More than 170 strains of BoNT-producing clostridial strains were analyzed by different molecular methods to evaluate the genetic diversity and understand the evolutionary history within this species. The conserved 16S rRNA gene sequences illustrate how the different serotypes are closely related to other clostridial species, the AFLP analyses are consistent with the 16S rRNA gene data but add significant resolution to the genomic background that contains the different neurotoxin genes, and finally, the diversity within and between the seven BoNT serotypes reveals a completely different phylogeny within this species that suggests intra- and interspecies transfer of these genes.
The taxonomy of the C. botulinum
species has historically been based on the identification and/or expression of botulinum toxin genes (38
). Since C. butyricum
and C baratii
strains that contain BoNT genes have been identified (2
), the taxonomy of the toxin-producing clostridia has become more complex. The dendrogram generated using 16S rRNA gene sequence data suggests that the different botulinum neurotoxins that define the species Clostridium botulinum
are actually contained in genomes from four different clostridial species. The 16S rRNA gene dendrogram demonstrates that BoNT/A-, BoNT/B-, and BoNT/F-producing strains are closely related to each other and to Clostridium sporogenes
and probably evolved from a common ancestor. However, the genomes for the BoNT/C-, D-, E-, and G-producing strains have 16S rRNA gene sequence profiles that closely align to distant clostridial relatives including C. novyi/C. haemolyticum
, C. baratii
, and C. subterminale
(Fig. ). The results reported here should not change the basic nomenclature for C. botulinum
in order to avoid confusion and because these taxonomic designations have been based on strong phenotypic as well as genotypic characteristics. However, the presence of related toxin genes in distantly related clostridia serves as a reminder that horizontal gene transfer has played a significant role in the evolution of Clostridium botulinum.
AFLP analysis of these strains illustrates clustering by group designation and by toxin serotype. The AFLP-based dendrogram divides the strains into clusters that follow the group I to group IV designations, which are based on physiological characteristics. AFLP analysis clearly separates the proteolytic and nonproteolytic groups and shows the relationship of the genomic backgrounds among strains that are usually defined by the expression of a single 3.8-kb BoNT gene into one of seven different serotypes. This AFLP analysis shows a close relationship of BoNT/A1 subtypes to the A1(B) strains that are distant from the BoNT/A2 and BoNT/A3 subtypes that lie within the BoNT/B1 and BoNT/B2 subtypes. AFLP also shows relationships among the proteolytic BoNT/B and BoNT/F isolates that are mirrored in the nonproteolytic BoNT/B and BoNT/F branches. AFLP analysis supports the group III clustering of BoNT/C and BoNT/D serotypes and the clustering of the group IV BoNT/G strains as distinct from the other serotypes.
Four out of the five bivalent strains in this study cluster together within a branch of the AFLP-based dendrogram that also contains strain A254, which produces BoNT/A3. This branch is also related to a branch containing three BoNT/A2-producing strains, including the remaining bivalent strain, Af695. These bivalent strains contain BoNT/A, B, and F genes and were all isolated from infant botulism cases in different geographic locations. The genomes of these isolates cannot be distinguished by AFLP, yet these strains contain different combinations of neurotoxin genes. The sequences of the individual neurotoxin genes show that the BoNT/B gene sequence in all of these strains is of the same subtype (not identical sequences) but that the BoNT/A genes differ, representing different subtypes. Ab149 contains a BoNT/A2 subtype sequence, but Ba207 contains a completely new BoNT/A subtype that we have termed BoNT/A4. These results suggest either that two lineages of a single strain already carrying the BoNT/B gene acquired the BoNT/A2 and BoNT/A4 genes horizontally or that two strains carrying BoNT/A2 and BoNT/A4 genes both acquired the same BoNT/B gene horizontally. Southern blotting or genome analysis of toxin gene integration sites would be necessary to distinguish between these possibilities.
Four of the five strains producing two BoNT serotypes (bivalent strains) were isolated from infants with botulism. This high proportion of bivalent strains found in infants might reflect sample bias within this collection, but this has been reported previously by others (4
). Of the 10 strains isolated from infants, 4 were found to be bivalent in this study. These 10 strains that affected infants are located in different branches of the AFLP dendrogram and include a C. butyricum
BoNT/E-producing strain (E543) from Italy. An examination of the sequences of the BoNT/A, B, and E genes from these strains from infants shows that the toxin gene frequently represents a unique cluster within the serotype. Within both the BoNT/A and the BoNT/B gene sequence-based dendrograms, three of the nine clusters in the trees contain BoNT produced by strains from infant cases; all of the bivalent strains producing BoNT/B form a unique cluster, as does the single strain (E543) producing BoNT/E. It must be noted, however, that the strains obtained from infants in this collection were deliberately chosen for their unusual characteristics and that a large collection of infant isolates may show higher percentages of the more common BoNT/A1 and BoNT/B1 subtypes.
