Development and evaluation of a dual-target glycoconjugate against Shigella sonnei

Abstract

Introduction

Shigellosis morbidity and mortality, combined with the increase in multidrug-resistant infections, make Shigella vaccine development a global imperative. Glycoconjugate vaccines that couple immunogenic O-antigen to protein derived from Shigella may provide broader protection across Shigella species and serogroups. Such an approach also circumvents immunotolerance arising from repeated use of the same carrier. Here we use bioconjugation, exploiting an oligosaccharyltransferase (OST) enzyme to couple O-antigen and carrier protein in vivo, to generate a Shigella-specific dual-target glycoconjugate.

Methods

Glycoconjugates were synthesised in E. coli SDB1 cells producing S. sonnei O-antigen, the OST PglS, and one of two Shigella carrier proteins or the standard bioconjugate carrier, ExoA. Recombinant glycoconjugate was purified using anion exchange chromatography and then used to immunise mice. Antibody responses were measured and compared by ELISA.

Results

When co-produced in E. coli, PglS was able to transfer the cloned S. sonnei O-antigen onto three carrier proteins, modified to accept glycans from the PglS transferase enzyme—ExoA and two immunogenic Shigella-specific outer membrane proteins, EmrK and MdtA. Production of MdtA or ExoA glycoconjugates for immunisation studies utilised successive rounds of anion exchange chromatography, to remove unglycosylated material and obtain highly purified glycosylated proteins. Analysis of murine sera following immunisation revealed an IgG response was raised against both carrier protein and the S. sonnei O-antigen for each glycoconjugate.

Discussion

A novel, conserved Shigella protein can be utilised as an effective carrier for the S. sonnei O-antigen to generate a Shigella-specific glycoconjugate that elicits IgG responses to both carrier protein and S. sonnei O-antigen.

1. Introduction

Glycoconjugate vaccines have been instrumental in preventing some of the most serious and debilitating childhood infections such as pneumonia and meningitis (1–4). Bacterial polysaccharides are regarded as key vaccine antigens but when delivered alone do not engage suitable T-cell responses, resulting in short-lived immunity and poor protection in children (5). Covalent linkage of polysaccharides to a protein carrier enables their presentation via major histocompatibility complex class II to T-cells, leading to T-cell activation and the generation of a robust, long-lasting immune response (5).

All currently licensed glycoconjugate vaccines are manufactured via chemical conjugation; polysaccharide and protein components are produced and purified separately before being chemically covalently linked (6). For decades, this approach has enabled the production of life-saving vaccines but is a costly and complex process requiring handling of pathogenic bacteria and multiple rounds of synthesis, purification, and quality control (6–8). Alternative means of synthesis such as bioconjugation could address these limitations via an in vivo method of glycoconjugate production. Here, glycoconjugate vaccines are produced heterologously in genetically modified strains of bacteria, such as E. coli, using bacterial oligosaccharyltransferase (OST) enzymes to couple polysaccharide and protein (Figure 1) (9–11). The simplicity of production compared to chemical conjugation has the potential to produce vast protein/glycan combinations and could substantially reduce vaccine manufacturing costs (7). This is especially important considering many bacterial infections present the highest burden in low- and middle-income (LMIC) settings. Furthermore, a sub-set of bioconjugates has now advanced into clinical trials, such as S4V-EPA, a quadrivalent vaccine against Shigella infections (12).

Figure 1.

Schematic of bioconjugation for the production of a Shigella sonnei-specific glycoconjugate. Three plasmids are co-expressed in E. coli, encoding a carrier protein, the oligosaccharyltransferase PglS and the S. sonnei O-antigen (SSOAg) (encoded by the Plesiomonas shigelloides O-antigen biosynthesis locus). The SSOAg (D-FucNAc4N-L-AltNAcA) is assembled on the cytoplasmic face of the inner membrane, then flipped into the periplasm, where it is polymerised then conjugated to a ComP-tagged carrier protein (green) by PglS (orange), to produce glycoconjugate. The red and white triangle and pink square are used here to represent D-FucNAc4N and L-AltNAcA, respectively. Schematic made using BioRender.

Shigellosis is a leading cause of diarrhoeal disease, globally. Of the four Shigella species, S. flexneri typically causes the majority of infections in LMICs, particularly in children under five years (13). S. sonnei is more prominent in high-income countries, although increasing global incidence driven in part by the spread of multidrug-resistant strains has led to its dominance in many regions (14). Shigellosis can range from self-limiting diarrhoea to acute bloody diarrhoea with fever and is estimated to cause 68,000 deaths in children under five, annually. For those who survive, many experience life-long consequences of infection (15, 16).

