INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

CAVEOLAE ARE non-clathrin-coated, 60- to 80-nm invaginations of the plasma membrane seen in many cell types. Most microvascular endothelia of the continuous type have an abundance of caveolae presumably for macromolecular transport from the circulating blood to underlying tissue cells. Early morphological studies suggest a role for caveolae in the transport of molecules across the endothelium (for review see Refs. 28 and 32). Endothelial cells can possess plasma membrane receptors for blood-borne molecules, such as insulin, transferrin, and albumins, presumably for specific transport. Other studies suggest that caveolae in various cells can also endocytose specific ligands such as cholera toxin and modified albumins for delivery to endosomal compartments (16, 19, 31).

In the last few years it has been shown that filipin, a cholesterol-binding agent known to disrupt the structure of caveolae, inhibits both the transcytosis of insulin and native albumin by endothelial cells as well as the endocytosis and degradation of modified albumins (31). These findings support the concept that caveolae play an active role in cellular ligand transport and processing. Further credence to this view comes from the finding that transcytosis and endocytosis of caveolar ligands is also inhibited by N-ethylmaleimide (NEM) (21, 26). NEM, a well-characterized inhibitor of vesicular transport in a variety of cell types (4, 6, 23), modifies an ATPase called NEM-sensitive fusion protein (NSF). Its catalytic activity was thought to be necessary for the fusion of the carrier vesicle with the target membrane (6). However, recent evidence suggests that NSF acts at a step before membrane fusion (2). The apparent involvement of NSF in the delivery of proteins via caveolae suggested the possibility that at least in endothelial cells caveolae-mediated transport utilizes factors already identified as essential in other vesicular transport pathways.

During intracellular trafficking, v-SNAREs (soluble NSF attachment receptors located on the vesicle), such as the vesicle-associated membrane protein (VAMP) family of proteins, must recognize and dock with receptor proteins or complexes on the target membranes (called t-SNAREs) before fusion of the target and vesicle membrane which permits ligand delivery (37). t-SNAREs can define a target or acceptor membrane and interact with several v-SNAREs expressed on different populations of approaching transport vesicles (11). VAMPs (also called synaptobrevins) are a family of 18- to 20-kDa membrane proteins first reported as being enriched in synaptic vesicles (3, 34, 36), which mediate vesicle docking and fusion with the plasma membrane. They are thought to form part of the ternary SNAP (soluble NSF attachment protein) receptor (SNARE) complex which consists of a four--helical bundle containing VAMP, syntaxin, and SNAP-25 in a 1:1:1 stoichiometry. Complex formation has recently been shown to involve a conformational change in the VAMP from an unstructured form to an -helical form (10), and currently it is thought that this conformational change initiates membrane fusion. VAMP is a critical component for trafficking because cleavage of VAMP can disrupt the organization and kinetics of vesicular trafficking pathways (15, 36).

Recently we have shown that endothelial cell caveolae purified directly from rat lung are enriched in VAMP-2, NSF, and SNAP (25). They also contain many signaling molecules, including various GTPases, kinases, and Ca²⁺ regulators (18, 27, 29). We have proposed that these components may enable caveolae to be regulated dynamic vesicular carriers capable not only of budding but also of docking and fusing with a recipient target membrane such as endosomes during their endocytosis (25, 29). Recently, we have found that dynamin, which forms a collar around the neck of caveolae, hydrolyzes GTP to mediate the fission of caveolae from the plasma membrane to form free transport vesicles (18, 28).

