Microscopy 62(1): 5162 (2013) doi: 10.

1093/jmicro/dfs083
........................................................................................................................................................................................................................................................

Review

Fluorescence imaging of synapse formation and remodeling

Shigeo Okabe*
Department of Cellular Neurobiology, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo Bunkyo-ku, Tokyo 113-0033, Japan *To whom correspondence should be addressed. E-mail: okabe@m.u-tokyo.ac.jp
..............................................................................................................................................................................................

Abstract

Brain function is based on proper connectivity between neuronal cells. In the developing brain, neurons extend axons and form synaptic connections with appropriate postsynaptic neurons. Molecular mechanisms underlying establishment of proper synaptic connections are one of the most important topics in the eld of developmental neurobiology. Dynamics of synaptic structure and local recruitment of synaptic molecules can be studied by live-cell imaging of neurons expressing uorescent probes of synaptic molecules. In this review, examples of live-cell uorescence imaging are presented and their contributions to our understanding about the molecular mechanisms of synapse formation and remodeling are discussed. Imaging of synaptic proteins in living neurons revealed rapid formation of individual synapses within hours and extensive remodeling of synaptic connections. Different types of neurons express unique protrusions from dendrites and axons, which play important roles in synapse formation and maturation. Rapid formation of synaptic structure is associated with continual assembly and disassembly of synaptic scaffolding proteins, which are essential building blocks of the presynaptic active zone and the postsynaptic density (PSD). Quantitative analyses of PSD scaffolding proteins further conrmed their essential roles in maintenance of the synaptic structure. These examples clearly indicate that uorescence-based live-cell imaging is an indispensable technique in the research on synapse development and its impact will further increase in combination with development of new light microscopic techniques in the future. live-cell imaging, uorescent proteins, synapse, postsynaptic density, glutamate receptors, two-photon microscopy, super-resolution microscopy 26 September 2012, accepted 9 November 2012; online 14 December 2012

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

..............................................................................................................................................................................................

Keywords Received

..............................................................................................................................................................................................

..............................................................................................................................................................................................

Introduction
In the central nervous system (CNS) of vertebrates, neuronal precursors are differentiated from the neuroepithelial cells, migrate within the nervous tissue and settle within specic cortical layers or nuclei. After migration, neurons extend long axonal processes and start to connect with their target neurons. The contact sites between presynaptic axons and postsynaptic dendrites start to differentiate into synapses, which are the structures important in information processing and memory storage.

Electrophysiological and structural properties of mature synapses were studied extensively in the late 20th century. However, appropriate techniques for the reliable detection and measurement of nascent synapses during development had not been available until recently. Therefore, researchers initially focused on several types of model synapses in the peripheral nervous system, such as the neuromuscular junctions (NMJs), which were larger than synapses in the CNS and easier to manipulate experimentally [1]. Nascent NMJs could be identied

........................................................................................................................................................................................................................................................

The Author 2012. Published by Oxford University Press [on behalf of The Japanese Society of Microscopy]. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

52

M I C R O S C O P Y , Vol. 62, No. 1, 2013

using specic uorescent probes, and the mechanisms of synaptic competition between multiple presynaptic axons targeted to the same postsynaptic muscle could be analyzed in detail. In the CNS, however, several factors prevented direct detection and analyses of developing synapses. First, synaptic structures in the CNS are much smaller than NMJs and the resolution of light microscopy is not sufcient for the detection of their detailed morphology. Second, the mammalian brain is covered by the cranium and direct access to the brain parenchyma requires invasive surgery. Because the brain tissue is fragile and easy to develop edema, highly sophisticated surgical procedures should be invented and applied to small animals. To circumvent this difculty, isolated preparations of the nervous system, such as dissociated neuronal cultures and slice cultures, have been developed and widely used for the analyses of synapse development within the CNS [2,3]. Imaging experiments in these reduced preparations in combination with the development of green uorescent protein (GFP)-based uorescent probes provided essential information about the behavior of nascent synapses and the time course of their differentiation. Another possible approach toward the understanding of synapse development in the CNS is the histological analyses of brain sections taken from different stages of brain development. Analyses of synapse development in vivo by electron microscopy revealed structural details of nascent synapses quantitatively [4]. Fluorescence imaging of living neurons and ultrastructural studies of the xed brain tissue provide us with complementary information and their integration has facilitated our understanding about CNS synapse development.

Structural components of synapses and molecules involved in the assembly of presynaptic and postsynaptic specializations
Detailed morphological characteristics of synapses have been studied by using transmission electron microscopy of chemically xed or quickly frozen samples of the nervous tissue (Fig. 1). Synapses are the sites of membrane adhesion between presynaptic axons and postsynaptic dendrites or cell bodies.

