Conserved Regions from Neisseria gonorrhoeae Pilin Are Immunosilent and Not Immunosuppressive

Abstract

PilE is the primary subunit of type IV pili from Neisseria gonorrhoeae and contains a surface-exposed hypervariable region thought to be one feature of pili that has prevented development of a pilin-based vaccine. We have created a three-dimensional structure-based antigen by replacing the hypervariable region of PilE with an aspartate-glutamine linker chosen from the sequence of Pseudomonas aeruginosa PilA. We then characterized murine immune responses to this novel protein to determine if conserved PilE regions could serve as a vaccine candidate. The control PilE protein elicited strong T-cell-dependent B-cell responses that are specific to epitopes in both the hypervariable deletion and control proteins. In contrast, the hypervariable deletion protein was unable to elicit an immune response in mice, suggesting that in the absence of the hypervariable region, the conserved regions of PilE alone are not sufficient for antibody production. Further analysis of these PilE proteins with suppressor cell assays showed that neither suppresses T- or B-cell responses, and flow cytometry experiments suggested that they do not exert suppressor effects by activating T regulatory cells. Our results show that in the murine model, the hypervariable region of PilE is required to activate immune responses to pilin, whereas the conserved regions are unusually nonimmunogenic. In addition, we show that both hypervariable and conserved regions of pilin are not suppressive, suggesting that PilE does not cause the decrease in T-cell populations observed during gonococcal cervicitis.

Neisseria gonorrhoeae is a gram-negative bacterium that causes the sexually transmitted disease gonorrhea, resulting in 339,000 reported and 700,000 estimated total cases in the United States in 2005 (12). Gonococcal cervicitis also increases human immunodeficiency virus (HIV) susceptibility and accelerates HIV disease progression (3). Prior to 2004, gonorrhea infections were commonly treated with broad-spectrum antibiotics from the fluoroquinolone and cephalosporin classes, but the emergence of fluoroquinolone-resistant N. gonorrhoeae isolates in the United States has led to the recommendation that the use of fluoroquinolones for the treatment of gonococcal cases be discontinued (11). The continuing emergence of antibiotic-resistant isolates has heightened the need for the development of new antibiotic and vaccine strategies for the treatment and prevention of gonococcal infections.

For N. gonorrhoeae, an early step in infection is attachment to host epithelial cells, mediated by type IV pili. Type IV pili are flexible filaments several nanometers long formed by the association of thousands of pilin subunits. They extend in all directions from the surfaces of gonococci. These organelles have a wide variety of functions which, in Neisseria species, include bacterial aggregation (26, 43), adhesion (40, 55), invasion (39, 46), host cell signaling (32, 38), surface motility (34), and natural transformation (1, 2, 5).

Exposure of pili at the cell surface, their necessary role for establishment of infection, and their strong antigenicity led to the idea that pili could make useful vaccine components. Indeed, for other pilus-bearing pathogens, notably Dichelobacter nodosus, pilus-based vaccines are successful in preventing disease (17). Historically, however, attempts to develop a pilus-based vaccine for N. gonorrhoeae have been hampered by antigenic variation within the pilin subunit (6, 8). This sequence variation results from gene conversion events in which information from several nonexpressed pilS loci is recombined into the pilE expression locus with extremely high frequency (16, 21, 22, 29, 48, 53, 56). The most variable region, located between invariant cysteines 121 and 151, is exposed on the surfaces of assembled pili, as evidenced by the fact that antibodies raised against peptides covering this region bind to the sides of pili (18). Antipilus sera from mice and rabbits, as well as sera from humans challenged with N. gonorrhoeae, have higher-titer responses to peptides from this region than to peptides from conserved PilE regions, indicating that the hypervariable region is immunodominant (18). Nonetheless, there is a limited antibody response to adjacent conserved regions (residues 110 to 120) as well as regions at the exposed ends of the pilus (residues 49 to 56) (18). In addition, sera against some peptides (69 to 84 and 135 to 151) are able to cross-react with pili from several strains (49). These and other observations lead to the hypothesis that in the absence of the immunodominant hypervariable region, other residues might be sufficiently immunogenic to elicit a protective antibody response able to bind native pilin (19, 49).

More than a decade ago, the three-dimensional crystal structure of N. gonorrhoeae pilin was solved, revealing that each monomer adopts a “lollipop”-like structure, with the stick formed by a long α-helix. The N terminus of this helix is hydrophobic and juts out from the rest of the protein, while the C-terminal half forms an αβ-roll fold by packing against an antiparallel β-sheet (42) (Fig. 1A). Two conclusions from this work affected the understanding of the immunogenicity of N. gonorrhoeae pili. First, the structure immediately suggested that pilus assembly is stabilized by the hydrophobic packing interactions of the conserved N-terminal half of the α-helix to form a hydrophobic core of the filament (19, 23, 42). This model helped to explain the invariant nature of this region of the protein as well as its poor immunogenicity. Second, the structure revealed that within the gonococcal pilin monomer, the hypervariable region forms a β-hairpin that is not an integral part of the αβ-roll fold (Fig. 1A) and predicted it to be surface exposed along the filament. The apparently modular nature of the β-hairpin explained how it can vary in size and sequence without preventing folding, assembly, or adherence properties (42). We employed this three-dimensional structural framework to design a PilE-based protein that lacks the hypervariable region in an effort to determine the immunogenicities of the conserved regions of PilE and to assess their capacities to serve as vaccine components. One might have considered a strategy by which PilE was cleaved into peptides or in which synthetic peptides were generated based on antibody recognition to conserved PilE regions (18). Such strategies have been applied to analyze the immunogenicities of isolated portions of pilin in an effort to find vaccine components for Vibrio cholerae (57) and Pseudomonas aeruginosa (9). Our rationale in generating a PilE protein that lacked the hypervariable region was to expose immune cells to all possible epitopes present in conserved regions of the PilE globular head.

FIG. 1.

