INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Pertussis toxin is a major virulence factor of Bordetella pertussis, the gram-negative bacterium that is the causative agent of whooping cough. It is a member of the AB₅ family of toxins, consisting of five subunits, S1, S2, S3, S4, and S5, present in a 1:1:1:2:1 ratio (18, 20, 27). S1 is the A or enzymatic subunit and catalyzes the ADP-ribosylation of G proteins in the target mammalian cell. Subunits S2 to S5 form the B pentamer, which delivers the S1 subunit to the target mammalian cell. Several of the mammalian cells targeted by pertussis toxin (including lymphocytes, macrophages, and neutrophils) are important effectors of the immune system, and toxin treatment compromises their ability to function, contributing to the severity of the disease (22, 23, 29).

Pertussis toxin assembly and secretion is a complex process. Each subunit is synthesized with a signal peptide (20), which mediates secretion to the periplasm via the equivalent of the Sec-mediated secretion machinery of Escherichia coli. Folding and assembly of the subunits occurs in the periplasm, and the ptl (pertussis toxin liberation) operon is required for efficient secretion of assembled toxin past the outer membrane (7, 9, 31). The ptl secretion machinery is specific for pertussis toxin, since secretion of other known virulence factors occurs normally in ptl mutant strains (31, 32). The secretion machinery appears to discriminate between assembled and unassembled pertussis toxin, since only assembled toxin is efficiently released from the bacteria. Since secretion involves substrate recognition, the Ptl secretion machinery must recognize some domain on pertussis toxin. In previous studies, S1 mutants were deficient for pertussis toxin secretion (26), implicating a role for S1 in pertussis toxin secretion. These mutations appeared to destabilize the molecule, and the strains were all deficient in S1 production, so no specific region of the S1 molecule was shown to play a role secretion. In this study we wanted to examine mutants capable of producing stable S1 and to define regions that may play a role in secretion.

The crystal structure of pertussis toxin has been determined. The active site in S1 appears to be in a cavity formed by an -helix bent over a -strand (termed the BA box) lined by -strands containing the catalytic residues (8, 27). Several mutants with changes in the BA box region of S1 have been generated. The mutations were serine 54 to glycine (S54G), serine 55 to glycine (S55G), serine 56 to glycine (S56G), or arginine 57 to lysine (R57K). These changes did not appear to affect catalytic activity, since recombinant S1 overexpressed and purified from inclusion bodies in E. coli was shown to possess ADP-ribosylation activity (13). In this study we introduced these mutations into B. pertussis and examined their effect on pertussis toxin production and secretion. No S1 or pertussis toxin could be detected in mutant S54G. The other three mutants produced perplasmic pertussis toxin but secreted reduced levels of toxin compared to the wild-type strain. The R57K mutant expressed normal levels of periplasmic pertussis toxin but was defective in pertussis toxin secretion. These results suggest that arginine 57 and the adjacent regions of S1 may be involved in pertussis toxin secretion.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. The B. pertussis strains used in this study are described in Table 1. B. pertussis were grown on Bordet-Gengou agar (BGA) medium (Difco, Detroit, Mich.) containing 15% sheep's blood (Colorado Serum, Denver, Colo.) or in Stainer Scholte minimal broth (SS broth) as previously described (30). E. coli strains were grown on L agar. When necessary, the following antibiotics at the indicated concentrations were added to the media: nalidixic acid, 30 µg/ml; gentamicin, 10 µg/ml (for maintenance of B. pertussis and E. coli strains) or 30 µg/ml (for selection of B. pertussis transconjugants); ampicillin, 100 µg/ml; and streptomycin, 300 µg/ml.

Reagents. Restriction enzymes and T4 DNA ligase were purchased from Gibco BRL (Gaithersburg, Md.) or New England BioLabs (Beverly, Mass.) and used according to the manufacturer's specifications. SeaKem and SeaPlaque (low-melting-point) agarose were obtained from FMC Bioproducts (Rockland, Maine). Antibiotics were purchased from Sigma Chemical Co. (St. Louis, Mo.). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) reagents were obtained from Bio-Rad Laboratories (Hercules, Calif.). Tissue culture media, antibiotic supplements, and fetal bovine serum (FBS) were acquired from Gibco BRL. DNA and protein molecular weight markers were purchased from Gibco BRL or Bio-Rad Laboratories. Plasmids were isolated by using either the Midiprep Kit or the Miniprep Kit from Qiagen (Valencia, Calif.).

