Identification of Bacillus anthracis specific chromosomal sequences by suppressive subtractive hybridization

Background Bacillus anthracis, Bacillus thuringiensis and Bacillus cereus are closely related members of the B. cereus-group of bacilli. Suppressive subtractive hybridization (SSH) was used to identify specific chromosomal sequences unique to B. anthracis. Results Two SSH libraries were generated. Genomic DNA from plasmid-cured B. anthracis was used as the tester DNA in both libraries, while genomic DNA from either B. cereus or B. thuringiensis served as the driver DNA. Progressive screening of the libraries by colony filter and Southern blot analyses identified 29 different clones that were specific for the B. anthracis chromosome relative not only to the respective driver DNAs, but also to seven other different strains of B. cereus and B. thuringiensis included in the process. The nucleotide sequences of the clones were compared with those found in genomic databases, revealing that over half of the clones were located into 2 regions on the B. anthracis chromosome. Conclusions Genes encoding potential cell wall synthesis proteins dominated one region, while bacteriophage-related sequences dominated the other region. The latter supports the hypothesis that acquisition of these bacteriophage sequences occurred during or after speciation of B. anthracis relative to B. cereus and B. thuringiensis. This study provides insight into the chromosomal differences between B. anthracis and its closest phylogenetic relatives.


Background
Bacillus anthracis, the etiological agent of anthrax, is responsible for a serious and often fatal disease of mammalian livestock and humans [8,22]. Its spore-forming capability and highly pathogenic nature have made it one of the most effective bioterrorism agents. Pathogenic strains of B. anthracis contain two plasmids that confer virulence; the toxin-encoding 181.6-kb pXO1 [21,27,29] and the capsule-encoding 94.8-kb pXO2 [12,27,42]. Loss of either plasmid attenuates virulence. Environmental isolates of virulent B. anthracis have been found to contain some pXO2cells indicating that spontaneous loss of this plasmid may occur outside the selective pressures of the host [11,41].
Bacillus cereus is commonly isolated from soil and has been implicated in food-related diseases [9]. Bacillus thuringiensis contains plasmids that encode an array of gene products including the insecticidal crystal proteins [39]. B. anthracis, B. cereus and B. thuringiensis are closely-related members of what has been referred to as the Bacillus cereus-group. Comparison of their 16S rRNA sequences place them within a range expected for members of the same species [3]. In addition, DNA-DNA hybridization [19,37] and pulsed field gel electrophoresis [14] studies have revealed great homology between their chromosomal DNAs. Recent data obtained from multilocus enzyme electrophoresis (MEE) and by multiple-locus sequence typing suggested that B. anthracis, B. cereus, and B. thuringiensis belong to one and the same species, and that salient distinguishing of each member is due to plasmid-borne genes [15].
Suppressive subtractive hybridization (SSH) is a PCRbased DNA subtraction method that enables the identification of genomic sequence differences between various strains or species of bacteria. Tester DNA is defined as that in which the differences are being sought relative to a driver DNA. For example, SSH has been successfully used to identify unique genomic sequences for specific strains of Helicobacter pylori [2], uropathogenic E. coli [18,43], and colonizing Campylobacter jejuni [1]. This technique has also been used to detect unique genomic sequences found in E. coli but not in Salmonella typhimurium [4], in pathogenic Burkholderia pseudomallei but not in nonpathogenic B. thailandensis [36], in pathogenic Burkholderia mallei but not in B. thailandensis [7] and in B. anthracis but not in B. cereus 14579T, B. thuringiensis Al Hakam, or B. cereus 3A [32]. This methodology has not only identified genetic differences among different strains or species of bacteria but also defined genes that contribute to the virulence of an organism. These virulence-related genes are often localized on the chromosome. Such clusters have been referred to as pathogenicity islands [13].
Herein we report the application of SSH to elucidate genomic differences of B. anthracis in comparison to B. cereus and B. thuringiensis, thereby identifying DNA sequences that are unique to the B. anthracis chromosome. In contrast with B. anthracis described as one of the most monomorphic species within the B. cereus group, B. cereus and B. thuringiensis present a vast range of genetic diversity. For this study we used the genomic DNA of B. cereus 14579T or B. thuringiensis 10792T as driver DNA because these are representive of the respective species. In the subsequent screening of the B. anthracis-specific chromosomal sequences, we used several Bacillus sp. strains more closely-related to B. anthracis [15,31,33]. The B. anthracis-specific chromosomal sequences found elucidate genomic differences of these three closely-related bacteria. These sequences can be exploited for the development of more stable and robust chromosome-specific markers important for accurate identification of B. anthracis and unique from those previously reported [30][31][32]. Forty four hundred Ba/Bc and Ba/Bt clones were prescreened for B. anthracis-specific sequences by comparing duplicate sets of colony filters. One set was probed with 32 P-labeled tester sequences, while the other set was probed with the corresponding 32 P-labeled driver sequences. Two hundred and twenty-four Ba/Bc and 189 Ba/Bt clones displayed more intense hybridization when probed with B. anthracis genomic DNA than with that of B. cereus or B. thuringiensis. Recombinant plasmid DNA of the 413 SSH clones selected was digested with RsaI, electrophoresed on agarose gels, Southern blotted, and then probed with the corresponding 32 P-labeled driver genomic DNA. Insert DNA sizes ranged between 100-2000 bp, as predicted from analysis of the frequency with which RsaI digests each of these Bacillus genomic DNAs. Ninety-six Ba/Bc and 59 Ba/Bt clones had inserts that were undetected by the driver DNA probe.

