Next-Generation Bacillus anthracis Live Attenuated Spore Vaccine Based on the htrA- (High Temperature Requirement A) Sterne Strain

Anthrax is a lethal disease caused by the gram-positive spore-producing bacterium Bacillus anthracis. Live attenuated vaccines, such as the nonencapsulated Sterne strain, do not meet the safety standards mandated for human use in the Western world and are approved for veterinary purposes only. Here we demonstrate that disrupting the htrA gene, encoding the chaperone/protease HtrA (High Temperature Requirement A), in the virulent Bacillus anthracis Vollum strain results in significant virulence attenuation in guinea pigs, rabbits and mice, underlying the universality of the attenuated phenotype associated with htrA knockout. Accordingly, htrA disruption was implemented for the development of a Sterne-derived safe live vaccine compatible with human use. The novel B. anthracis SterneΔhtrA strain secretes functional anthrax toxins but is 10–104-fold less virulent than the Sterne vaccine strain depending on animal model (mice, guinea pigs, or rabbits). In spite of this attenuation, double or even single immunization with SterneΔhtrA spores elicits immune responses which target toxaemia and bacteremia resulting in protection from subcutaneous or respiratory lethal challenge with a virulent strain in guinea pigs and rabbits. The efficacy of the immune-protective response in guinea pigs was maintained for at least 50 weeks after a single immunization.

(a zinc protease, which together with PA forms the exotoxin LT) and EF (an adenylate cyclase, which together with PA constitutes the exotoxin ET). An additional virulence factor is the poly D-glutamate anti-phagocytic capsule synthesized by enzymes encoded by genes located on the second native plasmid pXO2 3,4 .

The HtrA (High Temperature Requirement A) protein of B. anthracis
We have previously conducted functional-genomic and proteomic global surveys, combined with serological evaluation of B. anthracis antigens, for the identification of novel virulence factors as well as vaccine and diagnostic marker candidates among extracellular (secreted or membrane) proteins. These studies revealed a list of novel antigens (other than the classic exotoxins LT and ET) for further evaluation based on their immunogenicity, abundance under various culture conditions, and functional relatedness to infection [5][6][7][8][9][10][11][12] . HtrA, one of the B. anthracis proteins that emerged from these screens, is a secreted and membrane-associated bifunctional chaperone/protease encoded by a chromosomal gene. HtrA was subsequently shown to be abundantly expressed during infection and was consequently able to serve as an early disease-biomarker of anthrax 13 . Our recent study of the phenotype associated with directed disruption of the htrA gene 14,15 demonstrated that HtrA is necessary for tolerance to a variety of stress stimuli and for the secretion of several major extracellular proteins. Most notably, htrA gene disruption in the fully virulent toxinogenic and capsular Vollum strain resulted in a dramatic attenuation in the guinea pig model following subcutaneous or intranasal exposure in spite of the fact that the mutated bacteria were able to express the PA, LF and EF toxin subunits as well as the anti-phagocytic capsule 15 . The extent of the virulence attenuation associated with htrA disruption was significantly higher than that promoted by disruption of any other reported B. anthracis virulence determinants (other than the LT and ET toxins and the capsule biosynthetic operon themselves), which were suggested to play a role in bacterial pathogenesis (examples of such factors are listed in Table 1). As shown in Table 1, the impact on virulence associated with a particular mutation was not manifested in all the animal models employed and by all the routes of exposure. Here, it is shown that this is not the case with the htrA disruption: infection studies of the htrA-disrupted B. anthracis fully virulent Vollum strain in three rodent models of anthrax demonstrate the universality of the attenuated phenotype.
Immunization of guinea pigs with sub-lethal doses of the htrA-disrupted B. anthracis Vollum strain elicited an immune response that conferred protection to a subsequent lethal challenge 15 . These results suggested that HtrA is a major virulence factor essential for manifestation of B. anthracis pathogenesis and that htrA-gene disruption may represent the basis for the development of novel prophylactics against B. anthracis infection.

Anthrax vaccines
Because PA, the non-detrimental toxin subunit, is a very potent immunogen and its neutralization by the humoral immune response prevents the onset of the disease, all anthrax vaccine preparations contain PA 3,16 . The anthrax protein vaccines that are licensed for human use include partially purified PA preparations potentiated by various adjuvants.
