The Presynaptic Particle Web: Ultrastructure ... - Semantic Scholar

Neuron, Vol. 32, 63–77, October 11, 2001, Copyright 2001 by Cell Press

The Presynaptic Particle Web: Ultrastructure, Composition, Dissolution, and Reconstitution Greg R. Phillips,1 Jeffrey K. Huang,1 Yun Wang,5 Hidekazu Tanaka,6 Lawrence Shapiro,2 Wandong Zhang,1 Wei-Song Shan,1 Kirsten Arndt,1 Marcus Frank,1 Ronald E. Gordon,3 Mary Ann Gawinowicz,4 Yingming Zhao,7 and David R. Colman1,8 1 The Corinne Goldsmith Dickinson Center for Multiple Sclerosis Department of Neurology and The Fishberg Research Center for Neurobiology 2 Program in Structural Biology Department of Physiology and Biophysics 3 Department of Pathology The Mount Sinai School of Medicine One Gustave L. Levy Place New York, New York 10029 4 Protein Chemistry Core Facility Howard Hughes Medical Institute Columbia University New York, New York 10032 5 Kratos Analytical Inc. Chestnut Ridge, New York 10977 6 Department of Pharmacology Osaka University Medical School Osaka 565 Japan 7 Department of Biochemistry University of Texas Southwestern Medical Center Dallas, Texas 75390

Summary We report the purification of a presynaptic “particle web” consisting of ⵑ50 nm pyramidally shaped particles interconnected by ⵑ100 nm spaced fibrils. This is the “presynaptic grid” described in early EM studies. It is completely soluble above pH 8, but reconstitutes after dialysis against pH 6. Interestingly, reconstituted particles orient and bind PSDs asymmetrically. Mass spectrometry of purified web components reveals major proteins involved in the exocytosis of synaptic vesicles and in membrane retrieval. Our data support the idea that the CNS synaptic junction is organized by transmembrane adhesion molecules interlinked in the synaptic cleft, connected via their intracytoplasmic domains to the presynaptic web on one side and to the postsynaptic density on the other. The CNS synaptic junction may therefore be conceptualized as a complicated macromolecular scaffold that isostatically bridges two closely aligned plasma membranes. Introduction The concept of the synapse was originally framed in physiological terms, but it was also realized that there had to be a stable physical structure mediating the be8

Correspondence: [email protected]

havior of each synapse (Foster and Sherrington, 1897). Ultrastructural observations have since led to the current notion of a “characteristic” CNS synapse: pre- and postsynaptic membranes are rigorously parallel, in register with each other, and separated by a cleft of uniform distance (ⵑ25 nm, Peters et al., 1991). In orthogonal projection, electron dense “pegs” project into the synaptic cleft from thickenings of the pre-and postsynaptic membranes (Ichimura and Hashimoto, 1988), suggesting that the pre- and postsynaptic membrane linkage is mediated by molecules interacting within that space. These features of the CNS synaptic junctional complex are indicative of a strongly adhesive structure, and this conclusion is confirmed by cell fractionation experiments. Once formed, the pre- and postsynaptic membranes resist physical separation, such that even the strong shearing forces of nitrogen cavitation fail to separate them. The CNS synapse should therefore be thought of as a primary adhesive device—an adherens junction—upon whose scaffold has been superimposed, over the course of evolution, the molecular entities that mediate its physiology (Fannon and Colman, 1996; Colman, 1997). Both the adherens junction and the CNS synapse share similar features, such as an intercellular cleft of approximately the same distance, discrete concentrations of material, termed “densities,” held in register just below the cytoplasmic faces of the apposed membranes, and the presence of calcium-dependent cell-cell adhesion molecules—classic cadherins—which in all epithelia mediate strong, selective cell-cell adhesion (Colman, 1997; Shapiro and Colman, 1999). The CNS synapse, however, of course exhibits properties that reflect its complex neurophysiological functions. On the presynaptic side, proteins are coassembled that mediate synaptic vesicle clustering, exocytotic fusion, and membrane retrieval (Chen and Scheller, 2001; Slepnev and De Camilli, 2000), while the postsynaptic compartment of the junction harbors neurotransmitter receptors and their associated signaling components (Rao et al., 2000; Craig and Boudin, 2001). When viewed by electron microscopy, under optimal preparation conditions, differences between the preand postsynaptic densities are apparent. While the postsynaptic membrane exhibits a relatively continuous thickening below its membrane (the PSD, Peters et al., 1991), on the presynaptic side a prominent intermittent “gridwork” or “web” has been noted (Bloom and Aghajanian, 1968). This web was described as consisting of electron dense particles (ⵑ50 nm in diameter) arranged in a regular network just beneath the plasma membrane (Pfenninger et al., 1972), and it was speculated that these presynaptic particles participate in the alignment and docking of synaptic vesicles at the plasma membrane. More contemporary investigations have described a “presynaptic cytomatrix,” comprising fibrillar structures that originate at the presynaptic plasma membrane and extend deeply into the presynaptic compartment (Landis et al., 1988; Hirokawa et al., 1989; Garner et al., 2000)). The presynaptic cytomatrix also

Neuron 64

likely plays a role in tethering and guiding synaptic vesicles toward the presynaptic membrane. The PSD has been well characterized (Kennedy, 2000), and has generally been considered the primary organizer for the synaptic junction, an entity essentially independent of the presynaptic components of the synapse in terms of structural connections. However, the realization that pre- to postsynaptic adhesion is critical for the maintenance, stabilization, and physiology of the synapse has forced a reevaluation of this view. At the CNS synapse, pre- and postsynaptic membranes with their associated bona fide adhesion molecules align a dynamic transcellular structural “scaffold” whose elements penetrate both apposed membrane bilayers and protrude into the cytoplasms of the pre- and postsynaptic compartments (Tanaka et al., 2000). As at the neuromuscular junction (Hall and Sanes, 1993), the molecules necessary for inducing and processing neurotransmitter release are embedded within or linked to the scaffold, thus ensuring their positions in relation to the synaptic cleft. A major unanswered question is: how are the molecules which mediate physiological activity integrated with those that create the structural scaffold of the synaptic junctional complex? We have now experimentally determined conditions (pH 6) under which synaptic junctional complexes—with intact pre- to postsynaptic connections—can be isolated in high yield and high purity. Treatments that partially disrupt synaptic junction organization (1% Triton, pH 7–7.5) revealed ⵑ50 nm presynaptic particles connected to the PSD via filaments that are likely to participate, presumably indirectly, in cell-cell adhesion. Progressive elevation of the pH to 8 disrupts this interaction and solubilizes the presynaptic particles. Remarkably, subsequent lowering of pH to 6 via dialysis reconstitutes the particle web and permits its asymmetric reattachment to PSDs. Mass spectrometry of presynaptic web fractions demonstrates that the web is composed of a subset of proteins involved in vesicle recycling and membrane fusion. Thus, like the PSD, the presynaptic web should perhaps be recognized as an authentic “molecular machine,” integral to the complex functions of the presynaptic compartment. Our data also reveal the existence of a macromolecular scaffold which spans the synaptic cleft and holds the presynaptic web and PSD in strict register. Results A Stable Linkage between Pre- and Postsynaptic Specializations Is Maintained even after Membrane Solubilization Incompletely characterized adhesive linkages maintain the apposition of pre- and postsynaptic compartments in synaptosomal membrane preparations (Figure 1A) despite the high shear forces employed during homogenization of brain tissue. However, the presynaptic specialization is disintegrated by detergent treatment of synaptosomal membranes at pH 8, thus destroying the adhesive interactions and leaving only the PSD as an insoluble pellet (Figure 1B). This is an interesting observation in light of the demonstration that homophilic adhesive interactions (e.g., the classic cadherins) are oper-

