Phylogenetic Analysis of Clostridium botulinum Type A by Multi-Locus Sequence Typing

SUMMARY

The genus Clostridium comprises a heterogeneous group of organisms for which the phylogeny and evolutionary relationships are poorly understood. The elucidation of these evolutionary relationships necessitates the use of experimental methods that can distinguish Clostridium lineages, which are time and cost effective, and can be accurately and reproducibly employed in different laboratories. Multi-Locus Sequence Typing (MLST) has been successfully used as a reproducible and discriminating system in the study of eukaryotic and prokaryotic evolutionary biology, and for strain typing of various bacteria. In this study, MLST was applied to evaluate the evolutionary lineages in the serotype A group of Clostridium botulinum. C. botulinum type A has recently been shown to produce multiple subtypes, suggesting that it is not monophyletic as previously reported, but is comprised of distinct lineages. For MLST analysis, we initially evaluated fourteen housekeeping genes (gapdh, tuf, sod, oppB, hsp60, dnaE, aroE, pta, 23S rDNA, aceK, rpoB, 16S rDNA, mdh, and recA) for amplification and sequence analysis. In the first phase of the analysis, thirty C. botulinum type A strains producing subtype BoNTs A1 - A4 were examined. Results of this pilot study suggested that seven of the genes (mdh, aceK, rpoB, aroE, hsp60, oppB, and recA) could be used for elucidation of evolutionary lineages and strain typing. These seven housekeeping genes were successfully applied for the elucidation of lineages for 73 C. botulinum type A strains, which resulted in 24 distinct sequence types (STs). This strategy should be applicable to phylogenetic studies and typing of other C. botulinum serotypes and Clostridium species.

INTRODUCTION

Clostridium botulinum produces its characteristic botulinum neurotoxin (BoNTs), which is classified by the Centers for Disease Control and Prevention (CDC) as one of the six highest-risk threat agents for bioterrorism (“Category A Agents”) (Arnon et al., 2001). BoNTs have traditionally been immunologically distinguished into seven serotypes (BoNT/A-G), among which BoNTs A, B, E and F are known to cause human botulism (Hatheway & Johnson, 1998; Smith & Sugiyama, 1988). Type A is a representative of C. botulinum Group I proteolytic strains as it demonstrates close similarity to proteolytic type F and B strains (Collins & Lawson, 1994). It is of particular importance and interest since it causes the most severe human botulism (Woodruff et al., 1992), is considered to be the most significant bioterrorism threat, and has been increasingly used as a pharmaceutical modality (Johnson et al., 2006).

Recent findings have shown that BoNT/A has substantial sequence diversity and four subtypes have so far been identified (Arndt et al., 2006; Kozaki et al., 1995; Smith et al., 2005). Such sequence variation is not limited to type A, as more than 46 subtypes exist for serotypes A-G (Gimenez & Gimenez, 1995; Smith et al., 2005). With regard to bont/a, it has been represented by bont/a1 (NCBI accession numbers AF461539), bont/a2 (AY953275), bont/a3 (DQ185900), and bont/a4 (DQ185901).

Previous studies have also identified extensive genotypic and phenotypic diversity in these strains (Hatheway & Johnson, 1998; Johnson & Bradshaw, 2001; Kozaki et al., 1995; Smith et al., 2005). To date, phylogenetic analysis and typing of C. botulinum by genetic methods has mainly focused on pulsed field gel electrophoresis (PFGE), ribotyping (rRNA analysis), amplified fragment length polymorphism (AFLP), randomly amplified polymorphic DNA analysis (RAPD) and repetitive element sequence-based PCR (Rep-PCR) (Lindstrom & Korkeala, 2006). Although these techniques have utility, they also have disadvantages. For instance, rRNA analysis focuses on a single locus only and may not represent the diversity of the genome in the species. PFGE is based on restriction digest polymorphisms which is labor intensive, may be variable between different laboratories, and is difficult to use in identifying phylogenetic lineages (Hill et al., 2007; Johnson et al., 2005; Noller et al., 2003).

