Production of Bacillus anthracis protective antigen is dependent on the extracellular chaperone, PrsA

Protective antigen (PA) is a component of the Bacillus anthracis lethal and edema toxins and the basis of the current anthrax vaccine. In its heptameric form, PA targets host cells and internalizes the enzymatically active components of the toxins, namely lethal and edema factors. PA and other toxin components are secreted from B. anthracis using the Sec-dependent secretion pathway. This requires them to be translocated across the cytoplasmic membrane in an unfolded state and then to be folded into their native configurations on the trans side of the membrane, prior to their release from the environment of the cell wall. In this study we show that recombinant PA (rPA) requires the extracellular chaperone PrsA for efficient folding when produced in the heterologous host, B. subtilis; increasing the concentration of PrsA leads to an increase in rPA production. To determine the likelihood of PrsA being required for PA production in its native host, we have analyzed the B. anthracis genome sequence for the presence of genes encoding homologues of B. subtilis PrsA. We identified three putative B. anthracis PrsA proteins (PrsAA, PrsAB, and PrsAC) that are able to complement the activity of B. subtilis PrsA with respect to cell viability and rPA secretion, as well as that of AmyQ, a protein previously shown to be PrsA-dependent.


[SUMMARY]
Protective antigen is a component of the Bacillus anthracis lethal and edema toxins, and the basis of the current anthrax vaccine. In its heptameric form, PA targets host cells and internalizes the enzymatically-active components of the toxins, namely lethal and edema factors. PA and other toxin components are secreted from B. anthracis using the Secdependent secretion pathway. This requires them to be translocated across the cytoplasmic membrane in an unfolded state and then to be folded into their native configurations on the trans side of the membrane, prior to their release from the environment of the cell wall. In this study we show that recombinant PA (rPA) requires the extracellular chaperone PrsA for efficient folding when produced in the heterologous host, B. subtilis; increasing the concentration of PrsA leads to an increase in rPA production. To determine the likelihood of PrsA being required for PA production in its native host, we have analyzed the B. anthracis genome sequence for the presence of genes encoding homologues of B subtilis PrsA. We identified three putative B. anthracis PrsA proteins (PrsAA, PrsAB and PrsAC) that are able to complement the activity of B. subtilis PrsA with respect to cell viability and rPA secretion, and as well as that of AmyQ, a protein previously shown to be PrsA-dependent.

[INTRODUCTION]
Bacillus anthracis is the etiological agent of anthrax -a historically well-documented disease. Anthrax affects mainly herbivorous animals such as sheep, cattle and wild herbivores, although all mammals are susceptible. The bacterium is transmitted predominately via spores rather than vegetative cells. Infection is usually acquired through the uptake of spores from soil or infected animal products by inhalation, ingestion or cutaneous abrasions. Fully virulent strains of B. anthracis possess two major virulence factors: the anthrax toxins and a γ-polyglutamic acid (γ-PGA) capsule.
The anthrax toxin proteins, edema factor (EF), lethal factor (LF) and protective antigen (PA) are encoded respectively by genes cya, lef and pag, located non-contiguously on plasmid pXO1 (185 kbp) (1). PA binds to either EF or LF to produce the binary edema (EdTx) or lethal (LeTx) toxins, respectively (5,6). Mice challenge experiments with isogenic strains of B. anthracis expressing either LeTx or EdTx indicate that, although these toxins act synergistically, LeTx is the key virulence factor (1).
The other main virulence factor, the γ-polyglutamic acid (γ-PGA) capsule, is encoded on plasmid pXO2 (95 kbp)(1). The capsule is only weakly immunogenic, and is therefore not suitable for vaccine purposes (2,3). γ-PGA is synthesized by enzymes encoded by the capA, capB and capC genes (4). The γ-PGA capsule forms the outermost element of the B. anthracis cell where it inhibits phagocytosis (2) by providing a monotonous linear polymer.
The current UK and USA human anthrax vaccines, although differing slightly, are based on PA (83 kDa.). The UK anthrax vaccine consists of an alum-precipitated culture filtrate from an aerobic static culture of B. anthracis strain Sterne 34F2, whilst the USA vaccine consists of an alhydrogel-adsorbed cell-free culture filtrate of B. anthracis V770-NP1-R, grown anaerobically in a fermenter. Both strains, although avirulent, must nevertheless be handled as class III pathogens. In an attempt to reduce the costs associated with handling strains of B. anthracis (5,6), various alternative production systems have been explored. However, with the notable exceptions of Bacillus subtilis and Escherichia coli, these have met with little success (7,8). Cloning pagA into B. subtilis strain IS53 (9) resulted in the secretion of recombinant PA (rPA) to a concentration of about 40 µg/ml; approximately three-fold higher than that obtained with B. anthracis Sterne (~15 µg/ml). Strain B. subtilis WB600, a multiply extracellular protease deficient strain (10,11), has been used to reduce degradation of rPA by co-produced proteases.