ative between CNS synaptic membranes, and it indicates that the activation factors controlling transynaptic adhesion are likely to be different on pre- and postsynaptic sides. We first sought to identify conditions which might maintain pre- to postsynaptic attachment after selective membrane solubilization. Starting with synaptosomal membranes, we tested several extraction protocols, including a variety of detergents, salts, and pH titrations in order to find conditions under which we could isolate intact synaptic junctions from vesicular and other membranous components while preserving the adhesive linkage. Insoluble pellets derived from these experiments were examined by electron microscopy. We found that extraction with 1% TX-100 at pH 6 solubilizes almost all plasma membrane, but retains in the insoluble pellet paired pre- and postsynaptic structures, held in register by the 20–30 nm filaments or “pegs” that protrude into the cleft and which are observed in native synapses (Figure 1C; arrowheads in [F]). In addition, when viewed en face, perforated disks were frequently observed (Figure 1G) indicating that this morphology, when it occurs, is real, remarkably stable, and resistant to detergent extraction. pH 6 isolated synaptic junctions also exhibited a meshwork in both the pre- and postsynaptic densities that was absent in junctions treated at pH 8 (compare Figures 1F and 1G with 1D and 1E). pH 6-treated synaptic junctions exhibited few or no membrane attachments. These results reveal the existence of a stable, plasma membrane-independent proteinaceous scaffold which links and holds in register the pre- and postsynaptic densities. It is known that the classic cadherins participate in adhesive interactions across the synaptic cleft (Fannon and Colman, 1996; Uchida et al., 1996; Benson and Tanaka, 1998; Shapiro and Colman, 1999). We investigated whether cadherins were retained in isolated synapses obtained after treatment at pH 6. After immunogold-labeling of purified junctions with an N-cadherin antibody (Figure 2), we found label at or near the synaptic cleft, especially at sites where there was an apparent tight pre- to postsynaptic apposition (arrowheads). This supports the idea that cadherin-based adhesive interactions are integral components of synaptic junctional complexes (Fannon and Colman, 1996; Uchida et al., 1996). It should be noted that many junctions did not label with N-cadherin antiserum, as expected, since this cadherin is specific only to certain synaptic junctions (Fannon and Colman, 1996; Benson and Tanaka, 1998; Huntley and Benson, 1999), while different classic cadherins operate to mediate adhesion at other synapses in the CNS. In detergent-treated synaptosomal membranes incubated at increasing pH values from 6 to 8 in 0.5 unit increments, a stepwise disruption of the presynaptic specialization was observed. A relatively intact presynaptic specialization was observed at pH 6 (Figures 3A and 3E; arrowheads in [A]), while at pH 6.5, the presynaptic specialization was partially disrupted, revealing particles that appeared to be attached preferentially to the PSD (arrowheads in Figures 3B and 3F). Extraction at pH 7 more clearly revealed the array of 50 nm presynaptic particles (Figures 3C and 3G, arrowheads). At higher pH values (pH 7.5–8), the particles were completely solubilized, leaving only the PSD (Figure 3D, arrowheads). Furthermore, and of great interest, partial disruption of

Reconstitution of the CNS Presynaptic Web 65

Figure 1. The pH of TX-100 Solubilization of Synaptosomes Affects the Preservation of Synaptic Junctions in the Detergent-Insoluble Pellet Intact synaptosomes, showing vesicular and mitochondrial components and intact filamentous crossbridges (A) were treated with 1% TX-100, at pH 8 (B, D, and E) or pH 6 (C, F, and G), and the insoluble phase was pelleted, fixed, and processed for thin-section electron microscopy. PSDs were obtained when membranes were solubilized with 1.0% TX-100 buffered to pH 8 (B, D, and E) while paired electron dense profiles were observed when membranes were solubilized with 1% TX-100 at pH 6 (C, F, and G). High magnification images of single detergentextracted synapses are shown in panels (D)– (G). The synaptic material found in PSD preparations was elongated and retained the dimensions of PSDs observed in intact tissue, but the fine substructure was noticeably absent and appeared collapsed (D and E). In contrast, synaptic junctions observed in pH 6, TX-100 extracts were characterized by the presence of both pre- and postsynaptic densities that retained a file filamentous network of “loops” (F and G). In between the two densities, filamentous material appeared to connect the pre- and postsynaptic densities (arrowheads in [F]). Perforations (G) or disks were visible when synapses were viewed en face. Bar ⫽ 500 nm in (A)–(C) and 100 nm in (D)–(G).

the synaptic junction by extraction at an intermediate pH (pH 7) revealed filaments in the synaptic cleft that connected the presynaptic particles with the PSD (Figure 3H, arrowheads). Treatment of cortical tissue with ethanolic phosphotungstic acid (EPTA) revealed regularly spaced presynaptic particles as well as an intracleft dense line (see Figure 3I, and Pfenninger et al., 1972). Isolated junctions (TX-100, pH 6) treated with EPTA revealed that the PSD is in register with an array of presynaptic particles (Figures 3J and 3K) similar to what is observed in intact tissue, although EPTA-treated, detergent-isolated junctions did not display the intracleft dense line (Figure 3, compare I with J–K). Interestingly, in EPTA-treated junctions, extraction revealed fine filaments connecting presynaptic particles with the PSD (Figure 3L, arrowhead) remarkably similar to those observed in osmium-

treated, disrupted junctions (compare with Figure 3H, arrowheads). We therefore conclude that the particles are prominent, bona fide features of the presynaptic compartment where they underlie presynaptic membranes, appositional to the PSD. Both specializations are likely to be connected to the cytoplasmic domains of integral membrane proteins whose extracellular segments are bound to each other in the cleft. Together, this assemblage may be considered a cell-cell adhesive scaffold upon which is superimposed the functional elements of the CNS synapse. Separation and Identification of Preand Postsynaptic Components We showed above that treatment with TX-100 at pH 6 preserves the linkage of the presynaptic membrane specialization, including that of the presynaptic particles

Neuron 66

Figure 2. Immunogold Localization of N-Cadherin in pH 6 Detergent-Extracted Synaptic Junctions The insoluble pellet derived from extraction of synaptosomes with 1% TX-100 at pH 6.0 was fixed with paraformaldehyde and processed for thin-section electron microscopy without osmication. Antibodies to N-cadherin labeled the cleft of isolated junctions when the junctions were viewed from the side (arrowheads). N-cadherin antibody also labeled synapses sectioned en face (arrow). Bar ⫽ 500 nm.

to the PSD, while treatment with the same detergent at pH 8 disrupts the pre- to postsynaptic scaffold, solubilizing the presynaptic specialization and the particle web embedded in it, leaving only the PSD in the detergentinsoluble pellet. Therefore, we might expect that molecules involved in synaptic vesicle dynamics at the presynaptic membrane, if connected to the synaptic scaffold, would be retained in the detergent-insoluble pH 6 synaptic junctional pellet, but might be soluble at pH 8. Most postsynaptic proteins comprising the PSD would be predicted to be insoluble between pH 6–8 in TX-100. We might also expect that molecules loosely connected to the synaptic junctional scaffold might be solubilized in TX100 at pH 6. To evaluate these possibilities, we examined by immunoblot the soluble and insoluble fractions from intact, partially disrupted and completely disrupted synaptic junctions obtained by detergent extraction at increasing pH values (Figure 4). The representative markers were synaptophysin (vesicle); syntaxin (presynaptic membrane); Munc-18 and SNAP-25 (presynaptic membrane-associated); rim and bassoon (presynaptic cytomatrix); the glutamate receptor subunit NMDAR1 (postsynaptic); and the adhesion-related molecules N-cadherin, ␤-catenin, and the “extrasynaptic” neural cell adhesion molecule (N-CAM). Syntaxin, SNAP-25, and Munc-18 were present in the insoluble junctional fraction at low pH, and very little remained in this fraction at pH values above 7.5 (Figure 4, lanes 5–8), consistent with the localization of these proteins within the presynaptic membrane specialization, and possibly within the 50 nm presynaptic particles themselves. A significant proportion of the above proteins was solubilized at pH 6, however (Figure 4, lane 1), revealing synaptic and nonsynaptic pools of these molecules. Consistent with this, repeated extraction of isolated synaptic junctions with TX-100 (pH 6) did not solubilize the junction-associated pools of syntaxin,