The increase in antimicrobial resistance alongside the morbidity and mortality of shigellosis makes Shigella a priority pathogen for vaccine development, but despite numerous approaches, there remains no licensed candidate (12). This has in part been impeded by the lack of established correlates of protection to guide the design of an efficacious vaccine. It is widely viewed that serum IgG against the Shigella lipopolysaccharide (LPS) is associated with protection from infection, and as such, many vaccine candidates in clinical trials focus on targeting this antigen (17). These candidates include two chemically synthesised glycoconjugates, S. flexneri 2a and/or S. sonnei LPS conjugated to tetanus toxoid (18, 19) and S4V-EPA, a quadrivalent bioconjugate vaccine targeting S. sonnei and S. flexneri serotypes 2a, 3a, or 6, with LPS from each conjugated to Pseudomonas aeruginosa exotoxin A (ExoA) (12). Alternative vaccine designs based on the Shigella LPS include Invaplex, a high molecular mass complex of Invasion protein antigens (Ipa) IpaB and IpaC, with Shigella LPS (20, 21) and alt-Sonflex 1-2-3, a generalised module for membrane antigen (GMMA)-based vaccine (22). The latter uses a combination of outer membrane vesicles isolated from S. sonnei and three S. flexneri serotypes (1b, 2a, and 3a) (22).

Notably, all these vaccine candidates incorporate LPS from multiple Shigella serogroups, which is necessitated by the structural variability of the Shigella O-antigens (the outermost component of the LPS). S. sonnei harbours a single, conserved O-antigen structure, whilst there are multiple serotypes and sub-serotypes of S. flexneri (23). A vaccine targeting S. sonnei plus S. flexneri serotypes 2a, 3a and 6 is estimated to protect against approximately 80% of Shigella infections, based on the prevalence of the different serotypes as well as cross-reactivity between some serotypes of S. flexneri (12, 15, 24, 25).

Combining multiple glycan antigens to broaden serotype coverage is an established strategy in licensed vaccines against Streptococcus pneumoniae and Neisseria meningitidis. However, the need to incorporate numerous antigens does present a challenge in vaccine design and production, increasing costs (8, 26). This could be overcome by using glycoconjugate carrier proteins not only for glycan presentation but as conserved, serotype-independent immunogens, offering a means of expanding vaccine valency and dual stimulation of the immune response (27, 28). Furthermore, most existing glycoconjugate vaccines rely on a small pool of well-characterised carrier proteins. Overexposure to the same protein antigens risks the development of carrier-induced epitope suppression and blunting of the immune response; therefore, novel carriers could circumvent this (28, 29).

Bioconjugation complements the design of such “double-hit” glycoconjugates, as the site of glycosylation can be precisely controlled through positioning of the OST recognition sequon. This could help avoid masking important protein epitopes with the attached glycan (28, 30, 31). Although not yet widely explored, double-hit bioconjugates have already been investigated for S. pneumoniae, Group A Streptococcus and Klebsiella pneumoniae (32–34).

In this study, we cloned the complete S. sonnei O-antigen pathway in E. coli, and used two conserved, immunogenic Shigella proteins, EmrK and MdtA, as novel carriers for Shigella glycoconjugates. Both proteins, alongside the established bioconjugate carrier ExoA, could be conjugated to the S. sonnei O-antigen using the OST PglS. Evaluation of S. sonnei-specific glycoconjugates containing either MdtA or ExoA found both conjugates elicited a specific anti-protein and anti-O-antigen IgG response in a mouse infection model.

2. Materials and methods

2.1. Bacterial strains and plasmids

A list of bacterial strains used in this study can be found in Supplementary Table 1. Escherichia coli was routinely cultured using Luria–Bertani (LB) (Miller) agar or broth at 30°C, with shaking at 180 rpm for liquid cultures. Both LB and Terrific broth were used for large-scale glycoconjugate production, as detailed in Section 2.6. Where necessary, media was supplemented with the following antibiotics for plasmid maintenance: ampicillin (100 µg·mL⁻¹), tetracycline (20 µg·mL⁻¹), chloramphenicol (30 µg·mL⁻¹), and kanamycin (50 µg·mL⁻¹).

2.2. Construction of the Plesiomonas shigelloides O17 O-antigen plasmid

A list of plasmids and oligonucleotides used in this study can be found in Supplementary Tables 2, 3, respectively. The sequence of all plasmids was confirmed using long-read sequencing (Plasmidsaurus).

Cloning and synthesis of the S. sonnei O-antigen (SSOAg) used gDNA from Plesiomonas shigelloides O17 (NCTC10360), as the two harbour identical O-antigen structures (35). A 12.2 kb region of the P. shigelloides O17 O-antigen, from wzz to wbgZ, was amplified using Q5 polymerase with primers PsOFApal and PsORApal. Primers included ApaI restriction sites which were used for insertion of the O-antigen fragment into pBBR1MCS-3 using restriction ligation cloning, to generate pBPSO.

