The discovery that CDC 684 was not a B. megaterium strain but was rather B. anthracis, based on shared phenotypic features, prompted the use of the guinea pig model to determine its virulence. In a pilot experiment, groups of four guinea pigs injected i.m. with CDC 684 spores at doses of 114, 1,145, and 11,450 cfu/mL survived. These groups were then injected four days later with 1.29 × 10⁵, 1.29 × 10⁶ and 1.29 × 10⁷ cfu/mL, respectively, and again all survived. By comparison these identical spore preparation and treatment conditions produced LD₅₀ values for the virulent Ames and Vollum-1B strains of 175 and 306 spores respectively in the guinea pig model 16 17 .

This lack of lethality indicated that CDC 684 is significantly attenuated. In a second experiment to confirm attenuation, 10 guinea pigs injected i.m. with 1 × 10⁸ cfu/mL CDC 684 spores all survived. These results confirm that CDC 684 is highly attenuated with an LD₅₀ of >1 × 10⁸ spores in the guinea pig model.

The CDC 684 genome has been recently sequenced and assembled to closure at Los Alamos National Laboratory/J. Craig Venter Institute and is available on the NCBI Genome database [GenBank: CP001215.1]. The chromosome is 5,230,115 bp, pXO1 [GenBank: CP001216] is 181,773 bp and pXO2 [GenBank: CP001214] is 94,875 bp.

The use of comparative WGS defined an extremely conserved and accurate phylogenetic SNP tree for B. anthracis based on the analysis of 1,000 SNPs in 26 diverse isolates 18 . This analysis resulted in the hypothesis that only a few selected SNPs at key positions along five branches were needed to accurately place all B. anthracis isolates into one of 12 sub-clades. This notion was shown to be accurate when 13 canSNPs were subsequently used to accurately place more than 1,000 B. anthracis isolates into one of these 12 sub-clades 19 . In silico canSNP typing showed that CDC 684 falls along the lineage created by B. anthracis Vollum (A0488; [GenBank: ABJC00000000]). This sequenced Vollum strain is presumed to be a close relative of the British isolate that was tested as a biological weapon on Gruinard Island, Scotland, in the 1940s 20 .

The close phylogenetic relationship between CDC 684 and Vollum demonstrates that CDC 684 belongs to a highly virulent B. anthracis lineage. We were therefore interested in further determining the degree of relatedness between Vollum and CDC 684, given the marked differences in virulence between these two strains. An initial comparative in silico analysis of Ames Ancestor [GenBank: AE017334], CDC 684 and Vollum WGS uncovered ~ 390 SNP differences distinct from Ames Ancestor but common (i.e., derived) in both the CDC 684 and Vollum genomes. These results are consistent with other whole genome SNP comparisons of 128 B. anthracis isolates that suggest that the SNP genetic distance between Ames and Vollum is approximately 400 SNPs [Pearson, Schupp, Ravel and Keim, unpublished data].

Preliminary analysis of 30 SNPs that phylogenetically reside along a terminal position on the Vollum branch indicated that there were at least 10 new nodes along this branch, of which >100 Vollum-like isolates reside [Chung, Pearson and Keim, unpublished data]. In silico analysis of 10 new canSNPs along this branch indicated that CDC 684 was not in the terminal node created by the sequenced Vollum strain, but rather was located in a node midway between the sequenced strain and a branch point defined by the initial analysis of 100 Vollum-like strains, Figure 1.

This analysis also demonstrated that CDC 684 possessed 51 SNPs that appeared to be unique to this isolate. There were 15 isolates that shared the Vollum branch node with CDC 684. These isolates were predominantly recovered by the Centers for Disease Control during the 1950s and 1960s. While the incidence of lethal anthrax infections in the United States had been greatly reduced during the 20^th century 21 , it can be assumed that the majority of the CDC isolates labeled as B. anthracis would have come from sources containing virulent strains such as imported hides and/or animal deaths 22 .

Table 1 lists 27 non-synonymous chromosomal SNPs from 51 total that are unique to CDC 684 in comparison to the Vollum (A0488) strain. There are no obvious B. cereus or B. anthracis virulence factors on this list but the role for each of these proteins in CDC 684 may also be compromised by the large inversion event. It also needs to be reiterated that while these SNPs are unique in their relationship to the Vollum strain their status in 15 other un-sequenced isolates who shared the node along the Vollum branch are still undetermined. It is likely that most of these SNPs will be shared (i.e., no differences) with these 15 presumably virulent B. anthracis isolates.

