Optimization of the Cell Wall Microenvironment Allows Increased Production of Recombinant Bacillus anthracis Protective Antigen from B. subtilis

Abstract

The stability of heterologous proteins secreted by gram-positive bacteria is greatly influenced by the microenvironment on the trans side of the cytoplasmic membrane, and secreted heterologous proteins are susceptible to rapid degradation by host cell proteases. In Bacillus subtilis, degradation occurs either as the proteins emerge from the presecretory translocase and prior to folding into their native conformation or after the native conformation has been reached. The former process generally involves membrane- and/or cell wall-bound proteases, while the latter involves proteases that are released into the culture medium. The identification and manipulation of factors that influence the folding of heterologous proteins has the potential to improve the yield of secreted heterologous proteins. Recombinant anthrax protective antigen (rPA) has been used as a model secreted heterologous protein because it is sensitive to proteolytic degradation both before and after folding into its native conformation. This paper describes the influence of the microenvironment on the trans side of the cytoplasmic membrane on the stability of rPA. Specifically, we have determined the influence of net cell wall charge and its modulation by the extent to which the anionic polymer teichoic acid is d-alanylated on the secretion and stability of rPA. The potential role of the dlt operon, responsible for d-alanylation, was investigated using a Bacillus subtilis strain encoding an inducible dlt operon. We show that, in the absence of d-alanylation, the yield of secreted rPA is increased 2.5-fold. The function of d-alanylation and the use of rPA as a model protein are evaluated with respect to the optimization of B. subtilis for the secretion of heterologous proteins.

Bacillus anthracis, a gram-positive spore-forming bacterium, is the causative agent of anthrax. Expression of full virulence is determined by the production of two major virulence factors, the anthrax toxins and a poly-d-glutamic acid capsule, encoded by plasmids pXO1 and pXO2, respectively (14, 29). The products of three genes combine to form the two bipartite toxins of B. anthracis. Protective antigen (PA) and edema factor (EF) combine to form edema toxin, while PA combines with lethal factor (LF) to form lethal toxin.

It has long been known that an extracellular protective antigen is produced when B. anthracis is grown in vivo (6). PA alone is nontoxic and induces a protective immune response (45). The currently licensed vaccines against anthrax in both the United States and United Kingdom are based on precipitated cell extracts of B. anthracis (United States,V770-NP1-R absorbed onto aluminum hydroxide gel; United Kingdom, Sterne 34F₂ precipitated by alum) (15). Although the Sterne United Kingdom vaccine strain is attenuated, it is still categorized as an Advisory Committee on Dangerous Pathogens (ACDP) category III pathogen; therefore, vaccine production requires the use of expensive high-level containment facilities. Besides PA, these vaccines contain trace amounts of EF and LF (the other potentially toxic components) and a variety of medium-derived proteins, any of which may contribute to the short-lived side effects recorded in some individuals following vaccination (47).

To overcome these problems, attempts have been made to produce recombinant protective antigen (rPA) in a variety of heterologous hosts, including B. subtilis (20), Escherichia coli (39), Salmonella enterica serovar Typhimurium (12), Lactobacillus casei (54), and viral hosts, including baculovirus and vaccinia virus (19). Of the hosts tested so far, only B. subtilis has been found to give yields of rPA that are compatible with large-scale production (3, 4, 5).

Bacillus species in general, and B. subtilis in particular, are highly valued commercially for their ability to secrete proteins directly into the culture supernatant at gram-per-liter concentrations (13). The high secretion capacity, coupled with the ease of fermentation and impressive safety record, has identified B. subtilis as an important host for efficient large-scale secretion of a wide variety of heterologous proteins.

Secretion-competent preproteins are actively transported to and across the cytoplasmic membrane by the translocase complex, utilizing energy from ATP and the proton motive force (40, 46). Following translocation, signal peptidases process the preprotein, mediating the cleavage of the N-terminal signal peptide and the release of the unfolded mature protein on the trans side of the membrane. Many heterologous proteins are subject to rapid degradation by host cell-associated proteases before they are able to fold into their mature, usually more proteolytically resistant, fully folded conformation (20, 27, 43). Consequently, the speed and efficiency of in vivo folding of secreted proteins directly affect their overall yield.

