Differentiation of the Gene Clusters Encoding Botulinum Neurotoxin Type A Complexes in Clostridium botulinum Type A, Ab, and A(B) Strains

Abstract

We describe a strategy to identify the clusters of genes encoding components of the botulinum toxin type A (boNT/A) complexes in 57 strains of Clostridium botulinum types A, Ab, and A(B) isolated in Italy and in the United States from different sources. Specifically, we combined the results of PCR for detecting the ha33 and/or p47 genes with those of boNT/A PCR-restriction fragment length polymorphism analysis. Three different type A toxin gene clusters were revealed; type A1 was predominant among the strains from the United States, whereas type A2 predominated among the Italian strains, suggesting a geographic distinction between strains. By contrast, no relationship between the toxin gene clusters and the clinical or food source of strains was evident. In two C. botulinum type A isolates from the United States, we recognized a third type A toxin gene cluster (designated type A3) which was similar to that previously described only for C. botulinum type A(B) and Ab strains. Total genomic DNA from the strains was subjected to pulsed-filed gel electrophoresis and randomly amplified polymorphic DNA analyses, and the results were consistent with the boNT/A gene clusters obtained.

Botulinum neurotoxin (boNT) type A (boNT/A) is one of the seven (A to G) immunologically distinct toxins produced by Clostridium botulinum. The toxin may be synthesized by the microorganism in foods, under peculiar conditions, or directly in the human body. The ingestion of foods contaminated with preformed boNT causes food-borne botulism, whereas the absorption of boNT from the body site of synthesis causes either intestinal toxemia botulism (referred to as infant botulism in infants and infant-like botulism in adults) or wound botulism, depending on the route of entry of the boNT-producing clostridia.

All forms of botulism disease are characterized by flaccid paralysis of peripheral muscles resulting from the specific action of boNT at the neuromuscular junctions (16). boNT/A is the serotype most frequently involved in human botulism in the United States (3) and the second most common cause of botulism in Italy and Europe (36). boNT/A is also an effective therapeutic drug against a wide variety of involuntary muscle disorders and for pain relief, and it has been successfully used for aesthetic treatments (20).

The high potency and oral toxicity of boNT have been attributed to different factors, including the fact that it is synthesized by botulinum neurotoxigenic clostridia in combination with one or more associated nontoxic proteins as stable complexes (27). The nontoxic proteins include a nontoxic nonhemagglutinating (NTNH) protein and four hemagglutinin (HA) proteins with different molecular masses (∼54, 33, 20, and 14 kDa) (21). Three forms of boNT complexes are produced by clostridia, with the form depending on the serotype or strain (27): the medium (M) complex (∼270 kDa), which consists of the boNT protein (∼150 kDa) and NTNH protein (∼120 kDa); the large (L) complex (∼400 kDa), consisting of boNT, NTNH, and all four HA proteins; and the extra-large (LL) complex (∼900 kDa), consisting of 2 L subunits. Certain strains of C. botulinum type A produce the LL complex together with L and M complexes (19, 31); these strains are referred to as type A1. Other strains of C. botulinum type A that secrete the M complex only, with no HA proteins, are designated type A2; these strains have been predominantly described in Japan (5, 29, 35).

Genes encoding components of the boNT/A1 and boNT/A2 complexes are arranged in clusters which vary in structure and organization (Fig. 1) (21); three genes (ha33, ha17, and ha70) encoding the HA proteins are present in type A1 genetic clusters, whereas these genes are absent from type A2 gene clusters. On the other hand, a p47 gene, whose product plays an as-yet-undetermined role, although it is believed to be a regulatory one (23), is present only in type A2 gene clusters. Furthermore, the comparison of DNA sequences between boNT/A1 genes from various sources (GenBank accession numbers AF488749, AF461540, M30196, and X52066) (2, 6, 37, 39) and the boNT/A2 gene from the Kyoto-F strain isolated in Japan (GenBank accession number X73423) (38) has revealed a 94% sequence homology between boNT/A1 and boNT/A2 genes, which is lower than the 99% sequence homology observed among boNT/A1 gene sequences. Other differences between A1 and A2 genetic clusters include the orientation and position of the positive regulatory gene botR/A (26), and the sequence of the NTNH gene (24).

FIG. 1.

Schematic representation of type A toxin gene clusters of C. botulinum types A, A(B), and Ab. Genes in boldface type are those selected for PCR amplification in our strategy to identify the type A gene clusters. The computer-predicted XbaI restriction site in the boNT/A2 PCR gene fragment is shown. ¹See references 6 and 39; ²see reference 6; ³see references 7, 18, and 28; ⁴strain of C. botulinum type Ab from Japan, for which only a few segments of the boNT/A gene have been sequenced, has shown a boNT/A2 gene sequence (14).

