Characterization of haemagglutinin activity of Clostridium botulinum type C and D 16S toxins, and one subcomponent of haemagglutinin (HA1)

Kaoru Inoue1, Yukako Fujinaga1, Koichi Honke2, Kenji Yokota1, Tetsuya Ikeda3, Tohru Ohyama3, Kouichi Takeshi3, Toshihiro Watanabe4, Katsuhiro Inoue4 and Keiji Oguma1

Department of Bacteriology, Okayama University Medical School, 2-5-1 Shikata-cho, Okayama 700-8558, Japan1
Department of Molecular Medicine, Research Institute, Osaka Medical Center for Maternal and Child Health, 840 Murodo-cho, Izumi, Osaka 594-1101, Japan2
Hokkaido Institute of Public Health, N19, W12, Sapporo 060-0819, Japan3
Department of Food Science, Faculty of Bioindustry, Tokyo University of Agriculture, 196, Yasaka, Abashiri 099-2422, Japan4

Author for correspondence: Keiji Oguma. Tel: +81 86 235 7162. Fax: +81 86 235 7162. e-mail: kuma{at}med.okayama-u.ac.jp


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The 16S toxin and one subcomponent of haemagglutinin (HA), designated HA1, were purified from a type D culture of Clostridium botulinum by a newly established procedure, and their HA activities as well as that of purified type C 16S toxin were characterized. SDS-PAGE analysis indicated that the free HA1 forms a polymer with a molecular mass of approximately 200 kDa. Type C and D 16S toxins agglutinated human erythrocytes in the same manner. Their HA titres were dramatically reduced by employing erythrocytes that had been previously treated with neuraminidase, papain or proteinase K, and were inhibited by the addition of N-acetylneuraminic acid to the reaction mixtures. In a direct-binding test to glycolipids such as SPG (NeuAc{alpha}2-3Galß1-4GlcNAcß1-3Galß1-4Glcß1-Cer) and GM3 (NeuAc{alpha}2-3Galß1-4Glcß1-Cer), and glycoproteins such as glycophorin A and/or B prepared from the erythrocytes, both toxins bound to sialylglycolipids and sialoglycoproteins, but bound to neither neutral glycolipids nor asialoglycoproteins. On the basis of these results, it was concluded that type C and D 16S toxins bind to erythrocytes through N-acetylneuraminic acid. HA1 showed no haemagglutination activity, although it did bind to sialylglycolipids. We therefore speculate that binding to glycoproteins rather than to glycolipids may be important in causing haemagglutination by type C and D 16S toxins.

Keywords: Clostridium botulinum, haemagglutinin, glycolipid, glycoprotein

Abbreviations: CBB, Coomassie brilliant blue; HA, haemagglutinin; MLD, minimal lethal dose


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Clostridium botulinum strains produce immunologically distinct neurotoxins (types A to G) which inhibit the release of acetylcholine at the neuromuscular junctions and synapses. The molecular masses of all types of neurotoxins are approximately 150 kDa. The neurotoxins associate with non-toxic components in the culture fluid to form larger complexes which are designated progenitor toxins. In type A strains, three different-sized progenitor toxins with molecular masses of 900 kDa (19S), 500 kDa (16S) and 300 kDa (12S) are produced. Type B, C and D strains produce both the 16S and 12S toxins, whereas types E and F, and type G strains produce only the 12S or 16S toxin, respectively (Sakaguchi & Ohishi, 1984 ). In all these strains, the non-toxic components of the 19S and 16S toxins display haemagglutinin (HA) activity but that of the 12S toxin does not show HA activity. The 12S toxin is formed by association of a neurotoxin with a non-toxic component having no HA activity (designated here as non-toxic non-HA), while 19S and 16S toxins are formed by conjugation of the 12S toxin with HA. The non-toxic components (HA and non-toxic non-HA) are considered to be very important to the development of food poisoning because the non-toxic components protect the neurotoxin from the gauntlet of gastric juice when the progenitor toxins pass through the stomach (Sugii et al., 1977 ). Therefore, the larger the molecular size of the progenitor toxin of the same type, the higher the oral toxicity (Ohishi & Sakaguchi, 1980 ; Ohishi et al., 1977 ).

We purified different-sized progenitor toxins from C. botulinum type A and C culture fluids, and demonstrated that the HAs consist of subcomponents that have molecular masses of 52–53, 33–35, 19–23 and 15–17 kDa, which are designated here HA3b, HA1, HA3a and HA2, respectively (Fujinaga et al., 1994 ; Inoue et al., 1996 ). In the case of the type A strain, the HA and its one subcomponent protein, which had a molecular mass of 35 kDa (HA1), were purified in addition to the progenitor toxins. This indicated that HA and HA1 exist in a free state in the culture fluid. The free HA showed the same HA activity as the HA-positive toxins (16S and 19S toxins), whereas the free HA1 did not show HA activity (Inoue et al., 1996 ).

Experimental results obtained using the Ouchterlony test suggest that the antigenicity of the non-toxic components of types C and D is identical (Oguma et al., 1980 ). This hypothesis was confirmed by a comparison of the entire nucleotide sequence of the non-toxic components, which indicated that the two sequences were essentially identical (Fujinaga et al., 1994 ; Hauser et al., 1994 ; Ohyama et al., 1995 ;Tsuzuki et al., 1990 , 1992 ). The HA activity of type C and D cultures were investigated by Suzuki et al. (1986 ) and Balding et al. (1973 ) by employing crude HA or 16S toxin preparations. Suzuki et al. (1986 ) reported that the HA activity of type C crude HA was inhibited by adding gangliosides and fetuin to the reaction mixture, and also reduced by employing erythrocytes which had been treated with neuraminidase. Balding et al. (1973 ) reported that the HA activities of type C and D toxins were inhibited in the presence of MN active glycoprotein, as well as employing the neuraminidase-treated erythrocytes. Therefore, it is postulated that the sugar (N-acetylneuraminic acid; NeuAc) on the surface of erythrocytes is most likely the receptor for type C and D HA binding. However, it has yet to be established that purified erythrocyte glycolipids and glycoproteins will bind to HAs.