The simplest explanation for the attenuated phenotype for CDC684 would be the mutation of one or more of the virulence factors encoded on the pXO1 or pXO2 plasmids that altered expression or function. These virulence factors include the toxin gene complex on pXO1 (comprising genes encoding for protective antigen, edema factor, and lethal factor), the poly-D-glutamyl capsule gene complex on pXO2 (encoded by capA, capB, capC and acpA), and trans-acting transcription regulators on both plasmids 23 . However, in silico comparison of the completed sequences of the pXO1 and pXO2 plasmids from the CDC 684 strain to those of the Ames Ancestor and Vollum strains showed that all of the known virulence factors were intact. There was a single non-synonymous SNP found in pXO1 GBAA_pXO1_0019, a large gene of unknown function. Collectively we observed no putative functional differences in the plasmid-encoded virulence factors between CDC 684 and its closest relative, Vollum, which is a fully virulent strain 24 .

The most striking feature of CDC 684 genome is a massive inversion that reverses the orientation of 3.3 Mbp of the chromosome relative to the replication origin. The dimensions of the inversion have been graphically illustrated in a recent review of Bacillus anthracis genome variation 25 . This earlier report used Artemis software http://www.sanger.ac.uk/resources/software/artemis/ to illustrate the alignment and conserved gene order of four finished and closed genomes (B. anthracis Ames, B. anthracis Australia 94, B. anthracis CDC 684, and B. thuringiensis Al Hakam). While the fine-scale gene order in CDC 684 is precisely maintained as in the Ames chromosome, the large rearrangement has caused an inversion of a 3.3 Mbp region between the basepair coordinates 454 Kbp and 3,783 Kbp in the Ames Ancestor chromosome (see Figure 2).

The inversion appears to have been caused by an internal recombination event between homologous regions within two lysogenic lambda-like prophages (LambdaBa04 and LambdaBa02), which are found in all B. anthracis genomes 26 27 . The inversion can best be visualized at the molecular level by examining the orientation of the att (attachment) sites that flank the ends of these phages (Figure 2). Lysogenic bacteriophages possess cohesive ends (att), usually 12-13 bp repeats, which serve as both excision points and "sticky ends" that enable the phage to cirularize as it enters a lytic life cycle 28 . At first glance it seemed likely that the inversion may involve the att sites in these Lambda like prophages and that the exchange may have involved a site-specific recombination. But the two att sites were unique to each other, i.e., Lambda Ba04 and Ba02 contain distinct att sites (Figure 2B) that allow them to be distinguished from each other (Ba04, ATACAGCTCATGT and Ba02, TTTT(C/T)TTTACAC). In Ames Ancestor, pairs of these two distinct att sites define both the size (Ba04 = 37.3 kb; Ba02 = 44.0 kb) and boundaries of each prophage. In CDC 684 (Figure 2A), the external att sites (represented by black bars) are in relatively identical chromosomal positions to those in the Ames Ancestor. However, the internal att sites (represented by green and red bars) have dramatically exchanged positions between these genomes. In CDC 684, the right att site (red bar) for LambdaBa04 has moved to the left att position of Lambda Ba02, and likewise the left att site for Lambda Ba02 (green bar) has moved to the position occupied by right att site in Lambda Ba04. The net effect of this exchange is the creation of new hybrid prophages in CDC 684 (Figure 2B). These observations indicate that the large inversion event did not involve site-directed recombination but rather a homologous recombination event in the interior of both prophages.

A PCR approach was designed to detect the inversion sites in CDC 684 as a method that could test for the presence of the inversion in other isolates. Because of its size, the inversion is readily visible in the closed genome, but the molecular nature of the inversion is dependent on the proper alignment of two short regions (i.e. 165 bp) during the assembly of this genome. As illustrated in Figure 3, the 5' end of each of the rep sequences are distinct from each another and their positions are fixed at approximately the same positions in both genomes. However, the 3' end of the rep genes are highly homologous, with scattered SNPs the only distinguishing feature between these paralogs.