Pilin structure and sequences suggest design of DQ protein. (A) Three-dimensional structure of full-length PilE from N. gonorrhoeae MS11 (Protein Data Bank code 2PIL). Posttranslational modifications (phosphorylation and glycosylation) are displayed as line representations. (B) Sequence alignment of C-terminal regions of PilE from N. gonorrhoeae strain MS11 and PilA from P. aeruginosa strain K. Labeled above the alignments are the fourth strand of the antiparallel β-sheet (β4; cyan) that participates in the central αβ fold for each monomer and the deleted residues representing the hypervariable region in PilE from MS11 (β-hairpin; green). The C-terminal tail is in orange, and the conserved cysteines are shown in yellow. (C) Model of DQ protein. The hypervariable region has been replaced with DQ linker (green), and the N-terminal α1 helix has been truncated after residue 26. (D) SDS-12% polyacrylamide gel of purified MBP-DQ (lane 2), 12.9-kDa DQ protein (lane 3), MBP-re-Pilin (lane 3), and 14-kDa re-Pilin protein (lane 4).

We also took advantage of this engineered protein to investigate the role that pilin may play in immune suppression during N. gonorrhoeae infection. These additional experiments were motivated by the observation that during HIV infections, gonococcal cervicitis causes transient decreases in CD4⁺ and CD8⁺ T-cell counts (3, 31). Piliated gonococci may modulate initial CD4⁺ T-cell proliferation, thereby inducing secretion of interleukin-10 (IL-10), a T regulatory (T-reg) cell activator. Subsequent activation of T-reg cells could cause the weak immune responses to gonococcal infections (45). T-reg cells are a small subset of the T-cell population (about 1 to 2% in humans and 5 to 10% in mice) known to down-regulate immune responses, preventing autoimmune diseases, reducing transplant rejection, and suppressing immune responses to pathogens (35, 44, 50). Therefore, we have also assessed these PilE proteins' abilities to stimulate increases in T-reg cell populations and to actively suppress B- and T-cell responses.

In this study, we have characterized murine B- and T-cell responses to the conserved PilE regions and compared them to the responses elicited by the intact PilE globular domain. We demonstrate that conserved portions of the bacterial protein PilE are nonimmunogenic without being homologous to a self-antigen. Our data confirm that the hypervariable region is the major PilE subsequence that stimulates antibody responses. We demonstrate for the first time that neither PilE nor PilE nonimmunogenic subregions induce T-reg cell activation or suppress T- and B-cell responses to a positive stimulus.

MATERIALS AND METHODS

Cloning of DQ and re-Pilin proteins.

The gene encoding the hypervariable deletion (DQ) protein was constructed with splicing by overlap extension (SOEing) PCR (28) to amplify pilE without sequences for the hypervariable region or the N-terminal α-helix. The pilE gene was first amplified from N. gonorrhoeae MS11 genomic DNA and cloned into pET15b (Novagen). The N- and C-terminal coding sequences on either side of the hypervariable region were amplified in two separate reactions. The reverse primer for the N-terminal reaction encoded the aspartate-glutamine linker and 4 residues of the C terminus (K145-I142), while the forward primer for the C-terminal reaction encoded 4 residues of the N terminus (F120 to Q123) followed by the linker (N-forward SmaI, 5′-GCCTACCAACCCGGGACCGCCCGCGCGCAAGTTTCC-3′; N-reverse, 5′-CTTGGTGTCGATCTGGTCCTGTCCGCAGAACCATTTTACCGA-3′; C-forward, 5′-TTCTGCGGACAGGACCAGATCGACACCAAGCACCTGCCGTCA-3′; C-reverse HindIII, 5′-GACAGCTTATCATCGATAAGCTTTAATGCGGTAGTTTATCACAG-3′ [restriction sites in bold, splice overlaps underlined]). The two resulting overlapping DNA fragments were then amplified and fused in a third PCR, utilizing primers N-forward and C-reverse. These two primers were also used to amplify the truncated recombinant pilin (re-Pilin) protein from the pilE gene in pET15b.

The DQ SOEing product and the re-Pilin product were digested with SmaI and HindIII and ligated in separate reactions into similarly prepared pV13 (30). The pV13 expression vector encodes an N-terminal, histidine-tagged maltose-binding protein (MBP) under the control of an IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible T5 promoter, followed by a tobacco etch virus (TEV) protease cleavage site and a multicloning site. Plasmids were isolated from transformed Escherichia coli strain JM109, and the resulting clones were verified by automated DNA sequencing.

The above-mentioned cloning protocol yielded pV13 derivatives pJKH01 and pJKH02, encoding MBP-TEV-DQ and MBP-TEV-re-Pilin, respectively. To remove the coding region for five nonnative residues in pv13 (SKDPG at the N terminus of the cleaved recombinant protein), pJKH01 and pJKH02 were used as templates for an additional PCR amplification (forward primer, 5′-AAGGACGGATCCACCGCCCGCGCGCAAGTTTCC-3′ [BamHI site in bold]; C-reverse primer, described above). Digestion of these products and pV13 with BamHI and HindIII followed by separate ligations into the multicloning site of pV13 resulted in plasmids pJKH03 and pJKH04, bearing genes for recombinant proteins with only one additional N-terminal Ser residue after TEV cleavage. Finally, codons for a thrombin cleavage site were introduced into these plasmids between the TEV site and the beginning of the pilin genes by a further round of PCR amplification, digestion with BamHI and HindIII, and ligation (forward primer, 5′-AACAACCTCGGATCCAGCAGCGGCCTGGTGCCGCGCGGCAGCACCGCCCGCGCGCAAGTTTCCGAAGCCATCCTTTTC-3′ [BamHI site in bold, thrombin cleavage site underlined]; C-reverse primer, described above). These plasmids are pThrombinDQ and pThrombinpilin, for the DQ and re-Pilin proteins, respectively. These were used in all subsequent experiments described below.

Purification of the DQ and re-Pilin proteins.

