Dynamics of the Murine Humoral Immune Response to Neisseria meningitidis Group B Capsular Polysaccharide

Abstract

Immunization with Neisseria meningitidis group B capsular polysaccharide (CpsB) elicited responses in adult mice that showed the typical dynamic characteristics of the response to a thymus-independent antigen, in contrast to the thymus-dependent behavior of antibody responses to CpsC. The former had a short latent period and showed a rapid increase in serum antibodies that peaked at day 5, and immunoglobulin M (IgM) was the major isotype even though IgG (mainly IgG2a and IgG2b) was also detectable. This response was of short duration, and the specific antibodies were rapidly cleared from the circulation. The secondary responses were similar in magnitude, kinetics, IgM predominance, and IgG distribution. Nevertheless, a threefold IgG increase, a correlation between IgM and IgG levels, and dose-dependent secondary responses were observed. Hyperimmunization considerably reinforced these responses: 10-fold for IgM and 300-fold for IgG. This favored isotype switch was accompanied by a progressive change in the subclass distribution to IgG3 (62%) and IgG1 (28%), along with the possible generation of B-cell memory. The results indicate that CpsB is being strictly thymus independent and suggest that unresponsiveness to purified CpsB is due to tolerance.

The capsular polysaccharide (Cps) of Neisseria meningitidis group B (CpsB), the major cause of meningococcal disease in developed countries (38), is a linear homopolymer of α(2→8)-linked sialic acid on host sialogangliosides and sialoproteins (12, 16) causes immunological tolerance to sequential CpsB epitopes, with the anti-CpsB antibodies being mainly, if not solely, directed against conformational determinants preferably expressed by chains of eight or more residues (10). The conformational antigenic nature and metastable spatial structure of CpsB (10, 19), in combination with its neuraminidase sensitivity, tendency to internal lactonization, and intramolecular self-cleavage under mild acidic conditions (22, 29), were proposed to explain its poor immunogenicity (35). According to this hypothesis, the interaction of CpsB with B cells is transitory and therefore unable to elicit an antibody response (34). Alternatively, the high expression of longer sialic acid polymers (>12 residues), having the same α(2→8) linkage in polysialylated glycoproteins of vertebrate fetal tissues as well as limited areas of the adult neural system (21, 42), has been proposed to induce tolerance also to the conformational epitopes of CpsB (11). A feasible mechanism for inducing and maintaining tolerance, however, is not known. In any event, the poor immunogenicity of CpsB is associated with the α(2→8) linkage. Purified CpsC, a homopolymer of α(2→9)-linked sialic acid, has been shown to be immunogenic in mice (48).

Bacterial Cps complexed to protein carriers induces long-lasting immunoglobulin G (IgG) antibody responses in young children and mice, which is indicative of the Cps conversion to a T-cell-dependent (TD) antigen (18). In contrast, CpsB conjugated to tetanus toxoid (3, 8, 20) or complexed with meningococcal outer membrane proteins (OMPs) (23, 24) is able to induce only low levels of CpsB-specific IgM. In these responses, however, CpsB-specific IgG was detectable (3, 8, 23). Since in simple terms protection from these infectious agents is due to the presence of circulating specific antibodies (13) and bearing in mind that an artificial IgG immune response may initiate an autoimmune process (11), we studied the evolution over time of the serum antibodies and changes in isotype distribution obtained by immunization with the native form of CpsB—namely, live N. meningitidis—in order to further explore the underlying mechanisms in the generation of the immune responses to this peculiar autoantigen which has both epitopes disseminated in the host and epitopes of ontogenetic and topologically restricted expression, a situation reproduced in the mouse model.

MATERIALS AND METHODS

Animals.

BALB/cO1a, C57BL/10, C3H/HeN, and Swiss mice, 9 to 10 weeks old at the onset of experiments, were from our own colony at the Centro Nacional de Microbiología (CNM; Majadahonda, Madrid).

Bacterial strains and media.

Meningococcal strains were from the culture collection of the National Reference Centre for Neisseria (Centro Nacional de Microbiología) and were stored frozen in skim milk for later use. Overnight cultures in modified Frantz medium (27) were used in the preparation of Cps, and tryptic soy broth (Difco Laboratories, Detroit, Mich.) was used to obtain OMP vesicles. For mouse challenge, bacterial cultures grown for 16 h on Columbia blood agar plates were scraped and suspended in phosphate-buffered saline (PBS) to the required density (10⁹ CFU/ml corresponds to an optical density at 650 nm of ca. 1.0). The numbers of viable bacteria in the inoculum were confirmed by plating serial dilutions on blood agar plates. To prepare formalin-fixed bacteria, the suspensions were made up to 0.5% in formalin and incubated for 1 h at 20°C.

Bacterial antigens.

Purified CpsB, CpsC, and Cps29E from N. meningitidis B16B6 [B:2a:P1.2:L2(3)], 8148 (C:2b:P1.2), and MA5739 (29E), respectively, were prepared as described by Gotschlich et al. (15). The total sialic acid content of purified Cps was measured by the resorcinol-HCl method (47). Protein content (<0.05%) was measured by the bicinchoninic acid method (43), lipopolysaccharide (LPS) (<0.05%) was measured by silver staining of sodium dodecyl sulfate-polyacrylamide gels (49), and nucleic acids (<0.6%) were measured by absorbance at 260 nm. The OMP from N. meningitidis B16B6 was prepared by lithium chloride-sodium acetate extraction (7).

MAbs and antisera.