Therefore, a sequence-based system for assessing genetic relatedness among isolates would have utility in study of this pathogen. Multi-Locus Sequence Typing (MLST) is a method currently being implemented in many laboratories as a means of determining the degree of evolutionary relatedness among various strains of bacterial and eukaryotic species (Gatei et al., 2007; Jost et al., 2006; Lacher et al., 2007; Maiden, 2006; Vassileva et al., 2006; Zadoks et al., 2005). MLST was initially demonstrated in 1998 to be effective in studying the phylogeny of bacteria (Maiden et al., 1998). Since this pioneering study, MLST has been shown to be a useful method for bacterial typing as it has broad applicability in both the range of organisms that can be studied and the breadth of practical and conceptual problems that can be addressed (Urwin & Maiden, 2003). MLST combines advances in high-throughput sequencing, population genetics, and bioinformatics to provide a tool for the study of population and evolutionary biology of various organisms. In this study, we applied MLST analysis for assessing the genetic relatedness of C. botulinum type A strains.

MATERIALS AND METHODS

Bacterial strains and growth conditions

The C. botulinum strains included in this study (Table 1) are from EAJ’s laboratory strain collection. Cultures were grown in 10 mL of TPGY media (50 g/liter trypticase peptone, 5 g/liter Bacto peptone, 4 g/liter d-glucose, 20 g/liter yeast extract, 1 g/liter cysteine-HCl, at pH 7.4) for 2 days at 37°C.

Table 1.

List of strains used in the MLST analysis; their toxin subtype designation, sequencing types and alleles for 73 strains for the 7 selected housekeeping genes

Strain numbera	Strain nameb	recA	rpoB	oppB	hsp60	aceK	mdh	aroE	ST numberc	Subtyped
1*	ATCC 3502	8	9	10	9	13	6	13	1	A1
2*	PHLS5	7	8	9	8	7	8	10	2	A2
3*	A33	7	8	9	8	6	10	9	4	A(B)
4*	A588	6	8	5	8	11	5	11	10	A(B)
5*	14842	1	1	1	1	1	1	1	5	A(B)
6*	14931	7	6	9	7	6	10	9	6	A(B)
7*	519	7	8	9	8	6	10	9	4	A(B)
8*	A4831	8	9	10	7	13	6	8	11	A1
9*	14843	1	1	1	1	1	1	1	5	A(B)
10*	14849	7	8	9	8	6	10	9	4	A(B)
11*	MDa10	7	8	9	8	6	10	9	4	A(B)
12*	14860	7	6	9	7	6	10	6	12	A(B)
13*	657Ba	5	2	4	5	2	2	4	7	A4
14*	657Ba	5	2	4	5	2	2	4	7	A4
15*	A2	7	8	9	8	6	10	9	4	A(B)
16*	CDC/A3	3	3	2	3	5	3	3	3	A3
17*	A4475	6	8	9	6	11	5	12	8	A1
18*	14862	7	8	9	8	6	10	9	4	A(B)
19*	14851	7	6	9	7	6	10	9	6	A(B)
20*	10758	4	7	6	9	8	8	7	13	A1
21*	14853	7	8	9	8	6	10	9	4	A(B)
22*	A3(S)	7	8	9	8	6	10	9	4	A(B)
23*	14844	1	1	1	1	1	1	1	5	A(B)
24*	14050	5	2	4	5	2	2	5	14	A(B)
25*	5311A	6	8	5	8	11	8	13	15	A(B)
26*	A7	8	9	10	9	13	6	13	1	A1
27*	A4567	7	8	9	8	6	10	9	4	A(B)
28*	A1	8	9	10	9	13	6	13	1	A1
29*	A661222	7	9	8	8	4	7	13	16	A1
30*	A207	2	4	3	2	3	4	2	17	A1
31	5328A	5	2	4	4	2	2	4	18	ha−/orfx+ A1
32	667A	7	8	9	8	6	10	9	4	A(B)
33	A4447	6	8	9	6	11	5	12	8	A1
34	A69	8	9	10	7	12	10	13	19	A1
35	A73	8	9	10	9	13	6	13	1	A1
36	A4	7	8	9	8	6	8	10	22	sporo
37	A3	8	9	7	9	13	6	13	21	A1
38	A4438	8	9	10	9	13	6	13	1	A1
39	14847	7	8	9	8	6	10	9	4	A(B)
40	14850	7	8	9	8	6	10	9	4	A(B)
41	726	8	9	10	9	13	6	13	1	A1
42	WHOA	8	9	10	9	9	6	13	9	A1
43	H Lysine	8	9	10	9	13	6	13	1	A1
44	603A	8	9	10	9	13	6	13	1	A(B)
45	547A	8	9	10	9	13	6	13	1	A1
46	10745	8	9	10	9	13	6	13	1	A(B)
47	A14	8	9	10	9	13	6	13	1	A1
48	14869	8	9	10	9	13	6	13	1	A1
49	Hall A1	8	9	10	9	13	6	13	1	A1
50	H formaldehyde	8	9	10	9	13	6	13	1	A1
51	A6	8	9	10	9	13	6	13	1	A1
52	AB Seattle	7	8	11	8	10	10	13	23	A1
53	599A	8	5	10	9	13	6	13	24	A1
54	14906	8	9	10	9	13	6	13	1	A1
55	14853	8	9	10	9	13	6	13	1	A1
56	14846	8	9	10	9	9	6	13	9	A1
57	PHLS2	7	8	9	8	7	8	10	2	A2
58	9289	8	9	10	9	13	6	13	1	A1
59	A172	8	9	10	9	13	6	13	1	A1
60	A109	8	9	10	7	13	6	8	11	A1
61	A16	8	9	10	9	13	6	13	1	A1
62	13060	8	9	10	9	13	6	13	1	A1
63	11A	8	9	10	9	9	6	13	9	A1
64	77A	8	9	10	9	13	6	13	1	A1
65	1690	8	9	10	9	13	6	13	1	A1
66	A15	8	9	10	9	13	6	13	1	A1
67	14854	7	8	9	8	6	10	9	4	A(B)
68	56A	8	9	10	9	13	6	13	1	A1
69	A3955	8	9	10	9	13	6	13	1	A1
70	A608	8	9	10	9	13	9	13	20	A1
71	A5	8	9	10	9	13	6	13	1	A1
72	A9	8	9	10	9	13	6	13	1	A1
73	19A	8	9	10	9	13	6	13	1	A1

