INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Neisseria meningitidis has emerged as one of the most common causes of meningitis in children and young adults. Strains of N. meningitidis can be divided into serogroups on structurally distinctive polysaccharide capsules. Polysaccharide vaccines are available for prevention of meningococcal disease caused by serogroups A, C, Y, and W135, but group B polysaccharide is not included in these vaccines since it is not immunogenic in humans. Outbreaks of serogroup B epidemics and endemic cases have occurred periodically worldwide, and an effective vaccine for prevention of the disease remains a public health priority.

A number of experimental approaches have been used to develop such a vaccine (14, 21, 24); some are based on the use of proteins or lipopolysaccharides, and others are based on the use of the group B capsular polysaccharide (B PS). This polysaccharide is composed of a homopolymer of (2-8) N-acetyl neuraminic acid or polysialic acid (PSA). Identical polysaccharides are also present on Escherichia coli K1, Moraxella nonliquefaciens, and Pasteurella haemolytica A2. By itself it is poorly immunogenic, and a strategy to overcome the problem has been to couple it to a carrier protein, which did not result in improving its immunogenicity. This property is attributed to immunologic tolerance induced by the existence in mammals of polysialylated glycoproteins known as neural cell adhesion molecules (PSA-NCAM) bearing structurally identical polysaccharides.

A successful strategy proposed by Jennings et al. (18) to overcome the lack of immunogenicity has been to substitute the N-acetyl groups of the native B PS with N-propionyl (N-Pr) groups prior to its conjugation to a carrier protein.

A careful examination of the specificity of the antibodies induced by this conjugate and of their reactivity towards PSA-NCAM is a priority in establishing whether such a vaccine may be safe. This is even more important considering that group B meningitis often affects infants and young adults, making vaccination needed at early life stages when PSA-NCAM is still widely expressed (3, 31).

We used the N-propionyl polysaccharide (N-Pr PS) tetanus toxoid (TT) conjugate (N-Pr PS-TT) to immunize mice and to produce both a pool of antiserum and a series of immunoglobulin G (IgG) monoclonal antibodies. In this study we carefully evaluated their self-reactivity profiles by using a battery of tests chosen for their sensitivities and specificities to detect a recognition of PSA-NCAM and self-directed antibodies. We also searched for their perturbing effects in a functional assay for cell migration and differentiation, since PSA-NCAM is known to play a role in such events (28). Our rationale has been to analyze in a first series of experiments the reactions shown by the antiserum obtained from a group of hyperimmunized mice. In a second series of experiments, we selected three monoclonal antibodies (MAbs) based on their high levels of specificity for the N-Pr PS and their significant bactericidal activities against the group B meningococcus, and we compared their reactivities towards PSA-NCAM in the tests.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Production and characterization of mouse antibodies. (i) Preparation of the N-Pr PS-TT conjugate. Meningococcal group B polysaccharide was purified from N. meningitidis B11 according to the method of Gotschlich et al. (12).

(ii) Production of mouse polyclonal antibodies. All procedures involving the use of animals were performed in accordance with the European animal care guidelines and directives.

(iii) Production of anti-N-Pr PS monoclonal antibodies. Five 6- to 8-week-old CD1 mice were immunized i.p. three times with N-Pr PS-TT (3 µg of sialic acid per injection) as described above. Three days after a fourth injection (intravenous [i.v.], without adjuvant), the splenocytes were fused with X63 Ag 8653 myeloma cells. Supernatants of the resulting hybridomas were screened and selected according to their binding to N-Pr PS but not to B PS by enzyme-linked immunosorbent assay (ELISA). Their bactericidal activity against group B N. meningitidis (M986 strain) was evaluated. Three clones [IgG2b()] were selected based on their strict specificity for N-Pr PS and for their significant bactericidal activity. These clones are hereafter referred to as A74, A79, and A98. Ascites were prepared by injection into pristane-treated Swiss nude mice, and the respective MAbs were purified from ascitic fluid using protein G. Their IgG concentration was evaluated by ELISA using an IgG standard reference.

