Comparative Study of Immunogenic Properties of Purified Capsular Polysaccharides from Streptococcus suis Serotypes 3, 7, 8, and 9: the Serotype 3 Polysaccharide Induces an Opsonizing IgG Response

Streptococcus suis is an encapsulated bacterium and one of the most important swine pathogens and a zoonotic agent for which no effective vaccine exists. Bacterial capsular polysaccharides (CPSs) are poorly immunogenic, but anti-CPS antibodies are essential to the host defense against encapsulated bacteria. In addition to the previously known serotypes 2 and 14, which are nonimmunogenic, we have recently purified and described the CPS structures for serotypes 1, 1/2, 3, 7, 8, and 9.


FIG 2
Proinflammatory cytokine production by DCs in response to stimulation by S. suis or S. pneumoniae CPSs for 24 h. Native CPSs of S. suis serotypes 2, 1, 1/2, 3, 7, 8, and 9 or of S. pneumoniae serotype 19F (each at 200 g/ml) were incubated with DCs (10 6 cells/ml). After 24 h, supernatants were collected, and the IL-6 (A) and TNF (B) levels were quantified by ELISA. Cells stimulated with medium alone and with LPS (1 g/ml) served as negative (C-) and positive (Cϩ) controls, respectively. The data are expressed as means Ϯ the SEM for at least three experiments. Statistically significant differences versus medium alone (C-) are indicated (***, P Յ 0.001).
In contrast to these proinflammatory cytokines, all S. suis and S. pneumoniae CPSs at 200 g/ml induced significant release of the chemokines CCL2 and CCL3 after incubation with DCs (Fig. 3). Thus, we also investigated the involvement of PRRs in chemokine release by CPS-stimulated DCs by comparing wild-type (WT) DCs with either TLR2 Ϫ/Ϫ or MyD88 Ϫ/Ϫ DCs. The results showed no significant difference in CCL2 production (Fig.  3A). This indicates that CCL2 production by DCs is induced through a different pathway than the TLRs. However, CCL3 production was significantly reduced in TLR2 Ϫ/Ϫ and further reduced in MyD88 Ϫ/Ϫ DCs, although not abrogated (Fig. 3B). This indicates not only that CCL3 production by DCs is partially dependent upon TLR2 and MyD88 signaling but also that at least another pathway is involved in CPS-mediated activation of DCs. In addition, CCL2 and CCL3 production by CPS-stimulated WT DCs appears to be independent of the CPS structure, since the levels for both chemokines were all similar among samples from S. suis and S. pneumoniae (Fig. 3).
In vivo immunogenicity studies with purified CPSs 3, 7, 8, and 9. Despite the absence of cytokine production and the lack of differences in chemokine production patterns after stimulation of APCs with widely different CPSs, the immunogenicity of those newly obtained purified CPSs was evaluated in vivo. Native purified CPS 2 (included as control) or native CPSs 3, 7, 8, and 9 were emulsified with TiterMax Gold and administered to mice on days 0 and 21. CPSs 2, 7, 8, and 9 failed to induce a significant anti-CPS response (Fig. 4). Although CPS 2 is already known to be poorly immunogenic, additional control enzyme-linked immunosorbent assays (ELISAs) were performed for CPSs 7, 8, and 9 to ensure that the lack of anti-CPS response was not due to problems with the coating (see Fig. S1 in the supplemental material). First, ELISA plates coated with whole S. suis of either serotype 7, 8, or 9 were used to validate strong pneumoniae CPSs for 24 h. Native CPSs of S. suis serotypes 2, 1, 1/2, 3, 7, 8, and 9 or of S. pneumoniae serotype 19F (each at 200 g/ml) were incubated with either wild-type (WT; black bars), TLR2 Ϫ/Ϫ (blue bars), or MyD88 Ϫ/Ϫ (red bars) DCs (10 6 cells/ml). After 24 h, supernatants were collected, and CCL2 (A) and CCL3 (B) levels were quantified by ELISA. Cells stimulated with medium alone and with LPS (1 g/ml) served as negative (C-) and positive (Cϩ) controls, respectively. The data are expressed as means Ϯ the SEM for at least three experiments. Statistically significant differences versus the WT are indicated (*, P Յ 0.05; **, P Յ 0.01; ***, P Յ 0.001).
total Ig (IgG ϩ IgM) anti-S. suis responses in sera from hyperimmunized mice ( Fig. S1A to C). Second, these control hyperimmune mouse sera showed no anti-CPS response for serotypes 7 and 8 (see Fig. S1D and E) but an anti-CPS response for serotype 9 (see Fig.  S1F) using CPS-coated plates. Finally, sera collected at day 42 from mice immunized with purified CPSs 7, 8, and 9 showed no significant total Ig (IgG ϩ IgM) anti-S. suis responses (P Ͼ 0.05 versus placebo groups) using whole-bacterium-coated plates. The latter result further supports that these CPSs did not elicit antibodies in the immunized animals and that this lack of response was not related to coating technical issues (see Fig. S1G to I). Taken together, these data indicate that purified CPSs 7, 8, and 9 are not immunogenic in mice. In contrast, mice immunized with the native purified CPS 3 produced a strong antibody response as early as 7 days after a single dose, followed by a slight increase in the response after a boost on day 21 (Fig. 4B). A stronger response following boosting but also antibody isotype switching are good indicators of immunogenicity. Therefore, titers of the different anti-CPS antibody isotypes were determined in mice immunized with 25 g of CPS 3 adjuvanted with TiterMax Gold. As shown in Fig. 5A, not only strong IgM titers but also significant high levels of type 1 IgG subclasses (IgG2b, IgG2c, and IgG3) were observed (P Յ 0.05 versus the placebo group), whereas no type 2 IgG subclass (IgG1) was detected (P Ͼ 0.05 versus the placebo group). Based on these observations, the protective capacity of sera from CPS 3-immunized mice was evaluated using an opsonophagocytosis assay (OPA), a recognized correlate of immunity for encapsulated Gram-positive bacteria (38). As shown in Fig. 5B, sera from mice immunized with CPS 3 adjuvanted with TiterMax Gold induced high bacterial killing levels (P Յ 0.001 versus placebo group).