Axonal boutons are the cytoplasmic swellings formed at the contact sites between the axonal plasma membrane and the target neurons along the course of axonal trajectories. Axonal boutons contain synaptic vesicles, which are the storage sites of neurotransmitters, such as glutamate and -Aminobutyric acid (GABA). Arrival of action potentials to the presynaptic terminal activates voltage-gated calcium channels and triggers exocytosis of synaptic vesicles. Neurotransmitters released from the presynaptic membrane diffuse across the synaptic cleft and activate neurotransmitter receptors on the postsynaptic membrane. Most excitatory synapses in the forebrain pyramidal neurons are formed onto dendritic spines, small protrusions containing densely packed actin laments [5]. Another morphological feature of excitatory synapses is the presence of the postsynaptic density (PSD), located at the plasma membrane of the dendritic spine and apposed to the presynaptic active zone [6]. The PSD is composed of a variety of proteins, including membrane proteins, such as glutamate receptors and cell adhesion molecules, together with PSD scaffolding proteins and signaling molecules. In addition to the PSD, the postsynaptic cytoplasm contains several other microstructures, such as spine apparatus, endosomal membranes and mitochondria. The sizes of both presynaptic axonal boutons and dendritic spines are in the order of several micrometers. The PSDs are disk-like structures with diameters of 200500 nm and thicknesses of 3060 nm [7]. Synapse morphology and distribution of synaptic proteins are difcult to resolve without the aid of uorescence microscopy. Under optimized imaging conditions, structural distinction between different types of spines, such as thin, stubby and mushroom spines, can be reliably achieved and could be conrmed by retrospective electron microscopy [8,9]. In order to identify nascent synaptic structures, it is essential to develop techniques that enable expression of uorescent protein (FP)tagged synaptic molecules and subsequent detection of uorescence signals in living neurons. A number of FP-based probes for presynaptic and postsynaptic components have been developed and utilized for live-cell imaging. Formation and remodeling of presynaptic structures can be monitored

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

S. Okabe Fluorescence imaging of synapse formation

53

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

Fig. 1. Ultrastructure of the excitatory synapse. (a) Transmission electron micrograph of an excitatory synapse in the mouse hippocampus. Scale bar, 200 nm. (b) Major cytoplasmic components of the presynaptic and postsynaptic cytoplasm of the excitatory synapses.

using GFP-tagged synaptophysin or synaptobrevin2/ VAMP2 [10,11]. Both of these probes show selective accumulation in axonal boutons. Synaptophysin is one of the most abundant proteins of synaptic vesicles. Synaptophysin contains four transmembrane domains with cytoplasmic N- and C-termini. Synaptophysin tagged with GFP at its C-terminus can be properly recruited to the synaptic vesicles and is utilized as a reliable marker of local synaptic vesicle accumulation [11]. Synaptobrevin2/VAMP2 is also an integral synaptic vesicle protein with a single transmembrane domain and a cytoplasmic N-terminal domain. In the process of exocytosis, the assembly of a SNARE (soluble N-ethylmaleimidesensitive factor attachment protein [SNAP] receptor) complex from SNAP-25, syntaxin and synaptobrevin2/VAMP2 is proposed to drive the formation of fusion pore between vesicles and the plasma membrane. Synaptobrevin2/VAMP2 tagged with GFP at its N-terminus is functional and does not perturb synaptic vesicle exocytosis. GFP-tagged synaptobrevin2/VAMP2 is localized at the sites of synaptic vesicle accumulation and can be utilized as a reliable presynaptic marker [10]. Detection of the postsynaptic specialization can be achieved using FP-tagged PSD molecules. 2-amino-3-(3-hydroxy-5-methyl-isoxazol-4-yl)propanoic

acid (AMPA)-type and N-methyl-D-aspartic acid (NMDA)-type glutamate receptors are enriched in the biochemically puried PSD fraction and their postsynaptic localization was conrmed by immunoelectron microscopy. AMPA receptors and NMDA receptors are essential components of glutamate-mediated synaptic transmission and their presence at the postsynaptic sites can be detected by expression of FP-tagged receptors. In the case of AMPA receptors, their distribution on dendritic membrane is diffuse and local density of AMPA receptors on the postsynaptic membrane is not extremely high compared with extrasynaptic receptors [12]. Furthermore, it is proposed that nascent synapses may lack AMPA receptors (silent synapses) and local activation of NMDA receptors triggers recruitment of AMPA receptors to silent synapses [13]. From these considerations, FP-tagged AMPA receptor subunits, such as GFP-GluA1 and GluA2, have not been utilized frequently to detect nascent synapses. However, these GFP-tagged AMPA receptors are useful in the detection of activity-dependent modications of postsynaptic functions [14]. Recruitment of NMDA receptors to the postsynaptic sites takes place in the early stage of synaptogenesis. NMDA receptors are tetramers composed of two GluN1 subunits and two GluN2

54

M I C R O S C O P Y , Vol. 62, No. 1, 2013

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

Fig. 2. Clustering of PSD scaffolding proteins tagged with GFP at the postsynaptic sites and their turnover in living hippocampal neurons in culture. (a) A cultured hippocampal neuron expressing PSD-95-GFP (green) stained with a lipophilic dye DiI (red). Arrows indicate colocalization of PSD-95-GFP puncta (green) and dendritic spines (red). Arrowheads indicate a single straight axon devoid of PSD-95 puncta. Scale bar, 10 m for the main panel and 4 m for the lower small images. (b) Formation and elimination of the PSD detected by expression of GFP-tagged Homer1c in cultured hippocampal neurons. Imaging of the same dendritic segments with an interval of 24 h for 6 days revealed continual remodeling of the PSDs, with gradual increase in the total number of GFP-Homer1c clusters. Scale bar, 4 m. Reprinted from Okabe et al. [19].