In an attempt to increase rPA production from B. subtilis, we have examined the role of PrsA on rPA secretion. PrsA is an essential lipoprotein component of the B. subtilis protein secretion pathway, where it functions on the trans side of the cytoplasmic membrane as a post-translocational folding factor (12). PrsA has been shown to be rate-limiting for the high level secretion of α-amylase. Increasing the cellular concentration of PrsA results in a corresponding increase in the amount of α-amylase secreted into the culture medium (13).
Here we show that secretion of rPA, like that of α-amylase, is PrsA-dependent.
While the B. subtilis YacD, a PrsA paralogue, is not able to complement PrsA activity, we show here that three B. anthracis PrsA orthologues are able to do so, with respect to both viability and protein secretion in B. subtilis. Table I lists the bacterial strains and plasmids used. Strains were grown and maintained in Luria-Bertani (LB) medium (per liter: 10 g tryptone, 5 g yeast extract, 10 g NaCl), excepting for the determination of rPA production from strains MFJ683, MFJ943 and MFJ945, which were grown in complete anthracis (CA) medium (per liter: 35 g tryptone, 5 g yeast extract, 6 g Na 2 PO 4 .7H 2 O, 1 g KH 2 PO 4 , 5.5 g NaCl, 40 mg L-tryptophan, 40 mg L-methionine, 5 mg thiamine, 25 mg uracil). Antibiotics were used at the following concentrations: ampicillin, 50 µg/ml; chloramphenicol, 5 µg/ml; erythromycin, 1 µg/ml; kanamycin, 10 µg/ml. Unless stated otherwise, isopropyl β-D thiogalactopyranoside (IPTG) was added at 1 mM and xylose at 1% (w/v) to induce gene expression from the P spac and P xyl promoters, respectively. → Table I DNA Manipulation and Strain Construction -B. subtilis was transformed with plasmid DNA using the "Groningen" method (14). pRCW101 was constructed by amplifying a fragment (~300 bp) at the 5'-end of prsA, including ribosome binding site, from chromosomal DNA using oligonucleotide primers (Bsu-prsA-fwd 5'-CGCAAGCTTATTTGGAATGATTAGGAG -3' and Bsu-prsA-rev 5'-CGCGGA

TCCAGGGCAGTATATTGATCG-3') with restriction endonuclease sites (BamHI and
HindIII) incorporated at their 5'-termini. The fragment was cloned into pMUTIN4 (15)  Multiple alignment of protein sequences -The protein sequences were aligned using the ClustalW tool (http://www.ebi.ac.uk/clustalw). Calculation of the relative identity of the sequences was carried out using the Needleman-Wunsch global alignment algorithm, which is available within the EMBOSS suite of software at the Human Genome Project Resource Centre (http://www.hgmp.mrc.ac.uk).

RESULTS
Influence of PrsA on rPA secretion -Mutants of prsA encoding a defective PrsA protein exhibit a defect in the secretion of AmyQ, an α-amylase derived from B. amyloliquefaciens (12). The prsA3 mutation appears to affect PrsA folding, resulting in post-translocational proteolysis of PrsA and reduced PrsA activity (16). To determine whether PrsA influences the production of rPA, the pagA-encoding plasmid pYS5 (17)  To confirm that PrsA influences the yield of rPA, strains were constructed so that the level of PrsA could be controlled. The pMUTIN4 integration vector (15) was used to construct B. subtilis strain RCW201 in which expression of prsA was under the control of the IPTG-inducible P spac promoter. Since prsA is essential for viability (12), we confirmed that RCW201 had an absolute requirement for IPTG. At IPTG concentrations between 0.1 and 10mM, the growth rate and yield were similar to that of the wild-type strain, however, at lower concentrations, growth was characterized by a lag of several hours and the growth exclusively of IPTG-independent suppressor mutants. In contrast, the growth a strain in which the prsA homologue yacD was placed under P spac control was not IPTG-dependent (data not shown). RCW201PA (i.e. RCW201 with pPA101) was used to analyze the relationship between prsA expression and the yield of secreted rPA. RCW201PA was grown in LB broth in the presence of various concentrations of IPTG (0.5-10 mM). rPA production was determined by quantitative Western blotting analysis of culture supernatants. The data (Fig. 2) show a direct relationship between the level of prsA expression (i.e. IPTG concentration) and the yield of mature rPA. Colonies appearing on media lacking both IPTG and xylose were found to be IPTGindependent and were most probably suppressor mutants (19). → Fig. 5 Strain RCW303, carrying Ban-prsAA, exhibited smooth colony morphology in the presence of xylose; conditions favoring PrsA production. This compared with a rough morphology when the host PrsA was produced or co-produced with PrsAA either in the presence of IPTG alone (Fig. 6), or with IPTG and xylose (not shown). In contrast, wild-type B. subtilis, and the strains carrying Ban-prsAB, Ban-prsAC or a second copy of Bsu-prsA, exhibited a normal rough morphology in the presence of xylose (Fig. 6) and/or IPTG (not shown). These observations indicate that Ban-PrsAA has a different substrate specificity to that of the other PrsA proteins. → Fig. 6 The

