Functional Analysis of the Carboxy-Terminal Domain of Bacillus anthracis Protective Antigen

Abstract

Protective antigen (PA) is the common receptor-binding component of the two anthrax toxins. We investigated the involvement of the PA carboxy-terminal domain in the interaction of the protein with cells. A deletion resulting in removal of the entire carboxy-terminal domain of PA (PA608) or part of an exposed loop of 19 amino acids (703 to 722) present within this domain was introduced into the pag gene. PA608 did not induce the lethal-factor (LF)-mediated cytotoxic effect on macrophages because it did not bind to the receptor. In contrast, PA711- and PA705-harboring lethal toxins (9- and 16-amino-acid deletions in the loop, starting after positions 711 and 705, respectively) were 10 times less cytotoxic than wild-type PA. After cleavage by trypsin, the mutant PA proteins formed heptamers and bound LF. The capacity of PA711 and PA705 to interact with cells was 1/10 that of wild-type PA. In conclusion, truncation of the carboxy-terminal domain or deletions in the exposed loop resulted in PA that was less cytotoxic or nontoxic because the mutated proteins did not efficiently bind to the receptor.

Bacillus anthracis, a spore-forming bacterium, is the causative agent of anthrax. The two exotoxins are main virulence factors of the microorganism. These toxins are composed of three proteins: the protective antigen (PA, 83 kDa), the lethal factor (LF, 85 kDa), and the edema factor (EF, 89 kDa) (7, 15). Intravenous injection of the lethal toxin (PA plus LF) causes sudden death in animals (1). The edema toxin (PA plus EF) causes edema at the inoculation site (29). The components PA, LF, and EF are encoded by the pag, lef, and cya genes, respectively, and these genes are carried by virulence plasmid pXO1 (185 kbp) (3, 17, 20, 32). A mode of action has been proposed for anthrax toxins (13). PA binds to a ubiquitous proteinaceous cell receptor which has yet to be identified (6). It is then cleaved by a furin-like protease into PA63 and PA20 (a 20-kDa amino-terminal fragment) (11, 26). This processing facilitates the heptamerization of PA63 (19) and the subsequent binding of EF or LF. These toxic complexes are internalized by receptor-mediated endocytosis. The pH within the acidic vesicles decreases during intracellular trafficking, resulting in the insertion of PA into the membrane and the formation of a channel (2, 12, 18). EF and LF are further translocated into the cytoplasm to exert their catalytic effects. EF is a calmodulin-dependent adenylate cyclase (14). LF is a zinc metalloprotease which cleaves mitogen-activated protein kinase kinases 1 and 2 (5, 10, 31). The lethal toxin induces the lysis of macrophage cell lines such as RAW264.7 (8). PA is therefore a key protein, promoting both the binding of the toxins and the translocation of their enzymatic moieties (27, 33). Previous studies have suggested that the carboxy-terminal extremity of PA is involved in the recognition of the cell receptor (16, 28). The three-dimensional structure of the monomeric PA has been solved at 2.1 Å resolution (21) and consists of four folding domains. Domain 4 of PA encompasses the last 139 carboxy-terminal amino acids (596 to 735). We analyzed the function of domain 4 in toxicity by constructing various deletions in the pag gene, based on the structural organization of the molecule. The cytotoxicity of the resulting protein products in the presence of LF and their interaction with the receptor were tested.

Construction of carboxy-terminal domain mutant PA.