For expression, pThrombinDQ and pThrombinpilin were transformed into E. coli BL21-DE3 cells (Stratagene) and individual colonies were picked to inoculate 6-ml overnight cultures of Luria-Bertani medium with 100 μg ml⁻¹ ampicillin (LB-amp). These cultures were used to inoculate a 1-liter culture of LB-amp. Induction with 1 mM IPTG when the culture reached an optical density at 600 nm of 0.6 for 16 h at 20°C provided high-level expression of the fusion protein. Cultures were pelleted at 5,000 rpm for 15 min, and each pellet was resuspended with lysis buffer (50 mM NaPO₄H₂, 10 mM imidazole, 300 mM NaCl, pH 8.0) at a ratio of 1 g cell paste/5 ml. Resuspended cells were brought to a final concentration of 20% ethylene glycol and lysed by sonication, and the cell lysate was clarified by centrifugation (75,600 × g for 30 min.). The fusion protein was purified using immobilized metal affinity chromatography (IMAC) on Ni Sepharose high-performance resin (Amersham). Following binding in IMACA buffer (50 mM NaPO₄H₂, 10 mM imidazole, 500 mM NaCl, pH 7.5), a linear gradient from 100% IMACA to 60% IMACB (50 mM NaPO₄H₂, 350 mM imidazole, 500 mM NaCl, pH 7.5) over 100 column volumes was used to displace the protein from the column. Fusion protein-containing fractions were pooled and exchanged against IMACA overnight at 4°C in the presence of thrombin (50 units per 500 mg of fusion protein) to remove the histidine-tagged MBP. After cleavage, the protein was rerun over the 25-ml IMAC column to bind the MBP, leaving DQ or re-Pilin in the flowthrough. The pooled flowthrough fractions were exchanged against 50 mM NaPO₄H₂, pH 6.0, overnight at 4°C and loaded onto a weak cation exchange resin (CM Sepharose Fast Flow; Amersham). Bound DQ or re-Pilin was eluted using a linear salt gradient to 100% 1 M NaCl, 50 mM NaPO₄H₂, pH 6.0, over 20 column volumes. For DQ and re-Pilin, eluted fractions were pooled and filtered through a 30K-molecular-weight cutoff Amicon Ultra-15 centrifugal filter unit (Millipore) and then concentrated in a 5K-molecular-weight cutoff Amicon Ultra-15 centrifugal filter unit (Millipore). Protein concentration was determined with the calculated extinction coefficient. The identities of these proteins were verified by mass spectrometry and N-terminal sequencing (Utah State University).

Quantitation of free sulfhydryl groups.

The presence of free thiols in both DQ and re-Pilin was determined by measuring absorbance at 412 nm following reaction of 5,5′-dithio-bis-(2-nitrobenzoic acid) (DTNB; Pierce) with available thiols in the protein (47). Briefly, proteins in 1 mM EDTA, 100 mM NaPO₄H₂, pH 8.0 (stock buffer), and 25 μl of each protein were added to 5 μl of DTNB (10 mM DTNB dissolved in stock buffer; Pierce) and then further diluted in 250 μl of stock buffer to give the final volume of 280 μl. For the blank, 5 μl of DTNB was also diluted in 275 μl of stock buffer. The amount and concentration of free sulfhydryls in each sample were calculated from the molar extinction coefficient of 2-nitro-5-thiobenzoic acid (TNB; 14,150 M⁻¹ cm⁻¹). For cases in which absorbances were too low to calculate levels of free sulfhydryls, known concentrations of cysteine hydrochloride monohydrate (Pierce) were used to generate standard curves.

Experimental animals.

Female C57BL/6J and B10BR mice, 6 to 8 weeks old, were purchased from The Jackson Laboratory (Bar Harbor, ME) for immunization trials and suppressor cell assays. Animals were handled according to National Institutes of Health and University of Wisconsin—Madison Research Animal Resource Center guidelines and housed in university-approved facilities.

Isolation of sera for ELISAs.

C57BL/6J or B10BR mice were immunized by intraperitoneal (i.p.) injection with DQ or re-Pilin (50 or 100 μg/mouse), MS11 pili (50 μg/mouse), or phosphate-buffered saline (PBS). B10BR mice were immunized by i.p. injection with DQ (50 μg/mouse). Mice were given a booster injection at 4 weeks and sacrificed at 6 weeks. Sera were isolated and used for cytokine, general antibody, and immunoglobulin (Ig)-isotyping enzyme-linked immunosorbent assays (ELISAs).

Cytokine ELISAs.

IL-2, IL-4, IL-5, and gamma interferon ELISAs (OptEIA kit; BD Biosciences) were performed according to the manufacturer's instructions. A dilution series of individual serum samples (undiluted or diluted 1:10, 1:100, 1:1,000, or 1:10,000) from each immunization (PBS, DQ, or re-Pilin) was added to triplicate wells on Immulon-4 (Dynatech, Chantilly, VA) plates and incubated for 2 h at 25°C. Cytokines were detected with biotinylated anti-μ antibody plus Sav-horseradish peroxidase for 1 h. The assay was developed for 30 min at 25°C with a 3,3′,5,5′-tetramethylbenzidine (TMB) substrate reagent set (BD Pharmingen) in the dark and quenched with 1 M H₃PO₄. The absorbance at 450 nm was measured with an automated microtiter plate reader (VersaMax; Molecular Devices).

Antibody ELISAs.

A general antibody ELISA for IgM, IgG, and IgA was performed using a standard ELISA protocol. Immulon-4 (Dynatech, Chantilly, VA) plates were coated with DQ or re-Pilin (10 μg ml⁻¹) by incubation overnight at 4°C. A dilution series of individual serum samples (undiluted or diluted 1:10, 1:100, 1:1,000, or 1:10,000) from each immunization (PBS, DQ, or re-Pilin) and a naïve C57BL/6J mouse (not immunized; 6 to 8 weeks old) was added to duplicate wells and incubated for 2 h at 25°C. Antibodies were detected with horseradish peroxidase-conjugated rabbit anti-mouse IgGAM (ZYMED Laboratories Inc.) for 1 h. The assay was developed with a TMB substrate reagent set (BD Pharmingen), and absorbances were measured as described above.

