Assembly of Active Zone Precursor Vesicles

Neurotransmitter release from presynaptic nerve terminals is restricted to specialized areas of the plasma membrane, so-called active zones. Active zones are characterized by a network of cytoplasmic scaffolding proteins involved in active zone generation and synaptic transmission. To analyze the modes of biogenesis of this cytomatrix, we asked how Bassoon and Piccolo, two prototypic active zone cytomatrix molecules, are delivered to nascent synapses. Although these proteins may be transported via vesicles, little is known about the importance of a vesicular pathway and about molecular determinants of cytomatrix molecule trafficking. We found that Bassoon and Piccolo co-localize with markers of the trans-Golgi network in cultured neurons. Impairing vesicle exit from the Golgi complex, either using brefeldin A, recombinant proteins, or a low temperature block, prevented transport of Bassoon out of the soma. Deleting a newly identified Golgi-binding region of Bassoon impaired subcellular targeting of recombinant Bassoon. Overexpressing this region to specifically block Golgi binding of the endogenous protein reduced the concentration of Bassoon at synapses. These results suggest that, during the period of bulk synaptogenesis, a primordial cytomatrix assembles in a trans-Golgi compartment. They further indicate that transport via Golgi-derived vesicles is essential for delivery of cytomatrix proteins to the synapse. Paradigmatically this establishes Golgi transit as an obligatory step for subcellular trafficking of distinct cytoplasmic scaffolding proteins.

Synapses of the central nervous system are highly specialized asymmetric cell-cell contact sites mediating communication between neurons. A characteristic feature of synaptic junctions is the deposition of electron-dense meshworks of cytoskeletal and cytoskeleton-associated proteins at the pre-and postsynaptic plasma membrane. On the presynaptic side, this cytoskeletal protein matrix is called cytomatrix assembled at active zones (CAZ) 4 ; it defines the sites where synaptic vesicles dock and fuse to release neurotransmitter (1). The CAZ is thought to mediate pivotal events of synapse formation and function, including spatial restriction of neurotransmitter release to active zones and local recruitment of proteins and organelles (1)(2)(3)(4). CAZ-specific proteins include Munc13s, which are essential for neurotransmitter release (5,6); the Rab3-interacting molecules RIM1␣ and RIM2␣, which are scaffolding proteins regulating presynaptic events (7,8); the scaffolding protein CAST1/ERC2 (9 -11); and Bassoon and Piccolo/Aczonin. The latter two proteins are related giant CAZ components of 420 and 530 kDa, respectively, present at both excitatory and inhibitory synapses (12)(13)(14)(15)(16). The N-terminal 609 amino acids of Bassoon may be targeted to the Golgi apparatus and to synaptic vesicle clusters but are not required for either trafficking step. By contrast, a central region of Bassoon is essential for its anchoring within the CAZ, for synaptic transmission at a certain set of synapses, and for anchoring synaptic ribbons at active zones in retinal photoreceptor cells (17)(18)(19)(20).
The modes of CAZ assembly are currently being studied. Fluorescence imaging studies have revealed a stepwise incorporation of mobile units of Bassoon into nascent presynapses (21,22). In axons of immature neurons Bassoon and Piccolo are specifically associated with a class of 80-nm dense core vesicles, which have been termed Piccolo-Bassoon transport vesicles (PTVs) and which may represent cellular vehicles delivering packets of Bassoon to nascent synapses (23). Indeed it is an emerging view that presynaptic elements could be transported to synapses en bloc via active zone precursor vesicles whose exocytotic fusion might directly result in active zone generation (21,(23)(24)(25)(26). However, although such mechanisms are likely to exist, little is known about their importance in synapse formation. Moreover a need for vesicle-based transport of CAZ proteins, all of which are non-transmembrane proteins, is not directly obvious. Here we used cultured hippocampal neurons to ask: (i) what are the molecular mechanisms and subcellular sites of CAZ assembly and (ii) what is the importance of vesicular transport of CAZ proteins. Our data suggest that association within the trans-Golgi network is an obligatory step in transport of Bassoon and Piccolo to synapses.
