Presynaptic and Postsynaptic Scaffolds

NROXXX10.1177/1073858414523321The NeuroscientistZiv and Fisher-Lavie 523321research-article2014

Review

The Neuroscientist 2014, Vol. 20(5) 439–452 Presynaptic and Postsynaptic Scaffolds: © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav Dynamics Fast and Slow DOI: 10.1177/1073858414523321 nro.sagepub.com

Noam E. Ziv1 and Arava Fisher-Lavie1

Abstract The development of methods to follow the dynamics of synaptic molecules in living neurons has radically altered our view of the synapse, from that of a generally static structure to that of a dynamic molecular assembly at steady state. This view holds not only for relatively labile synaptic components, such as synaptic vesicles, cytoskeletal elements, and neurotransmitter receptors, but also for the numerous synaptic molecules known as scaffolding molecules, a generic name for a diverse class of molecules that organize synaptic function in time and space. Recent studies reveal that these molecules, which confer a degree of stability to synaptic assemblies over time scales of hours and days, are themselves subject to significant dynamics. Furthermore, these dynamics are probably not without effect; wherever studied, these seem to be associated with spontaneous changes in scaffold molecule content, synaptic size, and possibly synaptic function. This review describes the dynamics exhibited by synaptic scaffold molecules, their typical time scales, and the potential implications to our understanding of synaptic function.

Keywords synaptic tenacity, synaptic dynamics, synaptic scaffolds, FRAP, photoactivation

Chemical synapses are sites of cell-cell contact special- held in register by trans-synaptic cell adhesion molecules ized for the rapid transmission of signals between neu- and extracellular matrix proteins. rons and their targets: muscles, glands, or other neurons. Historically, much of what was known about CNS The vast majority of synapses in mammals are found in synapses and, in particular, about AZs, and PSDs was the central nervous system (CNS), where they typically based on electron microscopy (EM). However, an explo- connect the axon of one neuron to the dendrite or soma of sive amount of discoveries has been made during the last another neuron. Synaptic transmission is a directional 20 years on the molecular composition of CNS syn- process, and this directionality is manifested in the asym- apses—on the families, types, and subtypes of molecules metric structure of the presynaptic and postsynaptic com- found at excitatory and inhibitory synapses and their pri- partments. The axonal presynaptic compartment is mary roles: receptors, scaffolding molecules, proteins characterized by the presence of dozens to hundreds of involved in membrane trafficking, posttranslational mod- neurotransmitter-filled synaptic vesicles (SVs) and by ifications, protein synthesis and degradation, and many active zones (AZs), which are specialized regions of the more. It would be fair to say, however, that during the presynaptic axonal plasma membrane where SVs dock, initial years, a major concept carried over from the era of fuse, and release neurotransmitters into the synaptic cleft EM primacy was the notion that the synapse is a struc- (Fig. 1). The AZ is characterized by an electron-dense ture—a rather rigid and static molecular organization that matrix or lattice of proteins (the cytoskeleton of the AZ can undergo change when instructed to do so by physio- [CAZ]), which is thought to define the AZ as the site of logical signals but is otherwise quite stable and SV docking and fusion. The postsynaptic reception appa- unchanging. ratus is also characterized by an electron-dense thickening referred to as the postsynaptic density (PSD), the central function of which is to confine receptors of the 1Technion–Israel Institute of Technology, Haifa, Israel appropriate type in front of the AZ. At excitatory glutamatergic synapses, PSDs are typically found at the tip of Corresponding Author: small dendritic protrusions known as dendritic spines, Noam E. Ziv, Technion–Israel Institute of Technology, Faculty of Medicine and Network Biology Research Laboratories, Fishbach whereas PSDs of inhibitory synapses are typically located Building, Technion City, Haifa 32000, Israel on dendritic shafts or cell bodies. The PSD and CAZ are Email: [email protected]

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 440 The Neuroscientist 20(5)

Zhang 2009; Gundelfinger and Fejtova 2012; Kneussel and Loebrich 2007; Sheng and Hoogenraad 2007; Südhof 2012; Verpelli and others 2012). Instead, the review will discuss the dynamics of these molecules, the relationships of these dynamics with those of other molecules (such as neurotransmitter receptors), the relationships with other processes such as protein synthesis and degradation, and finally, the functional implications that these dynamics might have on synaptic function.

The Active Zone Presynaptic sites in the mammalian CNS typically appear as small varicosities (presynaptic boutons) distributed along axons (en passant synapses) in an irregular, near- random fashion (Hellwig and others 1994; Shepherd and others 2002 and references within) or, less commonly, at Figure 1. A schematic illustration of a mammalian central the tip of axonal branchlets. When observed at the EM nervous system, en passant glutamatergic synapse. The level, presynaptic boutons are found to contain clusters of presynaptic active zone, including molecules of the cytoskeleton SVs, opposed to the specialized region of the presynaptic 2+ of the active zone, voltage-gated Ca channels, and docked plasma membrane known as the AZ. In mammalian CNS vesicles, is shaded in gray. The active zone is juxtaposed and synapses, AZs are disc-like structures with diameters of connected to the postsynaptic density (PSD), located at the approximately 200 to 500 nm. The AZ is surrounded by a tip of a dendritic spine, via adhesion molecule pairs. Glutamate receptors are concentrated in the postsynaptic membrane via perisynaptic zone, which seems to be the major site of interactions with PSD scaffolding molecules. endocytosis, the essential regenerative upstroke of the SV cycle. Moreover, EM studies suggest that en passant presynaptic boutons lack obvious physical barriers that sepa- The advent of methods that allowed one to study the rate their contents from the cytoplasm and membrane of molecular dynamics of synaptic molecules, from large the axon proper (Shepherd and Harris 1998) or from those populations of synaptic molecules to single molecules of neighboring boutons. The lack of such barriers would within and outside of synapses, as well as long-term seem to challenge the ability of individual synapses to pre- imaging methodologies that allowed synapses to be fol- serve their complement of presynaptic molecules and SVs. lowed for many hours and days, has radically altered this Moreover, both spontaneous and, in particular, evoked notion (Choquet and Triller 2013). It is now clear that neurotransmitter release are associated with tremendous synapses are not structures in a strict sense but dynamic membrane trafficking, which involves SV mobilization, molecular assemblies at complex steady states, with syn- exocytosis, endocytosis, and SV reformation, suggesting aptic molecules moving in, out, and between synapses that activity further exacerbates this challenge. over a variety of time scales—from subseconds to The advent of new imaging techniques has allowed hours—and with synaptic compartments changing their these dynamics to be studied in living neurons and to eval- instantaneous contents and probably their fine functional uate how well this challenge is addressed. In this respect, characteristics on similar and longer time scales. several key techniques proved to be most valuable (Box 1). The main focus of this review is to describe the key Most of these are based on 1) the expression of fusion pro- concepts that have emerged over the last years concern- teins of specific synaptic molecules and green fluorescent ing the dynamic nature of synapses in the mammalian protein (GFP), one of its spectral variants or photoactivat- CNS. Specifically, the review will focus on synaptic able/photoswitchable derivatives; 2) confined photo- molecular complexes that are generally referred to as bleaching or photoactivation; and 3) time-lapse imaging synaptic scaffolds due to the roles that these are believed for monitoring changes in the fluorescence of these mole- to play in organizing presynaptic and postsynaptic orga- cules at individual synapses or changes in the distribution nization on a molecular scale. This generic role assign- of the fluorescently tagged molecules among nearby syn- ment clearly does not do justice to the specific properties apses (Kim and others 2010; Okabe 2012). and functions of these molecules; however, a comprehen- A large series of experiments using these and related sive review of their particular functions will not be pro- methods has revealed that under baseline conditions, vided here, and the reader is invited to refer to several SVs, traveling as individual vesicles or small packets, as recent reviews on these topics if so desired (Feng and well as SV-associated molecules, are continuously