Adjuvant effect on in vivo immunogenicity of purified CPS 3. The results showed that mice receiving on days 0 and 21 the same dose as before (25 g) of CPS 3 either non-adjuvanted (phosphate-buffered saline [PBS] only) or adjuvanted 1:1 (vol/vol) with Viable bacterial counts were determined after 2 h of incubation. To determine bacterial killing, viable bacterial counts from tubes incubated with sample sera were compared to those incubated with control naive mouse sera. Rabbit anti-S. suis type 3 strain 4961 serum was used as a positive control (Cϩ) and compared to control naive rabbit serum to determine bacterial killing. The results are expressed as percent bacterial killing for individual mice, with horizontal bars representing median. Individuals from the placebo group are indicated by a gray circle, whereas those of the CPS 3 group are indicated by an orange diamond. Statistically significant differences versus the placebo group are indicated (*, P Յ 0.05; **, P Յ 0.01; ***, P Յ 0.001).
Purified CPS 3 immunogenicity is not influenced by phenol-extractable immunestimulating ligands. Since we showed that CPS 3 induces a type 1 IgG antibody response in vivo, we decided to investigate the presence of potential TLR ligands in the native CPS 3 preparation. As shown above, no native S. suis CPS, CPS 3 included, led to the production of IL-6 or TNF by CPS-stimulated DCs, in contrast to the cytokine production observed with S. pneumoniae CPS 19F (Fig. 2). In our study, the production of both IL-6 and TNF was dependent upon MyD88 signaling but independent of TLR2, suggesting that the CPS 19F preparation used contains immune-stimulating TLR ligands (see Fig. S3). After phenol extraction, S. suis CPS 3 did not show structural alterations by dot blot and by 1 H nuclear magnetic resonance (NMR) analyses (see Fig.  S4 and S5). DCs stimulated with phenol-extracted CPS 3 secreted no IL-6 nor TNF in the supernatant, and their CCL2 and CCL3 production remained unchanged compared to its native counterpart (data not shown).
To further determine whether potential TLR ligands affected the antibody response in vivo, mice received on days 0 and 21 a dose of 25 g of either native or phenolextracted CPS 3 adjuvanted with TiterMax Gold. Kinetic analysis of total Ig (IgG ϩ IgM) antibody responses against native CPS 3 at days Ϫ1, 20, and 41 showed no significant differences (P Ͼ 0.05) between mice immunized with native and phenol-extracted CPS 3 (data not shown). Similarly, no significant differences were observed in anti-CPS 3 antibody isotypes (total Ig, IgM, and IgG) titers in sera collected on day 42 (P Ͼ 0.05; Fig.  6). These results suggest that immunogenicity of purified native CPS 3 in vivo is independent from phenol-extractable immune-stimulating ligands.
In vivo anti-CPS responses to heat-killed encapsulated intact S. suis serotypes 2, 3, and 9. Following the previous immunizations performed using soluble antigens, we also investigated the CPS-specific antibody response in the context of encapsulated intact bacteria. The titers of the different anti-CPS antibody isotypes were determined in mice that were hyperimmunized by multiple weekly intraperitoneal injections with 1 ϫ 10 9 CFU of heat-killed S. suis. Since mice hyperimmunized with S. suis serotypes 7 and 8 showed no anti-CPS responses (see Fig. S1D and E), these two serotypes were not further studied. Pooled serum from mice hyperimmunized with S. suis serotype 2 showed intermediate levels of IgM and low levels of IgG that are specific to CPS 2, with barely detectable levels of IgG2b and low levels of IgG2c as the only IgG subclasses detected (Fig. 7A). In contrast, pooled serum from mice hyperimmunized with S. suis serotype 3 possessed high levels of IgM and intermediate levels of IgG specific to CPS 3 ( Fig. 7B). Analysis of the anti-CPS 3 IgG subclasses showed high levels of IgG2c in contrast to barely detectable to low levels of IgG1, IgG2b, and IgG3. Finally, pooled serum from mice hyperimmunized with S. suis serotype 9 possessed intermediate levels of IgM and strong levels of IgG specific to CPS 9 (Fig. 7C). Analysis of the anti-CPS 9 IgG subclasses revealed high levels of IgG2b and IgG2c, in addition to low levels of IgG1 and IgG3. For all three serotypes, IgG2c was the dominant IgG subclass induced in hyperimmunized mice (Fig. 7).