Because NEM can modify and inactivate many proteins, including ion channels and some apparently involved in membrane fusion (1, 20, 22), it lacks sufficient specificity to allow dissection of molecular mechanisms. Therefore, we decided to utilize neurotoxins capable of specific cleavage of the v-SNARE VAMP at defined sites expressed on the cytoplasmic aspect of vesicular carriers. Past work shows that VAMP-specific botulinum neurotoxins prevent normal neurotransmission by cleaving VAMP to disrupt synaptic vesicle fusion required for the exocytic release of neurotransmitters (15, 36). Here, functional assays have been carried out to investigate whether the fidelity of caveolar trafficking can be maintained after VAMP cleavage in cultured endothelial cells. This report describes the subcellular distribution of VAMP-2 in rat lung tissue and in cultured vascular endothelial cells and the effects of the specific cleavage of VAMP-2 by neurotoxins on the intracellular trafficking of cholera toxin B (CTB) subunit by caveolae.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials. Early-passage rat aortic endothelial cells (RAEC) were donated by Dr. K. Guice (Duke University, Durham, NC) and early-passage bovine aortic endothelial cells (BAEC) were provided by Ken Baker (Sandoz, Boston, MA). Protein A-gold conjugates were a kind gift from Maria Ericsson (Harvard Medical School). Reagents and other supplies were obtained from the following sources: fetal calf serum (FCS), Dulbecco's modified Eagle's medium (DMEM), and antibiotics from GIBCO-BRL (Grand Island, NY); CTB subunit conjugated to fluorescein isothiocyanate (FITC), -actin-specific monoclonal antibody, heparin, endothelial growth supplement, cold-water fish skin gelatin (FSG), and methyl cellulose were purchased from Sigma Chemical (St. Louis, MO); BCA protein assay from Pierce Chemical (Rockford, IL); polyclonal antibodies to caveolin from Transduction Laboratories (Lexington, KY); monoclonal antibody Cl-69.1 was a gift from Dr. Reinhard Jahn (Yale University); reduced streptolysin-O (SLO) from Murex Diagnostics (Norcross, GA); chamber slides from Nunclon (Naperville, IL); Gelmount from Fisher Scientific (Pittsburgh, PA); botulinum A, D, and F toxins and CTB subunit conjugated to horseradish peroxidase (HRP) from Calbiochem (San Diego, CA); dithiothreitol (DTT) from Boehringer Mannheim Biochemicals (Indianapolis, IN); enhanced chemiluminescent substrate and Bolton and Hunter reagent from Amersham (Arlington Heights, IL); and all tissue culture plastic ware from Costar (Cambridge, MA) or Corning (Wilmington, DE). PD-10 columns were purchased from Bio-Rad (Melville, NY); polyethylene glycol (M_r 20,000) colloidal gold, uranyl acetate, and all electron microscopy processing and embedding reagents were obtained from Electron Microscopy Sciences (Fort Washington, PA).

Ultracryomicrotomy and immunogold localization of VAMP and caveolin on lung endothelium and BAEC. Each rat was anesthetized by injecting into the thigh muscles 0.1 ml of a 3:1 mixture of 10 mg/ml ketamine and 10 mg/ml xylazine per 100 g body weight. After thoracotomy the pericardium was removed and 0.2 ml of DMEM containing 30 µM freshly prepared nitroprusside (as a vasodilator) and 200 U heparin were injected into the right ventricle. Rat lungs were perfused in situ via the pulmonary artery as described by Schnitzer et al. (29) with 30 ml of phosphate-buffered saline (PBS) at 37°C supplemented with 14 mM glucose and 10 mg/ml bovine serum albumin (BSA) and then 4% paraformaldehyde in 0.1 M sodium cacodylate buffer, pH 7.4. The tissue specimens were infiltrated for 1-2 h with 2.3 M sucrose-0.1 M phosphate buffer, pH 7.4, containing 25% polyvinylpyrrolidone (35) and then mounted on metal nails and frozen in liquid nitrogen. Thin frozen sections were cut on tungsten-coated glass knives at 100°C using a Reichert FCS cryoultramicrotome. Sections were picked up in 2.3 M sucrose and transferred to formvar-carbon coated nickel grids and floated on PBS until immunogold labeling was done at room temperature. All antibodies were diluted in 0.5% cold-water FSG. Grids were floated on drops of 0.5% FSG for 10 min to block nonspecific labeling and then transferred for incubation with caveolin antibodies (1.5 µg/ml) for 45 min. The grids were then washed in four drops of PBS for a total of 15 min before labeling with a protein A-gold conjugate (10 nm) for 20 min and further washes in four changes of PBS for 15 min. The grids were incubated with 1% glutaraldehyde for 5 min, after which the glutaraldehyde was quenched by incubation in 0.2 M glycine (4 changes in 15 min). The sections were then incubated in the second antibody, Cl-69.1 (1:50), for 30 min and then washed for 15 min in PBS (4 changes) before labeling with a protein A-gold (5 nm) conjugate for 20 min. The grids were washed in four changes of PBS for 15 min and then in six changes of distilled water for 20 min. Contrasting embedding of the labeled grids was carried out on ice in 0.3% uranyl acetate in 2% methyl cellulose for 10 min. Grids were picked up with metal loops, and excess liquid was removed by streaking on filter paper, leaving a thin coat of methyl cellulose, and then dried. The grids were examined in a Phillips 300 electron microscope, and images were recorded at a primary magnification of ×20,000-30,000.

Morphometric analysis of VAMP-2 labeling in lung endothelium. Lung samples (1 mm³) were removed from the microvasculature of three different lobes of three rats (9 samples in all) and processed for ultracryomicrotomy. All were positive for VAMP-2 and caveolin by immunogold labeling but gave no labeling using control antibodies. Ultrathin cryosections immunolabeled with the VAMP-2 specific monoclonal antibody Cl-69.1 were examined from three different lungs and photographed using a Phillips 300 electron microscope. Gold particles (550) associated with the endothelium and epithelium were counted. Gold particles labeling the plasma membrane were distinguished from those apparently free in the cytoplasm and those associated with intracellular membranes. Linear membranes in the cytoplasm of unknown organelle type were scored as intracellular membranes. On close examination of the plasma membranes, gold particles associated with plasma membranes were distinguished from those associated with caveolae attached to the plasma membranes and the caveolae that were apparently detached (not in apparent physical contact either with the plasma membrane or with a vesicle attached to the plasma membrane). The number of gold particles associated with epithelial membranes was also counted.

