Genetic Homogeneity of Clostridium botulinum Type A1 Strains with Unique Toxin Gene Clusters

Abstract

A group of five clonally related Clostridium botulinum type A strains isolated from different sources over a period of nearly 40 years harbored several conserved genetic properties. These strains contained a variant bont/A1 with five nucleotide polymorphisms compared to the gene in C. botulinum strain ATCC 3502. The strains also had a common toxin gene cluster composition (ha−/orfX+) similar to that associated with bont/A in type A strains containing an unexpressed bont/B [termed A(B) strains]. However, bont/B was not identified in the strains examined. Comparative genomic hybridization demonstrated identical genomic content among the strains relative to C. botulinum strain ATCC 3502. In addition, microarray data demonstrated the absence of several genes flanking the toxin gene cluster among the ha−/orfX+ A1 strains, suggesting the presence of genomic rearrangements with respect to this region compared to the C. botulinum ATCC 3502 strain. All five strains were shown to have identical flaA variable region nucleotide sequences. The pulsed-field gel electrophoresis patterns of the strains were indistinguishable when digested with SmaI, and a shift in the size of at least one band was observed in a single strain when digested with XhoI. These results demonstrate surprising genomic homogeneity among a cluster of unique C. botulinum type A strains of diverse origin.

Clostridium botulinum is a gram-positive anaerobic bacterium that is identified by the production of botulinum neurotoxin (BoNT). BoNTs are extremely potent neurotoxins that induce flaccid paralysis and are potentially fatal because they can cause respiratory failure. There are seven serological types of BoNT (A to G) that are defined by the ability of serotype-specific equine antitoxins to neutralize the BoNT and prevent signs of botulism in the mouse bioassay.

The species C. botulinum is genetically diverse, containing four phylogenetic groups that have been identified based on 16S rRNA nucleotide sequence comparison (2). These groupings also coincide with specific metabolic differences. Group I strains are proteolytic and produce BoNT types A, B, and F. This group also includes Clostridium sporogenes, which is a nontoxic species that metabolically resembles C. botulinum. Group II contains strains that are nonproteolytic and produce BoNT types B, E, and F. Group III strains produce BoNT types C and D and typically cause botulism in animals. Finally, group IV strains produce the rare toxin type G and are generally referred to as C. argentinense. Rarely, strains of Clostridium butyricum and Clostridium baratii produce BoNT types E and F, respectively (5, 12).

The nucleotide sequences of the genes encoding BoNT/A-G differ by as much as 50%. Recently, Hill et al. (6) reported differences in the nucleotide sequence of the bont gene among type A, B, and E strains. Based on phylogenetic analysis of the bont/A sequence, four subtypes were identified (A1 to A4). The nucleotide sequences of the subtypes differ by up to 8%, and the predicted amino acid sequences differ by up to 16% (6). The bont genes are found in BoNT gene clusters, which include the genes for various regulatory and neurotoxin-associated proteins. The toxin gene clusters of subtypes A1 to A4 differ in their composition (7, 17). As shown in Fig. 1, the toxin gene cluster of subtype A1 includes the gene encoding the nontoxin nonhemagglutinin (ntnh), a regulatory gene (botR), and an operon encoding three hemagglutinins (ha70, ha30, and ha17). The A2 subtype BoNT gene cluster contains a ntnh gene with significant differences compared to ntnh/A1, botR, and several genes of unknown function (orfX1, orfX2, orfX3, and p47). Based on these differences in gene composition, we have designated the A1-like cluster ha+/orfX− and the A2-like cluster ha−/orfX+. Toxin gene clusters containing bont/A3 or bont/A4 are also ha−/orfX+ (7, 17). In type A strains containing an unexpressed type B BoNT gene [termed A(B) strains], the bont/A gene is associated with the ha−/orfX+ cluster, while the bont/B gene is associated with the ha+/orfX− cluster (15).

FIG. 1.

Toxin gene cluster organization. The toxin gene cluster organizations of A1, A2, and ha−/orfX+ A1 strains are shown. Also shown is the intergenic spacing between botR-orfX1 and orfX1-orfX3.

In 2006, the Centers for Disease Control and Prevention (CDC) identified type A BoNT in two bottles of commercially prepared carrot juice associated with cases of botulism in both Georgia and Florida (1). Bacterial isolates obtained from the implicated carrot juice in Georgia (strain CDC51303) and Florida (strain CDC51348) were found to be genetically distinct. PCR analysis demonstrated the presence of only bont/A in strain CDC51303 and both bont/A and bont/B in strain CDC51348. Only type A BoNT was detectable by mouse bioassay and enzyme-linked immunosorbent assay in cultures of either strain, indicating that bont/B was not expressed in strain CDC51348.

