Polysaccharide structure dictates mechanism of adaptive immune response to glycoconjugate vaccines

We had previously shown that a clear biomarker for Tcarb-dependent responses was a polysaccharide-specific antibody response after priming with a polysaccharide covalently linked to a carrier protein and boosting with the same polysaccharide linked to a different and unrelated carrier protein. To investigate the involvement of Tcarbs in the humoral immune response to different glycoconjugates, we performed priming and boosting immunization experiments with glycoconjugates made with Vi, GBSIb, Hib, and MenC CPSs. BALB/c mice were primed at the beginning of the experiment and boosted 14 d later with different antigen combinations. One week after the boost, serum levels of polysaccharide-specific IgG were determined. For Vi, GBSIb, and Hib glycoconjugates, boosting with a glycoconjugate containing the same polysaccharide but a heterologous carrier protein induced polysaccharide-specific IgG titers of the same magnitude as those seen after priming and boosting with glycoconjugates containing the same carrier (Fig. 1A). As reported with GBSIII and Pn3P (5, 6), this result supports a Tcarb-dependent mechanism for all three of these glycoconjugates.

In marked contrast to the above results, priming and boosting with MenC glycoconjugates containing heterologous carrier proteins induced significantly lower levels of MenC-specific IgG than did primary and secondary immunization with either a MenC-OVA (ovalbumin) conjugate or a MenC-CRM₁₉₇ (nontoxic mutant of diphtheria toxin) conjugate. This result was confirmed with a combination of different carriers, including MenC-TT (tetanus toxoid), MenC-HEL (hen egg lysozyme), MenC-OVA, and MenC-CRM₁₉₇ (Fig. 1B). We also found that MenC-specific IgM levels were similar in groups immunized with different conjugate combinations, a result suggesting a similar level of B cell activation independent of the carrier (SI Appendix, Fig. S1A). To exclude the possibility that we were seeing a T cell-independent response, we treated mice with CD4-specific blocking antibody during the interval between priming and boosting with MenC-CRM₁₉₇. The excellent booster IgG response observed in mice treated with isotype control antibody was significantly reduced in mice treated with anti-CD4 (SI Appendix, Fig. S1B). The failure of MenC conjugates prepared with different carrier proteins to induce a booster IgG response suggested that perhaps these glycoconjugates were not able to induce Tcarbs and that Tcarbs were not essential in regulating MenC polysaccharide-specific IgG responses.

On the basis of the previous results, we theorized that carrier protein/peptide-specific CD4⁺ T cells may be the only T cell subset essential to regulate the response to MenC conjugate vaccines (9, 10). To further test this hypothesis, we primed mice with a combination of MenC-CRM₁₉₇ plus unconjugated OVA protein, and we boosted these mice with MenC-OVA. We hypothesized that priming with this combination would yield MenC-specific memory B cells resulting from MenC-CRM₁₉₇ as well as OVA-specific memory helper T cells generated from the OVA protein component. Interestingly, after boosting with MenC-OVA, mice primed with the mixture of MenC-CRM₁₉₇ and OVA protein had MenC IgG levels similar to those in mice primed and boosted with MenC-OVA. The MenC IgG levels were significantly lower in mice primed with MenC-CRM₁₉₇ without OVA (Fig. 2A and SI Appendix, Fig. S2 A and B). This result suggests that in MenC glycoconjugates carrier protein-activated T cells are essential in inducing a humoral immune response and antibody class switching.

To further explore the requirements for cooperation between B and T cells, we performed another series of immunization experiments. We first primed mice with MenC-CRM₁₉₇ and then boosted them with either MenC-CRM₁₉₇ or MenC and CRM₁₉₇ protein (either alone or physically mixed but not conjugated). Booster IgG responses occurred only in mice that received the MenC-CRM₁₉₇ conjugate for both primary and secondary immunization. Similarly, priming of mice either with unconjugated MenC or CRM₁₉₇ protein alone or with a mixture of unconjugated MenC and CRM₁₉₇ protein did not support a robust secondary IgG response upon boosting with the MenC-CRM₁₉₇ conjugate (SI Appendix, Fig. S2C). This result suggested that a conjugate is required to stimulate MenC-specific memory B cells.

To directly examine the contribution of carrier-specific T cells to MenC-specific booster IgG responses, we performed adoptive transfer experiments (Fig. 2B). Donor groups of BALB/c mice were immunized with either MenC-CRM₁₉₇ or OVA, and splenic and lymph node B and CD4⁺ T cells from each group were purified. Three groups of recipient mice all received B cells from the MenC-CRM₁₉–immunized donors. Groups A and B also received CD4⁺ T cells from MenC-CRM₁₉₇–immunized donors, while group C received CD4⁺ T cells from OVA-immunized donors. One day after adoptive transfer, group A recipient mice were actively immunized with MenC-CRM₁₉₇, and groups B and C were immunized with MenC-OVA. The resulting IgG and IgM antibody titers in the three groups are shown in Fig. 2C. Groups A and C had significantly higher MenC-specific IgG titers than group B, while MenC-specific IgM levels were similar in the three groups. In a previously reported experiment supporting the role of Tcarbs in the immune response to glycoconjugates (6), Middleton et al. adoptively transferred B cells and CD4⁺ T cells from mice immunized with Pn3P-KLH (keyhole limpet hemocyanin) to recipient mice and then immunized the recipient mice with either Pn3P-KLH or Pn3P-OVA. The similar IgG levels in these two groups supported a Tcarb-mediated response. In contrast to these results with the Pn3P conjugate, our observations with MenC glycoconjugates suggested that a mechanism in which carrier protein/peptides are essential in recruiting T cell help is operative.

