Purification of Fully Activated Clostridium botulinum Serotype B Toxin for Treatment of Patients with Dystonia

Abstract

Clostridium botulinum serotype B toxins 12S and 16S were separated by using a beta-lactose gel column at pH 6.0; toxin 12S passed through the column, whereas toxin 16S bound to the column and eluted with lactose. The fully activated neurotoxin was obtained by applying the trypsin-treated 16S toxin on the same column at pH 8.0; the neurotoxin passed through the column, whereas remaining nontoxic components bound to the column. The toxicity of this purified fully activated neurotoxin was retained for a long period by addition of albumin in the preparation.

Clostridium botulinum strains produce immunologically distinct neurotoxins (serotypes A to G). The molecular masses of neurotoxin types A to G are approximately 150 kDa. The neurotoxins are produced as a single form and become dichain-form light (50-kDa) and heavy (100-kDa) chains by cleavage with proteases such as trypsin at about one-third of the distance from the amino terminus, and the toxic activity of the dichain form becomes fully activated (1). In culture fluid and food with acidic conditions, the neurotoxins associate with nontoxic components and form large complexes designated progenitor toxins. Under alkaline conditions, the progenitor toxins dissociate into neurotoxin and nontoxic components (10, 18). The progenitor toxins are found in three forms with molecular masses of 900 kDa (19S), 500 kDa (16S), and 300 kDa (12S) (14). The 12S toxin is composed of a neurotoxin and a nontoxic component having no hemagglutinin (HA) activity (designated nontoxic non-HA [NTNH]), whereas the 16S and 19S toxins are composed of a neurotoxin, NTNH, and HA. The serotype A strain produces three forms of toxins (19S, 16S, and 12S). Type B, C, and D strains produce the 16S and the 12S toxins. We purified different-sized progenitor toxins from serotype A, C, and D cultures (2, 4, 6, 12, 13) and demonstrated that (i) the 19S toxin is a dimer of the 16S toxin; (ii) HA consists of four subcomponents with molecular masses of 52 to 53, 33 to 35, 19 to 23, and 15 to 17 kDa, designated here as HA3b, HA1, HA3a, and HA2, respectively; and (iii) NTNH of the 12S toxin has a cleavage site(s) at the N-terminal region.

Recently, serotype A and B progenitor toxins have been used for treating patients with strabismus, blepharospasm, nystagmus, facial spasm, spastic aphonia, and many other forms of dystonia (9, 11). In both toxin types, progenitor toxins are used because they are easily obtained and are more stable than neurotoxin. The treatment is very effective but has a serious side effect for some patients in whom antiprogenitor toxins, including antineurotoxin antibodies, are produced after several injections. It seems that using neurotoxin alone is better than using the progenitor toxin (a complex of the neurotoxin and a nontoxic component). Furthermore, it has been reported that serotype B toxin, which is used therapeutically at present, is partially cleaved (16) and therefore the toxin is not fully activated. In this paper, we report a simple procedure for large-scale purification of botulinum serotype B progenitor toxin and neurotoxin, which have fully activated toxicity.

C. botulinum serotype B proteolytic strain Lamanna was cultured by the dialysis tubing method (17). The toxins were precipitated with 60% saturated ammonium sulfate, treated with protamine, and then applied to an SP-Toyopearl 650 M column (1.4 by 26 cm; Tosoh, Tokyo, Japan) equilibrated with 50 mM sodium acetate buffer (pH 4.2) in the same manner as for the purification of serotype A toxin (6). All the chromatography steps discussed in this paper were performed at room temperature. The proteins were eluted with an NaCl gradient (0 to 0.5 M), and 2.5-ml fractions were collected. Four protein peaks were eluted (Fig. 1). Peak 1 was eluted as a shoulder of peak 2. Peak 2 possessed HA activity, but peak 1 did not, and both had very low toxicity. The toxin titer (minimum lethal dose [MLD]/ml) was obtained by injecting the diluted preparation into three mice intraperitoneally, each receiving 0.5 ml. The HA titer was obtained by reacting twofold-diluted preparations with an equal volume of 1% neuraminidase-treated human erythrocytes in a microtiter plate (7, 8). Peak 3 possessed both toxicity (8 × 10⁷ MLD/ml) and HA activity. Peak 4, which eluted as a shoulder of peak 3, possessed high toxicity (2 × 10⁸ MLD/ml) but a low HA activity. Therefore, it was speculated that peak 3 and peak 4 were 16S and 12S toxins, respectively. However, the two toxins could not be clearly separated either by the cation-exchange column or by gel filtration, especially when the amount of preparation was large. Therefore, we decided to establish a new purification procedure.

FIG. 1.