^a* represents the initial 30 strains used in housekeeping genes screening study.

^bStrain names refer to Eric A. Johnson’s laboratory’s stocks.

^cST: Sequencing type.

^dA1 has BoNT/A1 gene (see NCBI publication number AF461539); A2 has the BoNT/A2 gene (see NCBI publication number AY953275), A(B) has BoNT/A1 gene and a silent BoNT/B gene (see NCBI publication number NCTC 2916), A3 has the BoNT/A3 gene (see NCBI publication number DQ185900), and A4 has the BoNT/A4 gene (see NCBI publication number DQ185901) and BoNT/B genes (see NCBI publication number EU341304). ha−/orfx+ A1 produces BoNT/A1 but with an ORFX cluster, sporo: Clostridium sporogenes

Total genomic DNA isolation

Total genomic DNA was isolated from the C. botulinum strains by lysozyme and proteinase K treatment as described previously (Dineen et al., 2003). DNA was then diluted to a concentration of 50 ng/& m;L and used for PCR amplification.

PCR amplification and sequencing

PCR amplifications were performed using the GeneAmp® High Fidelity PCR System (Applied BioSystems). PCR cycles were as follows: 95 °C for 2 minutes, followed by 25 cycles of 95 °C for 1 minute, an annealing step for 45 seconds, 72 °C for extension, followed by 1 cycle of 72 °C extension for 10 minutes. For annealing temperatures of the different genes see Table 2 and Table S1. Extension time depends on the lengths of different genes (see Table S1), and a general procedure of 1 minute was utilized to extend a 1kb fragment. Following amplification, all PCR products were isolated using the PureLink^™ PCR Purification Kit (Invitrogen). Sequencing preparations were produced using conditions advised by the University of Wisconsin Biotechnology Center for the ABI PRISM^® BigDye^™ Cycle Sequencing Kit (Applied BioSystems). Sequencing analysis was performed at the University of Wisconsin Biotechnology Center and final sequencing results were analyzed using the Vector NTI Suite Program (Invitrogen). Accession numbers for the resulting nucleotide sequences are as follows: rpoB (EU372261-EU372269), recA (EU372253-EU372260), oppB (EU372242-EU372252), mdh (EU372232-EU372241), hsp60 (EU372223-EU372231), aceK (EU372210-EU372222), and aroE (EU372197-EU372209).

Table 2.

