The transport of presynaptic proteins from soma through axon to their final destination of presynaptic terminals is complex and a subject of intense study [1, 2]. The presynaptic specializations consist of clusters of synaptic vesicles (SV) and active zone (AZ) cytomatrix, which are localized at sites of SV release. While the transport of AZ proteins has been described at both the light microscopy (LM) and electron microscopy (EM) levels [3, 4, 5], fewer EM studies have been carried out on SV proteins, especially in developing axons.
Fluorescence tagged SV proteins like GFP-synaptophysin  and GFP-VAMP/synaptobrevin  provided live observations of packets of tubular/vesicular structures carrying these SV proteins and moving through axons. Correlative LM immunolabeling of these mobile packets showed presence of other presynaptic proteins indicating that the packets contain many components required for the formation of the presynaptic terminal . Correlative EM of these mobile packets showed aggregates of tubular-vesicular structures, pleomorphic small vesicles, and dense core vesicles (DCV) . However, direct visualization of SV proteins on the various components via immunogold labeling by EM is lacking.
In axons younger than 3 days in vitro (DIV), GFP-tagged SV proteins have a diffused appearance with few stationary punctas, which represent nascent synapses . From 4 DIV onward, many mobile transport packets move up and down the axons, and the number and size of nascent synapses increase with time . The present study used pre-embedding immunogold EM to examine the distribution of various endogenous SV integral membrane proteins including synaptophysin , SV2 , VAMP/synaptobrevin , synaptotagmin , and SV-associated proteins including synapsin  and synuclein  in dissociated rat hippocampal neurons. Young axons at 3–6 DIV were chosen for easier identification of mobile transport packets which outnumber synapses at these developmental stages .
The present approach of examining the distribution of endogenous SV proteins provides a clear view of these proteins’ biogenesis and transport at the ultrastructural level. These observations illustrate the different routes taken by different SV and AZ proteins, and provide clues to their eventual incorporation into a nascent synapse.
Rabbit polyclonal antibody (rabbit pAb) against synaptophysin (1:250) was from DAKO (Glostrup, Denmark); mouse monoclonal antibody (mouse mAb) against SV2 (1:500) was a gift from Dr. Erik S. Schweitzer (UCLA, Los Angeles, CA); mouse mAb against VAMP (1:100, clone SP10) and SNAP–25 (1:250, clone SP14) were from Chemicon (Temecula, CA); mouse mAb against synaptotagmin (p65, 1:250, clone ASV30) and mouse mAb against Bassoon (1:100, clone SAP7F407) were from Stressgen (Victoria, BC, Canada); mouse mAb against -synuclein (1:100, clone 42) was from BD Biosciences (San Jose, CA); mouse mAb against synapsin I (1:250, clone 46.1) was from Synaptic Systems (Gottingen, Germany). Guinea pig polyclonal antibody against Piccolo (1:100) was a gift from Dr. Eckart Gundelfinger (Leibniz Institute for Neurobiology, Magdeburg, Germany). Controls for specificity of immunolabeling include omitting the primary antibody and using the different primary antibodies as controls for each other.
Most samples were from a previously published report on synaptic active zone cytomatrix proteins  and reexamined here for synaptic vesicle proteins. Briefly, cell cultures were prepared from embryonic 20-day-old rat fetuses by papain dissociation, and then plated with or without a glial feeder cultures, and experiments were carried out with 3–6 day-old cultures. No difference in labeling pattern was observed between the two types of cultures for any of the antibodies.
For optimal structural preservation, cells were fixed with 4% glutaraldehyde in 0.1 M cacodylate buffer at pH 7.4 for 30 min at room temperature and then stored at 4˚C. These samples were post-fixed with 1% osmium tetroxide in 0.1 M cacodylate buffer for 1 hr on ice, and stained with 1% uranyl acetate in acetate buffer at pH 5.0 overnight before dehydration and embedding for electron microscopy.
