Traffic of synaptic vesicle proteins in polarized and nonpolarized cells

Neurons are highly polarized cells that are specialized for the reception, integration and emission of signals. The perikaryal-dendritic region is specialized for receptive functions, the axon for the translocation of signals over long distances, and the axon terminal for the regulated secretion of signal molecules. Two main classes of organelles participate in regulated secretion from nerve terminals: large dense-core vesicles and synaptic vesicles (SVs) (Hdkfelt et al., 1986). Large dense-core vesicles store and secrete peptide neurotransmitters and may also contain amines. They are the neuronal equivalent of secretory granules of endocrine cells, which secrete amines and peptide hormones. Large dense-core vesicles have a scattered distribution in nerve terminals where their exocytosis is not restricted to the synaptic region and is preferentially triggered by trains of closely spaced action potentials. Secretion via large dense-core vesicles plays primarily a modulatory role at synapses (De Camilli and Jahn, 1990; Thureson-Klein et al., 1988). SVs store and secrete ‘fast’ non-peptide neurotransmitters and represent a highly specialized class of secretory organelles. They are homogeneous in size and they form large clusters in which SVs are connected to each other by a cytoskeletal matrix. These clusters, in turn, are docked to the region of the plasmalemma that faces the synaptic cleft. Exocytosis of SVs takes place selectively at this region and is stimulated with an extremely short latency (less than 1 millisecond) by nerve terminal depolarization. Secretion via synaptic vesicles is responsible for the fast point-to-point signalling typical of synapses as well as for modulatory signalling (Pappas and Purpura, 1972; De Camilli and Jahn, 1990).

The biogenesis of large dense-core vesicles is well established. Like endocrine secretory granules, large dense-core vesicles are assembled and loaded with their contents in the region of the irans-Golgi network (TGN) and are transported to the cell periphery as mature organelles (Tooze, 1991). The site where SVs are first generated remains unclear, but is well known that SVs undergo cycles of exo-endocytosis in nerve terminals and that at each cycle they are re-loaded with neurotransmitter content by transporters and enzymes present in the axon ending (Ceccarelli et al., 1973; Heuser and Reese, 1973; Betz and Bewick, 1992). Thus, these organelles can be continuously re-formed at a site distant from the TGN (Regnier-Vigouroux et al., 1991; Cameron et al., 1991; Lindstedt and Kelly, 1991).

The precise recycling pathway of SVs and the molecular mechanisms underlying this process have become the focus of intense investigation over the last several years (De Camilli and Jahn, 1990; Siidhof and Jahn, 1991; Trimble et al., 1991; Greengard et al., 1993; Kelly, 1993). The most widely accepted hypothesis is that SV recycling involves clathrin-coated vesicles and early endosomal intermediates (Heuser, 1989; Maycox et al., 1992), but a direct re-formation of SVs from the plasmalemma cannot be ruled out. If the re-formation of SVs involves fusion of coated vesicle-derived structures with early endosomes and budding from endosomes, the recycling of SVs can be seen as a special case of the receptor-mediated recycling pathway (Mell-man et al., 1987). By extension, it would also represent a special case of vesicular recycling, which takes place at all the stations of the secretory and endocytic pathway where anatomically distinct compartments are functionally interconnected by vesicular traffic. Thus, the study of the mechanisms involved in budding, docking and fusion of SVs and SV-derived structures may shed light on mechanisms of vesicular traffic in general. A unique advantage offered by the study of SVs as a prototypic vesicular carrier is the possibility of isolating them in high yield and purity (Huttner et al., 1983). This facilitates the identification and characterization of the main molecular components of their membranes, an important first step in elucidating molecular mechanisms. Many protein families present in SV membranes have already been identified. In order to establish how this information is relevant to other aspects of membrane traffic it is important to establish the precise relationship between SV recycling and other types of membrane recycling. This short review will address this relationship.

