Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Botulinum toxin, the most potent biological poison known, is responsible for botulism, a fatal paralytic disease of neuromuscular transmission (Burgen et al., 1949; Kao et al., 1976; van Ermengem, 1979; Simpson, 1981). The toxin specifically targets peripheral cholinergic nerve terminals. It does not cross the blood-brain barrier due to its large size (150 kDa), and thus, has no known action in the CNS. Exposure to botulinum toxin occurs through accidental ingestion of contaminated food, through wound contamination, or through its clinical use in the treatment of neuromuscular disorders such as blepharospasm (Scott et al., 1973; Brin, 1997; Jankovic and Brin, 1997). Currently, infant botulism is the most commonly reported form of botulinum poisoning. The objective of the current study was to determine whether the intracellular targets that mediate the mechanism of action of botulinum toxin at the mammalian neuromuscular junction are the same as those identified in nontarget tissues.

It has been demonstrated that botulinum toxin, a zinc-dependent metalloprotease, poisons neurotransmission through the cleavage of synaptic proteins that mediate transmitter release (Schiavo et al., 1992a, 1994a, 2000; Simpson et al., 1993). This intracellular action was elucidated by a series of elegant biochemical studies that used nontarget tissues such as subcellular brain preparations, recombinantly expressed proteins, and cell cultures. To date, three synaptic proteins have been identified as substrates for the proteolytic action of the seven botulinum toxin serotypes (A-G) in nontarget tissues (Schiavo et al., 1992b, 1993a,b, 1994b; Blasi et al., 1993a,b). They are SNAP-25, synaptobrevin II, and syntaxin I. These proteins are essential components of the synaptic vesicle-plasma membrane fusion complex (e.g., SDS-resistant SNARE complex), and their cleavage results in blockade of transmitter release (Söllner et al., 1993). The seven toxin serotypes, however, differ somewhat in their affinity for intracellular substrate. For instance, in rat brain, serotypes A and E cleave SNAP-25; serotype C cleaves syntaxin I; and serotypes B, D, F, and G cleave synaptobrevin II (Schiavo et al., 1992b, 1993a,b, 1994b; Blasi et al., 1993a,b). Serotype C has also been shown to cleave SNAP-25 in mouse spinal cord and bovine chromaffin cells (Foran et al., 1996; Williamson et al., 1996). Recently, Raciborska et al. (1998) examined the effect of botulinum toxin poisoning of frog muscle on the localization of SNAP-25, synaptobrevin, and syntaxin using fluorescent imaging techniques. Their findings indicated that botulinum toxin altered the dynamic interaction of these three proteins at the frog neuromuscular junction, suggesting that they had been cleaved by the toxin. However, the identity of such intracellular targets at the mammalian neuromuscular junction has not been investigated. Botulinum toxin has no known action in the CNS, therefore, it must be empirically determined whether the toxin targets the same proteins at both sites. To achieve this, studies need to be performed in mammalian neuromuscular preparations, using the information gained from nontarget tissue studies as guidelines. However, the use of neuromuscular preparations for biochemical analyses has been hampered by the relatively low level of expression of neural proteins in such tissues. Using procedures similar to those followed for obtaining synaptosomes from brain, we have overcome this technical difficulty and have successfully examined the botulinum toxin substrates at the mammalian neuromuscular junction. To the best of our knowledge, this is the first demonstration of the intracellular targets of botulinum toxin at a recognized mammalian target site, the neuromuscular junction.