In this paper, we established a new and simple method for purifying type D 16S toxin. It was also possible to purify free HA1 using the new purification method. We performed haemagglutination and haemagglutination-inhibition tests, and the direct-binding tests, by employing purified type C and D 16S toxins and type D free HA1, in order to determine how they bind to erythrocytes.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Bacteria and toxin.
The C. botulinum type C strain C-Stockholm and type D strain, D-CB16 were used. Type C 16S toxin was purified from culture fluid using procedures that have been described previously (Kimura et al., 1990 ; Tsuzuki et al., 1990 , 1992 ). Toxin titre (minimal lethal dose, MLD, ml-1) was determined in mice: 0·5 ml each preparation diluted in serial tenfold steps was intraperitoneally injected into three white mice. The type D 16S toxin and free HA1 were purified by a new procedure as follows. The organisms were cultured by a cellophane-tube procedure (Stern & Wentzel, 1950 ) at 35 °C for 5 d. The culture medium which possesed 104 MLD ml-1 was centrifuged (6000 g, 30 min), and the supernatant was fractionated with a 60% saturation of (NH4)2SO4. The precipitate collected by centrifugation at 15000 g for 10 min was dialysed overnight against 20 mM sodium phosphate buffer (pH 6·0). Later, it became apparent that the precipitate appearing during dialysis contained the 16S toxin, free HA1 and C3 enzyme, whereas the supernatant contained the 12S toxin. The precipitate was collected by centrifugation at 15000 g for 20 min, resuspended in 30 ml 150 mM sodium acetate buffer (pH 4·2), dialysed against the same buffer (150 mM sodium acetate buffer, pH 4·2), and then applied to an SP-Toyopearl 650M (Tosoh) column (1·4x26 cm) equilibrated with 150 mM sodium acetate buffer (pH 4·2). From this column, three protein peaks (peaks 1–3) were eluted by an exponential gradient of NaCl (0–0·8 M) in the equilibrating buffer (Fig. 1). Peaks 1 and 2 possessed weak HA activity (23) and very low toxic activity (less than 103 MLD ml-1). Peak 3 possessed both high toxic (more than 105 MLD ml-1) and HA activities (210). Based on the banding profile of SDS-PAGE (Ohyama et al., 1995 ), it was postulated that peak 3 contained the 16S toxin. The fractions of peak 3 were pooled and concentrated with 80% saturation of (NH4)2SO4. By applying it twice to a Toyopearl HW-65S (Tosoh) column (1·5x100 cm), the purified 16S toxin preparation was obtained. The fractions of peak 2 (Fig. 1) were pooled, concentrated with a 80% saturated (NH4)2SO4, and further purified by gel filtration on Toyopearl HW-65S. The purified peak 2 preparation showed a single band with a molecular mass of 25·5 kDa on SDS-PAGE. The N-terminal amino acid sequence of this protein band (AYSNTYQEFTNI----) was identical to that of type D (D-1873 strain) C3 enzyme (Moriishi et al., 1991 ). The peak 1 fractions were pooled, dialysed against 10 mM phosphate buffer (pH 6·0), and applied to a hydroxyapatite (Seikagaku) column (0·7x3 cm) equilibrated with 10 mM phosphate buffer (pH 6·0) (Fig. 2). After washing the column with 10 mM phosphate buffer (pH 6·0), the target protein was eluted with 10 mM phosphate buffer containing 1 M NaCl. The 16S toxin contaminating the peak 1 preparation was eluted with 0·2 M phosphate buffer (pH 6·0). The 16S toxin possessed both toxin and HA activities; however, the target protein eluted with 10 mM phosphate buffer containing 1 M NaCl did not show HA activity.



View larger version (13K):
[in this window]
[in a new window]
 
Fig. 1. SP-Toyopearl 650M ion-exchange column chromatography. The precipitate at pH 6·0 after ammonium sulfate fractionation from the culture fluid was applied to a column equilibrated with 150 mM sodium acetate buffer (pH 4·2). Elution was performed by increasing the concentration of NaCl to 0·8 M and 2 ml fractions were collected. The fractions indicated by solid bars were separately pooled. {circ}, A280. The dashed line represents the concentration of NaCl.

 


View larger version (16K):
[in this window]
[in a new window]
 
Fig. 2. Hydroxyapatite column chromatography. The pooled peak 1 fraction from SP-Toyopearl 650M ion exchange chromatography was applied to a hydroxyapatite column equilibrated with 10 mM phosphate buffer (pH 6·0). After washing with 1·5 column volumes of equilibrating buffer, the elution was performed successively with 10 mM phosphate buffer (pH 6·0) containing 1 M NaCl and 0·2 M phosphate buffer (pH 6·0). Fractions of 1 ml were collected. The fractions indicated by a solid bar were pooled. {circ}, A280; {blacktriangleup}, HA titre.

 
Glycolipids and glycoproteins.
Glycosphingolipids were purified from the following sources as described previously (Gasa et al., 1983 ): GM4 (NeuAc{alpha}2-3Galß1-Cer), GM1 (Galß1-3GalNAcß1-4(NeuAc{alpha}2-3)Galß 1-4Glcß1-Cer),GD1a[NeuAc{alpha}2-3Galß1-3GalNAcß1-4(NeuAc{alpha}2-3)Galß1-4 Glcß1-Cer], GD1b [Galß1-3Ga lNAcß1-4(NeuAc{alpha}2-8NeuAc{alpha}2-3)Galß1-4Glcß1-Cer] and SM4 (HSO3-3Galß1-Cer) from bovine brain; GM2 [GalNAcß1-4(NeuAc{alpha}2-3)Galß1-4Glcß1-Cer] from a Tay-Sachs brain; CMH (ceramide monohexoside, a mixture of Galß1-Cer and Glcß1-Cer) and LacCer (Galß1-4Glcß1-Cer) from horse erythrocytes; Gb3Cer (Gal{alpha}1-4Galß1-4Glcß1-Cer) and SM3 (HSO3-3Galß1-4Glcß1-Cer) from human kidney; Gb4Cer (GalNAcß1-3Gal{alpha}1-4Galß1-4Glcß1-Cer) and SPG (NeuAc{alpha}2-3Galß1-4GlcNAcß1-3Galß1-4Glcß1-Cer) from human erythrocytes; GM3 (NeuAc{alpha}2-3Galß1-4Glcß1-Cer) and GD3 (NeuAc{alpha}2-8NeuAc{alpha}2-3Galß1-4Glcß1-Cer) from rat liver. Fetuin was purchased from Wako Pure Chemical Industries. Asialofetuin and human glycophorin were purchased from Sigma.

SDS-PAGE and electroblotting.
SDS-PAGE was performed by the method of Laemmli (1970 ) using 12·5% acrylamide linear gel. Protein bands were stained with Coomassie brilliant blue (CBB) R-250. The bands separated by SDS-PAGE were electroblotted to PVDF membranes (ProBlott; Applied Biosystems) with a semidry blotting apparatus (Nippon Eido) according to the methods described by Hirano & Watanabe (1990 ).

