Functional Activity of Antibodies against the Recombinant OpaJ Protein from Neisseria meningitidis

Abstract

The opacity proteins belong to the major outer membrane proteins of the pathogenic Neisseria and are involved in adhesion and invasion. We studied the functional activity of antibodies raised against the OpaJ protein from strain H44/76. Recombinant OpaJ protein was obtained from Escherichia coli in two different ways: cytoplasmic expression in the form of inclusion bodies followed by purification and refolding and cell surface expression followed by isolation of outer membrane complexes (OMCs). Immunization with purified protein and Quillaja saponin A (QuilA) induced high levels of Opa-specific antibodies, whereas the E. coli OMC preparations generally induced lower levels of antibodies. Two chimeric Opa proteins, hybrids between OpaB and OpaJ, were generated to demonstrate that the hypervariable region 2 is immunodominant. Denatured OpaJ with QuilA induced high levels of immunoglobulin G2a (IgG2a) in addition to IgG1, whereas refolded OpaJ with QuilA induced IgG1 exclusively. These sera did not induce significant complement-mediated killing. However, all sera blocked the interaction of OpaJ-expressing bacteria to CEACAM1-transfected cells. In addition, cross-reactive blocking of OpaB-expressing bacteria to both CEACAM1- and CEA-transfected cells was found for all sera. Sera raised against purified OpaJ and against OpaJ-containing meningococcal OMCs also blocked the nonopsonic interaction of Opa-expressing meningococci with human polymorphonuclear leukocytes.

The obligate human pathogen Neisseria meningitidis primarily colonizes the nasopharynx in an asymptomatic manner. Only in a minority of cases does infection become systemic, resulting in life-threatening meningitis and sepsis. Virulent strains freshly isolated from the blood or cerebrospinal fluid are typically encapsulated. Classification into serogroups is based on structural differences in the capsular polysaccharide. While the group A, C, Y, and W-135 capsules can be used for effective vaccines, the serogroup B polysaccharide was found to be nonimmunogenic, probably due to the presence of identical α2,8-linked N-acetylneuraminic acid structures in developing fetal brain tissue (13, 16, 23, 46). Therefore, alternative strategies have been followed for the development of a vaccine against the common serogroup B, mainly focusing on the use of outer membrane proteins (OMPs) (reviewed by Jódar et al. in reference 24). Several clinical trials have been conducted with outer membrane vesicle (OMV)-based vaccines containing various combinations of OMPs (4, 12). OMVs contain the major porin proteins, PorA and PorB, lesser amounts of other proteins, including those with high molecular masses of 60 to 100 kDa, and lipopolysaccharide. Although efficacies in these trials range from 51 to 83%, improvements are clearly needed, in particular since the induced bactericidal antibodies tend to be highly type specific.

Additional studies are necessary to identify other OMPs that are important for inducing protective antibodies. All current vaccines aim to induce bactericidal serum antibodies, as this is known to correlate with protection against invasive disease (17). However, an alternative strategy would be to induce antibodies which block the infection at an earlier stage, at the level of mucosal attachment and invasion. Since Opa proteins are involved in adhesion and subsequent invasion into nasopharyngeal epithelial tissue (3, 34, 42), it is tempting to speculate that antibodies against Opa proteins block this crucial step in pathogenesis. Opa proteins are encoded by a family of homologous genes that undergo high-frequency phase and antigenic variation (18, 33). Both heparan sulfate proteoglycans and members of the CEACAM family of cell surface proteins have been identified as receptors used for Opa-mediated attachment to host cells (5, 27, 40, 42). As individual meningococcal strains are highly variable in the particular Opa repertoire expressed at any particular moment, studies of the ability of individual Opa proteins to induce a protective immune response are preferably performed by using heterologous expression systems. Additionally, by heterologous expression of the Opa protein, responses against other meningococcal surface antigens were prevented. In the present study, we describe the results of immunization experiments with mice with the invasion-associated OpaJ protein from N. meningitidis strain H44/76, expressed in an Escherichia coli background. Expression was achieved both in the outer membrane of E. coli and as cytoplasmic inclusion bodies from which OpaJ was purified and refolded into a native-like conformation (10). The resulting mouse sera were analyzed for total and isotype-specific antibody response. Functional activity of the antibodies was investigated, in particular their ability to interfere with Opa-mediated adhesion to CEACAM-expressing host cells.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

N. meningitidis strain H44/76 (B; 15; P1.7, 16) has been described previously (20). E. coli strain CE1265 was previously described by Korteland et al. (26), and strain PC2984 was obtained from NCCB (Phabagen collection, Utrecht, The Netherlands). The expression of OpaJ at the cell surface of E. coli strains CE1265 and PC2984 was realized by using the expression vector pMR05 (2) as previously described (10). Meningococci were grown overnight on GC agar plates (Difco Laboratories, Detroit, Mich.) supplemented with 1% IsoVitaleX at 37°C in a humidified 5% CO₂ atmosphere. The E. coli strains were grown at 37°C in Luria-Bertani (LB) medium (BIO 101, Carlsbad, Calif.) or on LB agar (Oxoid, Basingstoke, Hampshire, England) plates supplemented with 25 μg of chloramphenicol (Sigma, St. Louis, Mo.)/ml. For phagocytosis experiments, bacteria were heat killed at 56°C (for 30 min), after which they were washed in phosphate-buffered saline (PBS) supplemented with 1% bovine serum albumin (BSA, fraction V; Boehringer, Mannheim, Germany). Heat-killed bacteria (10¹⁰) in 0.5 ml of PBS were labeled with the pH-stable green dye Alexa488 (Molecular Probes, Leiden, The Netherlands), according to the manufacturer's protocol.