Iodination of CTB subunit. CTB was dialyzed against a borate-NaCl buffer, pH 8.4, and iodinated using the Bolton and Hunter reagent (Amersham Pharmacia Biotech). CTB (1 mg) was reacted with 1 mCi of the dried ester on ice for 30 min. An excess of glycine was added, and the reaction was continued further for 15 min to conjugate any unreacted ester. The labeled CTB was then separated from the hydrolysis products by gel filtration on a BSA-pretreated PD-10 column, and 1-ml fractions were collected and quantified by gamma-spectroscopy. Fractions containing the radiolabeled CTB were pooled, and the specific activity was determined.

Endothelial cell culture and permeabilized cell system. RAEC were grown at 5% CO₂ in low-glucose DMEM supplemented with 10% FCS, low-molecular-weight heparin (90 µg/ml), endothelial growth supplement (60 µg/ml), and 2 mM L-glutamine. BAEC were grown in high-glucose DMEM with 10% FCS, penicillin, and streptomycin. For permeabilization, the cells were cooled to 4°C for 10 min and washed once with cold buffer A (in mM: 20 HEPES, 110 NaCl, 5.4 KCl, 0.9 Na₂HPO₄, 10 MgCl₂, and 11 glucose, pH 7.4). Reduced SLO dissolved in buffer A at 0.6 U/ml was bound to cells at 4°C for 15 min before two washes in cold buffer A and incubation in buffer A at 37°C for 15 min to induce pore formation in the plasma membranes. For functional assays, the permeabilized cells were incubated in buffer A containing 2 mM ATP or in rat lung cytosol adjusted to a protein concentration of 5 mg/ml with cytosolic buffer (in mM, 25 KCl, 2.5 Mg-acetate, 5 EGTA, 150 K-acetate, and 25 HEPES, pH 7.4). Cytosol was prepared from rat lungs perfused as described previously (31). Briefly, rat lungs were flushed free of blood, perfused with cold protease inhibitors, and homogenized in cold cytosolic buffer at 4°C. The homogenate was centrifuged at 103,000 g for 1 h at 4°C, and the supernatant was snap frozen in liquid N₂ and stored at 80°C. The protein concentration was determined using a Micro BCA Protein Assay kit with BSA as a standard.

Cleavage of VAMP-2 in cultured endothelial cells. The optimum time and concentration of botulinum toxin D treatment necessary for cleavage of VAMP in permeabilized RAEC cells were assessed. The cells were seeded in a 96-well plate for 48 h and then permeabilized as described above before incubation at 37°C for 30 min with 0, 1, 10, 30, and 100 nM reduced botulinum toxin D in buffer A containing 2 mM ATP. The cells were washed three times in PBS and then processed for SDS-PAGE with electrotransfer to nitrocellulose for immunoblotting as in our past work (25). In a separate experiment, cells were incubated with 100 nM botulinum toxin D for 5, 10, 20, and 30 min or with 100 nM botulinum A or F toxin that had been incubated with 1.4 mM captopril or buffer A for 30 min, before processing for SDS-PAGE and immunoblotting. The immunoblots were probed for VAMP with the monoclonal antibody Cl-69.1, and the bands were detected using enhanced chemiluminescent substrate. Band intensities were quantified by densitometry of autoradiograms (Molecular Dynamics, Sunnyvale, CA) as described previously (25).

Endocytosis of CTB-FITC in toxin-treated endothelial cells. RAEC were grown in eight-well chamber slides and permeabilized as described above. Botulinum toxin D (30 µg) was reduced in 10 mM DTT for 30 min at 37°C and then diluted to 100 nM in 5 mg/ml cytosol. This toxin solution or cytosol alone was added to the cells for 30 min at 37°C before the cells were washed in cold PBS and CTB-FITC (3 µg/ml in 1% BSA) was added for 15 min at 4°C. The cells were washed in cold PBS (3 × 5 min) before cytosol was added and warmed to 37°C for different time periods. The cells were fixed with cold methanol for 10 min, washed in cold PBS and then distilled water before being mounted, viewed, and photographed using a Zeiss Axiophot fluorescence microscope.