The production of polyclonal ascitic fluid against N. meningitidis group B and of five IgM (MGB11, MGB12, MGB13, MGB14, and MGB15) and two IgG2b (MGB9 and MGB16) murine anti-CpsB monoclonal antibodies (MAbs) has already been described (5). The IgG2b MAbs were purified by protein A affinity chromatography, and the IgM MAbs were purified by precipitation with polyethylene glycol 6000 from ascitic fluids raised in BALB/c mice (37). The purified IgM was now free of contaminanting IgG as detectable by enzyme-linked immunosorbent assay (ELISA). The protein content of the purified material was assayed by the bicinchoninic acid method (43). Biotinylated goat polyclonal antibodies specific for each mouse light- and heavy-chain isotype were purchased from Southern Biotechnology Associates (Birmingham, Ala.). The specificities of these reagents were confirmed before use in ELISA by using plates coated with MAbs of all of the isotypes purified from culture supernatants.

Immunization protocols.

Live N. meningitidis group B strain B16B6 was used for all immunizations. In order to compare primary and secondary responses, age-matched and randomly distributed littermates of BALB/c mice were immunized intraperitoneally (i.p.) with 2 × 10⁸ CFU or 100 μl of PBS. Three weeks later, both groups were challenged with 2 × 10⁸ CFU. An additional group of mice primed with 2 × 10⁸ CFU received 10¹⁰ CFU in a second challenge in order to study dose effects. Mice were bled 5 and/or 6 days after each immunization.

The evolution of meningococcus-specific antibodies in serum was monitored by daily bleeds from the retro-orbital plexuses of 10 animals subjected to this immunization schedule. In order to avoid experimental artifacts due to the clearance of specific antibodies produced by serial bleedings, a similar experiment was performed by bleeding different groups of five mice each on every postimmunization day. The mean residence time of the anti-CpsB antibodies in serum was estimated graphically (17), with the peak of serum-specific antibodies being considered time zero. For comparative purposes, mouse groups were immunized with strain 8148 of N. meningitidis group C on the same schedule. On each day postimmunization, groups of three mice were splenectomized, and the spleens were homogenized, suspended in fetal calf serum supplemented with 10% dimethyl sulfoxide, and frozen in liquid nitrogen until tested by enzyme-linked immunosorbent spot assay. The persistence of infection was determined by plating 10-fold serial dilutions of blood, peritoneal washings, and liver and spleen homogenates collected at different times after immunization on blood agar.

The response to hyperimmunization was studied in mice immunized i.p. twice a week over a 4-week period with 2 × 10⁸ live CFU. All mice were bled at days 0, 4, 12, and 20, and following the last immunization (day 25), different random groups of five mice were bled on each day. After a 4-week rest period, the mice were boosted (day 52) with 2 × 10⁸ formalin-fixed CFU given intravenously (i.v.) or given intrasplenically (i.s.) as described by Spitz et al. (46). Groups were bled daily and splenectomized, and the spleens were frozen. As a control, groups of unimmunized littermates were subjected to the i.v. or i.s. bacterial immunization (day 52).

ELISA procedures.

The antibodies specific to Cps were detected by using Nunc Maxisorp plates sensitized for 1 h at 37°C with 0.5 μg of poly-l-lysine per well. The plates were washed five times with PBS, coated overnight at 4°C with 1 μg of CpsB or CpsC in PBS per well, and then blocked with 10% skim milk for 2 h at 37°C. Wells coated with Cps29E, an antigenically unrelated Cps, were used as antigen blanks in both cases. The washed plates had mouse sera, twofold serially diluted in PBS containing 0.05% Tween 20 and 1% bovine serum albumin (BSA), dispensed in blank and CpsB- or CpsC-coated wells of six replicate plates (one for each isotype). They were incubated overnight at 4°C. Since IgM is the major isotype in anti-CpsB antibody responses and at high concentrations inhibits the IgG binding to CpsB (5), the sera were diluted 1/5 in PBS with 1% BSA and 0.1 M 2-mercaptoethanol and incubated for 15 min at 20°C for the IgG determinations before being diluted and dispensed in wells (41). In contrast to those of IgG, the interchain disulfide bonds of the IgM structure are highly sensitive to mild reduction. The depolymerization of IgM produces a drastic drop in avidity so that even monomeric IgM and heavy-plus-light-chain fragments remaining in the sample become undetectable by ELISA. Thus, reduction of high-titer anti-CpsB sera eliminated their IgM anti-CpsB reactivity, while IgG reactivity increased. After five washes with PBS, 100 μl of the corresponding biotinylated polyclonal anti-mouse isotype Igs was added to the wells and incubated for 1.5 h at 37°C. The anti-IgM reagent was used at a dilution of 1/2,000, while anti-IgA and the different anti-IgG subclass reagents were used at their corresponding equivalence dilutions with an anti-κ secondary reagent (see below). The plates were further incubated for 1 h at 37°C with a 1/1,000 dilution of a streptavidin-horseradish peroxidase conjugate and developed with 0.15 M citrate-phosphate buffer (pH 5) containing 0.2% o-phenylenediamine and 0.003% H₂O₂. The reaction was stopped after 4 min by adding 50 μl of 4 N H₂SO₄ per well, and the absorbance was read at 492 nm.

The ELISAs described above were transformed into a quantitative assay by introducing appropriate standards in each plate. For anti-CpsB IgM quantitation, twofold dilutions of equimolar mixtures of five IgM anti-CpsB MAbs (MGB11 to MGB15), ranging from 1 to 250 ng/ml, were dispensed in CpsB-coated wells and used as standards. These were processed together with the sample wells.

