Specific Antibodies against the Zn2+-binding Domain of Clostridial Neurotoxins Restore Exocytosis in Chromaffin Cells Treated with Tetanus or Botulinum A Neurotoxin*

Although botulinum A neurotoxins are ineffective in cultured chromaffin cells, they will inhibit carbachol-induced release of noradrenaline provided they gain access to the cytosol either through artificial pores generated in the plasma membrane or by binding to incorporated exogenous gangliosides. The block of exocytosis persists for weeks followed by a slow recov- ery of cell function. When specific anti-botulinum A toxin antibodies are introduced into cells through pores after manifestation of the block by botulinum A neurotoxin, restoration of exocytotic function is accelerated and fully reestablished within 4 days. The same time course of restoration is seen with anti-tetanus toxin antibodies in cells poisoned by tetanus toxin. Since the light chains of the toxins are enzymatically active, we have introduced polyclonal and monoclonal anti-light chain antibodies into the cytosol. Of all light chain antibodies tested, only those directed against the peptide homologous to the zinc-binding sequence, which is present in both neurotoxins, restored exocytosis regard-less of which toxin caused the block. These


GTlb,
IV3NeuAc-I13(NeuAc),-GgOse4Cer. similar to motifs found in other metalloproteases. Tetx loses its toxicity when incubated with chelators (EDTA), indicating a pivotal role of its capability t o bind zinc (Sanders and Habermann, 1992;Schiavo et al., 1992~). The neurotoxins degrade with high specificity synaptic proteins that are involved in the fusion of transmitter-containing vesicles with the plasma membrane, thus inhibiting transmitter release (Schiavo et al., 1992b;Blasi et al., 1993). Whereas Tetx acts preferentially in spinal inhibitory neurons by blocking the release of glycine and y-aminobutyric acid (Curtis and de Groat, 1968), BoNtx inhibits synaptic transmission at the peripheral neuromuscular junction (Ambache, 1948). After intoxication, recovery of physiological functions is a slow process. It has been claimed that, in the case of botulism, nerve sprouting and a simultaneous appearance of newly synthesized peripheral acetylcholine receptors correlate with relief of clinical symptoms (Pestronk and Drachman, 1978;Duchen and Stritch, 1968). In the case of tetanus, hints about the mechanisms of physiological function recovery came from spinal cord neurons in culture. The toxin produced a state of neuronal hyperexcitability that faded eventually into synaptic quiescence (Bergey et al., 1987). The latter effect is due to an additional block of excitatory transmitter release that can be evoked at higher concentrations of toxin than those necessary to block glycine release. A partial restoration of excitatory activity occurred concomitantly with the intracellular degradation of Tetx. It was not clear whether these neurons resume synaptic transmission by sprouting or by complete degradation of the toxin, such that resynthesis of its target would no longer be counteracted by the proteolytic activity of the toxin (Habig et ai., 1986).
The intracellular targets of Tetx and BoNtx are present not only in nerve cells but also in certain other secretory cells, such as chromaffin cells (Ahnert-Hilger et al., 1989a;Mamen et al., 1989). However, since chromaffin cells lack gangliosides in their plasma membrane Marxen et al., 1991), they cannot take up the toxins unless artificial manipulations are performed. Nevertheless, chromaffin cells represent useful tools for studying the interference of toxins with exocytosis because they constitute a homologous cell population and offer an excellent access to structures controlling hormone release. There are several ways to introduce the toxins into the cytosol. By applying a strong electric field to the cells, pores are formed in the plasma membrane through which the toxins can diffuse into the cytosol (Bartels and Bigalke, 1992). Openings serving the same purpose can also be generated by cytolysins (Bittner et al., 1989;Ahnert-Hilger et al., 1989b). In addition, the toxins can be injected (Penner et al., 1986) or taken up by exogenous gangliosides incorporated into their plasma membrane . Plasma membrane pores offer access to the cytosol not only for the toxins but also for other molecules such as immunoglobulins (Lambert et al., 1990;Bartels and Bigalke, 1992;Marxen et al., 1990). Specific anti-tetanus and anti-botulinum A neurotoxin antibodies cannot promote the restoration of synaptic function in vitro, i.e. in spinal cord cultures (Habig et al., 1986), in mouse brain cultures (Habermann et al., 1988), or in chromaffin cells , nor in vivo by intravenous or intrathecal application (Olsen and Hiller, 1987) because antibodies cannot cross the intact plasma membrane. However, using electroporation, we were able to show that chromaffin cells recovered from poisoning by Tetx after application of specific anti-tetanus toxin antibodies (Bartels and Bigalke, 1992). Since chromaffin cells neither divide nor sprout in culture, they most likely resynthesize the toxin-cleaved substrate and thereby restore exocytosis. In this report we demonstrate an identical mechanism for the reconstitution of exocytosis blocked by BoNtx. We show that the length of recovery time is the same in Tetx-and BoNtx-poisoned cells and that antibodies directed against an amino acid sequence conserved in both toxins cannot only prevent the block but also reverse it for both toxins.