Epitope mapping.

Biotinylated oligopeptides of 18 residues, with a 14-amino-acid-residue overlap, spanning the hypervariable region 1 (HV1) and HV2 of OpaJ129 and OpaB128, were synthesized on a 10 μM scale by using an automated multiple-peptide synthesizer equipped with a 48-column reaction block (AMS 422; ABIMED Analysen-Technik Gmbh, Langenfeld, Germany) as described earlier (7). The peptides were used in a peptide enzyme-linked immunosorbent assay (ELISA) with avidin (5 μg/ml; Sigma)-coated Immunolon II 96-well microtiter plates (Dynatech) incubated for 1 h with 10 μg of peptide/ml. The reactivity of the monoclonal antibodies (MAbs) 15-1-P5.5 (kindly provided by Wendell Zollinger, Walter Reed Army Institute of Research, Washington, D.C.) (45) and MN20E12.70 (kindly provided by Betsy Kuipers, RIVM, Bilthoven, The Netherlands) was detected by using affinity purified goat anti-mouse immunoglobulins (GAM) (Southern Biotechnology Associates, Birmingham, Ala.) (1:5,000) conjugated to horseradish peroxidase (HRP).

Cloning and expression of OpaJ and OpaB and construction of chimeric Opa proteins.

The genes encoding OpaJ129 and OpaB128 were isolated from H44/76 by using Taq polymerase (Amersham, Piscataway, N.J.) and general opa primers (5′-CTTCTCTTCTCTTCCGCAGC-3′ and 5′-TCGGTATCGGGGAATCAGAA-3′), cloned into plasmid pCR2.1 (Topo TA cloning kit; Invitrogen, Carlsbad, Calif.), and subsequently sequenced with M13-forward and M13-reverse primers (Invitrogen). Plasmid pCR2.1 containing opaB128 or opaJ129 were used to amplify the DNA sequences encoding the mature OpaB128 or OpaJ129 proteins with Taq polymerase. The primers used (5′-AGCGCCCATGGCAAGTGAAG-3′ and 5′-GGCATCGGGATCCGGGAATCAG-3′) were based on the DNA sequences of opaB128 and opaJ129 of N. meningitidis strains H44/76 (unpublished data) and 190/87 (GenBank accession no. AF016285) (28). The primers contained base substitutions (underlined) to introduce NcoI and BamHI cleavage sites, respectively. The PCR product was cloned in plasmid pCR2.1. The NcoI-BamHI fragment was isolated from the resulting plasmid and ligated into the NcoI-BamHI-digested expression vector pET11d (New England Biolabs, Inc., Beverly, Mass.) downstream of the inducible T7 promoter. In the resulting construct, the codon for the first amino acid residue of the mature Opa protein was situated directly downstream of the ATG start codon. The sequence of the insert was checked by DNA sequencing by using the DNA sequencing kit and the ABI Prism 310 genetic analyzer according to the instructions of the manufacturer (Perkin Elmer Applied Biosystems, Warrington, Great Britain). Plasmids pET11d-opaB128 and pET11d-opaJ129 were used to transform the E. coli strain BL21(DE3).

OpaB128 and OpaJ129 expression at the cell surface of E. coli strain CE1265 was realized as previously described (10) by using the expression vector pMR05 containing the complete phoE gene (2). PCRs were performed on pCR2.1 containing either opaB128 or opaJ129 by using Taq polymerase and mutagenic primers (5′-ATAGATCTCGGGGAATCAGAAGCG-3′ and 5′-CTTCTCTTCTCTTCTGCAGC-3′) to generate a PstI site between the signal sequence and the mature portion and a BglII site behind the stop codon of opaB128 and opaJ129. The PstI-BglII fragments of opaB128 and opaJ129 were used to replace a PstI-BglII fragment of the phoE gene in pMR05, resulting in an in-frame fusion of opa to the signal peptide of phoE and expression from the phoE promoter. The resulting plasmids, pMR05-opaB and pMR05-opaJ, were isolated and purified with the Wizard Plus SV Miniprep kit (Promega, Madison, Wis.) and subsequently digested with PstI and AflIII (Boehringer). The fragments were electrophoretically separated on a 1% (wt/vol) agarose gel and isolated with the JETsorb gel extraction kit (Genomed, Bad Oeynhausen, Germany). The PstI/AflIII fragments coding for the semivariable (SV) region and HV1 were exchanged and religated by using T4 ligase in the appropriate buffer according to the instructions of the manufacturer (Boehringer). The resulting plasmids were used for transformation of CE1265 as described previously (10). The sequences of the inserts were checked by DNA sequencing. Surface expression of the two chimeras, OpaBj and OpaJb, was analyzed in a colony blotting experiment (37) with MAbs 15-1-P5.5 and MN20E12.70.

