motion compared with spontaneously recycling vesicles (Peng et al., 2012). Extensive vesicle mobility inside the central nerve terminals was also demonstrated independently in an SPT study that used Q-dots as labels for single synaptic vesicles, which also allowed use of real-time three-dimensional localization (Fig. 2B) (Park et al., 2012). This work, while focusing predominately on modes of vesicle fusion, showed that vesicles in hippocampal terminals undergo extensive and complex motion over several hundreds of nanometers towards the fusion sites. Movements in the range of 0.1 to 1 lm were the most common motion pattern exhibited by 70% of vesicles within the nerve terminals, with a small minority of vesicles also exiting the presynaptic boutons and traveling between the neighboring synapses. Only 27% of vesicles within the presynaptic boutons were found to have low, restricted mobility (spatial motion domain

A nanoscale resolution view on synaptic vesicle dynamics.

The ability of synapses to sustain neurotransmitter release during continuous activity critically relies on an efficient vesicle recycling program. De...
479KB Sizes 0 Downloads 12 Views