Cell Cultures-Primary cultures of rat hippocampal neurons were prepared, maintained, and transfected as described previously (17). Briefly cells from embryonic day 19 rat brains dissociated with trypsin were plated onto coverslips at a density of 60,000 cells/cm 2 and maintained in Neurobasal medium including B27 (Invitrogen), antibiotics, and glutamine. Neurons were transfected using the calcium phosphate method on day in vitro (DIV) 3 or 7 as described (17). Human embryonic kidney 293 cells were maintained in minimum Eagle's medium supplemented with 10% fetal calf serum and transfected using the calcium phosphate method.
Immunofluorescence and Fluorescence Imaging-Cells were fixed by incubation for 5-20 min in 4% formaldehyde or for 20 min in Ϫ20°Ccold methanol (13,17). Antibodies were applied in blocking solution including 10% horse serum, 2% serum albumin, 5% sucrose, and 0.3% Triton X-100. Coverslips were mounted using Mowiol and viewed with a 63ϫ numerical aperture 1.4 Plan-Apo objective in a Zeiss Axioplan II microscope equipped with a SpotRT cooled charge-coupled device camera (Visitron Systems). Examples of specimen were analyzed and verified in addition by data acquisition using a Zeiss Axiovert 200M motorized microscope followed by deconvolution using Autoquant/ Metamorph software (Visitron Systems) and Openlab software (Improvision). Confocal images were acquired using a Leica TSC4D confocal microscope. Analysis of digital images was performed using Metamorph software (Visitron Systems). Fusion constructs were detected using anti-GFP immunofluorescence when high detection sensitivity was necessary, e.g. for detection of large constructs after relatively short times of expression. For reasons of consistency, all images of fixed samples represent immunofluorescence data. Formaldehyde and methanol fixation yielded similar results for all markers. Cells expressing Bsn-GFP were fixed using methanol to remove soluble GFP, which is cleaved off from a fraction of Bsn-GFP (17). Constructs tagged at their N termini, which do not lose GFP, were fixed using either formaldehyde or methanol.
Quantitative Analysis of Synaptic Accumulation Bassoon-Neurons were double transfected on DIV 3 with mRFP-synaptophysin as a marker for presynaptic specializations and with either an inert construct or the Golgi-binding region of Bassoon. Cells were fixed on DIV 8 and immunostained for endogenous Bassoon. For analysis only axons that could clearly be assigned to a double transfected neuron were chosen. For determination of synapse density the number of mRFP-synaptophysin puncta/20 m of axon was determined in several regions of the distal axon. Proximal axon parts (prior to the first accumulation of synapses, or 50 m from the soma) were excluded because they were usually devoid of synaptic puncta.
Brefeldin A Treatment-Brefeldin A (BFA; Sigma) was stored at Ϫ20°C as a stock solution of 5 mg/ml in ethanol. Ethanol or a diluted stock solution of BFA (0.5 mg/ml in ethanol) was added directly to the culture medium of control or experimental neurons, respectively, at 1:500. The final concentration of BFA was 3.57 M (1 g/ml). Addition of ethanol had no effect on localization of Golgi markers, synaptic markers, and recombinant proteins as compared with untreated neurons.

RESULTS
Endogenous Bassoon and Piccolo at the Golgi Apparatus-To investigate where Bassoon may associate with cellular membranes and possibly with PTVs we immunostained cultured hippocampal neurons grown for 3-4 DIV using Bassoon antibodies. The neurons frequently exhibited, in addition to the punctate staining pattern in axons characteristic of PTVs, an accumulation of Bassoon immunofluorescence in a juxtanuclear region (Fig. 1A). To assess whether this juxtanuclear distribution represents an association of Bassoon with the Golgi apparatus, cells were treated for 30 min with BFA, a drug that disrupts the Golgi complex. Under these conditions the juxtanuclear immunofluorescence of Bassoon was dispersed as characteristic of Golgi-associated proteins (Fig. 1, B and C). By contrast, the maximum intensity of juxtanuclear Bassoon immunofluorescence was increased ϳ2.2-fold when cultures were incubated at 19°C for 45 min (Fig. 1, D and E), a treatment known to attenuate the exit of organelles from the Golgi apparatus (30). Under these conditions virtually all neurons displayed juxtanuclear Bassoon immunofluorescence. Neither the 19°C block nor the BFA treatment had any effect on the punctate distribution pattern of Bassoon in axons (data not shown). Together these data suggest that Bassoon becomes associated with membranes within the Golgi apparatus and that this step is perhaps a prerequisite to its transport into axons in association with PTVs.
