Immunization with outer membrane vesicles displaying conserved surface polysaccharide antigen elicits broadly antimicrobial antibodies

Abstract

Many microbial pathogens produce a β-(1→6)-linked poly-N-acetyl-d-glucosamine (PNAG) surface capsule, including bacterial, fungal, and protozoan cells. Broadly protective immune responses to this single conserved polysaccharide antigen in animals are possible but only when a deacetylated poly-N-acetyl-d-glucosamine (dPNAG; <30% acetate) glycoform is administered as a conjugate to a carrier protein. Unfortunately, conventional methods for natural extraction or chemical synthesis of dPNAG and its subsequent conjugation to protein carriers can be technically demanding and expensive. Here, we describe an alternative strategy for creating broadly protective vaccine candidates that involved coordinating recombinant poly-N-acetyl-d-glucosamine (rPNAG) biosynthesis with outer membrane vesicle (OMV) formation in laboratory strains of Escherichia coli The glycosylated outer membrane vesicles (glycOMVs) released by these engineered bacteria were decorated with the PNAG glycopolymer and induced high titers of PNAG-specific IgG antibodies after immunization in mice. When a Staphylococcus aureus enzyme responsible for PNAG deacetylation was additionally expressed in these cells, glycOMVs were generated that elicited antibodies to both highly acetylated PNAG (∼95-100% acetate) and a chemically deacetylated dPNAG derivative (∼15% acetate). These antibodies mediated efficient in vitro killing of two distinct PNAG-positive bacterial species, namely S. aureus and Francisella tularensis subsp. holarctica, and mice immunized with PNAG-containing glycOMVs developed protective immunity against these unrelated pathogens. Collectively, our results reveal the potential of glycOMVs for targeting this conserved polysaccharide antigen and engendering protective immunity against the broad range of pathogens that produce surface PNAG.

Keywords: glycoconjugate vaccine; immunization; infectious disease; oligosaccharide; synthetic biology.

Fig. 1.

Biosynthesis of glycOMVs bearing PNAG surface polysaccharide. Plasmid-based overexpression of the E. coli pga operon in a hypervesiculating strain of E. coli results in large quantities of rPNAG glycopolymer on the cell surface and on the corresponding OMVs shed by these cells. The native pgaABCD gene cluster encodes the cell envelope-spanning Pga machinery that coordinates the biosynthesis and secretion of PNAG (Inset). PgaC and PgaD are cytoplasmic membrane proteins that mediate PNAG polymerization, with the processive β-glycosyltransferase PgaC responsible for assembling chains of GlcNAc (blue circles) from activated UDP-GlcNAc precursor. The outer membrane porin PgaA translocates growing PNAG chains to the cell surface, while the outer membrane lipoprotein PgaB deacetylates PNAG during export, thereby introducing a limited amount of glucosamine into the polymer (white circles). Removal of additional N-acetyl groups is accomplished by heterologous expression of the S. aureus deacetylase IcaB in the periplasm, resulting in the formation of the rdPNAG glycoform. It should be noted that the occurrence of flagella, included in the drawing here, can vary from strain to strain; flagella are not present on JC8031 cells used in our studies but are likely to be found on other hypervesiculating strains.

Fig. 2.

Overexpression of pgaABCD yields PNAG-containing OMVs. (A) Dot blot analysis of OMV fractions derived from JC8031, JC8031 ΔpgaC, or JC8031 carrying plasmid pUCP18Tc-pga encoding the entire pgaABCD gene cluster (Upper). The same fractions from either JC8031 or JC8031 carrying pUCP18Tc-pga were treated with dispersin B (+) and compared with untreated (−) fractions (Lower). All OMV fractions were diluted 1:10 in PBS (where 1:1 dilution is 2 mg/mL of total protein), spotted on nitrocellulose membranes, and probed with mAb F598. (B) SDS/PAGE and Western blot analysis of OMV fractions derived from JC8031 or JC8031 carrying pUCP18Tc-pga. Membrane was probed with mAb F598. Molecular mass (MW) ladder is indicated on the left. (C) TEM analysis of OMVs produced by JC8031 (Top), JC8031 ΔpgaC (Middle), or JC8031 carrying pUCP18Tc-pga (Bottom). The lipid bilayers of OMVs were visualized by staining with uranyl acetate. The total protein concentration of OMVs used in this analysis was 100 µg/mL. (Scale bar: 200 nm.) (D) IEM analysis of OMVs derived from JC8031 carrying pUCP18Tc-pga. OMVs were immunostained with mAb F598 and subsequently visualized with gold-conjugated anti-human secondary. The total protein concentration of OMVs used in this analysis was 10 µg/mL. Simultaneous visualization of both the membrane and gold particles was not possible, and therefore, OMVs appear as a faint shadow on the transmission electron micrograph. (Scale bars: 100 nm.)

Fig. 3.