To achieve the objective of the present study, we used the mouse phrenic nerve-hemidiaphragm preparation, an established mammalian neuromuscular preparation that includes the relevant cholinergic target sites for botulinum toxin (Burgen et al., 1949; Simpson, 1981; Bandyopadhyay et al., 1987; Simpson et al., 1993). Enhanced chemiluminescent immunological detection methods were used to determine the presence of the putative target proteins SNAP-25, synaptobrevin II, and syntaxin I in the neuromuscular preparation. In addition, to determine whether these proteins serve as substrates for the toxin's proteolytic action in mammalian neuromuscular tissue, the intact neuromuscular preparation was incubated in botulinum toxin serotype A, C, or D and the effects of the toxins on the integrity of the putative intracellular substrates were examined. The results reported here demonstrate 1) that SNAP-25, synaptobrevin II, and syntaxin I proteins can be detected in a neuromuscular preparation; and 2) that they serve as selective substrates for botulinum toxin serotypes A, D, and C, respectively, at the mammalian neuromuscular junction.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Tissue Preparation. Hemidiaphragm and brain tissues were isolated from NIH Swiss adult male mice (30 g) following decapitation. All procedures were approved by the University's Institutional Animal Care and Use Committee. The hemi-preparation was removed from the mouse with muscle fibers remaining intact via connective tissue attachments on one end and bony attachments on the opposite end. In the first set of experiments to determine the presence of the putative substrates, the excised hemidiaphragm preparation was further modified as follows. After the initial isolation of the preparation, excess muscle tissue was carefully removed so that only tissue immediately surrounding the visible innervation zone of the phrenic nerve remained. The purpose of this modification was to minimize the potential contribution of non-neural proteins to the preparation, although clearly they could not be eliminated entirely. In the second set of experiments to determine the effects of toxin incubation on proteins, the hemidiaphragm preparation was not modified as described above, but remained intact during toxin incubation.

Synaptic Protein Preparation. Synaptic protein preparations from both brain and neuromuscular tissue were prepared according to previously published procedures with a few modifications (Coffield et al., 1997, 1999). All steps were performed on ice. The isolated tissues were minced in homogenization buffer containing 255 mM sucrose, 1 mM EDTA, protease inhibitor cocktail (Sigma, St. Louis, MO), and 20 mM Hepes (pH 7.4). The resulting suspension was homogenized with a hand-held electronic homogenizer (Omni 1000) at 15,000 rpm for 1 min. The homogenate was fractionated by centrifugation. Homogenates were initially spun at 1000g for 5 min using a tabletop centrifuge. The resulting supernatant (S₁) was then centrifuged at 10,000g for 10 min. The second supernatant (S₂) was centrifuged at 250,000g for 1 h in an Optima TLX ultracentrifuge (Beckman Coulter, Fullerton, CA) to obtain an enriched synaptosomal-containing fraction, P₃. The P₃ fraction was resuspended in homogenization buffer. Protein concentrations were determined by the modified Lowry method (Bio-Rad, Hercules, CA). Samples of the S₂, P₃, and S₃ fractions were resolved by SDS-polyacrylamide gel electrophoresis (1-75 µg of protein/lane).

Immunodetection. Subsequent to separation, proteins were electrophoretically transferred to polyvinylidene difluoride membranes in Tris-glycine transfer buffer. Blotted membranes were then blocked overnight (4°C) with 5% nonfat powdered milk in Tris-buffered saline. For identification of proteins, membranes were washed two times and incubated with the primary antibodies diluted in 0.5 to 1.0% milk. Following the primary antibody incubation, the membranes were washed and then incubated in a horseradish peroxidase-conjugated secondary antibody at room temperature. This incubation step was terminated with several washes and the immunoreactive protein bands were visualized using enhanced chemiluminescence (ECL Plus; Amersham Pharmacia, Arlington Heights, IL) according to manufacturer's instructions. Membranes were exposed to film (Hyperfilm-ECL) for times adequate to visualize chemiluminescent bands. Comparisons were made with known molecular weight standards.

Antibodies. For immunoblot detection, antibodies to SNAP-25 were purchased from Sternberger Monoclonals, Inc. (Baltimore, MD) and Sigma and used at dilutions ranging between 1:5,000 and 1:20,000. A rabbit polyclonal antibody to synaptobrevin II was purchased from Wako Chemicals (Richmond, VA) and used at dilutions ranging between 1:5000 and 1:8000. A monoclonal antibody to syntaxin I was purchased from Sigma and used at a working dilution of 1:8,000 to 1:10,000. Horseradish peroxidase-labeled secondary antibodies were purchased from ICN Pharmaceuticals, Inc. (Costa Mesa, CA) and Biosource (Camarillo, CA) and used at 1:5000. For immunocytochemistry, unlabeled goat anti-mouse IgG and goat anti-rabbit IgG were purchased from DAKO Corporation (Carpinteria, CA) and Sternberger Monoclonals, Inc. (Baltimore, MD), respectively, and used at dilutions ranging between 1:100 and 1:500. Peroxidase-antiperoxidase complex was purchased from Sternberger Monoclonals, Inc.