(iv) Control antibodies. Two different monoclonal antibodies, the mouse IgM anti-MenB (27) and the mouse IgG2a 30H12 (16), obtained after immunization with live group B meningococci, were also used in the study as positive controls. They specifically recognize the capsular B PS, and their characteristics have been published elsewhere. Monoclonal IgG2a and IgG2b with nonrelevant specificities were also used in the tests as negative controls.

(v) ELISA and competitive ELISA. The solid-phase ELISA was used to evaluate antibody binding to B PS or N-Pr PS. Microplates (Dynatech) were coated for 3 h at 37°C with a complex prepared with methylated human albumin (mHSA) and B PS (25 and 25 µg/ml) or mHSA and N-Pr PS (10 and 25 µg/ml, respectively). mHSA is used to improve the binding of negatively charged polysaccharides to microtiter plates (1, 5). Plates were subsequently saturated with 5% bovine serum albumin-phosphate-buffered saline (PBS) for 2 h at 37°C. Twofold serial dilutions of samples were made in PBS-Tween-1% bovine serum albumin, and 100 µl was incubated in the microplates for 1 h at 37°C. Plates were then incubated with an anti-mouse IgG coupled to alkaline phosphatase for 2 h at 37°C. After addition of the substrate (p-nitrophenylphosphate [PNPP]), the enzymatic reaction was monitored for 30 min and stopped with 1 N NaOH. Reading was performed at 405 nm on a Dynatech reader, and titers were calculated by comparison with reference sera.

(vi) Bactericidal assay and competitive bactericidal assay. The test was carried out in microtiter plates (Nunc) using baby rabbit serum (Aventis Pasteur preparation) as a source of complement. Group B N. meningitidis (M986 strain) was grown overnight on Mueller-Hinton agar and then for 3 h in Mueller-Hinton broth (Difco). The bacterial suspension was adjusted to 4,000 CFU/ml in Dulbecco's PBS (Difco) before use. Twenty-five microliters of each of the following components was added successively to the wells: the bacterial suspension and serial dilutions of the tested antibodies. The plates were then shaken for 20 min at 37°C prior to addition of 25 µl of complement. Following another 40-min incubation, an aliquot from each well was transferred onto Mueller-Hinton agar. The plates were then incubated at 37°C overnight under 10% CO₂, and the colonies were counted the following day. The bactericidal titer was expressed as the reciprocal of the highest dilution of the antibody tested at which 50% or more of bacteria were killed compared to the number killed with the complement control (bacteria plus complement). In addition, the minimal concentration of the tested antibody capable of killing 50% of the bacteria was calculated for monoclonal antibodies.

Mouse antibody reactivity to host polysialic acid. (i) Cell culture and immunofluorescence assay. The binding of antibody to PSA-bearing cells was assessed by fluorescence microscopy with two tumor cell lines (the human rhabdomyosarcoma TE671 and the AtT20 cells derived from a mouse anterior pituitary tumor) and on primary cultures of newborn mouse cerebellum.

(ii) Western immunoblotting assay. Brain membrane extracts from mice taken at different developmental stages were obtained using a procedure described previously (22). Proteins were separated under denaturing conditions by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and electrophoretically transferred to a nitrocellulose membrane that was probed overnight at 4°C with antibodies to be tested. Bound antibodies were detected after incubation with a peroxidase-conjugated antibody using an ECL (Roche) or a DAB detection kit (Vector laboratories).

(iii) Immunoprecipitation assay. (a) Purification and iodination of PSA-NCAM. PSA-NCAM was purified by affinity chromatography from detergent membrane extracts of embryonic mouse brains as described previously (25, 26). Purified PSA-NCAM was then radiolabeled with ¹²⁵I-Na (Amersham) using the Iodogen technique (30).

(b) Formation of immune complexes. Antibodies to be tested were incubated with protein A-Sepharose CL4B beads reacted with rabbit anti-mouse IgG antibodies for 4 h at 4°C under shaking. After being washed, the beads were incubated with ¹²⁵I-labeled PSA-NCAM for 14 h at 4°C under shaking. Complexes were then washed and mixed (vol/vol) with Laemmli buffer (0.125 M Tris-HCl, 4% SDS, 20% glycerol, 1% -mercaptoethanol, 0.001% bromophenol blue) and heated for 4 min at 100°C. Immunoprecipitated proteins and antibodies were separated from protein A beads by centrifugation, and the supernatant was subjected to 7% SDS-PAGE. Radiolabeled proteins were visualized by autoradiography.