The structure of the toxin gene clusters encoding components of the boNT complexes has also been elucidated for several strains of C. botulinum type A harboring a bont/B gene in their genome which is either expressed (C. botulinum Ab) or unexpressed [C. botulinum A(B)] (18, 28). These strains share the same boNT gene cluster organization (Fig. 1); the HA genes are in the bont/B gene cluster, whereas the p47 gene is in the bont/A gene cluster, as in type A2 strains. However, the C. botulinum Ab and A(B) strains studied to date display a bont/A1 gene sequence (5, 7, 18, 28), with the possible exception of an Ab strain from Japan for which three segments of the boNT/A gene have been sequenced and found to be closely related to the boNT/A2 gene from the Kyoto-F strain (14).

Information on boNT/A gene clusters is available for only a limited number of strains. In the present work, we describe the identification of boNT/A gene clusters in 57 strains of C. botulinum types A, Ab, and A(B) with different origins through the PCR detection of ha33 and p47 genes and analysis of boNT/A gene polymorphisms. Differentiation of the boNT/A gene clusters correlated well with the results of strain genotyping through pulsed-field gel electrophoresis (PFGE) and randomly amplified polymorphic DNA (RAPD) analysis.

MATERIALS AND METHODS

Clostridium isolates and growth conditions.

A total of 57 C. botulinum strains were used in this study (Table 1). Thirty-seven strains were from the culture collection of the Istituto Superiore della Sanità (Italy's National Health Institute), Rome, Italy, including 17 clinical isolates representing all of the clinical isolates from type A botulism in Italy since 1993 and 20 food isolates. Seven of the clinical isolates were from patients involved in an outbreak of food-borne botulism in Italy caused by the consumption of contaminated “mascarpone” cheese (1); 18 of the food isolates were from mascarpone cheese samples and other cheese products analyzed during that outbreak (12). The remaining 20 strains, including reference strain 62A, had been kindly provided by C. L. Hatheway of the Botulism Laboratory of the Centers for Disease Control and Prevention, Atlanta, Ga.

TABLE 1.

Strains of C. botulinum analyzed in this study and total results of PCR, PCR-RFLP, PFGE and RAPD experiments