For all IgG subclass and IgA anti-CpsB quantitations, the only standard used was an equimolar mixture of two anti-CpsB IgG2b MAbs (MGB16 and MGB9), ranging from 3 to 500 ng/ml, dispensed in twofold dilutions in CpsB-coated wells. The use of a single subclass standard to quantitate heterologous IgG subclasses is possible by using the approach of equivalence dilutions of secondary reagents, which was initially developed to quantitate IgG subclasses in systems lacking calibrated antigen-specific standards for each IgG subclass (5). The method was based on the precalibration of each secondary anti-isotype reagent, with respect to the anti-κ reagent, by ELISA with primary antibodies bearing κ light chains bound to their respective antigen. For the anti-IgG1 secondary reagent, a mixture of MAb IgG1κ titrated in plates coated with its specific antigen (unrelated to CpsB [e.g., tubulin]) was tested in parallel with several dilutions of the anti-IgG1 and anti-κ secondary reagents. The dilutions of the anti-IgG1 and anti-κ reagents which gave the same absorbance values and parallelism of the respective titration curves of the primary antibodies were considered the equivalence dilutions. This process was repeated for each anti-isotype reagent to obtain coupled equivalence dilutions between each anti-isotype reagent and the anti-κ secondary reagent. Considering that every IgG (monomeric Igs) on binding to antigen displays a similar array of its κ epitopes, a unique IgGκ standard (e.g., the monoisotypic IgG2b standard), irrespective of its subclass composition, will give the same signal with the precalibrated dilution of the anti-κ secondary reagents as an identical concentration of a particular isotype in the sample reacting with the corresponding equivalence dilution of the anti-isotype secondary reagent. Hence, in this quantitative ELISA, the monoisotypic IgG2b standard was incubated with the anti-κ secondary reagent, and the samples were incubated with the antisubclass specific secondary reagents, both at the equivalence dilutions for the isotype being quantitated. The concentrations of anti-CpsB antibodies of each isotype expressed in weight units were obtained by interpolation of the absorbance values, after subtracting antigen blank absorbances, into their respective standard curves constructed by using the logistic-log model (39). The serum IgG content is the sum of the concentrations of each subclass.

Quantitation of antibody concentrations specific for CpsC were carried out for each isotype by using the same method and standards as for anti-CpsB antibodies. Heterologous antigens for standards and samples tested have been used in other systems without excessively erroneous results (30). However, due to the greater avidity of the anti-CpsC antibodies (28), it is advisable to express anti-CpsC antibody concentrations obtained by interpolation into curves constructed with the CpsB-specific standards as microgram equivalents per milliliter. Plates coated overnight at 4°C with (per well) 1 μg of purified OMP from the B16B6 strain diluted in PBS were used to monitor the time courses of specific serum antibodies to TD bacterial antigens. For these isotype determinations, the polyclonal ascitic fluid against this strain was used as a standard in wells coated with purified OMPs. The anti-OMP antibody levels were expressed in arbitrary units, with 1 antibody OMP unit being considered the content of a particular isotype in the ascitic fluid standard. In other aspects, this protocol was similar to the quantitative CpsB ELISA.

Enzyme-linked immunosorbent spot assay.

Nunc Maxisorp 96-well plates coated with Cps as described above were blocked with 200 μl of PBS containing 1% BSA per well for 2 h at 37°C. Reagent volumes of 100 μl/well were used in all subsequent steps. After three washes with PBS, cell suspensions in RPMI 1640 (Difco) containing 1% BSA were added to the wells. The plates were covered and incubated at 37°C for 5 h in an atmosphere containing 5% CO₂. Usually four 1/3 duplicate serial dilutions starting from 2 × 10⁶ cells/well were tested. Cell suspensions used were from spleens which had been stored in liquid nitrogen. This enabled the simultaneous testing of samples collected on different days and minimized the effects of interassay variability. Cell viability averaged 80% as estimated by trypan dye exclusion. After cell incubation, the plates were washed five times with cold PBS containing 0.025% Tween 20 and incubated overnight at room temperature, in a humid chamber, with the biotinylated secondary reagents specific for IgM or IgG diluted 1/1,000 in PBS containing 1% BSA. The plates were washed again and incubated for 1 h at 37°C with a 1/1,000 dilution of the streptavidin-peroxidase conjugate. After another washing cycle, the enzyme reaction was developed with a modification of the peroxidase substrate solution described by Czerkinsky et al. (6). Briefly, a mixture of 10 mg of 3-amino-9-ethylcarbazole and 30 mg of 4-chloronaphthol was disolved in 1 ml of dimethylformamide and added to 40 ml of 0.05 M acetate buffer at pH 5. The solution was filtered (0.22-μm-pore-size filter) and made up to 0.001% H₂O₂. Introduction of 4-chloronaphthol in the substrate mixture produced purple spots. This increased the contrast and allowed a greater number of spots to be detected. The reaction was developed for 30 min, and the number of spots was recorded by examining the plates at a magnification of ×40. The mean secretory rate of specific spot-forming cells (SFC) was expressed as the quotient of the serum-specific antibody concentration on day n and the SFC number on day n − 1. The assay specificity was confirmed in three ways: (i) the number of spots diminished to 2% when the cell incubation was carried out at 4°C, (ii) cycloheximide (2 to 0.5 mM) treatment of cells abrogated spot generation, and (iii) the incorporation of CpsB (1 μg/well) into the cell supernatants drastically diminished spot diameters and the number of spots, by 46% for IgM and 96% for IgG. No specific spots were detected in wells coated with heterologous Cps.

Statistics.