Nucleotide sequencing of bont/A from strain CDC51303 demonstrated the presence of unique nucleotide polymorphisms compared to the established bont/A1 sequences in A1 and A(B) strains. PCR mapping of the BoNT gene cluster revealed that bont/A1 of strain CDC51303 resided within a ha−/orfX+ cluster.

These unusual genetic features prompted us to find other strains with similar properties. Franciosa et al. (4) previously reported that two C. botulinum type A strains, CDC1882 and CDC1903, lacked the ha genes but harbored the p47 gene. In addition, restriction fragment length polymorphism analysis of bont/A in C. botulinum strains CDC1882 and CDC1903 indicated that the gene was likely an A1 subtype (4). More recently, Hill et al. (6) reported the bont/A1 sequence of C. botulinum strain CDC297. This strain harbors all five of the bont/A nucleotide changes found in C. botulinum strain CDC51303. Similarly, C. botulinum strain CDC5328 was determined to have the ha−/orfX+ toxin gene cluster and a bont/A1 gene with the nucleotide changes observed in C. botulinum strain CDC51303 (7).

In the present study, we sought to confirm the toxin gene cluster organization and toxin subtype of the putative ha−/orfX+ A1 strains and examine their genetic relatedness by comparative genomic hybridization (CGH). We found that ha−/orf+ A1 strains isolated from different origins over several decades appear to be clonally related. Although the genetic diversity of C. botulinum has not been fully examined, this finding was unexpected since previous studies have demonstrated genetic heterogeneity among type A strains using techniques such as pulsed field gel electrophoresis (PFGE) and multiple-locus variable-number tandem repeat analysis (MLVA) (10, 13). Moreover, the unique bont/A1 allele found in ha−/orfX+ A1 strains appears to be present in strains that are genetically indistinguishable.

These findings elucidate genetic “signatures” for identification and subtyping of ha−/orfX+ A1 strains and suggest possible mechanisms by which these strains emerged.

MATERIALS AND METHODS

Growth of bacterial strains.

C. botulinum strains were grown anaerobically at 35°C on either egg yolk agar medium or in Trypticase-peptone-glucose-yeast extract (TPGY) medium. Stock cultures were stored in bovine brain medium at 4°C. Strains used in the present study are identified in Table 1.

TABLE 1.

Strains used in this study

Strain	Cluster type	Subtype	Yr isolated	Origin	Source	Source or reference
ATCC3502	ha+/orfX−	A1	∼1940	Unknown	Unkown	16
CDC21547	ha+/orfX−	A1	1976	Venezuela	Unknown	This study
FRI-H1A2	ha−/orfX+	A2	1994	New Jersey	Honey	8
CDC51303	ha−/orfX+	A1	2006	Georgia	Carrot juice	This study
CDC1882	ha−/orfX+	A1	1978	Arizona	Stool	4
CDC1903	ha−/orfX+	A1	1978	Oregon	Stool	4
CDC297	ha−/orfX+	A1	∼1969	New York	Liver paste	6
CDC5328	ha−/orfX+	A1	1985	Maryland	Tissue	This study

Genomic DNA extraction.

Ten-milliliter TPGY cultures of C. botulinum strains were incubated overnight at 35°C. Cultures were centrifuged for 10 min at 4,000 rpm in a swinging-bucket centrifuge to pellet the bacteria. The pellet was resuspended in 300 μl of Tris-EDTA (TE) containing 30 mg of lysozyme (Sigma, St. Louis, MO)/ml and incubated for 30 min at 37°C. A MasterPure DNA purification kit (Epicenter, Madison, WI) was used for subsequent extraction steps as follows: 300 μl of 2× tissue and cell lysis buffer containing 200 μg of RNase A (Qiagen, Valencia, CA) was added to the bacterial suspension, followed by 350 μl of MPC protein precipitation buffer. The suspension was centrifuged for 10 min at 4°C. Genomic DNA was precipitated by adding 1 ml of isopropanol and resuspended in TE buffer. DNA was stored at 4°C until analysis.

BoNT gene cluster PCR analysis.

PCR assays were designed to target internal fragments of the BoNT cluster genes ha17, ha34, ha70, orfX1, orfX2, orfX3, and p47 using primers listed in Table S1 in the supplemental material. In addition, the botR-orfX1 intergenic spacing region was amplified by using the primers A2upsrm-F and A2upsrm-R. The intergenic spacing of the orfX2-orfX3 region was amplified by using the primers ORFX2-F and ORFX3-R. PCR conditions consisted of an initial denaturing step at 94°C for 5 min, followed by 40 cycles of 94°C for 30 s and 50°C for 30 s, and an extension step at 72°C for 30 to 40 s (for PCR targeting ha17, ha34, ha70, orfX1, orfX2, or orfX3), 90 s (for PCR targeting p47 or botR-orfX1 spacing), or 3 min (for PCR targeting the orfX2-orfX3 spacing). All reactions included a final extension at 72°C for 5 min.