As the Golgi complex is rather condensed in young neurons, it was not possible to assess with which membranes within the Golgi stack Bassoon initially becomes associated. We therefore turned to older neurons (7-9 DIV), which contain a larger Golgi apparatus. Because the percentage of neurons containing a detectable juxtanuclear accumulation of Bassoon decreased with time in culture, we performed these experiments after a 45-min shift to 19°C. Under these conditions, both Bassoon and Piccolo co-localized with TGN38, a marker of the trans-Golgi network (TGN; Fig. 2). Bassoon and Piccolo immunofluorescence exhibited a somewhat granular staining pattern aligned with the rather continuous TGN38 staining. There was significantly less overlap of Bassoon with 58k protein (Fig. 2, G-I), another marker of the Golgi complex, than with TGN38 (for references to Golgi markers see "Materials and Methods"). A comparable separation of immunofluorescence signals was also observed for TGN38 and 58k (Fig. 2, J-L) demonstrating that the localization of Golgi proteins to distinct subcompartments can be separated by immunofluorescence. Hence co-localization of Bassoon and Piccolo with TGN38-positive structures may reflect association of the two proteins primarily with a TGN-related compartment (see also Fig. 4).
Subcellular Distribution of Recombinant Bassoon-To confirm the notion that Bassoon associates with the Golgi complex, we analyzed the distribution of GFP-tagged variants of Bassoon in the somata of transfected neurons. GFP-Bsn-(95-3938) and GFP-Bsn-(609 -3938), two N-terminally truncated versions of Bassoon that are targeted to synapses (17,21,22), accumulated at the Golgi complex in live and fixed neurons, indicating that N-terminal sequences, which are dispensable for presynaptic targeting (17), are also not required for Golgi association (Fig. 3). Similarly Bsn-GFP, a C-terminally tagged full-length construct of Bassoon that is targeted to synapses (17) Bsn-GFP was detected at the Golgi complex of all transfected neurons at all culture stages without 19°C block. Double immunofluorescence analysis revealed a pattern of Bsn-GFP fluorescence aligned with TGN38 (Figs. 3, H-J, and 4, A-C) in a way similar to the granular distribution of endogenous Bassoon at the Golgi complex. Moreover the degree of co-localization was highest for markers of TGN-related compartments, such as VAMP4 (Fig. 4, D-F) and syntaxin 6 (supplemental Fig. 1). There was less overlap with the cis-Golgi marker gm130 (Fig. 4, G-I) and with 58k ( Fig. 4, J-L) and no overlap with the endosomal markers transferrin receptor (Fig. 4, M-O) and EEA1 (data not shown). Interestingly juxtanuclear immunofluorescence of Bsn-GFP exactly co-localized with Piccolo ( Fig. 4, P-R). Together these data indicate that Bassoon and Piccolo associate with the same TGN compartment.