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 Ziv and Fisher-Lavie 441

Box 1. Measuring the Dynamics of Synaptic Scaffold Molecules

(A) Fluorescence recovery after photobleaching. This illustration shows a dendritic spine synapse in a neuron expressing a fluorescently tagged postsynaptic scaffold molecule (green circles). Some of these molecules are incorporated into the postsynaptic density (PSD), whereas others are loosely attached to cytoskeletal elements (such as actin filaments) within the spine, and the remainder is in the cytosol. At t , fluorescent molecules within the synapse are photobleached (gray circles) by spatially con- 0 fined strong illumination. During the first few minutes (t ), the photobleached molecules loosely attached to cytoskeletal elements 1 dissociate from these cytoskeletal elements and diffuse away and are replaced by fluorescent molecules from the cytosol. This is manifested as a rapid recovery of fluorescence in fluorescence recovery plots. Over subsequent minutes and hours (t ), photo- 2 bleached molecules within the PSD dissociate from the PSD and are replaced with fluorescent molecules from the cytosol and from the pool of molecules loosely attached to cytoskeletal elements. This is manifested as the slow recovery phase in fluorescence recovery plots. (B) Photoactivation and redistribution. This illustration shows an axon with several boutons along its length (gray circles) from a neuron expressing a photoactivatable fluorophore-tagged presynaptic scaffold molecule. At t , brief illumination at the appropri- 0 ate wavelength photoactivates the tagged molecules within one bouton (green). With time, the photoactivated molecules migrate out of the photoactivated bouton and move into neighboring boutons. This is manifested as a reduction of fluorescence in the photoactivated bouton (B ) and transient increases in the fluorescence in neighboring boutons (B ). 1 2 (C) Measuring molecular content constancy. This illustration shows a dendritic segment from a neuron expressing a fluorescently tagged postsynaptic scaffold molecule, which becomes incorporated into PSDs. The fluorescence of each PSD is then measured repeatedly at consecutive time points (gray rectangular regions of interest), with the fluorescence serving as a measure of that molecule’s content at each synapse. The degree to which these measures change over time is then used to quantify the constancy of this molecule’s content at individual synapses.

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 442 The Neuroscientist 20(5)

Figure 2. Activity can accelerate the redistribution of synaptic molecules. (A) An axon from a neuron expressing photoactivatable green fluorescent protein–tagged Synapsin Ia (a synaptic vesicle–associated molecule that tethers vesicles to other vesicles and to the actin cytoskeleton) (Box 2). A volume-filling fluorescent protein (cyan fluorescent protein) was also expressed in the same axon (not shown), allowing the tracing of the axon’s contour. At time t = 0, one bouton (magenta arrowhead) was briefly illuminated at 405 nm, leading to the rapid photoactivation of the tagged synapsin. About 1 minute later, the neuron was stimulated at 20 Hz for 30 seconds using field stimulation. This was followed by the dispersion of the tagged synapsin from the photoactivated bouton and the rapid appearance of the photoactivated molecules in nearby boutons (yellow arrowheads). Bar = 10 µm. (B) Time course of fluorescence loss and fluorescence increase from the photoactivated bouton and in a nearby bouton, respectively. Note the strong acceleration of the redistribution process by the stimulation procedure. (C) Average fluorescence loss rate from photoactivated boutons (five experiments). Data shown here were corrected for ongoing photobleaching. interchanged among neighboring presynaptic sites, where secretion (Staras and others 2010). Moreover, such exper- they become incorporated into local pools and function as iments indicate that synaptic activation further challenges native vesicles or molecules (Darcy and others 2006; the capacity of synapses to retain their individual con- Fernandez-Alfonso and others 2006; Fernandez-Alfonso tents as they show that stimulation leads to the temporary and Ryan 2008; Fisher-Lavie and others 2011; Herzog dispersal of SVs and SV-associated molecules into the and others 2011; Krueger and others 2003; ; Staras and axoplasm and membrane of flanking axonal segments others 2010; Tsuriel and others 2006; Westphal and oth- (Chi and others 2001; Fisher-Lavie and others 2011; Li ers 2008; Wienisch and Klingauf 2006; see also Ribrault and Murthy 2001; Star and others 2005; Tsuriel and oth- and others 2011; reviewed in Staras and Branco 2010) ers 2006). Moreover, this activation can also accelerate and even serve as source material for the formation of the rates of their redistribution among nearby synapses new synapses (Dobie and Craig 2011; Krueger and others (Fisher-Lavie and others 2011). An illustration of this 2003). Importantly, exchange rates can be surprisingly point is provided in Figure 2. high (estimated to be ~4% per minute), and thus, migrat- If SVs and presynaptic molecules migrate between ing vesicles can constitute a substantial fraction (esti- synapses at such high rates as these experiments suggest, mated at ~40% or more) of the SV pools available for a question arises as to if and how individual presynaptic