(iv) Immunohistochemistry assay. Samples of human muscles from patients suffering from progressive muscular dystrophy showing regenerative fibers expressing PSA-NCAM and a desmoplasic medulloblastoma (human neuroectodermic tumor) expressing PSA-NCAM were used.

(v) Cytotoxicity assay. AtT20 cells were seeded in 96-well plates at 10,000 cells/well, grown until confluence. In some experiments cells were treated with EndoN prepared in our laboratory (32) to remove PSA from the surfaces of the cells. Three hundred thousand counts per minute of ⁵¹chromium (sodium chromate, 350 to 600 mCi/mg; Amersham) was added to the culture medium for 16 h at 37°C. After being washed, cells were incubated with antibodies or serum to be tested for 45 min at 37°C. Cells were then washed and incubated with rabbit complement (Cedarlane) (1/10) for 1 h at 37°C. Supernatants containing ⁵¹Cr released from lysed cells were collected and counted for each well. Total ⁵¹Cr incorporated was estimated by counting individual wells after lysis of cells with 10% SDS. The percentage of cytotoxicity was calculated in the following way: [(experimental ⁵¹Cr release  spontaneous ⁵¹Cr release)/(total ⁵¹Cr incorporated  spontaneous ⁵¹Cr release)] × 100.

(vi) Effects of antibodies on neuroblast migration. (a) Culture of subventricular zone (SVZ) explants. SVZ explant cultures were performed as described by Wichterle et al. (33).

(b) Analysis of cell migration. After 48 h in culture, the explants were viewed using phase-contrast microscopy (Axiovert 35 M; Zeiss). Images of the explants were recorded with a video camera (Cool View; Photonic Science), digitized, and analyzed using image processing software (Visiolab1000; Biocom, Paris, France).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

CPA14 mouse antiserum. Immunoprecipitation of radioiodinated antigens is known to be a very sensitive technique. We immunopurified NCAM from embryonic mouse brains and submitted it to iodination as described in Materials and Methods. ¹²⁵I-labeled PSA-NCAM was immunoprecipitated both by a rabbit polyclonal antibody directed against the NCAM protein backbone (26) (Fig. 1, lane 1) and by the mouse anti-meningococcus group B MAb IgM directed against PSA (anti-MenB), characterized as described previously (27), and used as positive controls. Whereas anti-NCAM precipitated all the NCAM isoforms, anti-MenB antibody precipitated only the polysialylated forms which migrate as a broad band above 200 kDa (Fig. 1, lane 2). None of the tested antisera was able to immunoprecipitate PSA-NCAM (Fig. 1, lanes 3 to 5). These corresponded to antiserum obtained from mice immunized with N-Pr PS coupled to TT, named CPA14 (Fig. 1, lane 3), saline named B48/1J42 (Fig. 1, lane 4), or an uncoupled mixture of the N-Pr PS and TT named B48/2J42 (Fig. 1, lane 5).

View larger version (66K): [in this window] [in a new window]   FIG. 1.   Immunoprecipitation of 125I-labeled PSA-NCAM. The two positive controls tested, the polyclonal antibody anti-NCAM (lane 1) and the mouse MAb anti-MenB (lane 2), were able to precipitate all NCAM isoforms (polyclonal) or only the PSA-NCAM isoforms (anti-MenB). None of the antisera tested, CPA14 (lane 3), B48/1J42 (lane 4), and B48/2J42 (lane 5), was able to immunoprecipitate PSA-NCAM.