Details on primer sequences and the PCR conditions of 7 housekeeping genes used in MLST analysis

Gene name and primer sequences	Annealing temperature(°C)	Product length(bp)	Product location
rpoB	50	790	3717117–3717906
F 5′-ATG GTA CCC CTA AAG GAT TTA AGC A-3′
R 5′-CGA GAT AGC GAT GGA GGA ATA GAT A-3′
Mdh	48	693	329209–329901
F 5′-TTC TCA GTA ACA ATA GCT GGT GGT G-3′
R 5′-ATC TCC CTT ATC AAC CAC ATA TCC A -3′
aroE	48	749	2678553–2679301
F 5′-TTG TTT CAA TTT ATC GTC CTC CTT T-3′
R 5′-TTC AGA GTC ACC AGA AAT ACA CAA TAA–3′
hsp60	48	767	3508084–3508850
F 5′-TTG GTG GTG TAT TTT CTT TTT CTG G-3′
R 5′-AGC ACC AGG ATT TGG AGA TAG AAG-3′
aceK	44	835	500538-500562
F 5′-TAG GAG CAG ATA TTA ATT GGC ATG T-3′
R 5′-ACA AAT ACT TTT TCT ATA GCA TTT TCT-3′
oppB	44	811	1484690-1485500
F 5′-GGT AAC TTA TGT ATG TTA GAA GTT AAA AAT-3′
R 5′-TTA CTC TTT TAG AAG GAT GAT TTA TAG GAA-3′
recA	48	771	2532760-2533530
F 5′– TTT GCA TTT TCT CTT CCT TGT CCT A-3′
R 5′-AGC ATT AGG TAT AGG GGG AGT TCC T-3′

^aF: forward; R: reverse; aroE: shikimate dehydrogenase gene; hsp6: heat shock gene; mdh: malate dehydrogenase gene; recA: RecA gene; aceK: isocitrate dehydrogenase gene; rpoB: RNA polymerase subunit B gene; oppB: oligopeptide/dipeptide ABC transporter gene.

The C. botulinum Type A ATCC 3502 genome sequence from the Sanger Institute Genome Initiative (http://www.sanger.ac.uk/Projects/C_botulinum/) (Sebaihia et al., 2007) was used to identify various loci for MLST analysis. Genes were chosen for MLST analyses mainly based on their utility in the studies of other bacterial species. Once the fourteen genes were selected, the entire open reading frame of each was used to select primers for initial analysis.

In this initial analysis by MLST and its applicability to the C. botulinum group, 30 isolates (Table 1) were analyzed by performing PCR on 14 genes (Table S1) which were sequenced using an overlapping gene sequencing approach. The primers used to amplify the complete coding frame and the internal primers used in subsequent sequencing reactions are listed in the table. After this analysis was complete, specific regions of all 14 genes were analyzed for the 30 strains to determine which regions contained the optimum degree of divergence to allow for proper MLST analysis. While all 14 genes showed divergence, seven genes were chosen to facilitate further MLST analysis and enable the creation of subfamilies. Regions were evaluated and then entered into the Primer3 program (Rozen & Skaletsky, 2000) to amplify a 700–800 bp section (see Table 2 for primer sequence and gene-specific location). These primers were then used in the PCR amplification of designated gene products.

When final MLST primers were designed, PCR reactions were performed on 73 isolates (Table 1). Most of these isolates were known type A strains from the A1, A2, A3, A4, and A(B) groups. C. botulinum type A neurotoxin was not used as a gene for MLST analysis in order to make the procedure as broad as possible and allow for potential analysis of clostridium strains with other BoNT serotypes using the procedure.

Analysis of MLST data

Sequences were assembled from the resultant chromatograms using the ContigExpress program within Vector NTI (Invitrogen). For each of the seven loci, each sequence obtained was assigned as a distinct allele number. Each isolate is defined by an allelic profile consisting of seven integers, which corresponds to the allele numbers at the seven loci of recA, rpoB, oppB, hsp60, aceK, mdh and aroE. The unique allelic profiles were assigned a sequence type (ST). The resulting STs were analyzed using the programs Sequence Type Analysis and Recombination Tests (START) (Jolley et al., 2001) to organize the various data. Further analysis was conducted using the MEGA3 (Kumar et al., 2004) to identify relationships among the various strains. Also, the final data was compiled and submitted for hosting on the pubmlst.org website (Jolley et al., 2004).