Ig-isotyping antibody ELISA (ZYMED Laboratories Inc.) was performed according to the manufacturer's instructions. Immulon-4 (Dynatech) plates were coated with the target proteins (10 μg ml⁻¹) overnight 4°C and incubated the next day in triplicate with sera (1:10 dilution) from each immunization (PBS, DQ [50 μg], re-Pilin [50 μg], or MS11-pili [50 μg]). Coated plates were also incubated with mouse IgG1 (ZYMED Laboratories Inc.) as a negative control for nonspecific binding of serum IgGs to DQ or re-Pilin. The antibody ELISA plates were incubated with biotinylated rabbit anti-mouse IgG1, IgG3, IgG2a, IgM, or IgA (ZYMED Laboratories Inc.) for 15 min at 25°C. The assay was developed with a TMB substrate reagent set (BD Pharmingen), and absorbances were measured as described above.

Lymphocyte viability.

For routine cell cultures, lymphocyte viability was ≥95% at culture initiation, as determined by dye exclusion (trypan blue) analyses. Cells cultured with DQ did not display cytotoxicity or differential cell viability compared to cells cultured with medium alone or with other antigens. Further, since the addition of DQ or re-Pilin to mitogen-activated lymphocyte cultures did not alter the proliferative responses of responder cells, this was viewed as additional evidence that DQ or re-Pilin did not affect lymphocyte viability or function.

Isolation of spleen cells for T-cell-proliferation assays and flow cytometry.

C57BL/6J mice were immunized by i.p. injection with DQ or re-Pilin (100 μg or 50 μg/mouse) or PBS. Mice were given a booster injection at 4 weeks and sacrificed at 6 weeks. Spleen cells were isolated from immunized mice after sacrifice. For T-cell-proliferation assays, spleen cells (5 × 10⁶ cells ml⁻¹) were stimulated 24 h in vitro with medium, DQ (50 or 100 μg ml⁻¹), re-Pilin (50 or 100 μg ml⁻¹), or concanavalin A (ConA; 5 μg ml⁻¹; a positive control for the induction of T cells). After 48 h of incubation, [methyl-³H]thymidine ([methyl-³H]TdR; 1 μCi/well) was added, and the plates were incubated for an additional 24 h. The cultures were harvested onto nylon-backed glass fiber filters (Packard, Meriden, CT) using an Inotech cell harvester (Inotech Biosystems International, Lansing, MI) and counted in a Beckman Coulter LS6500 scintillation counter. For flow cytometry experiments, spleen cells were stimulated 24 h in vitro with medium, DQ (10 μg ml⁻¹), re-Pilin (10 μg ml⁻¹), or lipopolysaccharide (LPS; 10 μg ml⁻¹; a positive control for the induction of T-reg cells). Cells were stained with fluorescein isothiocyanate-conjugated rat anti-mouse CD4 (BD Pharmingen), allophycocyanin-conjugated rat anti-mouse CD25 (BD Pharmingen), R-phycoerythrin-conjugated rat anti-mouse CD233 (BD Pharmingen), and DAPI (4′,6′-diamidino-2-phenylindole); they were then analyzed using a BD LSR II flow cytometer.

Suppression assays.

Naïve cells were isolated from the spleens of nonimmunized C57BL/6J mice, and spleen cells including potential suppressor cells were isolated from the spleens of C57BL/6J mice immunized with PBS, DQ (50 μg), or re-Pilin (50 μg) after 24 h. The naïve cells (5 × 10⁶ cells ml⁻¹) were then stimulated for 24 h with LPS (10 μg ml⁻¹), ConA (5 μg ml⁻¹), or medium in the absence or presence of suppressor cells (1 × 10⁶, 3 × 10⁶, or 5 × 10⁶ cells ml⁻¹). In addition, naïve cells were stimulated for 24 h with LPS, ConA, or media in the presence of DQ (50 or 100 μg ml⁻¹) or re-Pilin (50 or 100 μg ml⁻¹) to determine if these molecules exerted any direct negative effects on B and T responder cells. After 24 h of incubation, [methyl-³H]TdR (1 μCi/well) was added, and the plates were incubated for an additional 24 h. The cultures were harvested onto nylon-backed glass-fiber filters (Packard, Meriden, CT) using an Inotech cell harvester (Inotech Biosystems International, Lansing, MI) and counted with a Beckman Coulter LS6500 scintillation counter.

Statistical significance.

The statistical significance of the differences observed was assessed by Student's t test. Differences were considered significant when P values of ≤0.05 were obtained. All experiments were performed at least three times.

RESULTS

Structure-based design of a pilin without its hypervariable region.

Analysis of the native MS11 PilE crystal structure (42) indicated that removal of the 18-amino-acid β-hairpin between Q123 and I142 should not perturb pilin's central hydrophobic core, although it would leave a 6.5-Å gap between the C terminus of the globular domain and the short C-terminal tail of pilin (Fig. 1A) (42). In order to bridge this gap, the hairpin was replaced with a linker from a structurally homologous protein. P. aeruginosa strain K pilin contains a four-stranded antiparallel β-sheet like MS11 PilE (14, 24) but has the 2-residue sequence aspartate-glutamine in place of the hypervariable β-hairpin (Fig. 1B). Thus, for these experiments, a PilE protein in which the hypervariable region was replaced with a DQ linker (Fig. 1C) was created and named for the one-letter code of this dipeptide. In addition to DQ, a re-Pilin protein with the intact hypervariable region was synthesized to allow us to assess the contribution that this region makes in the immunogenicity of PilE.