Screening of
The insert DNA from each of these 155 candidate testerspecific clones was isolated, 32 P-labeled and used to probe DNA blots containing the HindIII-digested genomic DNAs of (1) B. anthracis strain 7700, (2) B. thuringiensis strain 97-27, (3) B. cereus strain 14579T and (4) B. thuringiensis strain 10792T. B. thuringiensis strain 97-27 genomic DNA was used because it is extremely similar to that of B. anthracis [15,16,28,32,33]. Thirty-two of the Ba/ Bc clones and 20 Ba/Bt clones detected sequences of B. anthracis 7700 DNA only (data not shown). were probed with the 32 P-labeled insert DNA of each of the 52 clones found to be B. anthracis-specific after initial genomic DNA blot screening. An example of this more extensive analysis is provided in Figure 1, demonstrating detection of sequence homology in the four B. anthracis genomic DNA samples only when probed with the insert of Ba/Bt clone E4. Clone E4 did not detect any homology in any of the B. cereus, B. thuringiensis, or other Bacillus species included on this blot.
A summary of the 52 clones tested in this manner is presented in Table 2 (see Additional file 1). Thirty-nine of the 52 clones were observed to be completely specific for B. anthracis (highlighted in bold type). All probes but one consistently hybridized and detected a similar pattern of restriction fragments found in all four B. anthracis genomic DNA samples which included pathogenic B. anthracis strains RA3 and 7611. Clone 140 was the exception in that it detected some polymorphism in the restriction fragment pattern displayed in the B. anthracis genomic DNA samples (data not shown). Thirteen SSH clones did detect nucleotide sequences in strains other than B. anthracis. Clones 14, 34, 140, O4, Z6, L7, and O7 hybridized with one other; clones 151, 227, G and R3 hybridized with two other; and clones A4 and K4 hybridized with three other B. cereus and / or B. thuringiensis genomic DNAs on the survey blot, exclusive of the driver strains.

