Salmonella Typhimurium GMMA (obtained from 1418 ΔtolR mutant strain) and meningococcal B (MenB) GMMA (produced from four knock-out ΔsynX, Δctra, Δgna33, and ΔlpxL1 N. meningitidis strains) were produced and characterized as previously described (15, 23). Meningococcal serogroup A (MenA), meningococcal serogroup C (MenC), and H. influenzae type b (Hib) oligosaccharides were provided by GSK. Vi and streptococcal Group A Carbohydrate (GAC) PS were purified as previously described (24–27). MenA and Vi PS of reduced length were generated as previously described (26, 28).

Conjugates were synthesized as described below. The main characteristics of all the conjugates tested in this study are reported in Table 1 .

Conjugates were characterized by micro BCA for total protein recovery (21), while amount of saccharide antigen linked was determined by HPAEC-PAD after performing acid hydrolysis directly on GMMA as previously described (24, 32–36). It was verified that there was no interference from GMMA in the quantification of each saccharide. A known amount of the conjugated PS was physically mixed to GMMA, and it was verified that analysis by HPAEC-PAD gave results comparable with those obtained by testing the same amount of the PS alone. For MenA, MenC, Hib, and GAC saccharides, conjugate formation was also confirmed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE)/Western blotting as previously described (21). For MenA, MenC, and GAC conjugates, polyclonal sera internally generated in mice were used as primary antibodies, while for Hib conjugate, a commercial antibody (Bacto Hib DIFCO 2236-50-1) was used. NanoTracking Analysis (NTA) was used to count the number of GMMA particles in solution and estimate the number of PS chains per GMMA. NS300 Nanosight instrument (Malvern) equipped with a CMOS camera and a 488-nm monochromatic laser beam was used. Data acquisition and processing were performed using NTA software 3.2 build 3.2.16, and more details on the analysis can be found in De Benedetto et al. (23).

Percentage of free saccharide was calculated by solid-phase extraction (SPE) using a C4 cartridge (Vydac BioSelect) followed by HPAEC-PAD analysis for MenA, MenC, and Hib (33, 35), by high-performance liquid chromatography–size-exclusion chromatography (HPLC-SEC) (refractive index detection) for Vi and GAC conjugates.

All animal sera used in this study were derived from mouse or rat immunization experiments performed at the GSK Animal Facility in Siena or at Toscana Life Sciences Animal Facility (Siena, Italy), in compliance with the relevant guidelines (Italian D. Lgs. n. 26/14 and European directive 2010/63/UE) and the institutional policies of GSK. The animal protocols were approved by the Animal Welfare Body of GSK, Siena, Italy, and by the Italian Ministry of Health (Approval number 804/2015-PR) and Animal Welfare Body of Toscana Life Sciences and by the Italian Ministry of Health (Approval number 479/2017-PR).

CD1 5-week-old female mice were immunized subcutaneously (s.c) or intramuscularly (i.m.) at days 0 and 28. CD1 5-week-old female nude mice (devoid of mature T cells) were immunized s.c. at days 0 and 28 (37). Crl : CD 8-week-old female rats were immunized i.m. at days 0 and 28. Mice and rats were bled from the retromandibular plexus and the tail vein, respectively. Rats were preheated for 5 min in a warming cage at 37°C before bleeding. Final bleed in rats was performed under general anesthesia (alfaxalone 20 mg/kg + medetomidine 0.05 mg/kg + fentanyl 0.1 mg/kg). Blood was kept at 37°C up to 2 h or at RT up to 3 h in the untreated collection tubes and then centrifuged for 10 min at 2,851 rcf, 4°C before serum collection. Animal models, immunization routes, and schemes were selected according to the PS antigens tested. Anti-antigen-specific IgG levels were measured at days −1, 27, and 42 (40 for the study in rats) by enzyme-linked immunosorbent assay (ELISA) (38). Purified O-antigen from S. Typhimurium and Streptococcal Group A Carbohydrate conjugated to human serum albumin (HSA) were used for ELISA plate coating at 5 and 1 µg/ml, respectively, in carbonate buffer pH 9.6; purified Vi at 1 µg/ml in phosphate buffer pH 7.0; purified MenA and MenC capsular PS were used at 5 µg/ml in PBS pH 8.2; and Hib PS conjugated to HSA was used at 2 µg/ml in PBS pH 7.4. ELISA units were expressed relative to a mouse antigen-specific antibody standard serum curve, with the best five-parameter fit determined by a modified Hill plot. One ELISA unit is defined as the reciprocal of the dilution of the standard serum that gives an absorbance value equal to 1 in this assay. Each mouse serum was run in triplicate.

