O-acetylation of typhoid capsular polysaccharide confers polysaccharide rigidity and immunodominance by masking additional epitopes

Abstract

In this work, we explore the effects of O-acetylation on the physical and immunological characteristics of the WHO International Standards of Vi polysaccharide (Vi) from both Citrobacter freundii and Salmonella enterica serovar Typhi. We find that, although structurally identical according to NMR, the two Vi standards have differences with respect to susceptibility to de-O-acetylation and viscosity in water. Vi standards from both species have equivalent mass and O-acetylation-dependent binding to a mouse monoclonal antibody and to anti-Vi polyclonal antisera, including the WHO International Standard for human anti-typhoid capsular Vi PS IgG. This study also confirms that human anti-Vi sera binds to completely de-O-acetylated Vi. Molecular dynamics simulations provide conformational rationales for the known effect of de-O-acetylation both on the viscosity and antigenicity of the Vi, demonstrating that de-O-acetylation has a very marked effect on the conformation and dynamic behavior of the Vi, changing the capsular polysaccharide from a rigid helix into a more flexible coil, as well as enhancing the strong interaction of the polysaccharide with sodium ions. Partial de-O-acetylation of Vi revealed hidden epitopes that were recognized by human and sheep anti-Vi PS immune sera. These findings have significance for the manufacture and evaluation of Vi vaccines.

Keywords: Enteric; Glycoconjugate; Molecular modelling; Nuclear magnetic resonance; Vaccine; Vi polysaccharide.

Fig. 1

Chemical structure of Vi polysaccharide. Vi is a homopolymer of repeating units of α(1 → 4)-N-acetylated-D-galactosaminuronic acid (GalNAcA), O-acetylated at the C-3 position. The color scheme is the same as used in Fig. 7, Fig. 8: ring carbons (), N-acetyl (), O-acetyl (), and carboxyl ().

Fig. 2

Effect of ammonium hydroxide treatment on the O-acetylation level of Vi polysaccharide standards from C. fruendii and S. Typhi. The points from control and de-O-acetylated Vi standard from C. freundii (●) and S. Typhi (□) are the means of values determined in two to four independent assays. Confidence intervals are included on untreated and 1.0 M base-treated. Base treatment was performed at 37 °C for 18 h, and O-acetylation was determined by the Hestrin method.

Fig. 3

Effect of de-O-acetylation of Vi polysaccharide on the binding of anti-Vi IgG in mHSA and capture ELISAs. Post-immunization human 10/126 sera (A) and hyperimmune sheep sera (B) were used in mHSA ELISAs, and monoclonal mouse IgG was used in a capture ELISA (C). Data from control and de-O-acetylated Vi standard from C. freundii (●) and S. Typhi (□) are the mean of values determined in two independent assays (n = 4 data points).

Fig. 4

Effect of de-O-acetylation of Vi polysaccharide from C. freundii (A and B) and S. Typhi (C and D) on binding to human anti-Vi IgG 16/138 (A and C) and mouse monoclonal anti-Vi IgG (B, D) by PLL ELISA. Untreated (●), 0.05 M (△) and 1.5 M (○) ammonium hydroxide-treated Vi PS was co-coated with PLL to the plates. The results are the average of two assays and error bars represent standard deviations.

Fig. 5

Hydrodynamic behavior of Vi and meningococcal polysaccharide standards by SEC/MALS/Viscometry. The hydrodynamic volume (A) and intrinsic viscosity (B) of the PSs were determined during their chromatography on a TSKgel G5000PW_XL column in water (black bar), 20 mM Hepes, pH 7.4 (striped bar) or PBS, pH 7.4 (gray bar). Fifty µg of each PS standard were loaded in each case. Standard deviations of three replicate injections are shown by error bars.

Fig. 6

O- and N-acetylation signals detected by ¹H–¹³C HSQC NMR analysis of Vi polysaccharide solutions. Bi-dimensional NMR spectra of (A) untreated, (B) 0.05 M NH₄OH-treated, and (C) 1.5 M NH₄OH-treated C. freundii Vi solutions. The x- and y-axis contain the NMR resonances arising from ¹H and ¹³C, respectively. The insert in panel A has an expanded region of the acetyl crosspeaks.

Fig. 7

Conformations of the O-acetylated (left) and de-O-acetylated (right) Vi polysaccharide. The box shows representative structures for the dominant conformational families for the MD simulations 6RU: the acetylated strand was in a single helical conformation (b) for the entire simulation, while the de-O-acetylated strand alternated between the bent conformations shown in (c) and (d) and a helical conformation (d). The 6RU Vi PS helix can intercalate 1 to 2 sodium ions (g), whereas the more flexible de-O-acetylated strand intercalated up to 4 ions (h). The corresponding 20RU static models are shown for (a) Vi PS and (f) de-O-acetylated Vi. Atoms from the constituent groups are highlighted as follows: N-acetyl, ; O-acetyl, ; carboxyl, , with sodium ions, and the remaining atoms colored .

Fig. 8

Comparison of the simulation dynamics in O-acetylated (red) and de-O-acetylated (blue) Vi polysaccharide. (a) An example conformation of O-acetylated Vi with the C1-C4 end-to-end distance, r, and the φ and ψ dihedral angles labeled. Atoms are highlighted as follows: N-acetyl, ; O-acetyl, ; carboxyl, . Simulation time series of the φ and ψ dihedral angles show differing conformations for the glycosidic linkage in (b) O-acetylated Vi and (c) de-O-acetylated Vi, with the de-O-acetylated strand rotating more freely (d). The same is true for the chain extension: the r end-to-end distance time series is relatively constant for O-acetylated Vi () and much more variable in de-O-acetylated Vi ().