Preparation of antisera and immunoblotting.
Preparation of antisera against type C non-toxic components (a complex of non-toxic non-HA and HA) that was prepared previously (Oguma et al., 1980 ) in a rabbit was used. The protein bands separated by SDS-PAGE were electroblotted to a PVDF membrane as described above. The membrane was immersed overnight in PBS (pH 7·4) containing 10% (w/v) skimmed milk (blocking buffer 1). The membrane was incubated for 1 h with 1:1000-diluted anti-type C non-toxic components serum in blocking buffer 1 and then washed in PBS (pH 7·4) containing 0·05% (v/v) Tween 20 (PBS-Tween). After incubation with 1:1000-diluted peroxidase-labelled pig anti-rabbit IgG (Dako A/S) in blocking buffer 1 for 1 h, immunoreactive bands were detected by the ECL Western blotting system (Amersham).

Enzyme treatment of human erythrocytes.
Ten millilitres of 10% (v/v) washed erythrocytes were incubated with 0·01 unit Arthrobacter ureafaciens neuraminidase (Nacalai tesque), 80 mg trypsin (Difco), 80 mg papain (Wako Pure Chemical Industries), 8 mg proteinase K (Wako Pure Chemical Industries), or 100 mU endoglycoceramidase (EGCase II ACT, Takara Shuzo) in PBS (pH 6·0) at 37 °C for 1 h. After treatment, the erythrocytes were washed three times in PBS (pH 6·0).

Determination of HA titre and inhibition tests.
The HA titre was obtained by microtitration methods using multiwell plates. Type C and D 16S toxins or HA1 solution (100 µg ml-1, 35 µl) were diluted in serial twofold steps with PBS (pH 6·0) and mixed with an equal volume of 1% (v/v) suspension of washed human erythrocytes or enzyme-treated erythrocytes. After incubation at room temperature for 2 h, the reciprocal of the highest dilution that showed haemagglutination was used as the HA titre. All tests were performed in duplicate and repeated three times. Inhibition of HA activity of the type C and D 16S toxins with several different saccharides were examined as follows. Each diluted preparation (20 µl) was mixed with 20 µl of an appropriate concentration of saccharide solutions and incubated at 37 °C for 1 h. Thereafter, 40 µl 1% (v/v) suspension of erythrocytes was added to the mixtures. Following incubation at room temperature for 2 h, haemagglutination was assessed. All tests were performed in duplicate and repeated twice.

Binding to erythrocytes.
The binding of 16S toxins and HA1 to erythrocytes were analysed by two different methods. The first method was conducted using 96-well microtitre plates. Fifty microlitres of 16S toxin (50 µg ml-1) or HA1 (50 µg ml-1) were incubated in microtitration plates, and left to stand overnight at 4 °C. The plates were washed three times with 200 µl PBS (pH 6·0), and 150 µl 1% (w/v) BSA in PBS (pH 6·0) was added to the plates. After 2 h incubation at room temperature, 100 µl 1% (v/v) native or neuraminidase-treated erythrocytes in PBS (pH 6·0) containing 1% (w/v) BSA was added to the plates. For the inhibition tests, 100 µl 1% (v/v) native or neuraminidase-treated erythrocytes in PBS (pH 6·0) containing 1% (w/v) BSA and 100 mM monosaccharide were added to the plates. Plates were incubated for 30 min at room temperature and washed with 200 µl PBS (pH 6·0) six or seven times. Bound erythrocytes were lysed by adding 50 µl distilled water and the A405 was analysed. All tests were performed in duplicate and repeated three times. The second method used to assess the binding of the 16S toxin and HA1 to erythrocytes was performed using an anti-type C non-toxic components serum. Twenty microlitres of 16S toxin (20 µg ml-1) or HA1 (10 µg ml-1) were mixed with 20 µl PBS (pH 6·0), 20 µl 4% (w/v) BSA in PBS (pH 6·0) and 20 µl 1% (v/v) native or neuraminidase-treated erythrocytes. After incubation at 37 °C for 1 h, the erythrocytes were precipitated by centrifugation and washed three times with 150 µl PBS, pH 6·0. Precipitated erythrocytes were mixed with 10 µl SDS-PAGE sample buffer and heated at 100 °C for 5 min. Electrophoresis was performed on a 12·5% polyacrylamide linear gel. After the protein bands were electroblotted to a PVDF membrane, 16S toxins or HA1 bound to erythrocytes were detected by immunostaining with anti-type C non-toxic components serum. For the inhibition tests, 20 µl 16S toxins (20 µg ml-1) were mixed with 20 µl monosaccharide (400 mM) solution, 20 µl 4% (w/v) BSA in PBS (pH 6·0). After incubation at 37 °C for 1 h, 20 µl 1% (v/v) native or neuraminidase-treated erythrocytes were added to the reaction mixture, and the binding to the erythrocytes was analysed using the methods described above.

Preparation of glycolipids from erythrocytes.
Human blood (type A, 500 ml) was centrifuged and the serum was discarded. The erythrocytes were washed twice in 0·9% (w/v) NaCl. The washed cells were lysed by adding 1500 ml distilled water and centrifuged to sediment ghosts. The wet ghosts were suspended in distilled water and the ratio of chloroform/methanol/water was adjusted to 30:60:8 (by vol.) by the addition of chloroform and methanol. The lipid contents were extracted by sonication for 30 min. The solution was filtered through filter paper and evaporated. The lipids were dissolved in a minimum amount of chloroform/methanol/water (30:60:8, by vol., as above) and applied to a DEAE-Sephadex A-25 column (acetate form, Pharmacia Biotechnology) equilibrated with chloroform/methanol/water (30:60:8, by vol.). The neutral glycolipids were eluted with chloroform/methanol/water (30:60:8, by vol.). The monosialo-, disialo- and polysialoglycolipids were eluted with chloroform/methanol/water (30:60:8, by vol.) containing 40 mM, 100 mM and 300 mM ammonium acetate, respectively. Each evaporated fraction was dissolved in methanol–water (1:1, v/v) and applied to Sep Pak 18C (Waters). After washing the column with distilled water, the desalted glycolipids were eluted with methanol and a mixture of chloroform/methanol (2:1, v/v), and evaporated.