In general, live attenuated vaccines are considered to promote superior protective immunity compared to sub-unit vaccines owing to the notion that exposure of the host to the infective micro-organism in the context of infection (i) elicits a response against a wide range of antigens rather than a unique immunogen, and/or (ii) involves a variety of immune and inflammatory mechanisms rather than a limited humoral response to one particular antigen. Attenuated toxin-producing live bacteria (such as the nonencapsulated Sterne strain, pXO1 + ; pXO2 − ) have been used worldwide as effective veterinary vaccines since the 1930 s and in the past were used for human vaccination. While millions of people were vaccinated with the Sterne-like live vaccine (known as the STI strain) 17 between 1950 and 1980 in the former Soviet Union, the potential reactogenicity of the Sterne vaccine as indicated by animal experimentation, precluded use in humans in the Western world. Consequently, on grounds of toxicity, these live attenuated vaccines do not meet the safety standards mandated for human use and are suitable for veterinary purposes only. Yet, the protection conferred by administration of the Sterne attenuated live vaccine to experimental animals appeared to be superior to that provided by vaccination with PA in terms of the longevity of the protective response, efficacy against a broad spectrum of B. anthracis strains and the need for less administrations (1-2 injections, compared to the initial cumbersome 6-dose vaccination schedule with the subunit PA vaccine) 18 . Therefore, further attenuation of a Sterne vaccine strain could be a beneficial step in the development of next-generation live anthrax vaccines for human use. The present study demonstrates, in three animal models, that disruption of the htrA gene in the B. anthracis Sterne-vaccine platform is sufficient to generate a highly attenuated strain that preserves its ability to induce a protective immune response. The animal groups (10 mice or 5 rabbits per experimental group) were inoculated SC with increasing spore doses of the parental Vollum (black circles), mutated VollumΔ htrA (white squares) and transcomplemented VollumΔ htrA/HtrA (grey diamonds/grey dotted line) strains. Animal survival was followed for 3 weeks. In the experiment carried out in rabbits, only the 10 4 spore dose of the transcomplemented VollumΔ htrA/HtrA strain was tested, and all the animals in this experimental group succumbed to the infection. Upper-right inset: The results of a published previously similar experiment carried out with guinea-pigs 15 . A horizontal 50% survival dotted line in the "Mice" and "Rabbits" survival-profile panels indicates the estimated LD 50 values of the various strains. Lower-right inset: Anti-HtrA Western blot analysis of the secretomes of the three strains demonstrating complete elimination of HtrA expression in the mutated strain and restoration of the HtrA level in the trans-complemented strain is shown.

Results
Virulence attenuation of the Vollum ΔhtrA strain in various animal models of anthrax. To determine whether disruption of the htrA gene is associated with virulence attenuation not only in guinea pigs 15 (see also Fig. 1, inset) but also in other animal models, studies assessing the survival of experimental animals following administration of the VollumΔ htrA strain were carried out in mice and rabbits. The experiments consisted of subcutaneous (SC) administration of increasing doses of either one of 3 strains of bacteria: the parental wild-type Vollum strain (pXO1 + , pXO2 + ), its mutated Δ htrA derivative strain and the HtrA trans-complemented strain VollumΔ htrA/HtrA. The results depicted in Fig. 1 and summarized in Table 2 clearly demonstrate that in all three animal models the disruption of the htrA gene is associated with a substantial decrease of virulence. The attenuation indices (ratio between the LD 50 values of the mutant and the wild-type parental strains) promoted by the htrA disruption in the Vollum strain background ranged from 10 4 in mice to greater than 10 7 in rabbits. These attenuation values are comparable to those observed in guinea pigs ( Table 2). The extent of the virulence attenuation in mice is a rather unexpected result considering the fact that ICR out-bred mice exhibit high sensitivity to B. anthracis (as opposed to some other in-bred murine strains 19 ) and that mice in general are notoriously sensitive to the poly-D glutamic acid capsule, the synthesis of which is not affected in the Vollum mutant strain 15 . This observation strongly supports the notion that HtrA is essential for the manifestation of B. anthracis virulence and underlines the dominant nature of the htrA mutation as a means for virulence attenuation. In the rabbit system, which is considered to represent a suitable animal anthrax model that more faithfully emulates the human disease 3,20 , none of the animals succumbed upon administration of as much as 5 × 10 8 spores (higher doses were not attempted, therefore the LD 50 value is considered to be above this dose). The virulence attenuation phenotype of VollumΔ htrA was restored to levels comparable to those of the parental Vollum strain upon expression of HtrA in the trans-complemented strains VollumΔ htrA/HtrA (Fig. 1).
Generation and phenotypic characterization of the SterneΔhtrA strain. We reasoned that elimination of the htrA gene may extend the safety of the Sterne vaccine by decreasing its virulence. Consequently, a novel Sterne Δ htrA strain was generated by disruption of the htrA locus (NCBI locus tag GBAS_3995). The chromosomal context of the htrA gene in the Sterne strain is identical to that in the Vollum strain with htrA encoded as a monocistronic gene (Fig. 2a). Therefore, the gene targeting manipulation for disruption of the gene in the Sterne strain was carried out by allelic-exchange approaches (Fig. 1a, see also Materials and Methods) based on the upstream and downstream target homology sequences, which are identical to those employed for the disruption of the htrA gene in the Vollum strain 15 . The disruption was confirmed by direct sequencing of the mutated htrA locus as well as by Western blot analysis, which demonstrated the complete abrogation of HtrA expression (Fig. 2b). In addition, we have also generated a transcomplemented strain, SterneΔ htrA/HtrA, by introducing an extrachromosomal HtrA expression plasmid into the mutated strain, which restored HtrA synthesis to levels similar to those observed in the parental Sterne strain (Fig. 2b).