The structures of the GBSIII and MenC glycoconjugates were very different, though in both the polysaccharides were conjugated to carrier proteins through reductive amination. Whereas the GBSIII conjugate was cross-linked by oxidative creation of aldehydes on a fraction (∼30%) of the side-chain terminal sialic acid residues of the repeating unit, the MenC conjugate was linked at a single site on the reducing end of the polysaccharide (Fig. 3A). In addition, MenC depolymerization by mild acid hydrolysis and oxidation resulted in a size of 10–30 kDa, while GBSIII used for conjugation maintained its original size (>100 kDa). Thus, we hypothesized that conjugation chemistry might affect T cell recognition of carbohydrates. To increase the resemblance to the structure of a cross-linked glycoconjugate, we conjugated MenC through the carboxylic acid to derivatized carrier protein treated with adipic acid dihydrazide (ADH) linker (SI Appendix, Fig. S3) (11). This method activated a fraction of repeating units along the chain of MenC without reducing its molecular size (>100 kDa). The cross-linked MenC conjugate was significantly larger than the end-linked MenC conjugate, as shown by size exclusion chromatography (Fig. 3B).

We performed priming and boosting immunization experiments with the cross-linked MenC glycoconjugates containing either the same or different carrier proteins. Mice primed and boosted with MenC glycoconjugates containing the same carrier protein—either MenC-ADH-CRM₁₉₇ or MenC-ADH-TT—had strong booster MenC-specific IgG responses. However, priming and boosting with MenC conjugates containing different carrier proteins induced significantly lower titers of MenC-specific IgG (Fig. 3C). This result was similar to that observed with MenC linked at a single reducing end of the polysaccharide to either the same or different proteins (Fig. 1B). Therefore, conjugation chemistry does not seem to explain the failure of MenC glycoconjugates to induce Tcarbs.

We previously reported that, following immunization with a glycoconjugate vaccine, a depolymerized form of GBSIII (∼10 kDa) bound to an MHCII binding peptide can be presented and the carbohydrate recognized by the T cell receptor (TCR) (5). T cell recognition of carbohydrates was also reported for zwitterionic polysaccharides (12), which presumably bind directly to MHCII by electrostatic interactions. In this study, we speculated that MenC polysaccharide might fail to be presented in the context of MHCII on the antigen-presenting cell (APC) surface. One possible underlying reason could be the sensitivity to acidic hydrolysis of MenC, which is a linear homopolymer of α(2 → 9)-linked sialic acid. It is possible that the ketosidic linkage between sialic acid repeating units on MenC is hydrolyzed to an extremely small size in the endolysosome and therefore cannot be presented to or recognized by TCRs.

We conducted a flow cytometry analysis to assess this possibility by determining whether antigenically active MenC carbohydrates can be presented on the APC surface. Bone marrow-derived dendritic cells (BMDCs) from wild-type mice were incubated with unconjugated or conjugated GBSIII or MenC for 18 h. The cells were collected and stained with a monoclonal antibody specific for GBSIII or MenC at 4 °C and then incubated with a fluorophore-conjugated secondary antibody. Consistent with previously published results, the control saccharide (GBSIII) was detected on the surface of the cells incubated with the GBSIII conjugate. However, no BMDC surface presentation of MenC was detected after incubation of the cells either with unconjugated MenC or with conjugated MenC-TT or MenC-CRM₁₉₇ (Fig. 4A). Thus, processed MenC carbohydrate was not loaded onto the APC surface, was altered structurally, or was depolymerized to a very small size. Any of these possibilities might explain why the MenC antigenic epitope could not be detected on the surface of the APC by the monoclonal antibody and why it could not be recognized by the TCR, with consequent Tcarb induction.

