Identification of Peptides That Mimic the Pertussis Toxin Binding Site on Bovine Fetuin

Abstract

The introduction of acellular pertussis vaccines has greatly enhanced the safety profile of vaccines to prevent whooping cough. Pertussis toxin (Ptx) is one component produced by Bordetella pertussis that is contained in all of these vaccines, either in combination with other known pertussis virulence factors or as the sole pertussis component, combined with tetanus and diphtheria toxoids. A hydrogen peroxide toxoid of Ptx has been shown to be efficacious in preventing pertussis infections in a mass vaccination trial and is presently licensed in the United States and Europe (B. Trollfors, J. Taranger, T. Lagergard, L. Lind, V. Sundh, G. Zackrisson, C. U. Lowe, W. Blackwelder, and J. B. Robbins, N. Engl. J. Med. 333:1045-1050, 1995). The industrial production of Ptx can be performed through the cultivation of B. pertussis in well-defined growth media, in which the components can be well characterized and their origins can be documented. Once the bacteria are removed from the culture, Ptx can be isolated from the supernatant and purified by using the technique described by Sekura et al. (R. D. Sekura, F. Fish, C. R. Manclark, B. Meade, and Y. L. Zhang, J. Biol. Chem. 258:14647-14651, 1983). The only drawback of this procedure, which combines two affinity chromatography steps, one with Blue Sepharose and a second with matrix-bound bovine fetuin (BF), is the source and purity of the BF. Concern about vaccine preparations that may possibly risk contamination by material associated with bovine spongioform encephalopathy has continued to increase. We thus sought a replacement for the BF affinity chromatography and, more specifically, for the glycosidic moiety on BF. We describe here the identification of a seven-amino-acid peptide that mimics the glycosidic moiety on BF to which Ptx binds. Furthermore, we have constructed an affinity column containing this peptide that can be used to replace BF in Ptx purification. Finally, we used the X-ray crystallographic structure of Ptx bound to the oligosaccharide moiety of BF as a scaffold and replaced the oligosaccharide with the peptide.

Pertussis toxin (Ptx) is a key component in all acellular pertussis vaccines currently in use (4). The introduction of these acellular pertussis vaccines, which combine either Ptx as the sole pertussis component or several pertussis virulence factors with tetanus toxoid and diphtheria toxoid, has greatly improved the safety profiles over those of pertussis whole-cell vaccines, and the acellular vaccines have been found to be very efficacious (2, 4). Moreover, Ptx alone, combined with tetanus toxoid and diphtheria toxoid, has been shown to eliminate the burden of pertussis disease in a mass vaccination trial (15). To ensure the availability of such vaccines and their safety, we have sought to improve the production yield of Ptx (1) as well as to improve the isolation and purification of Ptx. We describe here the ability to replace bovine fetuin (BF), a substance often used in connection with affinity chromatographic purification of Ptx, with a peptide mimic which resembles the glycosidic moiety on BF to which Ptx adheres. This replacement would further refine a well-defined and characterized process and eliminate any possible contamination of ruminant origin.

Ptx is a well known AB-type toxin, with the A portion made up of the so-called S1 subunit having the ADP-ribosyltransferase activity and the B component comprising four similar polypeptides, S2 to S5, mediating the Ptx binding activity (7, 13, 14). Interestingly, it has been shown that sequences on both the S1 subunit and on S2 and S4 are required for secretion of the Ptx holotoxin (5). The ligands to which Ptx binds have been shown to contain oligosaccharides having the sialyllactosamine structure. Using Chinese hamster ovary (CHO) cells, which have been used to measure Ptx activity, Witvliet et al. demonstrated that the optimal binding of Ptx required a complete sialyllactosamine moiety on surface macromolecules (18). Such moieties can explain the interactions of Ptx with a variety of cells, such as chicken, horse, and goose erythrocytes, as well as glycosylated serum components, including haptoglobulin and BF. Stein et al. (12) have demonstrated the interaction of Ptx with these sugar complexes by X-ray crystallography.

Peptides that mimic nonproteinaceous structures were first demonstrated by Ward et al. (16) for phosphorylcholine and by Westerink et al. (17) for group C meningococcal polysaccharide. Subsequently, many groups have demonstrated that specific amino acid sequences can take on structures resembling specific carbohydrate structures. Luo et al. have recently presented a hypothesis of the mechanism for this mimicry (8). In our search to replace BF in the affinity purification of Ptx, we sought a substance that could be easily defined and characterized as well as having a binding mechanism and affinity similar to those of BF. Thus, from a phage display peptide library, we required that the phage not only bind to Ptx but also inhibit the binding of Ptx to BF. In this study, we report the successful identification and characterization of several peptides isolated from a phage display library that mimic the glycosidic moiety on sialylated BF. A synthetic peptide was constructed from these sequences and covalently bound to a solid chromatographic matrix. We demonstrated that this peptide affinity column can successfully substitute for that constructed from BF and is similar in its ability to bind as well as block the functional activity of pertussis toxin. Finally, we have constructed a molecular model of the peptide bound to Ptx.