TLC immunostaining.
Binding of 16S toxins and free HA1 to the glycolipids prepared from erythrocytes or the glycolipids purified as described above were analysed by TLC immunostaining. The glycolipids were developed on high-performance TLC aluminium sheets silica gel 60 plates (Merk) with chloroform/methanol/water (65:35:8, by vol.). The developed plates were dipped in 0·2% (w/v) polyisobutyl-methacrylate in n-hexan for 1 min and dried. The plates were incubated in PBS (pH 6·0) containing 1% (w/v) BSA (blocking buffer 2) for 1 h, followed by incubation with 5 µg ml-116S toxin or 10 µg ml-1 HA1 solution (in blocking buffer 2) for 1 h. The plates were then washed three times with PBS–Tween and reacted for 1 h with antiserum against type C non-toxic components diluted to 1:1000 with blocking buffer 2. After washing three times with PBS–Tween, the plates were incubated with a peroxidase-labelled anti-rabbit IgG antibody diluted to 1:1000 with blocking buffer 2 for 1 h. The immunoreactive bands that appeared were detected by ECL (Amersham). All of the procedures were performed at room temperature.

Binding to ghost membrane proteins or glycoproteins.
We analysed the binding of 16S toxins and free HA1 to ghost membrane proteins and glycoproteins. The latter were obtained commercially. The PVDF membranes blotted with proteins were immersed overnight in PBS (pH 6·0) containing 10% (w/v) BSA (blocking buffer 3) at 4 °C. The membranes were incubated in 5 µg ml-1 16S toxin or 10 µg ml-1 HA1 solution (in blocking buffer 3) for 1 h and then washed three times with PBS–Tween. After the membranes were reacted with antiserum against type C non-toxic components diluted to 1:1000 with blocking buffer 3 for 1 h, followed by a peroxidase-labelled anti-rabbit IgG antibody diluted to 1:1000 for 1 h, the immunoreactive bands were detected by ECL (Amersham).

Determination of amino acid sequence.
The N-terminal amino acid sequences of protein bands of ghost proteins which were bound by type C and D 16S toxins were determined. The protein bands electroblotted to a PVDF membrane were stained with CBB R-250, cut out and sequenced with a pulsed-liquid phase protein sequencer (model 477-A; Applied Biosystems).


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
SDS-PAGE profile of purified toxins and free HA1
Type D 16S toxin and HA1 preparation (Fig. 1) were purified by a new procedure as described in Methods. Both preparations in the buffer containing SDS and ß-mercaptoethanol were heated at 100 °C for 7 min and then SDS-PAGE was performed. The 16S toxin showed seven major bands with molecular masses of 140, 93, 53, 50, 33, 24–22 and 17 kDa (Fig. 3a, lane 4). On the basis of our previous report (Ohyama et al., 1995 ), it could be concluded that the 140 kDa band is the non-toxic non-HA, the bands of 93 and 50 kDa are the heavy and light chains of the neurotoxin, respectively. The remaining bands of 53, 33, 24–22 and 17 kDa are subcomponents of HA, which are designated HA3b, HA1, HA3a and HA2. HA3a consists of several proteins with slightly different molecular masses (three in Fig. 3a, lane 4). The preparation purified from peak 1 of the SP-Toyopearl 650M chromatography (Fig. 1) demonstrated a single band with a molecular mass of 33 kDa (Fig. 3a, lane 5). The N-terminal amino acid sequence of this band (SQTNANDLRN---) was identical to that of the HA1 of type D 16S toxin (Ohyama et al., 1995 ). Therefore, the peak 1 protein was concluded to be HA1 which exists in the medium in a free state as found in a type A culture (Inoue et al., 1996 ). When this preparation was subjected to SDS-PAGE without heat treatment, a single protein band with molecular mass of about 200 kDa was evident (Fig. 3a, lane 1). Type C 16S toxin showed seven major bands with molecular masses of 140, 100, 53, 50, 33, 23–22 and 17 kDa (Fig. 3a, lane 3) as has been reported previously (Fujinaga et al., 1994 ).



View larger version (85K):
[in this window]
[in a new window]
 
Fig. 3. SDS-PAGE pattern and immunoblot of purified type C and D 16S toxins and type D free HA1 preparation. (a) CBB-stained gel. The samples were mixed with sample buffer containing ß-mercaptoethanol and incubated at room temperature (lane 1) or heated at 100 °C for 7 min (lanes 2 to 5). Electrophoresis was performed on a 12·5% acrylamide gel. Lanes 1 and 5, free HA1; lane 2, standard proteins; lane 3, type C 16S toxin; lane 4, type D 16S toxin. The molecular-mass markers used were myosin (200 kDa), ß-galactosidase (116 kDa), phosphorylase b (97·4 kDa), BSA (66·2 kDa), ovalbumin (45 kDa), carbonic anhydrase (31 kDa), soybean trypsin inhibitor (21·5 kDa) and lysozyme (14·4 kDa). (b) Immunoblot analysis with the anti-type C non-toxic components serum. Heated (lanes 2 to 4) and unheated (lane 1) samples were separated by SDS-PAGE on a 12·5% acrylamide gel, blotted onto a PVDF membrane and successively reacted with the anti-type C non-toxic components and peroxidase-labelled anti rabbit immunoglobulin sera. Lanes: 1 and 4, free HA1; 2, type C 16S toxin; 3, type D 16S toxin.

 
Immunoblot analysis
The antiserum against type C non-toxic components was reacted with the purified type C and D 16S toxins and the purified type D free HA1 blotted onto the membrane after SDS-PAGE (Fig. 3b). The serum reacted with all subcomponents of HA and the non-toxic non-HA of both type C and D 16S toxins (Fig. 3b, lanes 2 and 3) and with the 33 kDa band of free HA1 (Fig. 3b, lane 4). In the preparation of non-heated free HA1, two bands of 200 and 33 kDa appeared (Fig. 3b, lane 1), although the 33 kDa band was not detected on SDS-PAGE by CBB staining (Fig. 3a, lane 1).

Haemagglutination and haemagglutination-inhibition tests
The minimum concentration of type C and D 16S toxins to give haemagglutination was determined by using enzyme-treated and non-treated (native) human erythrocytes (Table 1). The native erythrocytes were agglutinated by 0·05 µg type C and D 16S toxins ml-1, whereas the purified HA1 did not agglutinate the native erythrocytes even though 100 µg ml-1 was used. When the erythrocytes that had previously been treated with neuraminidase (37 °C for 1 h) were used, 25–50 µg type C and D 16S toxins ml-1 were needed to cause haemagglutination. No haemagglutination was observed in the erythrocytes that had received a prolonged incubation (2–3 h) with neuraminidase. The HA titre of type C and D 16S toxins was markedly reduced by employing the erythrocytes that had been treated with papain or proteinase K. Treatment of erythrocytes with trypsin also reduced HA titre of type C and D 16S toxins. However, the effect of the trypsin treatment was small when compared with those of the papain and proteinase K treatments. Endoglycoceramidase treatment of erythrocytes did not affect the HA titre of type C and D 16S toxins (data not shown).