Controls. Samples of mouse brain tissue were processed simultaneously with samples of neuromuscular tissue and served as the positive control. Furthermore, as a control for nonspecific background and the immunoblotting technique, companion blots were processed without the primary and/or secondary antibodies.

Toxin Incubation of Hemidiaphragm. Botulinum toxin types A (toxicity: 2.0 × 10⁷ mouse LD₅₀ units/mg), C (toxicity: 0.04 × 10⁷ mouse LD₅₀ units/mg), and D (toxicity: 5.0 ×10⁷ mouse LD₅₀ units/mg) were purchased from Wako Chemicals. Intact control and toxin-treated phrenic nerve-hemidiaphragm tissues were incubated individually at 32 to 34°C in an oxygenated physiological solution of the following composition: 137.0 mM NaCl, 5.0 mM KCl, 1.8 mM CaCl₂, 1.0 mM MgSO₄, 24.0 mM NaHCO₃, 1.0 mM NaHPO₄, and 11.0 mM d-glucose. This solution also contained 0.01% gelatin as an auxiliary protein to prevent nonspecific adsorption of the toxin to the incubation chamber. In an effort to achieve a measurable effect in a reasonable period of time, botulinum toxin was added to the tissues at a final concentration of 10⁸ M. Although this concentration is higher than that often used in purely electrophysiological studies, it is consistent with concentrations used to assess substrate proteolysis in previously published studies. In electrophysiological experiments all three toxins caused paralysis of neuromuscular transmission at 40 to 45 min (S. Kalandakanond and J. A. Coffield, unpublished observations). As a control for toxin activity, some tissues were incubated in toxin at 4°C. At the end of the incubation period, unbound toxin was removed by several washes with cold physiological buffer. Following removal of the bony attachments, the tissues were placed in homogenization buffer and processed for immunodetection of the protein of interest as described. In addition, as a test of substrate specificity, blots were reprobed with a cocktail of antibodies to the other two proteins. Differences in protein immunoreactivity between control and toxin-treated tissues were determined by scanning densitometry (Scion Image, Scion Corporation, Frederick, MD).

Toxin Incubation of Brain Synaptosomes. Brain synaptosomes were prepared as described above. The toxin was treated with 10 mM dithiothreitol for 30 min at room temperature. Botulinum toxin types A, C, or D (10⁸ M final concentration) was added to 5 to 10 µg of the synaptosomal preparation and incubated for 6 h at 32°C. Control and toxin-treated samples were analyzed by SDS-polyacrylamide gel electrophoresis, blotted, and then probed with antibodies to SNAP-25, syntaxin I, and synaptobrevin II as described above.

Statistical Analyses. A minimum of three treated and three untreated tissues was used per experiment. The data are expressed as the percentage of difference in densitometry between untreated and treated tissues. Statistical differences were determined using one-tailed Student's t test. P values <0.05 were considered significant.