Strain	Toxin type	Sourcec	ha33d	p47d	boNT/A RFLP type	boNT/A gene cluster type	No. of clusters generated bye:
							PFGE (Xhol)	PFGE (Smal)	RAPD (OPG5)
CL1Fa	A	c-FB, 1996, Naples	−	+	A2	2	NT	NT	1
CL206a	A	c-FB, 1996, Catanzaro	−	+	A2	2	NT	NT	1
CL3Na	A	c-FB, 1996, Naples	−	+	A2	2	NT	NT	1
CL4Ga	A	c-FB, 1996, Naples	−	+	A2	2	NT	NT	1
CL5Ca	A	c-FB, 1996, Naples	−	+	A2	2	NT	NT	1
CL6Pa	A	c-FB, 1996, Catanzaro	−	+	A2	2	NT	NT	1
CL7Za	A	c-FB, 1996, Catanzaro	−	+	A2	2	NT	NT	1
CL129	A	c-FB, 1997, Potenza	−	+	A2	2	NT	NT	1
CL130	A	c-FB, 1998, Verona	−	+	A2	2	1	1	ND
CL29	A	c-INF, 1998, Rome	−	+	A2	2	1	1	ND
CL103	A	c-INF, 1996, Brescia	−	+	A2	2	1	1	2
CL186	A	c-INF, 2001, Cosenza	−	+	A2	2	1	1	ND
CL260	A	c-INF, 2003, Rome	−	+	A2	2	1	1	ND
CL104	A	c-INF-LIKE, 1997, Novara	−	+	A2	2	2	2	3
CL127	A	F, honey, 1997, Padua	−	+	A2	2	1	1	ND
M 42b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 52b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 106b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 184b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 244b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 309b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 354b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 478b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 529b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 537b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 545b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 568b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 633b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 649b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 664b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 1001b	A	F, mascarpone, 1996	−	+	A2	2	NT	NT	1
M 867b	A	F, mozzarella cheese, 1996	−	+	A2	2	NT	NT	1
M 949b	A	F, cream cheese, 1996	−	+	A2	2	NT	NT	1
CL91	Ab	c-FB, 1993, Milan	+	+	A2	2	3	3	4
CL92	Ab	c-FB, 1993, Nuoro	+	+	A2	2	3	3	ND
CL87	Ab	c-FB, 1995, Catania	+	+	A2	2	4	4	5
S 157	Ab	F, salt, 1999, Bari	+	+	A2	2	3	3	ND
CDC-62A	A	c-FB (reference)	+	−	A1	1	5	5	6
CDC-1757	A	c-FB, 1977, Oreg.	+	−	A1	1	6	6	7
CDC-1839	A	c-FB, 1978, Tex.	+	−	A1	1	7	7	7
CDC-1903	A	c-FB, 1978, Oreg.	−	+	A1	3	8	8	8
CDC-1882	A	c-INF, 1978, Ariz.	−	+	A1	3	8	8	ND
CDC-2712	A	c-INF, 1979, Ariz.	+	−	A1	1	5	5	ND
CDC-2776	A	c-INF, 1979, Utah	+	−	A1	1	5	5	ND
CDC-3680	A	c-INF, 1981, Kans.	+	−	A1	1	6	6	ND
CDC-5196	A	c-INF, 1984, Mont.	+	−	A1	1	9	9	9
CDC-8213	A	c-INF, 1992, Alaska	+	−	A1	1	9	10	9
CDC-2153	A	c-INF-LIKE, 1978, Tex.	+	−	A1	1	10	11	9
CDC-3675	A	c-Unknown, 1980, Oreg.	+	−	A1	1	11	12	10
CDC-588	Ab	c-FB, 1976, Ohio	+	+	A1	3	12	13	11
CDC-1436	Ab	c-INF, 1977, Utah	+	+	A2	2	13	14	8
CDC-2171	A(B)	c-FB, 1978, P.R.	+	+	A2	2	14	15	12
CDC-4893	A(B)	c-FB, 1983, Ill.	+	+	A1	3	15	16	9
CDC-5277	A(B)	c-FB, 1984, W. Va.	+	+	A1	3	15	16	ND
CDC-5178	A(B)	c-FB, 1984, La.	+	+	A1	3	16	17	12
CDC-3517	A(B)	c-INF, 1980, Ariz.	+	+	A1	3	16	17	ND
CDC-1727	A(B)	F, unspecified, 1977, Alaska	+	+	A1	3	17	18	13

^aSee reference 1.

^bSee reference 12.

^cc-FB, clinical (food-borne botulism); c-INF, clinical (infant botulism); c-INF-LIKE, clinical (infant-like botulism); c-Unknown, clinical (form of botulism not known); F, food.

^d+, strains positive for indicated gene by PCR; −, strains negative for indicated gene by PCR.

^eNT, not typeable; ND, not done.

All strains had been shown to produce boNT/A at the time of isolation. Some of the isolates had also been previously characterized for the presence of boNT genes through PCR; of these strains, nine had been shown to contain, in addition to the boNT/A gene, a boNT/B gene in their genome which was either silent [C. botulinum A(B)] (CDC-2171, CDC-4893, CDC-5277, CDC-5178, CDC-3517, and CDC-1727) (9) or expressed (C. botulinum Ab) (CDC-588, CDC-1436, and CL87) (8, 9, 10).

Clostridium stock cultures were checked for purity on egg yolk agar plates (Oxoid, Basingstoke, England). Single colonies from the plates were transferred to TPGY broth (5% Trypticase, 0.5% peptone, 0.4% glucose, 2% yeast extract, 0.1% sodium thioglycollate [pH 7.0]) and grown overnight at 37°C under conditions of anaerobiosis (GasPack jars; Oxoid). One milliliter from each broth culture was centrifuged and washed with 1× TE buffer (10 mM Tris [pH 7.4], 1 mM EDTA), and the pellets were resuspended in 1 ml of sterile distilled water.

PCR assays.

boNT/A and boNT/B gene fragments of 2,280 and 1,284 bp, respectively, were detected through specific PCRs, as reported elsewhere previously (9). The primers H1 and H2 for the amplification of a 774-bp region of the ha33 gene were the same as those described by East et al. (7). Primers p47-A (5′-ACTTATGGTTGGGATATTGTTT-3′) and p47-B (5′-TATATTCTCACCTTCTTTAATTTGC-3′) were selected on the basis of the p47 gene sequence from the Japanese strain Kyoto-F (GenBank accession number X96493) to obtain a 1,236-bp PCR product with the aid of the Primer 3 program, which is available online (http://www.broad.mit.edu/cgi-bin/primer/primer3_www.cgi).