MATERIALS AND METHODS

Phage isolation.

A Ph.D.-7 phage display peptide library (New England Biolabs, Beverly, Mass.) was screened for phage that bound soluble Ptx. The media and procedures used for propagation of Ptx-bound phage were similar to those described in the manufacturer's instructions. Phage expressing peptide sequences having a high affinity for Ptx were identified as follows. The phage library was expanded to a density of 2 × 10¹¹ virions/10 μl according to the manufacturer's instructions. The phage were then diluted in 200 μl of 50 mM Tris-HCl (pH 7.5)-0.1% Tween 20 (TBST) (Sigma Chemical Co., St. Louis, Mo.) containing 600 ng of purified Ptx (100 μg/ml) and incubated at room temperature (RT) for 20 min. Thereafter, a Ptx-specific monoclonal antibody (MAb), designated 20.6, that recognizes the S1 subunit (9) was diluted 1:50,000 in TBST, added to the phage mixture, and incubated for an additional 1 h at RT. In a separate microcentrifuge tube, 100 μl of a 50% slurry of protein G-agarose beads (Pharmacia, Chicago, Ill.) was washed with 1 ml of TBST and centrifuged, and the supernatant was removed. The phage-Ptx-Ptx MAb mixture was transferred to a tube containing washed resin, mixed, and incubated at RT for 15 min. The resulting phage-Ptx-Ptx MAb-protein G complex was removed from the solution by centrifugation at 1,500 × g. This complex was washed 10 times with 1 ml of TBST each time, and the captured phage were eluted by addition of 1 ml of 0.2 M glycine-HCl (pH 2.2) containing 1 mg of bovine serum albumin (BSA) per ml, followed by neutralization with 1 M Tris-HCl (pH 9.1). The eluted phage were amplified according to the instructions of the manufacturer (New England Biolabs) and selected twice more by this procedure to obtain a phage population containing high-affinity Ptx binding. The phage were plaque purified on a lawn of susceptible Escherichia coli plated on Luria-Bertani agar, and the purified clones were amplified for single-stranded DNA isolation. The purified single-stranded DNA from the amplified phage was sequenced by using the dRhodamine dye terminator sequencing kit (Applied Biosystems, Foster City, Calif.) with the 28-mer gIII sequencing primer (New England Biolabs). The DNA sequencing reactions were analyzed with a ABI 310 genetic analyzer (Applied Biosystems). The phage DNA sequences were translated, and the resulting peptides were aligned by using the DNAsis sequence analysis package, version 2.0 (Hitachi Genetic Systems, Alameda, Calif.).

Detection of phage bound to Ptx.

Microtiter plates (Nunc Maxi-Sorp; Vangard International, Neptune, N.J.) were coated with 4 μg of BF per ml in 0.1 M carbonate buffer (pH 9.8) overnight at 4°C. The plates were washed with sodium acetate-Brij buffer (0.12 M C₂H₃O₂Na, 0.15 M NaCl, 0.005% Brij 35 [pH 7.5]), and nonspecific adherence was blocked by the addition of 0.2% fetal bovine serum (FBS)-phosphate-buffered saline (PBS) (5.35 mM KH₂PO4, 1.25 mM Na₂HPO4, 0.14 M NaCl) (pH 7.5). The plates were incubated for 1 h at RT and then washed three times as before. Purified Ptx (3 μg/ml) in PBS-0.08% BSA (PBS-BSA) was added, and the reaction mixture was incubated for 1 h at RT. The plates were washed as before, reacted with twofold dilutions of concentrated phage diluted in PBS-BSA, and incubated for 1 h at RT. The plates were further washed as before, and then an anti-M13 phage MAb-peroxidase conjugate (Amersham, Piscataway, N.J.) diluted 1:1,000 in PBS-Brij was added and incubated for 1 h at RT. The plates were washed, developed with 2,2′-azino-bis(3-ethylbenz-thiazoline-6-sufonic acid) (ABTS) substrate (Sigma Chemical Co) for 30 min at RT, and read at 405 nm with an MRX_TC ELISA plate reader with Revelation software package, version 4.0 (Dynex Technologies, Inc., Chantilly, Va.).

Inhibition of BF binding to Ptx with isolated phage.