influence of B. anthracis homologues on the production of AmyQ and rPA -B. subtilis
PrsA levels influence the yield of AmyQ (12) and rPA (Fig. 2). Consequently, strain RCW302 with a xylose-inducible copy of Bsu-prsA, and strains RCW303, RCW304 and RCW305 with xylose-inducible copies of Ban-prsAA, Ban-prsAB and Ban-prsAC, respectively, were used to determine their influence on the secretion of AmyQ and rPA.
The B. subtilis prsA complementation strains transformed with pKTH10, viz RCW302Amy, RCW303Amy, RCW304Amy and RCW305Amy, were grown in LB broth containing 1% xylose. The growth of the strains expressing P xyl -controlled B. subtilis or B.
anthracis prsA genes was similar to that of RCW101 expressing B. subtilis prsA from its native promoter (Fig. 7). In each case, AmyQ synthesis was induced during transition from exponential to stationary phase and the highest yield of AmyQ was observed in stationary phase. The stability of AmyQ in culture medium was confirmed by the maintenance of high α-amylase activities in 24 h culture supernatants. RCW101, in which B. subtilis prsA was expressed from its native promoter, displayed a yield of AmyQ several-fold higher than the strains in which the prsA genes were under xylose regulation. Of the latter strains, RCW302Amy (with Bsu-prsA) and RCW304Amy (with Ban-prsAB) exhibited a two to threefold higher yield of AmyQ than strains RCW303 (with Ban-prsAA) and RCW305 (with Ban-prsAC).
subtilis and B. anthracis prsA genes was similar to that of RCW102 in which the B. subtilis prsA was expressed from its native promoter (Fig. 8). In all strains, rPA production peaked during transition to the stationary phase, but then declined with different kinetics, presumably due to the presence of proteases in the culture medium (20). As was seen for AmyQ, the highest rPA yield was observed when B. subtilis prsA was expressed from its native promoter, in RCW102, while the maximal rPA yields of strains expressing B. subtilis or B. anthracis prsA genes from the P xyl promoter were between two-to three-fold lower. Strains expressing Ban-prsAA and Ban-prsAB produced slightly higher peak yields than those expressing Bsu-prsA or Ban-prsAC. However, as PrsA is required for B. subtilis viability, it is likely that at least one protein essential for cell wall synthesis is PrsA-dependent.
We investigated the role of PrsA in the secretion of B. anthracis PA, a key component of toxins EdTx and LeTx. Initial studies, using the prsA3 mutation that produces ~10% of the wild-type activity, indicated a role for PrsA in rPA production: unlike the wild-type control, no rPA was detected in the culture supernatant of the prsA3 mutant (Fig.1). The importance of PrsA on rPA production was confirmed using a strain in which the level of prsA expression induced from the P spac promoter was controlled by the addition of IPTG. A 5-fold increase in the concentration of IPTG in the medium (0.5 to 2.5 mM) resulted in a 2.5-fold increase in rPA production (Fig. 2); even at 10 mM IPTG, the amount of rPA was only about half that observed when Bsu-prsA was expressed from its native promoter (data not shown). Together these data indicate that the manipulation of PrsA synthesis could be a useful strategy for increasing the production of rPA.
Since rPA production in B. subtilis was PrsA dependent, we were interested to establish whether its native host, B. anthracis, encoded a PrsA homologue. Analysis of the raw DNA sequence data unexpectedly revealed the presence of three PrsA homologues in this bacterium, PrsAA, PrsAB and PrsAC. Using a complementation system in which the synthesis of the native B. subtilis PrsA or a B. anthracis homologue could be controlled independently, we were able to show that all three orthologues were functional in B. subtilis with respect to cell viability and protein secretion.