subunits. NMDA receptors tagged with GFP at their extracellular N-termini have been utilized for the detection of postsynaptic sites in cultured cortical neurons [15]. PSD scaffolding molecules are highly concentrated in the postsynaptic sites and are good candidates for postsynaptic markers. PSD-95 is one of the most abundant scaffolding molecules in the PSD [16]. PSD-95 can interact directly with NMDA receptors and also indirectly with AMPA receptors through its binding to transmembrane AMPA receptor regulatory proteins. PSD-95 interaction partners are not restricted to glutamate receptors, but include cell adhesion molecules, such as neuroligins and synCAMs, scaffolding molecules, such as guanylate kinase-associated protein (GKAP), and signaling molecules, such as synaptic GTPase-activating protein for Rac (SynGAP), neuronal nitric oxide synthase (nNOS) and spine-associated Rap-Gap (SPAR) [17]. Multiple binding partners of PSD-95 indicate its essential role in PSD organization. PSD-95 with its C-terminal tagged with GFP shows postsynaptic

distribution indistinguishable from that of endogenous PSD-95 and has been utilized as a reliable uorescent marker of the postsynaptic specialization (Fig. 2a) [18,19]. In addition to PSD-95, other GFP-tagged PSD scaffolding proteins, such as GFP-Homer and GFP-Shank, have also been utilized as probes for the PSD detection (Fig. 2b) [20,21]. The basic properties of these scaffolding proteins are similar, with quantitative difference in their local turnover rate and response to neuronal activity. Electron microscopic studies reported ultrastructural difference between asymmetrical (Gray type 1; correspond to excitatory glutamatergic synapses) and symmetrical (Gray type 2; correspond to inhibitory GABAergic synapses) synapses [22]. These structural differences are derived from the difference in thickness of the postsynaptic membrane specialization. Postsynaptic membranes of excitatory synapses develop thicker meshwork of protein assembly and are detected as prominent electrondense structures [6]. Typical PSD proteins, such as PSD-95, Homer and Shank, are specically

S. Okabe Fluorescence imaging of synapse formation

55

accumulated at excitatory asymmetrical synapses and their GFP-tagged probes can be utilized as specic markers of this type of synapses. In turn, GFP-tagged scaffolding proteins showing specic localization at inhibitory postsynaptic sites, such as gephyrin-GFP, can be utilized for the detection of symmetrical synapses in living neurons [23,24]. By using GFP-tagged presynaptic and postsynaptic molecules, time-lapse imaging of living neurons revealed the process of synapse formation and timing of recruitment of synaptic molecules. These analyses revealed basic principles of synapse formation and mechanisms of molecular assembly at the synaptic junctions [11,2527].

preparations and could quantitatively explain the overall rate of synapse density (Fig. 2b) [18,19]. It is possible to determine the time course of molecular assembly at presynaptic and postsynaptic sites simultaneously by using multicolor imaging of FP-tagged synaptic proteins. We expressed PSD-95 tagged with yellow uorescent protein (YFP) together with synaptophysin tagged with cyan uorescent protein (CFP) to determine whether presynaptic and postsynaptic molecules appear at synaptic sites with different time courses [11]. We found a strong temporal correlation between clustering of PSD-95-YFP and that of synaptophysin-CFP at the synaptic contact sites (Fig. 3d and e).
Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

Live-cell uorescence imaging of synaptic molecules and structure

Before GFP technology enabled us to visualize dynamics of synaptic molecules in living neurons, discussions on the time-course of synapse development had been based on the comparison of immunocytochemical and electron microscopic data of xed preparations. Because the density of synapses and dendritic spines increases gradually in both culture preparations and in vivo, it was widely accepted that the differentiation of individual synapses is also a slow process which may take a few weeks. However, time-lapse imaging of PSD-95 tagged with GFP in hippocampal pyramidal neurons in culture revealed rapid establishment of PSD-95-GFP uorescent clusters on time scales of several hours [19]. These PSD-95-GFP clusters were apposed to the presynaptic boutons and were selectively localized within dendritic spines, suggesting their identity as synaptic PSD structures [11]. A similar time course of postsynaptic molecular assembly was conrmed by several independent groups and has been taken as convincing evidence for rapid establishment of synaptic specialization [28,29]. If synapse assembly is a rapid process, how is relatively slow increase of overall synaptic density achieved? A possible explanation is that synapse addition and elimination take place simultaneously and the rate of addition is maintained to be moderately higher than the rate of elimination. Indeed, this difference in addition and elimination of PSD-95-GFP clusters existed in culture

Fig. 3. Simultaneous detection of multiple synaptic components by dual-color time-lapse imaging. (ac) Imaging of the PSDs by expression of PSD-95-yellow uorescent protein (YFP) (b) together with detection of spine structure by cyan uorescent protein (CFP) as a volume marker. (a) Overlay of PSD-95-YFP (red) and CFP (green) was also presented. (c) Formation of a PSD-95-YFP cluster (arrows) is coordinated with the enlargement of the spine head. Scale bar, 3 m. (de) Simultaneous imaging of synaptophysin-CFP (d) and PSD-95-YFP. (e) Accumulation of synaptic vesicles detected by synaptophysin-CFP is synchronized with clustering of PSD-95-YFP (arrows). Scale bar, 3 m. Reprinted from Okabe et al. [11,21].

56

M I C R O S C O P Y , Vol. 62, No. 1, 2013

Interestingly, appearance of synaptophysin-CFP clusters tends to precede clustering of postsynaptic PSD-95-YFP with time intervals of <1 h, indicating the existence of temporal orders in synapse differentiation. Immature hippocampal pyramidal neurons express dendritic lopodia, which are highly motile and have lifetimes shorter than those of spines. It has been proposed that dendritic lopodia serve as sensors of the surrounding tissue environment and are transformed into spines after physical interaction with nearby axons. To determine structural changes of immature dendritic protrusions and their relationship to clustering of PSD-95, we expressed both PSD-95-YFP and CFP in dissociated hippocampal neurons [11]. CFP was utilized as a volume label for morphological identication of dendritic protrusions. Imaging of PSD-95-YFP and CFP revealed the presence of lopodia-like protrusions on pyramidal neuron dendrites and appearance of PSD-95-YFP clusters in a small subset of these protrusions (Fig. 3ac). After establishment of PSD-95-YFP clusters, these protrusions were stabilized and increased their volume, indicating their morphological transformation into spine-like structures. Importantly, none of the PSD-95-YFP clusters within dendritic shafts induced the formation of dendritic protrusions. These observations indicate that lopodia-like protrusions from dendritic shafts play an important role in initial contact with nearby axons and subsequent differentiation of spines with synaptic contacts.