No growth differences were observed between the Δ htrA strain and the parental Sterne strain cultures (Fig. 2c). Furthermore, the Sterne and the SterneΔ htrA strains were indistinguishable with respect to their ability to germinate and sporulate ( Supplementary Fig. S1, online). The phenotype of the mutated strain was determined in vitro with respect to the characteristics and hallmarks previously determined for the htrA-mutated B. anthracis Vollum strain 15 and other bacteria [21][22][23][24] . These phenotypic characterizations, which included growth under temperature and oxidative stress conditions as well as expansion in cultured macrophages, are documented in Supplementary Figs S2 and S3 (online). Compared to the parental strain, the SterneΔ htrA strain demonstrated a limited ability to grow at high temperature, an increased sensitivity to oxidative stress caused by hydrogen peroxide ( Supplementary  Fig. S2, online) and a delay (of more than 8 hours) in the J774-macrophage infection assay ( Supplementary Fig. S3, online). All these phenotypic alterations could be restored to the parental-like levels by trans-complementation in strain SterneΔ htrA/HtrA, which demonstrates that these characteristics are directly related to the disruption of the htrA gene.
Production of toxins by the SterneΔhtrA strain. Considering the fact that HtrA is a chaperone involved in the secretion machinery 25,26 , the efficacy of a live vaccine based on the SterneΔ htrA strain could be compromised by a decreased ability of the mutated strain to synthesize and/or to secrete the B. anthracis toxins PA, LF and EF. Direct evaluation by Western blot analysis of the toxin level clearly demonstrated that the mutation did not appear to affect the level of PA, LF or EF produced by the SterneΔ htrA mutant strain (Fig. 2d). Both of the toxic subunits of the exotoxins, LF and EF, were secreted by the mutated strain and appeared to be indistinguishable from those secreted by the parental strain in a macrophage-lysis assay, which probes the functional integrity   In vivo pathogenicity studies in animal models of anthrax. The impact of the htrA deletion on the virulence of the B. anthracis Sterne strain was evaluated in mice, guinea pigs and rabbits by SC administration of increasing doses of mutant spores in comparison to administration of Sterne parental-strain spores. The results depicted in Fig. 3 demonstrated that the virulence of the SterneΔ htrA bacteria was significantly attenuated (see Table 2 for the LD 50 values of the mutated bacteria). The virulence attenuation index was almost 4 orders of magnitude in mice and guinea pigs. Furthermore, an increased dose-dependent mean-time to death (MTTD) was observed for the animals infected with the mutant SterneΔ htrA strain in the murine model. While for mice the MTTD for the Sterne parental strain was 4 days (for a dose of 10 5 spores) and 2.5 days (for a dose of 10 8 spores), in the case of the mutant SterneΔ htrA strain, the MTTD ranged from 8 days (for a dose of 10 5 spores) to 3 days (for a dose of 10 8 spores). Of note, the virulence attenuation could be alleviated by expression of HtrA in the transcomplemented strain SterneΔ htrA/HtrA, which was indistinguishable from the parental strain in the murine system (Fig. 3). In guinea pigs, 95% of the animals (19 out of 20) survived a dose of 10 9 spores SterneΔ htrA, but 100% of animals (10 of 10) survived a higher dose of 10 10 . We therefore estimate the LD 50 of SterneΔ htrA in guinea pigs to be greater than 10 10 . In rabbits, no animals succumbed upon administration of 5 × 10 9 spores (higher doses were not attempted for practical reasons); therefore, the exact LD 50 value could not be determined. (We note that the LD 50 value of the parental Sterne strain in rabbits is approximately 5 × 10 8 spores with occasional lethality at doses as low as 10 7 ). As high doses of the spores (e.g., 10 9 spores/animal) could be administered to guinea pigs, relatively high serum levels of PA could be reached in infected animals 72 hours after SC infection with SterneΔ htrA (Supplementary Figure S5, online). Similar high levels could not be attained following vaccination with the parental Sterne due to the limit of tolerable doses for immunization (less than 10 7 spores). The bacterial loads detected in the spleens of were inoculated via the SC route with increasing spore doses of the parental Sterne strain (black circles), the mutated SterneΔ htrA strain (white squares) and the transcomplemented SterneΔ htrA/HtrA strain (tested in mice grey diamonds/grey dotted line). The survival of the animals was followed for 3 weeks. Table 2 shows the LD 50 values calculated for the various strains in the three animal models. The experiments described employed the SterneΔ htrA IIBR-BA106 strain, but similar results were observed with the SterneΔ htrA IIBR-BA104K R strain (not shown). In the guinea pig experiment, 19 of 20 animals survived a dose of 10 9 SterneΔ htrA spores, but all the animals survived a dose of 10 10 spores.