S1 mutant constructs. The plasmids used in this study are described in Table 2. The original S1 site-directed mutations, S1/1E/pUC19 (S54G), S1/1F/pUC19 (S55G), S1/1G/pUC19 (S56G), and S1/1H/pUC19 (R57K), were generated by N. Burnette and V. Mar (13). In a previous study (7) we developed a suicide plasmid containing the pertussis toxin structural and secretion genes (the ptxptl operon). Plasmid pKC34 contains base pairs 0 to 13,025 of ptxptl operon in the EcoRI and PstI sites of pSP72. For these studies, the 307-bp NdeI/SphI fragment containing the S1 mutation was used to replace the corresponding wild-type sequences in pKC34 (6).

Conjugation. Each ptl construct containing the S1 mutation was introduced into B. pertussis by a triparental mating as previously described (3). Transconjugants were selected on BGA plates containing nalidixic acid and gentamicin. Nonhemolytic colonies, indicative of plasmid integration into the adenylate cyclase toxin locus, were selected for characterization.

PCR analysis of transconjugants. The presence of the pertussis toxin deletion in the chromosome of the BPRA transconjugants was verified by PCR. Chromosomal DNA was prepared by suspending a few bacterial colonies in 500 µl of 50 mM Tris (pH 7.6) containing 0.2 mg of proteinase K (Sigma Chemical Co.) per ml and incubating at 65°C for 45 min, followed by boiling for 20 min. Then, 2 µl of this preparation was used per 25 µl of PCR reaction. PCR was performed by using the Advantage-GC cDNA PCR Kit (Clontech Laboratories, Palo Alto, Calif.) according to the manufacturer's instructions. Primers were designed to span the deletion: 5'-CAAGATAATCGTCCTGCTCAACCGC-3' was used as the forward primer, and 5'-GTGAGGGCATAGGTCTGGAATGTGG-3' was used as the reverse primer. The appropriate 795-bp band was observed in BPRA, and in BP338 the appropriate full-length 3,535-bp band was observed.

Secretion assay. B. pertussis from 24-h BGA cultures were suspended to an optical density at 600 nm (OD₆₀₀) of 0.1 in SS broth. Six milliliters of the suspension was plated on a BGA plate containing nalidixic acid and gentamicin. After 30 h at 37°C, the culture was harvested from the plate, and its volume was noted. The cells were pelleted at 7,000 rpm in an SS34 rotor for 10 min. The supernatant was filter sterilized and stored at 20°C for use in a pertussis toxin CHO cell assay. The bacteria were suspended in phosphate-buffered saline (PBS) to the volume harvested, and the OD₆₀₀ values were measured. To assess intracellular toxin levels, a 1-ml aliquot of cells was pelleted at 6,000 rpm in a microcentrifuge and then suspended in 100 µl of 50 mM Tris and 50 mM EDTA containing 2 mg of lysozyme per ml. After 30 min at 37°C, 900 µl of PBS containing 0.05% Tween 20 was added. The cell debris was removed by centrifugation at 12,000 rpm in a microcentrifuge, and the supernatant filter was sterilized and stored at 20°C for use in a CHO cell assay.

CHO cell assay. The CHO cell assay was used to determine pertussis toxin activity (7, 12, 31). Pertussis toxin-treated CHO cells lose contact inhibition and clump together. Serial twofold dilutions of purified pertussis toxin (List Biologicals, Campbell, Calif.) and the unknowns were made in Ham's F-12 tissue culture medium containing 1% FBS, and 20 µl of the samples were added to 250 µl of CHO cells in 96-well plates. After 48 h, the cells were fixed and stained, and clumping was examined microscopically. The limit of detection for purified pertussis toxin was approximately 1 to 2 ng/ml, and the last positive well for an unknown sample was assigned that value. Each sample was assayed in duplicate. A student's t test was used to analyze the data.