Immunocytochemistry. Tissues were fixed in 10% formalin or 4% paraformaldehyde. Fixed tissues were infiltrated with either sucrose for frozen sectioning, or paraffin for microtome sectioning. Sections were mounted on slides, and processed for peroxidase-antiperoxidase immunocytochemistry using the same primary antibodies that were used in immunoblot detection. Immunoreactivity was visualized with diaminobenzidine as the chromagen.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Putative Intracellular Targets of Botulinum Toxin Are Present in the Mouse Diaphragm. Immunoreactive bands corresponding to SNAP-25, synaptobrevin II, and syntaxin I monomers were present in the phrenic nerve-hemidiaphragm preparation (Fig. 1). Based on total protein concentration, all three proteins were concentrated in the synaptosomal-enriched P₃ fraction; they were also detectable, to a lesser extent, in the more dilute S₂ fraction. For SNAP-25 and syntaxin I, the immunoreactivity was confined to discrete bands corresponding to molecular masses of approximately 25 and 35 kDa, respectively. Synaptobrevin II was also present as a single band corresponding to a molecular mass of approximately 13 kDa. Comparisons with mouse brain illustrate the relatively low abundance of the proteins in the hemidiaphragm preparation. The existence of various isoforms of these three proteins has been reported for a variety of tissues in the rat, including skeletal muscle (Ralston et al., 1994; Volchuk et al., 1994; Sumitani et al., 1995; Jagadish et al., 1996; Aguado et al., 1999); however, specific reports of the existence of these three particular isoforms in mouse muscle are lacking. In the present study, immunocytochemical examination of mouse diaphragm using the same primary antibodies to SNAP-25, synaptobrevin II, and syntaxin I revealed no specific staining in individual muscle fibers (data not shown).

View larger version (48K): [in this window] [in a new window]   Fig. 1.   Detection of SNARE proteins in mouse phrenic nerve-hemidiaphragm. Synaptic proteins from mouse phrenic nerve-hemidiaphragm and mouse brain were isolated by centrifugation and resolved by electrophoresis. Syntaxin I, SNAP-25, and synaptobrevin II were immunologically detected in the P3 fractions of both tissues using specific antibodies and enhanced chemiluminescence (diaphragm, 30 µg/lane; brain, 0.5-1.0 µg/lane).

Incubation of Intact Hemidiaphragm Tissue with Botulinum Toxin Serotype A Results in Specific Proteolysis of SNAP-25. After detecting the presence of SNAP-25, synaptobrevin II, and syntaxin I protein in the mouse hemidiaphragm preparation, we were interested in determining whether these proteins served as substrates of botulinum toxin action at the neuromuscular junction. In this series of experiments, the effects of incubation of the intact hemidiaphragm preparation in botulinum toxin serotypes A, C, and D on SNAP-25, syntaxin I, and synaptobrevin II immunoreactivity were examined. Intact phrenic nerve-hemidiaphragm preparations were used to further ensure that any proteolysis of protein resulted from receptor-mediated entry of toxin into nerve terminals, and not from nonspecific diffusion of toxin into a cut muscle fiber. Treatment of the intact preparation with botulinum toxin serotype A resulted in cleavage of the plasma membrane-associated protein SNAP-25, as demonstrated by a mean reduction of 67.43% in parent SNAP-25 immunoreactivity in the P₃ fraction compared with untreated tissue (Fig. 2A, top). The amount of proteolysis was dependent on the length of the incubation period with little change noted at 4 h, substantial proteolysis occurring at 6 h, and little additional proteolysis occurring after 6 h (Fig. 2B). Examination of the S₃ fraction revealed the presence of a second, slightly faster migrating band of SNAP-25 immunoreactivity, representing the remaining large fragment of SNAP-25 after cleavage (Fig. 2A, bottom). The much smaller (1-2 kDa) cleaved fragment of SNAP-25 was not observed. Incubation of this preparation in serotype A had little effect on the immunodetection of the other putative toxin substrates, synaptobrevin II and syntaxin I (Fig. 2C). Furthermore, SNAP-25 immunoreactivity in tissues incubated in serotype A at 4°C as a control for toxin activity was not different from controls (data not shown). The effects of incubation of mouse brain synaptosomes in serotype A for 6 h were examined for comparison (Fig. 5). As in the hemidiaphragm preparation, complete cleavage of SNAP-25 was not achieved in the brain preparation by 6 h. However, partial cleavage was observed as indicated by the appearance of a second, slightly faster migrating band of SNAP-25 immunoreactivity (~24kDa), representing the larger cleaved fragment of SNAP-25. As with the hemidiaphragm preparation, no effect on syntaxin I or synaptobrevin II immunoreactivity was observed in the serotype A-treated brain preparations.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Proteolysis of SNAP-25 in mouse phrenic nerve-hemidiaphragm by botulinum serotype A. A, immunoblot of synaptic proteins isolated from intact phrenic nerve-hemidiaphragm preparation treated with (+) or without () serotype A (108 M) for 6 h. Top, chemiluminescent immunodetection revealed a substantial reduction in SNAP-25 immunoreactivity in the P3 fraction of the toxin-treated preparation at 6 h; 25 µg/lane. Bottom, chemiluminescent immunodetection revealed the appearance of a second, lower band of SNAP-25 immunoreactivity in the S3 fraction of the toxin-treated preparation at 6 h; 75 µg/lane. B, intact phrenic nerve-hemidiaphragm tissues were incubated with botulinum serotype A (108 M) for 4, 6, and 20 h. Proteolysis of SNAP-25 was time-dependent, with no significant changes detected in the 4-h treatment, and substantial loss of SNAP-25 immunoreactivity at 6 and 20 h compared with untreated tissues. Data are expressed as a percentage of control immunoreactivity. **P < 0.01, n = 3 to 6 tissues. C, toxin-induced proteolysis was substrate-specific with little change in syntaxin I and synaptobrevin II immunoreactivity at 4 and 6 h. SNAP-25, ; syntaxin, ; synaptobrevin, . **P < 0.01, n = 3 to 6 tissues.