B. subtilis secretes seven extracellular proteases into the culture medium. Native secretory proteins which have coevolved alongside these proteases are, for the most part, resistant to their proteolytic activities. In contrast, heterologous proteins are often highly susceptible to these proteases, prompting attempts to engineer multiply-protease-deficient strains of B. subtilis, such as WB600 (53). However, despite the fact that six extracellular protease genes have been inactivated in this strain, so that it exhibits just 0.32% of the proteolytic activity of the wild type, the proteolytic degradation of heterologous proteins has not been eliminated (28, 42). Even proteins that are highly resistant to proteases in their fully folded state can be rapidly and substantially degraded following translocation, suggesting the existence of a degradation pathway associated with the cell wall. This activity has been attributed, in part at least, to the product of the wprA gene, which encodes a cell wall-associated serine protease (26). Consequently, a B. subtilis strain that was not able to express its wprA gene exhibited twice the α-amylase activity of the wild-type strain (43).

The identification of posttranslocational factors that participate in protein folding may facilitate the optimization of heterologous protein secretion. Factors known to influence the folding of secreted proteins include extracellular folding factors (e.g., PrsA) and the physicochemistry of the cell wall. PrsA is an extracytoplasmic molecular chaperone located on the trans side of the cytoplasmic membrane which functions as a peptidyl-prolyl-cis-trans-isomerase (PPI). PrsA catalyzes the conversion of proline residues from the cis configuration into the trans configuration. This is the rate-limiting step for the folding of many secretory proteins (50). Overproduction of PrsA has been shown to increase the production of α-amylase, proteases, and single-chain antibody fragments (23, 50, 52).

The efficient folding of a number of secretory proteins is dependent on the presence of divalent metal cations Fe³⁺ or Ca²⁺ (10, 25, 34, 44). The importance of Ca²⁺ to rPA stability has been shown by growing B. subtilis in minimal medium with and without CaCl₂ (10 mg/liter); in the absence of Ca²⁺, little if any rPA was detected by Western blotting (Leslie Baillie, personal communication). The availability of metal cations to act as folding effectors is determined by their concentration in the environment and the properties of the anionic polymers (teichoic and teichuronic acids) covalently attached to peptidoglycan in the cell wall.

Teichoic acids, including wall teichoic acids (WTA) and lipoteichoic acids (LTA), are essential components of the cation exchange system between the cell envelope and the surrounding environment. They confer a high density of negative charge on the cell wall and the capacity to attract metal cations such as Ca²⁺, Fe³⁺, and Mg²⁺ (7, 17, 34). The ion exchange activities of teichoic acids are mediated by the presence of ester-linked d-alanyl residues that control the overall negative charge of the cell wall (16). The free amino moiety of each d-alanyl residue is protonated at physiological pH, effectively neutralizing the negative charge of a neighboring phosphate residue on the teichoic acid molecule. Significantly, when extracellular cation concentrations are high, the effects of d-alanylation are marginal, whereas at low cation concentrations cell wall binding is significantly increased in the absence of alanylation (24).

The d-alanylation of LTA and WTA in B. subtilis is determined by the products of the dlt operon. This operon is composed of five genes, dltA to dltE. Insertional inactivation of genes dltA to dltD results in the complete inhibition of d-alanylation (32) but does not affect cell growth or metabolism. However, these mutants are marginally more vulnerable to cell lysis (51). Recent work has shown that inactivation of dltB or dltD markedly increases the stability of certain secretory proteins, especially those prone to proteolysis (18).