View larger version (73K): [in this window] [in a new window]   FIG. 2.   Western blot of extracts of membrane proteins prepared from embryonic (A and B) or adult (C and D) mouse brains. Antisera were diluted 1/50 and revealed with either anti-IgG (A and C) or anti-IgM (B and D) secondary antibodies. The mouse MAb (IgM class) anti-MenB (lanes 4) was used as a positive control. This MAb specifically recognized PSA-NCAM present in embryonic extract when revealed with anti-IgM. Among the antisera tested, CPA14 (lane 1), B48/1J42 (lane 2), and B48/2J42 (lane 3), only CPA14 reacted weakly with PSA-NCAM when revealed with anti-IgG (A) (lane 1) but not anti-IgM (B) (lane 1). CPA14 and B48/1J42 also bound to other bands in both embryonic and adult extracts when revealed with anti-IgM (A and B).

View larger version (175K): [in this window] [in a new window]   FIG. 3.   Immunofluorescence staining on AtT20 cells. The control mouse MAb IgM antibody anti-MenB was diluted 1/1,000 (A), the pooled antiserum B48/1J42 was diluted 1/50 (B), and CPA14 was diluted 1/50 (C) and 1/100 (D). Labeling was revealed with a fluorochrome-conjugated secondary antibody anti-IgM for MenB (A) and anti-IgG for the antisera (B, C, and D). A typical intense cell labeling was observed with MenB (A), and a weaker signal was observed depending on the dilution with CPA14 (C and D). B48/1J42 antiserum at a dilution of 1/50 showed background staining (B).

View larger version (119K): [in this window] [in a new window]   FIG. 4.   Immunohistochemical labeling of human medulloblastoma (A, B, and C) and regenerative muscle biopsies (D, E, and F). The control mouse MAb anti-MenB was diluted 1/1,000 (A and D), and antiserum B48/1J42 (B and E) and CPA14 (C and F) were diluted 1/100. On medulloblastoma sections, anti-MenB and CPA14 labeled islets of positive cells (arrows in A and C, respectively), whereas B48/1J42 weakly labeled all cell surfaces. In regenerative muscle, anti-MenB and CPA14 revealed regenerative fibers (arrows in D and F, respectively), with a higher intensity for anti-MenB. Both B48/1J42 (E) and CPA14 (F) exhibited nonspecific staining of cell nuclei.

Characterization of mouse monoclonal antibodies raised against N-Pr PS-TT. We isolated a series of monoclonal IgG antibodies produced against this immunogen and screened them for their recognition of N-Pr PS, their binding to B PS in ELISA, and their bactericidal activity. Three of them, all of the IgG2b subclass, representative of the screened population, were selected for further studies and named A74, A79, and A98. Competition ELISA experiments confirmed that they were not recognizing the native polysaccharide (B PS) (Table 2), in contrast to the control MAb 30H12 obtained after immunization with live group B meningococci (16), which was highly specific for B PS (IC₅₀, <0.01 µg/ml). Bactericidal assays indicated that A74 and A98 exhibited similar bactericidal titers (Table 3). Their bactericidal activity against the group B meningococcus was exclusively inhibited by N-Pr PS (IC₅₀ of B PS, >100 µg/ml) (Table 4). The bactericidal activity of A79, however, was inhibited by both B PS and N-Pr PS and was detected at a concentration five times lower than that required for A98 and A74. Despite these differences among the three antibodies, the cross-reactivity indexes of N-Pr PS to B PS were lower than 1% for all of them. 30H12, which is strictly specific to B PS (Table 4), showed the highest bactericidal power (activity detected at 0.12 µg/ml) (Table 3).

Recognition of PSA-NCAM by MAbs raised against N-Pr PS-TT. Each of the selected MAbs was evaluated for its recognition of PSA-NCAM in three of the tests that were also used to evaluate the polyclonal antiserum CPA14. These were immunoprecipitation, Western blotting, and immunofluorescence on live cells. None of the antibodies recognized ¹²⁵I-labeled PSA-NCAM in the immunoprecipitation test (not shown). In Western blot experiments, MAb A79 very specifically recognized PSA-NCAM in embryonic mouse tissue extracts (Fig. 5, lane 3) similarly to 30H12 (Fig. 5, lane 4), whereas the two others, A74 and A98 (Fig. 5, lanes 2 and 4, respectively), were negative. In the immunofluorescence tests performed on human TE671 cells expressing PSA on their surface, none of the three MAbs tested exhibited reactivity (shown for A79 and A98) (Fig. 6, H and J, respectively). Similar results were obtained with the AtT20 cell line and with primary culture of mouse cerebellum cells (not shown). The control antibodies, anti-MenB and 30H12, recognized PSA-NCAM (Fig. 6B and F, respectively), and removal of surface PSA with EndoN abrogated the reaction of anti-MenB (Fig. 6D).