PCRs with each primer pair were carried out separately in a programmable thermal cycler (model PTC100; MJ Research, Waltham, Mass.). For ha33 and p47 gene fragment amplifications, a final volume of 50 μl containing 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 2 mM MgCl₂, 0.2 mM each deoxyribonucleoside triphosphate (100 mM) (Amersham Pharmacia Biotech, Piscataway, N.J.), 0.5 μM primers, and 5 μl of cell suspensions was used; mixtures were preheated in the thermal cyclers at 99°C for 10 min to improve the cell lysis, and then 1.25 U of Taq polymerase (Applied Biosystems, Foster City, Calif.) was added. The cycling conditions were as follows: 35 cycles at 95°C for 45 s, 58°C for 45 s, and 72°C for 45 s, followed by a final extension phase at 72°C for 10 min for the amplification of the ha33 gene fragment and 35 cycles consisting of 1 min at 95°C, 1 min at 60°C, and 1 min at 72°C and a final cycle at 72°C for 10 min for the amplification of the p47 gene fragment.

Ten microliters of the PCR products was resolved by gel electrophoresis by using 1% agarose gels containing ethidium bromide and visualized under UV light. All experiments were performed at least twice.

Restriction fragment length polymorphism (RFLP) analysis of boNT/A gene PCR products.

The remaining boNT/A PCR product volume reactions were purified (QIAquick PCR purification kit; QIAGEN, Crawley, United Kingdom), and 15 μl was incubated with 10 U of XbaI (Roche Diagnostics, Mannheim, Germany) in a final volume of 20 μl containing the appropriate enzyme buffer at 37°C overnight. To verify the patterns obtained, all restriction analyses were performed at least twice.

DNA fragment length polymorphisms were distinguished by electrophoresis on a 2.5% agarose gel.

PFGE experiments.

Genomic DNA extraction was performed as described previously by Hielm et al. (17). DNA was digested with 20 U of either XhoI or SmaI enzyme (Roche Diagnostics) as recommended by the manufacturer. DNA fragments were separated by PFGE in a CHEF Mapper apparatus (Bio-Rad Laboratories, Hercules, Calif.). A constant temperature of 14°C was used, and the electrophoresis parameters were as follows: voltage of 6 V/cm, an angle of 120°, and switch times of 4 to 40 s (linear ramping factor) for 22 h. Gels were stained with ethidium bromide and visualized with a GelDoc 2000 apparatus (Bio-Rad Laboratories). The similarity between macrorestriction patterns was determined by using the Diversity Database software (Bio-Rad Laboratories). Clustering was performed by applying the Dice coefficient and the unweighted pair-group method using arithmetic averages; an intralinkage homology level of ≥80% (2% position tolerance) was chosen as the cutoff for defining the clusters. The results of the clustering analysis were confirmed by visual comparison. PFGE clusters were arbitrarily numbered.

RAPD analysis.

Forty-four selected strains of C. botulinum (Table 1) were subjected to RAPD analysis. The random primer OPG5 (5′-CTGAGACGGA-3′), previously shown to discriminate among C. botulinum strains, was used (11). RAPD amplification was performed with a 50-μl reaction mixture containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 2.5 mM MgCl₂, 200 μM each deoxynucleoside triphosphate, and 0.4 μM primer. Five microliters of each cell suspension, prepared as described above, was used as a template. A negative control in which cells were replaced with 5 μl of sterile distilled water was included. Mixtures were heated at 99°C for 10 min in the thermal cycler, and 1.25 U of Taq polymerase was added. The cycling program was as follows: 1 cycle of 94°C for 8 min; 40 cycles of 94°C for 1 min, 35°C for 1 min, and 72°C for 2 min; and l cycle of 72°C for 5 min. RAPD products were separated by electrophoresis on a 2.5% agarose gel. RAPD patterns were evaluated visually; only intense bands were accepted for the analysis, and minor variations (very faint bands) were neglected. One band difference was considered sufficient to discriminate between profiles. To ensure reproducibility, RAPD experiments were performed at least twice.

Mouse neutralization assay.

Three strains (CL91, CL92, and S157) yielding positive boNT/A and boNT/B PCR products were subjected to a mouse neutralization assay for detecting the concomitant production of neurotoxin types A and B, as previously described (8).

RESULTS

PCR detection of boNT/A and boNT/B genes and toxin production.