Microtiter plates (Nunc Maxi-Sorp; Vangard International) were coated with 4 μg of Ptx per ml in 0.1 M carbonate buffer (pH 9.6) overnight at RT. The plates were washed and blocked as before. The phage stocks were concentrated to 2 × 10¹¹ virions/10 μl by polyethylene glycol precipitation. The concentrated phage were then diluted 1:50 in PBS-BSA, and twofold serial dilutions in PBS-BSA were made across the plate. After the reaction mixture was incubated for 30 min, the plates were washed three times in sodium acetate-Brij, and then 100 μl of BF (4 μg/ml) in PBS-BSA was added to each well. After 1 h, the plates were washed as before and reacted with MAb 1-2-M diluted 1:1,000 in PBS-BSA. Originally this MAb was generated to gonococcal lipooligosaccharide (3), but it was shown to cross-react with oligosaccharides found on BF. The plates were than incubated for 1 h and washed as before. The bound MAb was detected by incubation with peroxidase-labeled goat anti-mouse immunoglobulin G (IgG) (Amersham) diluted 1:1,000 in PBS-Brij for 1 h at RT, developed with ABTS, and analyzed as described above.

Inhibition of Ptx binding to BF with phage.

Microtiter plates (Nunc Maxi-Sorp; Vangard International) were sensitized with BF (4 μg/ml), blocked, and washed as described above. In a separate microtiter plate, purified Ptx (3 μg/ml) was mixed with twofold dilutions of different peptide-expressing phage in PBS-Brij-0.02% FBS and incubated at 37°C for 1 h. The Ptx-phage mixture was then added to the BF-coated plates and incubated for 1 h at RT. After incubation, the plates were washed and reacted with Ptx-specific MAb 20.6 (9) diluted 1:3,000 in PBS-BSA for 2 h at RT. The plates were washed, and the secondary antibody, phosphatase-labeled goat anti-mouse IgG (heavy plus light chains) (Kirkegaard & Perry, Gaithersburg, Md.), diluted 1:3,000 in PBS-BSA was added to the plate and incubated for 2 h at RT. The plates were washed, developed with p-nitrophenyl phosphate (Sigma Chemical Co.) in a solution containing 0.1 M diethanolamine and 1 mM MgCl₂ (pH 9.8), and incubated for 1 h at RT before the absorbance at 405 nm was determined as before.

Determination of affinity of Ptx for the 3G2-based peptide and BF.

For determining the affinity of Ptx for the 3G2-based peptide, Reacti-Bind anhydride activated polystyrene microtiter plates (Pierce, Rockford, Ill.) were incubated overnight with 1 μM 3G2 peptide in 100 mM carbonate buffer (pH 9.6). Likewise, for determining the affinity of Ptx for BF, Nunc Maxi-Sorp microtiter plates (Vanguard International) were coated with 4 μg of BF (Sigma Chemical Co.) per ml. Initially, the plates were washed twice with sodium acetate-Brij, blocked for 1 h with 0.1% FBS, and washed twice, as before. A titration curve was established starting at 100 μM Ptx diluted in PBS-BSA, and serial twofold dilutions of the 100 μM stock was added across the plates. The plates were then incubated for 90 min at RT with shaking. The plates were then washed twice and incubated with the Ptx-specific MAb 20.6 (9) for 90 min. The plates were then washed twice and incubated for 90 min at RT with alkaline phosphatase-labeled goat anti-mouse IgG conjugate (Kirkegaard & Perry) diluted 1:2,000 in PBS-BSA. The plates were developed with p-nitrophenyl phosphate as before, and the K_m of the interaction was calculated using the Revelation software package, version 4.22 (Dynex Technologies, Inc.).

Purification of Ptx with a peptide affinity column.

We followed the procedure for the purification of Ptx first described by Sekura et al. (10). The initial step was performed by applying the filtrate from filtration (0.22-μm-pore-size filter) of a culture of Bordetella pertussis grown in modified Stainer-Scholte medium (11) to a Blue Sepharose Fast Flow XK50/30 column (Amersham Pharmacia Biotech) equilibrated with 0.25 M sodium phosphate (pH 6.0). Ptx was eluted from the column with 0.75 M MgCl₂-0.05 M Tris-HCl (pH 7.3). The eluted fractions containing Ptx were pooled, diluted 1:1 in distilled water, and applied to either the peptide column described below or to an affinity column made from BF, as previously described (10), equilibrated with 0.05 M Tris-HCl (pH 7.5). Bound Ptx was eluted from the BF column with 4 M MgCl₂ · 6H₂O-0.05 M Tris-HCl (pH 6.2).

The Ptx purified with the BF affinity column was compared with that purified with a peptide affinity column. The designed peptide, which contained two repetitive sequences of DGSFSGFG and a G-G-G spacer arm (GGGDGSFSGFGDGSFSGFG), was purchased from American Peptide Company (Sunnyvale, Calif.). The peptide was purified by high-pressure liquid chromatography to a purity of greater than 95%, and its composition was confirmed by both amino acid analysis and mass spectroscopy by the vendor (data not shown). Peptides were coupled at ligand densities of either 1.0 or 2.0 mg/ml of gel resin by using 50 ml each of NHS Sepharose (Amersham Pharmacia Biotech) according to the manufacturer's protocols. The resulting affinity columns were packed into XK 16 columns (Amersham Pharmacia Biotech), washed, and equilibrated with 0.05 M Tris-HCl (pH 6.0). The Ptx sample described above was applied to the columns, and the columns were washed with equilibration buffer until the absorbance at 280 nm returned to the baseline value. Bound Ptx was eluted from peptide columns with 3 M MgCl₂ · 6H₂O-1 M Tris (pH 7.4).