bont/A sequencing.

bont/A genes were sequenced in their entirety by PCR amplification of five overlapping fragments using High Fidelity Platinum Taq (Invitrogen, Carlsbad, CA). Each reaction was cleaned-up by using an UltraClean PCR clean-up kit (MoBio, Carlsbad, CA), and sequencing reactions were performed by using an Applied Biosystems 3730 DNA analyzer with the amplifying primers for each fragment and the additional sequencing primers shown in Table S1 in the supplemental material. Fragment 1 was amplified by using the primers A1Fa and A1Ra and sequenced with the primers A1Fa, A1Ra, A1Rs, and A2Fa. Fragment 2 was amplified using the primers A2Fa and A2Ra and sequenced with the primers A2Fa, A2Ra, A2Fs, and A2Rs. Fragment 3 was amplified using the primers A3Fa and A3Ra and sequenced with the primers A3Fa, A3Ra, A3Fs, and A3Rs. Fragment 4 was amplified using the primers A4Fa and A4Ra and sequenced with the primers A4Fa, A4Ra, A3Ra, and A5Fa. Fragment 5 was amplified using primers A5Fa and either A5Ra1 (for ha+/orfX− clusters) or A5Ra2 (for ha−/orfX+ clusters). This fragment was sequenced with the appropriate amplifying primers and primer A4Ra. Sequence data were assembled with CAP3 (http://pbil.univ-lyon1.fr/cap3.php), and multiple sequence alignments were generated by using MULTALIN (http://bioinfo.genopole-toulouse.prd.fr/multalin/). The GenBank accession numbers are given in Table 2.

TABLE 2.

Nucleotide polymorphisms in the bont/A1 gene

Strain	Nucleotide at positiona:					GenBank accession no.
	80	1621	3367	3425	3467
CDC51303	T	G	A	A	T	EU416225
CDC1903	-	-	-	-	-	EU416227
CDC297	-	-	-	-	-	EF028392
CDC5328	-	-	-	-	-	EU429475
CDC1882	-	A	-	-	-	EU416226
ATCC3502	C	A	G	G	G	AM412317

^aA dash indicates an identical nucleotide at the specified location compared to the C. botulinum CDC51303 strain.

PFGE.

Bacterial suspensions of C. botulinum strains were treated with 20 mg of lysozyme (Sigma)/ml and mixed with 1.2% agarose for casting into plugs for PFGE. Embedded bacteria were lysed essentially as previously described (9). Plugs were digested with either SmaI (at 25°C) or XhoI (at 37°C) for 4.5 h. Digested plugs were subjected to PFGE in a 1% agarose gel with 0.5× Tris-borate-EDTA buffer containing thiourea (4.3 mg/liter) for 19 h at 14°C. Switch times ranged from 0.5 to 40 s. with a 6-V/cm gradient.

flaA variable region sequencing.

The flaA gene was amplified by using primers modified from Paul et al. (14) and shown in the Table S1 in the supplemental material. The amplification product was cleaned up as described above, and sequencing was performed with each of the amplifying primers.

Comparative genomic hybridization.

Custom C. botulinum type A strain ATCC 3502 comparative genomic hybridization arrays were designed and synthesized by Nimblegen (Madison, WI). Coding sequence (CDS) numbering was based on the annotation data from the Sanger Centre (GenBank accession no. AM412317). The arrays consist of 384,771 unique 50- to 70-mer probes. The mean probe spacing (from start to start) is 8 bp. The arrays were cohybridized with the ATCC 3502 strain genomic DNA labeled with Cy5 random primers and each test strain genomic DNA was labeled with Cy3 random primers. Hybridizations were performed overnight at 42°C as described by the manufacturer's protocol. Arrays were washed using buffers I, II, and III supplied by Nimblegen. Finally, dried arrays were scanned at 635 and 520 nm by using a GenePix 4000B microarray scanner (Molecular Devices, Sunnyvale, CA), and TIFF images were analyzed by using Nimblescan version 2.3 (Nimblegen, Madison, WI). The data were visualized with SignalMap version 1.9 (Nimblegen) and are presented as normalized log₂ ratios of the fluorescence intensity of the reference strain/test strain.

Validation of predicted deleted regions using PCR mapping.

Microarray data were validated by assessing the presence or absence of 11 polymorphic regions by PCR targeting internal fragments of a representative CDS from such regions. Loci were chosen that were either absent in ha−/orfX+ A1 strains and present in ha−/orfX+ A2 strains or were present in ha−/orfX+ A1 strains and absent in ha−/orfX+ A2 strains. The PCR primers used to amplify the various loci are indicated in the Table S1 in the supplemental material. Similarly, PCR targeting internal fragments of CDSs flanking either upstream (CBO0791 to CBO0795, CBO0797, CBO0798, and CBO0800) or downstream (CBO0807, CBO0809, and CBO0810 to CBO0812) of the toxin gene cluster were assessed. PCR conditions consisted of an initial incubation at 94°C for 5 min, followed by 35 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min. PCR products were subjected to a final incubation at 72°C for 5 min.