Due to constraints on PCR amplicon size we used mismatch amplification mutation assays (MAMA, 29 ) to discriminate between the right and left ends of the large inversion in CDC 684 and Ames Ancestor. The rationale was to demonstrate the different ends of the inverted 3.3 Mbp fragment in CDC 684 by use of real time PCR assays. The MAMA system was designed to take advantage of polymorphic differences that characterize the left and right SNP signatures within the rep Lambda-like protein sequences relative to the Ames Ancestor genome. Both the left and right assay systems have common primers (CP, Table 2 and Methods) that are fixed because they are external to the 3.3 Mbp inversion site. The internal primers are nearly identical but they target mismatches at specific SNP sites; G on the left site and A on the right site of the Ames genome. These same internal primers (e.g., CP-Left-Inv-F and Left-Inv-R, Table 2) cannot amplify the same 400 and 500 bp products in CDC 684 because they are separated by 3.3 Mbp. But the reciprocal pairings of the internal primers do amplify products from CDC 684.

These MAMA were used to analyze several isolates within the Vollum branch. In addition, the SNPs flanking the inversion were compared to in silico analysis of other B. anthracis WGS to determine the configuration of this 3.3 Mbp region in other non-Vollum strains. Table 3 illustrates that only the CDC 684 isolate possessed the inverted genotype from among 17 genomes examined, indicating the inversion is not common in B. anthracis.

In E. coli the large ter region has been found to contain a specific substrate sequence, dif (for Deletion Induced Filamentation), which is used by two recombinases, XerC and XerD, to resolve chromosomal multi-mers and to allow daughter chromosomes to segregate before cell division 30 31 . It has been proposed that the dif site (a short palindromic sequence) is in fact a more likely site of termination than any specific ter sites for both the E. coli and B. subtilis chromosomes 32 . From the perspective of the CDC 684 genome, the dif sites in both γ-proteobacteria and Firmicutes appear to have an extremely close association with the maximum GC-skew in those genomes that have been analyzed 32 33 .

Dif sites have been defined in both B. subtilis 34 and a member of B. cereus sub-group 32 . A cursory survey of the palindrome from the B. subtilis and B. cereus dif site (AATATATATT) in the Ames Ancestor identified a 28-bp palindromic sequence 32 that is located at nearly the precise genomic site of the cumulative GC-skew. This sequence is conserved and positioned at the cumulative ~ 180° GC-skew position of every complete whole genome sequence in all of the GenBank entries for the B. cereus sub-group (Table 4). The one exception is the genome of CDC 684 where the conserved dif-like sequence and the GC-skew are oriented at ~ 120° in relationship to the origin of replication (Figure 4, Table 4).

The significant difference in the spatial orientation of the ori site and dif/GC skew sites in CDC 684 suggests that there could be an alteration in how the bi-directional replication of chromosome would proceed because of the unequal distances the opposite leading strands would need to travel. Because accumulated evidence indicates that genomes like those of E. coli and Bacillus sp do not tolerate significant changes between the spatial orientation of the ori and ter sites, we designed a growth experiment to compare the growth kinetics of CDC 684 to those of three wild type B. anthracis strains.

Growth curves for four strains of Bacillus anthracis: Ames, Ba_A0361 (a B branch isolate), Vollum and CDC 684 were grown in LB broth at 37°C (Figure 5). These cultures were grown in duplicate (Ames, BaA0361) or triplicate (Vollum, CDC 684) with growth measured by OD₆₀₀. The strains represent two major phylogenetic groups of B. anthracis. Note the relatively consistent growth curves for the three wild type isolates: Ames, Ba A0361 and Vollum, the closest relative to CDC 684. Two obvious differences between the CDC 684 and Vollum growth curves is a longer lag phase and slower mid log growth rate in CDC 684. These differences were noted despite careful efforts to exactly match inoculum sizes using direct plating viability counts. An extended lag phase would suggest that CDC 684 takes longer to adapt to the inoculum transfer process and/or to conditions necessary for growth and cell division. The slower mid log growth rate (~55 min in Vollum and ~80 min in CDC684) in this experiment suggests that even after revival from lag phase that CDC 684 has a cellular limitation to growth that does not exist in the wild type strains. These results provide a growth parameter that implies that the spatial change in the orientation of the origin of replication and the termination site in CDC 684 may have altered the growth of this isolate.