In vivo anti-CPS responses to live encapsulated intact S. suis serotypes 2, 3, and 9. While titration of sera from mice hyperimmunized with heat-killed S. suis provided interesting data, we also studied the CPS-specific antibody response following infections with live S. suis, since this is the most relevant model. In the present study, a sublethal infection model was used to better characterize the antibody responses in the absence of significant inflammation and severe systemic clinical signs. During pretrials, different bacterial doses were tested, and sublethal doses (producing little to no mortality and/or severe clinical signs of disease) were chosen for each strain (data not shown). Titration of the different anti-CPS antibody isotypes after primary infection with S. suis serotype 3 revealed strong levels of IgM and low levels of IgG (P Յ 0.01) in comparison to placebo mice ( Fig. 8A to C). A secondary infection with this serotype showed a significant boost for the IgM (P Յ 0.05) but not for the IgG (P Ͼ 0.05) response. In contrast to serotype 3, mice infected with S. suis serotype 2 showed no anti-CPS antibody titers following a primary infection, whereas a secondary infection induced low levels of IgM only (P Յ 0.01) (Fig. 8G to I). Sera from mice infected with S. suis serotype 9 displayed low levels of IgM and IgG that were not significant (P Ͼ 0.05) in comparison to placebo mice after a primary infection but nonetheless were boosted after a secondary infection (P Ͼ 0.05) (Fig. 8J to L). As such, while sublethal infections of mice with S. suis serotypes 2 and 9 induce low IgM anti-CPS responses only after a secondary infection, a sublethal infection with S. suis serotype 3 induce a high primary FIG 6 Isotyping of antibodies produced in mice immunized with phenol-extracted S. suis serotype 3 purified capsular polysaccharide (CPS 3) in TiterMax Gold. Mice (n ϭ 10) were immunized on day 0 and boosted on day 21 with 25 g of either native or phenol-extracted CPS 3. Placebo mice (n ϭ 15) were similarly injected with PBS adjuvanted with TiterMax Gold. Sera were collected on day 42. For titration of anti-CPS antibody isotypes, ELISA plates were coated with native CPS 3 and incubated with 2-fold serial dilutions of sera, and isotypes were detected using specific HRP-conjugated anti-mouse total Ig (IgG ϩ IgM), IgM, or IgG antibodies. The results are expressed as titers for individual mice, with horizontal bars representing the medians. Individuals from the placebo group are indicated by a gray circle, those of the native CPS 3 group are indicated by an orange diamond, and those of the phenol-extracted group are indicated by a blue triangle. Statistically significant differences between groups are indicated (ns, P Ͼ 0.05; **, P Յ 0.01; ***, P Յ 0.001).
production of anti-CPS IgM which was further boosted after a secondary infection. Anti-CPS IgG responses were only observed for serotypes 3 (as a primary response) and 9 (as a secondary response).

CPS 3 immunogenicity following sublethal infections is not influenced by the bacterial subcapsular domain.
Since we demonstrated that CPS 3 is immunogenic by itself (as a soluble purified antigen) but also when associated with the bacterial body (as a particulate antigen), we also opted to evaluate whether the immunogenicity of CPS 3 could be influenced by the composition and/or architecture of the bacterial subcapsular domain. Toward this aim, we used a CPS-switched SS2to3 mutant derived from the same S. suis serotype 2 strain previously used for the sublethal infections.
Titration of the different anti-CPS antibody isotypes in sera from mice that were sublethally infected with the SS2to3 mutant revealed strong levels of IgM (P Յ 0.001) and no significant levels of IgG (P Ͼ 0.05) after a primary infection in comparison to placebo (Fig. 8D to F). After a secondary infection with the SS2to3 mutant, a significant FIG 7 Isotyping of antibodies induced in control mice hyperimmunized with heat-killed S. suis serotype 2 strain S735 (A), serotype 3 strain 4961 (B), and serotype 9 strain 1273590 (C). Hyperimmune mice (n ϭ 3 to 4) were obtained by repeated immunization with 1 ϫ 10 9 CFU/ml heat-killed S. suis by intraperitoneal injection weekly during 4 to 8 weeks. 2 weeks after the last injection, sera were collected and pooled by serotype. For titration of anti-CPS antibody isotypes, ELISA plates were coated with native CPS 2, CPS 3, or CPS 9 and incubated with 2-fold serial dilutions of sera, and isotypes were detected using specific HRP-conjugated anti-mouse total Ig (IgG ϩ IgM), IgM, IgG, IgG1, IgG2b, IgG2c, or IgG3 antibodies. The results are expressed as titer, and the horizontal line behind the bars represents the cutoff; values found below are considered negative (titer Ͻ 100).