The presence of the boNT/A gene was confirmed in all 57 clostridial strains. Nine of the strains (eight from the United States and one from Italy) were also confirmed to be positive for the boNT/B gene (data not shown). We also found that an additional three strains from Italy (CL91, CL92, and S157), previously classified as C. botulinum type A, possessed both boNT/A and boNT/B genes (data not shown). These strains were demonstrated by the mouse neutralization assay to actually produce minor amounts of boNT/B (100 minimum lethal dose units/ml) together with boNT/A (8,000 minimum lethal dose units/ml) (A-to-B ratio, 80:1) and were thus reclassified as C. botulinum Ab (Table 1).

PCR amplification of ha33 and p47 gene fragments.

According to the scheme depicted in Fig. 1, we expected to amplify the ha33 gene from C. botulinum type A1 strains, the p47 gene from type A2 strains, and both the ha33 and p47 genes from the C. botulinum strains possessing both the boNT/A gene and the boNT/B gene (either expressed or unexpressed). To check the effectiveness of our approach, we carried out preliminary PCR experiments with primers specific for ha33 and p47 genes using the strains 62A (C. botulinum type A1 [6, 7]) and CDC-588 (C. botulinum Ab), whose boNT/A and boNT/B gene cluster organization has already been established (18), as templates for the amplification (Fig. 1). An ha33 amplification product of the expected size was obtained from strain 62A, whereas two products whose sizes were consistent with the ha33 and p47 gene fragments were amplified from strain 588 (Fig. 2), as expected.

FIG. 2.

ha33 PCR products (A) and p47 PCR products (B) amplified from C. botulinum type A strains 62A (lanes 1 and 11), CDC-2776 (lanes 2 and 12), CDC-1882 (lanes 3 and 13), CDC-1903 (lanes 4 and 14), CL130 (lanes 9 and 19), and CL29 (lanes 10 and 20); from C. botulinum type A(B) strains CDC-3517 (lanes 5 and 15) and CDC-5277 (lanes 6 and 16); and from C. botulinum type Ab strains CDC-588 (lanes 7 and 17) and CL91 (lanes 8 and 18). M, molecular size standard (1-kb DNA Ladder; Life Technologies, Gaithersburg, Md.).

The results of the amplification of the ha33 and p47 gene fragments for the remaining strains are reported in Table 1. Thirty-three C. botulinum type A strains from Italy were negative for ha33, while they yielded a positive p47 PCR gene product of the predicted size (about 1,236 bp). The remaining four strains of C. botulinum type Ab from Italy (Table 1) were positive for both the ha33 and p47 genes, as expected.

Ten of the C. botulinum type A strains from the United States, including reference strain 62A, were negative for the p47 gene and positive for the ha33 gene; the ha33 PCR products were 774 bp in length, as expected. The remaining two strains (CDC-1882 and CDC-1903) exhibited an ha33 and p47 gene pattern similar to that observed for the Italian strains (i.e., ha33 negative and p47 positive) (Fig. 2). Both genes were amplified from the eight strains of C. botulinum Ab and A(B) isolated in the United States (Table 1).

PCR-RFLP identification of boNT/A gene sequence type.

For restriction endonuclease selection, boNT/A1 and boNT/A2 gene sequences of C. botulinum type A strains available from the GenBank database (accession numbers AF488749, AF461540, M30196, X52066, and X73423) (2, 6, 37-39) were aligned (GCG software package) and analyzed with the Webcutter 2 restriction program (http://www.firstmarket.com/cutter/cut2.html). The theoretical analysis revealed a single XbaI restriction site at position 1536 of the 2,280-bp boNT/A2 gene region amplified by our boNT/A primer set which was absent from the analogous region of the boNT/A1 gene sequences. Restriction analyses of boNT/A gene PCR products with XbaI endonuclease would be expected to produce two restriction fragments of 1,536 and 734 bp from boNT/A2 gene sequences (boNT/A2 PCR-RFLP type) and no restriction products from boNT/A1 gene sequences (boNT/A1 PCR-RFLP type) (Fig. 1), thus allowing the boNT/A gene sequence type to be identified. The PCR-RFLP method to characterize the boNT/A gene sequence was validated on strains 62A and CDC-588, both of which are known to possess a boNT/A1 gene type (6, 18) (Fig. 3).

FIG. 3.

Results of XbaI restriction of boNT/A PCR products from C. botulinum type A strains 62A (lane 1), CDC-1757 (lane 2), CDC-1839 (lane 3), CDC-1882 (lane 4), CDC-1903 (lane 5), CL130 (lane 11), CL186 (lane 12), CL127 (lane 13), CL29 (lane 14), and CL104 (lane 15); from C. botulinum type A(B) strains CDC-3517 (lane 6) and CDC-2171 (lane 7); and from C. botulinum type Ab strains CDC-588 (lane 8), CL91 (lane 9), and S157 (lane 10). M, molecular size standard (DNA Ladder Mix; Genenco, Bakersfield, Calif.).

