Sequential Linkage of Carbohydrate Antigens to Mimic Capsular Polysaccharides: Toward Semisynthetic Glycoconjugate Vaccine Candidates against Streptococcus pneumoniae Serotype 14.

Vaccines based on isolated polysaccharides successfully protect humans from bacterial pathogens such as Streptococcus pneumoniae. Because polysaccharide production and isolation can be technically challenging, glycoconjugates containing synthetic antigens are an attractive alternative. Typically, the shortest possible oligosaccharide antigen is preferable as syntheses of longer structures are more difficult and time-consuming. Combining several protective epitopes or polysaccharide repeating units as blocks by bonds other than glycosidic linkages would greatly reduce the synthetic effort if the immunological response to the polysaccharide could be retained. To explore this concept, we bridged the well-understood and immunologically potent RU of S. pneumoniae serotype 14 (ST14) with an aliphatic spacer and conjugated it to the carrier protein CRM197. Mice immunized with the spacer-bridged glycan conjugates produced high levels of specific antibodies after just one or two vaccine doses, while the tetrasaccharide repeating unit alone required three doses. The antibodies recognized specifically ST14 CPS, while no significant antibody levels were raised against the spacer or unrelated CPS. Synthetic vaccines generated antibodies with opsonic activity. Mimicking polysaccharides by coupling repeating unit antigens via an aliphatic spacer may prove useful also for the development of other glycoconjugate vaccine candidates, thereby reducing the synthetic complexity while enhancing a faster immune response.

Bacterial capsular polysaccharides (CPS) comprised of repeating units have been identified as major virulence factors of bacterial pathogens.[1] Polysaccharide and glycoconjugate vaccines based on isolated CPS induce a protective immune response in people and prevent millions of deaths every year caused by pathogenic bacteria such as Streptococcus pneumoniae.[2] Almost all currently marketed vaccines rely on CPS isolated from bacterial culture. The isolation and purification of CPS from pathogens in sufficient quantities[3,4] can be challenging as other cellular polysaccharides are frequently found in CPS preparations even though their implications on the human immune system are not known. Certain CPS degrade during isolation or formulation, thus rendering the vaccine ineffective.[5−8] Semisynthetic glycoconjugate vaccines,[9,10] containing a synthetic oligosaccharide antigen resembling the CPS coupled to a carrier protein, have emerged as an attractive option with great potential for understanding glycan immunology and rationally designing efficacious bacterial vaccines.

A key consideration during synthetic vaccine design is antigen length. While CPS are many hundreds or thousands of monosaccharides in length, a majority of CPS repeating units vary fom two to six monosaccharides. Usually, just one or two synthetic repeating units are suffient to induce a protective immune response.[11] To produce specific antibodies against the CPS, B cells need to be activated and differentiated after binding to the glycoconjugate. The process relies primarily on B cell receptors (BCR) to sense the foreign antigen triggering a molecular cascade turning naïve B cells into mature antigen-specific B cells.[12] Numerous studies have shown that multivalent antigens are more efficient than monovalent antigens in promoting an antibody response, mainly because it induces BCR clustering leading to greater signal activation and consequently B cell differentiation.[13−16] Thus, earlier and stronger immune response improvement is crucial mainly when targeting newborns and infants.

Given the challenges associated with the synthesis of oligosaccharide antigens of increasing length, identifying minimal oligosaccharide epitopes is important. It remains to be seen whether an oligosaccharide chimera consisting of oligosaccharide repeating units connected via simple spacers is sufficient to induce a strong immune response associated with the specific response to polysaccharides. To address this question at the conceptual level, we focused on the capsular polysaccharide of S. pneumoniae serotype 14 (ST14). S. pneumoniae are Gram-positive bacteria that cause severe invasive pneumococcal diseases (IPDs) such as pneumonia, septicemia, meningitis, and otitis media.[17−20] ST14 is the most common of the more than 95 serotypes in the human population[21] and accounts for ≤29% of IPDs in children worldwide.[22] The ST14 CPS consists of tetrasaccharide [β-d-Galp-(1→4)-][→6)β-d-GlcpNAc-(1→3)-β-d-Galp-(1→4)-β-d-Glcp-(1→] repeating unit[23] that is the minimal structure required to induce ST14-specific antibodies.[24−26]