As a model secretory protein, B. anthracis protective antigen has advantages over other model proteins, such as α-amylases, in that it is subject to proteolysis at both cell-associated and extracellular locations. In this paper, we show that the degree of cell wall alanylation influences the secretion of rPA by B. subtilis. In particular, we show the kinetics of rPA secretion and the relative contributions of proteolytic events that occur before and after this protein is released from the cell. The analysis and optimization of rPA secretion are likely to provide insights into the key features of the B. subtilis secretion pathway, which could be adapted to improve the yield of secreted heterologous proteins.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains and plasmids are listed in Table 1. All strains were maintained on antibiotic medium 3 plates (Difco Laboratories Ltd., East Molesley, United Kingdom) or in liquid culture in 2xYT broth (1.6% tryptone, 1% yeast extract, 0.5% NaCl) buffered with 0.2 M MES (morpholineethanesulfonic acid, pH 6.0). Pulse-chase labeling experiments were carried out in Spizizen’s minimal salts medium (SMS) (41). For plasmid maintenance, growth medium contained ampicillin (100 μg/ml), erythromycin (1 μg/ml), kanamycin (5 μg/ml), and lincomycin (25 μg/ml), as necessary.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Description	Comments	Reference or source
Strains
B. subtilis 168	trpC2	Wild-type B. subtilis	1
B. subtilis DN1885	amyE amyR2	α-Amylase-negative derivative of B. subtilis 168	Novo Nordisk A/S
B. subtilis WB600	trpC2 ΔnprE ΔaprE Δepr Δbpf Δmpr ΔnprB	B. subtilis with six extracellular protease genes knocked out	53
B. subtilis JET1	amyE, pPA101, Kmr	DN1885 with rPA plasmid pPA101	This study
B. subtilis JET2	amyE, pPA101, IPTG-inducible dltA-E; Kmr Emr Lmr	DN1885 JET1 with pMUTin4dltAFP integrated at dltA	This study
E. coli XL1-Blue	recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F′-proAB lacIqZ M15ΔTn10 (Tctr)]		Stratagene
Plasmids
pPA101	rPA; Kmr	Derivative of pUB110 containing recomibinant pagA	20
pMUTin4	Ampr Emr LmrspoVG-lacZ Pspac		48
pMUTin4dltAFP	Kmr Ampr Emr Lmr	Integrated into the B. subtilis chromosome at dltA	This study

Stability of rPA and AmyL in spent culture medium.

Cultures of B. subtilis 168 and B. subtilis WB600 were grown for 24 h in 20 ml of 2xYT to allow the full expression of extracellular proteases. The cells were removed by centrifugation (10,000 × g, 30 min, 4°C), and the spent medium containing extracellular proteases was filtered through 0.45-μm-pore Acrodisc-32 filters (Pall Gelman Sciences, Ann Arbor, Mich.). Aliquots (5 μl of 1 mg/ml) of purified rPA (31) and AmyL (42) were added to 300 μl of culture supernatant and incubated at 37°C. Samples (20 μl) were removed at intervals over a 36-h period to determine rPA and AmyL levels by Western blotting.

DNA manipulations and PCR.

Plasmid DNA extraction, restriction endonuclease digestion, ligation, agarose gel electrophoresis, and transformation of electrocompetent E. coli cells were carried out as described previously (36). Enzymes, molecular size markers, and deoxynucleotides were purchased from Roche Diagnostics Ltd. (Lewes, United Kingdom) or Amersham Pharmacia Biotech Ltd. (Little Chalfont, United Kingdom). Extraction of B. subtilis DNA and transformation of B. subtilis by the Groningen method were done according to Bron (9). The PCR was carried out as described previously (37) with Pfu (Stratagene Europe, Amsterdam, The Netherlands) or Taq DNA polymerase (Promega UK Ltd., Southampton, United Kingdom), using the following cycling program: 1 cycle of 5 min at 94°C and then 35 cycles of 1 min at 94°C, 1 min at 60°C, and 1 to 4 min at 72°C, depending on the size of the PCR product (36).

Plasmid pJET1 was created by amplification of the 5′ end of the dltA gene by PCR, using primers DLT-F (5′-CGCGGATCCGGAGAGAGAATAACTATGAAAC-3′) and DLT-R (5′-CGCGGATCCTTCGTTTTCCAGCAGTTCTTCC-3′). These primers were designed to include BamHI restriction sites (shown in bold) to facilitate cloning. The recombinant plasmids were transformed into competent B. subtilis cells, and transformants were selected on antibiotic medium 3 plates containing erythromycin and lincomycin. The mutants were analyzed by PCR to confirm the integration of a single copy of the plasmid into the target gene on the chromosome, using a strategy similar to that described previously (37).

SDS-PAGE and Western blotting.

Cultures were grown in 20 ml of 2xYT broth buffered with 0.2 M MES (pH 6.0) at 37°C. Samples of culture supernatant (20 μl), taken hourly throughout growth, were mixed with 20 μl of 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer and then boiled for 5 min. The samples were run on a 10% polyacrylamide gel alongside prestained low-range molecular size markers (Bio-Rad, Hercules, Calif.). Proteins were transferred from the gel onto a 0.4-μm nitrocellulose membrane (Anderman Ltd., Kingston upon Thames, United Kingdom). rPA protein was detected using rabbit polyclonal anti-PA antiserum followed by anti-rabbit immunoglobulin-horseradish peroxidase conjugate (Dako Immunochemicals A/S, Glostrup, Denmark). Protein bands were visualized by immersion in freshly prepared Western blot staining solution containing 4-chloro-1-naphthol and hydrogen peroxide.