Endocytosis of CTB-HRP by permeabilized endothelial cells. RAEC were plated out in 96-well plates, and 48 h later they were permeabilized with 0.6 U/ml SLO as described above. The buffer was replaced with warm DMEM after which 100 µl CTB-HRP at 3.3 µg/ml (in 5 mg/ml cytosol) was added at 37°C. At the end of the incubation, the cells were washed three times in cold DMEM to remove unbound ligand and once in ELISA substrate buffer (0.05 M Na₂HPO₄, 0.025 M citric acid). Fifty microliters of 0.05% saponin was added to each well to lyse the cells followed by 100 µl OPD reagent (14 mg o-phenylenediamine dihydrochloride and 10 µl of 30% H₂O₂ in 12 ml substrate buffer). The reaction was stopped by the addition of 100 µl 4 M H₂SO₄, and the optical densities were read at 450 nm using a Thermo Max microplate reader (Molecular Devices).

Statistical analysis. All data values are presented as the means ± SD. When appropriate, a standard one-way analysis of variance was used to determine the significance of the difference of the means.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

VAMP colocalizes with caveolin in caveolae of endothelium. Sequential gold labeling of caveolae with the caveolin antibody and Cl-69.1 antibody for VAMP-2 was performed on rat lung ultrathin tissue cryosections (Fig. 1, A and B) and BAEC cells (Fig. 1, C and D) using different-sized gold labels (5 nm for Cl-69.1 and 10 nm for caveolin). Both labels could be seen clearly associating with the membranes of caveolae near the plasma membrane proper and also with those located deeper in the cytoplasm. VAMP-2 and caveolin labeling were also occasionally observed on the plasma membrane close to caveolae. Gold particles of both 5 and 10 nm were often seen labeling uncoated vesicular structures appearing free in the cytoplasm and not directly attached to the plasma membrane; less VAMP labeling of the epithelium and epithelial membranes was observed. Immunolabeling of rat lung sections with the VAMP-2 and caveolin antibodies separately also showed clear labeling of the caveolae with little labeling of the linear plasma membrane. Morphometric analysis of the distribution of the VAMP-2 label as shown in Table 1 revealed that almost 70% of the gold labeling of VAMP was detected on caveolae. Control experiments using nonspecific normal mouse IgG or no primary antibody in the immunolabeling procedures resulted in little to no labeling (data not shown). As in past work (25) antibodies specific for VAMP-1 showed no specificity for the caveolae. Finally, significant colocalization of VAMP and caveolin was also evident in cultured endothelial cells by immunofluorescence microscopy (data not shown).

View larger version (121K): [in this window] [in a new window]   Fig. 1.   Immunogold labeling of caveolin and vesicle-associated membrane protein (VAMP)-2 in endothelial caveolae. Immunogold labeling was performed on ultrathin cryosections of rat lung tissue (A and B) and cultured bovine aortic endothelial cells (BAEC; C and D). These sections were labeled with antibodies by sequential immunogold labeling of both caveolin (10 nm gold) and VAMP-2 (5 nm gold) (see MATERIALS AND METHODS). Location of plasma membrane is indicated (pm), and arrowheads show some of the labeled caveolae. Bar, 91 nm.

Characterization of cholera toxin uptake in permeabilized endothelial cells. To study the potential effects of VAMP-specific neurotoxins on the intracellular trafficking of caveolae in endothelial cells, a permeabilized cell system that we previously developed using cultured endothelial cells (30) was utilized to introduce the toxin molecules into the cytoplasm while maintaining the structural and functional integrity of the cells. We have previously shown in this system by both fluorescence and electron microscopy that intact and permeabilized endothelial cells could internalize CTB via caveolae. Electron microscopy showed that CTB conjugated to colloidal gold particles (CTB-Au) rapidly entered caveolae on the endothelial cell surface, and as early as 10 min was found in endosomes and multivesicular bodies within the cells (30). The B fragment of cholera toxin was chosen as a caveolar ligand due to its specificity and availability in free and conjugated form. Several laboratories have shown in various cell types, including endothelial cells, that CTB subunit binds GM₁ preferentially in caveolae for endocytosis and delivery to endosomes (16, 19, 30). GM₁ and VAMP have both been found to be enriched in isolated endothelial caveolae (25, 29). Thus we used CTB as a probe to follow the trafficking of caveolae in endothelial cells after the cleavage of caveolar VAMP by VAMP-specific neurotoxins.

View larger version (1518K): [in this window] [in a new window]   Fig. 2.   Time course of the binding and internalization of cholera toxin B (CTB) by permeabilized cells. Permeabilized rat aortic endothelial cells (RAEC) were incubated at 37°C with CTB-horseradish peroxidase (HRP) for indicated times and after washing CTB-HRP associated with the cells was quantified as described in MATERIALS AND METHODS. A: amount of cell-associated CTB-HRP in permeabilized cells is compared with that in control, unpermeabilized cells. Each point represents mean of 3 determinations with SD given as error bars. * Significantly different from each other, P = 0.01. B: accumulation of CTB-HRP by permeabilized cells is compared in presence of DMEM-1% BSA at 4°C and at 37°C vs. DMEM with rat lung cytosol at 37°C. These data represent 3 different experiments, n = 3. Results are given as mean value of 3 determinations within each experiment with SD shown as error bars. * Significantly different from each other, P = 0.01. The 4°C value is significantly different from DMEM value, P < 0.01, and cytosol value, P < 0.01.