Comparison of Ptx samples by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis.

Precast 4 to 12% bis-Tris gels, sample buffer, MES (morpholinepropanesulfonic acid) running buffer, and carbonate transfer buffer for blotting were purchased from Novex (San Diego, Calif.). The B. pertussis antigens used for positive controls were 69-kDa antigen (Staten Serum Institute, Copenhagen, Denmark), native dissociated fimbriae (Peter Fusco, Baxter Healthcare, Columbia, Md.), adenylate cyclase (Erik L. Hewlett, University of Virginia, Charlottesville), and filamentous hemagglutinin (FHA) (Juan L. Arciniega, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, Md.). The antibodies used for this study were anti-69 kDa MAb BPE3, anti-fimbria 2 and 3 MAb 3586A10, anti-FHA MAb X3C (from Juan L. Arciniega), rabbit polyclonal antibody to Ptx (Baxter Healthcare), and anti-adenylate cyclase (from Erik L. Hewlett). Ptx samples were standardized for protein content by using bicinchoninic acid (Pierce Chemicals) according to the manufacturer's protocol. The Ptx samples and controls were diluted with 4× sample buffer and heated at 100°C for 5 min. Fifteen microliters of each sample (containing 30 μg of Ptx per well and the titrated amount of control antigens) was loaded onto the precast 4 to 12% bis-Tris gel and electrophoresed according to the Novex protocols. The gels were stained by using a SilverXpress kit (Invitrogen). Gels for Western blotting, after electrophoresis, were rinsed with water and transfer buffer and electrotransferred onto polyvinylidene difluoride membranes. The membranes were blocked in 3% Carnation nonfat dry milk in PBS-0.5% Tween 20 for 1 h at RT. After being washed trice in the PBS-Tween buffer, the membranes were incubated with their corresponding primary MAbs and polyclonal sera, diluted 1:1,000 in PBS-Tween, for 2 h. The blots were washed again as before and incubated with either alkaline phosphatase goat anti-mouse IgG (Organon Teknika Co., Durham, N.C.) or alkaline phosphatase goat anti-rabbit IgG heavy and light chains (Kirkegaard & Perry) for 1 h at RT. The blots were rinsed as before and stained with the 5-bromo-4-chloro-3-indolylphosphate-nitroblue tetrazolium substrate solution from Kirkegaard & Perry. The blotting membranes were rinsed with water to stop the reaction after 30 min and allowed to air dry. The membranes were photographed with a model IS-1000 digital imaging system (Alpha Innotech Co., San Leandro, Calif.).

Comparative human inhibition ELISA.

Eight human serum samples from individuals inoculated with a licensed pertussis vaccine were available to study the interaction between Ptx and antibodies in the sera. A method was developed to determine the amount of antibodies in a particular serum which are directed to the surface-exposed epitopes of Ptx. In this way, the Ptx purified with the BF affinity column and that purified with the 3G2-based peptide affinity column were compared.

An inhibition enzyme-linked immunosorbent assay (ELISA) with control Ptx bound to the bottom of a BF-coated ELISA plate was used. In this assay, various concentrations of Ptx purified by using the BF affinity column or the 3G2-based peptide affinity column (ranging from 15.0 to 0.0007 μg/ml) were mixed in a separate V-bottom plate with one of the human serum samples at the half-maximal ELISA titer (1:250,000). After 2 h of incubation at RT, the mixture was centrifuged at 5,000 rpm for 10 min at RT in a Sorvall RT 6000B centrifuge to remove the antibody-antigen complexes. The free-antibody-containing supernatant was then placed on the BF-coated ELISA plate already containing 3 μg of bound Ptx per ml as described above. The available Ptx-bound antibodies were then detected with an alkaline phosphatase-labeled goat anti-human IgG conjugate, and the plates were developed as before. The following formula was used to calculate the percent inhibition of the specific antiserum at each dilution of antigen: 1 − (OD_antigen/OD_control), where OD_antigen is the average ELISA optical density (OD) from each set of wells that contained the antigen at each dilution and OD_control is the average ELISA OD from the set of wells that contained the diluted half-maximal antiserum without antigen. The OD_control should correspond closely with the half-maximal reading obtained previously.

CHO cell assay.