SNAP-25, or Munc-18 (data not shown). In contrast to the differential TX-100 solubility of syntaxin, SNAP-25, and Munc-18 with respect to pH, the presynaptic cytomatrix components rim and bassoon were primarily insoluble at all pH values, although small amounts of both proteins were solubilized at pH 8–9 (Figure 4, lanes 5–8). These results indicate that these proteins either remain associated with PSDs via interactions with other molecules and under conditions in which there is no visible presynaptic element, or, they become detached from PSDs by detergent extraction, but are insoluble under the extraction conditions and are therefore pelleted (see Discussion). In contrast, the synaptic vesicle protein synaptophysin was completely solubilized even at pH 6, indicating that the synaptic vesicle membrane is more labile than the presynaptic membrane specialization (Figure 4, lanes 1–4). There were incremental transfers of N-cadherin and ␤-catenin into the soluble fraction when the pH was increased from 6.0 to 9.0 in the presence of 1% TX100 (Figure 4, lanes 1–4) as both molecules decreased proportionally from the insoluble fraction (Figure 4, lanes 5–8). At the highest pH values (9), ⵑ50% of N-cadherin and ␤-catenin remained in the pellet, revealing that these molecules are components of the residual PSD fraction, although it is clear that, as expected (Fannon and Colman, 1996; Tanaka et al., 2000), N-cadherin is also found on the presynaptic side. NMDAR1 was completely insoluble at both low and high pH (Figure 4, lanes 5–8), consistent with its tight association with the PSD. N-CAM was completely solubilized at both low and high pH (Figure 4, lanes 1–4), as might be expected considering its “extrasynaptic” disposition (Tanaka et al., 2000). The EM data revealed that presynaptic web components are isolated together with PSDs by detergent solubilization at pH 6, and are separated from PSDs by detergent at pH 8. We reasoned, therefore, that a frac-


tion highly enriched in presynaptic components should be obtained in a final supernatant by sequential detergent extraction of synaptosomes, first at pH 6, and then at pH 8 (Figure 5A). The removal of nonsynaptic proteins and the enrichment of synaptic proteins by sequential extraction from pH 6–8 was assessed by SDS-PAGE and silver staining (Figure 5B). Synaptosomal membranes (Figure 5B, lane 1) were treated with 1% TX-100 at pH 6 to yield an insoluble intact synaptic junctional pellet fraction (Figure 5B, lane 2) and a soluble fraction (Figure 5B, lane 3). The junctions were then treated with 1% TX100 at pH 8 and pelleted. A subset of proteins originally found in the pH 6 pellet fraction was solubilized and recovered in the pH 8.0 supernatant (Figure 5B, lane 4), while in the pH 8.0 pellet, a largely different set of protein bands was revealed (Figure 5B, lane 5). The major proteins enriched in the pH 8 supernatant (presynaptic fraction) migrated with Mrs ⵑ180, 100, 70, 60, 55, and 45k. The major proteins enriched in the pH 8.0 pellet (postsynaptic fraction) had Mrs of 190, 160, 110, 90, 80, 70, and 50k. Other proteins present equally in both fractions migrated at 270 and 250k. We compared the proteins of the pre- and postsynaptic fractions by mass spectrometry. The results of this analysis are shown in Figure 5D. Consistent with the expectation that the pH 8 supernatant should consist mostly of presynaptic protein components, we found a striking enrichment of proteins which have been implicated in either synaptic vesicle exocytosis or recycling, including the clathrin heavy chain (MW ⫽ 180 kDa), two alternatively spliced forms of dynamin (MW ⫽ 98 and 97 kDa), the clathrin-uncoating ATPase (Schlossman et al., 1984) hsc70 (MW ⫽ 70), and the syntaxin binding protein (Hata et al., 1993) Munc-18 (MW ⫽ 64). In addition, a peptide sequence matching a segment of an expressed DNA sequence tag (EST) in the mouse EST database (MW ⫽ 55; accession number AA000408) was identified. The translated EST nucleotide sequence was found to correspond to the carboxyl terminus of what must be a novel member of the septin family of GTPases that is associated with membranes belonging to the presynaptic compartment. As a group, the septins have been implicated in the establishment of diffusion barriers which delineate distinct membrane compartments between the mother and daughter cells in budding yeast (Adam et al., 2000; Barral et al., 2000; Takizawa et al., 2000). This novel septin family member displayed almost complete identity to human KIAA0128 and mouse septin 6 (accession numbers D50918 and AB023622, respectively) across the length of the EST sequence, but differed in sequence with known septins within the peptide region identified by mass spectrometry (not shown). Tentatively, we have termed this protein septin 6A. In contrast to the prominent representation of proteins implicated in synaptic vesicle function in the supernatant fraction, the pellet fraction contained many cytoskeletal or structural proteins, most of which have been shown previously to be abundant within the PSD (Kennedy, 1998; Kennedy et al., 1983; Walikonis et al., 2000). The PSD proteins identified in the pellet included spectrin, the myosin V heavy chain, neurofilament M, neurofilament L, and the CamKII ␣ subunit (Figure 5C). The synapsins 1A/1B, which are actin binding proteins involved in synaptic vesicle targeting to presynaptic

membranes (reviewed in Hilfiker et al., 1999), were also present in the pellet fraction. Four other proteins found in these fractions, ATP synthase, hexokinase I (brain form), fumarate hydratase, and a putative motor protein (accession number NM_006839), may be derived from mitochondria which are found in both pre- and postsynaptic compartments (see Figure 1A), although the possibility exists that they are exported to synaptic sites (Soltys and Gupta, 1999). Of note, it has been reported that hexokinase I displays fast axonal transport that is inconsistent with a mitochondrial localization and so this molecule has been suggested to be directly associated with presynaptic terminals as well (Garner et al., 1996). It should be noted that in both fractions, spectrin ␣ and ␤ subunits were prominently represented, indicating that spectrin is likely to play structural roles on both sides of the synaptic junction. The Presynaptic Web Self-Assembles Evidence suggests that certain axon terminals, when appropriately triggered, can rapidly assemble a functional presynaptic membrane with associated vesicle clusters, suggesting that the presynaptic compartment exhibits a remarkable tendency to rapidly self-assemble (Ahmari et al., 2000; Scheiffele et al., 2000; Zhai et al., 2001). These data suggested to us that perhaps the presynaptic particles, once solubilized (in TX-100 at pH 8), might possibly be induced to re-form upon return to conditions that maintain their assembly (pH 6). Accordingly, after disruption of isolated junctions (pH 8), and removal of PSDs by pelleting, the supernatant was slowly dialyzed against pH 6. Remarkably, after dialysis, 50 nm particles were re-formed in the insoluble phase (Figure 6A). The reconstituted particles were interlinked by fine filaments that appeared to be similar to those observed linking the presynaptic web in synapses in situ (Pfenninger et al., 1972), and these also appeared to preserve the native spacing between particles. This morphology was reconstituted only by slow dialysis; a rapid shift from pH 8 to pH 6 yielded amorphous insoluble material (Figure 6B). From these results, we conclude that proteins comprising the presynaptic web, when isolated and then allowed to slowly coassociate, are able to reassemble structures that correspond to the presynaptic particles in native synapses. We were curious whether these reconstituted presynaptic particles might be capable of reattaching to PSDs in vitro. We disrupted pH 6-isolated junctions by solubilization at pH 8 and then reconstituted the presynaptic particles by dialysis (pH 6). We found that the 50 nm particles were able to reattach to the PSD (Figure 6C). This association appeared to be specific because the particles were always found to be asymmetrically arrayed along one side of the PSD. In contrast to the specific reassociation of presynaptic particles with the PSDs, when disrupted junctions were dialyzed against pH 6 buffer, no attachment of any material was observed when the dialysate was maintained at pH 8 (Figure 6D). The reconstituted particles shown in Figure 6E (right panel) closely resemble those found in native synapses from cortical tissue (Figure 6E, left and middle panels). In all cases, whether viewed in tissue from the side or

Neuron 68

Figure 3. Presynaptic Particles and Their Tethers to the PSD Are Revealed upon Partial Disruption of Isolated Synaptic Junctions or by Staining Intact Junctions with EPTA Presynaptic particles are obscured by other elements of the presynaptic specialization in pH 6 extracted synaptic junctions (A and E; arrowheads in [A] show intact presynaptic specialization), but are progressively revealed after treatment of synaptosomes with TX-100 at pH 6.5 (B and F, arrowheads), and more so at pH 7 (C and G, arrowheads). Extraction at pH 8 completely removes all presynaptic elements leaving only the PSD in the insoluble pellet (D, arrowheads). Upon extraction at pH 7, sufficient material is removed from pre- and postsynaptic densities to reveal the presence of tethers linking the particles to the PSDs (arrowheads in [H]). Arrows in (F) and (G) indicate subsynaptic “bodies.” EPTA treatment of intact isolated junctions (1% TX-100 at pH 6) reveals presynaptic particles remarkably similar to those seen by partial disruption at mildly alkaline pH (J–K). EPTA-stained particles in isolated junctions (J–K) were similar to those found in tissue (I). At high magnification, tethers linking particles to PSDs were observed in EPTA-treated isolated junctions (arrowheads, L) that were identical to those observed in partially disrupted osmium-treated junctions (arrowheads, H). Bar ⫽ 500 nm in (A)–(D), 250 nm in (E)–(G) and (I)–(K), and 100 nm in (H) and (L).