Results are presented as the geometric mean (± standard deviation) of individual concentrations, once preimmune levels were subtracted. Antibody levels were compared by using Student’s two-tailed t test. P values of <0.05 were regarded as significant.

RESULTS

Primary and secondary anti-CpsB antibody response to bacterial immunization.

The anti-CpsB antibody responses were compared in mice, of the same litter, receiving their primary or secondary bacterial immunization at 12 weeks of age. These responses were compared with those produced following primary immunization, 3 weeks before, in the group now receiving secondary immunization. At these ages the ability to produce an anti-CpsB antibody response is fully developed. The serum IgM anti-CpsB antibody level 5 days after secondary immunization at 12 weeks of age (48 ± 12 μg/ml) did not differ significantly from that produced by either the same mice following primary immunization at 9 weeks of age (46 ± 6 μg/ml; P = 0.62) or the group that received primary immunization at 12 weeks of age (54 ± 12 μg/ml; P = 0.64). Instead, the IgG levels after secondary immunization were about two- to threefold higher (0.54 ± 0.16 μg/ml) than those obtained on primary immunization at 9 weeks (0.21 ± 0.18 μg/ml; P = 0.025) or at 12 weeks (0.21 ± 0.1 μg/ml; P = 0.019). A similar increase in IgG was observed in sera collected 6 days after each immunization. Hence, the response to the CpsB expressed on the bacterial surface showed T-cell-independent (TI) behavior, similar to the response to purified Cps, except for increased IgG levels. Nevertheless, the purified CpsB was not immunogenic. One or two immunizations with 1 to 10 μg of CpsB did not significantly increase the preimmune anti-CpsB antibody level. A secondary i.s. immunization with 10 μg of the CpsB gave the same results. Therefore, the bacteria rendered the CpsB immunogenic.

Dose effect on the secondary antibody response.

A larger bacterial dose (10¹⁰ CFU) for secondary immunization elicited higher IgM and IgG anti-CpsB antibody responses than a lower one. This increase was 2- to 4-fold higher for the IgM levels (122 ± 30 μg/ml) and 10- to 15-fold higher for the IgG levels (3 ± 0.7 μg/ml) than those obtained following primary immunization.

Kinetics of serum anti-CpsB antibody response.

The profile of the anti-CpsB serum antibodies induced by an immunization with group B N. meningitidis clearly differed from that observed with anti-OMP in the same animals and from that of the anti-CpsC antibodies in mice of the same age immunized with group C N. meningitidis (Fig. 1). Moreover, the serum profiles of anti-OMP and anti-CpsC antibodies over time were similar, showing the typical characteristics of a secondary response to a TD antigen. These characteristics are a considerable increase in specific antibody levels, a shorter latent period, and an increased IgG/IgM ratio. The anti-CpsC secondary response was mainly IgG (>90%). Thus, the CpsC presented on the bacterial surface behaved like a TD antigen.

FIG. 1.

Kinetics of primary and secondary antibody responses to the meningococcal antigens CpsB, CpsC, and OMPs. The antibody concentration for each isotype and specificity in pooled sera collected at different times from the same animal groups (n = 10) after one or two i.p. immunizations (indicated by arrows) with 2 × 10⁸ CFU of live group B (upper panel) or group C (lower panel) N. meningitidis was estimated. Note that the IgG and IgM scales differ. eq, equivalents.

In contrast, the primary and secondary antibody responses to CpsB showed no differences in IgM antibody levels or the time course of specific Ig, thus confirming the TI behavior of CpsB (Fig. 1). The IgM anti-CpsB response had a short latent period (1 to 2 days) and reached a maximum only 5 days after each immunization (Fig. 1 and 2). Levels in serum on average tripled daily. IgG showed a similar profile but was delayed by 1 day with respect to IgM; it was detectable at 2 to 3 days with maximum levels by 6 to 7 days following both primary and secondary stimulations. The maximum anti-CpsB activity in serum coincided with the latent period of primary anti-CpsC and anti-OMP antibody responses (5 to 7 days). Thereafter the anti-CpsB antibody level decreased to one-third over the next 2 days. Thus, anti-CpsB IgM was no longer detectable after 14 to 16 days, and anti-CpsB IgG was no longer detectable after 12 to 15 days. Moreover, considering that once the maximum level in serum was reached, no more newly synthesized anti-CpsB antibody entered the bloodstream, the mean residence time can be estimated as 2.8 days for both IgM and IgG (IgG2a). Because SFC were detectable later than day 5 to 6 (see below), this value is still an overestimation. This anti-CpsB antibody profile was (i) observed in serial bleedings from the same (Fig. 1) or different (Fig. 2) mice, (ii) independent of the antibody level reached, (iii) independent of the bacterial strain or dose inoculated, (iv) independent of the ontogenetic response development, and (v) reproduced in C57BL/10, C3H, and Swiss mice (data not shown).

FIG. 2.

Kinetics of serum anti-CpsB IgM, IgG, and IgA (left) and IgG subclasses (right) after secondary immunization with 10¹⁰ CFU of N. meningitidis group B in BALB/c mice. Different groups of five mice each were used on each day postimmunization. Values are expressed as the geometric mean ± standard deviation for individual concentrations. For clarity, each IgG subclass is shown as the mean concentration in sera where the particular subclass was quantifiable.

Persistence of the meningococcal infection.