Microarray data accession number.

The microarray data were deposited into the Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo) under accession number GSE10223.

RESULTS

Identification of ha−/orfX+ type A1 C. botulinum strains.

Sequencing of the C. botulinum strain CDC51303 bont/A gene indicated that it was the A1 subtype. However, comparison of the bont/A1 gene with that of the C. botulinum strain ATCC 3502 revealed five nucleotide polymorphisms (Table 2) at positions 80, 1621, 3367, 3425, and 3467. All five nucleotide changes were nonsynonymous. Four of the five nucleotide changes were located in the heavy-chain encoding region, and a single base change was located in the light-chain encoding region. None of the nucleotide changes localize to regions encoding the HEXXH zinc-endopeptidase, membrane spanning, or ganglioside-binding motifs.

PCR mapping of several of the toxin gene cluster components (ha70, ha34, ha17, orfX1-3, and p47) revealed that C. botulinum strain CDC51303 harbored the orfX1-3 and p47 genes but not ha70, ha34, and ha17 (data not shown). This organization is similar to that observed in A2 subtype strains, as well as the bont/A associated cluster of A(B) strains (3, 15). Since these clusters differ with respect to the intergenic spacing of botR-orfX1 and orfX2-orfX3 (3), we determined the spacing between these genes in C. botulinum strain CDC51303. The CDC51303 strain did not contain a 1.2-kb insertion between botR and orfX1 (as found in A2 strains) but did contain a 0.6-kb insertion between orfX2 and orfX3. Since C. botulinum strain CDC51303 contained a bont/A1 subtype gene but displayed an unusual BoNT gene organization, we designated such strains “ha−/orfX+” A1.

We reexamined several putative ha−/orfX+ A1 strains that have been reported in the literature, including C. botulinum strains CDC1882, CDC1903, CDC297, and CDC5328. As shown in Table 2, C. botulinum strains CDC1903, CDC297, and CDC5328 harbor all five nucleotide changes in bont/A1 compared to C. botulinum strain ATCC 3502, consistent with the changes observed in C. botulinum strain CDC51303. However, C. botulinum strain CDC1882 harbors only four of the five identified nucleotide changes. The nucleotide at position 1621 is consistent with that of the reference strain, C. botulinum ATCC 3502. BoNT gene cluster analysis confirmed the ha−/orfX+ organization in C. botulinum strains CDC1882, CDC1903, CDC297, and CDC5328.

PFGE analysis of ha−/orfX+ A1 strains.

Since the five ha−/orfX+ A1 strains identified in Table 1 were isolated from diverse sources and at different times, we sought to determine whether they could be distinguished by using PFGE. Surprisingly, the strains were indistinguishable when digested with SmaI (Fig. 2A). However, when a second enzyme (XhoI) was used, a single major band difference in C. botulinum strain CDC1882 was evident (Fig. 2B). In addition, differences were noted with respect to closely spaced bands of smaller mass in this strain. Interestingly, this is the only ha−/orfX+ A1 strain harboring only four of the five nucleotide changes identified in C. botulinum strain CDC51303 compared to the reference strain (ATCC 3502). The PFGE profile of both SmaI- and XhoI-digested C. botulinum strain ATCC 3502 was clearly distinguishable from the ha−/orfX+ A1 strains.

FIG. 2.

PFGE profiles of ha−/orfX+ A1 strains. (A) PFGE of C. botulinum strains digested with SmaI; (B) PFGE of C. botulinum strains digested with XhoI. The lane labeled “Reference” is the XbaI-digested Salmonella braenderup strain H9812 used as a PFGE control and reference strain. An arrow points to increase in size of a single major band in XhoI-digested C. botulinum strain CDC1882.

flaA variable region sequencing.

The nucleotide sequence of the flagellin (flaA) gene was determined for C. botulinum strains CDC51303, CDC1903, CDC1882, CDC297, and CDC5328. All five strains contained identical flaA variable region (VR) nucleotide sequences (data not shown) consistent with the flaA VR group 2 sequence reported by Paul et al. (14). By comparison, the flaA gene sequence of C. botulinum strains ATCC 3502, FRI-H1A2, and CDC21547 were consistent with the flaA VR group 1 (data not shown).

Comparative genomic hybridization of ha−/orfX+ A1 strains.