The results of the RFLP analyses of the boNT/A gene PCR products obtained from the clostridial strains are summarized in Table 1. The boNT/A products amplified from all of the C. botulinum strains isolated in Italy, including the four strains of C. botulinum Ab, yielded two restriction fragments of the predicted sizes upon restriction with XbaI, in accordance with a boNT/A2 gene sequence type. No XbaI restriction fragments were obtained from the boNT/A amplification products of any of the C. botulinum type A strains isolated in the United States, including the reference strain 62A (Fig. 3). The boNT/A PCR products from strains CDC-1903 and CDC-1882 were not cleaved by XbaI, nor were the boNT/A gene amplification products from six of the remaining C. botulinum type Ab and A(B) strains from the United States (boNT/A1 PCR-RFLP type) (Fig. 3). The boNT/A gene PCR products from the last two strains [C. botulinum type Ab strain CDC-1436 and C. botulinum type A(B) strain CDC-2171] were the only ones among the strains from the United States to yield the two-band XbaI restriction profile of the boNT/A2 PCR-RFLP type.

PFGE analyses.

Of the 57 C. botulinum strains, only 31 were able to be characterized by PFGE (Table 1). Smeared DNA profiles were repeatedly obtained from the remaining 26 strains, all from Italy (data not shown); 25 of these 26 strains were the clinical and food isolates related to the mascarpone outbreak, and the other strain (CL129) was a clinical isolate from a single case of food-borne botulism. No heat and/or chemical (formaldehyde) treatment, as described by Hielm et al. (17), was sufficient to inactivate the DNases (data not shown).

Of the 31 typeable strains, 18 different clusters with a similarity level of ≥80% were generated by SmaI, and 17 were generated by XhoI (Table 1); PFGE clusters within the strains were consistent for the two enzymes, except for strain CDC-8213, which clustered with another strain (CDC-5196) when analyzed with XhoI and produced a unique profile when restricted by SmaI.

The 11 typeable strains from Italy produced four different PFGE clusters (two among six C. botulinum type A strains and the other two among four C. botulinum type Ab strains). Interestingly, of the latter strains, S157 from a sea salt sample produced SmaI and XhoI profiles that were indistinguishable from those of C. botulinum Ab clinical strains CL91 and CL92 (Table 1).

Thirteen and 14 different PFGE clusters were obtained with XhoI and SmaI, respectively, among the 20 strains from the United States. Eight clusters with XhoI and nine clusters with SmaI were recognized among the 12 C. botulinum type A strains, four with both enzymes among six C. botulinum type A(B) strains and two with both enzymes for the two type Ab strains CDC-588 and CDC-1436 (Table 1).

Representative PFGE profiles obtained with the XhoI enzyme from the clostridial strains are shown in Fig. 4.

FIG. 4.

XhoI PFGE profiles of C. botulinum type A strains CL103 (lane 1), CL186 (lane 2), CL260 (lane 3), 62A (lane 4), CDC-1839 (lane 5), CDC-5196 (lane 6), CDC-1882 (lane 7), and CDC-1903 (lane 8); C. botulinum type A(B) strains CDC-2171 (lane 9), CDC-5277 (lane 10), CDC-1727 (lane 11), CDC-3517 (lane 12), and CDC-4893 (lane 13); and C. botulinum type Ab strains CDC-588 (lane 14), CL91 (lane 15), and CL87 (lane 16). M, molecular size standard (Lambda Ladder PFG marker; New England Biolabs, Hitchin, United Kingdom).

RAPD analyses.

The 26 strains of C. botulinum type A which were nontypeable by PFGE were subjected to RAPD typing. Eighteen additional strains (4 from Italy and 14 from the United States) representative of the different PFGE clusters were included in the analysis. Thirteen distinct RAPD patterns were detected among the 44 total strains analyzed (Table 1). All 26 PFGE nontypeable strains from Italy shared the same RAPD profile. Another 12 patterns were recognized among the 18 additional strains analyzed. In general, RAPD profiles correlated well with PFGE profiles; however, in some cases, strains with different PFGE profiles shared the same RAPD profile. This result was in accordance with the higher discrimination capability of PFGE compared to RAPD, as is generally accepted (15). Some RAPD profiles are shown in Fig. 5.

FIG. 5.