Earlier efforts to express a pilin lacking the hypervariable region were unsuccessful, possibly due to poor translation, protein instability, or an assembly defect for this unusual subunit in N. gonorrhoeae (29). Therefore, we constructed DQ and re-Pilin genes without the first 27 codons of the highly nonpolar α1 helix and expressed the soluble PilE globular domains in E. coli. In the case of DQ, gene SOEing PCR (28) was used to replace codons for the hypervariable region with codons for the DQ linker. The truncated DQ and re-Pilin proteins were each expressed as fusion proteins with a thrombin-cleavable, N-terminal, histidine-tagged MBP. The fusion proteins were purified by IMAC, and the 12.7-kDa DQ and 14.3-kDa re-Pilin cleavage products were purified by IMAC and cation exchange (Fig. 1D).

In order to assess the overall folding of these recombinant proteins, we determined their free sulfhydryl content using Ellman's reagent (47). For both proteins, the concentrations of free thiol were below the detection limit of this assay, implying that Cys121 and Cys151 form the expected C-terminal disulfide bond.

The polydispersity of DQ or re-Pilin was assessed using dynamic light scattering. Re-Pilin had a polydispersity of 10.3% and an approximate hydrodynamic radius of 1.86 nm, resulting in a calculated molecular mass of 14.4 kDa, corresponding to a monomer. For DQ, polydispersity was higher (22.1%) and the approximate hydrodynamic radius was 2.5 nm, indicating that while this is capable of forming a dimer in solution, it does not form large, nonspecific aggregates.

Antibody responses in DQ- and re-Pilin-immunized mice suggest that the hypervariable region is necessary for TH-cell and subsequent B-cell activation.

Immunization trials were set up to analyze the TH-dependent humoral immune response to DQ and to compare it to the response elicited by re-Pilin protein with the intact hypervariable region. C57BL/6J mice were immunized with DQ, re-Pilin, or PBS, and serum ELISAs were used to test for the presence of antibodies specific for DQ or re-Pilin in immunized mice. In accordance with past human challenge trials (6) and epitope mapping results (18), mice immunized with re-Pilin had significant antibody responses to both target proteins (Fig. 2), demonstrating that antibodies raised in re-Pilin-immunized mice recognized the conserved epitopes in DQ as well as conserved and/or variable epitopes in re-Pilin. In contrast, DQ-immunized mice did not produce antibody responses to either target protein at levels above naïve- or PBS-immunized-mouse control levels (Fig. 2), suggesting that the hypervariable region is necessary for TH-cell activation and B-cell responses. Sera isolated from mice immunized with a wider range of antigen doses (both DQ and re-Pilin) and in the presence of the strong adjuvant TiterMax did not augment the antibody responses of mice compared to those shown here for both target proteins (data not shown).

FIG. 2.

DQ- and re-Pilin-specific antibody responses were detected in re-Pilin-immunized but not DQ-immunized mice. ELISA plates were coated with 10 μg ml⁻¹ DQ or re-Pilin (shown in parentheses) and incubated with straight sera (serum) or a 1:10, 1:100, 1:1,000, or 1:1,0000 dilution of sera (see key) isolated from mice immunized with nothing (naïve), PBS, 50 μg DQ (DQ50), 100 μg DQ (DQ100), 50 μg re-Pilin (re-Pil50), or 100 μg re-Pilin (re-Pil100). Values are means for two samples and are representative of three independent experiments. For some data points, the standard errors are so small that error bars are not visible on the graph.

Since the greatest antibody responses were in mice immunized with 50 μg of re-Pilin (Fig. 2), this experimental dose was used in Ig-isotyping experiments to determine which Ig isotypes in sera from DQ- or re-Pilin-immunized C57BL/6J mice had significant increases above the control PBS-immunized-mouse levels. Again, sera isolated from DQ-immunized mice did not produce antibody responses to either target protein above PBS-immunized-mouse levels (Fig. 3A). However, re-Pilin-immunized mice had significant IgG1, IgG2a, and IgG3 responses to both target proteins (Fig. 3A), indicating that both TH1 and TH2 cells had been activated. Collectively, these results support the hypothesis that the hypervariable region in re-Pilin was needed for appropriate TH-cell activation of B cells.

FIG. 3.

Ig isotyping of antibody responses. ELISA plates were coated with 10 μg ml⁻¹ of DQ, re-Pilin, or MS11 pili (listed in parentheses) and incubated with a 1:10 dilution of sera from mice immunized with either PBS, 50 μg DQ (DQ50), 50 μg re-Pilin (re-Pilin50), or 50 μg MS11 pili (MS11 Pili50) (see key) followed by isotype antibodies. Mouse IgG1 (Neg) not specific for any isotype was a negative control for isotype antibody binding to target proteins. Data are average values from three samples and are representative of three independent experiments. (A) C57BL/6J serum reactivity to DQ or re-Pilin. Significant IgG1, IgG2a, and IgG3 responses to each target protein were detected in re-Pilin (50 μg)-immunized mice. (B) Control responses as shown by B10BR DQ serum and C57BL/6J MS11 pilus serum reactivity to DQ, re-Pilin, or MS11 pili. Significant IgG1, IgG2a, and IgG3 responses to each target protein were detected in MS11 pilus (0.50 μg)-immunized mice. Although not shown, sera from C57BL/6J mice immunized with DQ did not respond to MS11 pili.

Important controls further this conclusion. First, sera from C57BL/6J mice immunized with 50 μg native MS11 pili had significant IgG1, IgG2a, and IgG3 responses to DQ, re-Pilin, and MS11 pili. Additionally, the lack of response to DQ was not mouse strain specific, as neither DQ-immunized C57BL/6J nor B10BR mice produced antibody responses capable of recognizing DQ, re-Pilin, or intact pilus fibers in these mouse strains (Fig. 3A and B and data not shown). Ig-isotyping screens were also performed to test for the presence of IgM and IgA. Neither IgM nor IgA was detected in the sera from these immunization trials (data not shown). Mice were injected in the peritoneal cavity, so low levels of the mucosal antibody IgA were expected. Furthermore, because mice were sacrificed 2 weeks after booster doses, at a time when IgM levels normally would have declined, low IgM serum levels were also expected.