Roles of dendritic and axonal protrusions in synaptogenesis

Neuronal networks in the mammalian cortex are composed of two types of neurons, excitatory pyramidal-shaped neurons and inhibitory neurons. Although glutamatergic excitatory synapses are formed on dendrites of both pyramidal neurons and interneurons, their postsynaptic morphology and molecular composition are distinct. The most obvious difference is the absence of dendritic spines in excitatory synapses on interneuron dendrites [30,31]. Dendritic lopodia from pyramidal neurons are thought to be the precursors of spines and important in searching nearby axons. It has not yet been claried whether interneuron dendrites

have any searching systems to contact nearby axons. Without such a mechanism to enhance the chance of contacting nearby axons, the ability of interneurons to increase synaptic density should be quite limited. However, previous electron microscopic studies reported that dendritic shafts of mature interneurons are densely covered with glutamatergic synapses, suggesting the presence of interneuron-specic strategy to increase synaptic contacts [32]. To solve this problem, time-lapse imaging of synapse formation on interneuron dendrites was performed and two important observations were made [33]. First, dendritic protrusive activity of interneurons was developmentally regulated. Although dendrites of mature interneurons had few protrusions, immature interneurons expressed numerous dendritic protrusions, which were longer than typical lopodia of pyramidal neurons. Second, PSD-95 clusters were frequently observed on these dendritic protrusions and showed slow translocation toward the base of protrusions (Fig. 4). These observations indicated that dendritic protrusions of interneurons serve as conduits for retrograde translocation of synaptic structure to the parental dendrites. This translocation was dependent on microtubules present within dendritic protrusions and was driven by dynein motor system. These experimental data indicate that the behavior of synaptic structure may differ between different types of synapses even within the same brain area. In the case of pyramidal neurons, the positions of individual synapses along dendritic shafts are precisely determined by the position of initial protrusive activity of dendritic lopodia. On the other hand, the positions of excitatory synapses along interneuron dendrites may be more exible and the synaptic connectivity can be increased effectively by using long protrusions as synaptic conduits. Assembly of synaptic structure is a stochastic process and the behavior of a large number of synapses with different stages of maturation is difcult to be classied and characterized. If synapse maturation can be synchronized, the time course of synapse development may be determined more precisely and molecular markers and structural features for each developmental stage may be characterized more easily. In the cerebellum,

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

S. Okabe Fluorescence imaging of synapse formation

57

remodeling and subsequently generated fewer axonal boutons. However, application of Cbln1 increased both motility of synapses and their subsequent transformation into axonal boutons. On the basis of the data obtained by using this articial synchronization system of synaptogenesis, we proposed bidirectional interaction model of synapse maturation between parallel bers and Purkinje cell dendrites. This model is based on our observation that a specic type of axonal protrusions with complex morphology appears during the intermediate stage of synapse formation and increase in the contact area between axons and dendritic spines enhances both Nrx-dependent presynaptic maturation and GluD2-dependent spine maturation. We learned from this study that time-lapse imaging of the synchronized formation of neural circuits is highly informative in dening precise stages of synaptogenesis.

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

Turnover of postsynaptic scaffolding molecules and their structural roles

We proposed that synapse addition and elimination take place simultaneously and the slight excess of added synapse over eliminated synapse determines the gradual increase of overall synapse density. This model indicates the importance of a regulatory mechanism that controls the rates of synapse formation and elimination through the assembly and disassembly of synaptic molecules. To determine the turnover rate of a given synaptic molecule at local synaptic sites precisely, uorescence recovery after photobleaching (FRAP) and uorescence activation were performed. With both techniques, the turnover rate of uorescent molecules in a local environment can be directly measured. Interestingly, FRAP analyses of four different PSD scaffolding molecules revealed distinct kinetic properties (Fig. 5) [20]. PSD-95, a scaffolding molecule directly interacting with the plasma membrane via its S-palmitoylation, showed low mobility by FRAP analysis. In contrast, Homer1c, a scaffolding molecule containing binding sites for actin, showed a much higher turnover rate. Because actin molecules in spines show very rapid exchange between soluble and polymerized pools [35], it is likely that dynamics of PSD scaffolding proteins is heterogeneous and scaffolding proteins

Fig. 4. PDS mobility along dendritic protrusions of interneurons. (a) Time-lapse images of an interneuron dendrite expressing PSD-95-GFP (green) and RFP (magenta) in a dissociated hippocampal culture. A PSD-95-GFP cluster (arrows) shows retrograde translocation. Scale bar, 3 m. (b) Time-lapse images of an interneuron expressing GFP-Homer1c and RFP in a cortical slice culture. A GFP-Homer1c cluster showed movement in a distal-to-proximal direction (arrows). Scale bar, 5 m. Reprinted from Kawabata et al. [33].

synaptogenesis between the axons of granule cells ( parallel bers) and the dendrites of Purkinje cells is regulated by a soluble factor Cbln1, a C1q family protein produced and secreted from granule cells. Cbln1 released from parallel bers binds to both presynaptic receptor neurexin (Nrx) and postsynaptic receptor glutamate receptor delta 2 (GluD2). It is likely that the tripartite Nrx-Cbln1-GluD2 complex affects synaptogenesis of both presynaptic and postsynaptic components. We took advantage of slice cultures derived from Cbln1 knockout (KO) mice and induced synaptogenesis in slices by exogenous application of Cbln1 [34]. Without Cbln1, parallel bers showed little morphological