Synaptic vesicles represent one of the best-characterized transport vesicles. A schematic drawing, which illustrates some of the main protein components that have been identified on SVs, is shown in Fig. 1 (for reviews see De Camilli et al., 1990; Trimble et al., 1991; Siidhof and Jahn, 1991; Bennett et al., 1992; Jessel and Kandel, 1993; Greengard et al., 1993).

One approach, which has been taken to elucidate the relationships between the trafficking of SV proteins and well-established traffic pathways of all cells, is to transfect SV proteins in cells that do not contain SVs and to monitor their targeting in this ectopic environment. The rationale behind this approach is that a SV protein will follow the intracellular trafficking route more closely related to the trafficking route of SVs in neurons. When synaptophysin, one of the most abundant SV membrane proteins (Wiedenmann and Franke, 1985; Jahn et al., 1985), was expressed in fibroblastic CHO cells, its intracellular distribution was virtually identical to that of transferrin and LDL receptors, two proteins that undergo constitutive recycling through the receptor-mediated recycling pathway (Johnston et al., 1989; Cameron et al., 1991; Linstedt and Kelly, 1991). Colocalization was demonstrated by a variety of complementary procedures including immunocytochemistry, immunoisolation, and equilibrium and velocity centrifugation. Additionally, brefeldin A (BFA) induced a redistribution into the same tubular network of both the transferrin receptor and synaptophysin (Fig. 2). BFA is a fungal metabolite that induces a massive fusion of early endosomes with each other and with elements of the TGN, resulting in the formation of an interconnected network of tubules where proteins of early endosomes and the TGN are intermixed (Lippincott-Schwartz et al., 1991).

Similar results were obtained when synaptoporin was expressed in CHO cells (Fykse et al., 1993). These findings indicate that at least two abundant SV proteins have targeting information, which leads to their accumulation in the receptor-mediated recycling pathway. They support the hypothesis that the recycling of SVs is closely related to the vesicular recycling between early endosomes and the plasmalemma.

The elucidation of the relationship between the SV recycling pathway and the receptor-mediated recycling pathway in neurons is complicated by the presence of distinct cellular compartments specialized for different functions. It would be useful to compare these two pathways in cells that contain SVs but that are not polarized. Cell lines derived from endocrine cells that secrete amines and/or peptide hormones have these properties.

Until recently, SVs were considered neuron-specific organelles. However, it is now clear that organelles closely related to SVs (referred to as synaptic-like microvesicles) are present in endocrine cells (De Camilli, 1991). Synaptic-like microvesicles have a membrane composition very similar to that of SVs and, like SVs, undergo exocytosis, endocytosis and recycling (Navone et al., 1986; Wiedenmann et al., 1988; Baumert et al., 1990; Cameron et al., 1991; Linstedt and Kelly, 1991; Regnier-Vigouroux et al., 1991). In addition, they appear to share with SVs the property of storing and secreting signalling molecules: synaptic-like microvesicles of pancreatic p-cell lines were shown to contain GABA (Thomas-Reetz et al., 1993) and those of the chromaffin cell-derived PC 12 cells were shown to contain acetylcholine (Bauerfeind et al., 1993). Thus, synaptic-like microvesicles of endocrine cell lines represent a useful model system with which to study basic aspects of SV traffic in a non-polarized cell.

In endocrine cells SV proteins are present primarily in two pools. One pool is localized in organelles with the same morphology and sedimentation characteristics as SVs, i.e. in bona fide synaptic-like microvesicles. The other pool is colocalized with the transferrin receptor and the LDL receptor in tubulovesicular structures with the morphology and sedimentation characteristics of early endosomes. The two pools of SV proteins present in a low-speed cell supernatants of endocrine cell lines (RIN cells and PC 12 cells) are dramatically demonstrated by velocity sedimentation in glycerol gradients (Cameron et al., 1991; Lindstedt and Kelly, 1991). In immunoisolation experiments carried out with PC 12 cells, about two-thirds of the transferrin receptor was recovered with synaptophysin-containing membranes (Cameron et al., 1991). Upon BFA treatment, a large fraction of synaptophysin is redistributed in the same tubular network, which is positive for transferrin receptor (our unpublished observations).