Bsn-GFP Is Not Transported to Synapses When Vesicle Biogenesis Is Blocked-Is the association of Bassoon and Piccolo with the Golgi complex important for axonal transport of these proteins? To investigate this question we aimed to express recombinant Bassoon in the presence or absence of a functional Golgi apparatus on DIV 8. After this time in culture, Bassoon is associated with axonal PTVs and synapses (21). Moreover axon length remains stable during a 14-h Golgi disruption Only cell somata are shown. A-F, Bassoon (green in C) and Piccolo (green in F) are closely associated with TGN38 (red). G-I, immunofluorescence for Bassoon (green) is distributed in a way complementary to immunofluorescence for Golgi marker 58k (red). J-L, likewise TGN38 immunofluorescence (green in L) is spatially discernible from that of 58k (red) suggesting that markers for two distinct Golgi compartments can be discriminated at the light microscopic level. The scale bar in L is 10 m for all panels. Insets are magnified 2-fold. paradigm using BFA in neurons of that culture stage (31). When neurons were transfected with Bsn-GFP on DIV 7 and fixed 18 h later, Bsn-GFP fluorescence was observed at the Golgi apparatus, as indicated by co-localization with syntaxin 6 in the soma, as puncta all along the axon and at synapses (supplemental Fig. 1 and Fig. 5G). Weak Bsn-GFP fluorescence was detected at the Golgi apparatus as short as 6 h after transfection, but 18 h were necessary for detection of fluorescence at synapses. When we applied BFA to DIV 7 neurons 4 h after transfection and fixed the neurons after an additional 14 h in the presence of the drug, Bsn-GFP was confined to the soma of transfected neurons (Fig.  5A). Bsn-GFP was always found in clusters (up to several micrometers in size) that also contained Piccolo (Fig. 5, C and D). Co-transfection of mRFP to visualize axons further supported this conclusion. Axons of BFA-treated neurons were devoid of recombinant Bassoon (Fig. 5, E and F, and Table 1), whereas puncta of recombinant Bassoon were abundant in axons of control neurons (Fig. 5, G and H, and Table 1). This indicates a significant impairment of axonal targeting of recombinant Bassoon in BFA-treated neurons. Immunostaining for endogenous Bassoon and Piccolo in untransfected, BFA-treated cultures revealed that the endogenous proteins co-localized in clusters in the soma too (Fig. 6). Under these conditions, extrasomatic puncta of Bassoon and Piccolo co-localizing with synaptophysin were still abundant, presumably representing pre-existing synapses (data not shown). However, the number of Bassoon puncta was reduced to 63% (p Ͻ 0.002), and the number of Piccolo puncta was reduced to 48% (p Ͻ 0.001). Moreover the intensity of these puncta was reduced to 57% for Bassoon (p Ͻ 0.002) and to 44% for Piccolo (p Ͻ 0.001). Together these experiments reveal a dramatic redistribution of Bassoon and Piccolo upon long term disruption of the Golgi apparatus. Moreover they corroborate the data obtained with Bsn-GFP and suggest that axonal transport of these proteins requires a functional Golgi apparatus.
To assess whether the effects observed after long term BFA treatment are specific and not due to a general perturbation of cell physiology we performed several control experiments. First, we expressed GFP-tagged versions of actin and tubulin in the presence of BFA, both of which were normally localized all along axons and dendrites 18 h after transfection (Table 1 and data not shown). Second, we expressed dsRed2-Mito, a plasmid targeting the red fluorescent protein dsRed2 to mitochondria in the presence of BFA. DsRed2-labeled mitochondria were observed all along neurites, which were usually 400 m or longer, 18 h after transfection. Mobile mitochondria moved with an average speed of 0.2 m/s when imaged in a time lapse series in living neurons (supplemental Fig.  2). Moreover the number of axons containing mitochondria that moved at least 50 m over a period of 5 min was similar between controls (nine cells of 11 analyzed) and BFA-treated cultures (eight cells of 12). Third, we co-transfected neurons at DIV 7 with either Bsn-GFP or GFP-actin and recombinant ARF1-Q71L. ARF1-Q71L is a constitutively active version of ARF1 that blocks vesicle exit from the Golgi apparatus in an ARF1-specific fashion (32). In co-transfected neurons Bsn-GFP was typically clustered in the soma and proximal dendrites, whereas GFPactin was found all along the axon (Table 1 and Fig. 7).
Finally we sought to block vesicle exit from the Golgi apparatus without the use of molecular or pharmacological means. To that end transfected neurons were cultured at 19°C for 18 h before examination. Strikingly Bsn-GFP immunofluorescence was confined to clusters in the soma of neurons (Fig. 8, A and C-E,   ( Table 1) and GFP-tubulin (Fig. 8B) were distributed all along the neurites. Together these data strongly suggest that blocking vesicle biogenesis at the Golgi apparatus impairs transport of Bassoon into axons.