To design a polysaccharide mimic consisting of two ST14 tetrasaccharide repeating units (RU) connected by a linker, molecular dynamics (MD) simulations[27,28] were employed to sample the conformational space of an oligosaccharide consisting of three ST14 RUs (Figure B). On the basis of that model, a linker was designed such that the resulting chimeric structure where the middle RU is replaced by the spacer corresponds approximately to that of the native glycan (Figure A). To assess the structural similarity between the synthetic ST14 mimic and the native glycan, the conformational ensembles of the two molecules were compared using all-atom MD simulations.

Figure 1

Molecular dynamcs simulations of the ST14 polysaccharide and ST14 polysaccharide mimic. (A and B) Simulation snapshots of the divalent ST14 mimic and a three-repeat unit ST14 polysaccharide, respectively. Comparison of the distance distributions from 1μs of simulation data for the distance between O2 and O3 corresponding to the effective linker lenght (C) and the radius of gyration (D). A complete set of the distance distributions between highlighted oxigen atoms (O1-O4) are shown in the Supporting Information.

On the basis of this design, just one tetrasaccharide instead of a dodecasaccharide has to be synthesized and is conjugated to another unit (Figure ). The spacer-bridged oligosaccharide derivatives were designed for conjugation to carrier protein CRM197 and in vivo immunological assessment. With proper spacing, the repeating units will interact with B cell receptors and result in a robust immune response.

Figure 2

(A) Structure of the ST14 capsular polysaccharide. (B) Spacer-bridged ST14 oligosaccharide mimic of the capsular polysaccharide.

Results and Discussion

Hybrid Design by Molecular Modeling

The ST14 dodecasaccharide consisting of three ST14 tetrasaccharide RUs was modeled with all-atom MD using the GLYCAMOSMO,14 force field.[27,28] In designing a linker/spacer to bridge the middle tetrasaccharide RU, we adopted a two-stage process. A flexible and immunologically silent linker was envisioned that would carry minimal functional groups but could be installed in one simple synthetic manipulation. Amide bond formation connecting two alkyl linkers was selected for an initial quick assessment using MacroModel 8.0[29] that identified the union of a pentenyl amine and a butanoic acid as a potential solution for the linker challenge at hand. The polysaccharide mimic was simulated with MD, using parameters from the Amber SB99 force field[30,31] and partial charges derived using the R.E.D. tools scripts[32] for the linker residue. On the basis of these simulations, the linker segment introduced in the ST14 mimic is much more flexible than the central tetrasaccharide it replaces, allowing the molecule to sample a greater conformational space, as reflected in the wider probability distributions for the distances shown in panels C and D of Figure (for a more detailed description, see the Supporting Information). The synthetic molecule reaches both more compact and more extended conformations than the comparatively rigid polysaccharide, and a significant overlap between the distributions for the two molecules remains. In addition, the conformational states within one tetrasaccharide repeat unit remain identical between the two molecules. Thus, the two saccharides in the ST14 mimic frequently access conformations that are very similar to those in the polysaccharide.

Synthetic Strategy

The synthesis of ST14 repeating unit oligosaccharides (1–3) with linkers at the reducing and nonreducing ends was based on key ST14 repeating unit tetrasaccharide 4 (Scheme ). Tetrasaccharide antigen 4 containing two linkers was assembled by the linear combination of linker 5, as well as building blocks 6,[33]7, 8,[34] and 9. Tetrasaccharide 10 that lacks the linker at the nonreducing end was prepared the same way using building block 11(35) instead of 9.

Scheme 1

Retrosynthetic Analysis of Capsular Oligosaccharide Derivatives 1–3 and 10

Oligosaccharide Assembly

Glucosamine building block 6, glucosyl 8, and galactosyl 11 were synthesized by following the reported procedures.[33−35] The synthesis of both galactosyl building blocks 7 and 9 commenced with ethyl 4,6-O-benzylidene-1-thio-β-d-galactopyranoside (see Scheme S1).