FIG 8
Anti-CPS antibody response after infections of mice with sublethal doses of S. suis serotype 3 strain 4961 (1 ϫ 10 6 CFU; A to C), a CPS-switched SS2to3 mutant of strain P1/7 (1 ϫ 10 6 CFU; D to F), serotype 2 strain P1/7 (1 ϫ 10 6 CFU; G to I), and serotype 9 strain 1135776 (1 ϫ 10 5 CFU; J to L). For the primary infection (1ϫ), mice were infected on day 0, followed by serum collection on day 21. For the secondary infection (2ϫ), mice were infected on day 0, reinfected on day 21, then sera were collected on day 42. Placebo mice were similarly injected once or twice with the vehicle solution (sterile Todd-Hewitt broth). Mouse groups were as follows: mice infected 1ϫ or 2ϫ with the serotype 3 and SS2to3 mutant and corresponding placebo mice (n ϭ 10); mice infected with the serotype 2 (1ϫ, n ϭ 4; 2ϫ, n ϭ 9) and corresponding placebo mice (n ϭ 5); mice infected with the serotype 9 (1ϫ, n ϭ 10; 2ϫ, n ϭ 9) and corresponding placebo mice (n ϭ 10). For titration of anti-CPS antibody isotypes, ELISA plates were coated with serotype-matched native CPS and incubated with 2-fold serial dilutions of sera, and isotypes were detected using specific HRP-conjugated anti-mouse total Ig (IgG ϩ IgM), IgM, or IgG antibodies. The results are expressed as titer for individual mice, with horizontal bars representing the medians. Individuals from the placebo groups are indicated by a gray symbol, whereas those of the infected groups are indicated by a colored symbol. Statistically significant differences between groups are indicated (ns, P Ͼ 0.05; *, P Յ 0.05; **, P Յ 0.01; ***, P Յ 0.001).
boost for the IgM (P Յ 0.05) and a significant IgG response (P Յ 0.05) were observed. Comparison of IgM and IgG responses in mice sublethally infected with either S. suis serotype 3 or the SS2to3 mutant after primary and secondary infections showed no significant difference between the two strains (P Ͼ 0.05), although the primary IgG response was lower in mice infected with the SS2to3 mutant ( Fig. 8C and F). As such, the immunogenicity of CPS 3 associated with the whole bacteria appears to be independent from the bacterial subcapsular domain.

DISCUSSION
The original idea behind this study stems from the fact that CPSs 2 and 14 of S. suis are particularly nonimmunogenic, even when associated at the bacterial surface in the course of a primary or booster experimental infection (12,26,27,(43)(44)(45). This low immunogenicity results in undetectable (CPS 14) to low IgM (CPS 2) titers and undetectable IgG titers for both CPSs (26,27). Also, this low immunogenicity cannot just be explained by the presence of sialic acid, known to possess immunomodulatory properties (25,27,46). Following recent reports by our team of new repeating unit structures for S. suis CPSs, especially serotypes 3, 7, 8, and 9, we decided to further investigate the potential immunogenicity of these serotypes, as they are structurally well diversified among themselves and from the previously described sialylated serotypes.
Purified polysaccharides are known to be TI antigens and thus are generally poorly immunogenic. By inducing multivalent B cell receptor (mIg) cross-linking, TI antigens can induce proliferation of B cells and limited IgM responses lacking isotype switching. However, in the presence of a second signal such as TLR signaling, B cells activated by TI antigens undergo isotype switching and secrete substantial amounts of Ig (15). A previous study by Sen et al. has demonstrated that pneumococcal polysaccharide preparations used for vaccine production contain TLR2 and TLR4 ligands and that these TLR ligands substantially enhance primary and secondary antipneumococcal polysaccharide (PPS) responses in mice, especially the type 1 IgG subclasses (21). As such, those unappreciated copurified and/or contaminating TLR ligands in PPS vaccine preparations play an important role in providing the second signal for the induction of an in vivo IgG TI humoral response to purified polysaccharides.
Despite the importance of CPS as protective antigens and their effect on the development of antibody responses, few studies have been dedicated to the characterization of CPS activity on APCs and to the corresponding signaling mechanisms. Moreover, of those that are available, one must be cautious, since many of these studies describing innate immune activity of various purified polysaccharides have failed to rigorously rule out contaminating immune-stimulating ligands (15). It would be logical to assume that structural features of CPSs, such as different repeating unit compositions or glycosidic linkages, are susceptible to producing different immune responses, although such mechanisms remain largely unknown. Our results demonstrated that S. suis purified CPSs of serotypes 1, 1/2, 3, 7, 8, and 9 do not induce the release of key proinflammatory cytokines, which confirms their poor stimulatory activity. However, all these CPSs, including S. pneumoniae CPS 19F, stimulated DC production of two members of the CC family of chemokines, i.e., CCL2 and CCL3. These two chemokines are known to play a major role in the selective recruitment of monocytes, macrophages, DCs, and lymphocytes to sites of inflammation and also known to be produced in the course of S. suis infections (25,47,48). After stimulation with purified CPSs of S. suis and S. pneumoniae, DC production of CCL2 was shown to be independent from TLR2 and MyD88 pathways, whereas CCL3 production was partially dependent upon both TLR2 and MyD88. These observations are in agreement with previous results obtained using purified CPSs of S. suis serotypes 2 and 14 and of GBS serotypes III and V (25), suggesting that DCs appear to respond to CPSs in a patterned manner, seemingly independent from their different structural features. This warrants further investigations into the involvement of members from the large family of lectin receptors as possible candidates (49)(50)(51)(52).