The rapid decrease in serum anti-CpsB antibody levels could be due to the persistence of the bacteremia or to a localized infection becoming established. Mice inoculated with >10⁸ CFU displayed within 1 h symptoms associated with septicemia, including lachrymation, lethargy, hypothermia, hypoxia, anorexia, and diarrhea, whose severity and rapidity of onset increased with the dose administered. However, most of the mice controlled their bacteremia within the first 8 h (<100 CFU). This seemed to be the critical period. At 16 to 24 h all hemocultures were negative and bacteria could not be isolated from spleen or liver homogenates. However, colonization of the peritoneal cavity remained in 20% of mice at 16 h postinoculation (10⁴ to 10³ CFU), being finally eliminated at 24 to 48 h. Following this period, neither signs nor symptoms of infection were manifest. Since the infection was controlled within the latent period of the anti-CpsB response, this was unlikely to have caused the antibody clearance.

For doses of >10¹⁰ CFU, the infection progressed more rapidly and nearly 80% of the mice died within 24 to 48 h. However, mice that had been subjected to a previous bacterial immunization with 2 × 10⁸ CFU experienced mild asymptomatic infection, which was controlled sooner. This indicated that protective immunity was elicited on primary infection.

Kinetics of the anti-CpsB antibody response during hyperimmunization.

The profile of development of anti-CpsB antibody levels in adult mice during hyperimmunization (Fig. 3) can be described as a succession of antibody waves, with a mean amplitude of 10 days and maxima 5 at 6 days after each immunization. This profile was similar to that obtained after a single immunization and was reproduced after a booster given 1 month after hyperimmunization ended. The anti-CpsB response after an i.s. boost seemed to appear sooner than that elicited by an i.v. boost, with the specific antibodies reaching a maximum at day 4 (Fig. 3).

FIG. 3.

Kinetics of anti-CpsB serum antibody response in adult mice during hyperimmunization. Nine-week-old BALB/c mice subjected to hyperimmunization with group B N. meningitidis (A) received 2 × 10⁸ formalinized CFU i.v. (B) or i.s. (C) 1 month later. Closed symbols, responses of control mice. Serial immunizations are indicated by arrows. Different groups of five mice were used each day. The anti-CpsB antibodies are included as the geometric mean of the individual concentrations ± standard deviation, and the anti-OMP antibodies are included as concentrations in pools of the same sera. The amplitudes of the antibody waves are indicated by transverse lines.

Hyperimmunization effects on the anti-CpsB antibody response.

The effect on the antibody responses induced by a prior antigenic stimulation was greater when hyperimmunization replaced a single primary immunization (Fig. 3). Five days after an i.v. challenge 1 month after the hyperimmunization regimen ended, serum IgM (85 ± 16 μg/ml) and IgG (15 ± 3 μg/ml) levels were 10- and 300-fold higher, respectively, than those obtained following a single immunization in control mice (8 ± 3 μg of IgM per ml, with barely detectable IgG [0.04 ± 0.1 μg/ml]). At the peak of the IgG anti-CpsB response (day 6), these levels were 800-fold higher. No anti-CpsB antibodies were found at the time of challenge; hence, differences in antibody levels between mouse groups were unlikely to be due to circulating anti-CpsB antibodies elicited during hyperimmunization. Finally, the hyperimmunization challenge was more evident if the boost was given i.s. instead of i.v. (Fig. 3), but IgM and IgG responses were lower when elicited by the former route. A single i.s. immunization induced very low levels of anti-CpsB IgM (1 ± 2 μg/ml) in control mice. This was particularly noticeable because i.s. inoculation elicited higher responses to the TD OMPs.

Splenic SFC response kinetics.

The time course of anti-CpsB antibody-secreting cells in the spleen was studied (Fig. 4). IgM SFC were already detectable at day 2 and rapidly increased in number to a maximum on day 4. In spite of the similarity to the IgG SFC profile, the appearance of IgG SFC seemed to be delayed, and their numbers did not increase until a considerably higher threshold number of IgM SFC was reached. At the peak, their numbers ranged from 11,000 IgM and 1,200 IgG SFC/spleen, following primary immunization with 2 × 10⁸ formalinized CFU (Fig. 4C), to 31,000 IgM and 8,000 IgG SFC/spleen after the secondary immunization with 10¹⁰ live CFU (Fig. 4A). Subsequently, their numbers decreased to the preimmune level of 2.3 ± 2 IgM SFC/10⁶ cells (170 ± 140 SFC/spleen) at day 6 to 7. No IgG SFC were detected in preimmune spleens. The reduction in SFC number was accompanied by an increase in spot diameter from day 4. This major peak coincided with the peak of total splenocyte numbers between days 2 and 6. Finally, when the assay times were further extended to day 7, a secondary peak of 840 ± 190 IgM SFC/spleen over the preimmune level was detected on day 10 (Fig. 4A) (P = 0.004).

FIG. 4.

Kinetics of the splenocytes secreting antibodies specific for CpsB (——) or CpsC (····). Time courses of the IgM SFC (○) and IgG SFC (•) expressed as the geometric mean ± standard error in spleens of BALB/c mice during the secondary response to i.p. immunization with 1 × 10¹⁰ CFU of live N. meningitidis group B (A) or 2 × 10⁸ CFU of live N. meningitidis group C (B), following a primary immunization with 2 × 10⁸ CFU of formalinized group B meningococci (C), and in hyperimmunized mice after the last immunizations (days −2 and 0) (D) and after an i.v. boost of formalinized group B meningococci 1 month later (E) are shown. The mice assayed are included in the dynamics of the serum antibodies depicted in Fig. 1 (B and C), Fig. 2 (A), and Fig. 3 (D and E).