Comparative genomic hybridization microarrays were used to compare the locations of mutations (such as deletions) among the ha−/orfX+ A1 strains relative to C. botulinum strain ATCC 3502. Several large deletions were observed at multiple genomic locations among the ha−/orfX+ A1 strains. The locations of many of the deletions overlapped those of an A2 strain, C. botulinum FRI-H1A2. Nonetheless, there were several ha−/orfX+ A1 strain-specific deletions. The deleted regions among the ha−/orfX+ A1 strains were identical, indicating that it was not possible to distinguish these strains by using such microarrays. Whole-genome tiling arrays indicated that ha−/orfX+ A1 strains failed to hybridize to ca. 6% of the oligonucleotides featured on the array (data not shown).

PCR amplification of a subset of either A2 strain-specific or ha−/orfX+ A1 strain-specific genes was used to validate the putative deleted regions indicated by the microarray data (see Fig. S1 in the supplemental material and Table 3). A total of 11 polymorphic regions were selected representing 4 A2 strain-specific deletions and 7 ha−/orfX+ A1 strain-specific deletions. All 11 regions were present in the reference strain, C. botulinum ATCC 3502 (Table 3). The four putatively deleted regions in the A2 strain consisting of CBO0310, CBO2156, CBO2293, and CBO2932 failed to produce amplification products using the appropriate primers. One of these loci is a component of a ferrichrome uptake system (CBO0310 to CBO0314) all of which appears to be deleted in the A2 strain examined (FRI-H1A2). Similarly, the seven putatively deleted regions in the ha−/orfX+ A1 strains consisting of CBO0640, CBO0652, CBO0881, CBO1050, CBO2468, CBO2615, and CBO3197 also failed to produce amplification products using the appropriate primers.

TABLE 3.

Validation of a subset of predicted polymorphic regions by PCR

Region (bp)a	CDS/gene	Annotated function	PCR resultb
			ATCC 3502	FRI-H1A2	CDC51303	CDC297	CDC1882	CDC1903	CDC5328
363530-364052†	CBO0310/fhuD1	Putative ferrichrome binding protein	+	-	+	+	+	+	+
736511-737071*	CBO0640	GntR-family transcriptional regulator	+	+	-	-	-	-	-
748682-749599*	CBO0652	Putative helicase	+	+	-	-	-	-	-
984713-985577*	CBO0881/bglC	β-Glucoside-specific IIabc component	+	+	-	-	-	-	-
1152354-1153119*	CBO1050	Composite ABC transporter	+	+	-	-	-	-	-
2300063-2300564†	CBO2156	Putative membrane protein	+	-	+	+	+	+	+
2449311-2449822†	CBO2293	Short-chain dehydrogenase	+	-	+	+	+	+	+
2601914-2602630*	CBO2468	Methyl-accepting chemotaxis protein	+	+	-	-	-	-	-
2763745-2764337*	CBO2615	Rhodanese-like protein	+	+	-	-	-	-	-
3121483-3121982†	CBO2932	Putative exported protein	+	-	+	+	+	+	+
3370351-3370902*	CBO3167	Putative ABC transporter permease	+	+	-	-	-	-	-

^aThe nucleotide location based on the genome sequence of ATCC 3502. *, region predicted to be present in C. botulinum strain FRI-H1A2 and absent in C. botulinum ha−/orfX+ A1 strains by microarray analysis; †, region predicted to be absent in C. botulinum strain FRI-H1A2 and present in C. botulinum ha−/orfX+ A1 strains by microarray analysis.

^bThe presence (+) or absence (-) of a band of appropriate size is indicated using PCR primers listed in Table S1 in the supplemental material under “putative polymorphic region validation.”

Analysis of the BoNT gene clusters associated with ha−/orfX+ A1 and A2 strains.

As expected, ha−/orfX+ A1 and A2 strains demonstrated poor hybridization to oligonucleotide probes representing the ha genes, consistent with initial PCR analysis of these strains. Probes to the gene immediately downstream of bont/A, which is annotated in the C. botulinum ATCC 3502 genome sequence as a transposase (CBO0807), also failed to hybridize with either the ha−/orfX+ A1 or A2 strains. This finding was predicted, as it has been previously reported (3, 17) that the gene downstream of bont/A in A2 strains (such as Kyoto-F) is lycA, which shows some homology to a phage-encoded protein. Array oligonucleotides corresponding to several genes flanking the toxin gene cluster in C. botulinum strain ATCC 3502 failed to hybridize to either the ha−/orfX+ A1 or A2 strains (Fig. 3). Five of the putatively absent upstream genes (CBO0793 to CBO0797) are annotated as hypothetical proteins.

FIG. 3.