Botulinum Toxin Serotype D Results in Proteolysis of Synaptobrevin II. Exposure of the intact hemidiaphragm preparation to botulinum toxin serotype D resulted in substrate-specific proteolysis of the vesicular protein synaptobrevin II (Fig. 3B). The loss of synaptobrevin II immunoreactivity occurred much faster than the loss of SNAP-25 observed with serotype A treatment. At 1 h of incubation, approximately 33% of the synaptobrevin II had been cleaved by serotype D, and by 3 h of incubation 72% of the synaptobrevin II had been lost (Fig. 3, A and B). The effects of toxin type D on SNAP-25 and syntaxin I immunoreactivity were also examined. Although examination of the synaptic membrane fraction (P₃) revealed a mean reduction of 33% in the immunoreactivity of membrane-associated SNAP-25 (Fig. 3B, top), this reduction was not statistically significant. Furthermore, total SNAP-25 immunoreactivity was unchanged in the S₂ fraction, which contains both cytosolic and membrane proteins (Fig. 3B, bottom), confirming the serotype specificity of toxin action on synaptobrevin II. Syntaxin I immunoreactivity was unchanged in either fraction.

View larger version (32K): [in this window] [in a new window]   Fig. 3.   Proteolysis of synaptobrevin II in mouse phrenic nerve-hemidiaphragm by botulinum serotype D. A, intact phrenic nerve-hemidiaphragm tissues were incubated with botulinum serotype D (108 M) for 1, 3, and 5 h. Proteolysis of synaptobrevin II was time-dependent, with some change detected in the 1-h treatment, and substantial loss of synaptobrevin II immunoreactivity at 3 and 5 h. Data are expressed as a percentage of control immunoreactivity. **P < 0.01, n = 3 to 6 tissues. B, immunoblot of intact phrenic nerve-hemidiaphragm preparation treated with (+) or without () serotype D (108 M) for 3 h. Top, chemiluminescent immunodetection revealed a substantial reduction in synaptobrevin II immunoreactivity in the P3 fraction of the toxin-treated preparation at 3 h; 35 µg/lane. Bottom, chemiluminescent immunodetection of SNAP-25 and syntaxin I in the S2 fraction of the serotype D-treated tissue at 3 h; 40 µg/lane.