RAPD profiles obtained with primer OPG5 from strains CL1F (lane 1), M354 (lane 2), M949 (lane 3), M867 (lane 4), CL129 (lane 5), CL103 (lane 6), CL104 (lane 7), CL91 (lane 8), CL87 (lane 9), and reference strain 62A (lane 10). M, molecular size standard (DNA Ladder Mix; Genenco).

DISCUSSION

We developed a strategy to identify the boNT/A gene clusters of 57 C. botulinum type A, Ab, and A(B) strains based on the combination of the ha33 and p47 genes and the RFLP analysis of the boNT/A gene sequence. A similar approach had previously been used by Cordoba et al. (5) to determine the presence or absence of HA and NTNH genes and the boNT/A gene polymorphisms in 25 C. botulinum strains. In our study, the ha33 and p47 genes were detected by PCR. For boNT/A gene sequence type identification, we performed computer-assisted alignment and restriction analysis of the available sequences of the boNT/A gene to select enzymes discriminating between the sequences of the boNT/A gene fragment amplified by our primer set. This fragment spans a region of the gene encoding part of the boNT/A heavy chain, where most of the amino acid differences between toxins A1 and A2 have been observed (38). boNT/A1 and boNT/A2 genes had previously been distinguished from each other by RFLP and DNA probes (5).

Amplification of the ha33 and p47 genes and boNT/A PCR-RFLP analyses were applied to strains of C. botulinum types A, Ab, and A(B) from various sources and with different geographical origins. All 33 strains of C. botulinum type A isolated in Italy displayed a combination consistent with the type A2 gene cluster; they were negative for the ha33 gene and positive for the p47 gene and revealed a boNT/A2 gene sequence type. Strains of C. botulinum type A2 were first described in Japan (29, 34); two additional strains were later identified by Cordoba et al. (one from the United Kingdom and one from Mauritius) (5). Considering that type A2 strains have rarely been described worldwide, the fact that all of the strains of C. botulinum type A isolated to date in Italy are type A2 is outstanding.

Ten of the 12 strains of C. botulinum type A from the United States showed a combination presumptive of the type A1 gene cluster (i.e., they were positive for the ha33 gene and negative for the p47 gene and yielded a boNT/A1 PCR-RFLP type). However, the two remaining strains of C. botulinum type A (strains CDC-1882 from infant botulism and CDC-1903 from food-borne botulism) were negative for ha33 and positive for p47 genes (like type A2 strains), whereas they showed a boNT/A1 PCR-RFLP type. This result was surprising since this combination has been reported only for the boNT/A gene clusters of C. botulinum type A(B) and Ab strains (5, 7, 18, 28) (Fig. 1). Based on these considerations, we hypothesize that a third boNT/A gene cluster exists among strains of C. botulinum type A which we have designated type A3. The fact that only 2 of the 45 C. botulinum type A strains investigated showed a type A3 boNT/A gene cluster suggests that this variant is rare.

The 12 strains of C. botulinum types Ab and A(B) were positive for both the ha33 and the p47 genes, as predicted on the basis of the boNT/A and boNT/B gene clusters (Fig. 1). Although our approach did not allow us to locate the positions of ha33 and p47 genes on the different boNT clusters, based on previous knowledge, only the p47 gene may be assumed to be present in the boNT/A gene cluster (7, 18, 28) (Fig. 1).

All type Ab strains from Italy revealed a boNT/A2 PCR-RFLP type. Among the C. botulinum strains possessing A and B toxin genes studied to date, only one strain from Japan has revealed a boNT/A2 gene sequence (14). Based on the combination p47-positive and boNT/A2 PCR-RFLP type, the type Ab strains from Italy were assigned to the type A2 gene cluster.

Five type A(B) strains and one Ab strain from the United States showed a boNT/A1 PCR-RFLP type; based on the combination p47-positive and boNT/A1 RFLP type, these strains were assigned to the type A3 gene cluster. The two remaining strains [strain CDC-1436 of type Ab and strain CDC-2171 of type A(B)] were the only strains from the United States that displayed a boNT/A2 PCR-RFLP type (type A2 gene cluster).

The results of our analyses showed that the boNT/A genetic clusters differed when the strains from Italy were compared to those from the United States; the type A2 gene cluster predominates among the isolates from Italy, whereas the type A1 gene cluster is predominant among the isolates from the United States. Previously, a similar distinction had been observed between C. botulinum type A strains isolated in Japan and those isolated in the United States based on the molecular sizes of toxin complexes produced by the strains (35). This result suggests that the population in Italy resembles that of Japan, and it might reflect different evolutionary lineages of the clostridia; although this possibility remains to be demonstrated, a similar geographic separation has been hypothesized for strains of neurotoxigenic Clostridium butyricum, which have been isolated only in Italy and China (30).