Separation of progenitor toxins by SP-Toyopearl 650 M cation-exchange column chromatography. An ammonium sulfate-precipitated preparation treated with protamine was applied to the column and eluted with an NaCl gradient. The HA activities of some fractions were determined. ○, protein; ▴, HA titer.

Recently, it was found that serotype A and B HA-positive toxins could bind to both erythrocytes and the epithelial cells of the small intestine mainly via HA1 and that these bindings were effectively inhibited by lactose (3, 8) (data concerning serotype B have not been published yet), indicating that HA-positive toxins can bind to lactose via HA1. Based on these data, we planned to separate HA-positive 16S toxin and HA-negative 12S toxin by use of affinity gel column-linked lactose; it was speculated that the 12S toxin would pass through the column whereas the 16S toxin would bind to the column. Fractions 70 to 79 and 80 to 90 in Fig. 1 were separately pooled and concentrated with 80% saturated ammonium sulfate. After centrifugation at 15,000 × g for 30 min, each precipitate was suspended, dialyzed against 10 mM sodium phosphate buffer (pH 6.0), and then applied to an aminophenyl beta-lactose gel column (1.0 by 6.0 cm; E-Y Laboratories Inc., San Mateo, Calif.) equilibrated with the same buffer. In both cases, the flowthrough fractions showed only high toxicity whereas the fractions eluted by the same buffer containing 0.2 M lactose showed both a high HA titer and toxicity, indicating that the former is 12S toxin and the latter is 16S toxin (Fig. 2). This was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and by determining the N-terminal amino acid sequence of each band that appeared. The banding profiles of the flowthrough fraction (Fig. 3, lane 1) and the eluate (Fig. 3, lane 2) were similar to those of 12S and 16S toxins, respectively, of other serotypes (2, 4, 6, 12, 13), and the N-terminal sequences were identical to those deduced from the nucleotide sequences of the serotype B toxin genes previously published (Fig. 3) (15, 19, 20, 21). Thus, it was concluded that the molecular compositions of serotype B 12S and 16S toxins are similar to those of other serotypes.

FIG. 2.

Separation of 12S and 16S toxins by beta-lactose gel affinity column chromatography. Fractions 80 to 90 in Fig. 1 were pooled and concentrated, and 8.4 ml of the sample (52 mg) was applied to the column at pH 6.0. Similar results were obtained by employing fractions 70 to 79. ○, protein; ▴, HA titer; ×, toxicity.

FIG. 3.

SDS-PAGE profiles of the purified progenitor toxins. Each separated toxin (the same as in Fig. 2) and protein standards were heated at 100°C for 7 min in sample buffer with 2-mercaptoethanol. Electrophoresis was performed on a 12.5% polyacrylamide gel. The gel was stained with Coomassie brilliant blue R-250. The N-terminal amino acid sequence of each band was also determined. The sequences corresponding to those deduced from the nucleotide sequences of genes are enclosed by open boxes. Lanes: M, molecular mass marker (myosin, 200 kDa; β-galactosidase, 116 kDa; phosphorylase b, 97.4 kDa; bovine serum albumin, 66.2 kDa; ovalbumin, 45 kDa; carbonic anhydrase, 31 kDa; soybean trypsin inhibitor, 21.5 kDa; lysozyme, 14.4 kDa); 1, unbound protein (12S toxin); 2, eluate (16S toxin).

The main fractions of peak 1 and peak 2 of Fig. 1 were also dialyzed and applied to the beta-lactose gel column. Both of them bound to the column and were eluted under the same conditions as those used for purification of the 16S toxin. By SDS-PAGE analysis, peak 1 proteins showed a single band with a molecular mass of 34 kDa, and peak 2 proteins showed four bands at 51, 34, 23, and 18 kDa (data not shown). Based on the N-terminal amino acid sequence, it was concluded that peak 1 and peak 2 proteins are HA1 and HA, respectively, as reported for serotype A toxins (6). Therefore, the beta-lactose gel was considered to be an affinity gel via at least HA1. The binding capacity of 1 ml of this matrix was estimated to be about 14 mg of 16S toxin. In one use of this purification method, 26 mg of the 12S toxin (3.2%) and 58 mg of the 16S toxin (7.2%) were obtained from the precipitates with 60% saturated ammonium sulfate of bacterial culture fluid (808.5 mg).