With four building blocks in hand, the stage was set to access tetrasaccharide 4. The protected amino linker was introduced at the reducing end by glycosylation of 5 and thioglycoside 6 promoted by N-iodosuccinimide (NIS) in the presence of triflic acid (TfOH), affording 12. Removal of the temporary fluorenylmethyloxycarbonyl (Fmoc) protecting group gave compound 13, which was further reacted with galactose 7, thus yielding the fully protected disaccharide 14. Cleavage of levulinoyl (Lev) ester with hydrazine hydrate yielded 15 as an acceptor that was glycosylated with building block 8 to furnish trisaccharide 16. Cleavage of the Fmoc group followed by coupling with thioglycoside 9 produced tetrasaccharide 4 (Scheme A).

Scheme 2

Synthesis of Tetrasaccharides 4 and 10

Tetrasaccharide 18 was prepared by glycosylation of trisaccharide 17 and galactose building block 11 activated with NIS/TfOH. Removal of Fmoc, benzoyl, and benzyl protecting groups yielded fully deprotected tetrasaccharide 10 (Scheme B).

The methyl ester at the nonreducing end of 4 was cleaved with a sodium hydroxide solution followed by addition of excessive sodium methoxide to remove all benzoyl esters, yielding compound 21. Subsequent hydrogenolysis catalyzed by palladium on carbon produced tetrasaccharide 1 (Scheme A).

Scheme 3

Synthesis of Oligosaccharide Derivatives 1–3

Hydrogenation of tetrasaccharide 4 with palladium on carbon yielded compound 22 that contains a free amino group at the reducing end. Amide bond formation to couple 21 and 22 proceeded well with benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate (PyBOP) and N,N-diisopropylethylamine (DIPEA) as 80% spacer-bridged divalent ST14 antigen 23 was produced (Scheme B). 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate[36] (HATU) also successfully catalyzed the coupling but produced some unidentified byproducts. Removal of all of the ester and ether protecting groups of 23 using the same procedure that was used for compound 4 yielded deprotected divalent ST14 2, containing 2 RU. Spacer-bridged trivalent ST14 3 was prepared by PyBOP-mediated coupling of 22 and 24 to afford 25 followed by global deprotection.

A spacer dummy conjugate BSA-35 was synthesized as a control to detect antibodies against the spacer (Scheme ). The synthesis of 35 commenced with phenyl 2-azido-2-deoxy-4,6-O-benzylidene-1-seleno-α-d-galactopyranoside 27, where the 5-methoxy-5-oxopentyl group was installed at position C3 to give 28 that was treated with NIS in the presence of water followed by 2,2,2-trifluoro-N-phenylacetimidoyl chloride and cesium carbonate to afford imidate 29. Glycosylation of linker 5 with 29 produced 30 before the resident azide was converted into an acetamino group with thioacetic acid in pyridine. Methyl ester 31 was hydrolyzed to yield 32 with a free carboxyl group, while hydrogenation resulted in 33 containing an amino group. Coupling of 32 and 33 with PyBOP furnished 34 that was fully deprotected to give 35. The conjugation of 35 and BSA was achieved using bifunctional p-nitrophenyladipate (Scheme B) and characterized via MALDI-TOF (see Figure S3).

Scheme 4

(A) Synthesis of spacer 35 and (B) Conjugation of 35 to BSA

Human Anti-ST14 CPS Antibodies Bind Synthetic Glycans

Synthetic glycans 1–3 and 10 and native CPS14 were covalently immobilized on glass slides (Figure A). A spacer dummy conjugate BSA-35 was used as a control to detect antibodies against the spacer. After incubation with a human reference serum of patients vaccinated with a CPS-based vaccine, the bound antibodies were detected using fluorescently labeled secondary anti-human antibodies. The human serum contained antibodies that bound all synthetic glycan derivatives 1–3. Spacer 35 was not significantly bound by antibodies present in human sera (Figure B).