Scientific RepoRts | 6:18908 | DOI: 10.1038/srep18908 infected animals (Supplementary Figure S5, online) indicate that the mutated bacteria can migrate to the spleens of the guinea pigs. However, 32 days after SC administration, no bacteria could be detected in the spleens of animals infected with doses as high as 10 9 spores. Immunization with SterneΔhtrA spores confers efficient protection against anthrax. To evaluate the potential use of the SterneΔ htrA strain as an efficient anthrax vaccine, both guinea pigs and rabbits were immunized with different doses, and their humoral immune response elicited by the immunization was evaluated. In addition, we evaluated the ability of the vaccinated animals to withstand a lethal challenge by a subcutaneous (SC) or respiratory (IN) exposure at a spore dose of 100 × LD 50 of the fully virulent B. anthracis Vollum strain (see schematic description of the experiments in Figs 4a and 5a). The experiments in guinea pigs included SC administrations of a single immunization (10 6 -10 9 spores/animal, Fig. 4) or two immunization doses 4-weeks apart (5 × 10 6 -10 9 spores/animal, Fig. 5). Both immunization regimens elicited protective immune responses as depicted in Figs 4-6. Following the single immunization, we found that all vaccinated animals developed substantial titres of antibodies directed against total B. anthracis proteins (Fig. 6a) as well as antibodies against PA (Fig. 6b). These effects were dose-dependent. Significant titres of toxin neutralizing antibodies (TNA) could be detected in animals vaccinated with 10 7 or more spores (Fig. 6c). In the double-immunization experimental groups, the humoral response was 1-2 orders of magnitude higher than that observed after a single immunization with the same dose (compare Fig. 6b-e and Fig. 6c-f). These observations underline the efficient boost effect promoted by a second dose of the vaccine spores administered 4 weeks after the initial immunization.
Vaccination of guinea pigs with the SterneΔ htrA strain could protect animals from a subsequent SC or IN challenge with 100 × LD 50 of the fully virulent B. anthracis Vollum strain (Figs 4 and 5). We observed that 80% of the animals vaccinated once with 5 × 10 8 spores survived IN challenge, and 100% of the animals vaccinated with 10 9 spores were protected (Fig. 4). Full protection was also observed in animals vaccinated twice with more than 10 7 spores (Fig. 5), a dose that was 3 orders of magnitude lower than the calculated LD 50 of the SterneΔ htrA vaccine strain in guinea pigs ( > 10 10 , Fig. 3 and Table 2).
The protective value of the vaccine was further confirmed in rabbits. We observed that administration of one dose of 5 × 10 8 spores of the SterneΔ htrA strain resulted in a strong humoral anti-toxin and antibacterial response, which promoted full protection of animals in a subsequent IN challenge (Fig. 7). Again this immunization dose is at least 10-fold lower than the lowest possible LD 50 dose of the SterneΔ htrA mutant (note that none of the rabbits died after exposure to the highest infection dose tested, see Table 2).
The longevity of the immune state was addressed in guinea pigs vaccinated with one dose (10 9 ) or two doses (10 8 , 4 weeks apart) of SterneΔ htrA spores. In these experiments, the antibody titres of randomly selected animals from the various vaccination groups were determined at different time points following the last immunization. The data in Supplementary Table S1 (online) show that the humoral response declined gradually yet was maintained at a sufficiently high protective level in both immunization modes, and provided complete protection to all animals (20 animals/group) against an IN lethal challenge (100 × LD 50 ) 52 weeks after vaccination.
The results presented above clearly demonstrate that disruption of the htrA gene can be implemented for the development of a safe live-attenuated anthrax vaccine without compromising the protective characteristic exhibited by the parental Sterne vaccine strain.