Figure 4. Differential Extraction of Synaptic Components Synaptosomes were extracted with 1% TX-100 at the indicated pH, and equal volumes of soluble and insoluble material were electrophoresed, transferred to nitrocellulose, and probed with antibodies to the indicated proteins. Presynaptic markers syntaxin, Munc-18, and SNAP-25 were present in the pellet at the lowest pH values but were completely extracted at pH values above 7, suggesting that these molecules are part of the presynaptic specialization. Repeated extraction of the pH 6 pellet with TX-100 at pH 6.0 never solubilized the insoluble pool of syntaxin, Munc-18, or SNAP-25 (data not shown). Presynaptic cytomatrix proteins rim and bassoon were insoluble at all pH values between 6 and 9, indicating that these proteins remain associated with the PSD under conditions in which there is an absence of a visible presynaptic element. Increases in the pH of extraction from 6.0 to 9.0, which gradually removed the presynaptic specialization, caused gradual increases in the solubility of N-cadherin and ␤-catenin, consistent with the localization of these proteins on both pre- and postsynaptic membranes. Vesicle proteins such as synaptophysin and extrasynaptic proteins such as N-CAM were completely extractable at all pH values between 6 and 9 (soluble fractions, lanes 1–4), suggesting that the structures in which these two proteins are found are not tightly linked to the synaptic scaffold. The NMDA receptor subunit R1 was always found in the insoluble fraction at both low and high pH consistent with its tight linkage to the PSD.

en face (Figure 6E, left and middle panels, respectively) or in the reconstituted particle fraction (Figure 6E, right panels), the particles were ⵑ50 nm in size (brackets in Figure 6E). We next investigated which of the proteins originally found in the presynaptic extract (see Figure 4, lane 4) are reconstituted as particles upon dialysis. Supernatants and pellets obtained from the reconstitution reaction in the absence of PSDs were electrophoresed and proteins were detected by silver staining (Figure 7A) or by immunoblotting (Figure 7C). Most of the proteins originally found in the pH 8 supernatant (compare with Figure 5B) were found in the insoluble pellet upon dialysis at pH 6 (Figure 7A, lane 2) and were largely depleted from the supernatant, except for a single protein band that was still prominently represented in the supernatant after dialysis (Figure 7A, lane 1, asterisk). Mass spectrometry revealed this to be hsc70 (Figure 7A, lane 1, asterisk, see also Figure 5). Most of the proteins found originally in the presynaptic extract were found to reconstitute into the 50 nm particles, but not, however, septin 6a, which appeared to have been degraded in processing. Of the presynaptic cytomatrix components, immunoreactivity for both rim and bassoon was observed exclusively in the reconstituted pellet, although degradation of these two large protease-sensitive proteins was also noted (Figure 7C). To evaluate whether any of the adhesion molecule groups known to be present at the synapse (Fannon and Colman, 1996; Song et al., 1999; Rao et al., 2000) might be involved in the attachment of the presynaptic particles to the PSD, we looked by immunoblot for the presence of adhesion-related molecules in the presynaptic particle fraction (Figure 7C). We found that N-cadherin, ␤-catenin, and a ␥-protocadherin, as well as CASK (proposed to mediate the cytoskeletal interactions of the neurexins and neuroligins) were all present in the particle fraction (Figure 7C, pellet). Multiple adhesive mechanisms probably play roles in the attachment of the particles to the PSD. This conclusion is supported by the fact that we observed some reattachment of particles to PSDs in the presence of EDTA (not shown), which unequivocally inactivates cadherin-mediated interactions. To verify the specificity of the reconstitution reaction, we determined whether reconstituted presynaptic particle components could be resolubilized and then subsequently reconstituted (Figure 7A, lanes 3 and 4). All major proteins found in the first reconstituted pellet (Figure 7A, lane 2) were also found in the second reconstituted pellet (Figure 7A, lane 4, arrows and arrowheads). Some additional minor contaminants were apparently removed during the second reconstitution reaction resulting in a “sharpening” of bands in the second pellet. This second purification of the particle fraction enabled the identification by mass spectrometry of additional protein constituents (Figure 7A, lane 4, arrows). These proteins included the N-ethyl-maleimide-sensitive factor (NSF), and the vacuolar ATP synthase (v-ATPase) subunit A as well as the cytomatrix protein synapsin 1b. Both NSF and v-ATPase play well-characterized roles in mediating vesicle fusion and recycling, respectively (Zinsmaier and Bronk, 2001; Forgac, 2000). The synapsin family is associated with the presynaptic cytomatrix and participates in the localization of synaptic vesicles and

Neuron 70

Figure 5. Segregation of Synaptic Proteins into Two Fractions Based on Their Differential Solubility (A) Schematic representation of the sequential extraction. (B) Silver-stained SDS-PAGE of fractions derived from the sequential extraction of synaptosomes. Synaptosomal membranes (lane 1) were solubilized with 1% TX-100 at pH 6 yielding an insoluble junctional pellet (pH 6 pellet, lane 2) and soluble nonsynaptic supernatant (pH 6 supernatant, lane 3). The junctional pellet was re-extracted in 1% TX-100 at pH 8.0. The supernatant (pH 8 supernatant, lane 4) and pellet (pH 8 pellet, lane 5) fractions contained different sets of proteins that were originally found in the intact junction pellet. Arrow indicates a nonsynaptic protein that was removed by solubilization at pH 6, arrowhead indicates a protein enriched in the pH 8 supernatant, and the asterisk indicates a protein enriched in the pH 8 pellet. (C) Fifty micrograms of the pH 8 soluble (lane 1) or insoluble (lane 2) fractions was electrophoresed on SDS-PAGE and stained with colloidal Coomassie blue. The bands identified by mass spectrometry are indicated by numbered arrows and their identity is shown in (D).

is probably not a component of the presynaptic web (see Discussion). The components that we identified in the present study as components of the presynaptic particle web are listed in Table 1. From these studies, it is clear that the presynaptic web forms a subset of interrelated proteins embedded in the presynaptic membrane specialization and it is composed of synaptic vesicle exocytosis and recycling proteins. Discussion In early EM studies of CNS tissue (Bloom and Aghajanian, 1968; Pfenninger et al., 1972), a grid or web-like

structure had been observed, consisting of ⵑ50–80 nm particles arrayed below the presynaptic membrane and connected by fine fibrils spaced ⵑ100 nm apart. We now show that these components can be isolated and reconstituted, and appear to be an important part of the cell-cell synaptic junctional scaffold. In our preparations derived from tissue fractionation, the particles, the spaces between particles, and the fibrillar network exhibit the uniform dimensions observed in situ. The 50 nm particles are generally separated by a distance of approximately 50–100 nm by fibrils of approximately 10 nm in width. Further support for the existence of the CNS presynaptic web comes from recent observations


Figure 6. Solubilized Presynaptic Particles Independently Reconstitute Their Structure and Reattachments to PSDs (A) Fifty nanometer presynaptic particles were formed by dialysis of the soluble presynaptic fraction against pH 6 buffer in the absence of PSDs. Reconstituted particles were regular in size and were connected by fine filaments (arrowhead). (B) A precipitate with no resemblance to presynaptic particles was formed when the presynaptic fraction was rapidly diluted 20-fold with 1% TX-100, 20 mM Tris pH 6.0, 1 mM CaCl2. (C) Reattachment of presynaptic particles to PSDs. The solubilized presynaptic fraction was separated from the PSD pellet. PSDs were resuspended in an equal volume of the same buffer. One twentieth of the original volume of PSDs was added back to the presynaptic fraction and the suspension was dialyzed against 1% TX-100, 20 mM Tris, pH 6.0, 1 mM CaCl2 (C), or against 1% TX-100, 20 mM Tris, pH 8.0, 1mM CaCl2 (D). In the pellet obtained from the pH 6 dialyzate, numerous PSDs were found to have 50 nm particles attached by visible filaments to only one side of the PSD (arrowheads, C). In contrast, the pellet from the pH 8 dialyzate contained mostly PSDs with no additional attachments (D). (E) Presynaptic particles in tissues viewed either from the side (left) or en face (middle) exhibit the same dimensions as reconstituted particles (right). Brackets in (E) denote a distance of 50 ␮m. Bar ⫽ 500 nm in (A) and (B) and 350 nm in (C) and (D).