The secondary splenic SFC anti-CpsB response showed minor differences with respect to the primary response. This could be summarized as a more rapid appearance of IgG SFC (day 2) and an increase in total SFC (Fig. 4A). Following hyperimmunization, the SFC also peaked 4 to 5 days after each immunization but maintained a higher background level (P = 0.0005) than the preimmune one (Fig. 4D). This was reproduced following a further boost, where the number of IgM SFC remained higher on day 6 and the IgG SFC peak shifted to day 5 (Fig. 4E). The secondary response to CpsC was earlier than that with CpsB, reaching a maximum of IgG and IgM SFC on day 3 (Fig. 4B). These SFC are specific for CpsC, with the number of anti-CpsB SFC detected (1.5 ± 0.6 IgM SFC/10⁶ cells) being similar to preimmune levels (P = 0.3). In contrast to the anti-CpsB response, the number of anti-CpsC IgG SFC (15,000/spleen) was higher (P = 0.009) than that of the IgM SFC (3,700/spleen). Nevertheless, despite the greater differences in serum antibody responses to CpsB and CpsC, around day 4 both their numbers and mean secretory rates (0.7 ng of IgM and 11 ng of IgG per splenocyte) were not spuriously higher than those obtained against CpsB with a similar schedule (1.6 ng of IgM and 0.7 ng of IgG per splenocyte). Finally, for mice subjected to the same schedule, a direct ratio between serum anti-CpsB antibody levels and SFC numbers or between IgM and IgG SFC can be obtained. However, the ratios were different depending on the schedule, suggesting variations in the secretory capacity of the anti-CpsB SFC elicited.

Distribution of Ig classes.

The antibody response to CpsB was mainly IgM, although there was a significant contribution of IgG which varied according to immune status and the postimmunization day considered. Thus, the percentage of IgM 5 days after stimulation showed a significant decrease (P < 0.0001) between primary (99.3 ± 0.5%) and secondary (97.6 ± 0.8%) responses and between days 6 (77 ± 6%) and 7 (90 ± 4%). The serum IgM and IgG levels were correlated for secondary (r = 0.901; P < 0.001; n = 25) but not for primary (r = 0.29; P > 0.1) responses. In the secondary response, the anti-CpsB levels increased with the dose but the IgM/IgG ratio remained constant. A similar correlation between IgM and IgG levels was found in hyperimmune sera (r = 0.860; P < 0.001; n = 25), but the regression line obtained with these sera collected 7 to 10 days after the last immunization showed a greater slope (b = 0.243) than that obtained 7 days after secondary immunization (b = 0.092), indicating a larger IgG contribution (18 ± 6%; P < 0.0001). Finally, IgA was detected in only 6% of sera, and its level seldom exceeded 1 μg/ml.

Distribution of IgG subclasses.

The IgG anti-CpsB response after primary or secondary immunization was mainly IgG2a (74%) and IgG2b (18%), whereas in hyperimmunized mice it was mainly IgG3 (62%) and IgG1 (28%), as previously reported (5). This change in subclass distribution was also observed progressively during hyperimmunization and was maintained in response to a single immunization 4 weeks later. When IgG2a and IgG2b were present in the same sera, these levels correlated neither with those observed following a single immunization (r = 0.18; P > 0.1) nor with those observed after hyperimmunization (r = 0.413; P > 0.1). In contrast, IgG1 and IgG3 levels correlated (r = 0.79; b = 1.02; P < 0.001) when both subclasses were present in hyperimmune sera (38%). In IgG1-restricted sera (15%), the levels were low (<1.5 μg/ml), but for IgG3-restricted samples, the levels were >4.5 μg/ml. Moreover, in sera containing three IgG subclasses (30%), the IgG2b levels correlated with IgG3 (r = 0.803; P < 0.05) but not with IgG1 (r = 0.242; P > 0.05). The sera with detectable IgG2b had higher (P = 0.008) IgG3 levels (7.4 ± 5 μg/ml) than IgG1 levels (1.2 ± 3.6 μg/ml), while sera with detectable IgG2a had similar (P = 0.6) IgG1 (4.3 ± 3.4 μg/ml) and IgG3 (3.6 ± 2.5 μg/ml) levels. In sera with detectable levels of four IgG subclasses (10%), the sum of IgG1 and IgG3 levels correlated with the sum of IgG2a and IgG2b levels (r = 0.74; P < 0.05). No specific pattern for the appearance and clearance of each IgG subclass in serum was observed over time (Fig. 2). These IgG distribution patterns suggest that the change in subclass distribution was from a prevalent IgG2a response, with variable contributions of IgG2b, to predominant IgG3 responses with subsequent decreases in IgG2a and increases in IgG1. These changes seemed to depend on the magnitude and efficacy of the antigenic stimulus.

Isotype distribution in other mouse strains.

The anti-CpsB isotype distributions in C57BL/10, C3H, and Swiss strains immunized with N. meningitidis group B were similar to that observed in BALB/c mice. However the Swiss mice produced lower levels of anti-CpsB antibodies, with a smaller contribution of IgG. There did not seem to be noticeable differences in IgG subclass distributions. The anti-CpsB IgA response was significant only in the C57BL/10 strain, where, on average, 4.7% of the total anti-CpsB antibody response was produced.

DISCUSSION

The purified CpsB, of high molecular weight, was unable to elicit a specific antibody response in mice, as previously reported (35). Nevertheless, in nature Cps is presented in combination with other bacterial components that provide ancillary signals which modulate the Cps immunogenicity and the properties of the antibody response elicited, thus enabling the participation of immune cells not implicated in the response to purified Cps (32, 45, 53, 54). The high anti-CpsB antibody responses obtained when N. meningitidis was used as the immunogen showed the critical importance of these ancillary signals in improving the response to TI antigens, indicating that in CpsB unresponsiveness a mechanism of anergy rather than clonal deletion of B cells is implicated.