View this table:
[in this window]
[in a new window]
 
Table 1. Haemagglutinin activity against native- and enzyme treated-erythrocytes

 
To elucidate the carbohydrate-binding specificity of the 16S toxins, haemagglutination-inhibition tests were performed. The HA activity of both type C and D 16S toxins was inhibited to approximately the same extent by NeuAc (Fig. 4). To inhibit 50% of the HA activity of both toxins, 12·5 mM NeuAc was required, whereas no inhibition was observed by 100 mM d-glucose, d-galactose, N-acetyl-d-glucosamine, N-acetyl-d-galactosamine, d-mannose, d-fucose or lactose.



View larger version (10K):
[in this window]
[in a new window]
 
Fig. 4. Effects of NeuAc concentrations on the haemagglutinin titre of type C and D 16S toxins. The haemagglutination was carried out in the presence of various concentrations of NeuAc as described in Methods. Haemagglutinin titre of type C ({circ}) and type D ({triangleup}) 16S toxins were determined.

 
Binding to erythrocytes
The interaction between type D free HA1 or 16S toxin and erythrocytes was analysed using two different methods (Fig. 5). In the method that was performed using microtitre plates, strong binding of erythrocytes to type D 16S toxin coating the plates was observed (Fig. 5a). On the other hand, the amount of native erythrocytes bound to free HA1 was significantly less than that obtained with 16S toxin even though the amount of HA1 coated on the plates was increased to 250 µg ml-1 from 50 µg ml-1. The binding of neuraminidase-treated erythrocytes to the 16S toxin was significantly reduced (Fig. 5a, black bars). NeuAc (100 mM) reduced the binding of erythrocytes to 16S toxin, while galactose (100 mM) did not. The limited binding of neuraminidase-treated erythrocytes to 16S toxin was also reduced in the presence of NeuAc. In the method which analysed the binding of 16S toxin or HA1 to erythrocytes by Western blotting, the same results described above were obtained. 1, type D 16S toxin strongly bound to erythrocytes (Fig. 5b, lane 4), however the HA1 scarcely bound (Fig. 5b, lane 2); 2, the binding of type D 16S toxin was dramatically reduced by treating erythrocytes with neuraminidase (Fig. 5b, lane 7); 3, NeuAc decreased the binding of 16S toxin (Fig. 5b, lane 5), but galactose did not (Fig. 5b, lane 6). The results obtained for the type C 16S toxin (data not shown) were the same as those obtained for the type D 16S toxin.



View larger version (27K):
[in this window]
[in a new window]
 
Fig. 5. Interaction between type D HA1 or 16S toxin, and erythrocytes. (a) The binding of erythrocytes to type D HA1 and 16S toxin was analysed using microtitre plates. Incubation was performed in the absence or presence of NeuAc or galactose (Gal). Bound native erythrocytes (white bars) or neuraminidase-treated erythrocytes (black bars) to type D HA1 or type D 16S toxin were lysed and the amount of bound erythrocytes was determined spectrophotometrically at 405 nm as described in Methods. Data represent the mean of three separate experiments. Error bars indicate SE. (b) The binding of type D HA1 and 16S toxin to erythrocytes was analysed by immunoblotting. Type D 16S toxin and HA1 bound to erythrocytes were separated by SDS-PAGE on 12·5% polyacrylamide gels. Proteins were detected with anti-type C non-toxic components serum. Type D HA1 was incubated with native erythrocytes (lane 2) or neuraminidase-treated erythrocytes (lane 3) at 37 °C for 1 h. Type D 16S toxin (lanes 4 to 9) was incubated in the absence (lanes 4 and 7) or presence of NeuAc (lanes 5 and 8) or galactose (lanes 6 and 9). After incubation at 37 °C for 1 h, native erythrocytes (lanes 4 to 6) or neuraminidase-treated erythrocytes (lanes 7 to 9) were added to the reaction mixture. Lanes 1 and 10 were 0·2 µg type D HA1 and 0·4 µg 16S toxin, respectively.

 
Binding of type C and D 16S toxins and HA1 to glycolipids
The results from the haemagglutination tests and inhibition tests suggest that polysaccharides on the surface of human erythrocytes play an important role in the binding of type C and D 16S toxins. To explore this hypothesis further, we analysed the direct binding of type C and D 16S toxins to glycolipids derived from human erythrocytes by TLC-immunostaining.

As shown in Fig. 6, type C and D 16S toxins mainly bound to two bands in the monosialoglycolipid fraction extracted from human erythrocytes (Fig. 6b, c, lane 2). Purified HA1 also bound to these bands in the monosialoglycolipid fraction (Fig. 7, lane 2). It is now well-established that GM3 and SPG constitute the major gangliosides found in the membranes of human erythrocytes (Watanabe et al., 1979 ). The mobility of two main bands bound by 16S toxins and HA1 was identical to those of GM3 and SPG, respectively.



View larger version (129K):
[in this window]
[in a new window]
 
Fig. 6. Binding of type C and type D 16S toxins to glycolipids extracted from human erythrocytes on a TLC plate. The glycolipids extracted from human erythrocytes were applied on each lane and the TLC plate was developed in the solvent system with chloroform/methanol/water (65:35:8, by vol.). (a) Glycolipids visualized by orcinol/H2SO4 reagent. (b), (c) The plates were incubated with type C (b) and type D (c) 16S toxins and then stained by TLC immunostaining as described in Methods. The glycolipids are: lane 1, neutral glycolipid fraction; lane 2, monosialoglycolipid fraction; lane 3, disialoglycolipid fraction; lane 4, polysialoglycolipid fraction.

 


View larger version (74K):
[in this window]
[in a new window]
 
Fig. 7. Binding of type D HA1 to glycolipids on a TLC plate. Purified glycolipids (1 nmol each) and glycolipid fractions extracted from human erythrocytes were applied on each lane and the TLC plate was developed in the solvent system with chloroform/methanol/water (65:35:8, by vol.). The plates were incubated with type D HA1 and then stained by TLC-immunostaining as described in Methods. Glycolipids are lane 1, neutral glycolipid fraction; lane 2, monosialoglycolipid fraction; lane 3, disialoglycolipid fraction; lane 4, polysialoglycolipid fraction; lane 5, mixture of GM1, GM2, GM3 and GM4; lane 6, SPG; lane 7, GD1a; lane 8, GD1b; lane 9, GD3.