Mouse immunizations demonstrated that purified CPS 3 when adjuvanted with TiterMax Gold, a water-in-oil emulsion, induces the development of strong opsonizing type 1 IgG (mainly IgG2b and IgG3) and IgM antibody responses, in stark contrast to CPSs 2, 7, 8, and 9 (this study) and CPS 14 (27). Immunization of purified CPS 3 alone or combined with Alhydrogel or with Quil-A as adjuvants failed to mount an antibody response. Nonetheless, this makes CPS 3 as the first immunogenic purified CPS of S. suis. The current model for polysaccharide-specific TI humoral immunity, as proposed by Snapper (15), requires two signals to ensure antibody secretion and class switching. Signal 1 is delivered through multivalent B cell receptor cross-linking by polymeric TI antigens. Signal 2 is delivered to CPS-specific B cells by innate cells through cytokine (such as IFN-␥, granulocyte-macrophage colony-stimulating factor, and IL-3) and/or BAFF/APRIL production, which are induced in response to PAMPs (such as TLR ligands). However, we showed that our purified CPS preparations are exempt from phenolextractable contaminants and from nucleic acids and proteins and also that they fail to induce cytokine production by DCs, at least those evaluated in our study. Another possible source for signal 2 in vivo could be from the adjuvants used. A previous study has shown that stimulation of porcine DCs with TiterMax Gold with no antigen induced IL-6 production, whereas stimulation with either Alhydrogel 2% or Quil-A induced IL-6 and/or IL-12 production, suggesting that these adjuvants by themselves can trigger signal 2 following immunization with purified CPSs (53). Yet, only the combination of CPS 3 with TiterMax Gold induced a potent protective response. This could be explained by the particulate form of the water-in-oil emulsion. Relative to soluble antigens, antigens in particulate form are phagocytosed more efficiently, produce higher levels of immune activation, and deliver concentrated antigens per particle than could be achieved through pinocytosis of soluble antigens (54,55). Of note, aluminum salt-based adjuvants, such as Alhydrogel 2%, are themselves particulate, but antigens adsorbed to them do not behave as particulates (54,55). Nevertheless, the adjuvant properties of TiterMax Gold do not benefit to the immunogenicity of other CPSs studied. Taken together, our results suggest that purified CPS 3 possesses unique structural and/or biochemical properties. In this regard, it was observed that water soluble CPS 3 was precipitated in the presence of the detergent sodium deoxycholate, suggesting an amphiphilic character for this polysaccharide (56). The hydrophobic element is most likely the di-N-acetylated D-bacillosamine, as acetylation increases hydrophobicity (such as is the case for chitin versus chitosan [57]), which is then linked to the negatively charged D-glucuronic acid, together forming the amphiphilic disaccharide repeating unit of CPS 3 (Fig. 1). In turn, this amphiphile could be copresented at the water-oil interface and/or associate with the amphiphilic block copolymer component of the TiterMax Gold emulsion in a similar fashion to its presentation at bacterial surfaces, and fully benefit from its adjuvant properties (58). Nonetheless, further studies into the structural and biochemical properties of purified CPS 3 and their effect on the induction of humoral immunity are warranted. The immunogenicity of CPS 3 was also demonstrated in mice injected with serotype 3 intact whole bacteria, either heat killed via hyperimmunization or live via sublethal infections. Mice infected/immunized with the serotype 3 bacteria produced a very robust IgM anti-CPS antibody response in addition to isotype-switched IgGs. In addition, and in contrast to soluble purified CPS, a very strong anti-CPS 9 IgG response was observed in mice hyperimmunized with heat-killed serotype 9, whereas mice sublethally infected showed a weak IgM and IgG response only after a secondary infection. Due to the virulence of the serotype 9 strain used for sublethal infections, lower doses had to be used, which might explain the lower response observed. This displayed immunogenicity for CPSs 3 and 9 in the context of whole intact bacteria is in agreement with the previously mentioned model for polysaccharide-specific TI humoral immunity: as CPSs are displayed at the bacterial surfaces, there are plenty of bacterial PAMPs, such as TLR2 ligands, to trigger signal 2 by APCs leading to stronger antibody secretion and isotype switching (15). Differences in these signals might also explain divergences in IgG switching, IgG subclasses, and/or presence/amplitude of memory (boost) anti-CPS responses between serotypes. Immunization/infection with intact serotype 2 whole bacteria led to the lowest anti-CPS response (in quantity and quality). However, a recipient serotype 2 strain displaying a serotype 3 CPS induced optimal anti-CPS 3 antibody response, confirming the poorly immunogenic nature of CPS 2 and that anti-CPS 3 response to whole bacteria is independent from the bacterial subcapsular domain. A previous study has reported an important role of bacterial subcapsular domain in the anti-CPS response against S. pneumoniae serotype 14 by studying an asialo mutant of GBS serotype III, which expresses an identical CPS (59). The absence of this effect in our model could be due to the fact that intraspecies serotype switching was used to better reflect serotype divergences within the S. suis bacterium.
Even though the predominant IgG subclasses differ between purified CPS and intact bacteria for serotype 3, or among serotypes for hyperimmunized mice, the anti-CPS response remains biased toward the type 1 subclasses of IgG (IgG2a/IgG2c, IgG2b, and IgG3), which are known to be superior in both opsonophagocytosis activity and complement activation than the type 2 (IgG1) subclass (60,61). Also, sublethal infections with serotypes 2, 3, and 9 led to boosted IgM and/or IgG titers following a secondary infection, a feature of T cell-dependent responses, in contrast to the TI response to purified CPS 3, which failed to show a significant boost following the second dose. Taken together, these observations demonstrate a number of consistent differences between intact bacteria in general versus purified CPSs, suggesting that different mechanisms are involved in the development of anti-CPS humoral responses (15,59,(62)(63)(64).