The genome of B. anthracis CDC 684: Chromosome [GenBank: CP001215.1]. pXO1 [GenBank: CP001216] and pXO2 [GenBank: CP001214] was sequenced at the Joint Genome Institute (JGI)/J. Craig Venter Institute using a combination of 3 kb and 8 kb DNA libraries. All general aspects of library construction and sequencing performed at the JGI can be found at http://www.jgi.doe.gov/. Draft assemblies were based on 59,691 total reads. The Phred/Phrap/Consed software package http://www.phrap.com was used for sequence assembly and quality assessment 43 44 . After the shotgun stage, reads were assembled with parallel Phrap (High Performance Software, LLC). Possible mis-assemblies were corrected with Dupfinisher 45 or transposon bombing of bridging clones (Epicentre Biotechnologies, Madison, WI). Gaps between contigs were closed by editing in Consed and by custom primer walking (Roche Applied Science, Indianapolis, IN). A total of 1955 additional custom PCRs were necessary to close gaps and to raise the quality of the finished sequence. The completed genome sequence of B. anthracis str. CDC 684 contains 62,606 reads, achieving an average of 10-fold sequence coverage per base with an error rate of < 10^-6.

Spores were prepared from B. anthracis CDC 684 as previously described 16 and female Hartley guinea pigs (660 g) were challenged intramuscularly (i.m.) with various spore concentrations (see 'Results') at USAMRIID as previously described 16 46 . Research was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to experiments involving animals and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals (National Research Council. 1996. Guide for the care and use of laboratory animals National Academy Press, Washington, DC.). The facility where this research was conducted is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International.

The thirteen canSNP alleles and the specific assays for each have been described previously 19 . TaqMan™ Minor Groove Binding (MGB) allelic discrimination assays were used to determine the precise canSNP grouping for every isolate used in this study 19 47 .

Additional SNP genotyping was conducted using the Mismatch Amplification Mutation Assay [MAMA] 29 , which is based on allele-specific PCR kinetics 48 , enhanced by penultimate mismatch primer design 29 49 . The MAMA approach was also used to distinguish the inverted 3.3 Mbp segment of CDC 684 from all other B. anthracis strains. MAMA assays were designed for both the 5' (left) and 3' (right) ends of the inversion; i.e., two sets of primer products separated by 3.3 Mbp. The sequences flanking the 3.3 Mbp inverted region were unique and common to both CDC 684 and the Ames genomes and were defined as Common Primers (CP). But the internal primers targeted nearly identical sequences and therefore used primers designed around mismatches that could distinguish and generate 400 and 500 bp PCR products. The primers were as follows (5' to 3'): Left-inv-R (TAAAGCATCCACATCTTTTAATGg

C

T

The MAMA assay system was also used to type 10 new canSNP sites that further define the Vollum lineage of B. anthracis. The primers for these sites are shown in Additional File 2 as a Table.

Each inversion SYBR MAMA reaction comprised 1X SYBR Green Master Mix (Applied Biosystems, Foster City, CA), 0.1 uM MAMA primer, 0.2 uM consensus primer, 0.08 U Platinum Taq polymerase (Invitrogen, Carlsbad, CA) and molecular grade H₂O to 9 uL. One uL of genomic DNA was added to each well to a final volume of 10 uL. Reactions were carried out in 384-well optical plates (Applied Biosystems) on an ABI Prism 7900 HT real-time instrument (Applied Biosystems) using the following thermocycling parameters: 2 min at 50°C, 10 min at 95°C, followed by 50 cycles of 15 s at 95°C and 1 min at 60°C. PCR products were subject to post-amplification dissociation (15 sec at 95°C, 15 sec at 60°C, 15 sec at 95°C) to confirm product specificity.

Additional File 1 provides an example of real time PCR profiles for the left inversion region using a fixed Common Primer (CP) that is located outside of the left boundary of the 3.3 Mbp inversion site in both CDC 684 and the Ames genomes. This figure demonstrates real time PCR kinetics for the detection of amplicons for the left boundary of the inversion site in both CDC 684 and the Ames Ancestor Genome using primer combinations described in Table 2.

A free software program, GenSkew http://genskew.csb.univie.ac.at/, was used to compute the cumulative skew for 15 complete WGS of B. anthracis, B. cereus and B. thuringiensis. These WGS data were downloaded from GenBank: http://www.ncbi.nlm.nih.gov/genbank/.

Stocks of B. anthracis Ames, B. anthracis Vollum (A0488), B anthracis A0361 (a B branch isolate), and B. anthracis CDC 684 were subcultured and grown for ~19 hours on LB agar. These cells were harvested and normalized to OD₆₀₀ densities that correspond to 10⁵ cfu/mL based on viable count estimates from previous experiments for each isolate. These measurements were used to precisely add 10⁵ cfu inoculums to create 3 ml culture tubes for each isolate. These cultures were grown at 37° C and OD₆₀₀ measurements were determined on a CO800 Spectrophotometer.