Figure 3

Native and synthetic oligosaccharides immobilized on a glycan microarray were incubated with the human reference serum of patients immunized with S. pneumoniae CPS. (A) Microarray in which glycans 1–3 and 10 were immobilized were bound by human IgG antibodies. (B) IgG antibody binding to synthetic glycans. No significant binding was detected against the spacer (BSA-35). A serum dilution of 1:100 was used in the analysis. Abbreviations: MFI, mean fluorescence intensity (mean ± standard deviation); PB, printing buffer; CPS, capsular polysaccharide; CWPS, cell wall polysaccharide; BSA, bovine serum albumin.

Preparation and Characterization of Glycoconjugates

The synthetic ST14-related glycans 1–3 and 10 were conjugated to carrier protein CRM197 using the homobifunctional reagent, adipate 4-nitro phenyl diester,[37] under mild conditions (Figure A). Coupling of 0.1 μmol of oligossaccharide to 1 mg of CRM197 resulted in glycoconjugates containing 10 (1), 11 (2), 7 (3), and 10 (10) glycans per molecule of CRM197. The loading was calculated by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) and confirmed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (see Figure S2).

Figure 4

Antibody levels of mice immunized with glycoconjugates. (A) Preparation of conjugates. (B) Immunization schedule. Mice were immunized on day 0 followed by two boosts on days 14 and 28. Blood was collected at each time point. (C) IgG antibody levels of mice immunized with synthetic glycans were measured using ELISA plates coated with CPS of S. pneumoniae serotype 14. Glycoconjugate CRM197-3 induced significantly higher antibody levels on day 14 and glycoconjugates CRM197-1 and CRM197-2 after the first boost (day 21) when compared to those of tetrasaccharide glycoconjugate CRM197-10, which increased markedly only after the second boost. Overall, CRM197-3 showed significantly higher antibody titer when compared to that of tetrasaccharide CRM197-10. The 1:100 serum dilution was used in the analysis. Abbreviations: A.U., absorbance units; PBS, phosphate-buffered saline. ***p < 0.001. **p < 0.01. *p < 0.05 (mean ± standard deviation).

Evaluation of the Immunogenicity of Glycoconjugates

Mice were immunized with the semisynthetic glycoconjugates to determine the immunogenicity of the synthetic ST14 glycans using aluminum hydroxide that is approved for human use as an adjuvant.[38] Immunizations with the vaccine formulation started on day 0, followed by a boost on days 14 and 28 (Figure B). Mice immunized with synthetic glycans produced specific antibodies against ST14 CPS. Interestingly, the antibody levels of mice immunized with ST14 glycans 3 were significantly higher than those with tetrasaccharide 10 already on day 14 (p < 0.05), while the levels of constructs 1 and 2 were significantly higher (p < 0.001) after the first boost (day 21) when compared to those with the synthetic tetrasaccharide 10 (Figure C). Overall, the trivalent derivative CRM197-3 group showed an antibody titer higher than that of the tetrasaccharide CRM197-10 group (p < 0.001) on days 14 and 35, supporting the concept that the increase in antigen length may lead to better BCR activation, eliciting better antibody production.[13]

Antibodies Raised against Synthetic Glycans Recognize the Specific Epitope of ST14 CPS

To test the specifity of binding of the antibody to ST14 CPS, mice sera from each group were incubated with the native capsular polysaccharide of serotype 2 and 14 and synthetic glycans on a microarray. Antibodies from individual mouse sera from each group recognized the synthetic structures of 1–3 and 10 and ST14 CPS, while no significant cross-reactivity was detected against ST2 CPS or the spacer. Antibodies that were produced following vaccinations with synthetic glycans were found to be specific to ST14 CPS (Figure B).

Figure 5

Glycan microarray incubated with sera of mice immunized with synthetic glycans conjugated to CRM197 and formulated with aluminum hydroxide. (A) Mice produced specific antibodies against native capsular polysaccharide CPS 14, synthetic glycan 10, and oligosaccharide derivatives 1–3. No significant binding to CPS 2 or spacer BSA-35 was detected. (B) The divalent and trivalent derivative glycans showed the highest level of binding throughout the groups. A serum dilution of 1:100 was used in the analysis. Abbreviations: MFI, mean fluorescence intensity; PBS, phosphate-buffered saline.