Botulinum Toxin Serotype C Results in Proteolysis of Syntaxin I. Incubation of the intact hemidiaphragm preparation in botulinum serotype C produced selective proteolysis of the plasma membrane protein syntaxin I (Fig. 4A). Similar to the proteolysis of SNAP-25 by serotype A, proteolysis of syntaxin I by serotype C was characterized by a prolonged latent period. Furthermore, the loss of syntaxin immunoreactivity was slower than the proteolysis of SNAP-25 and synaptobrevin II by serotypes A and D, respectively. Only 14% loss of syntaxin immunoreactivity was evident at 4 h (Fig. 4B). By 6 h, 30% of the syntaxin I immunoreactivity was lost, whereas a 56% reduction was observed at 20 h. Concurrent with the loss of the parent immunoreactive band in the membrane-bound P₃ fraction at 6 h was the appearance of a slightly faster migrating band of syntaxin immunoreactivity in the S₂ fraction (Fig. 4C). This lower band of syntaxin I immunoreactivity represents the cleaved fragment of syntaxin that was no longer membrane bound. No effect of serotype C on SNAP-25 immunoreactivity was evident in this preparation in the time frame examined. Incubation of mouse brain synaptosomes in serotype C for 6 h resulted in a decrease in the size of the parent syntaxin I band (~21.3%) and the appearance of two smaller cleavage bands (Fig. 5). Furthermore, neither SNAP-25 nor synaptobrevin II immunoreactivity was altered in the serotype C-treated brain preparations.

View larger version (38K): [in this window] [in a new window]   Fig. 4.   Proteolysis of syntaxin I in mouse phrenic nerve-hemidiaphragm by botulinum serotype C. A, immunoblot of P3 fractions from intact phrenic nerve-hemidiaphragm preparations treated with (+) or without () serotype C (108 M) for 6 h. Chemiluminescent immunodetection revealed a small but significant reduction in syntaxin I immunoreactivity in the toxin-treated preparation at 6 h; 20 µg of protein per lane. B, intact phrenic nerve-hemidiaphragm tissues were incubated with botulinum serotype C (108 M) for 4, 6, and 20 h. Proteolysis of syntaxin I was time-dependent, with no significant changes detected in the 4-h treatment, and an increased loss of syntaxin I immunoreactivity at 6 and 20 h compared with untreated tissues. Data are expressed as a percentage of control immunoreactivity. **P < 0.01, n = 3 to 6 tissues. C, immunoblot of S2 fractions from intact phrenic nerve-hemidiaphragm preparations treated with (+) or without () serotype C (108 M) for 6 h. Chemiluminescent immunodetection revealed the appearance of a second smaller band of syntaxin I immunoreactivity in the toxin-treated preparation at 6 h; 20 µg/lane.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Effects of botulinum serotypes A, D, and C on toxin substrates in mouse brain. Immunoblot of P3 fractions from mouse brain synaptosomal preparations treated with (+) or without () botulinum toxin (108 M) for 6 h; 2 µg/lane. Left, serotype A; middle, serotype D; right, serotype C. Chemiluminescent immunodetection revealed serotype-specific proteolysis of the parent substrate bands, with the appearance of cleavage fragments in the serotype A- and serotype C-treated preparations.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

A combination of pharmacological and electrophysiological investigations using the phrenic nerve-hemidiaphragm preparation has led to the development of a model of botulinum toxin action (Simpson, 1981). According to this model, botulinum toxin poisons neurotransmission by a multistep process that begins with binding of the toxin to a nerve terminal receptor, and ultimately culminates in the proteolytic inactivation of an intracellular substrate. Based on a number of biochemical studies, SNAP-25, synaptobrevin II, and syntaxin I were proposed as substrates of the intracellular action of botulinum toxin in a variety of species (Schiavo et al., 1992b, 1993a,b, 1994b; Blasi et al., 1993a,b; Williamson et al., 1996; Raciborska et al., 1998). In support of this, we have previously shown that serotype B cleaves synaptobrevin II from frog brain and that serotype C cleaves syntaxin I from human brain (Coffield et al., 1997, 1999). Although compelling, it is noteworthy that, with the exception of the work of Raciborska et al. (1998), these studies were conducted with nontarget tissues such as brain synaptosomal preparations, neuroendocrine cell lines, primary CNS cell cultures, and recombinantly expressed proteins, even though a clinically relevant action for this toxin in brain or neuroendocrine tissues has not been demonstrated. The use of nontarget tissues was necessitated by the perception that the expression of these proteins in a target tissue, such as a neuromuscular preparation, would be below detection limits. However, in the present study we have overcome this difficulty and have successfully examined the substrates of botulinum toxin A, C, and D at the mammalian neuromuscular junction. Thus, we demonstrate that 1) SNAP-25, synaptobrevin II, and syntaxin I, the botulinum toxin substrates in nontarget tissues, are present in the mouse phrenic nerve-hemidiaphragm, an established mammalian toxin target tissue; and 2) incubations of the intact mouse phrenic nerve-hemidiaphragm in botulinum toxin serotypes A, C, and D result in selective proteolysis of SNAP-25, syntaxin I, and synaptobrevin II, respectively. These data confirm that the botulinum toxin substrates initially identified in nontarget tissues are also substrates in a recognized mammalian target tissue. In the studies reported herein we did not address the relationship between substrate proteolysis and neuromuscular paralysis; however, this relationship has been examined in another study and communicated in a separate report (Kalandakanond and Coffield, 2001).