Neighbor-joining trees were constructed using the Kimura 2-parameter model of nucleotide substitution with the MEGA3 software and the inferred phylogenies were each tested with 500 bootstrap replications. Phylogenetic network analysis was conducted with the SplitsTree 4 program (Huson & Bryant, 2006) using the neighbor-net algorithm (Bryant & Moulton, 2004) and untransformed distances (p distance). The number of synonymous substitutions per synonymous site (d_S) and the number of nonsynonymous substitutions per nonsynonymous site (d_N) were estimated by the modified Nei-Gojobori method using MEGA3 (Kumar et al., 2004). Allelic sequences were fit to a nucleotide substitution model using the Datamonkey website and then either the single likelihood ancestor counting (SLAC) was used to fit a codon model to detect selection on individual codons (Pond & Frost, 2005). The SLAC method was also used to calculate the ratio of d_N to d_S and estimate the 95% confidence interval. The φw recombination test (Bruen et al., 2006) as implemented by SplitsTree 4 was used to distinguish recurrent mutation from recombination in generating genotypic diversity.

RESULTS

Initial MLST studies

Initial data indicated the presence of genetic linkages based on BoNT subtypes since several housekeeping genes had sequence variations consistent with BoNT/A subtype identity (A1, A2, A3, A4 and A(B)). Fourteen housekeeping candidate genes listed (Table S1) were initially amplified and sequenced. These were aroE, tuf, hsp60, mdh, recA, aceK, rpoB, 23S rDNA, oppB, pta, 16S rDNA, dnaE, sod, and gapdh. These genes were selected mainly because of their utility in previous bacterial MLST studies. The genes were then analyzed by standard PCR procedures and sequence analysis for 30 C. botulinum type A strains (Table 1 and Table S1). After analyzing the sequence data, 7 housekeeping genes (rpoB, mdh, aroE, hsp60, aceK, oppB, and recA) were selected because of their inherent genetic variability and placement in the C. botulinum genome as the MLST scheme would have an approximately even distribution throughout the genome (Fig. S1). Following this strategy, the 7 housekeeping genes were analyzed in 73 strains of C. botulinum and related clostridia with appropriate primers by sequencing analysis (Table 1 and 2).

Creation of MLST ST Profiles

MLST analysis of 73 Clostridium botulinum and related clostridia yielded locus frequencies that ranged from 8 to 13 alleles per locus (Table 1); 24 unique profile patterns or STs were identified. ST 1 encompassed 29 strains including subtype A1 and A(B) strains; ST 4 included 13 A(B) strains; ST 5 included 3 A(B) strains and ST 9 included 3 A1 strains. STs 2, 6, 7, 8, 11 were represented by two strains each covering a wide array of subtypes including A2, A(B), A4 and A1 (Table 1).

Evaluation of MLST ST profiles

To compare the level of sequence divergence as measured by MLST, we constructed a neighbor-joining (NJ) dendrogram showing the genetic relatedness among the 24 STs (Figure 1A). Bootstrap analysis classified the STs into 4 main groups with genetic distance greater than 0.01. The most divergent strains were ST-5 and ST-17, which was surprising since ST-5 comprised a set of A(B) strains and ST-17 was a BoNT/A1 producing strain. These strains were expected to be more closely related to those possessing the same subtype of BoNT. Their relatively large differentiation from the other strains may indicate that they had split from the main family of C. botulinum cells early in the overall evolution of the species. The fact that they have the same neurotoxin sequence as the other strains supports the view that the evolution of BoNT is not linked to the evolution of the species in toto. Four STs (3, 7, 14, and 18) formed a separate group; ST-7, ST-14 and ST-18 were closely related which was supported by 100% bootstrap analysis. This cluster of STs is interesting since they all were composed of strains with different BoNT sequences; these are BoNT/A(B), BoNT/A4, and BoNT/A1 with an A2 cluster, respectively. The 18 remaining strains form a closely related group with two subgroups; the largest group of nine STs (6, 12, 4, 2, 22, 23, 15, 8, 10) had diverse BoNT profiles, while the A2 and A(B) groups had 90% bootstrap support and seven STs (19, 9, 11, 1, 24, 20 and 21) largely composed of BoNT/A1 producing strains that had 88% bootstrap support (Figure 1A).

Figure 1.

Genetic relatedness among the 24 STs based on MLST sequence differences, A. The genetic relatedness based on Neighbor-joining (NJ) dendrogram without recombination. B. Tree-like Neighbor-net for the 24 STs based on SplitsTree analysis that allows for recombination. In both trees, genetic distance is measured by p-distance from pairwise comparison of sequence differences.