The anti-CpsB antibody responses elicited by live N. meningitidis group B displayed TI behavior. The response had a short latent period and showed a rapid increase in serum antibody levels, which reached a peak at day 5, and IgM was the major isotype. The secondary response had similar magnitude and kinetics. This specific antibody response profile agrees with the splenic anti-CpsB SFC kinetics, indicating that the spleen is a major organ in the murine response to CpsB. This is further supported by the greater yield of IgM anti-CpsB hybridomas obtained from fusions carried out 2 days after the last immunization. Yields progressively decreased when fusion was delayed to day 3 or 4, but only at these days can anti-CpsB IgG-secreting hybridomas be obtained (unpublished observations). It has been shown that B cells recently activated by antigen and initiating blast transformation are those which allow secreting hybridomas to be obtained (52). This rapid activation indicates TI antigenic presentation that requires neither antigenic processing nor activation and coordination of multiple cell populations. Thus, similar SFC kinetics have been observed on in vitro stimulation with haptenated TI antigens (36). Besides, CpsB access to the spleen does not seem to noticeably condition response kinetics. Moreover, i.s. immunization elicited lower anti-CpsB antibody responses than the i.v. or i.p. route, even though this does not clearly reflect on the number of anti-CpsB SFC elicited, suggesting that foci of high CpsB concentrations in the spleen suppress Ig secretion. Instead, the primary IgM anti-OMP response was more rapid and elicited higher antibody levels depending on the inverse order of the CpsB responses (i.s. >i.v.>i.p.), suggesting that systemic dispersion of the antigen diminished TD antigen presentation effectiveness. It could be concluded that the CpsB-reactive B cells are rapidly activated (maximum at day 2) and undergo differentiation in IgM-secreting cells to reach a maximum at day 4. During this period, a small proportion of activated B cells undergo isotype switching and differentiate into IgG-secreting B cells.

The anti-CpsB antibodies have a short half-life in serum. The wave of anti-CpsB antibodies elicited by immunization with live N. meningitidis did not plateau, with the descent phase being extraordinarily sharp and quite symmetrical to the exponential phase. This profile cannot be explained as a simple consequence of antibody clearance provided by the inoculum, since the mice recovered from infection during the latent period of the response. Moreover, similar antibody profiles were produced with formalinized bacteria. An explanation could be that the rapid differentiation of CpsB activated B cells during immunization is not accompanied by appreciable proliferation. Hence, the enlargement in SFC spot diameter after day 4 together with an abrupt fall in the number of splenic anti-CpsB SFC could be interpreted to result from increased Ig secretion, due to terminal differentiation of all CpsB-activated B cells to plasma cells. This could limit the persistence of antibody-specific secretion. In general, Cps elicits antibody responses whose levels slowly decrease. This has been associated with their low biodegradation rate and high antigenic stability. Conversely, the native CpsB is unstable (14, 22, 29). However, as has been observed in the response to purified Cps with a longer half-life in the host (4), a secondary peak of anti-CpsB SFC was observed around day 10. This shows that the CpsB persists in the organism in immunogenic form for long periods and provides evidence that CpsB instability cannot fully explain its low immunogenicity. Given this CpsB persistence, the formation of immunocomplexes with anti-CpsB antibodies could control the response by masking the CpsB to reactive B cells. The secondary peak appears when the anti-CpsB antibody levels are already low. Nevertheless, this process is unlikely to fully explain the rapid clearance. The anti-CpsB antibodies have low avidity (28). Therefore, a considerable antibody excess is needed to saturate the system. Moreover, despite the anti-CpsB IgG having lower avidity than IgM (40), the two isotypes show roughly similar serum clearance rates. Therefore the activation of some additional mechanism that favors clearance of anti-CpsB antibodies, e.g., the involvement of anti-idiotypic antibodies, could be hypothesized.

The anti-CpsC antibody responses elicited by bacterial immunization, in contrast, displayed characteristics and kinetics similar to those observed in the anti-OMP response, thus resembling TD behavior. This obviously does not involve antigenic presentation of the CpsC restricted to major histocompatibility complex class II, i.e., its TD antigenic nature (32). Nevertheless, TI type 2 antigens can activate T and non-T cells through multiple nonspecific mechanisms (32, 44). These cells produce soluble factors that reinforce TI responses and mediate B-cell maturation and differentiation. Furthermore, Cps-activated B cells specifically stimulate idiotypes recognizing T cells (2). Thus, purified CpsC can stimulate T cells displaying suppressor (Ts) and amplifier (Ta) activities that regulate CpsC-specific responses (2, 48). When Cps is presented on the bacterial surface, ancillary signals silence the initial suppressor mechanism, probably mimicking the Ta activity that is absent during the early poststimulation days and boosting the subsequent Ta activity elicited by Cps-activated B cells. Finally, the outer membrane vesicles released into the medium during growth or lysis, which have the same composition as the bacterial surface (35, 51), can act as carriers in the cognate collaboration between Cps-reactive B cells and T cells reactive with subcapsular antigens. This process will give antibody responses that have TD properties. It is likely that TI type 2 antigens are unable to induce Ig secretion from resting B cells in the absence of ancillary help (32). Thus, the unresponsiveness to purified CpsB and the TI behavior of the CpsB responses to whole bacteria can be considered to result from CpsB incompetence in T-cell activation by the last two specific mechanisms. The absence of specific Ta activity, which is maximal at day 4 to 5 in TI type 2 responses (2), could explain the rapid decline of the anti-CpsB response during this period. On the other hand, cognate T- and B-cell collaboration for CpsB-reactive B cells seems to be impaired. Thus, covalent CpsB-protein conjugates have low immunogenicity, with the neoepitopes produced during chemical coupling being immunodominant (3, 8, 20). In this context, the CpsB can be considered strictly a TI antigen.