To determine whether the 3G2-based peptide neutralized Ptx activity, a CHO cell assay was used. In this assay, CHO cells were grown to a confluent state in Ham's F-12 medium supplemented with 10% FBS (Invitrogen, Rockville, Md.). After the cells were trypsinized and adjusted to 1 × 10⁴ cells/ml in fresh medium, 100 μl of cells was added to each well of the 96-well polystyrene flat-bottom plates (Costar, Corning, N.Y.) and incubated overnight at 37°C in a 10% CO₂ incubator. The next day, Ptx (4 ng/ml) diluted in PBS was added to round-bottom plates and incubated with sequentially diluted 3G2-based peptide, BF, or a nonspecific 20-amino-acid peptide derived from a conserved region found in meningococcal PorB porins, at an initial concentration of 1 μg/ml. The mixtures were shaken for 30 min at RT, followed by incubation for 1.5 h at 37°C. Following incubation, the contents of each well were added to corresponding wells of the plate containing CHO cells, mixed, and then incubated at 37°C with 10% CO₂ for 24 to 48 h. The medium was then decanted, and the cells were dried, fixed with methanol, and stained with Giemsa stain. Inhibition of CHO cell clustering was observed under a microscope.

Computational docking of peptide PT5-3G2 into Ptx.

The ICM-Pro software package (MolSoft LLC, San Diego, Calif.) was used to model the binding of the 3G2 peptide (DGSFSGF) to Ptx. The binding sites on the protein target were derived from a crystal structure of sialyllactosamine-bound Ptx (Protein Database entry 1PTO); the three very similar sites were defined as those residues in the S2 subunit (chain B) and S3 subunits (chain C and chain I) within 4 Å of but not including the respective bound oligosaccharide. Before a docking simulation, the 3G2 peptide was divided into three-residue pieces (DGS, GSF, SFS, FSG, and SGF) with uncharged termini, and the atomic coordinates for each piece were constructed in silico. For each binding site on Ptx, a cube roughly 20 Å per side centered at the binding site was built, and five grid maps (Van der Waals [carbon atom probe], Van der Waals [hydrogen atom probe], hydrophobic, electrostatic, and hydrogen bond) were created over the protein surface within the cube. During a docking run (10 times normal length/time), a completely flexible peptide piece was fit into the set of maps describing a rigid binding site on Ptx, assuming that the protein subunit is already in a ligand-bound state. The conformation of the best fit (highest docking score) for each peptide piece into each of the three Ptx binding sites was kept as the respective docking solution.

RESULTS

The initial screening of the phage library for phage that bound Ptx was conducted by panning with Ptx bound to the bottom of a polystyrene culture plate. The several phage clones obtained from these initial experiments failed to inhibit the BF-Ptx interaction and had no similarity in amino acid sequence to each other (data not shown). We then turned to a liquid-phase binding format to determine whether this would improve our phage selection process. Phage from this peptide library were first incubated with purified Ptx in solution. The Ptx and any attached phage were then removed from solution with Ptx-specific MAb and protein G beads. After this protocol was repeated sequentially and the phage were subsequently amplified and cloned, several phage clones that appeared to bind Ptx were identified.

The next experiment was to determine whether any of these clones bound Ptx directly. In these studies, Ptx was attached to the microtiter plate, and the phage clones containing approximately equivalent amounts of phage particles were diluted across the plate. Phage clones demonstrating higher affinity, i.e., higher titers, were selected for further investigation, and the sequences of the peptide inserts were determined. These clones were additionally examined for their ability to bind to Ptx after the Ptx was bound to BF. In this case, microtiter plates were coated first with BF, then with the purified Ptx, and finally with the amplified phage clones. All of the clones were capable of binding to Ptx in a concentration-dependent manner (Fig. 1). Phage clone 3G5 (NGSFSGF) bound most strongly, followed by phage clones 3G8 (NGSFSGC) and 3G2 (DGSFSGF). The consensus sequence was XGSFSGX. As a negative control, an unrelated phage with the sequence TAPSRDL was tested similarly and showed no binding (data not shown). To determine if the phage could compete against BF for binding to Ptx, plate-bound Ptx was preincubated with the peptide-expressing phage, followed by the addition of BF. The phage inhibited binding of BF to plate-bound Ptx in a concentration-dependent manner. Phage clone 3G2, expressing the sequence DGSFSGF, showed the greatest inhibition, suggesting that it may closely resemble the Ptx binding site on BF (Fig. 2).

FIG. 1.

Direct binding of isolated phage to plate-bound Ptx. Microtiter plates were coated with 4 μg of BF per ml, and nonspecific adherence was blocked by the addition of 0.2% FBS-PBS. Purified Ptx (3 μg/ml) in PBS-BSA was then added to the plate. The bound Ptx was reacted with twofold dilutions of phage, with an initial concentration of 10⁹ phage/ml of diluent. An anti-M13 phage MAb-peroxidase conjugate was then added. The assay was developed with ABTS substrate and read at 405 nm. Values are averages of the ODs of the phage binding to Ptx for each twofold dilution performed in triplicate. The standard deviation measured for each dilution was <10%.