Serum bactericidal antibody (SBA) against meningococcal (MenA, MenC, and MenB) strains was tested using baby rabbit complement as previously described (29, 39). F8238 MenA, C11 MenC, NZ98/254, M08-0240104, and M01-0240320 MenB strains were used. Pooled sera from each group were tested by SBA.

Datasets were analyzed using two-tailed non-parametric Mann–Whitney test (for comparing the same time point for two different groups) or one-tailed non-parametric Wilcoxon matched-pairs signed rank test (for comparing different time points for a same group) with Prism (GraphPad Software). p-Values less than 0.05 were considered statistically significant.

PS with different structural features and size were conjugated to GMMA from different pathogens. MenA, MenC, and Hib oligosaccharides were terminally activated with SIDEA linker (29) and randomly conjugated to lysines of GMMA surface proteins ( Figure 1A ). Alternatively, the oligosaccharides were terminally derivatized with ADH and linked to LOS on oxidized GMMA by reductive amination ( Figure 1B ).

A similar approach was used for linkage of streptococcal GAC and S. Typhi Vi PS to GMMA. By playing with conjugation conditions, in particular by using different saccharide-to-protein molar ratios (as for meningococcal oligosaccharides) or different buffer pH (as for Vi), it was easy to modulate the number of sugar chains per GMMA particle ( Table 1 ). Formation of saccharide–GMMA conjugates was verified by Western blotting analysis ( Figure 2 ); and the amount of total saccharide and total protein were quantified by HPAEC-PAD and micro BCA, respectively. The saccharide-to-protein w/w ratio, coupled with an estimate of the number of GMMA particles per ml measured by NTA, allowed us to count the average number of saccharide chains per GMMA particle ( Table 1 ).

The impact that sugar length and linkage site on GMMA could have on the immune response induced by the conjugates was initially studied with MenA oligosaccharides linked to MenB GMMA. MenA oligosaccharides of different and non-overlapping length (polymerization degree (DP) equal to 5–12, 16–26, and >36) were conjugated to proteins or LOS on MenB GMMA ( Table 1 , constructs 1–6) and tested in mice. GMMA alone or physically mixed to MenA oligosaccharides were used as negative controls, while MenA–CRM₁₉₇ conjugate was the positive control. Conjugation to proteins or LOS on GMMA resulted in induction of a strong anti-MenA IgG response at a level comparable with that of MenA–CRM₁₉₇. We found that the sugar length did not influence MenA-specific serum IgG response, because no difference in antibody production was observed after immunization with the different MenA–GMMA conjugates ( Figure 3A ), regardless of whether conjugation was directed to LOS or proteins. Interestingly, the conjugates generated from saccharides attached to proteins invariably elicited a higher MenA-specific IgG response 2 weeks after second immunization compared with MenA oligosaccharides linked to LOS ( Figure 3A ). However, all GMMA conjugates, independently from the attachment site of meningococcal oligosaccharides to GMMA, induced antibodies with bactericidal activity against a homologous MenA strain ( Figure 3B ).

The ability of MenB GMMA to induce an immune response after attachment of MenA oligosaccharides was verified by testing the bactericidal activity of antibodies induced against three different MenB strains ( Figure 3B ). While bactericidal activity against UK320 and UK104 strains was not impaired by conjugation, the one against the New Zealand strain was negatively impacted. As bactericidal activity against this strain is mainly mediated by the PorA antigen on MenB GMMA (40, 41), we could speculate that the random conjugation of MenA oligosaccharides to proteins on GMMA could impact on PorA structure and conformation. However, the same was true for the glycoconjugates obtained by linkage of oligosaccharides to LOS on MenB GMMA, indicating that probably the saccharide chains masked some protein components shifting the immune response toward themselves.