Analysis of d-alanine esters of wall teichoic acids.

Cultures were grown in 20 ml of 2xYT broth buffered with 0.2 M MES (pH 6.0) at 37°C. Culture samples (1 ml) were removed at hourly intervals and centrifuged, and the resulting pellet was stored at −20°C. Cell disruption, alkaline hydrolysis of d-alanine from WTA, and colorimetric assays to determine liberated d-alanine concentrations were performed as described previously (18).

Pulse-chase experiments.

Bacterial cultures were grown to exponential phase in SMS, indicated by an optical density at 660 nm of 1.0. The culture was pulse-labeled with l-[³⁵S]methionine (561 kBq/μl; Amersham International Plc.), as described previously (43, 49). Following the chase, pairs of samples (750 μl) were taken at various times (0, 2, 5, 7, 10, 15, 30, 45, and 60 min). The first sample, containing cells and growth medium, represented total protein, while the second sample, obtained by filtration through a 0.45-μm-pore polyvinylidene difluoride filter (Whatman, Maidstone, United Kingdom), represented protein secreted into the culture medium. rPA in the samples was immunoprecipitated with rPA-specific antibodies, and the immune complexes were recovered by centrifugation. These were subjected to SDS-PAGE and analyzed by fluorography. The relative amounts of radiolabeled rPA at each time point were determined by phosphoimaging (PhosphorImager; Molecular Dynamics) and analyzed using ImageQuant software (version 3.22; Molecular Dynamics).

β-Galactosidase activity.

Cultures were grown in 20 ml of 2xYT broth at 37°C. Samples were removed to determine optical density at 660 nm and β-galactosidase activity as described previously (30).

Surface charge topology mapping.

A three-dimensional model of the surface topology of rPA was determined using the published crystallographic protein structure data of anthrax protective antigen (35) (Swiss-Prot accession number P13423). Loop regions for which the structure could not be determined were omitted, as well as the two bound Ca²⁺ ions. A three-dimensional solvent-accessible contour map of rPA was created using the Sybyl 6.7 program (Tripos Ltd., St. Louis, Mo.) using the Connolly surface method (11). The spatial distribution of electrostatic potential was determined using the Poisson-Boltzmann equation.

RESULTS

Stability of rPA in spent culture medium.

The onset of stationary phase triggers the synthesis by B. subtilis of a number of extracellular proteases (33). We therefore determined the effects of extracellular proteases released into the culture medium on the stability of rPA. Stability was determined in spent culture medium used for the growth of the wild-type B. subtilis 168 and derivative WB600, which is deficient in six extracellular proteases (53). A protease-resistant heterologous protein, Bacillus licheniformis α-amylase (AmyL), was included for comparison (43).

In the presence of spent culture medium from the wild type, rPA had a half-life of approximately 50 min (Fig. 1). In spent medium from the protease-deficient strain WB600, the half-life was extended to 12 h. The rapid proteolysis in the B. subtilis 168 supernatant is typical of a wide variety of heterologous proteins secreted by B. subtilis. AmyL, however, was markedly more stable; in the presence of wild-type levels of proteases, 80% of the initial protein was still present after 35 h, and this was increased to 90% in spent medium from the protease-deficient strain.

FIG. 1.

Stability of rPA (open symbols) and AmyL (closed symbols) in spent culture medium from B. subtilis 168 (squares) and WB600 (circles). The proteins were detected by Western blotting and are expressed as a percentage of the initial protein concentration.

Construction of a mutant with an inducible dlt operon.

To determine the effects of d-alanylation on the secretion and stability of rPA, we constructed a strain of B. subtilis in which the entire dlt operon was placed under the control of an isopropylthio-β-d-galactopyranoside (IPTG)-inducible P_spac promoter (Fig. 2). A 419-bp fragment from the 5′ end of the dltA gene was amplified from B. subtilis 168 chromosomal DNA using oligonucleotide primers DLT-F and DLT-R. The product was digested with BamHI, cloned in the pMUTin4 integrational vector (48), and used to transform E. coli XL1-Blue (Stratagene, La Jolla, Calif.). Diagnostic PCR, using vector- and insert-specific primers, was used to screen for the presence and correct orientation of the insert in the recombinant plasmids, and their structure was confirmed by sequencing the entire insert and vector sequences on either side of the cloning site (data not shown).