Sequence analysis of B. anthracis-specific clones
Sequence determination and chromosomal location of B. anthracisspecific clones The nucleotide sequence was determined for each insert of the 39 clones that were found to be B. anthracis-specific. Seventeen of the 39 clones demonstrated double, triple and as much as quadruple redundancy, finally sorting into 7 different RsaI-fragments (Table 2 and 3, see Additional file 1 and 2). Therefore, these 39 clones delineate 29 different RsaI fragments of the B. anthracis 7700 chromosome. BLAST analysis [24] sorted these 29 RsaI fragments into 6 different regions on the A2012 strain [34] chromosome ( Figure 2). Areas on the chromosome delineated either by 13 or more clones, 3 clones, or one clone were designated as regions, clusters, or orphans, respectively. Regions I and II encompassed eight and thirteen different B. anthracis-specific RsaI fragments that were identified by thirteen and sixteen clones, respectively. Three other locations were found to encompass two or three different B. anthracis-specific RsaI fragments. Clones 81/231 and F7 define Cluster I; clones 230, 84 and 135 define Cluster II; and clones 9/49 and 221 define Cluster III. Single B. anthracis-specific orphan clone M7 identified the remaining chromosomal area.

Annotation of Region I and II delineated by the B. anthracis-specific clones
Open reading frames (ORFs) overlapping and/or bordering B. anthracis-specific sequences were identified [26] and the amino acid sequences derived from these ORFs were compared to other protein sequences [25]. The results of these analyses are summarized in Table 3 (see  Additional file 2). Region I, delineated by 8 B. anthracis-specific RsaI-fragments and consisting of an area of 14.5 kb, contains 10 potential ORFs. Nine of these ORFs demonstrated  significant sequence similarity with genes encoding enzymes and proteins involved in cell wall biosynthesis or export of polysaccharides in other bacterial species. The G + C content of the DNA encompassing the SSH-selected sequences in this region was 32.1% which is not significantly different from that of the entire B. anthracis chromosome, reported to be 35.4% [35].
Region II, which encompassed an area of 24.5 kb flanked by clones H4 and 150b, contains 40 potential ORFs. Nineteen of these ORFs had sequence homology with a variety of bacteriophage-related gene products.  Cluster I contains a bacteriophage-related sequence that was detected by B. anthracis-specific clones 81/231, while clone F7 sequences overlapped with 3 potential ORFs encoding hypothetical proteins found to have no homology with any known proteins. These clones are contained within the area designated " B. anthracis phage lambda Ba02" of the B. anthracis Ames strain genome [35].    Table 2). The same set of strains was used to evaluate a B. anthracis chromosome-specific PCR assay that exploited polymorphisms found in the rpoB Gel electrophoresis analysis of PCR products generated after reaction of various Bacillus genomic DNAs with clone E4 derived primers. Figure 3 Gel electrophoresis analysis of PCR products generated after reaction of various    Figure 3. Region I contains ORFs that had significant similarity with genes encoding enzymes and proteins involved in cell wall polysaccharide biosynthesis. Typically in this process, phosphorylated sugar molecules are first activated by conversion into nucleotide sugar derivatives. Enzymes, such as epimerases and dehydrogenases, may then modify these nucleotide sugar precursors into the distinctive sugar subunit(s) characteristic of the cell wall of a particular bacterial species. Transferases may then catalyze the sequential transfers of the sugar residues from their nucleotide carriers to a membrane-bound acceptor, usually a lipid carrier, forming the characteristic carbohydrate repeating unit. This first stage of cell wall polysaccharide biosynthesis usually occurs in the cytoplasm at the cytoplasmic side of the plasma membrane. At this point, an ABC transporter cassette may permit the lipidlinked carbohydrate-repeating unit to pass through the membrane. This two-component system consists of a translocation permease protein with membrane-spanning domains and an associated ATP-binding protein that energizes the permease.