On the other hand, no correlation between the boNT/A gene clusters and the source of isolation (different clinical forms of botulism and diverse foods) was found, in accordance with findings by Cordoba et al. (5). This result is in contrast with previous observations that HA-negative (type A2) strains were more frequently associated with infant botulism (34).

The PFGE profiles of strains were generally consistent with the boNT/A gene clusters obtained. In particular, the two strains of C. botulinum type A exhibiting a type A3 gene cluster (CDC-1882 and CDC-1903) shared very similar profiles with each restriction enzyme used in the analyses (Fig. 4), indicating that the two strains are strictly related; profiles with both enzymes were different from those of all the remaining strains. The profiles of the Italian strains were distinguished from those of the United States, substantiating the geographic distinction between strains. Some strains of different clinical origins shared the same PFGE profiles, confirming the previous conclusion of no existing relationship between the genetic clones and the outcome of the disease. Twenty-five of the strains from Italy were not typeable by PFGE. Problems related to DNA smearing in PFGE analyses are due to the overproduction of DNases by some microbial strains and have specifically been reported for nonproteolytic C. botulinum strains (type E and some type B and F strains) (17) and for Clostridium spp. other than C. botulinum (33). We experienced the same problems with C. botulinum type A (proteolytic) strains. The strains that were not typeable by PFGE were thus subjected to RAPD analysis, a less discriminatory technique (15) which, however, has the advantage of overcoming DNase overproduction; strains representative of the different PFGE profiles obtained were included in the analyses. We found that all nontypeable strains produced the same RAPD profile, which was distinguished from the profiles of the other strains, suggesting a strict genetic relationship between them. This finding was interesting since most of the nontypeable strains had been isolated from an outbreak of botulism due to contaminated mascarpone consumption and from different samples of mascarpone and other cheeses analyzed during that outbreak (12). In total, the 37 Italian C. botulinum strains yielded a few banding patterns with both PFGE and RAPD techniques; therefore, since they had been isolated in different places and years, a low genetic diversity among strains distributed in Italy can be deduced.

In summary, we identified three different boNT/A gene clusters among 57 strains of C. botulinum which were consistent with PFGE and RAPD genotyping. These clusters potentially encode three different boNT/A complexes. It is already known that boNT/A1 and boNT/A2 complexes differ from each other in terms of molecular size, antigenicity of components, and boNT/A amino acid sequence; in fact, the neurotoxin moieties of the two complexes display a 10% amino acid difference, restricted to the heavy chain of proteins (29, 38), which is involved in the receptor recognition at the neuromuscular junctions (32). The third cluster that we identified theoretically leads to the production of an HA-negative medium-sized (M) complex similar to that produced by type A2 strains, except for the presence of boNT/A1 toxin.

The importance of the different complexes in pathology is still unclear. A number of roles have been attributed to the hemagglutinins (predominantly to the HA33 protein) of boNT complexes, specifically, (i) the linking of two L boNT/A complexes to form the LL complex (19), (ii) protection of boNTs from acidity and proteases of the stomach (31), (iii) the binding to the intestinal cells and translocation of boNTs through the intestinal mucosa (13, 19), and (iv) interaction with sugar moieties on red blood cells and delivery of boNTs into the bloodstream (4, 13). The latter two roles are still controversial since it has been demonstrated that both HA-positive and HA-negative complexes of the toxin can be absorbed from the gastrointestinal tract of mice and enter circulation (25). Moreover, the nontoxic proteins of complexes contribute to the stability and effectiveness of the boNT/A therapeutic agent; in fact, the larger boNT/A complex has been reported to be more effective as a therapeutic reagent, given its higher stability (20). Hence, variation in the composition of hemagglutinins in the boNT/A complexes might affect both the functionality and the potency of the neurotoxin itself.

On the other hand, differences between boNT/A1 and boNT/A2 gene sequences imply that the related boNTs have different amino acid composition and antigenic properties (22), with important implications for the production of specific antibodies against boNT/A; the receptor-binding abilities of boNT/A1 and boNT/A2 toxins should also be investigated.

Additional studies are being performed to sequence the different boNT/A gene clusters from some representative strains in order to confirm them and to identify any additional nucleotide differences that may exist between them. Comparisons of the activities of the related boNT/A toxin complexes in animal models and in cell culture systems will also be needed to eventually differentiate their mechanisms and roles.