EXPERIMENTAL PROCEDURES
Primary Culture of Chromaffin Cells-Bovine chromaffin glands were obtained from the local slaughterhouse, and chromaffin cells were prepared as described previously Livett, 1984).
Cells were seeded onto collagen-coated 35-mm Primaria dishes (2 x lo7 cells) in 1 ml of Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 6 mg of glucose/ml, 100 IU of penicillin/ml, 100 pg of streptomycidml, ciprofloxacin, cytosine arabinoside, fluorodeoxyuridine, and uridine (lo-, M each) (growth medium). The same cell preparation was also used for electroporation. The cultures were kept a t 37 "C in a humidified atmosphere of 90% air, 10% CO,.
Cell Permeabilization-Electroporation was performed by the capacitor discharge method (Bartels and Bigalke, 1992) in a Bio-Rad Gene Pulser a t 20 "C. The cells were suspended in growth medium without antibiotics and cytostatics; mixed with Tetx, BoNtx, and antibodies, as indicated; and transferred into a sterile Gene Pulser cuvette (800 pl). Permeabilization was achieved by using a capacitance of 960 microfarads and an electric field of 625 V/cm, which decreased to l/e (approximately 37% of this value) in 10 ms. The percentage of viable cells after poration was approximately 50%. The cells were seeded onto 24-well Primaria multiplates (2.5 x lo5 celldwell). The release experiments were routinely carried out 3 days after electroporation.
Catecholamine Release-Chromaffin cells were preloaded with 0.125 pCi/ml ~-[~Hlnoradrenaline in growth medium (250 pl/well) for 3 h. To examine [3Hlnoradrenaline release, cultures were washed three times at 10-min intervals in a buffer containing in mu: NaCI, 125.6; KCI, 4.8; CaCl,, 2.2; MgSO,, 1.2; KH,PO,, 1.2; glucose, 5.6; HEPES, 25; sodium ascorbate, 1; 0.2% bovine serum albumin, pH 7.3. Basal release was determined within the next 8-min period during incubation in the same buffer. Then, exocytosis was stimulated by carbachol (5 x lo-, M) during another 8-min period, and the amount of supernatant radioactivity was determined. The basal release was subtracted from the stimulated re- Summary of enzyme-linked immunosorbent assay tests Wells were coated with the respective antigen, and antibody-containing solutions were added after thorough washing.
Wells were then loaded with peroxidase-conjugated anti-IgG antibodies followed by the benzidine reaction. Pairs marked with ND were not tested for immunoreactivity. Positive and negative reactions are marked by + and -, respectively. Reactions just above the detection limit are marked by (+I. lease, and exocytosis was expressed as a percentage of the total radioactivity, i.e. the sum of basal release, stimulated release, and radioactivity remaining in the cells (approximately 5000 dpm, determined after extraction from the cells with 0.2% (w/v) sodium dodecyl sulfate) . In cells treated with either toxin, total radioactivity was about 20% higher than in control cells, probably due to inhibition of exocytosis during the [3Hlnoradrenaline loading procedure.