Refolding and purification of Opa protein.

In vitro folded and purified OpaJ protein was prepared as previously described (10). The purification of inclusion bodies was improved. Cultures of the E. coli strain BL21(DE3) containing either pET11d-opaB128 or pET11d-opaJ129, grown overnight at 37°C, were diluted 1/10 into fresh LB medium supplemented with 0.5% glucose (Fluka, Buchs, Switzerland) and ampicillin (100 μg/ml). When the culture reached an optical density at 660 nm (OD₆₆₀) of 0.6, isopropyl-β-d-thiogalactopyranoside (IPTG) (Boehringer) was added to a final concentration of 1 mM. After 3 h of incubation at 37°C, the cells were harvested by centrifugation (5,000 × g for 15 min at 4°C), the cell pellet was resuspended in 50 mM Tris-HCl and 40 mM EDTA (TE buffer). Sucrose (0.25 g/ml; Sigma) and lysozyme (0.2 mg/ml; Boehringer) were added prior to incubation for 1 h at 4°C. For osmotic shock treatment, 100 ml of TE buffer was added, and the mixture was incubated for 30 min at 4°C and sonicated with a macrotip (model 250 sonifier; Branson, Danbury, Conn.) three times (1.5 min; duty cycle, 50%; output, 9). The cell suspension was sonicated after the addition of 5 ml of 10% (vol/vol) Brij-35 (Sigma) and centrifuged (5,000 × g, 30 min at 4°C) and sonicated (duty cycle, 50%; output, 9) two times. Opa protein was dissolved in 8 M urea and 50 mM glycine (pH 8). Ultracentrifugation at 40,000 rpm for 2.5 h at 4°C (Centrikon T 1170 Rotor 45-94; Kontron Instruments, Milan, Italy) was used to remove residual membrane fragments. Opa protein (10 mg/ml) dissolved in 8 M urea and 50 mM glycine (pH 8.0) was diluted 100-fold in refolding buffer containing 328 mM ethanolamine (pH 12) and 0.5% n-dodecyl-N,N-dimethyl-1-ammonio-3-propanesulfonate (SB12; Fluka). An SP-Sepharose-HP column (volume, 15 ml) (Amersham Pharmacia Biotech Europe GmbH, Freiburg, Germany) was equilibrated with 10 mM Tris-HCl and 0.5% SB12 (pH 7.5) (bufferA), loaded with approximately 10 mg of refolded OpaJ129, and washed twice with buffer A at pH 7.5 and 8.5. The proteins were eluted with a linear gradient of NaCl from 0 to 1 M in 120 ml. OpaB was refolded and purified by a similar procedure as also described elsewhere (10). To check folding and purification, sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed under seminative and denaturing conditions (data not shown). The folded and purified proteins were stored at −20°C.

Preparation of samples for immunization.

The denatured OpaJ protein was obtained by heating the folded protein for 30 min at 100°C prior to immunization. To stimulate the immune response, 20 μg of Quillaja saponin A (QuilA) (Isotec, Luleä, Sweden) was added to half of the samples containing purified refolded and denatured OpaJ protein. All purified protein samples were diluted in 10 mM Tris-HCl (pH 8) and 0.2% (wt/vol) SB12 (Fluka) purified as described by Dekker et al. (11). The Opa-negative H44/76 and OpaJ-positive variants were selected by colony blotting (37) with MAbs 15-1-P5.5 and MN20E12.70. Outer membrane complexes (OMCs) were isolated by sarcosyl extraction according to the protocol described by Davies et al. (9). The OMCs were diluted in 10 mM Tris-HCl (pH 8.0). The protein concentration of all samples was determined with the Pierce (Rockford, Ill.) protein assay with BSA as the standard.

Immunizations of animals.

Female BALB/c/Rivm mice at 6 to 8 weeks of age were used for immunization. Individual mice within groups of eight of approximately equal weight were immunized subcutaneously with 5 μg of purified protein or OMCs on days 0, 14, and 28. Mice were terminally bled on day 42, and sera were stored at −20°C.

ELISA.

Whole-cell ELISAs were performed according to the method of Abdillahi and Poolman (1). For ELISA with purified protein, flat-bottom 96-well microtiter plates were coated overnight at 37°C with 100 μl of a 5-μg/ml purified Opa protein solution in PBS. Antibody titers were measured for each individual serum sample. The titer is defined as the dilution of the serum where 50% of the OD_max in the assay is reached. The starting dilution of all sera was 1:100. In the total immunoglobulin G (IgG) and crossreactivity ELISA, GAM-HRP (1:5,000) was used to detect antibody binding. In the subclass-specific ELISA, IgG1, IgG2a, IgG2b, and IgG3 GAM-HRP MAbs (Southern Biotechnology Associates) (1:5,000) were used for the detection of the isotype-specific antibodies. In the whole-cell ELISAs, where whole meningococcal cells were used for coating, the Opa expression was tested in colony blotting (37) with MAbs 15-1-P5.5 and MN20E12.70. The color reaction with 3,3′,5,5′-tetramethylbenzidine (Sigma) and H₂O₂ (Merck, Darmstadt, Germany) was performed as described previously (1). The geometric means were calculated to average the antibody titers found in the sera from mice immunized with the same samples.