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 Ziv and Fisher-Lavie 443 boutons manage to preserve their respective contents emphasize that these replacement processes, which we over long durations. A likely possibility is that the AZ refer to as exchange (and interchange) processes, are not represents a relatively stable focal point for presynaptic equivalent to protein synthesis and degradation (meta- organization, which defines, conserves, and maintains the bolic turnover). Rather, exchange rates primarily reflect location, size, and individual characteristics of each pre- the rates at which molecules are incorporated into and synaptic site, acting metaphorically (Tsuriel and others dissociate from their cognate molecular assemblies (as 2009) as “an island of stability in a sea of change.” explained in the section on metabolic turnover below). Indeed, ultrastructural studies, based on state-of-the-art As mentioned above, CAZ molecules tend to be huge, EM techniques, reveal that the AZ has an elaborate, pos- posing significant difficulties in creating and expressing sibly modular ultrastructure, which seems to organize fluorescently tagged variants of these molecules. presynaptic SV localization and distribution in the vicin- Nevertheless, FRAP experiments have been done with ity of the AZ and beyond (Siksou and others 2011). In several of these (Table 1), namely, Munc-13-1 (Kalla and terms of their molecular contents, AZs are now known to others 2006), Bassoon (Tsuriel and others 2009), and to contain several very large molecules—including Rab3- some extent Liprin-α2 and RIM1 (Spangler and others interacting molecule (RIM); RIM binding protein; Munc- 2013). Exchange rates of Bassoon were found to be rela- 13; glutamate (E)–, leucine (L)–, lysine (K)–, and serine tively slow, with time constants (which can be thought of (S)–rich protein (ELKS/ERC/CAST); Liprin-α; Piccolo; as the time at which approximately two thirds of the mol- and Bassoon—that form complexes often referred to as ecules are exchanged) on the order of 8 hours or more, the CAZ (Box 2). Further, CAZ molecules, most of which whereas Munc-13-1 exhibited somewhat faster exchange have multiple isoforms and splice variants, play impor- rates (on the order of 1.5 hours). The case of Munc-13-1 tant and often distinct roles in presynaptic function, from is particularly interesting as it was performed in neurons the regulation of SV docking, priming and release, to the from a knockin mouse in which the fusion protein was clustering of critical presynaptic molecules, such as volt- expressed from its endogenous genomic locus, resulting age-gated calcium channels. During the last decade, in no overexpression presumably due to the preservation much has been learned on these particular functions, on of endogenous expression control mechanisms. The the molecular interactions of CAZ molecules among exchange rates of Liprin-α2 and RIM1 have also been themselves, with other AZ molecules, and with SVs and explored and were reported to be on time scales of min- on their relationships with structural features of AZs as utes (Spangler and others 2013). The short duration of revealed by sophisticated EM and light microscopy tech- these experiments (5 minutes), however, precluded esti- niques (Gundelfinger and Fejtova 2012; Siksou and oth- mations of exchange rates for Liprin-α2 and RIM1 ers 2011; Südhof 2012). These important insights will not assembled into more stable AZ complexes, but prelimi- be addressed here. Instead, we will take it for granted that nary experiments indicate that these too are on the order these molecules play essential roles in presynaptic orga- of several hours (Zürner, Mittelstaedt, Schoch and Ziv, nization and function and consider instead the molecular unpublished data). It would thus seem that the AZ is dynamics of the CAZ and AZ molecules themselves. essentially a dynamic molecular assembly at steady state; Historically, CAZ molecules were defined as such yet, from the perspective of more freely diffusible synap- based on their high resistance to rather harsh biochemi- tic molecules and SVs, it also represents a relatively sta- cal preparative techniques, giving rise to the notion that ble core that could potentially conserve presynaptic they are part of very stable molecular complexes. As properties, at least over time scales of hours. mentioned above, the advent of new live imaging techniques provided new opportunities to examine this notion. One particularly useful technique in this regard is The Post Synaptic Density fluorescence recovery after photobleaching (FRAP) The PSD is a generic name for the complex of postsyn- (Kim and others 2010; Staras and others 2013). This aptic molecules located at the postsynaptic membrane of approach is based on the expression of a fluorescently CNS synapses. Its name comes from the fact that it tagged variant of a synaptic molecule, which is thereafter appears in EM images as an electron-dense thickening of selectively photobleached by strong illumination spa- the postsynaptic membrane, approximately 30 to 50 nm tially confined to the photobleached synapse (Box 1). As in width (for glutamatergic excitatory synapses; less for the bleaching of the fluorescent group is essentially irre- GABAergic inhibitory synapses), aligned with and jux- versible, the recovery of fluorescence at the photo- taposed against the presynaptic AZ. As its position and bleached synapse is indicative of the rate at which ultrastructural appearance imply, PSDs are composed of photobleached molecules within the photobleached syn- proteinaceous scaffolds that serve to anchor and confine apse are replaced by fluorescent molecules from sources neurotransmitter receptors at the postsynaptic mem- outside the photobleached synapse. It is important to brane. The molecular composition of these scaffolds

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 444 The Neuroscientist 20(5)

Box 2. Synaptic Scaffold Molecules

A schematic illustration of the molecular contents of presynaptic and postsynaptic scaffolds of a glutamatergic synapse. The active zone and postsynaptic densities are shaded. The sizes and spatial relationships of the various molecules are not to be taken literally, as the actual molecular structure and composition of these scaffolds are immensely more complex. Red checkmarks indicate the existence of published studies establishing the dynamic behavior of that molecule as detailed in the table below (Table 1). References marked with asterisks in Table 1 are studies in which molecular dynamics were explored for periods long enough to establish exchange rates of tightly incorporated molecules. Note that the dynamics of addi- tional molecules mentioned in this figure have been studied (synapsin, NMDA receptors, AMPA receptors, synaptic vesicle molecules, actin filaments); however, these are not strictly scaffold molecules and therefore are not listed here.

Table 1. Synaptic Scaffold Molecules for which Extensive Exchange Dynamics Have Been Reported.

Short-term imaging Long-term imaging Molecule Methods (seconds, minutes) (>1 hour) References Bassoon FRAP, PA + + *Tsuriel and others 2009 RIM FRAP + Spangler and others 2013 Liprin-α FRAP + Spangler and others 2013 Munc-13 FRAP + + *Kalla and others 2006 CASK FRAP + Spangler and others 2013 Syntaxin FRAP, SPT + Ribrault and others 2011 β-neurexin FRAP + Fu and Huang 2010 PSD-95 FRAP, PA + + *Gray and others 2006; *Kuriu and others 2006; Nakagawa and others 2004a; Nakagawa and others 2004b; *Nelson and others 2013; Okabe and others 2001; Sharma and others 2006; Yao and others 2003; *Zeidan and Ziv 2012 N-cadherin FRAP + Thoumine and others 2006 SAP97 FRAP + Nakagawa and others 2004a; Waites and others 2009 SAP102 FRAP + Zheng and others 2010 ProSAP/SHANK FRAP, PA + + Kuriu and others 2006; *Tsuriel and others 2006 Homer FRAP + Kuriu and others 2006; Okabe and others 2001 GKAP FRAP + Kuriu and others 2006; Yao and others 2003 CaMKII FRAP, PA + Lee and others 2009; Sharma and others 2006; Shen and Meyer 1999 α-actinin FRAP + Nakagawa and others 2004b Stargazin (TARP) FRAP + Bats and others 2007 Gephyrin FRAP + + Calamai and others 2009; Charrier and others 2010; *Vlachos and others 2013 Drebrin FRAP + Takahashi and others 2009

CASK = calcium (Ca2+/calmodulin [CaM])–associated serine kinase; FRAP = fluorescence recovery after photobleaching; GKAP = guanylate kinase–associated protein (also known as SAPAP); PA = Photoactivation; RIM = Rab3-interacting molecule; SPT = single particle tracking; TARP = transmembrane AMPA receptor regulatory protein.