FIG. 2.

Construction of an IPTG-inducible dlt operon in B. subtilis DN1885 by cloning a 5′ fragment of the dltA gene into the B. subtilis integration vector, pMUTin4. The open flag represents the native dlt promoter, P_dlt, while the IPTG-inducible P_spac promoter is depicted by a solid flag.

The resulting plasmid, pMUTin4dltAFP, was used to transform B. subtilis JET1 to generate B. subtilis JET2 dltA::pMUTin4. The recombinant strain was tested by PCR to verify the integrity of the construct. In this strain the expression of the dlt operon was under the control of the P_spac promoter, allowing the expression of the operon to be induced by the addition of IPTG to the growth medium (48).

Influence of the dlt operon on WTA d-alanylation and rPA secretion.

B. subtilis 168 and JET2 dltA::pMUTin4 were grown at 37°C in 2xYT, the latter with or without IPTG, until the onset of stationary phase. Samples were taken to monitor growth, the extent of WTA d-alanylation, and the yield of secreted rPA. All of the strains exhibited similar growth kinetics (Fig. 3A), confirming that the dlt operon has no obvious effects on cell growth (32).

FIG. 3.

Phenotypic effects of dlt operon modulation using B. subtilis 168 (▪) and B. subtilis JET2 dltA::pMUTin4 with (•) and without (⧫) 1 mM IPTG. (A) Growth kinetics in 2xYT-0.2 M MES (pH 6.5) determined at OD₆₆₀. (B) d-alanyl ester content of isolated wall teichoic acids, detected by a colorimetric reaction. (C) Extracellular secretion of mature rPA into the culture medium, detected via Western blotting.

The d-alanylation of WTA was measured to determine the activity of the dlt operon (Fig. 3B). In wild-type B. subtilis, the extent of d-alanylation increased during transition to stationary phase and then decreased during stationary phase. If the pH of the culture medium is increased to 8.1, the ester bond linking the d-alanyl residue to teichoic acid becomes labile (22) and as a result, no d-alanylated WTA could be detected (data not shown). Analysis of the non-IPTG-induced JET2 dltA::pMUTin4 showed, as expected, an absence of d-alanine residues in WTA. In contrast, alanylation was partially restored when 1 mM IPTG was added to the culture medium. However, the level of alanylation in the induced mutant (≈1 mg of alanine/unit of culture OD) was only about one fifth of that observed in the wild type. Increasing the concentration of IPTG to 10 mM did not restore the level of alanylation to that of the wild type (data not shown). The failure to restore the level of alanylation in the JET2 dltA::pMUTin4 to that of the wild type probably reflects the fact that the operator in this construct has been optimized to achieve maximum repression under noninducing conditions, with consequential limitations to the extent to which the P_spac promoter can be induced (48).

The secretion of rPA was determined by Western blotting (Fig. 3C). Wild-type B. subtilis secreted rPA to a maximum concentration of 2 mg/liter, while that of the noninduced B. subtilis JET2 dltA::pMUTin4 was 4.75 mg/liter. Induction of the dlt operon in the mutant with 1 mM IPTG resulted in the synthesis of an intermediate concentration of rPA (3 mg/liter), which was compatible with the degree of alanylation observed in this strain.

Transcriptional activity of the dlt operon.

The activity of the native dlt promoter was analyzed in B. subtilis JET2 dltA::pMUTin4, in which the native promoter was transcriptionally fused to a lacZ reporter gene, enabling the transcriptional activity to be monitored via β-galactosidase activity (Fig. 4). Transcription of the dlt operon remained low during exponential growth but, in agreement with previous reports (32), was transiently induced during transition into stationary phase. These data are consistent with the observed levels of WTA alanylation (Fig. 3B). The transcription of the native dlt promoter was independent of the presence of IPTG, indicating that this operon is not autoregulated by products of the dlt operon.

FIG. 4.

β-Galactosidase assay for transcriptional activity of the dlt operon using a dltA::lacZ transcriptional fusion. Growth (solid symbols) and β-galactosidase activity (open symbols) were measured in cultures of B. subtilis 168 (squares) and B. subtilis JET2 dltA::pMUTin4 with (diamonds) and without (circles) 1 mM IPTG.