of a very similar structure within motor terminals at the frog neuromusclar junction (Harlow et al., 2001). We conclude that the particle web at the presynaptic membrane is a real structure, intimately involved in synaptic vesicle localization at the presynaptic membrane. Our studies reveal a sensitivity of presynaptic structure to relatively small changes in pH, and demonstrate

that presynaptic particles can assemble and disassemble under defined conditions. These data have important implications for interpretation of work from other laboratories that argues for a remarkable propensity for spontaneous self-assembly of presynaptic components on suitable substrates. It was first shown that presynaptic structures are generated de novo in cultures when neurites

Neuron 72

Figure 7. Identification of Presynaptic Particle Components by Mass Spectrometry (A) Proteins from presynaptic particles reconstituted one time (lane 2) or two times (lane 4), and their supernatants (lanes 1 and 3) were electrophoresed and silver stained. Upon the first reconstitution, most of the proteins originally found in the presynaptic fraction were found in the pellet upon dialysis against pH 6 buffer (lane 2), although there was an enrichment of one protein band in the supernatant (asterisk, lane 1). The proteins in the first reconstituted pellet were resolubilized in 1% TX-100 at pH 8 and dialyzed again against pH 6 buffer. The same proteins were obtained in the pellet from the second reconstitution (lane 4) reaction as in the first (compare lane 4 with lane 2), although an enrichment or clarification of the bands was observed. The bands identified previously by mass spectrometry in the presynaptic supernatant are indicated by the arrowheads in lane 4. Mass spectrometry protein identification of the enriched bands in the second reconstitution reaction was performed (arrows). The indicated proteins in lane 4 are listed in (B). (C) Western blot of equal volumes of supernatants (supe) and pellets (pel) obtained after one reconstitution of the presynaptic extract.

from these cells came in contact with polyanionic-coated latex beads (Burry, 1980). By EM, it was demonstrated that these presynaptic organelles contained appropriate vesicle clusters, a presynaptic thickening, and a closely adherent presynaptic membrane engaging the bead surface. Most recently, Scheiffele et al. (2000) showed that granule cells will respond to exogenously expressed neuroligin on a cell surface by forming a similar presynaptic structure which can secrete in response to an appropriate stimulus (Scheiffele et al., 2000). Thus, it seems clear that those neurites destined to form presynaptic structures accumulate soluble constituents, and with the appropriate molecular trigger, can rapidly form a functional presynaptic compartment. Our data now show that constituents of the presynaptic particle web

can assemble in vitro to form structures which closely resemble presynaptic particles observed under ultrastructural examination of CNS tissue. The self-assembly of the presynaptic particle web, and possibly other presynaptic structures, is reminiscent of the rapid, protein synthesis-independent assembly of junctional complexes in epithelia. The notion that the presynaptic grid might participate in the alignment and docking of synaptic vesicles at the presynaptic membrane prior to their fusion was presented on the basis of EM alone (Pfenninger et al., 1972). This idea was based on the fact that the spaces between presynaptic particles, possibly maintained via the action of the connecting fibrils, are very similar to the size of synaptic vesicles. In freeze-etched preparations, it was

Table 1. Major Components of Presynaptic Particles Proteins by MS or Immunoblot

Mr

Function

References

Spectrin ␣ and ␤ Clathrin heavy chain Dynamin NSF V-ATP synthase subunit A Munc-18 Tubulin Actin N-cadherin Protocadherin-␥ CASK ␤-catenin

220–200 180 97–98 73 70 64 55 45 130 130 100 95

Deformable fibril network Coated vesicle formation Vesicle endocytosis Vesicle fusion complex Vesicle neurotransmitter loading Inhibition or promotion of vesicle fusion Cytoskeletal Cytoskeletal Adhesion Adhesion Adhesion-related Adhesion-related; link to cytoskeleton

Hainfield and Steck, 1977 Cremona and De Camilli, 1997 Cremona and De Camilli, 1997 Rothman, 1994; Scheller, 1995; Sudhof, 1995 Forgac, 2000 Aravamudan et al., 1999 Allison et al., 2000 Morales et al., 2000 Fannon and Colman, 1996; Uchida et al., 1996 Kohmura et al., 1998 Biederer and Sudhof, 2000 Fannon and Colman, 1996; Uchida et al., 1996


Figure 8. Model of the Presynaptic Web and Its Relationship with the Presynaptic Cytomatrix (A) A model for the organization of the presynaptic specialization and its attachment to the PSD is shown. Fibrillar components originating at the presynaptic membrane and projecting into the cytoplasm have been described in freeze-etch electron microscopy studies (Hirokawa et al., 1989; Landis et al., 1988). These fibrils may contain the presynaptic cytomatrix proteins (Garner et al., 2000). Shown superimposed upon the fibrils are the presynaptic particles which exhibit a visible attachment to the PSD (solid line, right panel). Under conditions in which presynaptic particles are stripped from the PSD (1% TX-100, pH 8) and hence solubilized, presynaptic cytomatrix components remain in the insoluble pellet along with the PSD, suggesting the possibility of a distinct adhesive system linking the cytomatrix with the PSD (dotted lines in right panel). (B) Three-dimensional model of the synaptic scaffold. Presynaptic particles are linked by adhesion molecules across the synaptic cleft to the PSD. The particles form an hexagonal array that may support the localization of synaptic vesicles to the membrane and their subsequent fusion. Presynaptic particles also contain components necessary for the retrieval of vesicle membrane proteins after their fusion with the plasma membrane. The particles may serve to immobilize endocytosis components at high concentrations next to the presynaptic membrane such that they might be readily available after vesicle fusion.

shown that within the spaces between particles, there are depressions in the presynaptic membrane which appear to be sites of vesicle attachment (Pfenninger et al., 1972). Upon synaptic stimulation, the distance between particles transiently increases, suggesting an insertion of material between the particles as a consequence of fusion of the synaptic vesicle with the plasma membrane (Triller and Korn, 1985). Thus the web appears to provide a framework to allow the docking and fusion of synaptic vesicles at the presynaptic membrane. However, it is not known how the protein components of the presynaptic web identified in the present study actively participate in the synaptic vesicle cycle. Our studies suggest that one important role of the presynaptic web may be in the sequestration of vesicle recycling proteins such as clathrin and dynamin in circumscribed presynaptic microdomains. Given that clathrin and dynamin exhibit the tendency to selfassemble endocytotic structures around naked lipid vesicles (Takei et al., 1998), a sequestering mechanism may be needed to prevent the precocious self-assembly of these molecules. The regular array of particles of defined size, interconnected by short fibrillar components, observed at the presynaptic membrane is highly reminiscent of the spectrin-based web underlying the erythrocyte plasma membrane (Hainfeld and Steck, 1977). Indeed, the presence of spectrin as a prominent component in our reconstituted presynaptic web fraction suggests the spacing between particles could be maintained by a flexible spectrin-based filament network and this structural framework might explain the dynamic changes in spacing of the presynaptic particles upon synaptic stimula-

tion. In agreement with this analogy, the spacing between web elements at the presynaptic and erythrocyte membranes appears to be consistent (50–100 nm). The presynaptic web may also exhibit analogies to another spectrin-based structure—the terminal web of intestinal epithelial cells (Hull and Staehelin, 1979). This structure contains a dense spectrin network that overlays a complex of intermediate filaments. This arrangement helps to keep actin bundles orthogonal to the membrane surface. Thus, the presynaptic web may participate in maintaining presynaptic components in the proper orientation orthogonal to the synaptic plane. In several other respects, the presynaptic web identified here resembles the “active zone material” (AZM) found at the presynaptic plasma membrane of the neuromuscular junction. Using electron microscope tomography, Harlow, et al. (2001) have resolved the AZM into individual physical components—“ribs,” “beams,” and “pegs”—that form a complex presynaptic framework whose intracellular position and structure suggest that its components function in docking and vesicle fusion. It is postulated that components of the AZM are linked to transmembrane molecules embedded in the synaptic membrane similar to the observed linkage for the presynaptic web to PSDs in our study. An Asymmetric Transcellular Synaptic Scaffold The central nervous system synapse, while optimized over the course of evolution for polarized cell-cell communication, nevertheless retains many of the properties of the adherens junction from which it appears to be derived (Colman, 1997; Shapiro and Colman, 1999). The adherens junction adhesively links two apposing parallel