FIG. 2.

Inhibition of BF binding to plate-bound Ptx with peptide-expressing phage. Microtiter plates were coated with Ptx (4 μg/ml) in 0.1 M carbonate buffer (pH 9.6), and nonspecific adherence was blocked by the addition of 0.2% FBS-PBS. The phage stocks were concentrated to 2 × 10¹¹ virions/10 μl by polyethylene glycol precipitation. The concentrated phage were diluted in PBS-BSA, and twofold serial dilutions in PBS-BSA were made across the plate. BF (4 μg/ml) diluted in PBS-BSA was added to each well and then reacted with MAb 1-2-M. The bound MAb was detected by incubation with peroxidase-labeled goat anti-mouse IgG, developed with ABTS, and read at 405 nm. The percent inhibition of each dilution of phage is shown and represents the average inhibition for each phage dilution performed in triplicate. The standard deviation measured for each dilution was <10%. The coefficient of variance of the assay was <10%. The assay was repeated twice with similar results.

To confirm that this peptide-expressing phage competed directly for the BF binding site on Ptx, the phage were preincubated with Ptx in solution and then added to BF-coated plates. These results showed that these clones could also block the binding of Ptx to plate-bound BF as well (Fig. 3). Clone 3G2 again showed the greatest amount of inhibition in this assay.

FIG. 3.

Inhibition of Ptx binding to plate-bound BF with peptide-expressing phage. Microtiter plates were coated with BF (4 μg/ml), and nonspecific adherence was blocked by the addition of 0.2% FBS-PBS. In a separate microtiter plate, purified Ptx (3 μg/ml) was reacted with twofold dilutions of different peptide-expressing phage. The Ptx-phage mixture was then incubated in BF-coated plates, followed by reaction with Ptx-specific MAb 20.6. The bound MAb was detected with phosphatase-labeled goat anti-mouse IgG (heavy plus light chains), followed by development with p-nitrophenyl phosphate and analysis at 405 nm. The percent inhibition for each dilution of phage is graphed and represents the average inhibition of each phage dilution performed in triplicate. The standard deviation measured for each dilution was <10%. The coefficient of variance of the assay was <10%. The assay was repeated twice with similar results.

The phage clone 3G2 demonstrated the greatest ability to inhibit the interaction between Ptx and BF, suggesting to us that its unique sequence mimicked the oligosaccharide binding site of Ptx on BF. We had synthesized a peptide having two repetitive sequences of the 3G2 sequence of DGSFSGF and a G-G-G spacer arm, GGGDGSFSGFGDGSFSGFG. The 3G2-based peptide was purified by high-pressure liquid chromatography, and its composition was confirmed by both amino acid analysis and mass spectroscopy (data not shown). The peptide was bound to a matrix similar to that used for the construction of the BF affinity column. Partially purified Ptx samples similar to that applied to the BF affinity column were used to test the peptide affinity column. Both the sample loading and elution were conducted in a manner similar to that for the BF column. The eluates were pooled and analyzed. The resulting silver-stained SDS-PAGE profiles of the eluates from both the peptide affinity column and the BF affinity column are compared in Fig. 4A. The SDS-PAGE demonstrates that the preparation is pure and free from any contaminating B. pertussis proteins, and it shows a band profile similar to that for the BF-purified Ptx. To confirm that the additional subunits that stain weakly with silver stain were present, Western blot analysis of the BF-purified Ptx and peptide-purified Ptx with rabbit polyclonal serum to Ptx was performed (Fig. 4B). The purified Ptx was also analyzed in Western blots for the presence of contaminating adenylate cyclase, fimbriae, pertactin, and FHA. The levels of these proteins were less than 15 ng/ml in both the BF- and peptide-purified preparations (for FHA they were 0.010 and 0.030% for BF and peptide purification, respectively, which represented less than 15 ng/ml). The upper, 98-kDa band present in the silver staining (Fig. 4A) represents contaminating FHA. However, the amount of FHA is less than 15 ng/ml and less than 0.03% of the total amount of Ptx. We also wanted to determine whether the peptide affinity column-purified Ptx was in anyway antigenically different than that isolated by BF. Using serum from a single individual with a significant antibody response to Ptx from immunization with Certiva that was purified with a BF affinity column, we compared the abilities of the two Ptx preparations to remove antibodies from this serum in an inhibition ELISA. The peptide-purified Ptx demonstrated an inhibition profile that was similar, milligram for milligram, to that of the BF-purified Ptx (Fig. 5). Similar results were obtained from seven other individuals immunized with Certiva. This further suggested to us that the Ptx eluting from the peptide affinity column was antigenically identical to the BF-purified Ptx under these conditions. To determine if the 3G2-based peptide could directly inhibit the functional activity of Ptx, an assay measuring CHO cell clustering was employed (Fig. 6). This assay showed 80 to 100% inhibition of CHO cell clustering when the peptide was used as an inhibitor. The MIC of the peptide was 100 pg/ml, which was similar to the MIC of the BF (Fig. 6). As a control, the unrelated peptide derived from the sequence of the meningococcal PorB porin showed no inhibition. These results were consistent with the relative affinity constants measured for Ptx binding to the 3G2 peptide and BF, which were determined to be 1.7 × 10⁻⁷ and 9 × 10⁻⁸ M, respectively.