Taken together these data suggest that in endocrine cells the traffic of SV proteins and the constitutive recycling of transferrin receptor are highly interrelated. A cartoon, which summarizes possible relationships of synaptic-like microvesicles to early endosomes in endocrine cells, is shown in Fig. 3. The model predicts that at each exo-endo-cytotic cycle, membrane proteins of synaptic-like microvesicles are internalized together with recycling receptors via clathrin-coated vesicles and are then delivered to early endosomes. Subsequently, SV proteins are sorted away from recycling receptors in early endosomes and are assembled into membrane microdomains, which bud off as synaptic-like microvesicles. Whether or not SLMVs may also be re-formed directly from the plasmalemma is not known at present.

In mature neurons, the presynaptic terminal is the cellular compartment specialized for the recycling of SV proteins. However, it is clear that endosome-plasmalemma recycling pathways operate both in axons and in dendrites. Until recently, little information was available regarding recycling traffic in the perikaryal-dendritic region of neurons. Given the complexity of neuronal tissue, an analysis of these traffic pathways cannot be practically performed on neurons in situ using either morphological or biochemical methods. However, central nervous system neurons can be grown in primary culture, and cultured neurons represent a powerful system for studying protein and organelle traffic using morphological techniques. It was shown that cultured hippocampal neurons faithfully express many of the properties of their in situ counterparts. These neurons establish axonal and dendritic polarity even when grown in isolation, thus allowing for the study of those features of cell polarity that are intrinsic to a neuron irrespective of its contact with neighboring cells (Goslin and Banker, 1989).

SV proteins are already present in hippocampal neurons in vitro at stages that precede the growth of an axon (stage 2, see Dotti et al., 1988). As soon as the axon begins to differentiate (i.e. one of the cell processes acquires a unique morphology and starts elongating at a faster rate), SV proteins become concentrated in this process. Their concentration in dendrites slowly declines, but never falls to undetectable levels (Matteoli et al., 1991, 1992; Mundigl et al., 1993). When viewed by immunofluorescence microscopy, immunoreactivity for SV proteins in isolated axons appears with a punctate distribution throughout the axonal arbor. When axons form synaptic contacts with other cells, these fine immunoreactive puncta coalesce into much brighter and larger puncta, which correspond to synaptic clustering of SVs (Fletcher et al., 1991; Matteoli et al., 1991).

Do SVs present in the processes of isolated neurons already undergo exocytosis and recycling? Typically, SV exocytosis is monitored by post-synaptic electrical recording, but this assay cannot be applied to isolated neurons. To answer this question, a morphological assay to detect SV exocytosis was developed (Matteoli et al., 1992). This assay, which is independent of the release of neurotransmitter, is based on the detection of lumenal epitopes of the SV protein synaptotagmin I, which are exposed to the cell surface as a result of exocytosis. Only a very small pool of synaptotagmin I is present at the cell surface under control conditions (Matteoli et al., 1992).

Application of this assay demonstrated a very active exo-endocytosis of synaptotagmin-I-containing vesicles in the developing axons of isolated neurons (Fig. 4) (Matteoli et al., 1992). Furthermore, it indicated a very long half-life of SVs in the axon. Binding of antibodies to the lumenal domain of synaptotagmin I did not appear to target the protein for lysosomal degradation and SVs were found to recycle for several days with the antibodies present in their lumen. In contrast, another endocytic marker, wheat-germ agglutinin (WGA), was taken up by the axon, but rapidly targeted to a population of vacuoles distinct from SVs and cleared from the axon by transport to the cell body within less than 24 hours (Matteoli et al., 1992).