Although we cannot say with certainty that all of the observed immunoreactivity represents protein of nerve terminal origin, several observations suggest that the majority of immunoreactivity originates from nerve tissue. First, an intact preparation was used to eliminate the possibility of toxin gaining intracellular access to proteins of muscle origin. Second, the toxin-mediated proteolysis was time-dependent, demonstrating an initial latent period that is characteristic of botulinum toxin-induced paralysis observed in electrophysiological studies. This latent period is indicative of the time necessary for the toxin to bind its receptor, become internalized, and reach its intracellular substrate. Although the latent period was somewhat longer than observed in most electrophysiological studies, this can be explained, in part, by the fact that in this particular study the tissue preparation was minimally stimulated (0.3 Hz). Higher rates of stimulation are known to enhance toxin entry and decrease the length of the latent period, as well as to enhance the availability of at least one substrate, synaptobrevin II, to toxin cleavage (Hughes and Whaler, 1962; Hua and Charlton, 1999). Third, the lack of specific muscle cell immunoreactivity on histological sections of diaphragm muscle strongly suggests that the majority of the immunoreactivity detected on immunoblots of the preparation originated from nerve tissue. Fourth, we demonstrate in a separate study that preincubation of the tissues in the universal botulinum toxin antagonist Triticum vulgaris lectin inhibited serotype A-mediated proteolysis of SNAP-25, indicating that the proteolysis by serotype A was dependent upon a toxin-receptor-mediated event (Kalandakanond and Coffield, 2001). Previous studies have reported that receptors for botulinum neurotoxin are not present on muscle cells (Black and Dolly, 1986a,b).

The proteolytic action of the toxin serotypes was substrate-specific as indicated by the findings that 1) each serotype cleaved only one of the three proteins; 2) the amount of substrate loss varied for each serotype; and 3) the cleaved fragments of SNAP-25 and syntaxin I differed in molecular weight, and were comparable in size to those observed in the mouse brain preparations, as well as to those previously reported by others (Blasi et al., 1993a,b; Williamson et al., 1996). Interestingly, the loss of synaptobrevin II immunoreactivity occurred much faster than the loss of either SNAP-25 or syntaxin I. Two factors potentially contribute to these findings. First, the reported LD₅₀ of the serotype D used in this study was considerably lower than the LD₅₀ values for serotypes A and C. Although all serotypes were tested at the same concentration, this difference in serotype potency may partially explain the temporal difference in substrate proteolysis. A second contributing factor is a potentially greater accessibility of serotype D to "cleavable" synaptobrevin II. Before cleavage can proceed, the substrate must have both binding and cleavage sites readily available (Pellizzari et al., 1996; Otto et al., 1997; Washbourne et al., 1997). The availability of these sites and, thus, the susceptibility to toxin action, are dependent upon the conformational state of the particular protein. For example, it has been reported that the majority of synaptobrevin II is found in an uncomplexed state (Otto et al., 1997). In this state the SNARE proteins have very little structure, making them much more susceptible to toxin cleavage than multimeric forms. Conversely, a larger population of SNAP-25 and syntaxin I may exist as heterodimers (Höhne-Zell and Gratzl, 1996) and also readily form SDS-resistant heterotrimeric complexes with synaptobrevin II (Hayashi et al., 1994). These multimeric complexes are more resistant to toxin action, probably because complex formation induces conformational changes that cause one or more of the required sites to be unavailable to the toxin. These differences in conformational states may account, in part, for the greater proteolysis of synaptobrevin II at a given time point in this study.