For immunogold labeling, cells were fixed with one of the following fixation conditions (optimal fixation conditions for each antibody are listed in Additional File 1): (1) 4% paraformaldehyde in phosphate buffered saline (PBS) for 45–60 min, (2) 4% paraformaldehyde and 0.02–0.05% glutaraldehyde for 30–60 min, (3) 2% acrolein in PBS for 1 min followed by 4% paraformaldehyde in PBS for 30–60 min. Fixed cells were washed and permeabilized/blocked with 0.1% saponin/5% normal goat serum in PBS for 1 hr, incubated with primary antibody for 1–2 hr, incubated with secondary antibody conjugated to 1.4 nm gold particles (1:250, Nanogold from Nanoprobes, Yaphand, NY) for 1 hr, silver enhanced (HQ silver enhancement kit, Nanoprobes), treated with 0.2% OsO4 in phosphate buffer for 30 min on ice, followed by 0.25% uranyl acetate in acetate buffer at pH 5.0 at 4˚C for 30 min–1 hr, dehydrated in a graded series of ethanol and embedded in epoxy resin. Thin sections were counterstained with uranyl acetate and lead citrate. Images were photographed with a bottom-mounted digital CCD camera (AMT XR–100, Danvers, MA, USA).
In dissociated hippocampal cultures that contain a mixture of neurons and glia, it is more difficult to distinguish the two cell types at 3–6 DIV than in older cultures. For example, the conspicuous structural differences of the nuclei between neurons and glia at 3 wk in culture  is not evident in young cultures used in the present study. However, because all antibodies used here are neuron-specific, a labeled cell body has to be of neuronal and not of glial origin.
Although it is more difficult to identify axons from dendrites by structural characteristics alone in younger cultures than in older cultures, axons can still be identified by immunogold labeling patterns because all the antibodies used here are specific for presynaptic proteins. Thus, the presence of label in neurites suggests that they are axons. This assumption is consistent with earlier reports of LM immunolabeling that synaptophysin and synapsin are sorted into axons early in development .
Dissociated hippocampal neuronal cultures at 3–6 DIV were fixed and labeled with antibodies against four different SV integral membrane proteins: synaptophysin, SV2, VAMP & synaptotagmin. Synaptophysin and SV2 antibodies produced the most consistent and high efficiency labeling under many different fixation conditions, and thus, were illustrated to a greater extent in the present study.
As expected of integral membrane proteins, labels for synaptophysin (Fig. 1a) and SV2 (Fig. 2a) were localized at the Golgi complex . In neuronal somas, labels for both antibodies were also specifically localized at membranous structures of various size and shape, scattered in the cytoplasm as individual entities (arrows in Figs. 1a & 2a). Many of these labeled vesicles/vacuoles became aggregated in the axons (arrows in Figs. 1b & 2b), but not in soma and dendrites. These labeled aggregates are termed as “SV membrane protein transport aggregate” in this paper. The overall size of the labeled aggregates ranged widely from several vesicles/vacuoles (~0.2 µm, arrows in Additional File 2a & b) to often exceeding 1 µm in length (arrows in Figs. 1b & 2b), and sometimes greater than 2 µm (Additional File 3a). Higher mag images of these labeled aggregates showed a mixture of tubular (arrows in Figs. 1c & 2c) and vesicular structures of variable size and shape.
Notably, clusters of vesicles of uniform size at ~40 nm were also labeled in axons (Fig. 1d; circled area in Fig. 2c). These vesicle clusters resemble synaptic vesicles (SV) clusters in the presynaptic terminals , and are termed as “clusters of SV-like vesicles” here. The number of vesicles in these clusters ranged from 4–30 in single sections, and a few examples from small to larger clusters are illustrated in Fig. 3. Interestingly, clathrin-coated vesicles were often present near these SV-like vesicle clusters (arrows in Fig. 3c), indicating occurrence of endocytoses .
Serial thin sections demonstrated that some labeled aggregates were clearly not part of synapses, and many axons containing these labeled aggregates did not even come in contact with dendrites (Fig. 4; Additional File 4). Thus, the aggregation of these labeled vesicles/vacuoles was intrinsic to axon and not induced by external contact with dendritic elements. Some of these aggregates consisted mostly of tubular vesicular structures (aggregate “a” in Fig. 4), and others mostly of SV-like vesicles (aggregate “c” in Fig. 4). Interestingly, clathrin-coated vesicles were often seen among both types of labeled aggregates (thick arrows in Fig. 4), indicating active endocytosis near both types. Due to crowding of the vesicles/vacuoles, it is difficult to discern whether clathrin-coated vesicles were specifically labeled. However, on occasion, some coated vesicles appeared to be labeled (arrow in Additional File 4, section #4).