Analysis of toxin gene cluster content among C. botulinum strains. The region representing the toxin gene cluster and flanking genes is shown. The normalized log₂ ratio of the fluorescence of the reference strain (C. botulinum strain ATCC 3502)/test strain is shown. The top track indicates the CDS location in the C. botulinum ATCC 3502 strain. Track A, C. botulinum strain CDC21547; track B, C. botulinum strain FRI-H1A2; track C, C. botulinum strain CDC51303; track D, C. botulinum strain CDC1882; track E, C. botulinum strain CDC1903; track F, C. botulinum strain CDC297; track G, C. botulinum strain CDC5328.

In order to confirm the absence of several CDS flanking the toxin gene cluster in ha−/orfX+ A1 and A2 strains, we performed PCR using primers targeting internal fragments of each of the CDS. As shown in Table 4, all flanking CDS were present in C. botulinum ATCC 3502. However, only CBO0791, CBO0792, and CBO0812 were amplified among the ha−/orfX+ A1 and A2 strains. CDSs that were predicted to be absent among the ha−/orfX+ A1 and A2 strains also failed to produce amplification products. More specifically, no PCR products were observed for the following CDSs located upstream of the ha genes in ha−/orfX+ A1 and A2 strains: CBO0793, CBO0794, CBO0795, CBO0797, CBO0798, and CBO0800. Similarly, CBO0807, CBO0809, CBO0810, and CBO0811 located immediately downstream of the bont/A gene were absent in ha−/orfX+ strains. Interestingly, the ha+/orfX− A1 toxin gene cluster (in strain ATCC 3502) is flanked by defective transposases including CBO0800 and CBO0807 to CBO0810, none of which are present in ha−/orfX+ A1 or A2 strains.

TABLE 4.

PCR analysis of regions flanking the toxin gene cluster

Region (bp)a	CDS/gene	Annotated function	PCR resultb
			ATCC 3502	FRI-H1A2	CDC51303	CDC297	CDC1882	CDC1903	CDC5328
893897-894538	CBO0791/dapB	Dihydrodipicolinate reductase	+	+	+	+	+	+	+
894874-895487	CBO0792/sseA	3-Mercaptopyruvate sulfurtransferase	+	+	+	+	+	+	+
896165-896751	CBO0793	Hypothetical protein	+	-	-	-	-	-	-
897235-897781	CBO0794	Hypothetical protein	+	-	-	-	-	-	-
898083-898255	CBO0795	Hypothetical protein	+	-	-	-	-	-	-
899062-899162	CBO0797	Hypothetical protein	+	-	-	-	-	-	-
899608-900116	CBO0798	Flagellin	+	-	-	-	-	-	-
901484-901785	CBO0800	Putative transposase (partial)	+	-	-	-	-	-	-
913897-914398	CBO0807	Putative transposase (pseudogene)	+	-	-	-	-	-	-
914528-914683	CBO0809	Putative transposase (pseudogene)	+	-	-	-	-	-	-
914982-915551	CBO0810	Putative transposase	+	-	-	-	-	-	-
916111-916781	CBO0811	Putative membrane protein	+	-	-	-	-	-	-
917964-918497	CBO0812	Putative aminohydrolase	+	+	+	+	+	+	+

^aThe nucleotide location based on the genome sequence of ATCC 3502.

^bThe presence (+) or absence (-) of a band of appropriate size is indicated using PCR primers listed in Table S1 in the supplemental material under “toxin gene cluster flanking region mapping.”

DISCUSSION

In this study, we examined the genetic characteristics of a group of C. botulinum A1 strains that contained an unusual toxin gene cluster and variant bont/A1 nucleotide sequence. Previous reports have either identified A1 strains with a unique toxin gene cluster organization (4) or the variant bont/A1 gene sequence (6). In the present study, a total of five such strains were characterized and compared to each other by using PFGE and comparative genomic hybridization microarrays. These unique type A1 strains may be identified by the presence of several bont/A1 nucleotide polymorphisms which could serve as a genetic “signature” for strains in this group.

This group lacks the ha genes usually associated with A1 strains but rather contains the orfX1-3 and p47 genes associated with A2 gene clusters. Therefore, we designated this group as ha−/orfX+ A1 to distinguish them from strains containing the archetypical A1 toxin gene cluster (such as in strain ATCC 3502), which would be designated ha+/orfX− A1. The BoNT gene cluster of ha−/orfX+ A1 strains is most similar in organization to that of the cluster associated with the bont/A gene among type A strains with an unexpressed bont/B [i.e., A(B) strains] (7, 15). Unlike A(B) strains, only bont/A was identified in the ha−/orfX+ A1 strains examined. The ha−/orfX+ A1 strains also differ with respect to A(B) strains in that none of the ha genes (which are associated with the bont/B cluster) were detected by PCR and were absent as shown by comparative genomic hybridization microarrays.