These observations raise the question as to why B. anthracis produces three PrsA proteins, while B. subtilis produces only one. An explanation might be found in the smooth colonies observed in B. subtilis expressing Ban-prsAA, rather than the rough morphology typical of wild-type B. subtilis expressing Bsu-prsA or one of the constructs expressing Ban-prsAB or Ban-prsAC. The simplest explanation is that Ban-PrsAA is inefficient at folding a specific PrsA-dependent protein required for normal cellular morphogenesis. It also suggests that individual PrsA proteins are specific for, or can distinguish between, different secretory protein substrates. We attempted to test this hypothesis by analysing the efficiency with which each PrsA protein was able to function in the secretion of two PrsA-dependent proteins, namely AmyQ and rPA (Figs. 7, 8). When individual B. anthracis and B. subtilis prsA genes were put under the control of the xylose-inducible P xyl promoter, AmyQ yields were similarly high in strains expressing Bsu-prsA and Ban-prsAB, while yields from strains expressing Ban-prsAA and Ban-prsAC were significantly lower (Fig.7). Interestingly, the PrsA homologue that appeared to be the most effective in mediating high AmyQ yields, PrsAB, is that with least amino acid similarity to B. subtilis PrsA.
We were not able to ascertain whether the amounts of PrsA synthesised by each of these strains was the same, since the only antibodies that were available were to Bsu-PrsA While it is reasonable to assume that homologous genes expressed from the same promoter produce similar amounts of their protein, the variations in AmyQ production may reflect differential amounts of functional PrsA due to factors such as: (a) variations in mRNA halflives of the various prsA genes; (b) variations in the recognition efficiency for the various prsA ribosome binding sites and; (c) differential stability of the various PrsA proteins in B.
subtilis. Alternatively, the various PrsA lipoproteins may exhibit differential substrate specificities with respect to AmyQ.
In a similar set of experiments rPA yields were examined under conditions when the various prsA genes were induced with xylose (Fig.8). rPA levels were generally higher with Ban-prsAA and Ban-prsAC, and lower with Bsu-prsA and Ban-prsAB, although the differences in yields were not as marked as with AmyQ. Interestingly, the kinetics of rPA degradation in strains expressing the various prsA genes followed different patterns, most likely due to their differential effects not only on the folding of rPA, but on that of the host extracellular proteases which are co-produced in stationary phase. The data (Fig. 8 mutants showed a rough, rather than the usual smooth colony morphology and a greatly increased LD 50 compared with wild-type. At least four of these bacterial species, B. anthracis 1 , Staph. aureus, Strept. gordonii and Strept. pneumoniae (22) also encode a second homologue of SecY, namely SecY2, indicating that they have a specialised transporter, SecY2-SecA2, for the export of a subset of secretory proteins. In the case of Strept. gordonii, this secondary transporter is required for the transport of a large serine-rich surface protein (GspB) that contributes to platelet binding. Similarly, B. subtilis encodes five signal peptidases, SipS, SipT, SipU, SipV and SipW (23). Although all are able to process secretory preproteins, only SipS and SipT are essential for viability: viability is maintained in the presence of either SipS or SipT, but not when both are deleted (24,25). The remaining signal peptidases have a minor role in protein secretion. The endoplasmic reticulum (ER)-type SipW, for example, appears to be required for the processing of two spore-associated preproteins, namely pre-TasA and pre-YqxM (26,27). Again these data indicate that, where they occur, paralogous components of secretory pathways are required for the processing of a subset of protein substrates.  5. Activity of the B. anthracis prsA orthologues in B. subtilis with respect to their ability to complement the lethal phenotype of prsA. All strains were derivatives of RCW201 (P spac Bsu-prsA): RCW301 (P spac Bsu-prsA; P xyl ); RCW302 (P spac Bsu-prsA; P xyl Bsu-prsA); RCW303 (P spac Bsu-prsA; P xyl Ban-prsAA); RCW304 (P spac Bsu-prsA; P xyl Ban-prsAB); RCW305 (P spac Bsu-prsA; P xyl Ban-prsAC). Strains were grown in LB broth with 0.1 mM IPTG, washed and then enumerated on LB agar supplemented as follows: unsupplemented (no shading); 1 mM IPTG (diagonal shading); 1mM IPTG and 1% xylose (wavy shading); 1 % xylose (horizontal shading); Colonies observed on unsupplemented LB agar were shown to be IPTG-independent (data not shown).