SDS-PAGE and immunoblotting. B. pertussis cells were grown and harvested as described for the secretion assay. SDS-PAGE was performed by the method of Laemmli (16) with the modifications of Peppler (21). Bacterial cells were centrifuged and concentrated to an OD₆₀₀ of 16 in PBS, and samples were added to 3 volumes of Laemmli's solubilization buffer. Samples were boiled for 7 min, and 5 µl of each was subjected to electrophoresis on a Mini Protean II gel system by using 0.75-mm spacers (Bio-Rad) with 12.5% polyacrylamide in the separating gel and 4% polyacrylamide in the stacking gel. Electrophoresis was conducted for approximately 45 min at 200 V until the dye-front left the gels. Gels were blotted onto nitrocellulose membranes (0.45-µm pore size; Schleicher & Schuell, Dassel, Germany) in a submarine apparatus (Trans-Blot Tank; Bio-Rad) using a modified Towbin buffer (28) (15.6 mM Tris-120 mM glycine-0.02% SDS-20% methanol) at 4°C. Blotting was performed for 2 h at 100 V without further cooling. Proteins were detected by probing with monoclonal antibody C3X4 to S1 (14) and visualized by chemiluminescence using the Dupont Western blot Renaissance kit (NEN Research Products, Boston, Mass.). Peroxidase-conjugated goat anti-mouse secondary antibody was purchased from Cappel (West Chester, Pa.). Apparent molecular weights were determined by comparison with prestained molecular weight markers (Bio-Rad) and purified pertussis toxin controls.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

To facilitate genetic analysis, we constructed a shuttle vector that would allow us to integrate the entire ptxptl operon into a known region of the B. pertussis chromosome (7). The 13-kb ptxptl operon was cloned into a suicide shuttle plasmid capable of replicating in E. coli but not B. pertussis (Fig. 1A). A cassette containing the cis-acting region of the OriT (origin of transfer) of P-incompatibility plasmid (to allow for mobilization from E. coli into B. pertussis), a gentamicin resistance marker, and a 2.1-kb fragment of the adenylate cyclase toxin gene, cycA (Fig. 1), was cloned into the plasmid. The adenylate cyclase toxin gene was chosen because adenylate cyclase toxin causes hemolysis on blood plates. Integration of the plasmid into this locus results in a nonhemolytic phenotype making it easy to identify plasmids that had integrated into this region of the chromosome (7).

View larger version (17K): [in this window] [in a new window]   FIG. 1.   pKC113, containing the 13-kb ptxptl operon in pSP72. White boxes, pertussis toxin structural genes; black boxes, ptl genes; hatched box, cyc/Gentr/oriT region. Pr, ptxptl promoter; E, EcoRI; P, PstI; Term, terminator; cyc, adenylate cyclase toxin gene; Gentr, gentamicin resistance; oriT, P-plasmid origin of transfer; pSP72, plasmid vector.

The plasmid can integrate into either the pertussis toxin operon or into the adenylate cyclase toxin locus. We found that integration at the adenylate cyclase locus occurred less frequently than integration into the ptxptl operon, as determined by comparison of hemolytic to nonhemolytic transconjugants, probably due to its smaller size (2.1 kb for cycA versus 13 kb for ptxptl), but multiple independent transconjugants could be obtained. PCR analysis verified that integration into the adenylate cyclase toxin locus did not affect the chromosomal pertussis toxin locus. The appropriate 795-bp band was observed in transconjugants of BPRA (Table 1, deleted for part of the pertussis toxin operon), and in BP338 (Table 1, wild-type strain) the appropriate full-length 3,535-bp band was observed (data not shown).

The nonhemolytic transconjugants were stable when maintained under antibiotic selection. We did not detect any hemolytic revertants out of 20,000 colonies examined after the first passage in the absence of gentamicin selection. However, after a second passage without gentamicin, hemolytic, gentamicin-sensitive revertants were observed at a frequency of about 1/5,000, suggesting that they had arisen by excision of the plasmid. While integration at either site should allow pertussis toxin expression from the cloned copy, other studies characterizing transconjugants that had integrated into the adenylate cyclase toxin locus allowed us to generate stable merodiploid strains to perform complementation studies (7). In these studies, pertussis toxin production and secretion was examined in parallel for at least three independent nonhemolytic transconjugants containing mutant plasmids, and at least three independent nonhemolytic transconjugants containing the wild-type plasmid control.