CPS immunogenicity might differ between species. For example, rabbits are used to generate the anti-CPS specific antisera used by diagnostic services to perform S. suis serotyping (65). Therefore, rabbits are able to generate anti-CPS antibodies to all known S. suis serotypes after hyperimmunization (66)(67)(68)(69). Nevertheless, our findings with the mouse model seem to correlate with the scarcely available data on CPS immunogenicity in the natural host, the pig. Some studies have demonstrated the generation of anti-CPS 2 antibodies in swine, albeit at very low levels, depending on the model used (12,44,45,70,71). Indeed, most studies were performed by either (i) immunization with whole killed bacteria or with CPS-conjugate preparations or (ii) using either natural infection (i.e., samples from the field) or experimental infection. One study has shown that purified CPS 2 can induce anti-CPS 2 specific antibodies in swine when using Freund's complete adjuvant (43). A single study has reported low levels of anti-CPS 7 in naturally exposed piglets (72). Therefore, swine seems to have a B cell repertoire able to recognize, at least, CPSs 2 and 7. Nevertheless, as shown in mice, immunogenicity is low unless immunization strategies such as CPS conjugates, strong adjuvants, or hyperimmunization are used. Future studies on swine anti-CPS antibody generation are warranted.
The results obtained in this study using either purified CPSs or whole bacteria have strong implications for the design of effective vaccines against S. suis and provide novel insights into the complexity of the anti-CPS response. Indeed, we have previously demonstrated that a serotype 2 glycoconjugate vaccine is able to induce a protective response in mice and pigs (12), thus highlighting the importance of CPS as a vaccine target. Consequently, understanding the development of the anti-CPS response to different S. suis serotypes is of utmost importance for vaccine development against this economically important pathogen and zoonotic agent. Our data further support the importance of the Th1 bias and/or the type 1 antibody response in the immunopathogenesis of S. suis infections and development of a protective response. Finally, this study provides important information on the divergent evolution of CPS serotypes with highly different structural and/or biochemical properties within S. suis and their interaction with the immune system.

MATERIALS AND METHODS
Bacterial strains and culture conditions. Bacterial strains that were used in this study are listed in Table 1. Isolated colonies on sheep blood agar plates were inoculated in 5 ml of Todd-Hewitt broth (THB; Oxoid, Nepean, Ontario, Canada) and incubated for 8 h in a water bath at 37°C with agitation at 120 rpm. Working cultures were prepared by transferring 10 l of 8-h cultures diluted 1:1,000 with phosphate-buffered saline (PBS) into 30 ml of THB, which was incubated for 16 h. Bacteria were washed three times and resuspended in PBS to obtain a concentration between 1 ϫ 10 8 to 1 ϫ 10 9 CFU/ml. Heat-killed bacterial cultures were obtained as previously described (73). Briefly, overnight cultures were washed three times with PBS and then resuspended in 30 ml of fresh PBS. A sample was taken to perform bacterial counts on Todd-Hewitt agar (THA). Bacteria were immediately killed by incubating at 60°C for 45 min and then cooled on ice. Bacterial killing was confirmed by absence of growth on blood agar for 48 h.
A CPS-switched 2-to-3 mutant (SS2to3) derived from the S. suis serotype 2 strain P1/7 was also used for the sublethal infections ( Table 1). The first step in the creation of this mutant consists of the genetic excision in this receiver strain of the whole cps-2 locus genes (responsible for CPS expression). For this, a DNA fragment comprising a chloramphenicol resistance gene (cat) cassette flanked by ϳ1 kbp of the upstream and downstream regions of the cps-2 gene cluster (in the receiver strain) was amplified by overlap-extension PCR. The receiver strain was transformed with the cat fragment to yield a nonencapsulated mutant (ΔCPS2). Then, the receiving ΔCPS2 mutant strain was transformed with a whole genome of a donor S. suis serotype 3 strain to yield the desired serotype-switched mutant SS2to3 through the replacement of the cat fragment with the cps-3 locus (M. Okura, unpublished data). Verification of this mutant by whole-genome sequencing (accession no. WABX00000000), transmission electron microscopy, serology, and NMR analyses of its purified CPS all confirmed its phenotype as a CPS 3-expressing well-encapsulated serotype-switched mutant (M. Okura, unpublished data). While expressing CPS 3 as its capsule, the subcapsular antigens of this mutant are from a phenotypically very different serotype 2 strain.
S. suis capsular polysaccharides: production and purification. The following methods for CPS production and purification were adapted from previous publications (23,(28)(29)(30)(31). Briefly, 6 liters of fresh THB was inoculated with the appropriate strain and incubated overnight. For more information on the strains used for CPS production, refer to Table 1.
Depending on the sensitivity of the specific CPS to hydrolysis, two methods of CPS extraction were used. For serotypes 2, 1, 1/2, and 3, the autoclave method was used (23,28,30). Briefly, bacterial cells from the 6-liter culture were pelleted by centrifugation at 10,000 ϫ g for 40 min, suspended by repeated pipetting in a buffer containing 33 mM phosphate and 145 mM NaCl (pH 8.0), and chilled. The bacterial suspension was autoclaved at 121°C for 15 min. The supernatant containing the crude CPS was recovered by centrifugation at 9,000 ϫ g for 50 min. For serotypes 7, 8, and 9, the water method was used (29)(30)(31). Briefly, the bacterial pellet from the 6-liter culture was resuspended in deionized water (ddH 2 O). The bacteria were killed by heating at 60°C for 45 min, as the standardized protocol for S. suis. The content of the tubes was freeze-dried for 72 h. Then, bacterial cells were stirred with ddH 2 O (3 g of dry cells in 100 ml) overnight at room temperature. The supernatant containing the crude CPS was recovered by centrifugation at 9,000 ϫ g for 50 min.