We then examined MenC polysaccharide processing in APCs. CPSs are taken up into the APC endosome and are depolymerized by oxidative agents such as reactive oxygen species and reactive nitrogen species (5, 12, 13) into smaller carbohydrates, generally ∼10 kDa in size. One possible explanation for lack of MenC detection on the APC surface is that MenC is digested through oxidative depolymerization or acidic hydrolysis to an extremely small size so that it cannot be recognized by the TCR. Accordingly, we assessed the size of the MenC saccharide within the APC endolysosome. We incubated radioactively labeled MenC {[³H]-MenC} with Raji B cells for 18 h, then isolated and lysed the endolysosome. We ran the lysate on a size-exclusion HPLC column and found that MenC was substantially depolymerized to a size similar to that of a monomer of sialic acid (Fig. 4B). We performed parallel experiments with a control polysaccharide—polysaccharide A (PSA) from Bacteroides fragilis—that is known to be processed to a size of ∼15 kDa (12). We compared the molecular size of the endosomally depolymerized MenC to that of similarly processed [³H]-PSA on Aquagel-OH 20 (SI Appendix, Fig. S4) and found that the size of the processed saccharides differed substantially: PSA was at least 7 kDa, and MenC appeared to be much smaller at ∼200 Da. To further define the mechanisms involved in the depolymerization of MenC, we incubated Raji B cells with [³H]-MenC in the presence of either a reactive oxygen species inhibitor (4-OH TEMPO) or an ATPase proton pump inhibitor (bafilomycin A1). Interestingly, less depolymerization of the polysaccharide was detected in the endolysosome after incubation with either inhibitor. In bafilomycin-treated cells, a small peak was seen at the void volume of the column that was not observed with 4-OH TEMPO treatment. When treated with a combination of the two inhibitors, a similar upward size shift was observed (Fig. 4C). This result suggested that both oxidative depolymerization and acidic hydrolysis contribute to MenC depolymerization, with acid hydrolysis perhaps being somewhat more important.

Six-week-old female BALB/c mice were purchased from Taconic Biosciences. All mouse experiments were approved by the Harvard Medical Area Standing Committee on Animals (Animal Protocol Is00000636).

Purified MenC, the MenC-CRM₁₉₇ conjugate, Vi polysaccharide, the Vi-CRM₁₉₇ conjugate, Hib polysaccharide, the Hib-CRM₁₉₇ conjugate, and CRM₁₉₇ protein were obtained from GSK Vaccines and GSK Vaccines Institute for Global Health. The Hib-OMPC (outer membrane protein complex) conjugate was purchased from Merck. GBSIb was isolated and purified from type Ib group B Streptococcus. Two conjugation methods were applied: cross-linked GBSIb and end-linked MenC conjugates were made through reductive amination, as previously described (8); cross-linked MenC and Vi conjugates were made using EDC [1-ethyl-3-(3-dimethylaminopropyl)carbodiimide], with an ADH linker to couple the carboxylic acid groups from the protein and the carbohydrate.

Groups of BALB/c mice were primed on day 0 and boosted on day 14 by i.p. injection of the antigen of interest (4–6 μg as polysaccharide content) in PBS mixed with 0.5 mg of alum hydroxide gel adjuvant. At least four mice were immunized in each experiment. Mice were bled from the tail vein 1 wk after boosting immunization. Levels of carbohydrate-specific antibodies in the serum were determined by solid-phase ELISA, as previously described (29).

Groups of donor BALB/c mice were primed and boosted with 4 µg of MenC-CRM₁₉₇ as saccharide content given i.p. at 3-wk intervals. Mice were killed 5 d after boosting immunization. CD4⁺ T cells were isolated from spleens and lymph nodes of mice immunized with either MenC-CRM₁₉₇ or OVA and negatively selected with a mouse CD4⁺ T cell isolation kit (130-104-454, Miltenyi Biotech). B cells from mice immunized with MenC-CRM₁₉₇ were isolated with a mouse B cell isolation kit (130-104-443, Miltenyi Biotech). CD4⁺ T cells (10⁷) from mice immunized with MenC-CRM₁₉₇ or OVA and B cells (10⁷) from MenC-CRM₁₉₇–immunized donors were adoptively transferred to recipient mice. The recipient mice were immunized 1 d after adoptive transfer with MenC-CRM₁₉₇ or MenC-OVA.

BMDCs from wild-type mice were incubated with antigen (GBSIII or the GBSIII-TT conjugate, MenC or the MenC-CRM₁₉₇ or MenC-TT conjugate) for 18 h at 37 °C. After incubation, cells were collected, washed five times with PBS, and labeled at 4 °C first with either GBSIII-specific monoclonal antibody or MenC-specific monoclonal antibody and then with a fluorophore-labeled secondary antibody. Surface staining was assessed by flow cytometry (MACSQuant Analyzer).

Raji B cells (10⁸) were cultured in the presence of 1 mg of [³H]-MenC for 18 h at 37 °C. Cells were then washed five times with PBS to remove unreacted [³H]-MenC. Cell lysis was performed by passage of cells through a 27-gauge needle in 250 mM sucrose with 10 mM Tris⋅HCl, pH 7.5. Differential centrifugation was then used to fractionate the lysed cells into endolysosome and cell-membrane fractions, as previously described (5, 12, 30). Endolysosomal fractions were solubilized by boiling for 20 min in 1% SDS and analyzed by size exclusion chromatography on either Agilent ProSEC 300S [MW (molecular weight) range, 1,500–800,000] or Aquagel-OH 20 (MW range, 100–20,000) with an UltiMate 3000 system.

Statistical significance was determined with the ordinary one-way ANOVA, using GraphPad Prism 7.0c. Data with P values of ≤0.05 were considered statistically significant (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001).