Raising and Preparation ofAntibodies-Horse antiserum to BoNtx A was obtained by immunization of an animal with 500 pg of toxoid (plus Ribi adjuvant) at 0,3, and 7 weeks. The animal was then boosted with 100 pg of native toxin (Ribi adjuvant) at 23 and 45 weeks. Sera obtained were titered in a mouse lethality bioassay using a World Health Organization antiserum as a standard. 1 IU of antiserum neutralizes 10,000 mouse LD,, of toxin. Peptide (HELIHK) was conjugated with keyhole limpet hemocyanin, bovine serum albumin, or thyroglobulin using l-ethyl-3-(3-dimethylaminopropy1)carbodiimide HCl. The reaction mixture was dialyzed against distilled water and then lyophilized. Rabbits were injected subcutaneously with HELIHK-keyhole limpet hemocyanin dissolved in a mixture of 0.6 ml of NaCl (0.9%) and 0.6 ml of Freund's complete adjuvant. Routinely, rabbits were boosted with the antigen (plus Freund's incomplete adjuvant) every 4 weeks. Mice were immunized with HELIHK-thyroglobulin according to the same protocol. Antisera against the individual chains of BoNtx were obtained by immunizing guinea pigs with heavy or light chains according to the same scheme. Two weeks after each boost, animals were bled and serum tested for immunoreactivity. Monoclonal antibodies against LC-Tetx and HC-Tetx were derived from a fusion of spleen cells from a mouse immunized with tetanus toxoid and the myeloma cell line P3X63 Ag8 (Kenimer et al., 1983). Enzyme-linked immunosorbent assay tests were performed using 96-well plates coated overnight with the antigen (Table  I) at 4 "C in sodium phosphate buffer (20 m) containing NaCl(150 mM).
After washing, the plates were incubated with bovine serum albumin (10 mg/ml) to block nonspecific binding. Sera diluted into phosphate buffer were added to the plates and incubated for 8 h. Wells were washed again and incubated with the appropriate anti-IgG peroxidase conjugate (rabbit, mouse, horse, guinea pig). After washing, the plates were developed with the peroxidase substrate benzidine, and the absorbance was measured in a n enzyme immunoassay reader at 450 nm. Rabbit IgG was further purified approximately 6-fold by affinity chromatography on protein G-Sepharose.
bodiimide HCl, and keyhole limpet hemocyanin were from Saxon Biochemicals GmbH, Hannover, Germany. Lev0-[7-~H]norepinephrine (14.2 Ci/mmol) was from New England Nuclear Research Products, Dreieich, Germany, and liquid scintillation mixture (Aqualuma) and scintillation minivials were from Baker Chemicals, Deventer, Holland. Trypsin (2.5%) was obtained from Boehringer, Mannheim, Germany. Protein G-Sepharose was from Pharmacia, Uppsala, Sweden, and anti-IgG from different animals conjugated with peroxidase was from Dako, Denmark. The ganglioside mixture was a gift from FIDIA Research Laboratories, Abano Terme, Italy.

RESULTS
When chromaffin cells were exposed t o an electric field in the presence of two concentrations of BoNtx or Tetx, carbacholstimulated release of [3Hlnoradrenaline decreased in a concentration-dependent fashion (Fig. 1). The toxins' concentrations determined not only the extent of the block of exocytosis but also its duration. When cells were exposed to lower concentrations of toxin, the restoration of hormone release was somewhat faster, but it was still a slow process. It took weeks before noradrenaline release was fully re-established (Fig. 1). During this time the toxins did not affect the viability of the cells.
Since the duration of exocytosis block depended on the toxin dose, it might be feasible to shorten the recovery period by neutralizing intracellular toxin. Since antibodies have a molecular weight similar to that of the toxins, they can also pass through the artificial pores (Bartels and Bigalke, 1992). Alternatively, clostridial toxins may enter chromaffin cells whose plasma membrane had been left intact but enriched with exogenous gangliosides. Carbachol-induced release of l3H1noradrenaline decreased by 50% when ganglioside-treated cells were exposed to 6.6 nM BoNtx (Fig. 2). This block was prevented when specific polyclonal anti-dichain BoNtx antibodies (160 IU/ml) (poly-DC-BoNtx) were introduced into the cytosol by electroporation 2 days before exposure to BoNtx (Fig. 2). The block was not prevented when the antibodies were applied without electroporation under otherwise identical conditions (not shown). This indicates that BoNtx reached the same cytosolic compartment as the antibodies, although toxin and anti- 2. Intracellular anti-DC-BoNtx antibodies prevent the inhibitory action of BoNtx. Chromaffin cells were exposed to an electric field in the presence of either electroporation solution alone (a, c) or electroporation solution containing poly-DC-BoNtx (b, d ) . Cells were then seeded onto culture plates. After 24 h, the solution in all dishes was replaced with a ganglioside-containing growth medium. After another 24 h, cells were exposed to low ionic strength solution with (c, d ) and without ( a , 6 ) BoNtx (6.6 nM) for 24 h, followed by a 24-h incubation in toxin-free growth medium. [3HlNoradrenaline release (ordinate), given as a percentage of total radioactivity, was initiated by stimulation with carbachol. Values are the mean of three determinations t S.D. The differences between bars a and c as well as c and d were statistically significant ( p < 0.001).