NEM inhibits CTB-HRP uptake. Because our previous work showed that NEM treatment of intact cultured endothelial cells greatly inhibited the endocytosis or transcytosis of ligands preferentially bound within caveolae (26), we decided to test our permeabilized cell system additionally using NEM. We found that the permeabilized endothelial cells responded to NEM treatment similarly to intact cells. A 2-min treatment with 1 mM NEM significantly reduced the cells ability to internalize CTB-HRP at 37°C as shown in Fig. 3. In the control untreated cells, the amount of cell-associated CTB-HRP reached an equilibrium after 10 min with very little increase observed up to 30 min. In the NEM-treated cells, very little of the CTB-HRP appeared to be internalized by the cells as the levels only just superceded those of the 4°C controls; by 20 min the amount of CTB-HRP associated with the NEM-treated cells was decreased by about 90% relative to the controls. Therefore the permeabilized endothelial cells in these experiments responded to NEM treatment similarly to intact cells and even vascular endothelium in situ (26). Cells incubated in the presence and absence of NEM at 4°C, a temperature that blocks all forms of endocytosis, showed very similar amounts of cell-associated CTB over the 30-min time course, indicating that NEM treatment did not influence the amount of CTB bound to the cell surface. When the amount of cell-associated CTB after NEM treatment at 37°C is compared with the curves obtained at 4°C it can be seen that after NEM treatment there was little signal beyond cell surface binding consistent with an almost total inhibition of endocytosis of the CTB.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   N-ethylmaleimide (NEM) inhibits CTB uptake. Time course of CTB-HRP internalization at 37°C and 4°C was compared in permeabilized RAECs exposed for 2 min to 1 mM NEM with control cells that had no NEM treatment. , NEM, 37°C; , +NEM, 37°C; , NEM, 4°C; , +NEM, 4°C. These data represent 3 different experiments, n = 3. Results are given as mean value of 3 determinations within each experiment with SD shown as error bars. * Significantly different from treatment in presence of NEM at 37°C (P < 0.001). The 4°C controls were not significantly different from each other (P > 0.1).

VAMP cleavage by botulinum toxins in permeabilized cells. Botulinum neurotoxins are zinc endopeptidases, each of which cleaves a specific peptide bond within the target molecule. Botulinum B, D, F, and G specifically attack VAMP, thereby apparently hindering its function as a vesicular docking receptor in synaptic vesicle targeting and fusion within the cell (15, 17). Experiments in which intact endothelial cells were incubated with unreduced botulinum D toxin for up to 12 h and then assessed for VAMP cleavage by immunoblotting, revealed that unlike neuronal cells, the endothelial cells were insensitive to the toxin (data not shown), perhaps due to an insufficient number of endocytosing cell surface receptors for the toxin. Thus the cultured cells had to be permeabilized with SLO, which was bound to the plasma membranes at 4°C, after which unbound SLO was washed away before raising the temperature to 37°C to trigger pore formation. This strategy avoids permeabilization of intracellular membranes (13) and undue impairment of cell transport and function as previously characterized.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Proteolysis of VAMP in cultured endothelial cells by botulinum D toxin. A: permeabilized RAEC were incubated for 30 min with 0-100 nM botulinum D toxin as described in MATERIALS AND METHODS or with 100 nM botulinum F () or botulinum A toxin, and remaining VAMP protein was detected by immunoblotting. B: permeabilized RAEC were incubated with 100 nM botulinum D for 5, 10, 20, and 30 min. After treatment, cell lysates were analyzed by immunoblotting with Cl-69.1, VAMP-2 specific monoclonal antibody and monoclonal antibody against -actin. Signals detected for VAMP-2 were quantified by densitometry. Data given are representative of 2 separate experiments.

VAMP-specific botulinum toxins inhibit CTB endocytosis. If VAMP-2 expressed on caveolae is important for the cellular trafficking of caveolar ligands, cleavage of VAMP-2 by neurotoxins might result in a demonstrable difference between CTB trafficking in control and toxin-treated cells. Having demonstrated VAMP cleavage in our permeabilized cell system, we investigated the effects of this cleavage in a functional assay measuring CTB endocytosis and accumulation within the cell. The ability of botulinum toxin-treated endothelial cells to endocytose CTB-HRP was investigated. Permeabilized cells were treated at 37°C with 100 nM botulinum D or F toxin for 30 min and then incubated with either ¹²⁵I-labeled CTB or CTB-HRP at 4°C or 37°C for 15 min. Figure 5 shows a reduction in ligand accumulated compared with untreated controls. Treatment with botulinum toxin D resulted in a 51% reduction in CTB cell association, whereas botulinum toxin F inhibited uptake by 63%. Treatment of cells with 100 nM botulinum toxin A had no inhibitory effect on CTB accumulation. In addition, botulinum treatments had no significant effect on the amount of CTB bound by the cells at 4°C (which was 32% of the total at 37°C over the 15-min time course). These findings demonstrated that the VAMP-specific neurotoxin treatments did not affect cell surface binding but did impede the ability of the cells to accumulate CTB at 37°C. If one subtracts the binding detected at 4°C from the total CTB associated at 37°C, then the inhibition of this estimate of internalization is 70-90%. Because of the presence of VAMP on caveolae, this inhibition of CTB accumulation in the cells at 37°C selectively by the toxins only cleaving VAMP, provides direct evidence for an important functional role for VAMP in transport mediated by caveolae.