View larger version (39K): [in this window] [in a new window]   FIG. 5.   Western blot with the selected mouse monoclonal antibodies on extracts of membrane proteins prepared from embryonic mouse brain. The purified antibodies (A74, A79, A98, and 30H12) were diluted at 5 µg/ml, and the ascitic fluid anti-MenB was diluted 1/400. The mouse MAb IgM anti-MenB and IgG2a 30H12 were used as positive controls. Anti-MenB (lane 1), 30H12 (lane 5), and A79 (lane 3) precipitated a large high-molecular-weight band characteristic of PSA-NCAM.

View larger version (80K): [in this window] [in a new window]   FIG. 6.   Immunofluorescence analysis with the human TE671 cell line and the selected mouse monoclonal antibodies. Purified monoclonal antibodies to be tested and the control 30H12 were diluted to a concentration of 25 µg/ml, and the ascitic fluid anti-MenB was diluted 1/500. Left panels show phase contrast. In the right panels, labeling was revealed with a fluorochrome-conjugated secondary antibody anti-IgM for anti-MenB (B and D) and anti-IgG for the other antibodies (F, H, and J). In panels C and D, cells have been treated with EndoN prior to labeling to remove PSA from their surfaces. The anti-MenB controls gave a positive signal of recognition (B) that was highly specific for PSA-NCAM, since the labeling was abolished by EndoN treatment (D). Among the IgG antibodies, only 30H12 gave a positive signal (F). A79 (H) and A98 (J) did not recognize PSA-NCAM on cell surfaces.

Search for functional perturbations due to cross-reaction with PSA-NCAM. The involvement of the PSA group in controlling neuroblast chain migration in vivo is well described (23). We set up an in vitro assay (6) in which this process could be evaluated in the absence or presence of agents such as EndoN or antibodies affecting PSA groups. Explants of mouse SVZ prepared as described in Materials and Methods and observed after 48 h of culture showed migrating neuroblasts organized as chains (Fig. 7A). All neuroblasts expressed NCAM and PSA (23), as confirmed by immunofluorescence analysis using anti-NCAM and anti-MenB (Fig. 7F and G, respectively). Removal of PSA by treatment of the explant with the enzyme EndoN, evidenced by the absence of anti-MenB binding, strongly perturbed the migration process. The exit of chains from the explant was delayed; these were smaller, and they migrated shorter distances (Fig. 7, compare panels A and B). When chains were occasionally formed, cell interactions between neuroblasts were modified, as were the overall morphologies of the cells (Fig. 7, compare panels A1 and B1). When the explants were cultured in the presence of either the anti-MenB or 30H12 antibody (Fig. 7C and D) showing positive PSA binding on live cells, the migration occurred normally. No differences could be found between the time cells took to exit from the explant, or in the size of the chains or the morphology of the cells, between antibody-treated explants and controls. There were no differences between either the anti-MenB and 30H12 or the A79 MAb, shown to recognize PSA in the Western blot test, and the two other antibodies (A98 and A74) showing no reaction in all of the previous tests performed (shown for A98 [Fig. 7E and E1]). Our data were not due to an absence of diffusion of the antibodies inside the explants, since we verified by immunofluorescence staining that anti-NCAM (Fig. 7F) as well as anti-MenB (Fig. 7G) or 30H12 (Fig. 7H) antibodies were indeed bound to the PSA-expressing cells.