In addition to the unusual BoNT gene cluster, the ha−/orfX+ A1 strains contain five nucleotide polymorphisms in the bont/A1 gene compared to the same gene from C. botulinum ATCC 3502, with the exception of C. botulinum strain CDC1882, which contains only four of these polymorphisms. All of the polymorphisms are nonsynonymous; however, the effects of the amino acid sequence changes on the toxin have not been investigated. Since the polymorphisms do not localize to protein domains with known function, it is not possible to make predictions regarding the effects these changes have on gene function.

The ha−/orfX+ A1 strains investigated in the present study have been isolated from distinct sources and geographical locations (Table 1). One strain, CDC297, was isolated nearly 40 years ago. Three of the strains (CDC51303, CDC1903, and CDC297) were associated with food-borne outbreaks of botulism, while one of the strains (CDC1882) was isolated from an infant botulism case.

Although there is no epidemiological association among the ha−/orfX+ A1 strains, various analyses (PFGE, flaA variable region sequencing, and comparative genomic hybridization) indicated that the strains were genetically indistinguishable. SmaI-digested PFGE profiles were identical among all of the ha−/orfX+ A1 strains. When the enzyme XhoI was used for PFGE analysis, a shift in the size of at least one band was observed in strain CDC1882. Whether this strain contains additional strain-specific DNA sequences is unknown and will require additional study using techniques such as genomic subtractive hybridization. Interestingly, we determined that this strain harbored only four of the five nucleotide polymorphisms in the bont/A gene associated with ha−orfX+ A1 strains.

Recently, Paul et al. (14) reported the use of nucleotide sequence polymorphisms in the variable region of the flagellin gene, flaA, among group I and II C. botulinum strains for strain identification. A greater number of flaA alleles were found among group I C. botulinum strains than group II. We also applied flaA variable region sequencing to determine whether sequence variations existed among the ha−/orfX+ A1 strains. Not surprisingly, analysis of the flaA gene demonstrated 100% sequence identity among the ha−/orfX+ A1 strains which clustered with the flaA VR group 2 (14). It is unlikely that the flaA VR group 2 is specific for ha−/orfX+ A1 strains since Paul et al. (14) reported both type A and type B strains in this group.

Since the ha−/orfX+ A1 strains appeared to be clonally related and harbored both an unusual bont/A nucleotide sequence and toxin gene cluster organization, we utilized comparative genomic hybridization microarrays to determine whether the strains contained strain-specific polymorphisms. For the present study, the array features were based on the genome sequence of C. botulinum strain ATCC 3502. The ATCC 3502 strain used as a reference in our hybridization experiments displayed some minor changes compared to the ATCC 3502 genome sequence, including the loss of the ∼16.3-kb pBOT3502 plasmid (not featured on the CGH arrays). The reference strain displayed poor hybridization to probes associated with two prophage elements located at approximately 1.82 to 1.86 Mbp and 2.46 to 2.52 Mbp. Nonetheless, this strain hybridized to the overwhelming majority of the features on the array providing valuable data for comparison of the test strains.

In our analysis, the genomic location of each probe was plotted to the normalized log₂ ratio of the fluorescence of the reference strain compared to the fluorescence of the test strain. Therefore, high log₂ ratios over a large region indicated significant deletions in the genome of the test strains compared to the reference strain. The majority of the deletions in the ha−/orfX+ A1 strains were shared with the A2 strain (FRI-H1A2) examined. These findings indicate that the genome structure of such strains may differ compared to the reference strain.

All of the ha−/orfX+ A1 and A2 strains examined demonstrated poor hybridization to probes representing the 5′ portion of the ntnh/A1 gene. These data are consistent with a predicted recombination site occurring in the middle of the ntnh gene from A1 strains (17) in which the 5′ end of ntnh/A1 shows significant divergence compared to ntnh from strains A2, A3, and A4.

Although comparative genomic hybridization analysis indicated that there were deletions specific to the ha−/orfX+ A1 strains compared to the A2 strain, we were unable to identify deletions that allowed differentiation among the ha−/orf+ A1 strains. This finding is not unexpected since the PFGE profiles of the strains were nearly indistinguishable and large strain-specific deletions would likely affect PFGE profiles. However, one limitation of CGH analysis is that a strain may contain DNA sequences that are not featured on the microarray and therefore the content of such genes cannot be assessed.

The bont/A associated toxin gene cluster of A1 subtypes includes bont/A, ntnh, botR, and the ha genes, while the clusters of the A2-A4 subtypes include bont/A, ntnh, p47, botR, and orfX1-3 genes (17). Recently, genome sequencing of the A2, A3, and A4 strains demonstrated the presence of a lycA gene immediately downstream of the bont/A2 gene and immediately upstream of the bont/A3 and bont/A4 genes (17). An arsC gene that encodes an arsenate reductase is located both upstream and downstream of the bont/A2 toxin gene cluster (immediately upstream of the orfX3 gene and immediately downstream of the lycA gene) (3, 17). Such differences in the flanking regions of the ha−/orfX+ toxin gene clusters of subtypes A2-A4 compared to the ATCC 3502 strain (subtype A1) may also account for the differences observed in the flanking regions of ha−/orfX+ A1 strains examined in the present study.