In the same neurons, the transferrin receptor was primarily restricted to the perikaryal-dendritic region at all stages of differentiation. This was shown both by immunostaining the receptor in fixed permeabilized cells with monoclonal antibodies, and by steady-state labeling of living neurons with fluorescent transferrin. In both cases the two staining patterns were identical, consistent with a constitutive exo-endocytotic recycling of transferrin receptor in dendrites. As soon as an axon emerges, the transferrin receptor is preferentially excluded from that process. In more mature axons the transferrin receptor is virtually undetectable (Cameron et al., 1991; Mundigl et al., 1993). The distribution of transferrin receptor immunoreactivity in dendrites is similar to the distribution of the dendritic pool of several SV proteins (SV2, protein p29, synaptotagmin, synaptobrevin, rab3A), and almost identical to that of synaptophysin (Cameron et al., 1991; Mundigl et al., 1993).

These observations suggest the existence of at least two distinct pathways of recycling between intracellular vesicles and the plasmalemma in neurons. The two pathways are anatomically separate and operate in different regions of the cell, the perikaryal-dendritic region and the axon. The transferrin receptor is present only in the perikaryal-dendritic recycling pathway. SVs proteins are enriched in the axonal recycling pathway.

To elucidate further the properties of these two recycling pathways the effects of BFA were investigated. Addition of 5-10 μg/ml BFA for 20 minutes to isolated hippocampal neurons induced a massive tubulation of perikaryal-dendritic early endosomes as indicated by the massive tubulation of transferrin receptor immunoreactivity. In addition, BFA induced a massive co-tubulation of dendritic synap-tophysin and transferrin receptor. Unexpectedly, it did not induce an identical co-tubulation of other SV proteins. Furthermore, the distribution of SV proteins in axons was virtually unaffected (Mundigl et al., 1993). These results strengthen the hypothesis that dendritic and axonal recycling pathways differ in some fundamental property. Additionally, they suggest that SV proteins are differentially sorted in the perikaryal dendritic region. While the significance of these findings remains unclear and requires future studies, they clearly indicate that SV proteins are not directly assembled into mature SV membranes when they first exit the Golgi complex as newly synthesized proteins. This is in agreement with recent results obtained in PC 12 cells indicating that newly synthesized synaptophysin is transported to the cell surface by the constitutive pathway (Regnier-Vigouroux et al., 1991).

In summary, the perikaryal-dendritic region of isolated hippocampal neurons developing in culture contain early endosomes where the transferrin receptor is colocalized with synaptophysin and with partial pools of other SV proteins. These endosomes may represent the neuronal equivalent of typical early endosomes, which are found in all cells. Recycling of SVs in axons seems to occur via endo-somal intermediates with unique properties. The two recycling compartments may be functionally interconnected via vesicular carriers, since transcytosis in both an anterograde and a retrograde direction has been documented in neurons (Kuypers and Ugolini, 1990). While the transferrin receptor is excluded from axons, SV proteins are greatly concentrated in these processes, where they have a long halflife, but are not excluded from dendrites. The striking concentration of SVs observed in axons may be the result of both selective targeting and selective retention.

The existence of functionally distinct early endosomal systems operating at two distinct poles of the cell has been documented in polarized epithelial cells (Partonet al., 1989; Rodriguez-Boulan and Powell, 1992). Several similarities have already emerged between the sorting of membrane components to the apical and basolaterals domains of epithelial cells and the sorting of membrane components to the axonal and dendritic plasmalemmal domains, respectively (Dotti et al., this volume). When hippocampal neurons in culture are infected with the vesicular stomatitis virus (VSV), the VSV glycoprotein, which is sorted to the basolateral domain in MDCK cells, is localized exclusively to cell bodies and dendrites. In contrast the hemaglutinin protein of the avian influenza fowl plague virus (FPV), which is an apically sorted protein in MDCK cells, is directed preferentially, but not exclusively, to the axon in neurons (Dotti and Simons, 1990). In addition, Thy-1, a glycosyl phosphotidylinositol anchored protein that is apically located in MDCK cells, is sorted to the axonal domain of cultured hippocampal neurons (Dotti et al., 1991).