body crossed the plasma membrane by different routes.
Antibodies could reverse a block previously established by ganglioside-mediated BoNtx entry, although recovery was slow (Fig. 3). Chromaffin cells electroporated in the absence of antibodies did not recover from the BoNtx treatment within 6 days after BoNtx application (Fig. 3). The degree of recovery 48 h after antibody treatment, when restoration is still incomplete (also see Fig. 11, was not enhanced by higher concentrations of the antibodies (Fig. 4, inset), though the restoration of catecholamine release depended on the concentration of specific antibodies present during the electroporation procedure (Fig.  4).
Since the proteolytic activity of both toxins is located in their light chains, we used the following light chain-specific antibodies in release experiments: 1) guinea pig polyclonal antibodies raised against the light chain of BoNtx (poly-LC-BoNtx), 2) mouse monoclonal antibodies against the light chain of Tetx (mono-LC-Tetx), 3) rabbit (poly-P1 and poly-P2) or mouse (poly-P3) polyclonal antibodies raised against a synthetic peptide identical to the zinc-binding motif (HELIH) found in the light chains of both toxins. Controls consisted of a mouse monoclonal antibody raised against Tetx heavy chain (mono-HC-Tetx) and guinea pig polyclonal antibody raised against BoNtx heavy chain (poly-HC-BoNtx). Each antibody was initially tested in enzyme-linked immunosorbent assay for immunoreactivity against its antigen. For the detection of anti-peptide antibodies, a different conjugate of HELIHK was used. Poly-P1 and poly-P2 had identical titers to the synthetic peptide (Fig. 5). Poly-P1 recognized DC-Tetx, DC-BoNtx, LC-Tetx, and LC-BoNtx (Fig. 5) while poly-P2 only recognized DC-Tetx and DC-BoNtx (Table I). Poly-P3 did not recognize LC-and DC-BoNtx nor LC-Tetx. Poly-LC-BoNtx did not react with DC-BoNtx. All other antibodies reacted as expected (Table I).
The capability of these antibodies to neutralize toxin was tested using three different experimental designs. In one set of experiments antibodies and toxins were mixed prior to addition to intact chromaffin cells incubated with gangliosides. With

. Concentration-dependent restoration of exocytosis by intracellular anti-DC-BoNtx antibodies. Chromaffin cells in mono-
layer culture enriched with gangliosides were exposed to low ionic strength solution in the presence (6.6 m) or absence of BoNtx for 24 h. Cells were washed twice with Ca2+-free phosphate-buffered saline and electric field in the absence or presence of the indicated concentrations dissociated with collagenase. The cell suspension was exposed to a n of specific anti-DC-BoNtx antibodies (abscissa). After 48 h (inset) and 96 h, the release experiment was performed. The release of PHInoraantitoxin ( c l ) , in the absence of both BoNtx and antitoxin (c2), and in drenaline from chromaffin cells in the presence of BoNtx and absence of the presence of BoNtx and antitoxin in the concentrations indicated is given as a percentage of total radioactivity (ordinate). Values are the mean of three determinations t S.D. this paradigm, only homologous antibodies against intact toxins or heavy chains prevented poisoning (Fig. 6). Antibodies against the light chains did not inhibit poisoning by toxin using the preincubation paradigm (Table 11, Fig. 6), although they did bind their respective antigens ( Table I). In a second set of experiments, the same mixtures were incubated with chromaffin cells during electroporation. With this protocol, mono-HC-Tetx and poly-HC-BoNtx lost their ability to prevent poisoning, suggesting that these antibodies inhibited binding or uptake of the toxin rather than its proteolytic activity (Table 11, Fig. 6). Mono-LC-Tetx and poly-LC-BoNtx did not inhibit the proteolytic activity of Tetx or BoNtx despite their ability to bind to their respective antigens although not to HELIH (Tables I and  11). In contrast, poly-P1 and -P2 were able to prevent poisoning by both toxins, and poly-P3 was able t o prevent poisoning by Summary of the neutralization tests in release experiments Release experiments followed three experimental designs. 1) Intact cells, preloaded with gangliosides, were exposed to toxin-antitoxin mixtures (extracellular). 