58

M I C R O S C O P Y , Vol. 62, No. 1, 2013

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

Fig. 5. Differential stability of PSD scaffolding proteins revealed by FRAP. (a) Organization of the PSD by layered distribution of multiple types of PSD scaffolding proteins. (b) Differential stabilization of four PSD scaffolding proteins measured by FRAP. Fluorescence recovery within 5 min was measured and quantitated. Scale bar, 1 m. Reprinted from Kuriu et al. [20].

enriched at the cytoplasmic surface of PSDs are dependent on actin cytoskeleton for their localization and stability. To further clarify the roles of PSD scaffolding proteins in the molecular architecture of PSDs, information about the molecular content of individual scaffolding proteins per single PSD is essential. Even if functional studies suggest importance of a given molecule by gene knockout and knockdown approaches, its importance as a structural component can be proved only by measuring its molecular content per synapses. To this end, we developed a technique of quantitative measurement of protein content per synapses by using uorescent microspheres as a calibration standard of GFP molecules

Fig. 6. Determination of absolute numbers of PSD scaffolding proteins in single synapses. (a) A confocal microscopic image of a dendritic segment expressing PSD-95-GFP. The neuron was chemically xed, reacted with anti-VGLUT1 antibody (red) and anti-MAP2 (blue) and sprinkled with uorescent microspheres (arrow). The uorescence signal from PSD-95-GFP clusters (arrowheads) was compared with the signal from uorescent microspheres. Scale bar, 2 m. (b) Calculation of the endogenous PSD-95 protein content by comparison of the immunouorescence intensity of control neurons and neurons infected with recombinant adenoviruses for the expression of PSD-95-GFP. After determination of the number of functional GFP fusions in a single postsynaptic site (363 molecules), the number of endogenous PSD-95 molecules per synapse was calculated to be 308, based on the ratio of total PSD-95 content between control and infected neurons (1:2.47). (c) Cumulative distribution of scaffolding protein numbers in single postsynaptic sites of hippocampal neurons on day 25 in culture. The synaptic contents of PSD-95 and total membrane-associated guanylate kinase proteins were 208 86.6 and 308 128 (mean standard deviation). Reprinted from Sugiyama et al. [36].

(Fig. 6) [36]. We rst determined the uorescence intensity of single GFP molecules and compared the uorescence of single microspheres with that of single GFP proteins. Our measurements indicated that the uorescence microspheres are equivalent to the uorescence intensity of about 4000 GFP proteins. By using these calibrated microspheres,

S. Okabe Fluorescence imaging of synapse formation

59

we next determined the number of GFP-tagged PSD scaffolding proteins present in single postsynaptic sites. We could successfully estimate the absolute numbers of four key scaffolding proteins, PSD-95, GKAP, Shank and Homer, per synapses. Interestingly, the synaptic contents of the four scaffolding proteins were quite similar and were in the range 100450 molecules per synapses. Because these four scaffolding molecules were known to interact with each other, we proposed that their binding stoichiometry is relatively simple. Furthermore, the total protein mass of four scaffolding proteins was calculated to be 100 MDa, which corresponded to about 10% of the total mass of single PSD (1.1 GDa). This suggests that only four types of PSD scaffolds occupy a substantial fraction of the total PSD mass. This conclusion is consistent with the idea that a small subset of PSD scaffolding proteins are responsible for the construction of the main PSD structure, which may function for anchoring and stabilization of other functional molecules, such as neurotransmitter receptors and cell adhesion molecules [16]. Several hundreds of PSD-95 molecules in a single synapse are sufcient to anchor glutamate receptors, which are estimated to be accumulated to single synapses in the range 10200 tetramers per synapse [37,38]. It is important to note that the number of PSD-95 per synapse and the number of glutamate receptor complexes per synapse are set to be in a similar range. This relationship may enable efcient regulation of glutamate receptor numbers through the synaptic content of PSD-95. This idea was supported by electrophysiological recordings of slices from PSD-95 knockout mice. In the absence of PSD-95 and related membraneassociated guanylate kinase proteins, synaptic transmission, which was mediated by AMPA-type glutamate receptors, could be detected, but its amplitude was signicantly down-regulated, suggesting PSD-95-dependent regulation of AMPA receptor content per synapse [39].

applications of new optical principles, light sources and sample preparations techniques to the analyses of synapse formation and remodeling. Especially, two important technologies, in vivo two-photon imaging and super-resolution microscopy, may greatly accelerate our understanding of synapse dynamics. Two-photon excitation laser scanning microscopy is a technique appropriate for high-resolution imaging of neuronal microstructure in vivo [40]. An infrared pulsed laser beam can penetrate the brain tissue more effectively than the shorter wavelength light and can be focused to the volume small enough to resolve submicron structures within the nervous tissue [41,42]. By using two-photon microscopy, activation of uorescence molecules is restricted to a small volume without unnecessary excitation above and below the focal plane, which may induce phototoxic effects after repeated imaging of a large volume of samples. Two-photon microscopy enabled researchers to resolve dendritic spines lled with GFP in the neocortex of living mice with time intervals of several days to months (Fig. 7) [43,44]. By comparing the images of dendritic segments and attached spines at different time points, the rate of formation and elimination of spines could be successfully measured in vivo. Initial studies revealed the highly stable nature of dendritic spines in the mature neocortex: more

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

Future directions of imaging research on synapse dynamics

Recent advancement of imaging technologies provided a wide range of opportunities for the

Fig. 7. In vivo imaging of layer 5 pyramidal neurons expressing GFP in the somatosensory cortex. (a) Two-photon imaging of the apical dendrites in the layer 1 of an adult transgenic mouse expressing GFP under the control of Thy1 promoter. (b) Dendritic spines visualized in vivo with an interval of 14 days. Arrow indicates an eliminated spine. Most of the imaged spines were stable over 14 days in vivo. Scale bar, 5 m.