To determine whether the integrated ptxptl operon was capable of directing pertussis toxin production, the plasmid was introduced into the wild-type strain BP338 and into the pertussis toxin secretion mutant BPM3171, possessing a Tn5 lac insertion in the PtlC gene (Table 1). Pertussis toxin secretion was monitored. The wild-type strain secreted 1,162 ng of pertussis toxin per ml (Fig. 2, column 1) while the ptl secretion mutant only secreted 17 ng of pertussis toxin per ml (Fig. 2, column 2). Introduction of the wild-type ptxptl operon restored secretion to the ptl secretion mutant (Fig. 2, column 3); BPM3171 with the integrated plasmid produced 1,033 ng of pertussis toxin per ml, a statistically significant increase from BPM3171 lacking the integrated plasmid but not statistically different from the wild-type strain. These studies verify that integration of the ptxptl operon restores pertussis toxin production to Ptl secretion mutants. The presence of two intact copies of the ptxptl operon in the wild-type strain did not result in more pertussis toxin secretion. These results are similar to previous studies where increased copy number did not lead to increased pertussis toxin production (2, 7).

View larger version (11K): [in this window] [in a new window]   FIG. 2.   The amount of pertussis toxin secreted after 30 h as determined by CHO cell toxin assay. Columns: 1, BP338/cyc::WT (wild type with two copies of the ptxptl operon); 2, BPM3171 (Ptl mutant); 3, BPM3171/cyc::WT (Ptl mutant with one copy of the ptxptl operon); 4, BPRA/cyc::WT (ptx deletion with one copy of the ptxptl operon); 5, BPRA/cyc::S1-54G (ptx deletion with intact copy of the ptxptl operon); 6, BPRA/cyc::S1-S55G (ptx deletion with S1 mutation in the ptxptl operon); 7, BPRA/cyc::S1-S56G (ptx deletion with S1 mutation in the ptxptl operon); 8, BPRA/cyc::S1-R57K (ptx deletion with S1 mutation in the ptxptl operon). *, P < 0.05 for mutant compared to wild type.

The intact ptxptl operon was also introduced into BPRA, a virulent hemolytic strain with a deletion of the pertussis toxin promoter and the S1, S2, S4, and S5 genes. This strain fails to express pertussis toxin and the ptl secretion genes (1, 15). The introduction of the wild-type ptxptl operon restored pertussis toxin expression and secretion to BPRA (Fig. 2, column 4). BPRA containing the intact ptxptl operon appeared to secrete less pertussis toxin than BP338 or BPM3171 containing the same construct, but the difference was not statistically significant compared to BP338 (P < 0.052). In addition, BP338 and BPM3171 are isogenic, but BPRA possesses a chromosomal streptomycin resistance mutation that could affect protein synthesis.

Mutational analysis of pertussis toxin subunit S1 mutants. Each of the S1 mutations was cloned into the ptxptl shuttle vector and introduced into BPRA, the pertussis toxin deletion strain. All of the mutants secreted statistically significantly less pertussis toxin than BPRA containing the wild-type ptxptl operon (Fig. 2, columns 5 to 8). One mutant, S54G, failed to secrete any detectable pertussis toxin (Fig. 2, column 5).

View larger version (16K): [in this window] [in a new window]   FIG. 3.   The amount of cell associated pertussis toxin after 30 h as determined by CHO cell toxin assay. Columns: WT, BPRA/cyc::WT; S54G, BPRA/cyc::S1-S54G; S55G, BPRA/cyc::S1-S55G; S56G, BPRA/cyc::S1-S56G; R57K, BPRA/cyc::S1-R57K. *, P < 0.05 for mutant compared to the wild type.

View larger version (34K): [in this window] [in a new window]   FIG. 4.   Western blot of intracellular S1. Lanes: 1, wild-type strain BP338; 2, pertussis toxin deletion strain BPRA; 3 and 8, BPRA/cyc::WT; 4, BPRA/cyc::S1-S54G; 5, S55G, BPRA/cyc::S1-S55G; 6, BPRA/cyc::S1-S56G; 7, BPRA/cyc::S1-R57K. The arrow denotes the migration point of intact 28-kDa S1 from purified pertussis toxin.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Pertussis toxin has the most complex structure of any toxin, with five subunits coming together with an unusual 1:1:1:2:1 stoichiometry. Toxin assembly occurs in the periplasm and is uncoupled from secretion past the outer membrane, since equivalent amounts of intracellular toxin accumulate in the absence of a functional ptl secretion operon (2, 6, 7, 31). It has been suggested that the nine Ptl proteins produce a gated channel in the outer membrane that allows specific passage of pertussis toxin (33). In support of this model, Ptl mutants are not deficient for secretion of any of the other known B. pertussis virulence factors (32). Thus, the Ptl-secretion machinery must specifically recognize its substrate, assembled pertussis toxin, before opening the gate. The goal of this study was to identify domains on pertussis toxin that are required for secretion.