Synaptic Targeting of Bassoon Is Impaired When Golgi Association Is Blocked-The observed Golgi association of Bassoon and the dependence of Bassoon localization on Golgi-derived vesicles are consistent with Bassoon associating with membranes at the level of the Golgi complex. If so, abrogation of Golgi binding should impair PTV association and axonal targeting of Bassoon. To test this prediction, we took advantage of our previous observation that two regions of Bassoon contain targeting signals exclusively for the Golgi apparatus. First, amino acids 1-97 are targeted to the Golgi apparatus via N-myristoylation. However, this region of Bassoon is not required for Golgi association (this study) or for presynaptic targeting (17). Second, amino acids 2088 -2564 tagged with GFP are confined to a juxtanuclear region (Ref. 17; see Fig. 10B). Large deletions including this region compromise presynaptic targeting of Bassoon (17,18). Here we found that the juxtanuclear localization of amino acids 2088 -2564, termed GFP-BsnGBR (Golgi-binding region), indeed reflects co-localization of the construct with TGN38 ( Fig. 9, A-C). Unlike endogenous and recombinant Bassoon, GFP-BsnGBR displayed a continuous staining pattern and exact co-localization with TGN38. When the GBR was deleted from recombinant Bassoon tagged with GFP at the N terminus, the resulting construct (GFP-Bsn⌬GBR) was distributed diffusely throughout transfected neurons (Fig. 9, D-F). This deletion construct still bound to CAST/ERC (supplemental Fig. 3), indicating that its general properties were not affected by the deletion. In particular, axons were diffusely stained and largely devoid of punctate staining (Fig. 9G) indicating that this mutant of Bassoon was not properly sorted out of the soma and into axons. Conversely GFP-Bsn-(609 -3938) and GFP-Bsn-(95-3938) produced the typical punctate pattern representative of PTVs (Fig. 9, H and I). Thus, perturbing the association of Bassoon with Golgi membranes by deleting Golgi-binding determinants drastically affects the intracellular transport of Bassoon.
Why does Bassoon cluster in the absence of a functional Golgi apparatus (Figs. 5, A and C, and 6, A and D), whereas a construct incapable of Golgi association is diffusely distributed (Fig. 9, D and G)? One explanation is that the deleted region itself promotes clustering when unable to bind to proper membrane receptors. Indeed when expressed in nonneuronal cells, GFP-BsnGBR formed aggregates similar to those formed by GFP-Bsn-(609 -3938) (Fig. 9, J and K), GFP-Bsn-(95-3938) and Bsn-GFP (data not shown), whereas the deletion construct (GFP-Bsn⌬GBR) was diffusely distributed (Fig. 9L). Together these data indicate that in the absence of appropriate binding partners either in non-neuronal cells

TABLE 1 Effects of blocking Golgi function on axonal localization of recombinant full-length constructs of Bassoon and actin
Constructs were detected in transfected neurons by anti-GFP immunofluorescence (IF). Axonal immunofluorescence of Bsn-GFP was always punctate. Data were obtained from transfected cells in three independent experiments. Neurons were maintained under the indicated conditions after transfection. Control cultures and BFA-treated cultures contained 0.005% ethanol, the solvent for BFA. This concentration of ethanol did not affect transport of any recombinant protein analyzed. ARF-Q71L is a constitutively active variant of ARF1 that impairs Golgi exit. To test the importance of GBR-mediated Golgi transit for trafficking of endogenous Bassoon we overexpressed the GBR in neurons, hypothesizing that it might compete with Bassoon for Golgi association and act as a dominant-negative tool. We used a red fluorescent version of the synaptic vesicle protein synaptophysin (mRFP-synaptophysin) as a marker for synapses in co-transfected neurons (Fig. 10, A, C, and E). Quantitative analysis revealed that the immunofluorescence of Bassoon at synaptic sites, indicated by mRFP-synaptophysin, was reduced on average to 72% (316 puncta analyzed, p Ͻ 0.001) in cells co-expressing mRFP-synaptophysin and GFP-BsnGBR compared with control cells co-expressing mRFP-synaptophysin and a palmitoylated CFP, which is targeted to the Golgi apparatus and the plasma membrane. Analysis of the distribution of fluorescence intensities revealed that expressing GFP-BsnGBR increased the number of synapses containing relatively low amounts of Bassoon (Fig. 10G). The number of mRFP-synaptophysin puncta per axon section was not significantly affected (control, 2.6 Ϯ 1.1/20 m; GFP-BsnGBR, 2.4 Ϯ 1.4/20 m), consistent with the lack of change in synapse number in Bassoon transgenic mice (18). However, 18% of mRFP-synatophysin puncta did not stain for endogenous Bassoon in GFP-BsnGBR-expressing neurons, although in control neurons, only 1.9 % of mRFP-synaptophysin puncta were Bassoon-negative. Thus, these data indicate that the GBR of Bassoon interferes with trafficking of Bassoon and suggest that endogenous Bassoon is targeted to synapses via GBR-mediated Golgi transit.