Discussion
The currently licensed human anthrax vaccines contain filtered supernatants of avirulent B. anthracis strains potentiated by various adjuvants 3,16 . The active ingredient of these formulations is the toxin subunit PA. Although these vaccines were demonstrated as efficient and safe, they suffer from a cumbersome vaccination schedule that involves multiple initial administrations and yearly boost immunizations. Efforts to improve the sub-unit vaccine by bacterial additives or adjuvants 3 are paralleled by the alternative approach, which relies on live attenuated strains as a basis for a next-generation vaccine. For example, several studies explored the possibility of using non-virulent mutated strains of B. anthracis as vaccines [27][28][29] or even heterologous platforms, such as vaccinia virus, Bacillus subtillis or Salmonella typhimurium [30][31][32][33] , expressing recombinant forms of PA. In the past, we developed a nontoxinogenic and nonencapsulated recombinant spore vaccine based on a pXO1 and pXO2 cured variant of the nonproteolytic V770-NPI-R strain [34][35][36] . In the present study, we report on a novel live attenuated vaccine strain based on the existing Sterne veterinary vaccine strain (currently precluded from human use in the Western world, see Introduction) that is attenuated for virulence through deletion of the htrA gene. HtrA is a bifunctional chaperone and protease that was shown to be essential for the manifestation of B. anthracis virulence possibly owing to its role in protein quality control, which is necessary for overcoming the stress insults encountered by the bacteria in the host 15,37 . Here, we demonstrated in three animal models that extensive attenuation of the virulent Vollum strain, as well as that of the Sterne vaccine strain, can be achieved by disruption of htrA (Figs 1 and 3 and Table 2). The success of a live attenuated vaccine encompasses a delicate balance between low virulence (hence amenable with human use) and the ability to promote a protective immune response. Virulence attenuation may lower the pathogenicity of the micro-organism to the extent that its interaction with the host results in an immune response insufficient to promote protection in a subsequent encounter with the fully virulent strain (a central theme in vaccine development 38 ). The observations reported here clearly demonstrate that in spite of its markedly reduced virulence, the SterneΔ htrA strain elicits robust protection against a lethal challenge with the fully virulent Vollum strain administered via the SC or IN routes in vaccinated guinea pigs and rabbits (Figs 4-7 and Supplementary Table S1, online). Furthermore, inspection of the anti-PA, anti-bacterial proteins and toxin-neutralizing antibody titres in vaccinated animals demonstrates that the immune response following vaccination with the Sterne Δ htrA strain (Figs 6 and 7 and Supplementary Table S1, online) targets both the toxin and other bacterial antigens. Therefore, this response addresses both toxaemia and the bacteremia, which are both associated with B. anthracis infection. Of note, full-protection of rabbits vaccinated with a PA sub-unit vaccine correlates with at least 3-fold higher levels of neutralizing antibodies 39,40 compared to the antibodies in the fully protected animals following SterneΔ htrA vaccination. Similarly, in the guinea pig immunization experiments (Fig. 6), some animals exhibited  Table S1, online).
The observations discussed above indicate that the novel SterneΔ htrA vaccine exhibits (i) significantly improved safety as manifested by its dramatic virulence attenuation, and (ii) high efficacy in protecting against infection by a fully virulent B. anthracis strain. The balance between these two parameters determines the value of the live attenuated vaccine, which may be quantified and fully appreciated by determining the "therapeutic window" of the vaccine in guinea pigs. This is schematically visualized by comparing the LD 50 and the PD 50 (Protective Dose 50%) as exhibited in guinea pigs by the parental Sterne strain and the novel SterneΔ htrA strain (Fig. 8). The conclusion from this presentation is that in the case of the Sterne vaccine, survival from a lethal challenge can be achieved after immunization at doses that are close to the LD 50 value (thus with a minimal "therapeutic window" ratio of LD 50 to PD 50 of only 8.0). However, in the case of the SterneΔ htrA vaccine, administration of as little as 10 −2 × LD 50 (for single dose immunization) or even 10 −3 × LD 50 (for the double-dose immunization) is sufficient to promote over 50% survival after challenge (Fig. 8). This calculation estimates that the "therapeutic window" of the novel SterneΔ htrA is at least two orders of magnitude higher compared to that of the parental Sterne veterinary vaccine, which supports the conclusion that the former is a substantially safer vaccine.
In conclusion, the immunization of guinea pigs and rabbits by a single or double administration of this novel SterneΔ htrA live-attenuated spore vaccine at doses well below the LD 50 of the vaccine strain elicits full protection against subcutaneous or airway infection of a fully virulent anthrax strain. Therefore, the SterneΔ htrA strain can serve as an appropriate platform for a next-generation human anthrax vaccine. For the identification of CO 2 -induced proteins, the cells were grown at 37 °C in NBY supplemented with 0.9% NaHCO 3 in hermetically sealed flasks without agitation (referred hereafter as NBY-CO 2 ). NBY-CO 2 media promotes efficient toxin production 9,14,42 . The spores were prepared in Schaeffer's sporulation media (SSM) at 34 °C for 72 hours with vigorous shaking. In all the cases, the cultures were initiated with a starter inoculum diluted to a final OD (660 nm) of 0.1. For the H 2 O 2 stress test, the cells were allowed to grow under optimal conditions for 2.5-3 hours (typically the time needed to enter logarithmic phase) and were then split into BHI cultures containing the indicated final concentrations of H 2 O 2 . The growth rate μ (expressed in hr −1 ) was calculated as 60 × [ln(OD 2 /OD 1 )]/(t 2 -t 1 ), where OD 1 and OD 2 are the optical densities (at 600 nm) measured at two time-points (t 1 and t 2 expressed in minutes) along the logarithmic growth phase.