In the process of trying to genetically produce pertussis toxoid for vaccine purposes, several laboratories independently produced dozens of strains expressing mutant forms of S1 (4, 13, 17, 18, 24, 25). In many cases, only recombinant S1 produced by E. coli was characterized. In one of the few studies where the S1 mutations were introduced into B. pertussis and expressed in the context of the entire ptxptl operon, Pizza et al. implicated a role for the S1 subunit in secretion (24). However, these mutants all had the same phenotype in B. pertussis: a deficiency in S1 expression. As a result it was not possible to identify specific domains on S1 that were involved in secretion. In contrast, the same investigators isolated one S1 mutant (9K/129G) that was devoid of enzyme and toxin activity but was secreted at normal levels (25). This result suggests that secretion does not require enzymatic activity, and the amino acids at position 9 and 129 are not required for secretion.

To narrow the search for regions on pertussis toxin that may be required for secretion, we hypothesized that such a region would not be necessary for catalytic activity and should be exposed on the surface of pertussis toxin, where it would be accessible to the Ptl secretion machinery. The crystal structure of pertussis toxin has suggested that the catalytic site is in a cavity formed by an -helix bent over a -strand (termed the BA box), surrounded by -strands containing the catalytic residues (8). Mutants in the BA box (S54G, S55G, S56G, and R57K) were of special interest to us because they retained ADP-ribosylating activity (13). This region is partially conserved among other AB₅ toxins with ADP-ribosylating activity (8), suggesting that it may have some functional importance aside from toxicity. In examining the crystal structure of pertussis toxin, serine 54 is buried, but amino acids 55 through 58 project out from a cavity formed by the intersection of S1 with S3 and S4. The close proximity of this region of S1 to two different regions in the B subunit could provide a means to distinguish assembled toxin from free subunit.

We introduced these mutations into the pertussis toxin operon and examined secretion in B. pertussis. Like many of the previously characterized S1 mutants, the S54G mutant protein was unstable when expressed in B. pertussis. Low levels of antigenic S1, but no functional pertussis toxin could be detected in this mutant. In contrast, the other three mutants produced stable periplasmic pertussis toxin that retained toxin activity in the biological CHO cell assay following release from the periplasm. Slightly less biologically active toxin was recovered from the S55G and S56G mutants. The wild-type strain produced 266 versus 92 ng/ml for the S55G mutant and 122 ng/ml for the S56G mutant. The limit of detection is the CHO cell assay is about 1 ng/ml, so the phenotype of these two mutants is clearly different from the S54G mutant, for which no toxin activity was detected. When antigenic S1 was examined, the amount of full-length S1 produced by these mutants seemed similar to that produced with the wild type, but there was reduced expression of a breakdown product. The S1 mutants, like the Ptl secretion mutants characterized in previous studies (2, 6, 7, 31), accumulate about as much periplasmic pertussis toxin as the wild-type strain. However, when secreted toxin is also considered, the total amount of toxin expressed by the secretion mutants is much less than that produced by the wild-type strain. One explanation is that a feedback mechanism could shut down toxin synthesis when the periplasm becomes full. Alternatively, a more efficient periplasmic degradation pathway could be induced. The different phenotype of the mutants suggests that different mechanisms could be operating in each case.

In contrast, the R57K mutant expressed as much antigenic and functional toxin in the periplasm as the wild-type stain. However, the R57K mutant had a dramatic defect in secretion, and secreted only 8% as much toxin as the wild-type strain. For comparison, BPM3171, the first mutant we isolated in the ptl operon, also produced normal levels of periplasmic toxin; however, it secreted 2% as much toxin as the wild-type strain. These results suggest that arginine 57 and perhaps the adjacent regions of S1 may be involved in pertussis toxin secretion. Connell et al. (5) identified a mutation in the B subunit of cholera toxin (another member of the AB₅ family of toxins) that affected toxin secretion. Since the secretion machinery can distinguish assembled toxin from unassembled subunits, it is likely that there are regions on the B subunit that are necessary for secretion in addition to the domain that we have identified on S1. In future studies we hope to identify B subunit domains needed for secretion and to select for the second site mutations in the Ptl proteins that recognize the R57K mutant and restore secretion.