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 Ziv and Fisher-Lavie 445

(which differs greatly between glutamatergic excitatory alterations in PSD content or size, but other than that, and glycinergic/GABAergic inhibitory synapses) is rich postsynaptic scaffolds are viewed as relatively stable and intricate, with new information on scaffold contents structures. Indeed, early studies in which FRAP of fluo- and organization emerging continuously. Here too, we rescently tagged molecules was used to estimate the will refrain from describing particular scaffolding mole- exchange rates of postsynaptic scaffolding molecules led cules and their putative roles (reviwed in Feng and Zhang to the somewhat misleading conclusion that synapses 2009; Kneussel and Loebrich 2007; Sheng and contain two pools of these molecules: a relatively small Hoogenraad 2007; Verpelli and others 2012). Instead, we pool of “mobile” or “dynamic” scaffolding molecules focus on the dynamics of these molecules as exemplified that fully exchange over the course of a few minutes and by prominent members of this family. a significant “immobile” or “stable” pool that exhibits As mentioned above, PSD scaffolds serve to localize practically no exchange dynamics. In these early studies, and confine neurotransmitter receptors at postsynaptic however, experiment durations were rather short (sec- membranes. It is now apparent that such receptor con- onds to a few minutes), and thus, exchange dynamics of finement processes are much more complex than initially the “immobile” pool, believed to represent molecules perceived: Live imaging studies indicate that major neu- tightly integrated into the PSD, were missed. When rotransmitter receptors, namely AMPA-type glutamate FRAP and photoactivation experiments were extended receptors at excitatory synapses as well as glycine and to longer time scales (i.e., hours), however, it became GABA receptors at inhibitory synapses, continuously apparent that the “immobile” pool is also dynamic, albeit move in and out of synapses at rapid rates. Initially, it was slower, with molecules moving in and out of PSDs over thought that these processes involved mainly receptor time scales of several hours. Furthermore, it now seems endocytosis and exocytosis. However, it is now quite that the “mobile” pools represent molecules only weakly clear that these processes occur primarily through lateral attached to the PSD or associated with dynamic cyto- diffusion in the plane of the postsynaptic membrane, skeletal components, of which the most notable are actin where receptors are occasionally confined through tran- filaments (Kuriu and others 2006; Zheng and others sient interactions with PSD scaffolding molecules 2010). These experiments thus suggest that even scaf- (Choquet and Triller 2013). In fact, when individual folding molecules tightly integrated into the PSD, such receptors are tracked using single particle tracking meth- as PSD-95, move in, out, and between PSDs at consider- ods (Kim and others 2010), receptors show intermittent able rates (Gray and others 2006; Mondin and others transitions between confined states, during which they 2011; Specht and Triller 2008; Tsuriel and others 2006; pause and show little mobility, to unconfined states, dur- Zeidan and Ziv 2012), implying that scaffold stability is ing which they diffuse within PSDs and between neigh- a relative term. boring synapses, resulting in synaptic residence times Beyond the migration of scaffold molecules in and that can be as short as a few seconds (but are often longer: out of synapses, older studies (Fischer and others 1998) e.g., Ashby and others 2006; Kerr and Blanpied 2012). It as well as recent ones (Kerr and Blanpied 2012) consis- is believed that pauses reflect transient binding events to tently indicate that postsynaptic densities of excitatory specific postsynaptic scaffolding molecules and perhaps synapses undergo continuous, actin-driven “morphing” also less specific steric interactions with “obstacles” in or “contortions” on time scales of minutes. Moreover, the membrane plane. Thus, momentary receptor contents several very recent studies based on “super-resolution” of individual synapses reflect the population outcome of light microscopy techniques revealed that within indi- receptor lateral diffusion kinetics, the extent of steric vidual synapses, scaffolding molecules are often clus- interferences, the number of binding sites created by scaf- tered into “nanodomains,” that is, very small (<100 nm) fold proteins, the binding affinities to these sites, extra- patches of scaffold molecules that tend to co-localize synaptic receptor concentrations, membrane trafficking with their cognate receptors (Fukata and others 2013; dynamics, and intracellular receptor pool sizes. MacGillavry and others 2013; Nair and others 2013; In this description, receptor confinement at the post- Specht and others 2013). At glutamatergic synapses, synaptic membrane is conceptualized as a diffusion- such nanodomains morph on time scales of minutes reaction process, with postsynaptic scaffolds providing (MacGillavry and others 2013), with the vast majority and defining a certain number of “slots” to which recep- of nanodomains exhibiting life spans of less than 1 hour tors can bind (Bats and others 2007; Jacob and others (Nair and others 2013). Thus, postsynaptic scaffolds are 2005; Lisman and Raghavachari 2006; Opazo and others not only subject to a continuous exchange of their 2012; Yudowski and others 2013). Physiological signals molecular constituents but also to continuous changes in can thus alter synaptic properties by changing receptor their higher level organization, further challenging the affinities to postsynaptic scaffolds (e.g., by phosphoryla- notion that postsynaptic scaffolds are essentially stable tion) or by changing the number of “slots” through “structures.”

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 446 The Neuroscientist 20(5)

Scaffold Molecule Metabolic Scaffold Dynamics and Synaptic Turnover Tenacity As mentioned above, the dynamics of presynaptic and Individual synapses contain, on average, tens to hundreds postsynaptic scaffolding molecules described so far of copies of specific scaffold molecules (Sheng and reflect the rates at which synaptic molecules are incorpo- Hoogenraad 2007). How do the dynamics of scaffolding rated into and dissociate from their cognate molecular molecules affect the constancy of their contents at spe- assemblies, not the metabolic turnover of these molecules cific synapses? It could be argued that the continuous (i.e., the rates at which these are synthesized and exchange and interchange of presynaptic and postsynap- degraded). How do exchange dynamics of scaffold mol- tic scaffolding molecules have no net effects, and thus, ecules compare with their metabolic turnover rates? synaptic scaffold contents (and by extension, synaptic Historically, older studies based on radiolabeling functional properties) remain constant over behavioral methods indicated that turnover rates of some presynap- time scales. It might also be argued, however, that these tic proteins can be remarkably slow, resulting in half- same dynamics continuously challenge the capacity of lives (i.e., the time over which one half of an initial individual synapses to preserve their molecular contents protein quotient is degraded) of many days and even over long time scales, and consequently, the constancy of weeks (Baitinger and Willard 1987; Petrucci and others synaptic properties is compromised. This is a crucial mat- 1991). More recent studies, however, have reported half- ter: Activity-induced modification of synaptic connec- lives for presynaptic and postsynaptic scaffolding pro- tions (“synaptic plasticity”) is widely believed to teins in the range of several hours (Ehlers 2003; Yao and represent a major mechanism for modifying the func- others 2007). These high turnover rates appear surpris- tional properties of neuronal networks, possibly providing, as they are not entirely congruent with the slow ing a cellular basis for phenomena collectively referred to transport rates of these proteins from somata to remote as “learning and memory.” If synaptic tenacity, the capac- axonal and dendritic synaptic sites (Scott and others ity of individual synapses to maintain their individual 2011; Tang and others 2013; Tsuriel and others 2006). properties over time, is inherently limited, spontaneous Moreover, wherever examined, steady state levels or changes, occurring independently of physiologically rel- recovery rates in FRAP experiments of fluorescently evant input, would cause spurious changes in network tagged synaptic scaffold molecules were not affected by function or undo physiologically relevant ones, placing the pharmacological inhibition of protein synthesis over severe constraints on synaptic plasticity as a mechanism many hours (Herzog and others 2011; Kalla and others for realizing long-term directed changes in network func- 2006; Tsuriel and others 2006). Recently, state-of-the-art tion. How tenacious are synapses in this respect? metabolic labeling methods combined with mass spec- Over the last few years, several groups have used fluo- troscopy were used to systematically measure the turn- rescently tagged scaffolding molecules and long-term over rates of synaptic and neuronal proteins. These imaging to examine the constancy of synaptic scaffold studies revealed that turnover rates of synaptic scaffold- molecule contents at individual synapses. Practically all ing proteins are quite slow, with half-lives on the order of these studies observed large fluctuations in scaffold mol- 2 to 5 days in cell culture (Cohen and others 2013b) and ecule contents at individual synapses when these were probably three to four times longer in adult mice (Price followed for many hours and days (Box 1c). Thus, for and others 2010). Comparisons of exchange rates (e.g., example, long-term imaging of individual postsynaptic measured by FRAP) and metabolic turnover rates for the compartments expressing fluorescently tagged variants same molecules indicate that the former are one to two of postsynaptic scaffolding molecules such as guanylate orders of magnitude greater than the latter (Cohen and kinase–associated protein (GKAP), ProSAP/Shank, others 2013b). It thus seems that at first approximation, Homer, and most notably PSD-95 revealed that the syn- the dynamics of many synaptic scaffold molecules are aptic contents of these molecules change considerably predominated by protein exchange and mobilization over the time course of hours (Fisher-Lavie and Ziv 2013; rather than protein synthesis and degradation, with the Kuriu and others 2006; Okabe and others 2001; Zeidan latter primarily serving to maintain local, shared pools of and Ziv 2012) and days (Cane and others 2014; Gray and these proteins (Tsuriel and others 2006). It should be others 2006; Kaufman and others 2012; Minerbi and oth- mentioned, however, that mass spectroscopy–based ers 2009). Importantly, although fluctuations were most analyses are biased toward relatively abundant proteins, pronounced in active networks, suppressing all activity and thus, it remains possible that some less abundant had only a modest impact on the extent of these fluctua- scaffold proteins exhibit higher synthesis and degrada- tions. Similarly, very significant fluctuations were tion rates, in particular in response to specific physiolog- observed in the synaptic contents of the presynaptic AZ ical signals (Waites and others 2013). molecules Bassoon and Munc-13 (Fisher-Lavie and