The crude CPSs were further purified from the extracted material by solvent extraction with chloroform to remove lipids, precipitation with 25% (vol/vol) ethanol and 0.1 M CaCl 2 to remove nucleic acids, precipitation of CPS and other macromolecules with 80% (vol/vol) ethanol, dialysis, and freezedrying. The final purification step to remove proteins was performed by gel filtration chromatography as previously described (30). Briefly, an XK-26/100 column packed with Sephacryl S-400 HR (GE Healthcare Life Sciences, Uppsala, Sweden) was used and eluted with 50 mM NH 4 HCO 3 (for serotypes 2, 1, 1/2, and 3) or with 50 mM NaCl (for serotypes 7, 8, and 9) at a flow rate of 1.3 ml/min, using an ÄKTA Purifier 10 system (GE Healthcare Life Sciences), including a UV-900 monitor, and equipped with a Knauer Smartline 2300 RI detector (Knauer, Berlin, Germany) connected to the system via an AD-900 analog/digital converter (GE Healthcare Life Sciences). Fractions giving a positive RI signal but no absorption at 280 and 254 nm were pooled and freeze-dried. Dot blot confirmation of CPS-containing fractions with reference antisera has already been performed previously (23,(28)(29)(30)(31). To desalt, the purified material was dissolved in water, dialyzed against ddH 2 O for 24 h at 4°C, and finally freeze-dried.
S. pneumoniae capsular polysaccharide. Purified pneumococcal polysaccharide of type 19F was manufactured by Pfizer and purchased through the American Type Culture Collection (ATCC; 84-X).
CPS quality controls. Each CPS was subjected to rigorous quality control tests as previously described (25) on 1-mg/ml samples. Nucleic acids were quantified using an ND 1000 spectrophotometer (NanoDrop, Wilmington, DE). The absorbance was measured at 230 and 260 nm on CPS samples. Calculations were done with the NanoDrop software. According to the manufacturer, results are reproducible between 2 and 100 ng/l. Proteins were quantified on CPS samples using the modified Lowry protein assay kit (Thermo Fisher Scientific, Saint-Laurent, Quebec, Canada) in a 96-well microplate according to the manufacturer's instructions. The calculated limit of detection (P Յ 0.05) was 5 g/ml for the modified Lowry assay. Proteins were also quantified on CPS samples using a NanoOrange protein quantitation kit (Molecular Probes, Eugene, OR) according to manufacturer's instructions by heating samples into microtubes, followed by transfer to a 96-well microplate for fluorescence reading. The calculated limit of detection (P Յ 0.05) was 4 g/ml for the NanoOrange assay. To perform the titration of mouse Ig isotypes, day-42 serum was serially diluted (2-fold) in PBST, and antibodies were detected using either HRP-conjugated goat anti-mouse total Ig [IgG ϩ IgM] as mentioned above, goat anti-IgG (Fc␥ fragment specific; Jackson Immunoresearch), goat anti-IgM diluted 1:1000, goat anti-IgG1, goat anti-IgG2b, goat anti-IgG2c, or goat anti-IgG3 diluted 1:400 (Southern Biotech, Birmingham, AL). For mouse serum titration, the reciprocal of the last serum dilution that resulted in an OD 450 of Յ0.2 (as a preestablished cutoff for comparison purposes) was considered the titer of that serum. For representation purposes, negative titers (less than or equal to the cutoff) were given an arbitrary titer value of 50.
To control interplate variations, an internal reference positive control was added to each plate. For titration of mouse antibodies, this control was a pool of sera from hyperimmunized mice (produced as described above). Reaction in TMB was stopped when an OD 450 of 1.0 was obtained for the positive internal control. Optimal coating conditions for the anti-CPS and anti-S. suis ELISAs, optimal dilutions for the positive internal control sera, and the HRP-conjugated anti-mouse antibodies were determined during preliminary standardizations.
Opsonophagocytosis assay for serotype 3. Instead of using a cell line or a single cell type, the OPA was standardized using whole blood from naive mice (12,77). This model considers all blood leukocytes and thus represents a more realistic model of the complex interactions between all immune cells and the bacteria during a systemic infection, as is the case for S. suis.
Murine whole blood OPAs were performed as previously described (12,77). Blood was collected by intracardiac puncture from naive C57BL/6 mice, treated with sodium heparin, and then diluted to obtain 6.25 ϫ 10 6 leukocytes/ml in complete cell culture medium. All blood preparations were kept at room temperature. Using washed bacterial cultures of S. suis serotype 3 strain 4961 (Table 1) grown as described above, final bacterial suspensions were prepared in complete cell culture medium to obtain a concentration of 1.25 ϫ 10 6 CFU/ml. The number of CFU/ml in the final suspension was determined by plating samples onto THA. All bacterial preparations were kept on ice.