T-cell-proliferation assays and cytokine ELISAs suggest that re-Pilin activates a low level of T cells necessary for antibody production.

In addition to analyzing antibody responses, we characterized T-cell and cytokine responses in C57BL/6J mice immunized with DQ, re-Pilin, or PBS. Since re-Pilin immunization resulted in re-Pilin- and DQ- specific antibodies, we hypothesized that T cells from re-Pilin-immunized mice would proliferate differently in response to DQ or re-Pilin from T cells isolated from DQ-immunized mice. However, isolation of spleen cells and subsequent stimulation with either DQ, re-Pilin, or ConA (T-cell stimulator) (13) yielded equivalent results for DQ, re-Pilin, and PBS immunizations. T cells from all immunizations proliferated in response to the positive-control ConA (counts at or above 20,000), but immune cells from DQ- and re-Pilin-immunized mice were unable to proliferate in response to DQ or re-Pilin (counts below 2,000).

In addition, we used ELISAs to examine sera isolated from immunized mice for TH1 (gamma interferon and IL-2) or TH2 (IL-4 and IL-5) responses. We did not detect these cytokines in sera from DQ-, re-Pilin-, or PBS-immunized mice (data not shown). Our data suggest that neither DQ nor re-Pilin strongly activates T-cell populations or cytokine responses in immunized mice. Since we can detect re-Pilin- and DQ-specific antibody responses in re-Pilin-immunized mice, we assume that a low level of T-cell activation must be occurring in response to re-Pilin to activate B-cell responses.

Neither DQ nor re-Pilin immunization induces suppressor cells, nor does exposure to exogenous DQ or re-Pilin inhibit T- and B-cell proliferation.

Since sera from DQ immunizations did not have significant increases in any of the antibody isotypes over PBS serum levels (Fig. 3), we hypothesized that the conserved PilE regions in DQ were either nonimmunogenic or perhaps actively suppressing immune responses in mice. Suppressor cell assays were utilized to investigate whether spleen cells isolated from DQ- or re-Pilin-immunized mice suppress the responses of naïve immunocompetent cells and to determine if exogenously added DQ or re-Pilin directly alters host responses.

To examine responses mediated through antigen-primed cell/cell interactions, naive spleen cells were isolated from a nonimmunized mouse and stimulated for 2 days with media, LPS (B-cell stimulator) (10), or ConA (T-cell stimulator) (13). These naïve cells were incubated in the presence of potential suppressor cells isolated from the spleens of mice immunized with PBS, DQ, or re-Pilin. If cells from DQ- or re-Pilin-immunized mice had suppressor cell capabilities, then we would expect B- or T-cell proliferation in response to LPS or ConA, respectively, to drop significantly below the naïve-cell-proliferation levels (36, 51, 58). However, the addition of cells from DQ- or re-Pilin-immunized mice did not suppress T- or B-cell proliferation (Fig. 4). Instead, as more cells from DQ-, re-Pilin-, or PBS-immunized mice were added to cultures, T- and B-cell proliferation increased as measured by [³H]TdR incorporation (Fig. 4). Potential suppressor cells did themselves proliferate in response to LPS or ConA at levels similar to those for naïve-cell proliferation and, therefore, are also not impaired in their abilities to be stimulated. As a positive control for suppression, spleen cells isolated from Trypanosoma brucei rhodesiense-infected mice were also added to naïve cells. Spleen cells from T. brucei rhodesiense-infected mice contain activated macrophages that secrete nitric oxide and prostaglandins that suppress T-cell proliferation (36, 51, 58). As expected, the addition of these cells inhibited proliferation of naïve cells in response to ConA and LPS stimulation (Fig. 4).

FIG. 4.

Neither DQ nor re-Pilin immunization induces suppressor cells. Naïve cells were stimulated 24 h with LPS (10 μgml⁻¹), ConA (5μgml⁻¹), or media only (see key) and/or in the presence of potential suppressor cells at 1e6, 3e6, or 5e6 cells ml⁻¹. Suppressor cells were isolated from PBS-, DQ (50 μg)-, or re-Pilin (50 μg)-immunized mice or from T. brucei rhodesiense (TBR)-infected mice as a positive control. B- and T-cell proliferation was measured by [methyl-³H]TdR incorporation (counts). Suppressor cells themselves were also stimulated for 24 h with LPS, ConA, or media (see key) without naïve cells (DQ control and re-Pilin control). Data are the averages for three samples and are representative of three independent experiments. Addition of T. brucei rhodesiense suppressor cells (3 × 10⁶ or 5 × 10⁶ cells ml⁻¹ for B cells and 1 × 10⁶, 3 × 10⁶, or 5 × 10⁶ cells ml⁻¹ for T cells) significantly suppressed proliferation of naïve B and T cells to LPS and ConA, respectively.

A second suppressor cell assay examined the direct effects of exogenously added DQ or re-Pilin on the proliferation of spleen cells in response to ConA or LPS. The addition of these proteins did not significantly suppress the proliferation of naïve B or T cells during LPS or ConA stimulation, indicating that neither DQ nor re-Pilin directly alters host responses (Fig. 5). For this assay, the control T. brucei rhodesiense cells again inhibited proliferation of naïve cells in response to ConA or LPS stimulation (Fig. 5). From these data, we conclude that neither the conserved PilE regions in DQ nor the conserved and hypervariable PilE regions in re-Pilin exert direct or indirect suppressor effects on T- or B-cell populations in this assay system.

FIG. 5.

Exposure to DQ or re-Pilin does not inhibit T- and B-cell proliferation. Naïve cells were stimulated 24 h with LPS (10 μg ml⁻¹), ConA (5 μg ml⁻¹), or media (see key). Exogenous DQ (DQ50; 50 μg ml⁻¹), exogenous re-Pilin (re-Pilin50; 50 μg ml⁻¹), or suppressor cells (1e6, 3e6, or 5e6 cells ml⁻¹) isolated from T. brucei rhodesiense (TBR)-infected mice were added to naïve cells. B- and T-cell proliferation was measured by [methyl-³H]TdR incorporation (counts). Data are the average values from three samples and are representative of three independent experiments. Addition of T. brucei rhodesiense control suppressor cells (3 × 10⁶ or 5 × 10⁶ cells ml⁻¹) significantly suppressed proliferation of naïve B and T cells to LPS and ConA, respectively.