FIG. 4.

Comparison of BF-purified Ptx with peptide-purified Ptx by SDS-PAGE and Western blot analysis. (A) Silver-stained bis-Tris gel of Ptx purified by using the 3G2-based peptide or BF. Molecular mass markers (in kilodaltons) are shown on the left. (B) A similar gel was blotted to a polyvinylidene difluoride membrane, probed with polyclonal serum from a rabbit immunized with BF-purified Ptx, and developed with alkaline phosphatase-conjugated goat anti-rabbit IgG (heavy plus light chains) Arrows indicate the Ptx subunits S1 to S5.

FIG. 5.

Comparison of the antigenic equivalency of BF-purified and peptide-purified Ptx by using serum from an individual immunized with Certiva. An inhibition ELISA with control Ptx bound to the bottom of a BF-coated ELISA plate was used. Various concentrations Ptx purified with the BF affinity column or the 3G2-based peptide affinity column were mixed in a separate V-bottom plate with one of the human serum samples at the half-maximal ELISA titer (1:250,000). The mixture was centrifuged to remove the antibody-antigen complexes. The free-antibody-containing supernatant was then placed on the BF-coated ELISA plate, which already contained 3 μg of bound Ptx per ml. The available Ptx-bound antibodies were then detected with an alkaline phosphatase-labeled goat anti-human IgG conjugate, developed with p-nitrophenyl phosphate, and analyzed at 405 nm. The formula 1 − (OD _antigen/OD_control) was used to calculate the percent inhibition of the specific antiserum at each dilution of antigen. The percent inhibition for each dilution of peptide-purified and BF-purified Ptx is shown and represents the average for each Ptx dilution performed in triplicate. The standard deviation measured for each dilution was <10%. The coefficient of variance of the assay was <10%. The assay was repeated twice with similar results. Certiva is a licensed acellular pertussis vaccine purified by BF column chromatography.

FIG. 6.

Inhibition of Ptx cytotoxicity in CHO cells with the 3G2-based peptide. CHO cells were grown to a confluent state in Ham's F-12 medium supplemented with 10% FBS. The cells were trypsinized and adjusted to 10⁴ cells/ml in fresh medium, and 100 μl of cells was added to each well of polystyrene flat-bottom plates and incubated overnight. Ptx (4 ng/ml) diluted in PBS was then added to round-bottom plates and incubated with sequentially diluted 3G2-based peptide, BF, or a nonspecific 20-amino-acid peptide derived from a conserved region found in meningococcal PorB porins. The Ptx-peptide mixtures were then added to the flat-bottom plate containing CHO cells. The medium was decanted, and the cells were dried, fixed with methanol, and stained with Giemsa stain. Inhibition of CHO cell clustering was observed under a microscope. (A) 3G2-based peptide (15 pg/ml; 8.8 × 10⁻⁶ μM) plus Ptx (1 ng/ml; 8.4 × 10⁻⁶ μM); (B) porin peptide (15 ng/ml; 6.0 × 10⁻³ μM) plus Ptx (1 ng/ml; 8.4 × 10⁻⁶ μM); (C) BF (100 pg/ml; 2.1 × 10⁻⁹ μM) plus Ptx (1 ng/ml; 8.4 × 10⁻⁶ μM); (D) Ptx (1 ng/ml 8.4 × 10⁻⁶ μM).

The sialyllactosamine binding subunits of Ptx, S2, and S3 are very homologous in both sequence (∼70% identical) and structure (same fold; 1.44-Å all-atom root-mean-square deviations). The 1PTO crystal structure contains two copies of Ptx in the asymmetric unit, or four possible unique sialyllactosamine binding sites (two S2 and two S3 subunits), and three of these sites in the crystal structure hold an oligosaccharide. The bound oligosaccharides lie along the main chain of a particular beta strand in these subunits, with the closest saccharide bracketed by the side chains of residues His101, Tyr102, Tyr103, and Ser104 in each subunit. Molecular modeling, specifically the docking of the 3G2 peptide into these three homologous binding sites, was undertaken to attempt to understand Ptx affinity. Unfortunately, the docking of longer and flexible peptides to a protein surface, especially without a pronounced pocket, is just past the capability of any current modeling software. For this reason, the 3G2 peptide was split into three-residue pieces near the flexible docking limit of 10 to 15 torsion angles. This method of splitting a longer peptide into shorter pieces allows more accurate docking simulations and details the set of residues in the particular peptide interacting with the protein. The three-amino-acid piece of the 3G2 peptide with the best fit or highest score at each of the three Ptx sialyllactosamine binding sites was SFS (Fig. 7).