Bactericidal assays.

Serum bactericidal activity was determined as described previously (39) with some modifications. In short, twofold dilutions of heat-inactivated sera (30 min at 56°C) were incubated with baby rabbit complement (Pel-Freez Biologicals, Rogers, Ark.) (final concentration, 20% [vol/vol]) and 2.5 × 10² CFU of bacteria for 60 min at 37°C. Titers are expressed as the final dilution giving at least 90% killing of the inoculum.

Blocking assays.

The HeLa-Neo, -CEACAM1, and -CEA transfectants were a gift from F. Grunert (Genovac AG, Freiburg, Germany) and were described previously (5, 8). The HeLa transfectants were cultured in RPMI 1640 medium (Gibco, Logan, Utah) supplemented with 5% fetal calf serum (FCS) (Gibco) and 0.75 mg of Geneticin (Calbiochem Biosciences, Inc., La Jolla, Calif.)/ml. Cells were grown for 2 days to near confluency in 24-well tissue culture plates (Greiner, Cellstar, Kremsmünster, Austria) in 1-ml portions of RPMI 1640 medium with 10% FCS. PC2984-pMR05-opaB and PC2984-pMR05-opaJ were grown overnight on LB agar plates as described above (see “Bacterial strains, plasmids, and growth conditions”), swapped from the plate, and resuspended in PBS (SVM, Bilthoven, The Netherlands). The OD₆₂₀ of the bacterial suspension was measured and adjusted to 1.0. The sera were pooled, and 5 μl of pooled serum was added to 50 μl of bacteria. This suspension was immediately added to the cells and incubated for 4 h at 37°C, in a final volume of 500 μl of RPMI supplemented with 10% FCS and 25 μg of chloramphenicol/ml. After the incubation of bacteria and sera with the transfected HeLa cells, the cells were washed three times with PBS to remove nonadherent bacteria, lysed with 0.5 ml of 1% saponin (Sigma). Saponin fractions were plated on LB agar plates supplemented with 25 μg of chloramphenicol/ml. The incubation with nonbinding MAb (15-1-P5.5 in the assays using OpaJ⁺ E. coli and MN20E12.70 in the assays using OpaB⁺ E. coli) was referred to as the condition under which 0% blocking occurred to determine the anti-Opa-specific blocking.

Phagocytosis assays.

To characterize the H44/76 Opa⁺ and Opa⁻ variants used in this assay, 10⁸ heat-killed bacteria in a 10-μl volume were incubated with 5 μl of 10 μg of MAbs/ml (MN5C11G [anti-P1.16] [35], B306 [anti-Opc] [43], MN20E12.70 [anti-OpaB and -OpaD], and 15-1-P5.5 [anti-OpaA and -OpaJ] [45]) for 30 min at 4°C, washed twice, and incubated with CY3-labeled rabbit anti-mouse IgG MAbs (Jackson Immunoresearch). After washing, the resulting fluorescence of 10⁴ gated bacteria was determined by fluorescence-activated cell sorter (FACS) analysis in a FACSCalibur flow cytometer (Becton Dickinson, San Jose, Calif.).

The phagocytosis assay was performed as described by Vidarsson et al. (41). In short, human neutrophils (polymorphonuclear leukocytes [PMN]) were isolated with Ficoll (Pharmacia)-Histopaque (Sigma) gradients followed by hypotonic lysis in water (for 30 s at 4°C) of residual red blood cells. Sera were serially diluted in 1-ml polypropylene tubes (Micronics, Lelystad, The Netherlands). PMN (10⁵) were added along with 5 × 10⁶ Alexa488-labeled bacteria (see above) in a final volume of 100 μl and incubated at 37°C for 30 min. After washing, samples were resuspended in 300 μl of FACS buffer (PBS supplemented with 1% BSA and 0.1% azide), and the fluorescence intensities of PMN were measured by flow cytometry. For all experiments, cells from an FcγRIIa R/R131, FcγRIIIb-NA1/NA2 individual (PCR allotyped) were used.

Statistical analysis.

ELISA results are expressed as geometric means with the standard error of eight independent observations. The results of the blocking assay are expressed as geometric means with the standard deviation of three independent observations. The data were statistically analyzed by a one-tailed Student's t test; differences were considered significant at P < 0.05.

RESULTS

Opa repertoire of N. meningitidis strain H44/76.