Immunization with glycOMVs yields rPNAG-specific IgGs. ELISA-determined IgG antibody titers to rPNAG in sera of immunized BALB/c mice at 6 wk (gray bars) and 8 wk (white bars) after the first immunization (t = 0; black bars). Groups of mice (n = 8) were immunized as follows: (A) PBS, unconjugated PNAG derived from E. coli Top10 cells carrying pUCP18Tc-pga (rPNAG), OMVs from plasmid-free JC8031 ΔpgaC cells (empty OMVs), OMVs from JC8031 cells carrying pUCP18Tc-pga (rPNAG-glycOMVs), or OMVs from JC8031 carrying pUCP18Tc-pga and pTrc-IcaB (rdPNAG-glycOMVs); and (B) PBS and 5GlcNH₂-TT conjugate. Mice immunized with OMVs were injected s.c. at t = 0 and boosted at 3 and 6 wk with 10 µg total protein as measured by total protein content in OMV fraction. Mice immunized with 5GlcNH₂-TT were injected s.c. at t = 0 and boosted at 3 and 6 wk with 10 µg purified conjugate, which was administered with incomplete Freund’s adjuvant. Purified rPNAG isolated from E. coli Top10 cells carrying pUCP18Tc-pga was used as immobilized antigen. Bars represent means, and error bars indicate SD of mean IgG titers. One-way ANOVA allowed for rejection of the null hypothesis that all immunization groups have the same mean IgG titer (P < 0.001). *Statistical significance (P < 0.05) by Tukey–Kramer honest significance difference (HSD); **Statistical significance (P < 0.01) by Tukey–Kramer HSD.

Fig. 4.

glycOMVs elicit opsonically active antibodies and protective immune responses against S. aureus. (A) Opsonic killing activity against S. aureus CP8 strain MN8 by different dilutions of serum from BALB/c mice (n = 8 except for 5GlcNH₂-TT, where n = 4) immunized with PBS, empty OMVs, rdPNAG-glycOMVs, or 5GlcNH₂-TT. Bars represent means, and error bars indicate SD of the mean. The mAb F598 served as positive control and was tested at 10 μg/mL (black bars) and 5 μg/mL (gray bars). The mAb F429, which is specific for the P. aeruginosa alginate antigen, was tested at 10 μg/mL and served as a negative control to which all data were normalized. Negative control serum from mice receiving PBS was only tested at a dilution of 1:30. All of the sera were tested in the absence of either human PMNs (no PMNs) or human complement (no C′) or in the presence of heat-inactivated complement (HIC′), and the average of this entire dataset is shown. (B) Kaplan–Meier survival analysis of groups of BALB/c mice (n = 16 except for 5GlcNH₂-TT, where n = 12, and both detoxified OMV groups, where n = 8) that were immunized with PBS, empty OMVs, rPNAG-glycOMVs, rdPNAG-glycOMVs, detoxified rdPNAG-glycOMVs, or 5GlcNH₂-TT and subsequently challenged at 56 d after the first immunization with 1 × 10⁹ cfu/mL (10 × mLD₅₀) of S. aureus strain Rosenbach in 200 μL sterile PBS via tail vein injection. Mice immunized with OMVs were injected s.c. at t = 0 and boosted at 3 and 6 wk with 10 µg total protein as measured by total protein content in the OMV fraction. Mice immunized with 5GlcNH₂-TT were injected s.c. at t = 0 and boosted at 3 and 6 wk with 10 µg purified conjugate, which was administered with incomplete Freund’s adjuvant. The groups receiving rdPNAG-glycOMVs and detoxified rdPNAG-glycOMVs showed statistically significant protection over the PBS group (P < 0.05) as determined by a log rank test.

Fig. 5.

GlycOMVs elicit bactericidal antibodies and protective immunity against F. tularensis LVS. (A) Serum bactericidal activity against F. tularensis LVS by antibodies in the serum of BALB/c mice (n = 8 except for 5GlcNH₂-TT, where n = 4) immunized with PBS, empty OMVs, rdPNAG-glycOMVs, or 5GlcNH₂-TT. Survival data are derived from standard SBA, where dilutions of serum from immunized mice were tested against F. tularensis LVS Iowa in the presence of baby rabbit complement. All data were normalized to the killing measured for baby rabbit complement alone. Bars represent means, and error bars indicate SD of the mean. The mAb F598 served as positive control and was assayed according to the same dilution scheme, with 1:1 being equivalent to 1 mg/mL. Negative controls included mAb F598 in the absence of human complement (no C′) or in the presence of heat-inactivated complement (HIC′). (B) Kaplan–Meier survival analysis of groups of BALB/c mice (n = 5) that were immunized with PBS, detoxified empty OMVs, detoxified rdPNAG-glycOMVs, or 5GlcNH₂-TT and subsequently challenged at 70 d after the first immunization with F. tularensis LVS Iowa via i.p. injection of 500 cfu (∼500 × LD₅₀). Mice immunized with OMVs were injected s.c. at t = 0 and boosted at 2 and 4 wk with 10 µg total protein as measured by total protein content in OMV fraction. Mice immunized with 5GlcNH₂-TT were injected s.c. at t = 0 and boosted at 2 and 4 wk with 10 µg purified conjugate, which was administered with incomplete Freund’s adjuvant. The groups receiving detoxified rdPNAG-glycOMVs showed statistically significant protection over the PBS and detoxified empty OMV groups (P < 0.05) as determined by a log rank test. The 5GlcNH₂-TT group was not significantly higher than the PBS group or significantly lower than the detoxified rdPNAG-glycOMVs group.