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 Ziv and Fisher-Lavie 447 others 2011; Fisher-Lavie and Ziv 2013; Matz and others raising questions as to their general applicability to the 2010) as well as SV contents (Fisher-Lavie and others intact brain. 2011; Herzog and others 2011; Zeidan and Ziv 2012). Very few studies have examined the long-term func- In most of the aforementioned studies, long-term tional stability of specific synaptic connections, but measurements were made of either presynaptic or post- where performed, connection strength was found to fluc- synaptic scaffold molecule contents but not both simul- tuate significantly (Loebel and others 2013; Tsukamoto- taneously, and thus, the degree to which fluctuations in Yasui and others 2007; Usami and others 2008). Thus, for scaffold molecule contents on one side of the synapse example, in an electrophysiological study performed in correlate with fluctuations in scaffold molecule contents rat cortical slices (Loebel and others 2013), paired record- across the synaptic cleft was not known. Given the pre- ings performed twice from the same neurons at an inter- cise juxtaposition of presynaptic and postsynaptic mem- val of 12 hours revealed that synaptic connection strengths brane specializations observed in EM micrographs, it between these neuron pairs changed over a 10-fold range might be expected that changes in presynaptic and post- during this time interval. Moreover, by performing com- synaptic scaffold molecule contents would go hand in pound stimulation paradigms and fitting responses to a hand. Interestingly, however, when comparisons are model, it was concluded that these mainly reflected well- made by quantitative immunohistochemistry, correla- matched changes in both presynaptic and postsynaptic tions between presynaptic and postsynaptic scaffold pro- properties. tein contents, while positive, are often quite imperfect. More direct evidence for strong relationships between While such imperfections might simply reflect measure- changes in presynaptic scaffold content and synaptic ment inaccuracies, they might indicate that presynaptic function was provided in a study by Matz and colleagues and postsynaptic molecular contents fluctuate asynchro- (2010). This study showed a good correspondence nously, and thus, when comparisons are made at single between the synaptic content of the AZ molecule Bassoon time points (e.g., in fixed tissue), correlations between and SV release probability at the same synapses. presynaptic and postsynaptic contents are underesti- Moreover, spontaneous changes in Bassoon content were mated. Alternatively, they might imply that presynaptic/ associated with corresponding changes in neurotransmit- postsynaptic “stoichiometries,” that is, relative quanti- ter release probability. Interestingly, we too have observed ties of presynaptic and postsynaptic scaffold molecules, an excellent temporal correspondence between changes differ significantly from one synapse to another. In a in Bassoon content and changes in total SV pool size recent study (Fisher-Lavie and Ziv 2013), the degree to (Fig. 3), again suggesting that spontaneous fluctuations in which changes in scaffolding molecule contents corre- scaffold molecule contents reflect general changes in late across the synaptic cleft was examined. Concomitant, synapse properties. long-term imaging of the contents of the AZ molecule Finally, it seems that the recent aforementioned Munc-13-1 and the PSD molecule PSD-95 at the same findings on the nanodomain-based organization of synapse revealed that changes in the contents of these postsynaptic scaffolds have important functional molecules were generally well correlated, indicating that implications. Simulations suggest that this form of fluctuations in the contents of scaffolding molecules in clustered organization strongly shapes and impacts one synaptic compartment are at least partially indicative postsynaptic responses (MacGillavry and others 2013; of general changes in synaptic properties. Interestingly, Nair and others 2013). If nanodomain organization is ratios of Munc-13-1 to PSD-95 contents at individual as dynamic as these studies suggest, then postsynaptic synapses, which varied over a fourfold range among dif- responses might be expected to change significantly ferent synapses, were conserved quite well, indicating over similar time scales. that in spite of spontaneous changes in scaffold molecule In comparison to the large body of work on synaptic contents, synapses are capable of conserving particular scaffold molecule dynamics performed in cell and tissue presynaptic/postsynaptic “stoichiometries,” at least over cultures, in vivo studies of this topic (Cane and others time scales of many hours. 2014; Gray and others 2006; Herzog and others 2011) are These findings thus suggest that the dynamics exhib- scarce. One notable study is that of Gray and colleagues ited by scaffolding molecules are not without a net effect (2006), which studied the dynamics of PSD-95 in the and that these dynamics are associated with significant intact cortex of mice. This study points to several inter- (spontaneous) changes in synaptic properties. Yet, it esting findings: First, PSD-95 was shown to move in, out, should be noted that 1) while the aforementioned studies and between synapses at high rates (with typical resi- reported large fluctuations in synaptic contents of partic- dency times of <1 hour). Second, PSD-95 contents at ular molecules, the impact of such fluctuations on synap- individual synapses were found to change considerably tic function remained mostly unknown, and 2) most of over time scales of days (a finding recently confirmed these studies were carried out in ex vivo preparations, and greatly extended by Cane and others, 2014). Third, a

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 448 The Neuroscientist 20(5)