The neurotoxin gene sequence comparisons of all of the toxin serotypes (serotypes A to G) suggest that the BoNT gene has evolved separately in different genomic backgrounds. The dendrogram indicates that the seven BoNT genes form three distinct clusters: a large cluster consisting of the A, E, and F neurotoxins; a second cluster comprised of the B and G toxins; and a third cluster comprised of the C and D toxins. These relationships are different from the group I to group IV designations supported by the 16S rRNA gene sequences and AFLP analysis. This discordant phylogeny suggests gene transfer among different clostridial species and suggests that C. botulinum
has contributed to the movement of the BoNT gene into various genetic backgrounds. The nucleotide differences within these neurotoxin genes are probably a result of both natural variation and selection pressure. Recombination events, similar to that illustrated in Fig. , where an A1/A3 recombination created BoNT/A2, can also be found within other toxin gene lineages, including many C/D and D/C interserotype recombination events that have previously been reported (31
). Several recombination events within the nontoxic nonhemagglutinin genes of A1, B, and F strains have also been described (11
The current analysis of 134 BoNT/A, B, and E toxin genes significantly increases our understanding of the extent of subtype variability within these three serotypes. The neurotoxin sequences demonstrate that there is more diversity within these toxin serotypes than previously known (summarized in reference 39
). Two new BoNT/A genes, one new BoNT/B gene, and two new BoNT/E genes were identified. The two new BoNT/A genes clearly represent new BoNT/A subtypes that we have termed BoNT/A3 and BoNT/A4. Subtypes have historically been defined by the differential binding of monoclonal antibodies (13
), and the 15% and 11% amino acid differences between BoNT/A1, A3, and A4 would certainly result in differential binding of some BoNT/A monoclonal antibodies (39
). The toxins encoded by the new BoNT/B gene (BoNT/B3) and the new BoNT/E genes (BoNT/E2 and E3) differ from BoNT/B1 and BoNT/E1 by 4%, 1%, and 2% at the amino acid level, respectively. It is not clear whether these new toxins represent new toxin subtypes using the historical standard of monoclonal antibody binding. While single amino acid changes can cause a loss of antibody binding, whether the amino acid differences in these toxins are large enough to result in differential monoclonal antibody binding is unknown and must await the completion of studies using panels of monoclonal antibodies. However, lacking monoclonal antibody studies, subtypes could also be defined based on nucleotide or, more appropriately, amino acid differences, especially where multiple members are identified from different strains. This is the case for the BoNT/E2 and E3 genes.
Accurate analyses and understanding of the recombinations between toxin genes of different serotypes and subtypes may be more helpful for identifying potential vaccines and therapeutic antibodies than relying on phylogenetic dendrograms or overall pairwise sequence distances. For example, BoNT/A2 represents a recombination of the 5′ end of the BoNT/A1 light-chain gene with the 3′ end of the BoNT/A3 gene. This analysis permits the identification of regions of BoNT that could be used to generate antibodies that can cross-react with all three subtypes. Similarly, knowledge of the recombination site between BoNT/C and D will allow the identification of targeted regions for the generation of antibodies that cross-react with chimeric BoNT/C and D.
In conclusion, the neurotoxins produced by Clostridium tetani
, C. butyricum
, and C. baratii
are as similar, or more similar, to C. botulinum
neurotoxins as the various serotypes of BoNT are to each other (36
). Historically, the expression of these neurotoxins has been used to taxonomically identify these clostridia as C. botulinum
or C. tetani
. The presence of these toxins in different genetic backgrounds suggests their movement both within the species and among other species. Most of these bacteria are distributed throughout the world, yet there is no known geographical relationship to the genetic diversity. Environmental niches, geographic distribution, and gene transfer mechanisms among these spore-forming clostridia must all interact to produce the sequence diversity observed in one of the most lethal neurotoxins known. The BoNTs produced by these clostridial species show sequence differences both within and between serotypes. Identifying the extent of these differences is the crucial first step in the development of improved diagnostics and therapeutics for the treatment of botulism.