Pulse-chase analysis of rPA processing.

The kinetics of rPA secretion were determined by coupled pulse-chase/immunoprecipitation experiments. Wild-type B. subtilis and JET2 dltA::pMUTin4, with and without 1 mM IPTG, were grown in Spizizen’ exis minimal medium to exponential phase (OD₆₆₀ of approximately 1) and pulse-chased with l-[³⁵S]methionine. Radioactively labeled rPA was immunoprecipitated, separated by SDS-PAGE, and visualized by autoradiography (Fig. 5A). The precursor form of rPA (mature protein plus signal peptide), with a molecular mass of 85.8 kDa, was clearly distinguishable from that of the mature form (82.7 kDa).

FIG. 5.

Secretion of rPA from exponentially growing cultures of B. subtilis in the presence and absence of the products of the dlt operon. Representative data are shown for pulse-chase experiments carried out on strains B. subtilis 168 and B. subtilis JET2 dltA::pMUTin4 with and without 1 mM IPTG. (A) Autoradiographs of pulse-chased rPA following immunoprecipitation with anti-rPA polyclonal antibodies and SDS-PAGE. The top gels in panel A indicate precursor (p) and mature (m) rPA immunoprecipitated from whole-culture samples, and the bottom gels indicate mature rPA released into the culture medium. (B) Quantification by phosphoimaging of the different forms of rPA at different times following the chase. ▪, rPA precursor; ⧫, mature rPA in whole-culture samples; •, mature rPA released into the growth medium. The amount of each form of rPA is expressed as a percentage of the total rPA (precursor plus mature) synthesized during the pulse.

In the wild type, the processing of rPA from precursor to mature form (Fig 5A) was relatively slow compared to other secreted Bacillus proteins (e.g., AmyL) (43). Following the addition of the nonradioactive chase (T₀), only 50% of the radioactively labeled rPA was in the mature form. The maximal amount of labeled mature rPA was recorded approximately 5 min following the chase, with the labeled precursor form disappearing after approximately 10 min. The amount of mature rPA decreased rapidly; at 30 min postchase, half had been degraded, and virtually no mature rPA was detected after 60 min (Fig. 5A). The secreted mature rPA, representing 20% of the maximum synthesized during the pulse, reached a peak at 7 min and was completely degraded after approximately 40 min. These data confirm that rPA is subject to extensive degradation during or shortly after translocation through the cytoplasmic membrane and after secretion into the growth medium.

Processing of precursor rPA in B. subtilis JET2 dltA::pMUTin4 (Fig. 5B and 5C) appeared to be slower than that of the wild type, and significant amounts of the precursor form were detected 15 min after the pulse. In the absence of induction (i.e., no alanylation), the maximal level of labeled mature rPA was detected approximately 15 min after the chase and remained relatively stable for the next 30 min. Even at 60 min postchase, significant amounts of rPA were detectable (Fig. 5B), both in whole culture samples and secreted into the culture medium.

Induction of the dlt operon in B. subtilis JET2 dltA::pMUTin4 (Fig 5C) reduced the stability of rPA to a level that was intermediate between that of the wild-type and the noninduced mutant strain (Fig 5B). Processing of the precursor protein into mature protein occurred at an intermediate rate. Mature rPA peaked after 7 min at 50% of the total synthesized during the pulse (cf. B. subtilis 168, 15%, and B. subtilis JET2 dltA::pMUTin4 without IPTG, 60%). Similarly, the secretion of rPA into the culture medium peaked at 30% of the maximum at 7 min, and it was degraded relatively slowly thereafter. In both the wild type and the mutant, there was a consistent difference between the amount of mature rPA detected in the whole culture and that secreted into the culture medium. This difference is likely to be accounted for by rPA trapped in the matrix of the cell wall.

Characterization of the surface topology of rPA.

Protective antigen from B. anthracis has a calculated pI of 5.58. In order to determine the distribution of electrostatic potential on the surface of rPA, a three-dimensional solvent-accessible model was generated. The model shows clear variations in charge distribution; the surface is predominantly negatively charged, but with the strongly positively and negatively charged regions projecting from the main body of the protein. An exception is a concentration of negatively charged residues in domain 1 (35) and a strongly electropositive region corresponding to the convergence of three lysine residues.