View larger version (138K): [in this window] [in a new window]   FIG. 7.   Effect of PSA perturbation on neuroblast migration. Purified monoclonal antibodies (A74, A79, A98, and 30H12) were diluted at 5 µg/ml, the ascitic fluid anti-MenB was diluted 1/200, and EndoN was diluted 1/2,500 of a solution at 0.5 U/ml and used in the cultured explants as described in Materials and Methods. Blockade of PSA groups by anti-MenB or 30H12 (C and D, respectively) did not modify the migration of neuroblasts out of explants of mouse SVZ (C and D) by comparison to the control (A), nor did it influence the morphology of the migrating chains (C1 and D1 compared to A1). By contrast, removal of a PSA group by EndoN perturbed migration (B) and chain formation (B1). Note that cells were mainly isolated from one another. Experiments conducted in the presence of A98 antibody (E) gave results similar to those of the control (A), and chains could be observed (E1). Antibody penetration inside the explant and fixation on PSA-expressing neuroblasts were controlled by immunofluorescence with the polyclonal anti-NCAM (1/500), anti-MenB (1/200), and 30H12 (20 µg/ml) antibodies (F, G, and H, respectively).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The examination of the antibody response from a group of mice hyperimmunized with N-Pr-PS coupled to TT (antiserum CPA14) showed that this candidate vaccine is able to elicit antibodies specific to N-Pr PS and to a much lower extent to B PS. In agreement with the fact that N-Pr PS is used conjugated to a protein carrier, a population of these antibodies was of the IgG class. These antibodies also elicited complement-mediated bactericidal activity against the group B meningococcus. Our data corroborate results of previous studies conducted with a similar immunogen (2, 19). This places this immunogen as a potential candidate vaccine to prevent group B meningococcal disease or to obtain selected antibodies to be used for treatment of individuals infected by the bacteria. Nevertheless, one important safety concern with such a vaccine is the potential antihost activity of the antibodies elicited. Indeed, B PS and PSA-NCAM share common epitopes, since antibodies raised against live meningococcus group B strains are able to recognize PSA-NCAM. This is well illustrated by the reactivity of the anti-MenB antibody (27) used in the present study and so far considered one of the best available probes for recognizing PSA-NCAM in mammalian tissues. Indeed, the protein NCAM is the major and only identified carrier of PSA groups in mammals (27), as confirmed by the more recent observation that NCAM-deficient mice do not express detectable PSA (7).

It cannot be excluded that autoantibodies have the potential either to evoke autoimmune disease in vaccinated individuals or to cross the placenta and interfere with neural cell migration in the developing fetus. A flurry of studies with mammals showed that the PSA-NCAM isoforms are expressed mainly during ontogeny and in early postnatal stages and remain expressed in the adult only in discrete areas of the central nervous system that are normally not accessible to circulating antibodies and on very small populations of cells in other areas of the body (3, 31). In pathological situations, PSA is also expressed on the most aggressive tumors of neuroectodermic origin.

Our present study was designed to analyze thoroughly the possible existence of populations of PSA-NCAM cross-reactive antibodies in the immune response induced by the N-Pr PS-TT immunogen and to examine the functional perturbation potentially elicited by the binding of such antibodies on cells expressing PSA-NCAM.

Because the nature of the response might vary somewhat from one individual to another, we first tested an antiserum composed of a pool of sera obtained from a group of hyperimmunized mice. The reactivity of this antiserum, CPA14, was compared with controls and in particular with a pool of sera obtained from mice immunized with either saline or uncoupled carrier and hapten. Since uncoupled N-Pr PS is not immunogenic by itself, differences of reactivities between these three batches could be attributed to antibodies induced by the coupled hapten.

The serum CPA14 showed weak reactivity towards PSA-NCAM, which was detectable only in some of the tests. CPA14 was not able to immunoprecipitate iodinated PSA-NCAM. Nevertheless, a weak but specific binding of CPA14 to PSA-NCAM in embryonic brain extracts was detected in a Western blot and on the surfaces of live cells by immunofluorescence. These differences among the tests suggest that the antiserum contains several populations of antibodies differing in their abundance, thermodynamic characteristics, and epitope binding. The same behavior observed for the selected MAb A79, which recognizes B and N-Pr PS and CPA14, suggests that a population of antibodies showing the same reactivity as MAb A79 exists in CPA14, but it must be minor. Furthermore, our results point out the differences in reactivities between cross-reactive antibodies produced by individual clones. The data also suggest a low affinity of the antibodies induced by the vaccine candidate towards the antigen PSA-NCAM and/or a structural constraint of the antigen for its recognition by the induced antibodies. Experimental conditions which involved harsh washings of the antigen-antibody complexes, such as in immunoprecipitation, would not allow detection of the low-affinity populations. It is also possible that the denaturation of PSA tertiary structure by detergents prevents its recognition in some instances. In any case, this indicates that the use of a single test is not secure enough to exclude a cross-reaction.