The features observed with the anti-CpsB antibody responses permit the hypothesis that the production of an IgM anti-CpsB response basically depends on the simultaneous presence of a nonspecific costimulus favoring Ig secretion and, to a lesser extent, the proliferation of B cells stimulated by the CpsB. Such a mechanism would elicit IgM responses of a magnitude directly related to both the antigenic stimulus (CpsB) and the costimulus, i.e., the bacterial dose. Although LPS and meningococcal porins have known mitogenic capacities for murine B cells (25, 53), which obviously have important effects on the response, the ability of bacterial lipoproteins to reinforce IgM secretion and proliferation of B cells activated by multivalent antigens (45) fits more accurately to the predictable properties of the costimulus. The outer membrane composition during infection differs noticeably from that expressed by the same strain in culture. One of the differences is in the composition of inducible lipoproteins (1). This different composition could partially explain the greater immunogenicity of live bacteria than of formalinized organisms grown in culture, as well as the poor primary IgM responses elicited by immunization with complexes of CpsB and OMPs (24, 35). The last is a noticeable difference with respect to the above-described anti-CpsB responses to live bacteria.

A small population of CpsB-activated B cells could receive additional ancillary signals favoring the isotype switch. The lack of correlation between the individual concentrations of anti-CpsB IgG and anti-CpsB IgM elicited during the primary response suggests that these signals are unrelated to previous CpsB stimulation. Moreover, the IgG response elicited was not associated with the level of natural anti-CpsB IgM (data not shown). Factors such as the course of infection and/or immune status with respect to subcapsular antigens could be involved. However, immunity to OMPs seems to be of limited importance, considering the delay in the appearance of IgG anti-OMPs with respect to the IgG anti-CpsB during the primary response. In contrast, during the secondary response, the anti-CpsB IgG elicited was reinforced and its level correlated with the IgM anti-CpsB level, suggesting a greater importance of the immune elements generated during primary immunization against subcapsular antigens. The IgG subclass distribution obtained agrees with this. The primary or secondary IgG responses were mainly IgG2a with an important IgG2b contribution. Cytokine production by T and non-T cells during infection might be the cause of this unusual anti-Cps IgG distribution. Thus, T cells have been reported to nonspecifically influence the in vivo B-cell response to TI antigens in a way restricted to IgG2a and, to a lesser extent, IgG2b (33). LPS and meningococcal porins increase the expression of the costimulatory ligand CD86 in B cells that then activate T cells, which release cytokines (53). Activated T cells for subcapsular antigens during priming could amplify this process during challenge.

Hyperimmunization substantially increased the frequency of isotype switch events. This is probably due to clonal expansion and accumulation of CpsB-activated B cells during hyperimmunization, as can be inferred from several observations. First, the anti-CpsB response was not exhausted, even though immunizations were closely spaced, indicating that not all reactive B cells undergo complete differentiation. Moreover, the number of splenic anti-CpsB SFC following later immunization was maintained over preimmune levels. Second, the IgG distribution changed to a prevalence of IgG3 and IgG1. In the absence of T cells, the response to TI type 2 antigens showed a distribution in which the relative proportion of each isotype was correlated directly with the 5′→3′ positions of their respective Igh-C genes, i.e., IgM > IgG3 > IgG1 > IgG2b > IgG2a (33), as we observed for the anti-CpsB response of hyperimmunized mice. Finally, these properties were maintained in the response to a boost given 1 month after the hyperimmunization ended, indicating that this IgG distribution is not a simple consequence of the continuous CpsB stimulus. These changes in the CpsB response can be interpreted kinetically. Thus, the rapid kinetics of B-cell activation during the anti-CpsB response could become useless ancillary signals from cells whose activity starts or raises the maximum further than day 4. However, during hyperimmunization, the signals produced following an immunization could act efficiently on B cells activated by CpsB in the subsequent immunization (2 to 4 days later). Taking into account this kinetic condition, the clonal expansion induced, and the direction of the isotype switch, this ancillary help could mimic the Ta activity that displays these activities in the response to purified Cps (2).

The results favor the hypothesis of tolerance to CpsB conformational epitopes. The properties displayed by the anti-CpsB response agreed with a defect in the immune system associated with this Cps which is concerned with the proliferation, differentiation, and isotype switch in B cells activated by CpsB. These defects are only partially overcome by ancillary signals which may be directly or indirectly derived from the subcapsular antigens. We therefore suggest that the poor immunogenicity of CpsB is due to the functional absence of Ta (and Ts)-specific activities, in combination with an accelerated clearance of anti-CpsB antibodies. As a result, the anti-CpsB response is necessarily of short duration, an undesirable property for vaccines. On the other hand, our results emphasize the critical importance of incorporating components that complement and amplify the CpsB-initiated response. The outer membrane vesicles have these properties (9, 31, 50, 53). However, their composition must have major effects on the response. Nevertheless, although breaking tolerance to an autoantigen is not necessarily followed by a clinical autoimmune process, the absolute safety of anti-CpsB antibodies still needs to be proven before CpsB-based vaccines can be administered to humans.