 
Direct binding of type C and D 16S toxins, and type D HA1 to purified glycolipids was examined (Figs 7, 8). Type C and D 16S toxins strongly bound to GM4, GM3, GM2, GM1, GD3, GD1a, GD1b and SPG (Fig. 8b, c; lanes 1 to 8), however they did not bind to neutral (CMH, LacCer, Gb3Cer, Gb4Cer) (data not shown) nor to sulfated (SM3, SM4) glycosphingolipids (Fig. 8b, c; lanes 9 and 10). They did not bind to neuraminidase-treated GM4, GM3, GM2, GD3 and SPG (data not shown). Type D HA1 also bound to gangliosides (Fig. 7, lanes 5 to 9), however it did not bind to neutral (CMH, LacCer, Gb3Cer, Gb4Cer) nor sulfated (SM3, SM4) glycosphingolipids (data not shown).



View larger version (38K):
[in this window]
[in a new window]
 
Fig. 8. Binding of type C and D 16S toxins to purified glycolipids on a TLC plate. Purified glycolipids were applied on each lane and the TLC plate was developed in the solvent system with chloroform/methanol/water (65:35:8, by vol.). (a) Glycolipids visualized by orcinol/H2SO4 reagent. (b), (c) The plates were incubated with type C (b) and D (c) 16S toxins and then stained by TLC-immunostaining as described in Methods. Glycolipids (1 nmol each) are lane 1, GM1; lane 2, GM2; lane 3, GM3; lane 4, GM4; lane 5, SPG; lane 6, GD1a; lane 7, GD1b; lane 8, GD3; lane 9, SM3; lane 10, SM4.

 
Binding of type C and D 16S toxins to glycoproteins
To investigate the binding of type C and D 16S toxins to glycoproteins, the binding test was performed with a membrane preparation of ghosts prepared from human blood cells or neuraminidase-treated human blood cells, as well as fetuin, asialofetuin and human glycophorin (Fig. 9). The ghost membrane preparation demonstrated multiple protein bands on SDS-PAGE by CBB staining (Fig. 9a, lane 5). The results show only two bands were bound by type C and D 16S toxins (Fig. 9b, c; lane 5). However, no bands were detected from ghost proteins prepared from neuraminidase-treated erythrocytes (Fig. 9b, c; lane 6). The N-terminal amino acid sequence of these protein bands was found to be identical (L--TEVAMH---) and in turn matched that of glycophorin A or B of the human red cell membrane reported by Tate & Tanner (1988 ). In the present study, we additionally analysed the binding of type C and D 16S toxins to glycophorin purchased from Sigma. Two protein bands were mainly bound by the 16S toxins, probably glycophorin A and/or B (Fig. 9b, c; lane 4). The mobility of two bands bound by the 16S toxins, both in commercially obtained glycophorin and the ghost membrane preparation, was identical. The type C and D 16S toxins also bound to fetuin; however, they did not bind to asialofetuin (Fig. 9b, c; lanes 2 and 3) nor neuraminidase-treated glycophorin (data not shown). By contrast, the free HA1 showed no specific binding to any of the glycoproteins (data not shown).



View larger version (133K):
[in this window]
[in a new window]
 
Fig. 9. Binding of type C and D 16S toxins to glycoproteins and membrane proteins from human erythrocytes. Glycoproteins and membrane proteins were electrophoresed, electroblotted to a PVDF membrane and the binding of toxins was analysed. Part of the gel (a) was stained with CBB R-250. The PVDF membrane was incubated with type C (b) and D (c) 16S toxins and immunostained as described in Methods. Standard proteins are lane 1. Glycoproteins (10 µg each) are in lane 2, fetuin; lane 3, asialofetuin; lane 4, human glycophorin. Membrane proteins from human erythrocytes (300 µg) are in lane 5 and membrane proteins from neuraminidase-treated human erythrocytes (300 µg) are in lane 6.

 

   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Type D 16S toxin was purified by a newly established method. Until now, the 16S and 12S toxins had to be precipitated with (NH4)2SO4 from the culture supernatant and then separated by column chromatography (Miyazaki et al., 1977 ; Ohyama et al., 1995 ). The recovery of the purified toxins was suboptimal because of cross-contamination between the two toxins. In the new procedure reported here, the 16S and 12S toxins were separated by dialysing the ammonium sulfate precipitate against a phosphate buffer at pH 6·0, resulting in optimal recovery of the purified toxins. In the preliminary experiment, however, this procedure could not be applied for purifying type C 16S toxin because type C 16S toxin did not precipitate at pH 6·0. Type D free HA1 and C3 enzyme were also precipitated at pH 6·0 as well as 16S toxin. Free HA1 was completely separated from 16S toxin by hydroxyapatite column chromatography. Purified free HA1 showed a single band with a molecular mass of 33 kDa, the same as the HA1 subcomponent of the 16S toxin, on SDS-PAGE with heat treatment, but showed a band with a molecular mass of about 200 kDa on SDS-PAGE without heat treatment. The immunoblot analysis with anti-type C non-toxic components serum revealed two bands with molecular masses of 200 kDa and 33 kDa, indicating that the free HA1 may exist as a polymer in culture fluid. When HA1 purified from type A culture fluid (Inoue et al., 1996 ) was subjected to SDS-PAGE without heat treatment, three protein bands with molecular masses of 70, 35 and 30 kDa were seen (unpublished data). These results suggest that any types of free HA1 may readily form polymers, although the molecular mass of those polymers may be different depending on the type of toxin.