Escherichia coli strains (Supplementary Table S2, online) were used for plasmid construction. The antibiotic concentrations used for selection in Luria-Bertani (LB, Difco) agar/broth were as follows: for E. coli strains, ampicillin (Ap, 100 μ g ml −1 ); for B. anthracis strains, kanamycin (Km, 10 μ g ml −1 ), chloramphenicol (Cm, 7.5 μ g ml −1 ) and erythromycin (Em, 5 μ g ml −1 ). The calculated "therapeutic window" was determined from the ratio between the safety (LD 50 ) and the efficacy (PD 50 ) values of the two vaccine strains. Note that for SterneΔ htrA, only a lower limit of the LD 50 was determined ( Fig. 3 and Table 2), and therefore, the therapeutic window may be even wider than has been estimated in this figure.
Scientific RepoRts | 6:18908 | DOI: 10.1038/srep18908 Plasmid and strain construction. The plasmids and oligonucleotide primers used in this study are summarized in Supplementary Table S2 (online). The oligonucleotide primers were designed according to the genomic sequence of B. anthracis "Ames ancestor" strain (accession number AE017334, GI 50082967) and were prepared using the Expedite synthesizer (Applied Biosystems). The sequence identity of the htrA locus in the Vollum and Sterne strains was verified by whole-genome sequencing. Genomic DNA was extracted as previously described 42,43 . The PCR amplifications were performed using the Taq (Qiagen) or Expand High Fidelity (Roche) systems. The DNA sequences were determined by the ABI rhodamine termination reaction (ABI310 Genetic Analyzer, Applied Biosystems). The plasmid used for generation of the htrA-disrupted strains and the trans-complementation plasmids for extrachromosomal HtrA expression were previously described 15 . All the plasmids transformed into the Vollum or Sterne strains were first propagated in the methylation-deficient E. coli strain GM2929. B. anthracis cells were electro-transformed as previously described 34 . To disrupt the htrA gene by one-step homologous recombination, an allelic exchange technique was performed as follows: plasmid pEO-htrA was introduced into competent cells of the Sterne strain, and the transformants were selected for Km R at 30 °C. The integrants were recovered by growing the transformants in LB broth at 30 °C for 1.5 h, shifting to 38 °C (non-permissive temperature) for 6 h, plating serial dilutions on LB plates containing Km and were then incubated at 38 °C for 12-16 h. Single colonies were selected, resuspended in 0.1 ml LB broth, spotted (5 μ l) on duplicate LB plates containing Km or Em, and then the twin plates were incubated either at 42 °C or 30 °C for 12-16 h. The deletion mutants were isolated as Km R Em S clones that failed to grow at 42 °C (as expected from the phenotype associated with htrA-gene disruption). The deletion of the internal fragment of htrA and insertion of the Km R cassette into the chromosome were confirmed by PCR using flanking chromosomal primers and primers derived from the Km cassette (primers used: HTRA9/KANA2C and KANA1/HTRA12C for the upstream and downstream integration htrA/Km junctions, respectively) and by sequencing of the htrA locus. The above approach generated a SterneΔ htrA strain that carries kanamycin resistance (IIBR collection BA104K R ). A two-step gene-disruption technique was implemented for the generation of an additional htrA-disrupted strain devoid of antibiotic resistance. This technique, which is based on the activity of the I-SceI recombinase, was previously reported 44 . The B. anthracis cells are first transfected with the plasmid pEGS-htrA, which contains the same upstream and downstream htrA gene homology regions as the plasmid pEO-htrA, an I SceI recombinase recognition site at the 5' of the downstream htrA homology region, an erythromycin resistance gene and a GFPuv gene 45,46 . The transfection results in the generation of a strain exhibiting a single cross-over integrant that is further resolved to the desired double cross-over markerless disrupted strain by expression of the I SceI recombinase (from the transfected pXX-I SceI plasmid). The above marker-less allelic exchange approach generated a SterneΔ htrA strain devoid of antibiotic resistance (designated in the IIBR collection BA106). The two SterneΔ htrA strains, BA104K R and BA106, were phenotypically indistinguishable. Throughout this report, the strain used for the various experiments is indicated in the relevant figure legend (see Fig. 2a for a schematic description of the disrupted htrA gene). Complementation was accomplished by transforming pASCHtrA into the htrA mutant strain.