60

M I C R O S C O P Y , Vol. 62, No. 1, 2013

than 90% of spines imaged in the adult mouse neocortex could survive when they were re-examined several months later [43,45]. This stability is in contrast with rapid turnover of synapses in culture conditions. There are two possibilities that can explain this discrepancy. First, synapse dynamics in culture conditions may be articially upregulated by the presence of factors specic to the culture condition or the absence of factors present in the native tissue environment. For example, culture medium and serum supplement may upregulate synapse elimination in vitro. It is also possible that mature astrocytes may protect newly formed synapses and the lack of glial support may explain the higher turnover rate of synapses in culture [41]. The second possibility is that rapid turnover of synapses in vitro reects the situation of synapses in the early phase of synapse development in vivo. Indeed, quantitative analyses of spine density in the early postnatal period of rat visual cortex revealed that signicant increase in spine density occurs between postnatal days 6 and 21 [46]. Application of in vivo two-photon imaging in the early postnatal period may clarify this point. However, this experiment has not yet been accomplished, mainly because of the technical difculty of imaging nascent synapses in vivo. At present, most of the in vivo synapse analyses rely on dendritic spines as a structural marker of synapses. In the future [47], application of GFP-tagged postsynaptic proteins, such as PSD-95-GFP, will increase the precision of identifying nascent synapses in vivo and will reveal the degree of synapse turnover in the early postnatal period. Another important advancement in the eld of synapse imaging is application of super-resolution imaging to molecular distribution and dynamics of synaptic proteins [48]. Recently, two papers reported signicant improvement in spatial resolution of postsynaptic molecular movement and PSD structure by using photoactivated localization microscopy (PALM) and stochastic optical reconstruction microscopy (STORM) techniques [49,50]. PALM is the imaging technique based on photoactivation of FPs and STORM is based on stochastic activation of organic uorescent dyes attached to appropriate antibodies. Recent application of PALM to the behavior of single actin molecules revealed

the presence of actin ow in spines and conrmed the results of previous photoactivation studies of actin tagged with photoactivatable GFP [50]. Furthermore, these PALM studies revealed heterogeneity in the speed of actin movement within single spines. In the cytoplasm close to the PSD, actin velocity was specically higher than in the other area. These PALM experiments of actin mobility in spines suggested that overall organization of actin laments in spines is based on short laments with less aligned orientation and heterogeneous velocity. It is also possible to evaluate molecular distribution of multiple synaptic proteins at the synaptic junctions by using super-resolution imaging techniques. Dani et al. reported distribution of PSD scaffolding proteins and glutamate receptors within the postsynaptic membrane of xed brain sections labeled with antibodies against presynaptic and postsynaptic molecules [49]. They could successfully detect differential distribution of AMPA and NMDA receptors in individual synapses by using the STORM technique. In the future, precise organization of glutamate receptors and scaffolding proteins can be determined by using the PALM/STORM techniques and their intrasynaptic redistribution associated with synaptic plasticity may be directly visualized.

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

Conclusions
Before the introduction of live imaging techniques of synaptic structure and function, electrophysiological recording was the only possible technique to analyze functional modication of synapses. Although electrophysiological recording is an indispensable technique to record activity of synaptic receptors and ion channels, it usually lacks the information of single synapses and is not suitable for long-term monitoring of synapse behaviors. Introduction of GFP technology and its application to the research on synapse formation led to direct monitoring of single synapse behavior for a long time with minimal perturbation. These imaging studies in the last decade provided a clear picture of synapse formation in the mammalian neocortex and hippocampus. Two-photon excitation microscopy, combined with various gene expression technologies, such as generation of transgenic mice and

S. Okabe Fluorescence imaging of synapse formation

61

.......................................................................................................................

infection of viral vectors, further extended the possibility of imaging synapse formation and remodeling within tissue environment. Especially, two-photon imaging of neocortical neurons through optical windows on the rodent cranium enabled researchers to monitor the lifetime and the behavior of single synapses in vivo. In the near future, using in vivo imaging technologies combined with techniques of manipulating neuronal activity, we can expect that the basic principles of neural connectivity in living animals can be extracted with rm experimental evidences.

.......................................................................................................................

11 Okabe S, Miwa A, and Okado H (2001) Spine formation and correlated assembly of presynaptic and postsynaptic molecules. J. Neurosci. 21: 61056114.

.......................................................................................................................

12 Passafaro M, Piech V, and Sheng M (2001) Subunit-specic temporal and spatial patterns of AMPA receptor exocytosis in hippocampal neurons. Nat. Neurosci. 4: 917926.

.......................................................................................................................

13 Nusser Z, Lujan R, Laube G, Roberts J D, Molnar E, and Somogyi P (1998) Cell type and pathway dependence of synaptic AMPA receptor number and variability in the hippocampus. Neuron 21: 545559.

.......................................................................................................................

14 Shi S H, Hayashi Y, Petralia R S, Zaman S H, Wenthold R J, Svoboda K, and Malinow R (1999) Rapid spine delivery and redistribution of AMPA receptors after synaptic NMDA receptor activation. Science 284: 18111816.