View larger version (53K): [in this window] [in a new window]   Fig. 5.   Botulinum D and F but not A inhibit CTB endocytosis. Permeabilized endothelial cells were treated with 100 nM botulinum D, F, or A for 1 h and then incubated with CTB-HRP or 125I-labeled CTB at 37°C or 4°C in presence of cytosol. Cells were washed with cold DMEM before processing for HRP detection or gamma spectroscopy. Results are mean values obtained in 3 separate experiments and are expressed as percentage of amount of CTB internalized by (37°C) or bound to (4°C) non-toxin-treated cells. * Significantly different from A (P < 0.001). Each data point represents mean of 9 determinations.

Electron microscopy of botulinum D-treated endothelial cells. We next investigated the effects of VAMP cleavage on endothelial cell morphology. Transmission electron microscopy of cells treated with botulinum toxin D revealed the presence of accumulated caveolae-sized spherical vesicles as well as larger irregular vesicular structures inside the cell, mostly near the cell surface (Fig. 6, A and B) and sometimes deeper within the cell. In many instances, the vesicles had lost their spherical nature and had become elongated (see arrows in Fig. 6, A and B). The caveolae of non-toxin-treated cells, shown for comparison (Fig. 6C), are located at the cell surface as flask-shaped invaginations. In the botulinum toxin D-treated cells, such accumulation of vesicles near the plasma membrane may have resulted from the failure of budded caveolae to dock and fuse with their cognate target membranes, such as an endosome. These findings were consistent with the reduced accumulation of CTB in the cells as previously described. The lack of proper docking and fusion with endosomes would prevent normal ligand trafficking and block efficient delivery to endosomes, ultimately reducing overall cellular CTB accumulation and perhaps even recycling of caveolae back to the cell surface. Treatment of the cells with botulinum D did not otherwise appear to affect the structure of the cell monolayer or the integrity of the cell junctions.

View larger version (130K): [in this window] [in a new window]   Fig. 6.   Morphology of botulinum D-treated RAEC. Permeabilized RAECs were incubated with 100 nM botulinum D in rat lung cytosol for 30 min as described in MATERIALS AND METHODS and then processed for transmission electron microscopy. A: in botulinum D-treated cells, small vesicles as well as elongated vesicular structures (see arrows for examples) can be seen accumulating in cytoplasm. Note that few caveolae are obviously attached to plasma membrane. B: presence of elongated vesicular structures (arrows) and small vesicles near plasma membrane of botulinum D-treated cells. C: plasma membrane and associated caveolae (arrowheads) of permeabilized but non-toxin-treated cells. Bar, 120 nm.

Botulinum toxin prevents CTB delivery to intracellular organelles. Various morphological studies show that CTB binds preferentially to caveolae and with time accumulates in endosomes (16, 19). More recently we have shown by electron microscopy and fluorescence microscopy that intact and permeabilized cultured endothelial cells utilize caveolae to internalize CTB (18, 30). Here we use this system and fluorescence microscopy to assess visually the internalization of CTB-FITC by caveolae in the permeabilized cells. In these experiments CTB-FITC was bound at the cell surface at 4°C followed by washing at 4°C and then warming to 37°C to allow internalization of the specifically membrane-bound CTB. CTB-FITC bound at 4°C gave a fine punctate cell surface staining pattern (Fig. 7A). This punctate staining became more defined and intense on warming to 37°C for 5 min. After 20 min (Fig. 7B) and 30 min (Fig. 7C), the staining became predominantly intracellular and near the nucleus, indicative of significant internalization and delivery to endosomes as shown previously (18, 30).