Type C and D 16S toxins agglutinated human erythrocytes. Previously, Suzuki et al. (1986 ) and Balding et al. (1973 ) proposed that the NeuAc on the surface of erythrocytes may act as the receptor for types C and D HAs on the basis of haemagglutination and haemagglutination-inhibition tests with crude preparations of types C and D HA or crude type C 16S toxin. They found that haemagglutination titre was reduced by employing erythrocytes pretreated with neuraminidase. In addition, Suzuki et al. (1986 ) reported that the type C HA activity was inhibited by adding gangliosides and fetuin to the reaction mixture. Balding et al. (1973 ) reported that both type C and type D HA activities were inhibited by adding MN active glycoprotein isolated from human red blood cells, indicating that not only glycolipids, but also glycoproteins containing NeuAc, were potential receptors for HAs. In the present study, we confirm these predictions on the basis of the following tests and their inhibition tests with purified type C and D 16S toxins and type D HA1. The experiments examined haemagglutination activity, binding to erythrocytes, and binding to the glycolipids and glycoproteins prepared from erythrocytes. The HA activities of these toxins were reduced by employing the neuraminidase-pretreated erythrocytes. Furthermore, the HA titres of type C and D 16S toxins were reduced by adding NeuAc to the reaction mixtures (Fig. 4). Balding et al. (1973 ) reported that 20 mM NeuAc did not inhibit the HA activities of the crude type C progenitor toxin and of the crude non-toxic components of type D. In the present study, we demonstrate that the presence of NeuAc at concentrations greater than 20 mM dramatically inhibits the haemagglutination caused by both type C and D 16S toxins (Fig. 4). HA titres of type C and D 16S toxins were reduced by the treatment of erythrocytes with proteases (Table 1), but not reduced by endoglycoceramidase treatment. These results suggest that glycoproteins, rather than glycolipids, are critical to the binding of type C and D 16S toxins to the surface of erythrocytes that is associated with haemagglutination. Previously, Balding et al. (1973 ) reported that the HA activity of type C crude progenitor toxin was inhibited by the treatment of erythrocytes with papain, while the HA activity of the non-toxic components of type D progenitor toxin was slightly activated. They also reported that the treatment of erythrocytes with trypsin did not change type C and D HA titres. This discrepancy with the findings reported here might be attributable to differences in the concentration of proteases used, difference in the strains and (or) the purity of the samples used in the experiments. Since the sequence of the non-toxic components of C-ST and D-CB16 progenitor toxins were found to be essentially identical (Fujinaga et al., 1994 ; Hauser et al., 1994 ; Ohyama et al., 1995;Tsuzuki et al., 1990 , 1992 ), it seems reasonable to suppose that both type C and D HAs behave in a similar manner. The interaction between type C and D 16S toxins with erythrocytes was reduced by employing neuraminidase-pretreated erythrocytes and by adding the NeuAc to the reaction mixture (Fig. 5). Binding studies with erythrocyte extracts, purified glycolipids and commercially purchased glycoproteins demonstrated that the type C and D 16S toxins bound to glycolipids containing sialic acid (gangliosides) (Figs 6, 8) and sialoglycoproteins (fetuin and glycophorin) (Fig. 9). However, they did not bind to neutral glycolipids or sulfated glycolipids. The binding of 16S toxin to GM4, GM3, GM2, GD3 and SPG was abolished by treating these gangliosides with neuraminidase. Therefore, it was postulated that the presence of NeuAc in the carbohydrate structure is important for binding of 16S toxins to glycolipids. Furthermore, they did not bind to neuraminidase-treated glycophorin and asialofetuin. These data also support the hypothesis that type C and D 16S toxins bind to both glycolipids and glycoproteins if they contain NeuAc in their carbohydrate structure. Glycophorin is the major sialoglycoprotein of the membrane of human erythrocytes (Marchesi et al., 1972 ), and the majority of the sialic acid existing on the surface of the human erythrocytes is carried not by glycolipids but by glycophorin molecules (Alberts et al., 1995 ). Therefore, it is speculated that type C and D 16S toxins agglutinate human erythrocytes by binding to glycophorin on the surface of erythrocytes.

Type D free HA1 showed no HA activity. The interaction between type D HA1 and erythrocytes was significantly lower than that observed between the 16S toxin and erythrocytes. Type D HA1 was able to bind to sialylglycolipids, but could not bind to sialoglycoproteins. This also supports the hypothesis that sialoglycoproteins, rather than sialylglycolipids, are important for the binding of type C and D 16S toxin to erythrocytes associated with haemagglutination. Tsuzuki et al. (1990) reported that the gene products from E. coli which were transferred by the plasmid containing HA1 gene showed HA activity. However, this plasmid contained the genes for HA2 and part of HA3 in addition to HA1. To account for these results, we speculate that: 1, the formation of a complex involving HA1, HA2 and part of HA3 may be required to cause haemagglutination; 2, the active binding region of free HA1 to erythrocytes may be obscured by the formation of polymer; 3, free HA1 might be processed at C-terminal region, which is required for its binding to erythrocytes to cause haemagglutination; 4, the higher-order structure of free HA1 may be different from that of HA1 genetically produced by E. coli (or HA1 contained in the 16S toxin). Previously, we purified type A HA1 from culture medium (Inoue et al., 1996 ). The type A HA1 did not show HA activity as also seen for the type D HA1 described in this manuscript. However, Fu et al. (1998 ) reported that the type A HA1 possessed HA activity. This discrepancy may be caused by the difference of strain or purification methods. We think the purity of free HA1 is very important because the pooled HA1 fractions after ion-exchange column chromatography showed both weak HA activity and toxic activity (described in Bacteria and toxin section of Methods), which was considered to be caused by contaminated 16S toxin, but these activities were eliminated by purifying this preparation over a hydroxyapatite column (Fig. 2).

It has been postulated that the orally ingested botulinum progenitor toxins are absorbed from the upper intestine into the lymphatic system (Sugii et al., 1977 ). Recently, we proposed that type C HA plays a critical role in the absorption of the progenitor toxins from the small intestine of guinea pigs (Fujinaga et al., 1997 ). On the contrary, Maksymowych & Simpson (1998 ) suggested that neurotoxin itself might be absorbed in the case of type A after performing in vitro experiments with human carcinoma cells. In this paper, we demonstrate that the type C and D 16S toxins bound to both sialylglycolipids and sialoglycoproteins. In our experiment with guinea pigs, type C 16S toxin, but neither 12S toxin or neurotoxin, bound to the epithelial cells of the small intestine: this binding was reduced by pretreatment of the epithelial cells with neuraminidase (Fujinaga et al., 1997 ). The mechanisms of how HA-positive and -negative toxins (including neurotoxin alone) are absorbed and enter the lymphatic vessels through the epithelial cells should be analysed in the near future.


   ACKNOWLEDGEMENTS
 
We thank Professor Shinsei Gasa (Sapporo Medical University) for preparation of the SPG used. This work was supported by grants from the Ministry of Education and Ministry of Health and Welfare of Japan.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K. & Watson, J. D. (1995). The Molecular Biology of the Cell, 3rd edn. New York: Garland Publishing.