Multivesicular bodies (MVB, open arrows in Fig. 2b; Additional File 2c; Additional File 4, section # 4 & 5), which are classified as late endosome that may be en route to fuse with lysosome , were often present among the labeled aggregates of vesicles/vacuoles. However, the great majority of these MVBs were not labeled for SV integral membrane proteins. Notably, no late stage lysosomes, such as lipofuscin bodies or vacuoles containing multilamellar structures or electron dense material , were observed in the vicinity of labeled SV protein transport aggregates.
Labels for two other SV integral membrane proteins, VAMP/synaptobrevin and synaptotagmin, were also localized at the Golgi complex (images not shown), and at individual and aggregated vesicles/vacuoles in the axons (Additional File 2c, d). Thus, the four SV integral membrane proteins studied here had similar distribution patterns in soma and in axon. However, the present preembedding immunogold labeling method does not allow double labeling, and thus, cannot determine whether these four proteins are colocalized in the same vesicle/vacuoles. Notably, not all vesicles/vacuoles were labeled even when they were in the vicinity of the labeled aggregates (boxed area in Figs.1b, 2c and Additional File 2).
In contrast to labels for SV integral membrane proteins, labels for two SV-associated proteins, synapsin I and -synuclein, were not localized at the Golgi complex (Figs. 5a & 6a) in the soma, but dispersed in cytoplasm, not associated with any membranous structures. The synapsin I antibody used here produced better labeling efficiency than the -synuclein antibody, and thus, generated more detailed observations here.
In mature synapses, labels for synapsin I and -synuclein are localized to clusters of SVs in the presynaptic terminals . Here in young neuronal cultures before robust synaptogenesis occurs, most labels for synapsin I and -synuclein were cytosolic in somas and axons, but became concentrated on clusters of SV-like vesicles in axons (arrow in Figs. 5b & 6b). However, no other membranous structures, such as tubular or pleomorphic vacuoles were conspicuously labeled for synapsin I in the axons. Thus, other than being colocalized in SV-like vesicles, the distribution of labels for SV-associated proteins were very different from those of the SV integral membrane proteins.
SNAP–25, synaptosomal associated protein of 25 kDa, is part of the SNARE complex involved in exocytosis of synaptic vesicles . In mature neurons, label for SNAP–25 is polarized to axon and localized to plasma membrane along the entire axon [19, 20]. Here, in young neuronal cultures, labeling pattern of SANP–25 was compared to those of the SV proteins illustrated above. Label for SNAP–25 was localized at the Golgi complex in soma (Fig. 7a), and became sorted to axolemma (Fig. 7b, c) as early as 4 DIV. In contrast to SV proteins, clusters of SV-like vesicles were clearly not labeled (Fig. 7c).
Nascent synapses were seen as early as 3 DIV (Fig. 8a) and synaptic profiles appeared more frequently in subsequent days. The immature synapses contained fewer number of SV than mature synapses [5, 21], but already showed the characteristic synaptic cleft with a uniform gap at ~20 nm, and the postsynaptic density of variable prominence (Fig. 8). As in the case of mature synapses , the SVs were labeled for both the SV integral membrane proteins (Figs. 8a & b) and SV-associated proteins (Fig. 8c). Thus, even though SV integral membrane proteins and SV-associated proteins were transported via different routes from soma through axons, they ended up in the same final compartment, the SVs, at the presynaptic terminal upon synapse formation. Furthermore, labels for AZ cytomatrix proteins were also localized at active zone in these nascent synapses  (Fig. 8d).
Dense core vesicles (DCV) are more frequently seen in young axons than in mature samples, both in animals  and in cell cultures [3, 5, 7]. These DCVs sometimes existed in groups, intermingled with some SV-like vesicles (Fig. 9). Because occurrence of these vesicle mixtures of multiple DCVs and SVs were relatively infrequent compared to the occurrence frequencies of SV or AZ protein transport aggregates, it is difficult to capture them for serial section analysis. Thus, it cannot be determined whether these vesicle mixtures are part of a developing presynaptic specialization or existed in isolation in the absence of dendritic contact. While the SV-like vesicles labeled for all SV proteins, the vesicular membranes of DCV only labeled for some, such as SV2 (Fig. 9a) and synaptotagmin (Fig. 9b), but not for all SV membrane proteins. For example, DCV were mostly negative for synaptophysin (Fig. 9c) or VAMP (image not shown), and labels for AZ cytomatrix proteins were localized to dark material outside of DCV (Figs. 9d) .