Discussion
In Region I potential ORFs BA-0356/7/8 and BA_0361/2, demonstrating sequence homology to genes encoding an epimerase and a dehydrogenase, respectively, were identified by SSH clones 190/N, Y4 and 142/I6. Potential ORFs BA_0365/6 and BA_0371, detected by SSH clones 55/67/ 212/235, 66 and P3, demonstrated sequence similarity to genes encoding different types of transferases. Clone 190/ N and Y4 inserts also overlapped with potential ORFs BA_0359 and BA_0360, which displayed significant sequence similarity to other known ABC transporter cassette proteins. The best match was with B. subtilis genes tagG and tagH, involved in the translocation of cell wall component teichoic acid. However, the corresponding B. anthracis potential ORFs cannot encode equivalent functional homologs. B. subtilis cell wall-associated teichoic (TA) and teichuronic acid (TUA) are absent in B. anthracis [23]. The remaining potential ORFs (BA_0363/4 and BA_0368/9 and BA_0370 detected by clones 201, 55/67/ 212/235, O3, G and P3) in this region were found to share homology with proteins involved in cell wall synthesis whose specific functions are currently unknown. These same ORFs were also found to be specific for B. anthracis A2012 by whole genome comparison with B. cereus 14579T [17].
The Gal-NAG polysaccharide, derived from sugar components D-galactose and N-acetyl-D-glucosamine, has been demonstrated by monoclonal antibodies to be uniquely associated with the cell wall of B. anthracis and a few B. cereus strains [10]. This result was also obtained for B. anthracis pX01-and pX02-strains, indicating that the enzymes and proteins necessary for synthesis and export of this B. anthracis-specific polysaccharide are located on the chromosome. It is possible that the Gal-NAG polysaccharide, or other components associated uniquely with the cell wall of B. anthracis, are synthesized by enzymes and exported by proteins encoded by genes specific to B. anthracis. Such specific genes might be detected by SSH as the genes associated with Region I have been. Substantiation of this hypothesis will require functional analysis of the Region I genes. It is interesting to note though that genes encoding cell wall biosynthesis proteins have been selected in many other subtractive hybridization experiments [1,4,5,7,18,36]. In conclusion, the present study revealed differences between the chromosome of B. anthracis and those of eight strains of B. cereus and B. thuringiensis. The general differences were seen in genes that may encode proteins related to cell wall synthesis and bacteriophages as well as ORFs encoding proteins of unknown function.

Bacterial strains and genomic DNA preparation
The B. anthracis, B. cereus and B. thuringiensis strains used in this study are listed in Table 2. The bacterial genomic DNA was extracted by a method described elsewhere [6].

Plasmid DNA preparation and insert fragment isolation
Plasmid DNA was isolated from 1-5 ml LB or TB/kanamycin cultures by either the boiling miniprep method [38] or the Plasmid Mini Kit (Qiagen, Valencia, CA). Insert fragments to be used as probes for genomic DNA blots were separated from vector sequences on 0.7% agarose gels and extracted by electroelution [38].

DNA blot analysis
One microgram of DNA was digested with the appropriate restriction enzyme (Life Technologies, Rockville, Md.), subjected to electrophoresis in 0.7% (w/v) agarose gels and transferred to Genescreen Plus nylon membranes (New England Nuclear Life Science Products, Inc., Boston, MA) as per manufacturer's instructions. DNA blots were prehybridized and hybridized at 65°C in 330 mM sodium phosphate, pH 7, 10 mM EDTA, 5% (w/v) SDS, 10% (w/ v) dextran sulfate and salmon sperm DNA at 25 µg/ml. Filters were washed twice for 30 min. intervals at 65°C in 2 × SET, 0.5% SDS and autoradiographed.

Sequencing and bioinformatics
Each clone was sequenced in duplicate with the M13 forward and reverse primers using an ABI377 automated sequencer (Applied Biosystems, Boston, MA) and the BigDye terminator ready reaction kit (Perkin-Elmer Applied Biosystems, Foster City, CA). The nucleotide sequences were edited and assembled with the Sequencing Analysis 3.0 and AutoAssembler 3.1.2 programs. The SSH DNA sequences were subsequently localized on the A2012 strain chromosome [24] and annotated [25,26].