Diluted whole blood (5 ϫ 10 5 leukocytes) was mixed with 5 ϫ 10 4 CFU of S. suis (multiplicity of infection [MOI] of 0.1) and 40% (vol/vol) of sera from naive or immunized mice in a microtube to a final volume of 0.2 ml. The tube tops were pierced using a sterile 25-gauge needle, and then the microtubes were incubated for 2 h at 37°C with 5% CO 2 , with gentle manual agitation every 20 min. After incubation, viable bacterial counts were performed on THA. Tubes with addition of naive rabbit serum or rabbit anti-S. suis type 3 serum (65) were used as negative and positive controls, respectively. The bacterial killing percentage was determined by using the following formula: percentage of bacteria killed ϭ [1 Ϫ (bacteria recovered from sample tubes/bacteria recovered from negative-control tubes with naive mouse sera)] ϫ 100. Final OPA conditions were selected based on several pretrials using different incubation times and MOIs (data not shown).
Phenol extraction of S. suis serotype 3 CPS. Phenol extraction of CPS 3 was conducted as described by Sen et al. (21) with some modifications. Native CPS 3 was dissolved in water containing 0.2% triethanolamine (TEA) to a final CPS concentration of 5 mg/ml. Sodium deoxycholate was omitted, since it caused CPS 3 to precipitate. Water-saturated phenol (0.5 ml) was added, and the tube was vortexed intermittently for 5 min. The solution was left standing for 5 min at room temperature for phase separation, then was placed on ice for an additional 5 min. The material was centrifuged for 2 min at 10,000 ϫ g at 4°C. The upper aqueous layer was transferred to another tube. The phenol layer was re-extracted with 0.5 ml of water containing 0.2% TEA. The aqueous layers were pooled and re-extracted with 1 ml of water-saturated phenol. Pooled aqueous layers were adjusted to 80% ethanol containing 30 mM sodium acetate and left overnight at 4°C to precipitate. The material was then centrifuged for 10 min at 10,000 ϫ g at 4°C. The precipitate was washed with 1 ml of ice-cold absolute ethanol and then air-dried. CPS 3 was redissolved in 1 ml of water and freeze-dried to be weighed subsequently.
Dot blotting. Dot blot analyses were performed essentially as described previously by Van Calsteren et al. (28). Ten microliters (each at 1 mg/ml) of both native and phenol-extracted CPS 3 and of CPS 2 (used as negative control) were blotted on a polyvinylidene difluoride Western blot membrane (Bio-Rad, Hercules, CA). The membrane was blocked for 1 h with a solution of Tris-buffered saline (TBS) containing 2% skim milk, followed by 2 h of incubation with rabbit anti-S. suis type 3 serum. The membrane was washed three times with TBS, and anti-rabbit HRP-conjugated antibody (Jackson Immunoresearch) was added for 1 h. The membrane was washed three times with TBS and revealed with a 4-chloro-1-naphthol solution (Sigma-Aldrich).
NMR spectroscopy. Both native and phenol-extracted CPS 3 were exchanged in 33 mM phosphate buffer pD 8.0 in D 2 O (99.9 atom% D), freeze-dried, and dissolved in D 2 O (99.96 atom% D). NMR spectra were acquired on CPS samples at concentrations of 0.6 to 0.7% at 11.75 T using a Bruker Avance 500 spectrometer equipped with a 5-mm triple resonance TBI probe with 1 H, 13 C, and 109 Ag-31 P channels at 60°C using standard Bruker pulse sequences at the Centre Régional de Résonance Magnétique Nucléaire (Department of Chemistry, University of Montreal). For conventional 1 H spectra, 32 K complex data points were acquired after a 30°pulse with a digital resolution of 0.18 Hz/point and processed off-line using SpinWorks v4.2.8.0 (Kirk Marat, home.cc.umanitoba.ca/~wolowiec/spinworks/) by exponential multiplication with a 0.2-Hz line broadening factor, zero filling, complex Fourier transform, phase correction, and fifth-order polynomial baseline correction. 1 H chemical shifts ␦ in ppm were referenced to internal deuterated 2,2-dimethyl-2-silapentane-5-sulfonate at ␦ 0 as recommended by Wishart et al. (78).
Experimental mouse sublethal infections. Six-week-old female C57BL/6 were injected intraperitoneally with sublethal doses of the S. suis strains used (1 ϫ 10 6 CFU for serotype 2, serotype 3, and CPS-switched SS2to3 mutant; and 1 ϫ 10 5 CFU for serotype 9). For more information on the strains used for experimental infections, refer to Table 1. These sublethal doses were chosen since they produced mild clinical signs and low mortality in our infection model in order to follow the antibody response. For the primary infection model, groups of mice (n ϭ 4 to 10) were infected on day 0 followed by serum collection on day 21. For the secondary infection model, groups of mice (n ϭ 9 or 10) were infected on day 0, reinfected on day 21; sera were then collected on day 42. Placebo groups of mice (n ϭ 5 to 10) were similarly injected once or twice with the vehicle solution (sterile THB). Infected animals were monitored at least twice daily for the first 7 days after each infection and then daily until the end of the study. Blood bacterial titers were assessed 24 h following the first injection on day 0 by collecting 5 l of blood from the tail vein. Proper dilutions in PBS were plated onto THA.
Statistical analyses. Parametric data are expressed as means Ϯ standard errors of the means (SEM) and were analyzed for significance using one-way analysis of variance (ANOVA) followed by the Tukey test. Nonparametric data are shown with the median and were analyzed for significance using either the Mann-Whitney Rank Sum Test (to compare two groups) or the Kruskal-Wallis ANOVA on ranks, followed by the Dunn procedure (to compare three groups or more). Statistical analyses were performed using Systat SigmaPlot v11.0.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.4 MB.