Balding, P., Gold, E. R., Boroff, D. A. & Roberts, T. A. (1973). Observations on receptor specific proteins. II. Hemagglutination and hemagglutination-inhibition reactions of Clostridium botulinum types A, C, D, and E hemagglutinins. Immunology 25, 773-782.[Medline]

Fu, F.-N., Sharma, S. K. & Singh, B. R. (1998). A protease-resistant novel hemagglutinin purified from type A Clostridium botulinum. J Protein Chem 17, 53-60.[Medline]

Fujinaga, Y., Inoue, K., Shimazaki, S. & 8 other authors (1994). Molecular construction of Clostridium botulinum type C progenitor toxin and its gene organization. Biochem Biophys Res Commun 205, 1291–1298.[Medline]

Fujinaga, Y., Inoue, K., Watanabe, S., Yokota, K., Hirai, Y., Nagamachi, E. & Oguma, K. (1997). The haemagglutinin of Clostridium botulinum type C progenitor toxin plays an essential role in binding of toxin to the epitherial cells of guinea pig small intestine, leading to the efficient absorption of the toxin. Microbiology 143, 3841-3847.[Abstract]

Gasa, S., Makita, A. & Kinoshita, Y. (1983). Further study of the chemical structure of the equine erythrocyte hematoside containing O-acetyl ester. J Biol Chem 258, 876-881.[Abstract/Free Full Text]

Hauser, D., Eklund, M. W., Boquet, P. & Popoff, M. R. (1994). Organization of the botulinum neurotoxin C1 gene and its associated non-toxic protein genes in Clostridium botulinum C 468. Mol Gen Genet 243, 631-640.[Medline]

Hirano, H. & Watanabe, T. (1990). Microsequencing of proteins electrotransferred onto immobilizing matrices from polyacrylamide gel electrophoresis: application to an insoluble protein. Electrophoresis 11, 573-580.[Medline]

Inoue, K., Fujinaga, Y., Watanabe, T., Ohyama, T., Takeshi, K., Moriishi, K., Nakajima, H., Inoue, K. & Oguma, K. (1996). Molecular composition of Clostridium botulinum type A progenitor toxins. Infect Immun 64, 1589-1594.[Abstract]

Kimura, K., Fujii, N., Tsuzuki, K., Murakami, T., Indoh, T., Yokosawa, N., Takeshi, K., Syuto, B. & Oguma, K. (1990). The complete nucleotide sequence of the gene coding for botulinum type C1 toxin in the C-ST phage genome. Biochem Biophys Res Commun 171, 1304-1311.[Medline]

Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680-685.[Medline]

Maksymowych, A. B. & Simpson, L. L. (1998). Binding and transcytosis of botulinum neurotoxin by polarized human colon carcinoma cells. J Biol Chem 273, 21950-21957.[Abstract/Free Full Text]

Marchesi, V. T., Tillack, T. W., Jackson, R. L., Segrest, J. P. & Scott, R. E. (1972). Chemical characterization and surface orientation of the major glycoprotein of the human erythrocyte membrane. Proc Natl Acad Sci USA 69, 1445-1449.[Abstract]

Miyazaki, S., Iwasaki, M. & Sakaguchi, G. (1977). Clostridium botulinum type D toxin: purification, molecular structure, and some immunological properties. Infect Immun 17, 395-401.[Medline]

Moriishi, K., Syuto, B., Yokosawa, N., Oguma, K. & Saito, M. (1991). Purification and characterization of ADP-ribosyltransferases (exoenzyme C3) of Clostridium botulinum type C and D strains. Infect Immun 173, 6025-6029.

Oguma, K., Syuto, B., Iida, H. & Kubo, S. (1980). Antigenic similarity of toxins produced by Clostridium botulinum type C and D strains. Infect Immun 30, 656-660.[Medline]

Ohishi, I. & Sakaguchi, G. (1980). Oral toxicities of Clostridium botulinum type C and D toxins of different molecular sizes. Infect Immun 28, 303-309.[Medline]

Ohishi, I., Sugii, S. & Sakaguchi, G. (1977). Oral toxicities of Clostridium botulinum toxins in response to molecular size. Infect Immun 16, 107-109.[Medline]

Ohyama, T., Watanabe, T., Fujinaga, Y., Inoue, K., Sunagawa, H., Fujii, N., Inoue, K. & Oguma, K. (1995). Characterization of non-toxic-nonhemagglutinin component of the two types of progenitor toxin (M and L) produced by Clostridium botulinum type D CB-16. Microbiol Immunol 39, 457-465.[Medline]

Sakaguchi, G. & Ohishi, I. (1984). Structure and function of botulinum toxins. In Bacterial Protein Toxins, pp. 435-443. Edited by J. E. Alouf, F. J. Fehrenbach, J. H. Freer & J. Jeljasawicz. London: Academic Press.

Stern, M. & Wentzel, L. M. (1950). A new method for the large scale production of high titer botulinum formol-toxoid types C and D. J Immunol 65, 175-183.[Medline]

Sugii, S., Ohishi, I. & Sakaguchi, G. (1977). Correlation between oral toxicity and in vitro stability of Clostridium botulinum type A and B toxins of different molecular sizes. Infect Immun 16, 910-914.[Medline]

Suzuki, N., Syuto, B. & Kubo, S. (1986). Purification and characterization of hemagglutinin of Clostridium botulinum type C strain stockholm. Jpn J Vet Res 34, 269-278.[Medline]

Tate, C. G. & Tanner, M. J. A. (1988). Isolation of cDNA clones for human erythrocyte membrane sialoglycoproteins {alpha} and {delta}. Biochem J 254, 743-750.[Medline]

Tsuzuki, K., Kimura, K., Fujii, N., Yokosawa, N., Indoh, T., Murakami, T. & Oguma, K. (1990). Cloning and complete nucleotide sequence of the gene for the main component of hemagglutinin produced by Clostridium botulinum type C. Infect Immun 58, 3173-3177.[Medline]

Tsuzuki, K., Kimura, K., Fujii, N., Yokosawa, N. & Oguma, K. (1992). The complete nucleotide sequence of the gene coding for the non-toxic-nonhemagglutinin component of Clostridium botulinum type C. Biochem Biophys Res Commun 183, 1273-1279.[Medline]

Watanabe, K., Powell, M. E. & Hakomori, S. -I. (1979). Isolation and characterization of gangliosides with a new sialosyl linkage and core structures. II. Gangliosides of human erythrocyte membranes. J Biol Chem 254, 8223-8229.[Abstract]

Received 6 January 1999; revised 26 April 1999; accepted 11 June 1999.