T-reg cells are not involved in murine responses to DQ and re-Pilin.

In addition to determining if DQ or re-Pilin had global suppressor abilities on T and B cells, we wanted to determine if these proteins could induce a subpopulation of T-reg cells which display CD4⁺, CD25⁺, and CD223⁺. This experiment was partially motivated by published research in which IL-10 secretion from T cells was increased by piliated N. gonorrhoeae in the presence of IL-2 or anti-CD3 (45). The authors suggest that the pilus-dependent induction of IL-10 may be inducing a T-reg phenotype, which could account for poor gonococcal T- and B-cell responses during infection (45).

In order to test whether either the conserved PilE regions in DQ or the conserved and hypervariable regions of re-Pilin activate T-reg cells, spleen cells were isolated from DQ- or re-Pilin-immunized mice and stimulated for 24 h in the presence of media, DQ, re-Pilin, or LPS (T-reg cell stimulator) (10). CD4, CD25, and CD223 (33) are used to characterize T-reg cells. Cells were quadruple stained for these markers, and live-cell population numbers were determined by flow cytometry. In all immunizations, there were increases in the numbers of CD4⁺ CD25⁺ CD233⁺ cells after LPS stimulation, while cells stimulated with DQ or re-Pilin displayed CD4⁺ CD25⁺ CD233⁺ cell counts equivalent to those of unstimulated cells (data not shown). If DQ or re-Pilin activated T-reg cells, we would have expected DQ- and re-Pilin-primed mice to have at least higher CD4⁺ CD25⁺ cell counts after either DQ or re-Pilin stimulation than after LPS stimulation. Since cell counts were equivalent in all immunizations and only slightly increased in the presence of LPS, it was concluded that T-reg cells are not involved in DQ and re-Pilin responses. Because DQ and re-Pilin are recombinant proteins expressed and purified from E. coli, there is a possibility that the posttranslational modifications on native pilin from N. gonorrhoeae could have suppressor abilities that we have not addressed in this study. From these data, we conclude that the soluble conserved and hypervariable regions of PilE do not stimulate increases in T-reg cell populations.

DISCUSSION

We have used the crystallographic structure of PilE from N. gonorrhoeae to design a protein lacking both the N-terminal, hydrophobic α-helix and the hypervariable region and assessed whether this protein could serve as a gonorrhea vaccine component. We show that the conserved PilE regions in our DQ protein are unable to elicit PilE-specific antibody responses in immunized mice and conclude that these regions are nonimmunogenic. In addition, we show that immunization with either the conserved or the hypervariable regions of MS11 PilE does not cause an increase in T-reg cell populations and does not suppress T- or B-cell responses.

Past human challenge studies assessing the abilities of pilin immunizations to protect volunteers from N. gonorrhoeae challenge showed that a pilin vaccine afforded protection only when volunteers were challenged with the same gonococcal strain (6, 8). Since these landmark trials, antibody epitope mapping has shown that serum isolated after infection preferentially reacted to pilin peptides which undergo sequence variation (18). Crystallographic analysis of MS11 PilE and pilus fiber models have shown that these immunodominant hypervariable residues form a surface-exposed β-hairpin (15, 42). Since the β-hairpin does not contribute to the central fold of the pilin monomer and makes relatively few interactions with it, we hypothesized that this region could be replaced with the DQ linker from P. aeruginosa strain K pilin (Fig. 1). An alternative could have been a linker from P. aeruginosa strain K122-4 PilA, which has KA in this position (4). In addition to the DQ protein, we constructed re-Pilin, which includes the hypervariable region but does not contain posttranslational modifications in the αβ-region, due to expression in E. coli. Our rationale in using re-Pilin as a positive control, instead of purified pili, was to assess the contributions of the hypervariable region to the immunogenicity of the protein and to avoid contributions from posttranslational modifications, low-abundance pilin-like proteins (26, 59), or highly antigenic outer membrane proteins that accompany pili when they are sheared from gonococci. We justified this choice by showing that DQ serum responses do not recognize native MS11 pili, while MS11 pilus serum recognizes all target proteins.

We analyzed antibody responses to both re-Pilin and DQ to determine if conserved PilE regions in DQ elicit a broad antibody response to N. gonorrhoeae pilin. Sera isolated from control re-Pilin- or MS11 pilus-immunized animals had strong antibody responses to re-Pilin (Fig. 2 and 3A) or MS11 pili (Fig. 3B), respectively, and in both cases responses comparable to those for DQ (Fig. 2 and 3). Thus, B-cell responses to re-Pilin or purified pili elicit an antibody repertoire that recognizes conserved pilin epitopes. A response to variable epitopes has previously been demonstrated for MS11 pili (18). Although it is likely that the re-Pilin response also recognizes variable epitopes, our data do not address this point directly. Unexpectedly, immunization with DQ yielded no serum antibody responses to target proteins, even with a wide range of dosages, in the presence of adjuvant or when an additional booster dose was added to the immunization regimen (data not shown). Matrix-assisted laser desorption ionization-time of flight analysis of the protein profiles in urine samples collected 16 to 20 hours after DQ immunization revealed that mice do not simply excrete DQ, which could have levels reduced sufficiently to prevent antibody responses.