Domain 4 of PA contains an exposed loop of 19 amino acids which begins at position 703. Mutations affecting the whole of domain 4 of PA or the exposed loop were produced (Fig. 1). We thereby created stop codons after the sequences encoding arginine 592 (PA592) or aspartate 608 (PA608), removing the whole of domain 4 (PA592) or retaining the first 16 amino acids of this domain (PA608). Two short, in-frame deletions resulting in the removal of 9 (PA711) or 16 (PA705) amino acids from the loop were created after the nucleotides encoding residues 711 and 705, respectively. Mutants PA proteins were obtained by site-directed mutagenesis with a PCR. Plasmid pACP41 was used as the template for the PCR (23). The divergent sequences of the primers used were 5′-AAGAAAATTTTAATCTTTTCTAAAAAAGGC-3′ and 5′-GTTTTCTTTAGTAACAGCATATACATTTAC-3′ for PA705, 5′-ACTAGGATTAATAATAGTGTTTTCTTTAG-3′ and 5′-ATCAAGAAAATTTTAATCTTTTCTAAAAAAGGC-3′ for PA711, and 5′-CTCACTAATCCGCCCCAACTGCTATG-3′ and 5′-GGCTATGAGATAGGATAAGGTAATTCT-3′ for PA608. For the construction of PA592, a stop codon was introduced by using a site-directed mutagenesis kit (Stratagene, La Jolla, Calif.). The mutated pag genes, cloned into shuttle plasmid pAT28 (30), were transferred by heterogramic mating as previously described (22) into PA-deficient B. anthracis RP31 (23), and the production of PA-related proteins was tested. The three mutated proteins, PA608, PA705, and PA711, were purified by fast protein liquid chromatography as previously described (23) from the supernatants of B. anthracis transconjugants grown in R medium (24) containing tryptone (5 g/liter). The molecular masses of the corresponding proteins were as expected (approximately 83 kDa for PA711 and PA705 and 70 kDa for PA608) (Fig. 2). In contrast, PA592 was not found in the supernatants of B. anthracis, suggesting that the amino acids between asparagine 592 and aspartate 608 are required for stability of the molecule.

FIG. 1.

Schematic representation of the mutagenesis of domain 4 of PA. The three-dimensional structure of the carboxy-terminal domain of PA (amino acids 592 to 735), according to Petosa et al. (21), is shown. We truncated the PA molecule by introducing a stop codon after the codon encoding arginine 592 or aspartate 608. Two in-frame deletions resulting in the removal of 9 and 16 amino acids (residues 711 to 721 and 705 to 722, respectively) were also created in the exposed loop of 19 amino acids (residues 703 to 722). Point mutations are indicated by dashed arrows; deletions are indicated by plain arrows.

FIG. 2.

Trypsin treatment of mutant PA. Wild-type and mutant PA proteins were purified from B. anthracis supernatants and studied in native form (−) or after (+) cleavage by trypsin. Samples (2 μg) were subjected to electrophoresis in a sodium dodecyl sulfate–12% polyacrylamide gel and stained with Coomassie blue. The values on the left are molecular masses in kilodaltons.

Cytotoxicity of mutant PA in the presence of LF.

We tested the ability of PA mutant proteins to induce LF-mediated lysis of murine macrophage cell line RAW264.7 (Fig. 3). A 20-ng/ml concentration of wild-type PA was required for lysis of half of the macrophages (50% effective concentration [EC₅₀]). PA608 was completely inactive against macrophages (EC₅₀, >2 × 10⁴ ng/ml). PA711 and PA705 were mildly toxic, with EC₅₀s of 2 × 10² and 6 × 10² ng/ml, respectively (EC₅₀ of PA, 20 ng/ml). To determine which step of the intoxication process was affected, we analyzed the interaction of the mutant PA with LF and with the cell receptor.

FIG. 3.

Cytotoxicity assay for mutant PA. Wild-type or mutated PA protein (from 1 × 2 × 10⁴ to 2 × 10⁴ ng/ml) was serially diluted into a 96-well plate containing RAW264.7 cells in the presence of a constant concentration of LF (2 × 10⁴ ng/ml). The cells were incubated for 3 h, and their viability was assessed by using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (9). The experiment was carried out at least three times for each mutant PA protein.

Interaction of mutant PA with LF.