We found that several genes flanking the BoNT gene cluster of the ATCC 3502 strain were absent among ha−/orfX+ A1 and A2 strains. These genes included several hypothetical and defective transposase genes located both upstream and downstream of the BoNT gene cluster. Therefore, we propose that some or all of the ha−/orfX+ A1 toxin gene cluster may constitute a genomic island and that the region is significantly larger than previously recognized (∼22 kb). These findings suggest that the genomic structure of ha−/orfX+ A1 and A2 strains are likely to be different from that of the ATCC 3502 ha+/orfX− A1 genome. Such differences may be due in part to potential genetic mobility of the toxin gene cluster.

Analysis of the toxin gene cluster sequence of ha−/orfX+ A1 strains demonstrated that the orfX1-3, botR, and ntnh genes were most similar to that of the bont/A associated cluster in the A(B) strain, NCTC 2916 (7). Given such similarity in the toxin gene cluster sequences of these strains, it is possible that A(B) strains could have arisen by two separate recombination events involving the type A and type B toxin gene clusters. From our data, it is not possible to determine whether ha−/orfX+ A1 strains arose prior to the A(B) strains or resulted from a loss of the unexpressed type B gene cluster. In any event, the recombination event resulting in ha−/orfX+ A1 strains likely occurred in a strain related to the A4 subtype since MLVA indicated a clustering of strain CDC297 with the A4 subtype strain (10). In addition, Jacobson et al. (7) reported that multilocus sequence typing analysis of strain CDC5328 and the A4 subtype strain showed that they differ by only one of seven alleles sequenced.

The type A and B toxin gene clusters of the bivalent (Ba) A4 strain are located on a plasmid (11, 17). Although strains CDC297 and CDC5328 clustered closely with the A4 strain (7, 10), Marshall et al. (11) demonstrated that the bont/A gene was located on the chromosome in at least one of the ha−/orfX+ A1 strains (CDC5328).

There is evidence of genetic divergence among ha−/orfX+ A1 strains since strain CDC1882 contains only four bont/A nucleotide changes, while the remaining ha−/orfX+ A1 strains contain five bont/A nucleotide changes. The minor difference in the PFGE profile of strain CDC1882 compared to the other ha−/orfX+ A1 strain further supports the notion of genetic divergence within the ha−/orfX+ A1 strains. It is unknown whether strain CDC1882 is ancestrally related to the remaining ha−/orfX+ A1 strains or whether it diverged from them.

The high degree of genetic similarity among the ha−/orfX+ A1 strains of diverse origins was unexpected since previous studies have found that type A strains are genetically heterogeneous. For example, Nevas et al. (13) observed 11 SacII macrorestriction fragment PFGE profiles, several of which consisted of only a single strain each, among 28 type A strains. Using a MLVA approach, Macdonald et al. (10) distinguished 53 type A1 and A(B) strains into 30 types. Even among the 37 type A1 strains analyzed, there were 17 genotypes.

In addition, the variant bont/A1 sequence of ha−/orfX+ A1 strains is noteworthy in that the bont/A1 allele is considered to be highly conserved among C. botulinum type A strains (6). Of 54 type A1 strains examined, three bont/A1 alleles were identified consisting of the allele typically associated with ha+/orfX− A1 strains (e.g., strain ATCC 3502), the allele associated with A(B) strains varying by two nucleotides (compared to ATCC 3502), and the allele associated with ha−/orfX+ A1 strains.

The ha−/orfX+ A1 strains examined in the present study were isolated from diverse locations throughout the United States over several decades, suggesting that such strains are likely to be widely distributed in the environment. Our findings indicate that these strains are clonally related which may impact the ability to distinguish isolates associated with an outbreak of botulism from concurrent sporadic cases. Therefore, higher-resolution methods may be required to distinguish unrelated ha−/orfX+ A1 strains. Potentially, single nucleotide polymorphisms among ha−/orfX+ A1 strains could be used to further subtype these strains.

Genomic sequencing of ha−/orfX+ A1 strains may provide distinguishing single-nucleotide polymorphisms for these strains. Comparison of the genome sequence of ha−/orfX+ A1 strains to the genomes of other type A subtypes or A(B) strains may also yield insights into genome plasticity and models for the acquisition of different BoNT gene clusters. Finally, comparison of ha−/orfX+ A1 strains with ha+/orfX− A1 strains provides an opportunity to understand how differences in the toxin gene cluster affect botulinum subtype A1 toxin production and/or potency.