We next investigated the abilities of DQ and re-Pilin to induce T-cell proliferation. Spleen cells from DQ- or re-Pilin-immunized mice were unable to proliferate to either DQ or re-Pilin (data not shown). We have observed that these cells proliferate in response to DQ or re-Pilin with anti-CD3 costimulation, which would suggest that primed T cells recognize epitopes in both proteins. However, this proliferation is nonspecific as these spleen cells also proliferate at levels similar to those for a negative control, hen egg lysozyme. Consequently, the previously observed IL-2-primed human T-cell proliferation in response to anti-CD3 and piliated N. gonorrhoeae (45) and the presently observed T-cell proliferation in response to anti-CD3 and DQ or re-Pilin are both consistent with nonspecific costimulation by anti-CD3. In our hands, pili nonspecifically activate naïve T cells in the absence of anti-CD3 (data not shown), calling into question the wisdom of using anti-CD3 as a mechanism for boosting specific immune responses to gonococcal proteins or cells. We believe that immunization with re-Pilin must induce a low level of T helper cell activation that is sufficient for antibody production but is not robust enough to be detected by T-cell-proliferation assays.

Since we were not able to detect DQ-specific T-cell responses and as serum-antibody levels from DQ-immunizations were equivalent to control serum levels from PBS-immunized mice (Fig. 2 and 3), we hypothesized that the conserved PilE regions in DQ were either nonimmunogenic or actively suppressing the immune responses in mice.

N. gonorrhoeae infections have been suggested to down-regulate mucosal immune defenses in order to promote bacterial survival in the human host. Recent work has linked gonococcal cervicitis to decreases in T-cell populations thought to increase HIV susceptibility and disease progression (3). Studies have also shown that interactions between opacity protein-expressing gonococci and the carcinoembryonic antigen-related cellular adhesion molecule on immune cell surfaces inhibit T-cell responses (7) and induce B-cell apoptosis, halting antibody production (41). In addition, previous work has shown that costimulation of T cells by piliated gonococci and IL-2 or anti-CD3 can induce IL-10 secretion, hinting at a role for pili (or pilus function) in N. gonorrhoeae-mediated immune suppression (45). To determine if portions of PilE can suppress immune-cell responses, we conducted suppressor cell assays evaluating the direct and indirect effects of DQ and re-Pilin on B- and T-cell responses. If binding of pilin proteins to immune cell surfaces could activate T-reg cells during gonococcal cervicitis, these cells would inhibit other T-cell responses, accounting for decreases in CD4⁺ and CD8⁺ cells (3). Neither the addition of exogenous DQ or re-Pilin nor that of spleen cells isolated after DQ or re-Pilin immunization suppresses T- or B-cell populations (Fig. 4 and 5). Likewise, analysis of CD4⁺ CD25⁺ CD223⁺ cell populations by flow cytometry suggests that neither DQ nor re-Pilin actively suppresses responses by stimulating T-reg cells. Since CD4⁺ CD25⁺ CD223⁺ cell counts were equivalent in all immunizations (data not shown), we concluded that neither DQ nor re-Pilin activates T-reg cell responses.

These data suggest that immunization with re-Pilin leads to helper T-cell and subsequent B-cell activation, resulting in antibody production, rather than T-reg cell activation and immune suppression. However, neither the antibody nor suppression data explain the basis for the DQ protein's immunosilence. Analysis of past literature reveals that self-antigens, such as tumor-associated antigens, have been documented as nonimmunogenic (27). Consequently, we performed BLAST searches of the DQ protein in mouse and human protein databases and found that the DQ protein is not homologous to any known self-antigens. To the best of our knowledge, this is the first study showing that a portion of a bacterial protein is immunosilent without any sequence homology to mammalian proteins. One possible explanation for our data is that the nonimmunogenicity of DQ is a result of altered antigen-processing events. Presentation on major histocompatibility complex II molecules is necessary for TH-cell activation and subsequent B-cell differentiation into memory and antibody-producing plasma cells. It is possible that the conserved PilE regions that make up DQ are not being processed and presented to murine T and B cells. Future studies evaluating DQ processing and presentation on major histocompatibility complex II molecules may provide answers to these questions.

The ability of N. gonorrhoeae to evade host immune responses has been attributed to this bacterium's capacity to vary the expression and/or sequence of surface antigens such as pili (54), lipooligosaccharide (52), opacity proteins (37), and porin (20). In addition to phase and antigenic variation, N. gonorrhoeae avoids host responses by having evolved surface antigens that do not activate the TH2 cells needed for humoral responses during uncomplicated genital infections, thereby avoiding the immunological memory necessary for protection against subsequent infections (25). Our data on PilE are consistent with this model and suggest how N. gonorrhoeae infects a human host without generating immunological memory to pilin. Immunization with re-Pilin induces an antibody response capable of recognizing epitopes in DQ and re-Pilin (Fig. 2 and 3). However, the hypervariable and conserved PilE regions of re-Pilin are unable to activate T helper cell populations or stimulate significant cytokine responses that are necessary for memory cell induction. In addition, no PilE-specific antibody responses are activated by immunization with the conserved regions of DQ.

We have incorporated the conserved PilE regions into a rationally designed antigen and shown that these regions are nonimmunogenic. We conclude from this study that the DQ protein, which contains conserved PilE regions, is not an appropriate vaccine candidate for N. gonorrhoeae. If the hypervariable PilE region is present in the recombinant protein, we detect DQ- and re-Pilin-specific antibodies, suggesting that the hypervariable region is critical for activating PilE-specific antibody responses. Since antibodies from re-Pilin-immunized mice bind DQ epitopes, the addition of a T-cell epitope from a different gonococcal protein in place of the linker could activate T and B cells, allowing for the production of antibodies to the conserved PilE regions. Further studies will have to address whether such antibodies can bind intact pilus fibers from different N. gonorrhoeae strains and/or facilitate complement-mediated lysis of bacteria in serum bactericidal assays. Our data show that the conserved PilE regions with the addition of the DQ dipeptide cannot stimulate antibodies capable of binding intact pili. We also show that conserved and hypervariable regions of PilE are not T-reg cell activators and do not down-regulate T- and B-cell responses, leading us to conclude that PilE protein sequences are not responsible for the observed decreases in CD4⁺ cells in gonococcal cervicitis (3).