The interaction between PA and LF requires the proteolytic processing of PA by the furin-like protease. In vitro, trypsin mimics this action, cleaving PA into PA20 and PA63. Treatment of PA705, PA711, and PA608 with trypsin (600:1) for 30 min at 30°C released PA20 and a PA63 mutant polypeptide of the expected size (Fig. 2). Thus, the trypsin-sensitive site was exposed in the mutant molecules. Moreover, PA705 and PA711 formed heptamers after processing, as previously observed for wild-type PA (19) (data not shown). An assay was developed to detect the binding of LF to the mutant PA molecules. Purified LF protein (1 μg per well) was subjected to electrophoresis in a nondenaturing polyacrylamide gel and transferred to a nitrocellulose membrane (Hybond-C; Amersham, Buckinghamshire, England). Wild-type or mutated PA proteins (2 μg/ml), freshly treated with trypsin, were incubated with the LF blots for 16 h at 4°C. The PA63-LF complex was detected by Western blotting using rabbit anti-PA serum, followed by the addition of goat anti-immunoglobulin G coupled to peroxidase (ECL kit; Amersham). Only the trypsin-treated forms of the PA proteins recognized LF on the membrane (Fig. 4). A double signal was routinely observed which corresponded to the two isoforms of LF. The three mutant proteins interacted with LF in a manner similar to that of wild-type PA. Thus, (i) the conformation of the LF-binding region of PA was not significantly changed and (ii) domain 4 of PA was not required for the binding of LF, consistent with previous reports (16, 28).

FIG. 4.

Interaction of mutant PA proteins with LF. Purified LF protein was subjected to electrophoresis in a 10% nondenaturing polyacrylamide gel and transferred to a nitrocellulose membrane under nondenaturing conditions. Wild-type or mutated PA proteins (2 μg), untreated (−) or treated (+) with trypsin, were added. The filter was incubated for 16 h at 4°C, and the LF-PA63 complex was detected with anti-PA serum.

Binding of mutant PA to cells.

The binding of wild-type and mutant PA proteins to cells was tested on CHO-K1 cells, each of which has approximately 10,000 PA receptor molecules on its surface (6). An enzyme-linked immunosorbent assay was used to determine protein binding. The CHO-K1 cell line was seeded into 24-well plates (Costar) containing glass slides at a density of approximately 10⁵ cells per well. The cells were incubated for 16 h at 37°C in a 5% CO₂–95% air atmosphere, cooled for 15 min at 4°C, washed with cold phosphate-buffered saline (PBS), and incubated for 3 h at 4°C in RPMI 1640 with wild-type or mutant PA protein. Unbound proteins were removed by washing the cells with cold PBS. Cells were then fixed and then incubated with anti-PA serum (1/4,000) for 45 min at 37°C in PBS-skim milk powder (2%). After washing, goat anti-immunoglobulin G coupled to β-galactosidase (1/10,000) was added and incubated with cells in PBS-skim milk powder (2%) for 45 min at 37°C. The glass slides were washed with PBS and incubated for 30 min at 37°C with methylumbelliferyl-β-galactoside (MUG). The enzymatic reaction was stopped by addition of 100 mM glycine (pH 10.4). The hydrolysis of MUG by the β-galactosidase resulted in the formation of a fluorescent component. The level of fluorescence associated with cells was assessed on a fluoroscan fluorimeter (Titertek-Fluoroskan, Labsystem).

The threshold for detection of wild-type PA was about 10 ng/ml, and the receptors were saturated at a concentration of 5 μg/ml (Fig. 5). The truncated protein, PA608, did not bind to cells, even at high concentrations (10⁴ ng/ml), and the level of binding of PA711 and PA705 to cells was significantly lower (1/10) than that of the wild-type protein. Therefore, the low cytotoxicity of domain 4 mutant PA is due to a defect in the receptor interaction of these molecules. The 19-amino-acid loop of PA is structurally similar to the 15-amino-acid (from residue 516 to residue 530) exposed loop in the binding domain of the diphtheria toxin (4). Replacement of Lys 516 and Phe 530 with Ala results in mutant diphtheria toxins that are 22 and 10 times less cytotoxic than the wild-type protein, respectively, because the molecules do not efficiently bind to the receptor (25). The exposed loops of PA and the diphtheria toxin may have similar functions. They may be involved in stabilization of the toxin-receptor complex or may participate in receptor recognition.

FIG. 5.

Assay of binding of wild-type or mutant PA proteins to CHO-K1 cells. Various concentrations of wild-type or mutant PA proteins (from 10 to 1 × 10⁴ ng/ml) were incubated with CHO-K1 cells for 3 h at 4°C. The binding of PA proteins to cells was detected with anti-PA serum and a secondary antibody coupled to β-galactosidase. MUG was added as the substrate for β-galactosidase, and fluorescence was quantified with a fluorimeter. The experiment was carried out at least three times for each mutant PA protein.