Among the four Opa genes from strain H44/76, two could be identified as OpaB128 and OpaJ129 (M. I. De Jonge, H. J. Hamstra, L. van Alphen, J. Dankert, and P. van der Ley, unpublished data) which display very different versions of hypervariable region 1 and 2 (Fig. 1). This difference in amino acid sequence resulted in a different binding specificity, with OpaJ recognizing only CEACAM1 and OpaB recognizing both CEACAM1 and CEACAM5 (De Jonge et al., unpublished). The remaining two genes from H44/76, opaA and opaD, encode proteins that are nearly identical to either OpaJ or OpaB, respectively. These four Opa proteins are recognized by two different MAbs, MN20E12.70 and 15-1-P5.5. We determined the minimal epitopes recognized by these MAbs with 18-mer overlapping biotinylated peptides based on HV1 and HV2 of OpaB and OpaJ. The minimal epitope for 15-1-P5.5 was found to be GGPIIQ on HV2 of OpaJ, which is consistent with previous results from Hobbs et al. (19,) and for MN20E12.70, we determined the minimal epitope to be GIWQELK on HV1 of OpaB (Fig. 1).

FIG. 1.

Amino acid sequences of the variable regions of OpaB128 and OpaJ129. Minimal epitopes of MAbs MN20E12.70 and 15-1-P5.5 are underlined.

In order to target the Opa proteins to the outer membrane of E. coli, an in-frame fusion of opaJ and opaB was constructed by using the promoter- and signal sequence-encoding segment of the phoE gene of E. coli, as previously described (10). The Opa-expressing E. coli strains were used for immunization experiments, as test strains for Opa-specific antibodies in ELISAs and in an in vitro infection assay.

Immunogenicity of purified refolded and denatured OpaJ.

After purification of OpaJ, we immunized mice with the refolded and denatured form of the protein with and without the adjuvant QuilA. Opa-specific antibody titers were determined in whole-cell ELISAs with E. coli strain CE1265 expressing OpaJ and the homologous strain not expressing any Opa proteins as the immobilized antigen. Neither refolded nor denatured OpaJ protein without adjuvant evoked a significant humoral murine immune response. However, in the presence of QuilA, high Opa-specific antibody titers were obtained (Fig. 2). The antisera from the mice immunized with denatured OpaJ plus QuilA induced higher total IgG titers than refolded OpaJ plus QuilA, although this difference is not significant (P = 0.084). Similar levels were obtained by the positive control, H44/76 OpaJ⁺ OMCs (Fig. 2).

FIG. 2.

(A) Total IgG response measured in whole-cell ELISA with E. coli strain CE1265 expressing OpaJ (CE1265-opaJ) as the immobilized antigen and wild-type CE1265 as the negative control (CE1265).The data shown represent the geometric means and standard errors of 8 mice per group. (B) Total IgG response measured in whole-cell ELISA with H44/76 expressing Opa proteins detectable with 15-1-P5.5 MAb (H44/76 P5.5+) and H44/76 expressing Opa proteins detectable with MN20E12.70 MAb (H44/76 D2+) and H44/76 not expressing any Opa protein as a negative control (H44/76 Opa−). (C) Total IgG response measured in ELISA with purified OpaB and OpaJ. The data shown represent the geometric means and standard errors of 8 mice per group. Mice immunized with OMCs containing Opa protein giving lower titers than mice immunized with OMCs without Opa proteins are defined as nonresponders. Two of eight mice immunized with OMCs containing OpaJ did not respond; the titers from these nonresponders were included.

Immunogenicity of OpaJ presented in E. coli and N. meningitidis OMCs.

Besides immunizing with purified recombinant OpaJ protein, OpaJ was also presented as part of OMCs isolated from either E. coli or N. meningitidis. For E. coli, strain CE1265 with and without OpaJ expression was used for OMC isolation. We screened for Opa-negative and OpaJ-positive variants of N. meningitidis strain H44/76 by using the characterized MAbs in colony blotting and used them for OMC isolation.

In order to measure only anti-Opa antibodies and to avoid a nonspecific reaction in the analysis of sera from mice immunized with the E. coli OMCs, microtiter plates were coated with either 15-1-P5.5-positive, MN20E12.70-positive, or Opa-negative H44/76 variants. All total IgG titers with E. coli OMCs appeared to be strikingly low, approximately 10-fold lower than those obtained with either OpaJ⁺ meningococcal OMCs or purified OpaJ (Fig. 2). Some reaction was also found with the Opa-negative H44/76 variant, probably due to residual OpaJ expression. Although all opa genes are expected to be out of frame in the Opa-negative variants, a high frequency (10⁻³) of independent on and off switching has been observed during growth (18), making it difficult to obtain truly Opa-negative meningococci. Significant differences in the reaction of the sera against the three H44/76 variants could not be identified due to high variation within the groups, caused by a relatively high number of nonresponders (Fig. 2B). The specificity of the antibodies raised by the OpaJ-containing E. coli OMCs was confirmed by an ELISA with purified OpaB and OpaJ as the immobilized antigen (Fig. 2C).

Cross-reactivity with OpaB and antibody reaction against chimeric Opa proteins.