Neuron 74

plasma membranes and locks them into a transcellular scaffold, half of which is contributed by each participating cell. The scaffold consists of transmembrane adhesion molecules that interlock in the intercellular cleft with cytoplasmic domains that protrude into the intracellular milieu. Anchored to these cytoplasmic domains in each cell are signaling molecules and cytoskeletal elements that organize and cluster beneath the junctional membrane plates (Aberle et al., 1996). In terms of adhesive forces, the adherens junction is symmetrical and is held together by homophilic cadherin-mediated interactions. On the other hand, while cadherin-based adhesive interactions are undoubtedly important for the maintenance of the rigorously parallel synaptic cleft, their approximately equal partitioning and differential solubility in both pre- and postsynaptic fractions suggests that these molecules may have differential roles in organizing or aligning synaptic elements on both the pre- and postsynaptic side (Tanaka et al., 2000). On the postsynaptic side, our study suggests that cadherin molecules are firmly anchored to the PSD via interactions of their cytoplasmic domains either directly or indirectly with PSD structural proteins. In contrast to the tight anchorage of postsynaptic cadherins, the presynaptic cadherins are more easily released from the junction by detergent, are more dynamic, and thus may subserve other roles. It is interesting that we have noted that the classic cadherins possess within their cytoplasmic domains two tyrosinebased consensus sorting sequences (YXXø; ø ⫽ large aliphatic) for the AP-1 adaptor proteins (Vincent et al., 1997) involved in clathrin-mediated endocytosis. Thus it is possible that presynaptic cadherins may have a role in organizing or localizing the presynaptic endocytosis complex which may correspond to the presynaptic particles (Figure 8B). The machinery that guides synaptic vesicles to their appropriate location within the presynaptic compartment prior to docking and fusion (the presynaptic cytomatrix; Garner et al., 2000) appears to be linked to the synaptic junction by a different mechanism than that of the presynaptic particles (Figure 8A). Prior to docking, synaptic vesicles are aligned at their position within the presynaptic compartment by structural cytomatrix elements which involve proteins such as the synapsins, bassoon, and piccolo (Cases-Langhoff et al., 1996; Hilfiker et al., 1999; tom Dieck et al., 1998). These molecules appear to be very tightly linked to the PSD across the synaptic cleft, as evidenced by their presence in the most detergent-resistant fraction along with the PSD— conditions where no presynaptic elements are detected by electron microscopy (see Figures 3 and 4). These results are consistent with the finding that some of the cytomatrix proteins are highly enriched in PSD fractions (tom Dieck et al., 1998; Wang et al., 1999). Alternatively, the presence of these presynaptic vesicle targeting molecules in the PSD fraction may simply reflect their relative insolubility when compared to the rest of the presynaptic machinery and not their specific linkage to the PSD. Regardless of their linkage mechanisms to the PSD, the differential solubility of the presynaptic cytomatrix and the presynaptic web elements suggests that these are two distinctly organized, yet interrelated, substructures within the presynaptic compartment (Figure 8A).

The presynaptic cytomatrix most likely emanates into the cytoplasm from the presynaptic membrane as fibrils that extend well into the presynaptic compartment as has been observed in freeze-etched preparations (Hirokawa et al., 1989; Landis et al., 1988). These fibrils might tether or guide synaptic vesicles via the action of the synapsins. While clearly distinct from the presynaptic particles, the fact that the cytomatrix proteins rim and bassoon are found in the reconstituted presynaptic particle fraction (see Figure 7) suggests that these proteins may interact with proteins found in the particles. It is intriguing to speculate that the cytomatrix fibers, firmly anchored to the presynaptic membrane, provide nucleation sites for the components of the presynaptic particles (Figure 8). The presynaptic particle proteins would then be available at high concentrations directly underneath the presynaptic membrane. In conclusion, the results of this study shed light on the organization of the synaptic junctional complex and expand the framework upon which to address the interrelationships of its protein components. Experimental Procedures Antibodies The N-cadherin antiserum has been described (Tanaka et al., 2000). A polyclonal antibody to ␥-protocadherins, corresponding to the conserved portion of the cytoplasmic domain (Wu and Maniatis, 1999), was raised in rabbits. Some antibodies (synaptophysin, synaptotagmin, Munc-18, SNAP-25, rim, and ␤-catenin) were from Transduction Laboratories (Lexington, KY). NMDA receptor R1 and N-CAM antibodies were obtained from Pharmingen (San Diego, CA). Bassoon antisera were obtained from Stressgen Biotechnologies (Victoria, BC). All other antisera were from Sigma (St. Louis, MO). Synaptosome Preparation Synaptosomes were prepared using a one-step synaptosome preparation method based on the known isopycnic densities of various cellular components (Cohen et al., 1977). Cortices from five adult male rats (5–6 g total wet weight) were homogenized in 15 ml of solution (0.32 M sucrose, 0.1 mM CaCl2, 1 mM MgCl2, 0.1 mM PMSF) at 4⬚C using a Teflon-glass homogenizer. The homogenate was brought to a final sucrose concentration of 1.25 M by the addition of 2 M sucrose (70 ml) and 30 ml 0.1 mM CaCl2 and divided into six 25 ⫻ 89 mm ultracentrifuge tubes. The homogenate was overlaid with 10 ml 1.0 M sucrose, 0.1 mM CaCl2, and with 5 ml homogenization solution and centrifuged (100,000 ⫻ g; 3 hr; 4⬚C). After centrifugation, a band representing synaptosomal membranes was collected at the 1.25 M/1.0 M sucrose interface. Solubility Analysis of Synapses and Synaptic Proteins Synaptosomes were diluted 1:10 with ice cold 0.1 mM CaCl2 and divided into 100 ␮l aliquots. An equal volume of 2⫻ solubilization buffer was added (2% TX-100, 0.2 mM CaCl2, 40 mM Tris buffered to various pH values; see Results); the samples were then mixed and incubated (20 min on ice). The samples were then centrifuged at 10,000 ⫻ g for 30 min at 4⬚C. Supernatants were collected and pellets resuspended in an equal volume (200 ␮l) of 0.1 mM CaCl2. An equal volume of 2⫻ SDS-PAGE sample buffer was added and the samples were loaded onto 7.5% SDS gels. Gels were transferred to nitrocellulose and the membranes were probed with antibodies. Electron Microscopy For EM, rats were anesthetized and perfused transcardially (4% paraformaldehyde, 2% glutaraldehyde; Hank’s solution with 15 mM Hepes buffer, pH 7.0; 1 mM CaCl2). Brains were removed and postfixed. Cortical sections were treated with 2% OsO4, 2 hr, and stained en bloc with uranyl acetate, dehydrated, and embedded in Epon (Embed 812, Electron Microscopy Sciences). Alternatively, cortical sections were treated with 1% phosphotungstic acid in absolute