Figure 3. Changes in the synaptic contents of the active zone molecule Bassoon and synaptic vesicles (SVs) are temporally correlated. (A) An axon of a neuron expressing a variant of Bassoon tagged with a green fluorescent protein and the SV molecule SV2A tagged with a red fluorescent protein (mCherry). Bar = 10 µm. (B) Changes in Bassoon and SV2A contents measured over several hours in two presynaptic boutons. (C) Changes in Bassoon and SV2A over time. Note that in one case, the correlation (Pearson correlation) between the temporal changes in the fluorescence of the two molecules is very high (0.90), whereas in the other, it is lower (0.43). (D) Distribution of temporal correlation values for all boutons followed in these experiments compared to the correlations measured for all possible combinations in which the data for Bassoon and SV2A were taken from different boutons. Data shown here were corrected for ongoing photobleaching. Mean correlations are shown as r. negative correlation was found between PSD-95 dynam- Although in vivo studies on synaptic scaffold dynamics and synaptic size, and fourth, PSD-95 dynamics were ics are scarce, quite a few in vivo studies published over shown to subside (but not disappear) as animals matured the last decade reported that the volume of dendritic spines (a similar maturation-dependent stabilization was also fluctuates considerably over time scales of many hours reported for Gephyrin, the major scaffold molecule of and days (Holtmaat and others 2006; Loewenstein and glycinergic/GABAergic inhibitory synapses) (Vlachos others 2011). As dendritic spine volume is assumed to and others 2013). reflect synaptic size, and by inference, synaptic scaffold

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 Ziv and Fisher-Lavie 449 molecular content and synaptic strength (Holtmaat and Funding Svoboda 2009), the observed changes in spine volume The author(s) received the following financial support for the fluctuations are in full agreement with both in vivo and ex research, authorship, and/or publication of this article: This vivo observations on the (in)constancy of synaptic scaf- work was supported by funding from the United States Israel fold molecule contents at individual synapses. Binational Science Foundation (2007425), the European Union Seventh Framework Programme under grant agreement HEALTH-F2–2009–241498 (“EUROSPIN”), the Deutsch- Summary Israelische-Projektkooperation German-Israeli Project The development of methods to follow the dynamics of Cooperation Foundation, and the Intel Collaborative Research synaptic molecules in living neurons has radically altered Institute for Computational Intelligence (ICRICI). our view of the synapse, from that of a relatively static structure to that of a dynamic molecular assembly at steady References state. In this regard, presynaptic and postsynaptic scaffold Ashby MC, Maier SR, Nishimune A, Henley JM. 2006. Lateral molecules, which undoubtedly confer some degree of sta- diffusion drives constitutive exchange of AMPA receptors bility to these assemblies over time scales of hours and at dendritic spines and is regulated by spine morphology. J days, are themselves subject to significant dynamics, albeit Neurosci 26:7046–55. slower than those of more mobile molecules such as neu- Baitinger C, Willard M. 1987. Axonal transport of synapsin rotransmitter receptors, and cytoskeletal elements. As dis- I-like proteins in rabbit retinal ganglion cells. J Neurosci 7:3723–35. cussed above, these dynamics are probably not without Bats C, Groc L, Choquet D. 2007. The interaction between effect; wherever studied, these seem to be associated with Stargazin and PSD-95 regulates AMPA receptor surface spontaneous changes in synaptic properties, manifested as trafficking. Neuron 53:719–34. changes in scaffold molecule content, synaptic size, and Calamai M, Specht CG, Heller J, Alcor D, Machado P, Vannier possibly synaptic function. Thus, it would seem that the C, Triller A. 2009. Gephyrin oligomerization controls GlyR tenacity of individual synapses is inherently limited. mobility and synaptic clustering. J Neurosci 29:7639-48. How can these findings be reconciled with the prevail- Cane M, Maco B, Knott G, Holtmaat A. 2014. The Relationship ing notion of synaptic plasticity as a major mechanism for between PSD-95 Clustering and Spine Stability In Vivo. J directed modification of neuronal network function? Neurosci 34:2075-86. Several possibilities exist. Perhaps dynamics differ Charrier C, Machado P, Tweedie-Cullen RY, Rutishauser D, between synapses; thus perhaps, scaffold molecular Mansuy IM, Triller A. 2010. A crosstalk between β1 and β3 integrins controls glycine receptor and gephyrin trafficking dynamics might be slower for certain synaptic popula- at synapses. Nat Neurosci 13:1388-95. tions, such as large synapses (Gray and others 2006; Chi P, Greengard P, Ryan TA. 2001. Synapsin dispersion and Yasumatsu and others 2008); perhaps they slow down reclustering during synaptic activity. Nat Neurosci 4:1187– with maturation (Gray and others 2006; Grillo and others 93. 2013; Mostany and others 2013; Vlachos and others Choquet D, Triller A. 2013. The dynamic synapse. Neuron 2013). Alternatively, fluctuations at the level of individual 80:691–703. synapses might “average out” wherever neuron to neuron Cohen L, Koffman N, Meiri H, Yarom Y, Lampl I, Mizrahi connections are based on multiple synapses. Yet, the pos- A. 2013a. Time-lapse electrical recordings of single neu- sibility exists that the dynamics described here inherently rons from the mouse neocortex. Proc Natl Acad Sci U S A limit the tenacity of synaptic connections. If so, perhaps 110:5665–70. continuous retuning of synaptic properties might be Cohen LD, Zuchman R, Sorokina O, Müller A, Dieterich DC, Armstrong JD, and others. 2013b. Metabolic turnover of required to maintain network function for long durations synaptic proteins: kinetics, interdependencies and implica- (Cohen and others 2013a). Conversely, reduced synaptic tions for synaptic maintenance. PLoS One 8:e63191. tenacity might be associated with the deterioration of net- Darcy KJ, Staras K, Collinson LM, Goda Y. 2006. Constitutive work function during aging (Grillo and others 2013; sharing of recycling synaptic vesicles between presynaptic Mostany and others 2013). Whatever conclusions might boutons. Nat Neurosci 9:315–21. be drawn from these findings, the dynamics of scaffolding Dobie FA, Craig AM. 2011. Inhibitory synapse dynamics: molecules; their migration in, out, and between synapses; coordinated presynaptic and postsynaptic mobility and the and changes in their contents over time should be kept in major contribution of recycled vesicles to new synapse for- mind when considering synaptic plasticity phenomena mation. J Neurosci 31:10481–93. and the permanency of changes that these induce. Ehlers MD. 2003. Activity level controls postsynaptic composition and signaling via the ubiquitin-proteasome system. Nat Neurosci 6:231–42. Declaration of Conflicting Interests Feng W, Zhang M. 2009. Organization and dynamics of PDZ- The author(s) declared no potential conflicts of interest with respect domain-related supramodules in the postsynaptic density. to the research, authorship, and/or publication of this article. Nat Rev Neurosci 10:87–99.

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 450 The Neuroscientist 20(5)