In order to measure the cross-reactivity of anti-OpaJ antibodies with OpaB, whole-cell ELISA against E. coli CE1265 expressing OpaB was also performed (Fig. 3). Cross-reactivity was generally low compared to total IgG titers against OpaJ. Very low cross-reactive antibody titers were found in the sera of mice immunized with refolded OpaJ plus QuilA; higher titers were found for denatured OpaJ and the H44/76 OpaJ⁺ OMCs (the titers were 761 and 845, respectively) (Fig. 3).

FIG. 3.

Antibody reaction against chimeric Opa proteins measured in whole-cell ELISA. The construction of the chimeric opa genes is depicted in panel A. The antibodies present in the sera of mice, immunized with Opa-negative and OpaJ-positive meningococcal OMCs and with purified refolded (Ref.) and denatured (Den.) OpaJ with QuilA, were tested for reaction with E. coli bacteria expressing OpaBj and OpaJb (B). The data shown represent the geometric means and standard errors of 8 mice per group.

To identify the most immunogenic region of the OpaJ protein, hybrids of the two opa genes, opaB and opaJ, were made and expressed in the outer membrane of E. coli. The ability of the immune sera to recognize the chimeric Opa proteins was determined after immunization with H44/76 Opa⁺ OMCs and refolded or denatured OpaJ with QuilA (Fig. 3B). In all cases, the highest levels of IgG antibodies raised against OpaJ were directed against the OpaJb protein containing HV2 from OpaJ and SV and HV1 from OpaB (Fig. 3A). A significantly higher titer was found against OpaJb than against OpaBj with the antisera raised against purified denatured OpaJ protein. (P = 0.004). The differences found with the other sera were not significant. Apparently, the antibodies were mainly directed against the HV2 of OpaJ, since low cross-reactivity with OpaB was found (Fig. 3B).

Antibody isotype distribution and bactericidal activity of sera.

The antibody isotype distribution of the response to OpaJ was determined by ELISA with anti-mouse IgG isotype-specific conjugates. Immunization with purified, refolded, and denatured OpaJ plus QuilA induced high levels of IgG1 subclass antibodies; the titers were 5,853 and 7,488, respectively (Fig. 4). A significant difference (P = 0.028) in the induction of IgG2a subclass antibodies between the immunizations with refolded and denatured OpaJ protein was observed, with much higher titers for the denatured protein. Surprisingly, high titers of the IgG3 subclass were detected in the antisera directed against E. coli OMCs but not with purified OpaJ (Fig. 4). The highest subclass titer in the antisera directed against E. coli OMCs containing OpaJ was found to be the IgG3 subclass (Fig. 4).

FIG. 4.

Antibody subclass-specific response to meningococcal OMCs and purified OpaJ plus QuilA with whole cells of CE1265-OpaJ as the immobilized antigen (A) and E. coli OMCs containing OpaJ with whole cells of H44/76 meningococci as the immobilized antigen (B). The data shown represent the geometric means and standard errors of 8 mice per group.

Despite the fact that denatured OpaJ with QuilA induced a significant IgG2a titer, 90% complement-mediated killing was only found at fivefold or lower serum dilutions. No differences in bactericidal titers were measured after immunization with H44/76 Opa⁻ and OpaJ⁺ OMCs, indicating that bactericidal antibodies were primarily directed against components other than OpaJ.

Antibody blocking activity in an in vitro infection assay.

To determine the blocking activity of anti-Opa antibodies, we incubated Opa-expressing E. coli bacteria with stably transfected HeLa cells expressing different CEACAM receptors. We found that the OpaJ protein specifically binds to CEACAM1 while OpaB binds to both CEACAM1 and CEA (De Jonge et al., unpublished). The antisera raised against those formulations of OpaJ giving the highest titers, i.e., H44/76 OpaJ⁺ OMCs and refolded or denatured OpaJ plus QuilA, were added to OpaB- or OpaJ-expressing cells of E. coli strain PC2984 before incubation with the transfected HeLa cells. For each group, sera from nonresponding mice were left out and the others were pooled. To select for specific anti-Opa blocking activity, the reference (0% blocking) was defined as the average of the number of CFU recovered after incubation with a nonbinding MAb. The sera raised against Opa-negative H44/76 OMCs was used as negative control. The blocking activity was measured as the percentage of nonrecovered CFU after incubation with sera, reflecting the reduction in HeLa cell association (adhesion and/or invasion) of the Opa-expressing E. coli bacteria. At a 1:100 dilution, over 80 to 90% blocking was found for the pooled sera from all three groups (Fig. 5). A clear concentration-dependent effect was observed since a strong reduction in the blocking activity was found after an additional 10-fold serum dilution. Strikingly, all the sera appeared to have cross-reactive blocking activity, since the sera raised against OpaJ could block the binding not only between OpaJ and CEACAM1 but also between OpaB and either CEACAM1 or CEA (Fig. 5).

FIG. 5.

Antibody blocking activity in an in vitro infection assay with sera raised against refolded and denatured OpaJ and meningococcal OMCs containing OpaJ. Two dilutions (1/100 and 1/1,000) of all sera were tested for blocking the binding between OpaJ and CEACAM1 (A), OpaB and CEACAM1 (B), and OpaB and CEA (C). The data shown represent the geometric means and standard errors of three independent experiments.