ethanol containing a trace of water (2 drops of 95% ethanol per 10 ml) for 2 hr at 50⬚C. The tissue was then rinsed in ethanol and Epon-embedded. For isolated synaptic junctions, pellets were fixed with 4% glutaraldehyde; phosphate buffer-pH 7.4. For standard EM, pellets were treated with 2% osmium tetroxide in acetate buffer, stained en bloc with uranyl acetate, dehydrated through ethanols, and embedded in Epon. EPTA treatment of isolated junctions was done as above for rat cortex. Ultra-thin sections were cut and, for osmicated specimens, stained with uranyl acetate-lead citrate and viewed in an electron microscope at 8,300–50,000⫻ magnification. Immunogold labeling was performed essentially as described (Phend, et al., 1995). Separation of Pre- and Postsynaptic Proteins Synaptosomes were extracted sequentially with 1% TX-100 at pH 6.0 followed by pH 8.0. Synaptosomes (4 ml, prepared as described above) were diluted 1:10 with ice cold 0.1 mM CaCl2. The suspension was brought to a final concentration of 20 mM Tris, pH 6.0 and 1% TX-100. The membranes were extracted (30 min) and the insoluble material was pelleted (40,000 ⫻ g, 30 min). The supernatant was decanted and proteins precipitated with 10 volumes acetone at ⫺20⬚C and recovered by centrifugation at 15,000 ⫻ g for 30 min. The pellet was resuspended in 10 ml 20 mM Tris, pH 6.0 and 1% TX100; a small aliquot was taken for gel electrophoresis, reextracted, precipitated, and pelleted as above. The insoluble pellet was resuspended (10 ml of 20 mM Tris, pH 8.0, 1% TX-100), extracted, centrifuged, and the supernatant reprecipitated as above. The insoluble pellets as well as the recovered supernatants from all three extractions were dissolved in 5% SDS and the protein concentrations were determined using the BCA protein assay (Pierce; Rockford, IL). Reconstitution of Presynaptic Particles from Solubilized Fractions The presynaptic fraction was prepared as above and subsequently dialyzed against 1% TX-100, 20 mM Tris, pH 6.0, 0.1 mM CaCl2 for three changes at 10 hr each. Control experiments included dialysis for the same time but at pH 8.0. The dialysates were centrifuged at 15,000 ⫻ g and equal volumes of the supernatant and pellets were electrophoresed on SDS-PAGE and silver-stained, Coomassie stained, or transferred to membranes for Western blotting. Pellets were fixed and processed as above for EM. In some cases, purified PSDs were added to the reconstitution reaction. The PSD pellet was resuspended in 1%TX-100, 20 mM Tris, pH 8, 0.1 mM CaCl2, in a volume equal to that of the supernatant. One twentieth of the original volume of PSDs was added back to the presynaptic supernatant prior to dialysis. MS/MS Analysis Twenty to fifty micrograms of each protein sample (see above) was loaded onto a 16 ⫻ 24 cm 7.5% SDS-PAGE gel and electrophoresed at 50V overnight. Proteins were visualized by staining with silver or with colloidal Coomassie blue (Novex; San Diego, CA). Gel slices containing the bands of interest were excised from the Coomassiestained gel, destained in 25 mM NH4HCO3 buffer (pH 8.0; methanol:water, 1:1), cleaned in a fixation solution (water:acetonitrile:acetic acid, 45:45:10) for 20 hr, swollen with water for 2 hr, and then equilibrated with digestion buffer (50 mM NH4HC03, pH 8.0) for 4 min. Tryptic digestion was carried out at 37⬚C for 2 hr in 10 to 20 ␮l of digestion buffer including 0.2 ␮g of sequencing grade modified trypsin (Boehringer Mannheim, Indianapolis, IN). The tryptic peptides were extracted twice with 70% acetonitrile in water and concentrated in a speed vac concentrator (Savant; Holbrook, NY). The dried sample was dissolved in 6 ␮l of HPLC solution A (water:acetonitrile:acetic acid, 97.5:2:0.5) for mass analysis. HPLC-MS/MS analysis was performed in a LCQ (Finnigan MAT) coupled online with a capillary HPLC system (Microm BioResources). Two microliters of the peptide solution was loaded on capillary HPLC connected with a C18 column (10 cm length, 75 mm internal diameter). The peptides were sequentially eluted from the HPLC column with a gradient of 5%–90% of HPLC solution B (acetonitrile:water:acetic acid, 2:97.5:0.5) at a flow rate of 0.7 ml/min and were sprayed directly from the tip of the capillary column to a LCQ mass spectrometer for MS/MS analysis. LCQ was operated in a

data-dependent mode where the machine measured the intensity of all peptide ions in the mass range of 400 to 1400 (mass to charge ratio) and isolated the peptide peak with the highest intensity for collision-induced dissociation. Thus, the masses of both the peptide and its daughter ions were detected. The accurately measured masses of the tryptic peptide and its fragments were used to search for protein candidates in the protein sequence database with the program “PepFrag” (Fenyo et al., 1998) (http://prowl1.rockefeller. edu/prowl/pepfragch.html). MALDI-MS Analysis Gel bands were prepared for digestion by washing twice with 200 ␮l 0.05 M Tris, pH 8.5/50% acetonitrile for 20 min with shaking. After removing the washes, the gel pieces were dried for 30 min in a Speed-Vac concentrator. Gels were digested by adding 0.05 ␮g modified trypsin in 13–15 ␮l of 0.025 M Tris, pH 8.5. The tubes were placed in a heating block at 32⬚C and left overnight. Peptides were extracted twice with 50 ␮l 50% acetonitrile/2% TFA and the combined extracts were dried and resuspended in matrix solution. Matrix solution was prepared by making a 10 mg/ml solution of 4-hydroxy␣-cyanocinnamic acid (Sigma; St. Louis, MO) in 50% acetonitrile/ 0.1% TFA and adding two internal standards, angiotensin and bovine insulin, to the matrix solution. The dried digest was dissolved in 3 ␮l matrix/standard solution and 0.7 ␮l was spotted onto the sample plate. When the spot was completely dried, it was washed twice with water to remove buffer salts. MALDI mass spectrometric analysis was performed on the digest using a PerSeptive Voyager DE-RP mass spectrometer in the linear mode. Peptide masses were used to search the protein database with the program ProFound (http://129.85.10.192/profound_bin/WebProFound.exe). Acknowledgments This work was supported by grants from the NIH (NS20147 and NS041687) and the NY State Spinal Cord Injury Research Program (C016883) to D.R.C. G.R.P. was supported by an NRSA (NS10836) from the NIH and W.Z. by NS37731. Received March 19, 2001; revised July 20, 2001. References Aberle, H., Schwartz, H., and Kemler, R. (1996). Cadherin-catenin complex: protein interactions and their implications for cadherin function. J. Cell. Biochem. 61, 514–523. Adam, J.C., Pringle, J.R., and Peifer, M. (2000). Evidence for functional differentiation among Drosophila septins in cytokinesis and cellularization. Mol. Biol. Cell 11, 3123–3135. Ahmari, S.E., Buchanan, J., and Smith, S.J. (2000). Assembly of presynaptic active zones from cytoplasmic transport packets. Nat. Neurosci. 3, 445–451. Allison, D.W., Chervin, A.S., Gelfand, V.I., and Craig, A.M. (2000). Postsynaptic scaffolds of excitatory and inhibitory synapses in hippocampal neurons: maintenance of core components independent of actin filaments and microtubules. J. Neurosci. 20, 4545–4554. Aravamudan, B., Fergestad, T., Davis, W.S., Rodesch, C.K., and Broadie, K. (1999). Drosophila UNC-13 is essential for synaptic transmission. Nat. Neurosci. 2, 965–971. Barral, Y., Mermall, V., Mooseker, M.S., and Snyder, M. (2000). Compartmentalization of the cell cortex by septins is required for maintenance of cell polarity in yeast. Mol. Cell 5, 841–851. Benson, D.L., and Tanaka, H. (1998). N-cadherin redistribution during synaptogenesis in hippocampal neurons. J. Neurosci. 18, 6892– 6904. Biederer, T., and Sudhof, T.C. (2000). Mints as adaptors. Direct binding to neurexins and recruitment of munc18. J. Biol. Chem. 275, 39803–39806. Bloom, F.E., and Aghajanian, G.K. (1968). Fine structural and cytochemical analysis of the staining of synaptic junctions with phosphotungstic acid. J. Ultrastruct. Res. 22, 361–375.

Neuron 76

Burry, R.W. (1980). Formation of apparent presynaptic elements in response to poly-basic compounds. Brain Res. 184, 85–98.

protein” is a subunit of a calmodulin-dependent protein kinase. Proc. Natl. Acad. Sci. USA 80, 7357–7361.

Cases-Langhoff, C., Voss, B., Garner, A.M., Appeltauer, U., Takei, K., Kindler, S., Veh, R.W., De Camilli, P., Gundelfinger, E.D., and Garner, C.C. (1996). Piccolo, a novel 420 kDa protein associated with the presynaptic cytomatrix. Eur. J. Cell Biol. 69, 214–223.

Kohmura, N., Senzaki, K., Hamada, S., Kai, N., Yasuda, R., Watanabe, M., Ishii, H., Yasuda, M., Mishina, M., and Yagi, T. (1998). Diversity revealed by a novel family of cadherins expressed in neurons at a synaptic complex. Neuron 20, 1137–1151.

Chen, Y.A., and Scheller, R.H. (2001). SNARE-mediated membrane fusion. Nat. Rev. Mol. Cell Biol. 2, 98–106.

Landis, D.M., Hall, A.K., Weinstein, L.A., and Reese, T.S. (1988). The organization of cytoplasm at the presynaptic active zone of a central nervous system synapse. Neuron 1, 201–209.

Cohen, R.S., Blomberg, F., Berzins, K., and Siekevitz, P. (1977). The structure of postsynaptic densities isolated from dog cerebral cortex. I. Overall morphology and protein composition. J. Cell Biol. 74, 181–203.

Morales, M., Colicos, M.A., and Goda, Y. (2000). Actin-dependent regulation of neurotransmitter release at central synapses. Neuron 27, 539–550.

Colman, D.R. (1997). Neurites, cadherins and synapses reconciled. Mol. Cell. Neurosci.

Peters, A., Palay, S.L., and Webster, H.D.F. (1991). The fine structure of the nervous system (New York: Oxford).

Craig, A.M., and Boudin, H. (2001). Molecular heterogeneity of central synapses: afferent and target regulation. Nat. Neurosci. 4, 569–578.