DISCUSSION
Although Bassoon is required for synaptic transmission only at a subset of synapses it is nevertheless located at the majority of central nervous system synapses (18,20). Here we made use of the fact that Bassoon is a well established CAZ marker and that recombinant Bassoon has been extensively characterized as a general synapse marker in expression studies in cultured neurons (11,17,21,22).

Role of Golgi-derived Vesicles in Correct Subcellular Localization of Bassoon and Piccolo: Golgi Disruption
Paradigms-Our previous studies have shown that in axons and growth cones Piccolo and Bassoon are associated with the surface of 80-nm dense core vesicles termed PTVs (21,23). The first question addressed in this study was: is vesicle-based transport one of several modes of CAZ protein delivery to synapses, or is it a crucial pathway? The redistribution of Bassoon and Piccolo into somatodendritic clusters in the absence of a functional Golgi apparatus suggests that PTVs have an essential role in transporting these CAZ proteins to sites of synapse formation at least during the initial period of synaptogenesis. This conclusion is based on a set of experiments aimed at blocking the secretory pathway while leaving non-vesicular modes of protein transport intact. Three assays were used to block Golgi-derived vesicle formation: (i) BFA-dependent dispersion of the Golgi complex, (ii) specific interference with the monomeric GTPase ARF1 using a recombinant construct, and (iii) incubation of neurons at 19°C. All three tools affected the localization of Bassoon when applied in standard  short term paradigms. In addition, all three tools lead to ectopic accumulation and clustering of Bassoon synthesized in the absence of a functional secretory pathway in long term paradigms, i.e. a 14-h suppression of vesicle biogenesis at the Golgi apparatus. These long term paradigms, although crucial in addressing the question of the importance of vesicles in synaptic targeting of CAZ proteins, are expected to impose stress to the neurons analyzed. The conclusion that Bassoon and Piccolo need vesicle association to be taken to synapses thus relies critically on the specific actions of these tools.
The temporal paradigm of 14 h was chosen because it represented the minimum time required for transfected neurons to synthesize GFP-Bassoon and transport detectable amounts of it to axons and presynapses. A pulselike expression of GFP-Bassoon was necessary to test what happens to protein synthesized in the absence of vesicle biogenesis. It has been shown that axons of stage 4 neurons, as used in our study, stop growing but do not shrink when incubated for 14 h with BFA (31). Moreover neurons remain capable of selective protein sorting after a 72-h period of inhibited endosomal trafficking (33). We therefore felt confident to test the feasibility of a 14-h BFA treatment of stage 4 neurons as an experimental paradigm. Indeed axon length was not reduced; soluble and cytoskeletal proteins including GFP, GFP-actin, and GFPtubulin were localized throughout neurons including their tips; and mitochondria maintained mobility in axons, indicating that the general mechanisms underlying diffusion, transport of cytoskeletal proteins, and microtubule-based organelle transport remained intact.