Effect of OpaJ-specific antibodies on phagocytosis.

Phagocytosis was measured in vitro by using ex vivo human PMN and heat-killed meningococci. Both H44/76 Opa⁻ and Opa⁺ variants expressed comparable amounts of PorA P1.16 but no Opc as measured by FACS analysis. The Opa⁺ variant expressed Opa proteins reactive with both MAbs MN20E12.70 and 15-1-P5.5 while we confirmed that the Opa⁻ variant did not express any Opa protein (Fig. 6a). It should be noted that in this assay no distinction can be made between different types of interaction of bacteria and PMN, i.e., only adhesion or also phagocytosis.

FIG. 6.

(a) Characterization by FACS analysis of PorA P1.16, Opc, OpaB, and OpaD (MAb MN20E12.70) and OpaA and OpaJ expression (MAb 15-1-P5.5) of the meningococcal H44/76 Opa⁺ and Opa⁻ variants used in the opsonophagocytosis assay. α, anti. (b) Opsono- and nonopsonophagocytosis of heat-killed Alexa 488-labeled meningococcal H44/76 Opa⁻ (A and C) and Opa⁺ variants (B and D) by human PMN in the presence of antisera raised against purified refolded and denatured OpaJ (A and B) and against H44/76 Opa⁺ and Opa⁻ OMCs (C and D). As a positive control for opsonophagocytosis, the two meningococcal H44/76 variants were incubated with PMN in the presence of MAb HMN12H2 IgG1, a humanized MAb specific for PorA. The experiments were repeated two times with similar results.

In the absence of sera, low phagocytosis was found for Opa⁻ variants but high phagocytosis was found for Opa⁺ variants, compatible with the Opa-CEACAM1 interaction. Sera raised against either OMCs or purified OpaJ had little or no effect on phagocytosis of Opa-negative H44/76. However, a dramatic reduction of the phagocytic index for Opa-positive bacteria was observed with sera from mice immunized with purified OpaJ or OpaJ-containing OMCs. The dominant effect measured in this assay is therefore the blocking of Opa-mediated association with PMN by antibodies against OpaJ protein rather than the opsonic activity of these antibodies (Fig. 6b).

DISCUSSION

The Opa proteins in N. meningitidis are encoded by a family of four genes that share a conserved framework interspersed by an SV region and two HV regions. Different functions have been attributed to the opacity proteins, but their main function is to mediate adhesion of meningococci to epithelial cells, which can subsequently lead to host cell invasion (14, 44). Although the Opa proteins have been shown to be able to induce bactericidal antibodies in humans (30), their high degree of sequence variability is a major impediment to vaccine application, as these bactericidal antibodies are generally highly type specific. However, despite this hypervariability, the number of receptors specifically targeted by the meningococcal Opa proteins is quite limited, as over 80% of them recognize human CEACAM1 (14, 31). The variable regions are located within the two largest surface-exposed loops, which are potentially involved in interaction with the extrabacterial environment. It can thus be expected that the sequence variation of these regions might be restricted by the need to preserve the specific receptor binding sites. At present, not much is known about the detailed molecular structure of the Opa-receptor complex. However, it is conceivable that antibodies directed against these specific receptor binding sites might block adhesion mediated by a broad range of distinct Opa proteins. In a previous study, a method for obtaining individual Opa proteins in highly purified form and native conformation has been described (10). We now used the OpaJ protein from N. meningitidis H44/76 obtained in this way to investigate its immunogenicity in mice and the functional activity and specificity of the resulting antisera.

Purified, refolded, and denatured OpaJ samples were prepared with and without the addition of the nontoxic adjuvant QuilA, since it is known that purified proteins need adjuvant stimulation in order to evoke an immune response (32). Without QuilA, no significant IgG titers were measured by ELISA, whereas in its presence, both refolded and denatured OpaJ protein induced high levels of total IgG. In contrast, immunization with the same OpaJ protein in an E. coli background formulated as OMCs resulted in approximately 10-fold-lower titers of Opa-specific IgG, in spite of high expression levels of Opa at the cell surface (10). Although the OMCs derived from E. coli and N. meningitidis contained the same amount of protein, the meningococcal OMCs elicited a much stronger response. The reduced immunogenicity of the E. coli OMCs is not due to the quantity of OpaJ but rather to the context or delivery. This might result from interference by immunodominant E. coli OMPs, e.g., OmpA (22). In any case, the heterologous surface expression in E. coli was designed as a tool to study the immunogenicity and functional activity of individual Opa proteins and not as an alternative antigen presentation system with vaccine applications.

The results of the whole-cell ELISA with the chimeric OpaBj and OpaJb proteins indicated that antibodies raised against OpaJ protein, both in OMCs and as purified protein, were mainly directed against the HV2 region. Apparently, the longest loop as predicted in a two-dimensional topology model (28, 36) contains the most immunogenic region. The fact that the highest antibody titer was found against this most variable region explains the low cross-reactivity with the heterologous OpaB, showing an approximately 10-fold-lower IgG titer.