SDS-PAGE, Western blot analyses, antibodies, and ELISA. SDS-PAGE was carried out
on 4-12% NuPAGE ® Bis-Tris gels (Invitrogen). For comparisons between proteins secreted by different strains, equal amounts of protein (50 μ g) from the corresponding strains were loaded side by side on SDS-PAGE. The Western blots were generated using the Nitrocellulose Western iBlot Gel Transfer Semi-dry system (Invitrogen). Visualization of the immunoreactive bands was accomplished using an ECL (electrochemiluminescence) reaction (Pierce SuperSignal ® West Pico Chemiluminescent substrate kit, Thermo Scientific) mediated by peroxidase-conjugated secondary antibodies (Amersham) detected by the FUJIFILM LAS-3000 detection system. The following antibodies were used in this study: (i) Specific anti-HtrA antibodies, which were obtained by DNA immunization with the pCI-htrA plasmid in female outbred ICR mice using the Helios Gene Gun System (Bio-Rad) as previously described 47 ; (ii) Rabbit anti-PA antibodies 35,48 ; (iii) Anti-LF and (iv) anti-EF mouse antibodies 34,36 . The primary and secondary antibodies were used at 1:1000 and 1:5000 dilutions, respectively.
The ELISA tests for the quantification of PA in sera samples and the detection of anti-PA antibodies in the serum of infected animals were carried out as previously described 34,36,[48][49][50] .
The ELISAs for the quantification of anti-bacterial antibody titres in the circulation of vaccinated animals were carried out on microtitre plates coated with a B. anthracis urea-extracted proteinous preparation 8 as previously described 50 . The proteinous extracts were prepared from a nonvirulent, non-toxinogenic and non-encapsulated Δ Vollum strain (pXO1 − , pXO2 − ). Therefore, the ELISA test monitors the humoral response elicited against B. anthracis chromosomally encoded exposed proteins and does not include the anti-toxin response nor the capsular antigens (any antigen encoded by pXO1 or pXO2 is absent).
Quantification of the toxin-neutralizing antibody (TNA) titres in the circulation of vaccinated animals was carried out by a J774A.1 macrophage-lysis inhibition assay in the presence of excess PA and LF as previously described 36 .
Dilution drop assays. Ten-fold serial dilutions were generated from logarithmic growth phase bacteria (0.5 OD 600 ) and were dropped (50 μ l) in duplicate on LB agar plates that were incubated overnight at 37 °C or 42 °C. LF and EF functional assay. Bacteria (0.1 OD, 600 nm) were used to inoculate 1 ml cultures of DMEM (Biological Industries, Bet Haemek, Israel) supplemented with 10% FCS in 24-well microtitre plates and were incubated for 16 hrs at 37 °C in a 10% CO 2 atmosphere. The culture supernatant was collected by centrifugation (5 min; 4,000 × g) and was filtered twice through a 0.22-μ m filter. LF activity in the culture was assayed by Scientific RepoRts | 6:18908 | DOI: 10.1038/srep18908 layering the two-fold dilutions of the filtered supernatant in the presence of 0.25 μ g/ml PA on sub-confluent monolayers of J774A.1 cells (plated in 96-well microtitre plates at a density of 5 × 10 4 cells/ml one day prior to assay). The cells were incubated for approximately 4 hrs at 37 °C under 5% CO 2 until the cytopathic effect was observed microscopically. Visualization of the cell lysis was achieved by addition of XTT-reagent (Cell Proliferation Kit, Biological Industries, Bet Haemek, Israel), and the lysis was quantified using a spectrophotometer (Novaspec Plus, Amersham). A luminescent ATP-depletion EF functional assay based on the calmodulin-dependent adenylate cyclase activity of EF was developed. This assay is based on the strict dependence of a luminescent luciferase reaction on ATP, which can be efficiently converted to cAMP by adenylate cyclases. In brief, the DMEM culture supernatants (as above) were concentrated 5 times using the Microcon concentration/buffer exchange system (Millipore YM-10, Merck, Ireland). For the ATP depletion, 10 μ l of the concentrated supernatant (containing 3-5 ng of EF as quantified by ELISA) was added to 50 μ l of the luciferase assay reagent (Promega, Madison, WI, USA) containing 1 mM Calmodulin (Sigma) and 0.5 mM CaCl 2 (Merck) and was incubated in a polysorp fluoro-Nunc microtitre plate (Nunc) at 30 °C for 2.5 hrs. QuantiLum recombinant luciferase (Promega) was added to the reaction mixture, and the luminescence was measured immediately with a luminometer (VICTOR 3 TM, Perkin Elmer).

Evaluation of sporulation and germination efficiency.
To determine the sporulation efficiency, bacteria were sampled from cultures grown in SSM (at 34 °C with shaking at 200 rpm), and the percentage of spores was determined by viable plating before and after a 20 min heat incubation step (70 °C). While both vegetative cells and spores are equally efficient in generating colonies, after heat incubation only the heat-resistant spores are able to generate colonies. The spore percentage was calculated as the ratio between the CFU count after heat treatment divided by the CFU count before heat treatment (x 100). The germination efficiency was determined in BHI media inoculated with heat-treated spores at a final OD 600 of 0.5 by spectrophotometrically following the decrease in the optical absorbance at 600 nm, which is characteristic of spore depletion in the culture (Alvarez et al., 2010). The percentage of spores was calculated by comparing the optical absorbance before ( = 100% spores) and at the various time points after the initiation of germination. The results were confirmed by viable counting before and after a heat treatment step (70 °C, 20 min).