Of the tools used, the Golgi-disrupting drug BFA may be the one with the broadest spectrum of targets, raising questions as to the specificity of this drug. BFA prevents activation of ADP ribosylation factor (ARF) isoforms 1-5, monomeric GTPases that recruit coat proteins to membranes and thus induce vesicle budding from a donor compartment (34 -36). In addition, a second mode of action of BFA has been described involving inhibition of CtBP/BARS, a protein involved in Golgi vesicle fission (37). Thus, the cell biological consequences of BFA effects, be it via blocking ARF isoforms or CtBP/BARS, ultimately result from an impairment of vesicle generation. However, our finding that ARF1-Q71L expression produced effects similar to those of BFA, strongly suggests a major role for Golgi-derived vesicles in subcellular transport of Bassoon. ARF1-Q71L is a constitutively active version of ARF1 that permanently associates with Golgi membranes, thus specifically preventing fission of vesicles from the Golgi apparatus (32,38). Together these data suggest that Bassoon, as well as Piccolo, requires Golgi-derived vesicles for delivery to synapses. This conclusion is cor-  Bassoon deletion constructs were localized by immunofluorescence in DIV 6 neurons (A-I) and human embryonic kidney 293 cells (J-L). A-C, a construct encoding the GBR (amino acids 2088 -2564) fused to GFP (GFP-BsnGBR) co-localizes with TGN38. D-F, when the GBR is deleted from fulllength Bassoon tagged with GFP at the N terminus (GFP-Bsn⌬GBR), the resulting recombinant protein is distributed diffusely in the soma. G, the Golgi association-deficient mutant is diffusely distributed in axons, indicating that Golgi association via the GBR is essential for targeting of Bassoon to PTVs. H and I, N-terminally deleted constructs, which were shown to associate with the Golgi apparatus, give the typical PTV staining pattern in axons, indicating that N-myristoylation is not essential for targeting to PTVs. J-L, GFP-BsnGBR (J) and constructs containing the GBR, exemplified here by GFP-Bsn-(609 -3938) (K), produce aggregates upon expression in non-neuronal cells, whereas GFP-Bsn⌬GBR is diffusely distributed, indicating that the GBR promotes aggregation in non-neuronal cells (L). The scale bar in J represents 10 m for A-F and J-L and 3 m for G-I. roborated by the importance of Golgi association via the GBR for trafficking of Bassoon.
Golgi Association of Bassoon-Piccolo and Bassoon are so far the only specific markers of active zone precursor vesicles or PTVs (21,23). To understand the mechanisms of PTV assembly it is important to know at what stage of subcellular transport Bassoon and Piccolo do associate with vesicles. Both proteins co-localize at the Golgi apparatus. The granular appearance characteristic of both recombinant and endogenous Bassoon as well as Piccolo may reflect concentration of Bassoon and Piccolo in TGN exit sites and may be expected for non-resident proteins of the Golgi complex. It has been shown that cargo segregation can be visualized by light microscopy at the level of the TGN (30,39). The granular appearance may thus represent Bassoon and Piccolo associating with clusters of budding vesicles or tubular transport containers, which originate from specialized regions of the TGN (40).
Several lines of evidence suggest that the compartment where Bassoon and Piccolo co-localize and associate with membranes is indeed the TGN. Bassoon immunofluorescence showed substantial overlap with TGN-related markers, whereas no significant co-localization was observed with endoplasmic reticulum, 5 cis-Golgi, and early and recycling endosomal markers. Moreover, GFP-BsnGBR, a construct targeted to the Golgi apparatus but not detected outside the soma, colocalized exactly with TGN38. In particular, this construct, which is apparently unable to leave the Golgi complex, was characterized by a homogeneous distribution entirely congruent with TGN38, whereas all constructs capable of exiting from the soma displayed a discontinuous pattern of immunofluorescence along TGN38. Interestingly the second of the two regions of Bassoon mediating Golgi targeting, comprising N-myristoylated amino acids 1-97, co-localizes with syntaxin 6, another TGN marker (17). This suggests that Bassoon may be generally recruited to TGN membranes via N-myristoylation and the region termed BsnGBR followed by concentration into TGN exit sites via distinct sequences.