The present study used immunogold EM to examine distribution of endogenous synaptic vesicle (SV) proteins in young axons in dissociated hippocampal cultures at 3–6 DIV to document these proteins’ biosynthesis, axon transport, and eventual sorting into the SV. The main focus here is the contrast between the two types of SV proteins—SV integral membrane proteins vs. SV-associated proteins.
In the neuronal soma where proteins are synthesized, labels for SV integral membrane proteins (synaptophysin, SV2, VAM/synaptobrevin, and synaptotagmin) were localized at the Golgi complex and other membranous structures in the cytoplasm. In contrast, labels for SV-associated proteins (synapsin and synuclein) were not localized at the Golgi, consistent with earlier LM results . Here at the ultrastructural level, labels for synapsin I and -synuclein were dispersed in the cytoplasm, not associated with membranous vesicles/vacuoles.
SV integral membrane and SV-associated proteins were both polarized into axons early in development , but transported differently [1, 23]. SV integral membrane proteins are transported as a mixture of tubular-vesicular structures [6, 7] by fast axonal transport, predominantly carried by Kinesin 3 family . However, different SV membrane proteins may be sorted into different cargos  as synaptophysin and SV2 are transported separately in spinal nerve bundles  and in differentiated PC12 cells . On the other hand, axonal transport for SV-associated proteins (synapsin and synuclein) is even more complicated [1, 23] because these proteins are not constantly, but reversibly associated with SV membranes. The membrane-associated form may be part of the fast component of axonal transport, while the ones not associated with membranes may be in the slow component [1, 23]. The present study provided structural evidence that label for synapsin I was mostly cytosolic, and only became associated with vesicular membranes after clusters of SV-like vesicles were formed in axons. This finding is consistent with the notion that synapsin I plays a role in clustering the SV vesicles .
SVs with a full complement of their specific proteins are not formed in soma but only in axon, and can form in the absence of dendritic contact. These observations are consistent with earlier reports that SVs are formed only after undergoing exo- and endocytosis through specialized sorting at recycling endosomes in axons, and can form without dendritic contact [1, 6, 26, 27]. The common occurrence of clathrin-coated vesicles near the SV transport aggregates provides structural evidence of robust endocytosis at these locations. The fact that many clathrin-coated vesicles were of a similar size to the clusters of SV-like vesicles nearby are consistent with the possibility that these coated vesicles could shed the clathrin coating and become SV-like vesicles . Furthermore, the axolemmal labeling of SNAP–25, a part of the SNARE complex involved in exocytosis , is consistent with the idea that exocytosis can occur all along the axons, not just restricted to presynaptic active zone . Thus, the present study showed ultrastructural and immunogold illustrations that young axons could be capable of localized exocytosis and endocytosis, resulting in clusters of SV-like vesicles at non-synaptic sites.
In addition to SV proteins, the active zone (AZ) cytomatrix proteins, such as Bassoon and Piccolo, also have to be transported through axons to reach their final destination at the synapses [2, 3, 4, 5]. These AZ transport aggregates consist of 1–2 dense core vesicles (DCV) and 4–5 SV-like vesicles in single sections, and the average size of these AZ transport aggregates (~ 0.2 µm)  is much smaller than the SV membrane protein transport aggregates reported here, which often exceeds 1 µm in length (Additional File 3). Although LM immunolabeling showed partial colocalization of SV and AZ transport cargos [3, 28], the bulk of the SV membrane proteins has to be transported via the much larger sized SV membrane protein transport aggregates due to the sheer abundance of SV proteins contained in these aggregates.