Based on these data, we decided to determine if sources of genetic variation other than classic evolution could have caused the observed variation among the genes. Specifically, we evaluated the role that recombination had in the genetic relationships among sequence types. To examine this, we constructed and examined a phylogenetic network (Figure 1B) based on SplitsTree analysis (Bryant & Moulton, 2004). This analysis does not force the sequence data into a bifurcating tree and allows for numerous parallel paths indicative of the presence of phylogenetic incompatibilities in the divergence of STs. Such incompatibilities could arise from recombination or recurrent mutations in the MLST loci. To detect recombination, we used the φ_w test, which has been shown in SplitsTree analysis to discriminate between recurrent mutations and recombination in a variety of circumstances (Bruen et al., 2006). In application to the concatenated sequences of the 241 STs, there were 151 informative sites and the φ_w test was found to show statistically significant evidence of recombination (p <0.001). Interestingly, although there is significant recombination among very clearly related STs, there is no evidence of recombination among the more distantly related strains matched by STs 3, 5, 7, 14, 17, and 18 (Figure 1B). This result is intriguing considering the observation that there is a low degree of similarity between the strains in these STs at the BoNT level. In fact all but 2 of these STs have their own unique BoNT profile as only STs 5 and 14 have A(B) profiles. The remainder of the STs all represent unique strains with different neurotoxin profiles ranging from strains with a BoNT/A1 gene to more distinct strains like those having a BoNT/A3, BoNT/A4 or a BoNT/A1 gene with an A2 cluster.

DISCUSSION

MLST was utilized for analysis of the genetic diversity and phylogenetic analysis within C. botulinum serotype A, in which four distinct subtypes of BoNT have recently been elucidated (Arndt et al., 2006). The genes chosen in this study for MLST analysis were selected because of their utility in previous MLST studies with other bacterial species and their distinct distribution in the C. botulinum chromosome. Initial analysis of 14 candidate genes were combined with the genomic location of each gene estimated from the genome sequence of C. botulinum ATCC 3502 to determine seven genes that provided a representation of the genetic relatedness of the strains on a genome scale. The loci used in the final analysis were hsp60, rpoB, oppB, mdh, recA, aceK, aroE.

Based on MLST studies of other pathogenic bacterial species, it is common to include a virulence gene as one of the loci used to determine genetic relationships within a bacterial group. In our analysis, this strategy was not followed since the BoNT gene does not have sufficiently distinct loci that could be used for MLST analysis. Further, it would limit the applicability of the MLST system to other C. botulinum serotypes and neurotoxigenic clostridia of different species. At this time, there are 5 alleles of this gene known (A1, A2, A3, A4, and A1 in A(B) strains), which would not provide adequate genetic variation for MLST analysis.

There are several other genes associated with BoNT that could also be used to study genetic variation, but were also not chosen for analysis. These include genes within the distinct toxin gene clusters in type A (Jacobson et al., manuscript in press). There are two primary types of neurotoxin clusters in C. botulinum type A (Jacobson et al., manuscript in press). The difference between the two basic clusters is substantial as three of the four genes consist of either a set of hemagglutinin (HA) genes or a set of genes of unknown function called “orfx”s. The only gene that is common among the clusters is the nontoxic nonhemagglutinin gene (ntnh). However, ntnh would pose problems as a MLST locus since two copies are present in strains with bivalent designations, such as A(B). Therefore, we selected seven housekeeping genes spread evenly across the genome for the MLST analysis and did not include the BoNT gene or genes within the BoNT clusters.

Several interesting results emerged from this study, specifically that there appears to be a significant amount of genetic association between the A1 and A(B) strains since there was intersection between the two types in several ST groups, particularly ST-1 (Table 1). There are several possible hypotheses to explain this diversity. The most likely is that the evolution of the BoNT/A genes and their respective gene clusters differs significantly from the evolution of the species in toto. This hypothesis is supported by other observations related to evolution of BoNT/A and toxin gene clusters. Since BoNT acts solely on the nervous systems of higher eukaryotes, the selective pressure for its evolution is enigmatic. The BoNT gene and the structure of BoNT/A and BoNT/B have a highly mosaic composition (Arndt et al., 2006). The acquisition of the toxin gene clusters may have resulted by acquisition of eukaryotic genes, e.g. by viral infection, and by gene transfer that occurred during evolution of C. botulinum (DasGupta, 2006; Johnson and Bradshaw, 2001). This would explain how members of this species, which differ widely in genetic and phenotypic properties, possess BoNT/A genes and an associated protein cluster that are highly conserved. It also explains why certain C. botulinum isolates such as those possessing the BoNT/A3 and BoNT/A4 have such a low degree of relatedness compared to strains possessing BoNT/A1 gene, while the BoNT/A genes have relatively high homology of 80–90%.