The progenitor toxins dissociate into a neurotoxin and a nontoxic component under alkaline conditions, 12S toxin dissociates into a neurotoxin and NTNH, and 16S and 19S toxins dissociate into a neurotoxin and a complex of NTNH and HA (10, 18). For purification of the neurotoxin, the purified 16S toxin thus obtained was first dialyzed against 10 mM sodium phosphate buffer (pH 8.0) to dissociate it to a neurotoxin and a nontoxic component, and then it was applied to the lactose gel column equilibrated with the same buffer (Fig. 4A). The neurotoxin passed through the column, whereas the nontoxic component bound to the column (the latter was eluted by the same buffer containing 0.2 M lactose). This was confirmed by SDS-PAGE; the flowthrough fraction demonstrated that the bands were consistent with the intact neurotoxin (158 kDa) and its heavy chain (106 kDa) and light chain (50 kDa) (Fig. 4B, lane 1). These results also indicated that the neurotoxin is not fully activated, even though this strain is proteolytic. Thus, we tried to purify the fully activated 16S and neurotoxin as follows. The partially activated 16S toxin preparation was incubated with bovine pancreatic trypsin (Sigma Chemical Co., St. Louis, Mo.) at pH 6.0 for 1 h at 37°C with a toxin-to-enzyme ratio of 100 to 1 in order to activate its toxicity. The preparation was then applied to the lactose gel column at the same pH. As expected, the fully activated 16S toxin (Fig. 5, lane 1) bound to the column, but the trypsin was washed out. The fact that trypsin does not bind to the column was confirmed independently (data not shown). This fully activated 16S toxin was dialyzed against the 10 mM sodium phosphate buffer (pH 8.0) and then layered on the column equilibrated with the same buffer. The fully activated neurotoxin (Fig. 5, lane 2) appeared in the unbound fractions. More simply, the fully activated neurotoxin could be obtained by applying the fully activated 16S toxin on the beta-lactose gel column at pH 6.0 and then changing the pH of the column with 0.1 M sodium phosphate buffer (pH 8.0) to dissociate the neurotoxin from the nontoxic component on the column.

FIG. 4.

(A) Separation of the neurotoxin by beta-lactose gel affinity column chromatography from the 16S toxin. The purified 16S toxin (3.5 mg) dialyzed against the buffer (pH 8.0) was applied to the column equilibrated with the same buffer. After the unbound protein was collected, the bound protein was eluted with the same buffer containing 0.2 M lactose. The HA and toxin titers of some fractions were determined. ○, protein; ▴, HA titer; ×, toxicity. (B) SDS-PAGE profiles of the flowthrough protein (neurotoxin) shown in Fig. 4A. Electrophoresis was performed on a 12.5% polyacrylamide gel. Lanes: M, molecular weight marker; 1, unbound protein (neurotoxin).

FIG. 5.

SDS-PAGE profiles of fully activated 16S toxin and neurotoxin purified by beta-lactose gel affinity column chromatography. The 16S toxin was first incubated with trypsin for 1 h at 37°C and then purified by the lactose gel column at pH 6.0. By applying this fully activated 16S toxin on the same column at pH 8.0, the fully activated neurotoxin was obtained in the unbound fraction. Lanes: 1, fully activated 16S toxin; 2, fully activated neurotoxin.

Another problem in using the neurotoxin for treatment is the low stability of the neurotoxin in long-term storage. It was reported that the pH of the buffer affects the stability of toxicity of the serotype B progenitor toxin (16) and that albumin can be used to stabilize the serotype A toxin (5). Therefore, the fully activated neurotoxin was filtered with a 0.45-μm-pore-size membrane filter and then diluted to 1,000 MLD/ml in sterilized 20 mM sodium phosphate buffer with different pHs (6.0, 7.0, and 8.0), with or without human serum albumin (0.5 mg/ml). After storage of these preparations (0.5-ml volume each, in vials) at 4 or −80°C for 40, 90, or 180 days, the level of remaining toxicity was determined in an assay using mice. Regardless of temperature or pH, no reduction of toxicity was observed for the albumin-containing samples that had remained in storage for at least 6 months (Table 1). On the other hand, the samples without albumin dramatically lost their toxicity within 40 days.

TABLE 1.

Stability of fully activated neurotoxin under different storage conditions^a

Buffer	Temp (°C)	pH	Toxin titer (MLD/ 0.5 ml) after storage for days indicated
			40	90	180
20 mM sodium phosphate	4	6.0	<1
		7.0	<1
		8.0	<1
	−80	6.0	1
		7.0	1
		8.0	1
20 mM sodium phosphate	4	6.0	500	500	500
+ albumin		7.0	500	500	500
		8.0	500	500	500
	−80	6.0	500	500	500
		7.0	500	500	500
		8.0	500	500	500

^aThe fully activated neurotoxin was diluted to 500 MLD/0.5 ml in buffer at different pHs with and without human serum albumin. After storage, the toxin titer remaining was determined for each sample.

These results may contribute to the development of new dosages of botulinum neurotoxin for the treatment of patients with dystonia. We are now investigating the effects of fully activated neurotoxin in animals electrophysiologically.