The distribution of surface charge is likely to influence the interaction of rPA with the negatively charged cell wall of B. subtilis during secretion. The negatively charged face of rPA is likely to be repelled by the cell wall, whereas the positively charged projections may facilitate cell wall binding (44). The latter may be responsible for the discrepancy between the amount of rPA synthesized and that released into the culture medium (Fig. 5).

DISCUSSION

Obtaining high yields of secreted rPA from B. subtilis has been hampered by marked proteolytic degradation (5). Attempts to eliminate this degradation using B. subtilis WB600, a protease-deficient strain in which six out of seven extracellular protease genes have been inactivated (4, 53), have only been partially successful. Here we report a 15-fold increase in the half-life (from 0.8 to 12 h) of rPA in spent culture medium from a protease-deficient strain compared with the wild type. This was in marked contrast to B. licheniformis α-amylase, which was stable over extended periods, even in spent culture medium derived from the wild type. rPA therefore represents a useful model for proteins that are subject to rapid proteolysis by B. subtilis extracellular proteases.

The recent observation that incompletely folded secretory proteins are more susceptible to proteolysis that their natively folded counterparts has focused attention on the identification of cotranslocational and posttranslocational events that influence the secretion of proteins from B. subtilis (8, 18, 43). Proteins may be most susceptible to proteolysis as they emerge from the translocase, and factors that increase the rate at which they achieve their native state are likely to promote protein stability and hence increase the overall recoverable yields. Thus, characterization of this environment and factors involved in efficient protein folding may provide insights for the optimization of heterologous proteins.

The microenvironment on the trans side of the cytoplasmic membrane into which proteins are secreted is controlled and maintained by the physicochemical properties of the membrane and the cell wall. The net charge in this region is modulated by the extent of d-alanylation of wall and lipoteichoic acids, which in turn affect the concentration of metal cations such as Ca²⁺, Fe²⁺, and Mg²⁺ at the cell membrane/wall microenvironment (7, 17, 18). These cations are required as folding factors for a variety of secreted proteins, including α-amylase and levansucrase (10, 25, 34, 44). The presence of calcium ions in the native structure of anthrax protective antigen means that metal ions are also likely to be important for the folding of this protein (35). In the absence of d-alanylation, the negative charge density in the wall increases, as does its affinity for metal cations (24). This in turn increases the availability of these cations to act as folding effectors and hence increases the rate of protein folding (18).

In contrast to previous studies in which the dltB and dltD genes have been insertionally inactivated (18), we have constructed a mutant in which the expression of the dlt operon has been placed under the control of the IPTG-inducible P_spac promoter (48). In comparison to the wild type, the yield of secreted rPA, as detected by Western blotting, was significantly increased in the noninduced mutant.

Analysis of rPA secretion kinetics by pulse-chase showed that rPA was stabilized in the absence of d-alanylation, leading to the release of increased amounts of mature rPA secreted into the culture medium. However, there was a consistent difference between the total amount of rPA synthesized during the pulse and the amounts released into the culture medium. This difference probably represents cell-associated degradation by wall- and/or membrane-associated proteases occurring shortly after translocation of the protein across the cytoplasmic membrane. In addition, the rate of precursor processing and cell-associated degradation is higher in the wild type than in the nonalanylated mutant.

The pulse-chase experiments also highlight a discrepancy, of approximately 10%, between the amount of mature rPA in the whole culture and that secreted into the culture medium. This may represent an interaction between a portion of the rPA protein and the cell wall. Physical-chemical interactions between the cell wall and secreted proteins have been suggested to inhibit the release of certain heterologous proteins (38). Although there is no definitive in vivo evidence for such interactions, except for native wall-associated proteins (2), studies using chimeric α-amylases with modified net charge have shown that positively charged proteins may be retarded by electrostatic interactions with the cell wall (21, 44). Analysis of the surface charge topology of rPA indicated that one face was negatively charged, while the opposing face had a number of positively charged projections that may facilitate interactions with the cell wall.

The use of B. subtilis as a host provides many advantages over the currently licensed method for the production of anthrax protective antigen. However, rPA, like many other heterologous proteins, is subjected to proteolytic attack by host cell proteases. The use of rPA as a model protein for investigating cell-associated and extracellular factors affecting protein stability is likely to prove valuable, not only for the production of the next generation of anthrax vaccines, but also for the secretion of other heterologous proteins of commercial and medical importance from B. subtilis.