Our data also indicated that besides likely being of low affinity, the population of cross-reactive antibodies in the pool of antisera is minor. Indeed, in all tests where a reaction could be evidenced, the reaction was always very weak and close to the limit of sensitivity of the technique used. This is true for immunoblotting but also for immunofluorescence on live AtT20 cells or for binding on human tissues where PSA-NCAM is heavily expressed, such as in medulloblastoma. In each of these tests, a reaction could be detected only in conditions where very low dilutions of the antibodies were used. Immunoblotting indicated that at such low dilutions of the antiserum, several molecules of the self were revealed by antibodies of the IgM class. These molecules were not antigens structurally related to PSA, because they were also revealed by the negative control antiserum B48/1J42. This also indicated that these IgM antibodies were not specifically induced by the conjugate N-Pr PS-TT. Since we observed on tissue sections a strong background staining in cell nuclei, it is possible that the cross-reactive nuclear antigens are in fact never in contact with circulating antibodies in physiological situations. In any case, it remains possible that in many vaccination procedures, such a phenomenon exists without being detrimental.

The immunofluorescence and Western blot data both indicated that the PSA cross-reactive antibodies were of the IgG class. This is in agreement with previous data on this immunogen showing that when used as a hapten, N-Pr PS shifted the response towards IgG (18), whereas live bacteria induced mainly antibodies from the IgM class against the B PS, as shown in sera from patients with group B meningitis disease (20, 13).

Notice that the analysis of the MAbs clearly indicated that IgG antibodies strictly specific for N-Pr PS could be obtained and that even in the absence of a reaction with the purified B PS, they were able to bind to it when it was presented on the bacteria and, most importantly, to lyse the bacteria in the presence of complement, although at a lower degree than the anti-B PS antibodies. These antibodies, such as MAb A74 and A98, were not showing a cross-reaction with PSA-NCAM in any of the tests. Therefore, there are some structural differences, so far undescribed but not unexpected, between PSA bound to the protein NCAM and B PS on the bacterial capsule.

Both the cytotoxicity and migration tests were instrumental in generating information on the potential deleterious effects of the antibodies cross-reacting with PSA-NCAM. The cytotoxicity test indicated that CPA14 antiserum induced no more cell lysis than the negative controls. Thus, no cell lysis could be attributed to the antibody binding to PSA-expressing cells. This was clearly confirmed by the use of EndoN, which allowed us to unambiguously decide which part of the lysis was due to binding of the antibodies to PSA.

The migration test revealed an interesting and new piece of data on how perturbation of PSA might have different effects depending on whether it is masked by antibodies or its expression is suppressed (EndoN treatment). Indeed, as was already shown by members of our group (6) and others (15, 23), removal of PSA strongly perturbs chain migration of normally expressing PSA neuroblasts. Here we show that antibodies masking PSA and likely preventing its interactions did not induce such perturbations. This might be interpreted in the light of what is known on the PSA mode of action and function. Indeed, whereas no molecule interacting specifically with PSA (receptor) has been reported so far, PSA is believed to modulate cell-cell interactions by creating a coat around cells, thus preventing their close contact and communication to occur (29). If this is the case, it is understandable that the presence of an antibody will merely increase this steric effect of PSA but will not modify its function. Whatever the interpretation, this observation might suggest that the PSA cross-reactive antibodies induced by this candidate vaccine might not be detrimental at least to this particular action of neuroblasts, even if they are able to cross the placental barrier or the blood-brain barrier and reach the developing nervous system of the fetus or infant.