The toxin-induced proteolysis observed in the intact hemidiaphragm was qualitatively similar to that observed in the mouse brain synaptosomal preparations. However, some differences were noted. For instance, in the botulinum serotype C-treated brain tissue, two cleavage fragments of syntaxin I were detected, compared with a single cleavage fragment in the hemidiaphragm preparation. The syntaxin I antibody used in this study recognizes both syntaxin 1A and 1B. Both syntaxin 1A and 1B are found in the brain and are cleaved by botulinum toxin serotype C. Syntaxin 1A is cleaved between Lys₂₅₃ and Ala₂₅₄, whereas syntaxin 1B is cleaved between Lys₂₅₂ and Ala_253. Thus, it is likely that the higher molecular weight cleavage fragment noted on the brain immunoblots originates from syntaxin 1A, and the lower from 1B. This was supported by the disappearance of the lower band when brain immunoblots were probed with a syntaxin antibody that recognized only syntaxin 1A (data not shown). Conversely, in the hemidiaphragm immunoblots, only the lower molecular weight cleavage fragment was detected. This is interesting in light of recently published evidence that the syntaxins 1A and 1B are differentially expressed in the peripheral nervous system, with 1B localized to nerve terminals of the neuromuscular junction, and 1A to perivascular nerve endings (Aguado et al., 1999). Another apparent difference was in the degree of SNAP-25 proteolysis noted in the brain synaptosomal preparation compared with the intact hemidiaphragm. The reduction in SNAP-25 immunoreactivity as measured by scanning densitometry of the parent SNAP-25 band was only 15% in the brain preparation compared with 67% in the hemidiaphragm. This was due to the greater amount of SNAP-25 protein found in brain, combined with the limited ability to accurately measure two closely apposed protein bands with scanning densitometry. Because the brain P₃ fraction was treated with toxin after homogenization and centrifugation, this fraction contained both the membrane-associated parent protein (25 kDa) as well as the larger cleavage fragment (24 kDa). In the hemidiaphragm, which was treated with toxin before homogenization and centrifugation, the P₃ fraction contained only the membrane-associated parent protein, allowing a more accurate measurement of loss of immunoreactivity by scanning densitometry.

The prolonged toxin incubation times necessary to measure significantly reduced immunoreactivity, and the incomplete substrate proteolysis reported here are at variance with some previous reports of permeabilized cell systems or recombinant proteins (Schiavo et al., 1992b, 1993a,b, 1994b; Blasi et al., 1993a,b; Lawrence et al., 1997). However, the current findings are consistent with other studies that used intact tissue or cell preparations (Foran et al., 1996; Williamson et al., 1996; Raciborska et al., 1998). These differences are most likely due to reduced accessibility of the toxin to its substrate in an intact preparation. Subcellular preparations may concentrate the substrates, facilitating the toxin's access to a critical mass of substrate. Furthermore, the use of permeabilized cells and recombinant proteins bypasses the plasma membrane endocytotic and endosomal translocation steps, theoretically enhancing the toxin's access to substrate.

These findings provide the first direct confirmation that the action of botulinum toxin at the mammalian neuromuscular junction is mediated by the selective inactivation of nerve terminal proteins that participate in the molecular machinery mediating neurotransmitter release. To the best of our knowledge this is the first time that studies of this nature on the intracellular action of botulinum toxin have been extended to a recognized mammalian target tissue preparation. These findings contribute significantly to understanding more fully the molecular mechanisms of botulinum toxin in its target tissue, and are important for the future development and testing of more effective toxin analogs and antagonists.