View larger version (144K): [in this window] [in a new window]   Fig. 7.   Fluorescence microscopy of processing of CTB-FITC in control and botulinum-treated cells. Permeabilized RAEC were either treated with botulinum D toxin in rat lung cytosol for 30 min at 37°C (D, E, F) or with rat lung cytosol alone (A, B, C). Cells were washed and then incubated with CTB-FITC at 4°C for 15 min. After being washed, cells were either immediately fixed (A and D) or warmed to 37°C for 5 min (inset, A), 20 min (B and E), or 30 min (C and F) before being fixed and processed for fluorescence microscopy.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This study describes the identification of VAMP-2 in endothelial cells in culture and in rat lung endothelium and demonstrates that it is an essential protein for the targeted intracellular trafficking of cholera toxin, a caveolar ligand. The immunogold localization of VAMP-2 in rat lung endothelium reveals that it is located on the endothelial caveolae, both at the cell surface and within the cell. The directed delivery of CTB by caveolae to intracellular organelles such as endosomes or multivesicular bodies appears to depend on the presence of intact VAMP. When cleaved by VAMP-specific neurotoxins, caveolae lack VAMP and thereby may lose their ability to bind VAMP's cognate t-SNARE, which appears necessary for the completion of delivery by the docking and fusion of caveolae with an intracellular endosomal compartment.

The development of the permeabilized cell system for endothelial cells has enabled us to take advantage of the specific cleavage properties of clostridial neurotoxins. The pores created in the plasma membrane by SLO treatment are large enough to allow the passage of macromolecules in and out of the cells but too small to allow the loss of vesicular components (60-100 nm) from the cells. The protocol used was a modification of that of Miller and Moore (13, 14), which was developed in CHO cells to study the constitutive secretion via the biosynthetic pathway. In these studies, normal cell morphology, function, and kinetics of cellular processes was observed for at least 90 min after perforation of the plasma membrane. In our experiments similar observations were made when comparing intact and permeabilized cells. The process of endocytosis of ligands via caveolae appeared normal; the kinetics and degree of CTB-Au internalization by the cells were very similar (30) and that of CTB-HRP showed a similar time course and only a diminution of 20-30% in the observed uptake. This reduction was significantly redeemed by the addition of cytosol to the incubation medium. These findings are similar to our previous findings (30) and other reports comparing endocytosis or exocytosis in the permeabilized cell systems (13, 33) and affirm the suitability of the system for the experiments described.

The striking inhibition of CTB internalization by NEM in the permeabilized endothelial cells mirrors the effect seen in intact cells (26) and is consistent with the normal functioning of these cells. NEM has been reported to inhibit vesicular trafficking in cells by alkylating NSF, which is an essential component of the vesicular trafficking pathway. Inhibition of the endocytosis of a caveolar ligand by NEM supports the notion that NSF is required for caveolar transport. However, NEM can modify many proteins and is not specific for NSF (26). Also, other NSF-like proteins involved in vesicular trafficking are NEM sensitive (1, 20, 22). We were interested therefore in a more specific manipulation, possibly by focusing on other proteins of the docking and fusion machinery mediating vesicular trafficking.

VAMP is a vesicle-associated protein integral to the SNARE complexes mediating docking and fusion (3, 34, 36). Cleavage of VAMP expressed on synaptic vesicles of rat brain by serotypes B, D, F, and G of botulinum neurotoxins has been shown to be specific and to block neurotransmitter exocytic release by preventing vesicular docking and fusion with the plasma membrane (15, 39). It seemed possible that analogous mechanisms were operative in caveolae-mediated trafficking of ligands in endothelial cells because VAMP had already been found to be enriched in endothelial cell caveolae isolated from rat lungs (25). In this report, our electron microscopy data show that VAMP-2 is expressed in intact tissue on the plasma membrane and on caveolae, predominantly in lung vascular endothelial cells in vivo. Morphometric analysis of membrane labeling in the endothelium showed almost 70% of the label associated with the caveolae and 21.5% labeling the plasma membrane. The level of plasma membrane labeling may reflect the exchange of v-SNAREs onto the target membrane during the docking and fusion stages of vesicular transport.

The demonstration by SDS-PAGE of VAMP cleavage by botulinum toxin D in the endothelial cells (Fig. 4) implicated VAMP-2 as the protein responsible for the botulinum toxin-impaired caveolar trafficking and accumulation of the CTB subunit seen in the fluorescence microscopy study (Fig. 7). If VAMP-2 expressed on caveolae was acting in docking and/or fusion, possibly as a v-SNARE in caveolar trafficking to other organelles, cleavage of the VAMP-2 would prevent accumulation of ligand inside the cell by reducing delivery to target endosomes. Without a v-SNARE (VAMP) to bind its cognate t-SNARE on the target organelle, the efficacy of docking and fusion would be reduced sufficiently to cause a significant diminution in cellular accumulation of ligand in large intracellular target compartments such as endosomes. Hence, the passage of the caveolar ligand from an early plasmalemma-associated or free small cytoplasmic compartment to a further stage in the endocytic pathway would be inhibited. Consistent with these observations, electron microscopy of endothelial cells treated with botulinum D toxin showed an accumulation of small and elongated vesicles near but apparently no longer attached to the plasma membrane. This was in marked contrast to the normal morphology of the control endothelial cells and would be expected to account for a noticeable change in membrane trafficking. This accumulation of caveolae-sized vesicles seemingly free in the cytoplasm is consistent with the effective budding of caveolae and a block in their targeted delivery to and fusion with intracellular endosomal compartments.