The selective localization of the transferrin receptor in dendrites and perikarya of hippocampal neurons in primary cultures (Cameron et al., 1991; Mundigl et al., 1993) is in agreement with the selective localization of this protein in basolateral endosomes of epithelial cells (Fuller and Simons, 1986). Polarized epithelial cells do not contain SVs. If mechanisms that target proteins to recycling compartments of axons are similar to those that target proteins to the apical region of epithelial cells, SV proteins may be at least partially directed to the apical pole if expressed by transfection in polarized epithelial cells. To test this hypothesis, MDCK cells that stably express synaptophysin were prepared and the localization of synaptophysin in these cells was investigated.

Laser confocal microscopy analysis of confluent monolayers of transfected MDCK cells after synaptophysin immunostaining revealed a punctate fluorescence distribution apically, intracellularly and along the lateral plasma membrane (Fig. 5). Synaptophysin colocalized in part with the 135 kDa glycoprotein (gp!35), which is localized only to the apical cell surface (Ojakian and Schwimmer, 1988), and in part with the p-subunit of the Na/K-ATPase (not shown), which is localized at the basolateral plasma membrane (Smith et al., 1987). However, an additional intracellular pool of synaptophysin was also visible (Fig. 5, and our unpublished observations).

To assess further the cell surface domains at which synaptophysin is exposed, transfected cells were surface-labeled using a membrane-impermeant biotin analog (sulfo-NHS-biotin; Graeve et al., 1989). Cells were grown to con-fluency on polycarbonate filters to allow for the selective biotinylation of either the apical or the basolateral plasma membrane domains. Synaptophysin became labeled when the biotinylation reagent was added to the apical or to the basolateral side of the monolayer at 0°C, demonstrating that synaptophysin is exposed to both plasmalemmal domains (Fig. 6 and our unpublished observations). Additionally, following both apical or basolateral biotynylation, labeled synaptophysin became internalized. This was demonstrated by results obtained with the cleavable biotin analog NHS-SS-biotin (Graeve et al., 1989). The basolateral or the apical plasma membranes were selectively biotinylated with this compound at 0°C, and then incubated at 37°C to allow endocytosis to occur. Biotin remaining at the cell surface was stripped by reduction with glutathione and then alkylated using iodoacetamide. Endocytosis of synaptophysin was measured by the increase in the amount of biotinylated synaptophysin that became resistant to reduction by extra-cellular glutathione over time (Graeve et al., 1989; and our unpublished observations).

Taken together these results suggest that synaptophysin is targeted to both basolateral and apical recycling compartments of the cell. These findings indicate a parallel with the distribution of the protein in isolated hippocampal neurons in primary culture: in these neurons synaptophysin as well as other SV proteins are concentrated in axons, but not excluded from dendrites. MDCK cells may represent an useful experimental system with which to identify the motifs of SV proteins that are required for axonal targeting.

It will be of interest to determine whether synaptophysin is targeted apically in MDCK cells via a direct route or by following transcytosis from the basolateral surface. Such an indirect route for apical delivery has been documented in polarized epithelial cells (Rodriguez-Boulan and Powell, 1992). It will also be of interest to determine whether dendritic synaptophysin can be targeted to axons by an analogous transcytotic route.

SVs are highly specialized secretory organelles by which neurons secrete non-peptide neurotransmitters at synapses. Their recycling pathway appears to be closely related to the receptor-mediated recycling pathway that functionally interconnects early endosomes and the plasmalemma in all cells. In non-polarized cells that express SV proteins, the recycling of SVs partially overlaps with the recycling of transferrin receptor. In neurons, SV proteins are progressively segregated away from transferrin receptors in parallel with the establishment of dendritic and axonal polarity. This segregation correlates with the organization of two anatomically and functionally distinct recycling pathways, which may be closely related to the two recycling pathway present at the apical and basolateral domains of polarized epithelial cells.

We thank Dr T. Siidhof for the gift of the synaptophysin cDNA clone, which was used to prepare stably transfected MDCK cell lines.