.......................................................................................................................

15 Washbourne P, Bennett J E, and McAllister A K (2002) Rapid recruitment of NMDA receptor transport packets to nascent synapses. Nat. Neurosci. 5: 751759. 16 Okabe S (2007) Molecular anatomy of the postsynaptic density. Mol. Cell Neurosci. 34: 503518.

Funding
This work was supported by Grants-in-Aid for Scientic Research (18200025, 20019013, 21220008, and 22650070 to S.O.) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. A part of this study was the result of Development of biomarker candidates for social behavior carried out under the Strategic Research Program for Brain Sciences by the Ministry of Education, Culture, Sports, Science and Technology of Japan (S.O.), and also by Takeda Science Foundation (S.O.).

.......................................................................................................................

.......................................................................................................................

17 Kim E and Sheng M (2004) PDZ domain proteins of synapses. Nat. Rev. Neurosci. 5: 771781. 18 Ebihara T, Kawabata I, Usui S, Sobue K, and Okabe S (2003) Synchronized formation and remodeling of postsynaptic densities: long-term visualization of hippocampal neurons expressing postsynaptic density proteins tagged with green uorescent protein. J. Neurosci. 23: 21702181. 19 Okabe S, Kim H D, Miwa A, Kuriu T, and Okado H (1999) Continual remodeling of postsynaptic density and its regulation by synaptic activity. Nat. Neurosci. 2: 804811.

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

.......................................................................................................................

.......................................................................................................................

.......................................................................................................................

20 Kuriu T, Inoue A, Bito H, Sobue K, and Okabe S (2006) Differential control of postsynaptic density scaffolds via actin-dependent and -independent mechanisms. J. Neurosci. 26: 76937706. 21 Okabe S, Urushido T, Konno D, Okado H, and Sobue K (2001) Rapid redistribution of the postsynaptic density protein PSD-Zip45 (Homer 1c) and its differential regulation by NMDA receptors and calcium channels. J. Neurosci. 21: 95619571.

References
....................................................................................................................... .......................................................................................................................

.......................................................................................................................

1 Sanes J R and Lichtman J W (1999) Development of the vertebrate neuromuscular junction. Annu. Rev. Neurosci. 22: 389442.

22 Peters A, Palay S L, and Webster H deF (1991) The Fine Structure of the Nervous System, (Oxford University Press, New York, NY).
.......................................................................................................................

.......................................................................................................................

2 Banker G A and Cowan W M (1977) Rat hippocampal neurons in dispersed cell culture. Brain Res. 126: 342. 397.

.......................................................................................................................

3 Stoppini L, Buchs P A, and Muller D (1991) A simple method for organotypic cultures of nervous tissue. J. Neurosci. Methods 37: 173182.

.......................................................................................................................

23 Dobie F A and Craig A M (2011) Inhibitory synapse dynamics: coordinated presynaptic and postsynaptic mobility and the major contribution of recycled vesicles to new synapse formation. J. Neurosci. 31: 1048110493.

.......................................................................................................................

4 Fiala J C, Feinberg M, Popov V, and Harris K M (1998) Synaptogenesis via dendritic lopodia in developing hippocampal area CA1. J. Neurosci. 18: 89008911. 5 Yuste R (2010) Dendritic Spines, (The MIT Press, Cambridge, MA).

.......................................................................................................................

24 Meier J and Grantyn R (2004) A gephyrin-related mechanism restraining glycine receptor anchoring at GABAergic synapses. J. Neurosci. 24: 13981405.

.......................................................................................................................

.......................................................................................................................

6 Ziff E B (1997) Enlightening the postsynaptic density. Neuron 19: 11631174.

.......................................................................................................................

25 Fujii R, Okabe S, Urushido T, Inoue K, Yoshimura A, Tachibana T, Nishikawa T, Hicks G G, and Takumi T (2005) The RNA binding protein TLS is translocated to dendritic spines by mGluR5 activation and regulates spine morphology. Curr. Biol. 15: 587593.

.......................................................................................................................

7 Petersen J D, Chen X, Vinade L, Dosemeci A, Lisman J E, and Reese T S (2003) Distribution of postsynaptic density (PSD)-95 and Ca2 +/calmodulin-dependent protein kinase II at the PSD. J. Neurosci. 23: 1127011278.

.......................................................................................................................

26 Iki J, Inoue A, Bito H, and Okabe S (2005) Bi-directional regulation of postsynaptic cortactin distribution by BDNF and NMDA receptor activity. Eur. J. Neurosci. 22: 29852994. 27 Yamagata Y, Kobayashi S, Umeda T, Inoue A, Sakagami H, Fukaya M, Watanabe M, Hatanaka N, Totsuka M, Yagi T, Obata K, Imoto K, Yanagawa Y, Manabe T, and Okabe S (2009) Kinase-dead knock-in mouse reveals an essential role of kinase activity of Ca2 +/calmodulin-dependent protein kinase IIalpha in dendritic spine enlargement, long-term potentiation, and learning. J. Neurosci. 29: 76077618. 28 Bresler T, Shapira M, Boeckers T, Dresbach T, Futter M, Garner C C, Rosenblum K, Gundelnger E D, and Ziv N E (2004) Postsynaptic density assembly is fundamentally different from presynaptic active zone assembly. J. Neurosci. 24: 15071520.

.......................................................................................................................

8 Nagerl U V, Kostinger G, Anderson J C, Martin K A, and Bonhoeffer T (2007) Protracted synaptogenesis after activity-dependent spinogenesis in hippocampal neurons. J. Neurosci. 27: 81498156.