Pfenninger, K., Akert, K., Moor, H., and Sandri, C. (1972). The fine structure of freeze-fractured presynaptic membranes. J Neurocytol 1, 129–149.

Cremona, O., and De Camilli, P. (1997). Synaptic vesicle endocytosis. Curr. Opin. Neurobiol. 7, 323–330.

Phend, K.D., Rustioni, A., and Weinberg, R.J. (1995). An osmiumfree method of epon embedment that preserves both ultrastructure and antigenicity for post-embedding immunocytochemistry. J. Histochem. Cytochem. 43, 283–292.

Fannon, A.M., and Colman, D.R. (1996). A model for central synaptic junctional complex formation based on the differential adhesive specificities of the cadherins. Neuron 17, 423–434. Fenyo, D., Qin, J., and Chait, B.T. (1998). Protein identification using mass spectrometric information. Electrophoresis 19, 998–1005. Forgac, M. (2000). Structure, mechanism and regulation of the clathrin-coated vesicle and yeast vacuolar H(⫹)-ATPases. J. Exp. Biol. 203, 71–80. Foster, M., and Sherrington, C.S. (1897). A Text Book of Physiology, Part III: The Central Nervous System, 7 Edition (London: Macmillan). Garner, J.A., Linse, K.D., and Polk, R.K. (1996). Type I brain hexokinase: axonal transport and membrane associations within central nervous system presynaptic terminals. J. Neurochem. 67, 845–856. Garner, C.C., Kindler, S., and Gundelfinger, E.D. (2000). Molecular determinants of presynaptic active zones. Curr. Opin. Neurobiol. 10, 321–327. Hainfeld, J.F., and Steck, T.L. (1977). The sub-membrane reticulum of the human erythrocyte: a scanning electron microscope study. J. Supramol. Struct. 6, 301–311.

Rao, A., Harms, K.J., and Craig, A.M. (2000). Neuroligation: building synapses around the neurexin-neuroligin link. Nat. Neurosci. 3, 747–749. Rothman, J.E. (1994). Mechanisms of intracellular protein transport. Nature 372, 55–63. Scheiffele, P., Fan, J., Choih, J., Fetter, R., and Serafini, T. (2000). Neuroligin expressed in nonneuronal cells triggers presynaptic development in contacting axons. Cell 101, 657–669. Scheller, R.H. (1995). Membrane trafficking in the presynaptic nerve terminal. Neuron 14, 893–897. Schlossman, D.M., Schmid, S.L., Braell, W.A., and Rothman, J.E. (1984). An enzyme that removes clathrin coats: purification of an uncoating ATPase. J. Cell Biol. 99, 723–733. Shapiro, L., and Colman, D.R. (1999). The diversity of cadherins and implications for a synaptic adhesive code in the CNS. Neuron 23, 427–430.

Hall, Z.W., and Sanes, J.R. (1993). Synaptic structure and development: the neuromuscular junction. Cell Suppl. 72, 99–121.

Slepnev, V.I., and De Camilli, P. (2000). Accessory factors in clathrindependent synaptic vesicle endocytosis. Nat. Rev. Neurosci. 1, 161–172.

Harlow, M.L., Ress, D., Stoschek, A., Marshall, R.M., and McMahan, U.J. (2001). The architecture of active zone material at the frog’s neuromuscular junction. Nature 409, 479–484.

Soltys, B.J., and Gupta, R.S. (1999). Mitochondrial-matrix proteins at unexpected locations: are they exported? Trends Biochem. Sci. 24, 174–177.

Hata, Y., Slaughter, C.A., and Sudhof, T.C. (1993). Synaptic vesicle fusion complex contains unc-18 homologue bound to syntaxin. Nature 366, 347–351.

Song, J.Y., Ichtchenko, K., Sudhof, T.C., and Brose, N. (1999). Neuroligin 1 is a postsynaptic cell-adhesion molecule of excitatory synapses. Proc. Natl. Acad. Sci. USA 96, 1100–1105.

Hilfiker, S., Pieribone, V.A., Czernik, A.J., Kao, H.T., Augustine, G.J., and Greengard, P. (1999). Synapsins as regulators of neurotransmitter release. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 354, 269–279.

Sudhof, T.C. (1995). The synaptic vesicle cycle: a cascade of proteinprotein interactions. Nature 375, 645–653.

Hirokawa, N., Sobue, K., Kanda, K., Harada, A., and Yorifuji, H. (1989). The cytoskeletal architecture of the presynaptic terminal and molecular structure of synapsin 1. J. Cell Biol. 108, 111–126. Hull, B.E., and Staehelin, L.A. (1979). The terminal web. A reevaluation of its structure and function. J. Cell Biol. 81, 67–82. Huntley, G.W., and Benson, D.L. (1999). Neural (N)-cadherin at developing thalamocortical synapses provides an adhesion mechanism for the formation of somatopically organized connections. J. Comp. Neurol. 407, 453–471. Ichimura, T., and Hashimoto, P.H. (1988). Structural components in the synaptic cleft captured by freeze- substitution and deep etching of directly frozen cerebellar cortex. J. Neurocytol. 17, 3–12.

Takei, K., Haucke, V., Slepnev, V., Farsad, K., Salazar, M., Chen, H., and De Camilli, P. (1998). Generation of coated intermediates of clathrin-mediated endocytosis on protein-free liposomes. Cell 94, 131–141. Takizawa, P.A., DeRisi, J.L., Wilhelm, J.E., and Vale, R.D. (2000). Plasma membrane compartmentalization in yeast by messenger RNA transport and a septin diffusion barrier. Science 290, 341–344. Tanaka, H., Shan, W., Phillips, G.R., Arndt, K., Bozdagi, O., Shapiro, L., Huntley, G.W., Benson, D.L., and Colman, D.R. (2000). Molecular modification of N-cadherin in response to synaptic activity. Neuron 25, 93–107.

Kennedy, M.B. (1998). Signal transduction molecules at the glutamatergic postsynaptic membrane. Brain Res. Brain Res. Rev 26, 243–257.

tom Dieck, S., Sanmarti-Vila, L., Langnaese, K., Richter, K., Kindler, S., Soyke, A., Wex, H., Smalla, K.H., Kampf, U., Franzer, J.T., et al. (1998). Bassoon, a novel zinc-finger CAG/glutamine-repeat protein selectively localized at the active zone of presynaptic nerve terminals. J. Cell Biol. 142, 499–509.

Kennedy, M.B. (2000). Signal-processing machines at the postsynaptic density. Science 290, 750–754.

Triller, A., and Korn, H. (1985). Activity-dependent deformations of presynaptic grids at central synapses. J. Neurocytol. 14, 177–192.

Kennedy, M.B., Bennett, M.K., and Erondu, N.E. (1983). Biochemical and immunochemical evidence that the “major postsynaptic density

Uchida, N., Honjo, Y., Johnson, K.R., Wheelock, M.J., and Takeichi, M. (1996). The catenin/cadherin adhesion system is localized in


synaptic junctions bordering transmitter release zones. J. Cell Biol. 135, 767–779. Vincent, V., Goffin, V., Rozakis-Adcock, M., Mornon, J.P., and Kelly, P.A. (1997). Identification of cytoplasmic motifs required for short prolactin receptor internalization. J. Biol. Chem. 272, 7062–7068. Walikonis, R.S., Jensen, O.N., Mann, M., Provance, D.W., Jr., Mercer, J.A., and Kennedy, M.B. (2000). Identification of proteins in the postsynaptic density fraction by mass spectrometry. J. Neurosci. 20, 4069–4080. Wang, X., Kibschull, M., Laue, M.M., Lichte, B., Petrasch-Parwez, E., and Kilimann, M.W. (1999). Aczonin, a 550-kD putative scaffolding protein of presynaptic active zones, shares homology regions with Rim and Bassoon and binds profilin. J. Cell Biol. 147, 151–162. Wu, Q., and Maniatis, T. (1999). A striking organization of a large family of human neural cadherin-like cell adhesion genes. Cell 97, 779–790. Zhai, R.G., Vardinon-Friedman, H., Cases-Langhoff, C., Becker, B., Gundelfinger, E.D., Ziv, N.E., and Garner, C.C. (2001). Assembling the presynaptic active zone. A characterization of an active zone precursor vesicle. Neuron 29, 131–143. Zinsmaier, K.E., and Bronk, P. (2001). Molecular chaperones and the regulation of neurotransmitter exocytosis. Biochem. Pharmacol. 62, 1–11.