Antibodies Raised against Synthetic Glycans Show Opsonophagocytic Killling Activity In Vitro

Protective immunity against S. pneumoniae is mainly antibody-mediated.[39] To prove the functional activity of antibodies raised against synthetic glycans, we performed an opsophagocytic killing assay (OPKA) in vitro. Pooled sera of mice immunized with CRM197-10, CRM197-1, CRM197-2, CRM197-3, or PBS with aluminum hydroxide after three doses (Figure A) were incubated with HL-60 cells, baby rabbit complement, and ST14 bacteria. The human anti-pneumococcal reference serum (007sp) was used as a standard reference.[40] The antibodies raised against synthetic antigens showed very similar antibacterial activity of 007sp, while sera from the PBS group showed no antibacterial activity (Figure B). A four-parameter logistic model was applied to establish the serum dilution point at which 50% of the bacteria are killed (opsonic index). The results are in the range of the accepted confidence interval of reference serum,[40] and the opsonic indices of 007sp serum, CRM197-10, CRM197-1, CRM197-2, and CRM197-3 were in a very similar range (Figure C). These results are expected because the tetrasacchairde RU alone has already been proven to elicit antibodies with opsonic activity.[24] The combination of two or three RU with a spacer should not change the antibody activity because the tetrasaccharide RU remains the same. Thus, the sequential linkage of ST14 RU does not affect the opsonic activity of generated antibodies against ST14.

Figure 6

Opsonophagocytic killing activity in vitro of CRM197-10, CRM197-1, CRM197-2, and CRM197-3 vaccine-generated antibodies. (A) Anti-CPS14 IgG titer of mice vaccinated with semisynthetic CRM197-10, CRM197-1, CRM197-2, and CRM197-3 as well as with PBS adjuvanted after three vaccine doses (day 35) used in the assay. Data were analized by analysis of variance (mean ± standard deviation of five immunized mice). (B) Opsonophagocytic activity of polled sera of mice vaccinated with semisynthetic CRM197-10, CRM197-1, CRM197-2, CRM197-3, and PBS against ST14. Sera 007sp was used as the reference standard. Two independent experiments in duplicate were used in the analysis (mean ± standard deviation). Semisynthetic vaccines showed a pattern similar to that of the 007sp reference serum. (C) The opsonic index (OI), where 50% of the bacteria are killed, was based on a four-parameter logistic model of a generated cruve with four points from two independent assays. The 007sp reference serum was used as a standard, and the black dashed line represents the mean OI with a 95% confidence interval (CI) marked as dottted gray lines.[40] The OI values of semisynthetic vaccines are similar and in the range of the 95% CI of reference values. The sequential linkage of synthetic ST14 RU does not affect the opsonic killing acitivity against ST14. **p < 0.01.

Significance

Mono-, di-, and trivalent ST14 derivatives 1–3, respectively, were rapidly synthesized by bridging RUs with an aliphatic 10-carbon spacer. The ST14 oligosaccharide derivatives were conjugated with carrier protein CRM197 to form semisynthetic neo-glycoconjugates. These neoglycoconjugates were formulated with aluminum hydroxide and immunologically evaluated in mice. Both glycan array and ELISA analyses of the immune sera demonstrated that the aliphatic spacer decreased neither antigenicity nor immunogenicity. Interestingly, the ST14-specific antibody responses against derivatives 1–3 were significantly higher after only one or two immunizations when compared to that of the branched tetrasaccharide RU only. Most importantly, the sequential linkage of ST14 repeating units does not impair the opsonic killing activity of generated antibodies.

The strategy of spacing synthetic repeating unit glycan antigens with linkers of the appropriate length to mimic capsular polysaccharides is promising for the design of semisynthetic glycoconjugate vaccine candidates and results in a faster, specific antibody response. Higher antibody levels holding opsonic activity in the earlier immunization phase are important for constraining pneumococcus infection at an early stage.