The MLST analysis also supports that recombination was a prominent driving force contributing to the relatedness of strains tested in this study. However, there were a few outliers (STs 3, 5, 7, 14, 17, and 18). Given the uniqueness of most of these strains’ BoNT sequences, it suggests that these strains may have been geographically or ecologically isolated, such that there was limited interaction with other strains within the C. botulinum species. This could in turn explain why the BoNT profiles of these strains are unique, as it is possible that they may have evolved to different degrees with respect to eukaryotic gene acquisition, gene transfer, and recombination. This hypothesis is supported by the unique ST-3 pattern of BoNT/A3 strains, as only one outbreak of botulism has been attributed to BoNT/A3, whereas BoNT/A1 and BoNT/A2 have been involved in numerous botulism outbreaks.

Another set of interesting outliers in this study were STs 5 and 17. At this time, there is little known about the strains possessing these STs. The three in ST-5 are A(B) strains, while the one ST-17 is an A1 strain, and each is distinct from other analyzed strains. Initial experiments performed in our laboratory have indicated that the ST-5 bacteria possess a unique neurotoxin cluster arrangement compared to the standard clusters observed in the literature (Jacobson et al., unpublished data).

Additionally, strain 5328A (ST-18) has a BoNT/A1 gene associated with an orfx cluster arrangement that is similar to that seen in C. botulinum strains having the BoNT/A2 gene. This type of cluster arrangement is present in about half of BoNT/A1 strains, but only when a BoNT/B silent gene is also present and associated with a HA cluster. In strain 5328A, the BoNT/B silent gene cluster is absent and has only a BoNT/A1 cluster. This appears to be unusual but becomes more revealing when compared with a A(B) strain and the BoNT/A4 strain, which also has a BoNT/B gene cluster. The implications of both the lack of a cryptic BoNT/B and its relatedness to these other strains have yet to be completely explained and will require further analysis.

In summary, MLST is a nucleotide sequence-based approach with many advantages for subtyping and phylogenetic analysis of various organisms. We show in this study that MLST is an efficient and discriminatory method for strain differentiation and phylogenetic analysis of Clostridium botulinum. Twenty-four unique ST lineages were identified from analysis of 73 C. botulinum type A strains. In future studies, we will expand this MLST procedure to other BoNT-producing bacteria including serotypes B -G. This will be of value in further elucidating and understanding the genetic relatedness in the diverse species C. botulinum. Lastly, this strategy may also be applicable to phylogenetic studies of other Clostridium species.

Table 3.

Genetic variability of 7 genes in 24 STs. The number of variable sites ranges from 39 to 73. The synonymous rate of substitution (d_S) and non-synonymous rate of substitution (d_N) range among alleles and average to 5.40 and 0.69 among all sites. Estimates of d_N /d_S and CI were calculated by method of Pond and Frost (2005). Both oppB and aroE have larger d_N/d_S suggestion these alleles are under more relaxed selection.

Locusa	# of sites	# of variable sites	# of alleles	dS × 100 (mean ± SE)	dN × 100 (mean ± SE)	dN/dS [95% CI]
recA	423	39	8	4.77 ± 0.82	0.14 ± 0.08	0.015 [0.005, 0.035]
rpoB	407	46	9	7.63 ± 1.14	0.08 + 0.07	0.003 [0.000, 0.012]
oppB	447	66	11	4.89 ± 0.85	1.23 ± 0.29	0.194 [0.135, 0.270]
hsp	498	51	9	3.72 ± 0.70	0.63 ± 0.18	0.050 [0.031, 0.075]
aceK	453	60	13	7.06 ± 1.27	0.88 ± 0.25	0.059 [0.037, 0.088]
mdh	447	52	10	5.31 ± 0.97	0.39 ± 0.13	0.035 [0.018, 0.060]
aroE	411	73	13	4.23 ± 0.77	1.52 ± 0.26	0.184 [0.132, 0.247]
Total	3186	387	24	5.40 + 0.37	0.69 + 0.08	0.067 [0.056, 0.079]