FIG. 7.

Sialyllactosamine binding site in an S3 subunit of Ptx (chain I of PDB entry 1PTO). The Ptx subunit is displayed as a yellow ribbon, with selected residues (His101, Tyr102, Tyr103, Ser104, and Arg125) drawn in a ball-and-stick style colored gold. (A) The bound oligosaccharide is represented as a ball-and-stick model colored by atom. (B) A conformation of the computationally docked SFS piece in the 3G2 peptide, represented as a ball-and-stick model colored by atom. The ICM-Pro software package was used to model the binding of the 3G2 peptide (DGSFSGF) to Ptx. The binding sites on the protein target were derived from a crystal structure of sialyllactosamine-bound Ptx (PDB entry 1PTO); the three very similar sites were defined as those residues in the S2 subunit (chain B) and S3 subunits (chain C and chain I) within 4 Å of but not including the respective bound oligosaccharide. Before a docking simulation, the 3G2 peptide was divided into three-residue pieces (DGS, GSF, SFS, FSG, and SGF) with uncharged termini, and the atomic coordinates for each piece were constructed in silico. The conformation of the best fit (highest docking score) for each peptide piece into each of the three Ptx binding sites was kept as the respective docking solution.

DISCUSSION

The ability to purify on a large scale bacterial vaccine components that retain both functional activity and, most importantly, their immunochemical structure remains an important process in the production of vaccines. In the case of Ptx, the process first described by Sekura et al., (10) combines two sequential affinity columns to purify the Ptx holotoxin. The first column is constructed of the dye Cibracon blue. This dye is though to structurally resemble NAD and probably serves as a substrate analogue to which the ADP-ribosyltransferase S1 subunit of Ptx binds (7). The subsequent second affinity column, constructed of BF, was based on initial studies by Irons and MacLennan showing that Ptx bound to the glycoprotein haptoglobulin (6) and culminated in the X-ray crystallography data of Stein et al. (12). Sekura et al. showed that commercially available BF, a sialoglycoprotein, could be formulated as a resin, and they used this affinity matrix to purify toxin that was indistinguishable from toxin purified from B. pertussis by other methods (10). The beauty of using these two columns sequentially is that it ensures that both the A and B subunits, i.e., S1 to S5 in Ptx, are present as a holotoxin and functional. However, recently the use of animal-derived material or components in vaccine production or processing has become a major concern, especially with the uncertainty of ruminant products and their exposure to infectious agents, such as that causing bovine spongioform encephalopathy. In order to both simplify the affinity ligand used in Ptx purification and replace the BF with a well-defined and characterized compound, our group sought to isolate peptide mimotopes of the glycosidic moiety on BF that bound Ptx.

The biopanning technique was modified to purify Ptx from solution and resulted in the isolation of numerous clones. Subsequent selection and cloning of the phage library resulted in seven clones, three of which showed interaction with Ptx as well as inhibition of Ptx binding to BF. Of these clones, clone 3G2 seemed to react with soluble or bound Ptx with the highest affinity and to inhibit the binding of BF. This inhibition suggested that the unique peptide expressed by this clone more closely resembles the conformation of the sialylated BF (Fig. 2 and 3).

Confirmation of the utility of this data was established by using an affinity column constructed from the deduced unique sequence of clone 3G2 for the purification of Ptx. The peptide affinity purification chromatography was compared to that with BF. Ptx eluted from both columns in a similar profile with a concentration gradient of MgCl₂. The resulting purified Ptx retained an SDS profile similar to that of BF-purified Ptx (Fig. 4A), a similar Western blot profile (Fig. 4B), and the same degree of purity as BF-purified Ptx (see Results), which suggests that both the structural integrity of the molecule and the purity of the preparation are not altered by the process. The ELISA inhibition data suggested that the two Ptx preparations were also antigenically identical. The inhibition of the functional activity by the peptide in a manner similar to that for BF suggests that the parts of the peptide could structurally and functionally mimic the oligosaccharide groups on BF.

Computer modeling of peptide 3G2 docked to the sialyllactosamine binding sites on Ptx implies that the center section of the peptide (SFS) carries the Ptx affinity, which is very comparable to the experimental result of a consensus or core sequence necessary for Ptx binding (XGSFSGX). Future studies will determine whether the peptide-purified Ptx vaccine elicits the same level of functional antibodies in mice that is induced by the presently licensed vaccine.