Next, Hib oligosaccharides were conjugated to MenB GMMA by targeting proteins or LOS (constructs 7–8, Table 1 ), and the Hib-specific serum IgG response was measured in rats, in comparison with Hib oligosaccharides mixed to GMMA and Hib-CRM₁₉₇ conjugate ( Figure 4 ). As observed for MenA conjugates, also in this case, the conjugate obtained by linkage of Hib to proteins induced a stronger anti-Hib IgG response than the conjugate produced by linking oligosaccharides to LOS. Both GMMA conjugates elicited anti-Hib PS IgG titers significantly higher than Hib simply mixed to GMMA and comparable with Hib-CRM₁₉₇.

After having investigated the impact of saccharide length and attachment via proteins or LOS on GMMA, we interrogated the effect of the density of saccharide conjugated to GMMA particles, which is another important feature for glycoconjugate vaccines. We produced conjugates differing for the average number of MenA or MenC oligosaccharides linked per GMMA particle (constructs 9–12, Table 1 ). No major impact of oligosaccharide density on anti-PS IgG response ( Figure 5A ) and functionality of the sera induced in mice was found ( Figure 5B ). However, control of glycosylation density could be useful to fully preserve the immune response induced by GMMA per se. In fact, by testing the bactericidal activity of the sera against a panel of different MenB strains, we assessed that the largest was the number of meningococcal oligosaccharide chains conjugated per GMMA particle, and the highest was the impact on the functionality of the elicited sera (this was particularly evident for MenB NZ98/254 strain). Therefore, linkage of fewer sugar chains per GMMA particle seems preferable.

To further explore the effect of glycan length and density with a larger-size PS, conjugates formed by S. Typhi Vi PS attached to S. Typhimurium GMMA were generated (constructs 13–16, Table 1 ). Linkage of Vi to LPS on GMMA allowed to introduce a different number of PS chains to GMMA, while conjugation to proteins resulted in few Vi chains per GMMA particle only. As previously verified for MenA and MenC ( Figure 5 ), we did not find impact of antigen density on Vi-specific serum IgG response. However, the saccharide length in this case generated a significant effect, because longer Vi PS (48.5 kDa) were able to induce significantly higher Vi-specific serum IgG titers than the shorter Vi (3.8 kDa) ( Figure 6A ). All conjugates induced anti-S. Typhimurium O-antigen IgG response, preserving immunogenicity of GMMA per se ( Figure 6B ).

The display of PS on GMMA, especially if in high-density modality, generates repetitive epitope moieties on GMMA surface that can facilitate cognate B-cell receptor cross-linking, which could lead to B-cell activation in the absence of T-cell help. This shift toward a strong and fast T-independent B-cell stimulation would not promote germinal center formation and the consequent generation of long-lived plasma cells secreting high-affinity antibodies and memory B cells. Therefore, the T-independent B-cell response can have a negative impact on the efficacy of the humoral immune response, especially in infants or young children, and a detrimental effect on immunological memory and persistence of the antibody response. To investigate any potential T-independent nature of the humoral immune response induced by PS conjugated on GMMA, we evaluated different PS–GMMA conjugates immunizing wild-type and nude mice, devoid of mature T cells. We used MenC, Vi, and GAC PS conjugated to S. Typhimurium GMMA, so as to test PS with different structural features (constructs 12 and 17–19, Table 1 ).

MenC–GMMA conjugate was unable to induce a significant MenC-specific antibody response in nude mice compared with wild-type animals, clearly confirming the need of the T-cell help to promote a humoral response against MenC oligosaccharide ( Figure 7A ). On the contrary, Vi-GMMA induced a strong anti-Vi-specific serum IgG response in wild-type as well as nude mice, revealing that this PS promoted a T-independent humoral immune response ( Figure 7B ). Antibody levels induced in wild-type mice were high after the first dose, with no booster after re-injection. The same was verified by linking Vi PS to LPS or proteins on GMMA. Using GAC–GMMA conjugate, we observed an intermediate situation, since nude mice immunized with GAC–GMMA conjugate generated a GAC-specific serum IgG response, which was significantly lower than that induced in wild-type mice ( Figure 7C ). Thus, GMMA-conjugated GAC PS elicit a weak T-independent saccharide response.