General Importance of Golgi Association for Subcellular Transport of CAZ Proteins-We also asked whether Golgi association is an essential step in subcellular targeting of Bassoon. The diffuse distribution of FIGURE 10. Overexpressing the Golgi-binding region of Bassoon interferes with synaptic targeting of endogenous Bassoon. A-F, experimental paradigm for quantitative analysis of synaptic Bassoon immunofluorescence. The panels show a neuron co-expressing mRFP-synaptophysin and GFP-BsnGBR after fixing and immunostaining for endogenous Bassoon using Alexa647 secondary antibodies. A, mRFP autofluorescence. B, GFP autofluorescence. mRFP-synaptophysin is located in the Golgi region as well as in axonal hot spots, most of which represent synapses and contained endogenous Bassoon. GFP-BsnGBR is confined to the Golgi apparatus. C, magnified version of the box highlighted in A. D, immunofluorescence of endogenous Bassoon in that region. E, areas of mRFP-synaptophysin fluorescence selected for quantitative analysis are highlighted in pink. The boundaries of the areas are highlighted in blue for clarity. F, the sum of all pixel gray values (total gray level) of Bassoon immunofluorescence was determined for each area. G, histogram of total gray levels of synaptic Bassoon in neurons expressing GFP-BsnGBR or CFP-Mem (control), a palmitoylated CFP targeted to the cytoplasmic side of the Golgi apparatus and the plasma membrane. The total gray level of Bassoon was determined for 316 mRFP-synaptophysin puncta from 25 control and 25 GBR-expressing cells, respectively. x axis, percentage ("fraction") of analyzed puncta. y axis, binned total gray levels of synaptic Bassoon immunofluorescence. The number of synapses containing low levels of Bassoon is increased upon expression of the Golgi-binding region of Bassoon.
Golgi binding-deficient constructs in transfected neurons suggests that recruitment of Bassoon to Golgi membranes, although most likely transient, is nevertheless essential for subsequent steps in subcellular transport. This strongly suggests that transport of Bassoon to axons and synapses depends on prior membrane association and TGN transition and that loading on PTVs or PTV precursors at the Golgi complex is an obligatory step in this transport process. This is further supported by the fact that overexpressing the GBR of Bassoon reduced the amount of endogenous Bassoon at synapses.
The fact that Bassoon and Piccolo co-localize already at the level of the Golgi complex raises the possibility that PTVs may be loaded with two key CAZ proteins immediately upon generation. The apparent importance of Golgi association and vesicle-based transport for axonal transport and for synaptic delivery of Bassoon and Piccolo is intriguing in view of the nature of CAZ proteins as non-transmembrane proteins for which a necessity for vesicle association is not obvious per se. One reason for this pathway could be that these proteins evolved due to their potential as large scaffolding proteins whose tendency to aggregate had to be counteracted during evolution by association with membranous vehicles. Another possibility would be that active zone generation demanded an event of simultaneous packaging of CAZ molecules during vesicle biogenesis, e.g. to ensure a certain ratio of distinct molecules per vesicle. The fact that vesicle-based transport seems to be crucial for axonal targeting of two prototypical CAZ proteins, at least at the culture stage analyzed, lends strong support to our hypothesis that active zones are generated from exocytotic fusion of PTVs (21)(22)(23). Moreover the implication of the Golgi complex as an obligatory stage in subcellular transport of Bassoon and a site of co-localization of Bassoon and Piccolo raises the possibility that these two and potentially additional CAZ proteins are assembled on nascent PTVs already at the TGN. In this scenario, CAZ formation would be comprised of two spatially and mechanistically distinct events, i.e. assembly of CAZ molecules into a complex at the Golgi apparatus and deposition of preformed CAZ complexes at nascent synapses most likely from less than five PTVs (21,22). This would call for the existence of signals mediating CAZ protein targeting to sites of assembly at the TGN as well as trafficking signals leading PTVs to nascent synapses.