IgG1 was found to be the dominant isotype in the response to purified OpaJ, as was also described for Opc (25). However, a striking difference was found in the induction of IgG2a between folded and denatured protein, as only denatured OpaJ also induced a relatively high IgG2a titer. This is in contrast to the results of Jansen et al. (21), who found a higher IgG2a titer with refolded PorA than with denatured PorA, also in the presence of QuilA. The refolding of OpaJ has been studied extensively (10), but correct folding is not a guarantee for the induction of bactericidal antibodies, as many other factors also play a role, for example, the accessibility or density of epitopes. Recently, an immunomodulatory function has been described for gonococcal Opa proteins. Their binding to CEACAM1 arrests the activation of T lymphocytes (6). Although the sequence divergence between human and murine CEACAM1 is probably too high to allow OpaJ binding to the mouse equivalent, it cannot be excluded that it still displays a similar immunomodulatory function in mice through binding to another receptor, thus explaining the striking difference in isotype distribution between native and denatured OpaJ. In agreement with this interpretation, it was previously found that refolded OpaJ is a more effective CEACAM1 binder than the denatured form (10). Only very low bactericidal activity was found for the sera raised against purified OpaJ, in spite of the high IgG2a titer, which is known to be an effective complement activator. Also, there were no differences in the bactericidal titers of the sera against H44/76 Opa⁺ and H44/76 Opa⁻ OMCs. Apparently, the OpaJ protein is not such an effective inducer of bactericidal antibodies as, for instance, PorA, at least not in mice. This agrees with the results of Sacchi et al. (31a), who found no differences in bactericidal titers in sera from mice immunized with OMV preparations differing in the presence or absence of class 5 proteins (Opa and Opc).

Blocking of the Opa adhesin function might be an alternative protective mechanism to bactericidal action. Therefore, the antibody blocking activity was measured in a highly defined in vitro infection assay. This assay has been used to determine the binding specificity of OpaB and OpaJ from N. meningitidis H44/76, both expressed on the cell surface of E. coli, through their binding to transfected HeLa cells expressing the different CEACAM proteins. Whereas in the infection assay OpaJ only binds to CEACAM1, OpaB recognizes both CEACAM1 and 5 (unpublished data). The sera with the highest anti-OpaJ titers, obtained either with purified OpaJ or OpaJ in H44/76 OMCs, all displayed clear blocking of the OpaJ-CEACAM1-mediated adhesion. Interestingly, cross-reactive blocking of the heterologous OpaB protein with both CEACAM 1 and 5 was also found, in spite of the approximately 10-fold-lower whole-cell ELISA titers against OpaB than against OpaJ. Apparently, antibodies directed against epitopes common between these distinct Opa proteins are mainly responsible for the observed functional blocking.

Our in vitro adhesion assay, although highly defined, uses Opa expression in E. coli and CEACAM expression in HeLa cells. We also studied the interaction of meningococci with professional phagocytes, which may be a more natural setting. As expected, only Opa-expressing meningococci associated with human neutrophils in a nonopsonic manner. An important role for Opa proteins in the interaction of meningococci with human monocytes has been described previously (15, 29). In our assay, we used human neutrophils homozygous for the FcγRIIa-R131-encoding allele, which has a relatively high affinity for mouse IgG1 (38). It might therefore be expected that our anti-OpaJ sera should not only be capable of blocking the Opa-CEACAM1 interaction but also be able to mediate opsonophagocytosis via the human Fc receptors. However, our results indicated that addition of the sera resulted only in blocking of the nonopsonic interaction between the bacteria and the neutrophils. As the number of neutrophil-associated bacteria was much higher for the OpaJ⁺ strain than for the OpaJ⁻ strain, the observed blocking must work through the interaction between OpaJ and its neutrophil receptor, presumably CEACAM1. With the sera raised against the meningococcal OMCs, blocking activity was much higher for the OpaJ⁺ than the OpaJ⁻ OMCs, again demonstrating the dominant role for Opa in the nonopsonic phagocytosis. Only at higher serum concentrations could blocking antibodies against other outer membrane components than Opa or Opc (not expressed by our strains) come into play. The difference in phagocytic activity between the sera raised against pure OpaJ and against OMCs containing OpaJ is also probably due to the opsonophagocytic activity of antibodies raised against other OMPs of N. meningitidis.

In conclusion, we demonstrated that purified recombinant OpaJ protein with QuilA is highly immunogenic in mice, whereas OpaJ presented in E. coli OMCs is less immunogenic. Although hardly any bactericidal and opsonophagocytic antibodies were found, OpaJ could induce antibodies with blocking activity, both for homologous and heterologous Opa-CEACAM interactions. This implies the presence of conserved surface-exposed epitopes on Opa proteins with distinct HV1 and HV2 regions. If similar cross-reactive blocking antibodies can also be induced with purified Opa protein at mucosal surfaces, this would constitute a novel protective mechanism for meningococcal vaccines by blocking the adhesion to the nasopharyngeal epithelium.