To quantify the effects of VAMP-2 cleavage on the endocytosis and accumulation of CTB, permeabilized toxin-treated cells were assayed for their ability to internalize the ligand compared with untreated controls. We had planned to investigate the effect of VAMP-2 cleavage on the ability of the cells to degrade internalized ¹²⁵I-labeled CTB. However, experiments designed to study the degradation of ¹²⁵I-labeled CTB by both NRK fibroblasts and RAEC cells failed to detect any degradation of the ¹²⁵I-labeled CTB even after 37°C incubations as long as 4 h. This observation can be explained by the absence of or very slow trafficking of the toxin to the lysosomal system after internalization by cells. This phenomenon has been observed for cholera toxin by other investigators (9) in several cell types. It appears that CTB is sequestered into an endosome-like compartment where it may remain for prolonged periods of time before its eventual passage to and degradation in the lysosomes. Our past work showed significant CTB accumulation in endosomes and then multivesicular bodies (30). Here we found that both botulinum toxins D and F, which cleave VAMP at adjacent sites on the molecule (39), significantly impaired the delivery of CTB to these intracellular compartments, which reduced the overall uptake of CTB-HRP by cultured endothelial cells. This reduction in ligand accumulation in the toxin-treated cells appears to be caused by targeting inefficiency disrupting normal delivery and thus accumulation within endosomes. It is not due to the depletion of cell surface receptors for the CTB, as botulinum treatment did not appear to change the amount of CTB cell surface binding over the time course of the experiment. It is likely as in other vesicular trafficking pathways that an equilibrium exists between caveolae incoming to the endosomes and a recycling component returning to the plasma membrane. Thus, through an apparent process of maintaining membrane balance, impairment of the incoming delivery may affect the recycling component, which ultimately impairs overall uptake. It appears reasonable to predict that such mechanisms exist to regulate the amount of cargo in intracellular compartments and also the percentage of plasma membrane that is internalized at any given time.

In the fluorescence microscopy studies of permeabilized endothelial cells endocytosing CTB-FITC, the ligand specifically bound at the cell surface at 4°C was internalized by the cells after warming to 37°C. Internalized CTB could be visualized as punctate structures inside the cells, the distribution of which became predominantly perinuclear after 20 min. This is identical to the distribution of the ligand seen in intact cells (16, 18, 19, 30). Furthermore, we found that the degree of endocytosis of CTB could be enhanced in permeabilized cells by incubation of the cells with ligand in the presence of cytosol, a qualitative observation subsequently supported by biochemical data that demonstrated potentiation of CTB internalization in the presence of cytosol to levels similar to that seen in intact cells. This is likely to be due to the replenishment of key cytosolic factors for caveolar trafficking, such as dynamin (18), some of which may have been lost from the permeabilized cells by the washing procedures involved in the processing of the cells.

It was clear that after treatment of the endothelial cells with botulinum toxin D, the trafficking of the CTB was impaired and the cells seemed unable to perform their normal targeted delivery of the CTB from the plasma membrane to endosomes within the cells. Treatment of the cells with the botulinum toxin appeared not to inhibit the cell surface binding of CTB in the caveolae; the budding of caveolae is probably not affected (an increase in the number of apparently free vesicles was found). However, ligand delivery to and accumulation within endosomal compartments were severely inhibited. These observations suggest that movement of CTB into the cell occurred but that the subsequent normal delivery to or fusion with recipient organelles (which results in a strong punctate perinuclear staining), was compromised by the neurotoxin treatment. Electron microscopy shows the accumulation of small vesicular structures inside the botulinum-treated but not control cells, which is consistent with an interruption in the dynamics of the vesicular trafficking pathway caused by the botulinum toxin treatment. This observation, coupled with the biochemical demonstration of VAMP-2 cleavage in endothelial caveolae by neurotoxins specific for VAMP-2, indicates that VAMP-2 is important for the maintenance and normal functioning of the caveolar trafficking pathway in endothelial cells.

Caveolae contain key elements constituting molecular machinery facilitating dynamic and targeted transport in cells. So far, this ranges from caveolin-dependent formation of caveolae to their dynamin-mediated fission from the plasma membrane to form free transport vesicles to now the discovery of VAMP-dependent targeted intracellular delivery. We find here that the normal intracellular trafficking of ligands preferentially transported by caveolae is impaired by the cleavage of VAMP. The demonstration of VAMP on caveolae supports the hypothesis that the transport of caveolar ligands requires caveolae to maintain a mechanism for target recognition to ensure fidelity in ligand transport. Such a requirement is necessary for vesicles participating in docking and fusion maneuvers and is supportive of the description of caveolae as dynamic transport vesicles in protein transport.