Notably, DCVs are consistently present in AZ transport aggregates, and AZ proteins like Bassoon and Piccolo are associated with the outside of the DCV membrane . It has been proposed that a nascent presynaptic active zone can be formed by the exocytosis of a few DCV  or AZ transport aggregate . It is likely that the exocytosis of these DCVs would deposit the externally associated AZ material onto the cytosolic side of the plasma membrane, forming an AZ-like structure. Whether such AZ-like structures precede dendritic contact is still unresolved. If so, such “orphan” active zones would have Bassoon or Piccolo-labeled dark material localized to the cytoplasmic side of axonal plasma membrane without an apposed dendritic element. No such “orphan” AZ-like structures were seen in young axons 3–6 DIV by EM examination , but they could exist in cultures older than 10 DIV, where Bassoon-labeled “orphan” punctas are present by LM evidence . Finally, many more DCVs are present in developing than in mature axons [3, 5, 7, 21], and multiple DCVs are sometimes seen at nascent presynaptic terminals  but rarely in mature ones . The depletion of DCVs in mature axons suggests that DCVs are exocytosed during development, and could possibly play a role in synaptogenesis .
Interestingly, multivesicular body (MVB), a vacuole of the late endosome category , was frequently seen in close association with the SV protein transport aggregate. This observation is consistent with LM observations on axons from young hippocampal cultures that ~85% of anterogradely transported SV punctas colocalize with lysosome-related punctas . The lysosome-related marker used in that study is Lamp1, which labels MVBs even before their fusion with lysosomes . Thus, the MVB seen in the present study near SV membrane transport aggregates may represent the Lamp1-labeled “lysosome-related vesicles” . In that study, loss of the lysosomal kinesin adaptor led to accumulation of SV and AZ proteins in the soma and a decrease of these proteins in the presynaptic sites, suggesting that a lysosome-related organelle may be involved in presynaptic biogenesis . The present finding that these MVBs did not labeled for SV proteins suggests that SV proteins may not traffic through these MVBs.
In summary, the present findings provide ultrastructural basis which are consistent with the views that (1) SV integral membrane proteins (synaptophysin, SV2, VAMP, synaptotagmin) and SV-associated proteins (synapsin I and synuclein) are transported in axons via different routes, the former in aggregates of tubular-vesicular structures and the latter mostly cytosolic, (2) SV-associated proteins only become membrane-associated after SV-like vesicles with a uniform size at ~40 nm are formed, (3) clusters of SV-like vesicles are not formed in soma but in axon, (4) these clusters of SV-like vesicles contain a full complement of SV-specific proteins, and can form in young axons prior to dendritic contact. The present study also provides additional evidence that the SV transport aggregates are distinct from clusters of SV-like vesicles or AZ transport aggregates, and that the bulk of SV proteins are transported via SV transport aggregate.
I thank Rita Azzam, and Virginia Crocker for expert EM technical support, Christine A. Winters for hippocampal dissociated cultures, Dr. Ayse Dosemeci for critical reading of the manuscript.
Supported by National Institute of Neurological Disorders and Stroke (NINDS) intramural funds.
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.
This is a solo author manuscript. The author read and approved the final manuscript.
The animal protocol was approved by the National Institute of Neurological Disorders and Stroke Animal Use and Care Committee (Animal protocol Number: ASP1159) and conforms to NIH guidelines.
The author declares that she has no competing interests.
life cycle. Prog Neurobiol. 2006; 80: 177-217.
proteins in a preassembled multi-vesicle transport aggregate. Neuroscience. 2007; 150: 575-84.
vesicle and plasma membrane proteins in living axons. J Cell Biol. 1998; 140: 659-74.
or walk-on role in synaptic vesicle exocytosis? Bioessays. 2004; 26: 445-53.
Biol. 2006; 7: 631-43.
synapse function and plasticity. Prog Neurobiol. 2010; 91: 313-48.
2015; 5: 699-713.
and synaptophysin in hippocampal neurons developing in culture. J Neurosci. 1991; 11:1617-26.
along the axolemma: an immunogold study of intact neurons. J Neurocytol. 2000; 29: 67-77.
Neuroscientist. 2014; 20: 71-81.
recycling of synaptic vesicles in developing processes of cultured hippocampal
neurons. J Cell Biol. 1992; 117: 849-61.
active-zone transport vesicles and are diminished following synaptogenesis in
mature hippocampal slices. Neuroscience. 2006; 141: 2097-106.