2) Electroporated cells (intracellular) were exposed to toxin-antitoxin mixtures (preventing). 3) Electroporated cells were exposed to antitoxin alone (restoring) after ganglioside and toxin exposure. In the second set, antitoxin neutralized both light chain toxins and DC toxins. Pairs marked with ND were not tested for neutralization. Positive neutralization is marked by + and the failure to neutralize by -, respectively.  (Table 11). The concentration dependence of the protection is shown in Fig. 7a. The amount of poly-P1 contained in approximately 10 mg of IgG was equivalent to 40 IU of poly-DC-Tetx in preventing the block of exocytosis. Like the polyclonal DC-toxin antibodies, polyclonal antibodies P1 a n d P2 were also able to reverse a block that had been established in the paradigm where ganglioside-treated cells were exposed to toxin followed by antibody electroporation 1 day later (Table 11). Fig. 7b shows the concentration dependence for reversal of BoNtx poisoning by poly-P1. 10 mg of IgG was as effective in restoring exocytosis as 120 IU of poly-BoNtx.  (Bartels and Bigalke, 1992) large enough to allow proteins such as Tetx and BoNtx and immunoglobulins to diffuse through these openings. When the pores are closed by membrane fusion a short time later, the proteins remain trapped inside the cells. Since chromafin cells lack binding sites for clostridial neurotoxins , the toxins cannot have gained access to the cytosol by any other route than by diffusion through the electrogenerated pores. Exocytosis is blocked by either toxin in a concentration-dependent fashion. Toxicity indicates that the disulfide bonds connecting the heavy and light chains have been cleaved within the cells because only the reduced light chains are capable of proteolytic activity (Ahnert-Hilger et al., 1989a, 1989b) and that the light chains of both toxins are able to reach their intracellular substrates and degrade them. Exocytosis is blocked by Tetx and BoNtx over a long period of time. The recovery of exocytotic function is a slow process, and the rate of substrate resynthesis may be decisive for the progress to restoration. Since the toxins will presumably degrade substrate proteins as long as they are active, a restoration of noradrenaline release will not occur until the amount of intracellular active toxin declines to a level at which the degradation of the substrate is superseded by its synthesis. The decline in the activity of the toxins can be accelerated by neutralizing them with specific anti-toxin antibodies that, like the toxins, are able to pass through the plasma membrane by way of the electrogenerated pores. Thus, the duration of the block of exocytosis can be shortened. Despite the presence of antibodies, the recovery of exocytotic function proceeds at a slow pace and reaches an upper rate that cannot be Values are the mean of three determinations f S.D. b shows concentration-dependent restoration of exocytosis by intracellular anti-HELIH antibodies in chromafin cells poisoned with BoNtx by ganglioside-mediated transport. Chromafin cells in monolayer culture enriched with gangliosides were exposed to low ionic strength solution in the presence (0.6 m) or absence of BoNtx for 24 h. Cells were then washed twice with Caz*-free phosphate-buffered saline and dissociated with collagenase. The cell suspension was exposed to an electric field in the absence or presence of the indicated concentrations of specific anti-HELIH antibodies (abscissa). After 72 h, the release experiment was performed. The release of [3Hlnoradrenaline from chromafin cells in the presence of BoNtx and absence of antitoxin (cl), in the absence of both BoNtx and antitoxin ( c~) , in the presence of BoNtx and poly-BoNtx (c3), and in the presence of BoNtx and IgG containing poly-P1 in the concentrations indicated is given as a percentage of total radioactivity (ordinate). Values are the mean of three determinations 2 S.D. accelerated by higher antitoxin concentrations. Thus, the rate of resynthesis of the substrates following neutralization may govern restoration. Neutralization itself occurs rapidly because exocytosis is not blocked if antibodies are present in the cytosol before the toxins (for Tetx, see Bartels and Bigalke (1992)). This can be achieved by electroporating the cells in the presence of antibodies and subsequently incubating them with gangliosides that will be incorporated into the plasma membrane and serve as binding sites a n d vehicle of entry for the toxins. After the neutralization with specific antibodies, BoNtx-poisoned cells showed a time course of recovery nearly identical to those treated with Tetx ( Bartels and Bigalke, 1992), although the substrates acted on by the toxins differ from each other (Schiavo et al., 1992a(Schiavo et al., , 1992b(Schiavo et al., , 1992cMcMahon et al., 1993). This could mean either that the rates of resynthesis are similar for different substrates or that a cellular structure must be resynthesized that contains both substrates. The substrates for Tetx are synaptobrevin I1 and cellubrevin (Schiavo et al., 1992c;Link et al., 1992; McMahon et al., 19931, and that for BoNtx is SNAP25 (Blasi et al., 1993). All are constituents of the vesicular membrane and are involved in the fusion of the vesicles with the plasma membrane (Siidhof and Jahn, 1991;Jahn and Siidhof, 1993). Thus it is possible that recovery of exocytosis may occur after restoration of vesicles with intact fusion proteins. In nerve cells in uiuo, recovery may not necessarily be restricted to the sprouting of new axons as demonstrated after injection of therapeutic doses of BoNtx (Alderson et al., 1991) but may result from restoration of intact synaptic vesicles.

Extracellular
The binding of zinc to the light chain of Tetx is crucial for its proteolytic activity (Sanders and Habermann, 1992;Schiavo et al., 1992~). Thus, recombinant mutants of the light chain of Tetx, in which zinc cannot bind due to replacement of histidine 234 or glutamic acid 235 with alanine, lose their efficacy to block exocytosis with chromaffin cells.' The antibodies (poly-Pl-P3) raised against a synthetic peptide resembling the zinc binding site (234 HELIH) were capable of neutralizing Tetx. The same antibodies also interfered with BoNtx, which carries an identical amino acid sequence, indicating that this domain is involved in the toxicity of BoNtx as well. The effectiveness of the antibodies was dependent on how the cells were treated with the toxins. Though poly-Pl-P3 bind with high affinity to the synthetic peptide and the sera from the rabbits containing poly-P1 and -P2 have identical titers, the three sera differ markedly at recognizing the zinc binding epitope as the intact toxin molecule or light chain. The weak interaction of the anti-HELIH antibodies with intact toxin may explain why these antibodies cannot prevent the action of either toxin when preincubated with toxins followed by addition to ganglioside-enriched chromaffin cells (this paper) or to the phrenic hemidiaphragm p r e p a r a t i~n .~ The differences in binding affinities of poly-Pl-P3 to the holotoxins or their light chains may indicate that these anti-HELIH antibodies do not recognize an identical structure within the peptide. The zinc-binding domain is probably hidden within the intact molecule, a concept supported by the fact that specific polyclonal anti-Tetx and anti-BoNtx antibodies, respectively, are not active against the HELIH peptide in an enzyme-linked immunosorbent assay. Physiologically, anti-HELIH antibodies will probably not be present within the pool of polyclonal anti-toxin antibodies. The latter are capable of neutralizing extra-and intracellular toxins by forming complexes with the toxins. Apparently, these complexes are not taken further up into intact cells or, if formed in the cytosol, prevent the active site of the enzymes from recognizing their substrates. The inability of either monoor poly-anti-LC toxin antibodies to neutralize intracellular light chains may be due to a formation of less stable complexes or a lack of complex for-H. Bigalke and H. Niemann, unpublished observations. H. Bigalke, unpublished observations. mation. Alternatively these antibodies may bind to epitopes on the light chain that are irrelevant to their proteolytic activity. However, upon exposure of the zinc-binding domain of either Tetx or BoNtx within the cell, anti-HELIH antibody can bind to the proteolytic site and prevent proteolysis of the natural substrate.