.......................................................................................................................

9 Zito K, Scheuss V, Knott G, Hill T, and Svoboda K (2009) Rapid functional maturation of nascent dendritic spines. Neuron 61: 247258.

.......................................................................................................................

10 Ahmari S E, Buchanan J, and Smith S J (2000) Assembly of presynaptic active zones from cytoplasmic transport packets. Nat. Neurosci. 3: 445451.

62

M I C R O S C O P Y , Vol. 62, No. 1, 2013

AMPA receptors by a family of MAGUK scaffolding proteins. Neuron 52: 307320.

.......................................................................................................................

29 Friedman H V, Bresler T, Garner C C, and Ziv N E (2000) Assembly of new individual excitatory synapses: time course and temporal order of synaptic molecule recruitment. Neuron 27: 5769.

.......................................................................................................................

.......................................................................................................................

.......................................................................................................................

30 Freund T F and Buzsaki G (1996) Interneurons of the hippocampus. Hippocampus 6: 347470.

.......................................................................................................................

40 Helmchen F and Denk W (2005) Deep tissue two-photon microscopy. Nat. Methods 2: 932940. 41 Nishida H and Okabe S (2007) Direct astrocytic contacts regulate local maturation of dendritic spines. J. Neurosci. 27: 331340. 42 Umeda T, Ebihara T, and Okabe S (2005) Simultaneous observation of stably associated presynaptic varicosities and postsynaptic spines: morphological alterations of CA3-CA1 synapses in hippocampal slice cultures. Mol. Cell Neurosci. 28: 264274.

.......................................................................................................................

31 Markram H, Toledo-Rodriguez M, Wang Y, Gupta A, Silberberg G, and Wu C (2004) Interneurons of the neocortical inhibitory system. Nat. Rev. Neurosci. 5: 793807. 32 Gulyas A I, Megias M, Emri Z, and Freund T F (1999) Total number and ratio of excitatory and inhibitory synapses converging onto single interneurons of different types in the CA1 area of the rat hippocampus. J. Neurosci. 19: 1008210097.

.......................................................................................................................

.......................................................................................................................

.......................................................................................................................

33 Kawabata I, Kashiwagi Y, Obashi K, Ohkura M, Nakai J, Wynshaw-Boris A, Yanagawa Y, and Okabe S (2012) LIS1-dependent retrograde translocation of excitatory synapses in developing interneuron dendrites. Nat. Commun. 3: 722.
.......................................................................................................................

.......................................................................................................................

43 Grutzendler J, Kasthuri N, and Gan W B (2002) Long-term dendritic spine stability in the adult cortex. Nature 420: 812816.

34 Ito-Ishida A, Miyazaki T, Miura E, Matsuda K, Watanabe M, Yuzaki M, and Okabe S (2012) Presynaptically released Cbln1 induces dynamic axonal structural changes by interacting with GluD2 during cerebellar synapse formation. Neuron 76: 549564.

.......................................................................................................................

44 Trachtenberg J T, Chen B E, Knott G W, Feng G, Sanes J R, Welker E, and Svoboda K (2002) Long-term in vivo imaging of experience-dependent synaptic plasticity in adult cortex. Nature 420: 788794. 45 Kondo S, Kohsaka S, and Okabe S (2011) Long-term changes of spine dynamics and microglia after transient peripheral immune response triggered by LPS in vivo. Mol. Brain 4: 27.

.......................................................................................................................

.......................................................................................................................

35 Star E N, Kwiatkowski D J, and Murthy V N (2002) Rapid turnover of actin in dendritic spines and its regulation by activity. Nat. Neurosci. 5: 239246.
.......................................................................................................................

.......................................................................................................................

46 Miller M (1981) Maturation of rat visual cortex. I. A quantitative study of Golgi-impregnated pyramidal neurons. J. Neurocytol. 10: 859878.

Downloaded from http://jmicro.oxfordjournals.org/ by guest on March 24, 2013

36 Sugiyama Y, Kawabata I, Sobue K, and Okabe S (2005) Determination of absolute protein numbers in single synapses by a GFP-based calibration technique. Nat. Methods 2: 677684.
.......................................................................................................................

47 Gray N W, Weimer R M, Bureau I, and Svoboda K (2006) Rapid redistribution of synaptic PSD-95 in the neocortex in vivo. PLoS Biol. 4: e370.
.......................................................................................................................

37 Nusser Z (1999) A new approach to estimate the number, density and variability of receptors at central synapses. Eur. J. Neurosci. 11: 745752.

.......................................................................................................................

48 Huang B, Bates M, and Zhuang X (2009) Super-resolution uorescence microscopy. Annu. Rev. Biochem. 78: 9931016.

.......................................................................................................................

38 Tanaka J, Matsuzaki M, Tarusawa E, Momiyama A, Molnar E, Kasai H, and Shigemoto R (2005) Number and density of AMPA receptors in single synapses in immature cerebellum. J. Neurosci. 25: 799807.
.......................................................................................................................

.......................................................................................................................

49 Dani A, Huang B, Bergan J, Dulac C, and Zhuang X (2010) Superresolution imaging of chemical synapses in the brain. Neuron 68: 843856.

39 Elias G M, Funke L, Stein V, Grant S G, Bredt D S, and Nicoll R A (2006) Synapse-specic and developmentally regulated targeting of

50 Frost N A, Shroff H, Kong H, Betzig E, and Blanpied T A (2010) Single-molecule discrimination of discrete perisynaptic and distributed sites of actin lament assembly within dendritic spines. Neuron 67: 8699.