Fernandez-Alfonso T, Kwan R, Ryan TA. 2006. Synaptic vesi- Kerr JM, Blanpied TA. 2012. Subsynaptic AMPA receptor dis- cles interchange their membrane proteins with a large sur- tribution is acutely regulated by actin-driven reorganization face reservoir during recycling. Neuron 51:179–86. of the postsynaptic density. J Neurosci 32:658–73. Fernandez-Alfonso T, Ryan TA. 2008. A heterogeneous “rest- Kim SA, Sanabria H, Digman MA, Gratton E, Schwille P, ing” pool of synaptic vesicles that is dynamically inter- Zipfel WR, and others. 2010. Quantifying translational changed across boutons in mammalian CNS synapses. mobility in neurons: comparison between current optical Brain Cell Biol 36:87–100. techniques. J Neurosci 30:16409–16. Fischer M, Kaech S, Knutti D, Matus A. 1998. Rapid actin- Kneussel M, Loebrich S. 2007. Trafficking and synaptic based plasticity in dendritic spines. Neuron 20:847–54. anchoring of ionotropic inhibitory neurotransmitter recep- Fisher-Lavie A, Zeidan A, Stern M, Garner CC, Ziv NE. tors. Biol Cell 99:297–309. 2011. Use dependence of presynaptic tenacity. J Neurosci Krueger SR, Kolar A, Fitzsimonds RM. 2003. The presynaptic 31:16770–80. release apparatus is functional in the absence of dendritic Fisher-Lavie A, Ziv NE. 2013. Matching dynamics of contact and highly mobile within isolated axons. Neuron presynaptic and postsynaptic scaffolds. J Neurosci 40:945–57. 33:13094–100. Kuriu T, Inoue A, Bito H, Sobue K, Okabe S. 2006. Differential Fu Y, Huang ZJ. 2010. Differential dynamics and activity- control of postsynaptic density scaffolds via actin-depen- dependent regulation of alpha- and beta-neurexins at devel- dent and -independent mechanisms. J Neurosci 26:7693– oping GABAergic synapses. Proc Natl Acad Sci U S A. 706. 107:22699-704. Lee SJ, Escobedo-Lozoya Y, Szatmari EM, Yasuda R. 2009. Fukata Y, Dimitrov A, Boncompain G, Vielemeyer O, Perez Activation of CaMKII in single dendritic spines during F, Fukata M. 2013. Local palmitoylation cycles define long-term potentiation. Nature 458:299-304. activity-regulated postsynaptic subdomains. J Cell Biol Li Z, Murthy VN. 2001. Visualizing postendocytic traffic 202:145–61. of synaptic vesicles at hippocampal synapses. Neuron Gray NW, Weimer RM, Bureau I, Svoboda K. 2006. Rapid 31:593–605. redistribution of synaptic PSD-95 in the neocortex in vivo. Lisman J, Raghavachari S. 2006. A unified model of the pre- PLoS Biol 4:2065–75. synaptic and postsynaptic changes during LTP at CA1 syn- Grillo FW, Song S, Teles-Grilo Ruivo LM, Huang L, Gao G, apses. Sci STKE 2006:re11. Knott GW, and others. 2013. Increased axonal bouton Loebel A, Le Bé JV, Richardson MJ, Markram H, Herz AV. dynamics in the aging mouse cortex. Proc Natl Acad Sci U 2013. Matched pre- and post-synaptic changes under- S A 110:E1514–23. lie synaptic plasticity over long time scales. J Neurosci Gundelfinger ED, Fejtova A. 2012. Molecular organization and 33:6257–66. plasticity of the cytomatrix at the active zone. Curr Opin Loewenstein Y, Kuras A, Rumpel S. 2011. Multiplicative Neurobiol 22:423–30. dynamics underlie the emergence of the log-normal distri- Hellwig B, Schüz A, Aertsen A. 1994. Synapses on axon col- bution of spine sizes in the neocortex in vivo. J Neurosci laterals of pyramidal cells are spaced at random intervals: 31:9481–8. a Golgi study in the mouse cerebral cortex. Biol Cybern MacGillavry HD, Song Y, Raghavachari S, Blanpied TA. 2013. 71:1–12. Nanoscale scaffolding domains within the postsynaptic Herzog E, Nadrigny F, Silm K, Biesemann C, Helling I, Bersot density concentrate synaptic AMPA receptors. Neuron T, and others. 2011. In vivo imaging of intersynaptic 78:615–22. vesicle exchange using VGLUT1 Venus knock-in mice. J Matz J, Gilyan A, Kolar A, McCarvill T, Krueger SR. 2010. Neurosci 31:15544–59. Rapid structural alterations of the active zone lead to sus- Holtmaat A, Svoboda K. 2009. Experience-dependent struc- tained changes in neurotransmitter release. Proc Natl Acad tural synaptic plasticity in the mammalian brain. Nat Rev Sci U S A 107:8836–41. Neurosci 10:647–58. Minerbi A, Kahana R, Goldfeld L, Kaufman M, Marom S, Ziv Holtmaat A, Wilbrecht L, Knott GW, Welker E, Svoboda K. NE. 2009. Long-term relationships between synaptic tenac- 2006. Experience dependent and cell-type-specific spine ity, synaptic remodeling, and network activity. PLoS Biol growth in the neocortex. Nature 441:979–83. 7:e1000136. Jacob TC, Bogdanov YD, Magnus C, Saliba RS, Kittler JT, Mondin M, Labrousse V, Hosy E, Heine M, Tessier B, Levet Haydon PG, and others. 2005. Gephyrin regulates the F, and others. 2011. Neurexin-neuroligin adhesions capture cell surface dynamics of synaptic GABAA receptors. J surface-diffusing AMPA receptors through PSD-95 scaf- Neurosci 25:10469–78. folds. J Neurosci 31:13500–15. Kalla S, Stern M, Basu J, Varoqueaux F, Reim K, Rosenmund Mostany R, Anstey JE, Crump KL, Maco B, Knott G, Portera- C, and others. 2006. Molecular dynamics of a presynaptic Cailliau C. 2013. Altered synaptic dynamics during normal active zone protein studied in Munc13-1-enhanced yel- brain aging. J Neurosci 33:4094–104. low fluorescent protein knock-in mutant mice. J Neurosci Nair D, Hosy E, Petersen JD, Constals A, Giannone G, Choquet 26:13054–66. D, and others. 2013. Super-resolution imaging reveals that Kaufman M, Corner MA, Ziv NE. 2012. Long-term relation- AMPA receptors inside synapses are dynamically orga- ships between cholinergic tone, synchronous bursting and nized in nanodomains regulated by PSD95. J Neurosci synaptic remodeling. PLoS One 7:e40980. 33:13204–24.