Macrophage infection assay.
The macrophage infections were carried out as previously described 43,49 . In brief, J774.1 murine macrophages grown in DMEM supplemented with 10% foetal calf serum (37 °C; 5% CO 2 atmosphere) were seeded at a concentration of 10 5 cells/well in a 24-wells plate one day prior to infection. The cells were incubated for 1 hour in the presence of 5 × 10 5 B. anthracis spores/well, were washed extensively with DMEM supplemented with 2.5 μ g/ml −1 gentamycin, were incubated for 3 hrs in the presence of 2.5 μ g/ml gentamycin and were washed again and layered with fresh media. Under these conditions, essentially all the bacteria that are not internalized are removed (as tested by viable bacterial counts). The supernatants from identically treated wells were harvested at various time points, and the viable bacterial counts were determined. Macrophage lysis as a consequence of the multiplication of the bacteria was monitored by assaying the accumulated lactate dehydrogenase (LDH) released in the media from the damaged cells at the harvesting times using a standard LDH-L kit (Thermo-Fisher Scientific, Middletown, VA, USA).

Dissemination of bacteria to spleens of infected guinea pigs. Female Hartley guinea pigs (Charles
River Laboratories), weighting 220 to 250 g were infected SC with various doses of Sterne or Sterne Δ htrA spores. At various time points (1-35 days) post-infection three randomly selected animals were euthanized and their spleens were removed, homogenized and plated in 10-fold serial dilutions for determining the total bacterial load. Similar samples were plated after heat-shock (70 °C, 20 minutes) to determine the percentage of spores in the organs.
Infection of experimental animals. Female Hartley guinea pigs (220 to 250 g, Charles River Laboratories), ICR mice (20-25 g, Harlan) and NZ rabbits (2.5 kg, Charles River Laboratories) were infected with various spore preparations. Prior to infection, the spore preparations were heat-shocked (70 °C, 20 min) to kill the residual vegetative bacteria and were serially diluted 10-fold in saline to produce spore suspensions within the range 10 2 -10 11 per ml. A 0.1 ml spore dose volume was administered subcutaneously (SC) to each guinea pig or mouse. A total of 5-10 guinea pigs per dose-strain were used. The rabbits were injected subcutaneously with 1 ml of PBS containing the desired amount of spores. The remaining spore dose suspensions were plated for total viable counts (CFU ml −1 ) to confirm the dose administered to the animals. The attenuation index is defined as the ratio between the LD 50 exhibited by the mutated strain and the WT parental ancestor. At seventy-two hours after infection, some of the guinea pigs were bled by cardiac puncture to determine the levels of PA in the peripheral circulation. The animals were observed daily for 21 days. Four weeks post-infection, the surviving guinea pigs or rabbits were bled by cardiac puncture or from the ear vein, respectively, for serological studies. In the guinea pig and rabbit immunization experiments, at 5 weeks post-infection (or 52 weeks in the immune longevity experiments) the animals were challenged SC or IN with the indicated lethal dose of the parental B. anthracis Vollum strain (SC-LD 50 = approximately 100 spores, IN-LD 50 = approximately 10 4 spores). The IN infection was performed following anaesthesia of the guinea pigs or rabbits by nasal instillation of 100 μ l PBS (guinea pigs) or 500 μ l PBS (rabbits) containing the desired dose (guinea pigs: 50 μ l/nostril; rabbits 250 μ l/nostril). The spore lethal dose required to kill 50% (LD 50 ) of the animals was calculated by non-linear fit regression using the GraphPad Prism (version 5) statistical analysis software (San Diego, CA).
Longevity of the immunological status in vaccinated guinea pigs. Anti-PA, anti-bacteria and toxin-neutralizing (TNA) titres in the blood samples of vaccinated guinea pigs were determined at various time points post-immunization. The blood samples were collected from 5 randomly selected animals from each group, and their antibody titres were determined. Therefore, the titre values (in all time-points except the last time-point preceding challenge, Supplementary Table S1, online) represent the response of 5 different animals from the same experimental group. At fifty weeks post-immunization, blood samples were collected from all animals for the evaluation of antibody titres. At fifty-two weeks post-immunization, all the animals were challenged with 100 × LD 50  Statistical analysis. Kaplan-Meier analyses and dose-dependent survival analyses were used for survival evaluation. The survival data were analysed using GraphPad Prism (version 5) statistical analysis software. A t-test was used to compare the mean survival time between the groups. A two-tailed log rank test was used to determine the statistical significance of differences between groups. A p value of < 0.05 was considered statistically significant.
All the animal experiments were approved by the IIBR committee for animal research. The experimental animals were handled according to the National Research Council 1996 Guide for the Care and Use of Laboratory Animals and regulations of the IIBR Animal Use Committee.