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 Ziv and Fisher-Lavie 451

Nakagawa T, Futai K, Lashuel HA, Lo I, Okamoto K, Walz T, Specht CG, Izeddin I, Rodriguez PC, El Beheiry M, Rostaing and others. 2004a. Quaternary structure, protein dynamics, P, Darzacq X, and others. 2013. Quantitative nanoscopy and synaptic function of SAP97 controlled by L27 domain of inhibitory synapses: counting gephyrin molecules and interactions. Neuron 44:453-67. receptor binding sites. Neuron 79:308–21. Nakagawa T, Engler JA, Sheng M. 2004b. The dynamic turn- Specht CG, Triller A. 2008. The dynamics of synaptic scaf- over and functional roles of alpha-actinin in dendritic folds. Bioessays 30:1062–74. spines. Neuropharmacology 47:734-45. Star EN, Newton AJ, Murthy VN. 2005. Real-time imaging of Nelson CD, Kim MJ, Hsin H, Chen Y, Sheng M. (2013) Rab3a and Rab5a reveals differential roles in presynaptic Phosphorylation of threonine-19 of PSD-95 by GSK-3β is function. J Physiol 569(Pt 1):103–17. required for PSD-95 mobilization and long-term depres- Staras K, Branco T. 2010. Sharing vesicles between central sion. J Neurosci 33:12122-35. presynaptic terminals: implications for synaptic function. Okabe S. 2012. Molecular dynamics of the excitatory synapse. Front Synaptic Neurosci 2:20. Adv Exp Med Biol 970:131–52. Staras K, Branco T, Burden JJ, Pozo K, Darcy K, Marra V, and Okabe S, Urushido T, Konno D, Okado H, Sobue K. 2001. others. 2010. A vesicle superpool spans multiple presyn- Rapid redistribution of the postsynaptic density protein aptic terminals in hippocampal neurons. Neuron 66:37–44. PSD-Zip45 (Homer 1c) and its differential regulation Staras K, Mikulincer D, Gitler D. 2013. Monitoring and by NMDA receptors and calcium channels. J Neurosci quantifying dynamic physiological processes in live neu- 21:9561–71. rons using fluorescence recovery after photobleaching. J Opazo P, Sainlos M, Choquet D. 2012. Regulation of AMPA Neurochem 126:213–22. receptor surface diffusion by PSD-95 slots. Curr Opin Südhof TC. 2012. The presynaptic active zone. Neuron 75:11– Neurobiol 22:453–60. 25. Petrucci TC, Macioce P, Paggi P. 1991. Axonal transport kinet- Takahashi H, Yamazaki H, Hanamura K, Sekino Y, Shirao T. ics and posttranslational modification of synapsin I in Activity of the AMPA receptor regulates drebrin stabili- mouse retinal ganglion cells. J Neurosci 11:2938–46. zation in dendritic spine morphogenesis. 2009. J Cell Sci Price JC, Guan S, Burlingame A, Prusiner SB, Ghaemmaghami 122(Pt 8):1211-9. S. 2010. Analysis of proteome dynamics in the mouse Tang Y, Scott D, Das U, Gitler D, Ganguly A, Roy S. 2013. Fast brain. Proc Natl Acad Sci U S A 107:14508–13. vesicle transport is required for the slow axonal transport of Ribrault C, Reingruber J, Petković M, Galli T, Ziv NE, synapsin. J Neurosci 33:15362–75. Holcman D, and others. 2011. Syntaxin1A lateral diffusion Thoumine O, Lambert M, Mège RM, Choquet D. 2005. reveals transient and local SNARE interactions. J Neurosci Regulation of N-cadherin dynamics at neuronal contacts 31:17590–602. by ligand binding and cytoskeletal coupling. Mol Biol Cell Sharma K, Fong DK, Craig AM. 2006. Postsynaptic pro- 17:862-75. tein mobility in dendritic spines: long-term regulation by Tsukamoto-Yasui M, Sasaki T, Matsumoto W, Hasegawa A, synaptic NMDA receptor activation. Mol Cell Neurosci. Toyoda T, Usami A, and others. 2007. Active hippocam- 31:702-12. pal networks undergo spontaneous synaptic modification. Scott DA, Das U, Tang Y, Roy S. 2011. Mechanistic logic PLoS One 2:e1250. underlying the axonal transport of cytosolic proteins. Tsuriel S, Fisher A, Wittenmayer N, Dresbach T, Garner CC, Neuron 70:441–54. Ziv NE. 2009. Exchange and redistribution dynamics of Shen K, Meyer T. 1999. Dynamic control of CaMKII transloca- the cytoskeleton of the active zone molecule bassoon. J tion and localization in hippocampal neurons by NMDA Neurosci 29:351–8. receptor stimulation. Science 284:162-6. Tsuriel S, Geva R, Zamorano P, Dresbach T, Boeckers T, Sheng M, Hoogenraad CC. 2007. The postsynaptic architecture Gundelfinger ED, and others. 2006. Local sharing as a of excitatory synapses: a more quantitative view. Annu Rev predominant determinant of synaptic matrix molecular Biochem 76:823–47. dynamics. PLoS Biol 4:e271. Shepherd GM, Harris KM. 1998. Three-dimensional structure Usami A, Matsuki N, Ikegaya Y. 2008. Spontaneous plasticity and composition of CA3 -->CA1 axons in rat hippocampal of multineuronal activity patterns in activated hippocampal slices: implications for presynaptic connectivity and com- networks. Neural Plast 2008:108969. partmentalization. J Neurosci 18:8300–10. Verpelli C, Schmeisser MJ, Sala C, Boeckers TM. 2012. Shepherd GM, Raastad M, Andersen P. 2002. General and Scaffold proteins at the postsynaptic density. Adv Exp Med variable features of varicosity spacing along unmyelinated Biol 970:29–61. axons in the hippocampus and cerebellum. Proc Natl Acad Vlachos A, Reddy-Alla S, Papadopoulos T, Deller T, Betz H. Sci U S A 99:6340–5. 2013. Homeostatic regulation of gephyrin scaffolds and Siksou L, Triller A, Marty S. 2011. Ultrastructural organization synaptic strength at mature hippocampal GABAergic post- of presynaptic terminals. Curr Opin Neurobiol 21:261–8. synapses. Cereb Cortex 23:2700–11. Spangler SA, Schmitz SK, Kevenaar JT, de Graaff E, de Wit Waites CL, Leal-Ortiz SA, Okerlund N, Dalke H, Fejtova A, H, Demmers J, and others. 2013. Liprin-α2 promotes the Altrock WD, and others. 2013. Bassoon and Piccolo main- presynaptic recruitment and turnover of RIM1/CASK to tain synapse integrity by regulating protein ubiquitination facilitate synaptic transmission. J Cell Biol 201:915–28. and degradation. EMBO J 32:954–69.

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015 452 The Neuroscientist 20(5)

Waites CL, Specht CG, Härtel K, Leal-Ortiz S, Genoux D, Li of active zone protein RIM1 regulates synaptic vesicle D, Drisdel RC, and others. 2009. Synaptic SAP97 isoforms release. Cell 130:943–57. regulate AMPA receptor dynamics and access to presynap- Yasumatsu N, Matsuzaki M, Miyazaki T, Noguchi J, Kasai H. tic glutamate. J Neurosci 29:4332-45. 2008. Principles of long-term dynamics of dendritic spines. Westphal V, Rizzoli SO, Lauterbach MA, Kamin D, Jahn R, J Neurosci 28:13592–608. Hell SW. 2008. Video-rate far-field optical nanoscopy dis- Yudowski GA, Olsen O, Adesnik H, Marek KW, Bredt sects synaptic vesicle movement. Science 320:246–9. DS. 2013. Acute inactivation of PSD-95 destabilizes Wienisch M, Klingauf J. 2006. Vesicular proteins exocytosed AMPA receptors at hippocampal synapses. PLoS One and subsequently retrieved by compensatory endocytosis 8:e53965. are nonidentical. Nat Neurosci 9:1019–27. Zeidan A, Ziv NE. 2012. Neuroligin-1 loss is associated Yao I, Iida J, Nishimura W, Hata Y. 2003. Synaptic localiza- with reduced tenacity of excitatory synapses. PLoS One tion of SAPAP1, a synaptic membrane-associated protein. 7:e42314. Genes Cells 8:121-9. Zheng CY, Petralia RS, Wang YX, Kachar B, Wenthold RJ. Yao I, Takagi H, Ageta H, Kahyo T, Sato S, Hatanaka K, 2010. SAP102 is a highly mobile MAGUK in spines. J and others. 2007. SCRAPPER-dependent ubiquitination Neurosci 30:4757–66.

Downloaded from nro.sagepub.com at TECHNION 34965 IND on November 3, 2015