2.3. Construction of carrier protein and pglS expression plasmids

Carrier proteins used for glycoconjugate production were composed of an N-terminal DsbA sequence for trafficking to the periplasm, protein coding sequence, C-terminal ComP tag (amino acids 29 to 145) for glycosylation with the PglS OST (36) and a final C-terminal 6XHistag for protein purification (Supplementary Figure 1). The detoxified, C-terminal fragment of ExoA used in this study has been described elsewhere (37, 38). The sequence of ExoA used to construct pEC415_ExoAComP also contains two inserted glycosylation sequons recognised by the PglB OST (37). These were inactivated in this study by replacing the glycosylated asparagine residues with glutamine (39). The amino acid sequences of EmrK and MdtA were derived from S. sonnei and modified to remove the N-terminal native signal peptide, resulting in expression of amino acids 38 to 387 for EmrK and 66 to 453 for MdtA.

Carrier proteins were initially designed with two ComP tags, flanking the protein coding sequence at both the N- and C-termini. Protein constructs were ordered as gBlocks with additional 5’ and 3’ overhangs for insertion into pEC415 via Gibson assembly, and the pEC415 backbone was amplified using primers pEC415DsbA_f and r. gBlocks did not include the DsbA signal sequence as this was already present within the pEC415 plasmid backbone, located nine base pairs downstream of the ribosome binding site. Due to absent or very low protein expression from these constructs (data not shown), the N-terminal ComP tag was removed by inverse PCR; plasmids were then ligated using the KLD Enzyme Mix (NEB). Primers for inverse PCR were as follows: ExoA_NComPdel_f and r for pEC415_ExoAComP, EmrK_NComPdel_f and r for pEC415_EmrKComP and MdtA_NComPdel_f and r for pEC415_MdtAComP. ExoA and MdtA with no ComP tag(s) were ordered as gBlocks and inserted using Gibson assembly into pEXT20, amplified using primers pEXT20_noComP_f and r (Supplementary Figure 1).

To generate pEXT22_pglS, the nucleotide sequence of Acinetobacter baylyi pglS was ordered as a gBlock. A minor number of nucleotide changes were required to permit gBlock synthesis, but these did not alter the amino acid sequence. EcoRI and BamHI restriction sites were included at the 5’ and 3’ ends of the sequence, respectively, for insertion into pEXT22 using restriction ligation cloning.

2.4. Chromosomal deletion of mdtA from E. coli W3110

mdtA was deleted from the E. coli W3110 chromosome using lambda red recombineering (40). The deletion cassette was composed of two 120 bp homology arms, representing the DNA sequences immediately upstream and downstream of mdtA. Within the homology arms was a chloramphenicol cassette flanked by two dif recombinase sites (41). This construct was ordered as a gBlock and inserted using Gibson assembly into pKD4, amplified using primers pKD4MdtA_f and r. Gibson reactions were dialysed into dH₂O using 0.45 µm membrane filters before transformation into electrocompetent TransforMax E. coli EC100D (pir+) (Biosearch Technologies). Transformants were selected on LB agar with chloramphenicol and ampicillin, and extracted plasmids were confirmed with long-read sequencing, to produce pKD4_mdtA.

pKD4_mdtA was used as the template for amplification of the deletion cassette, from the 5’ end of homology arm one to the 3’ end of homology arm two, using primers MdtAcassette_f and r. Purified amplicon was used to transform electrocompetent E. coli W3110 harbouring the lambda red recombineering plasmid, pKD46. Transformants were selected on LB agar with chloramphenicol, and colonies were screened for cassette integration by PCR, using primers MdtAscreen_f and r.

PCR-confirmed, chloramphenicol-resistant integrants were passaged on LB agar only at 37°C, to facilitate loss of the temperature-sensitive pKD46 and removal of the chloramphenicol cassette, using the endogenous E. coli recombination system, Xer, which recognises the dif recombinase site (41). Chloramphenicol- and ampicillin-sensitive colonies were screened for loss of the chloramphenicol cassette by PCR, using the same screening primers as above, and confirmed using long-read sequencing.

2.5. Small scale glycoconjugate production and purification

To screen for SSOAg production, strains were grown overnight in 5 mL LB, shaking at 37°C then harvested by centrifugation at 17,000×g for 10 min at 4°C. To screen for glycoconjugate production, a starter culture was grown for 16 h–18 h and used to inoculate 40 mL LB at a 1:100 dilution, then incubated shaking at 30°C. When OD₅₉₅ reached 0.5 to 0.7, 0.5 mM IPTG was added to induce OST expression. Following overnight incubation, cultures were induced with 0.2% L-arabinose for carrier protein expression then incubated at 25°C for a further 8 h. Cells were harvested by centrifugation at 17,000×g for 10 min at 4°C.

For protein extraction and purification, cell pellets from each 40 mL culture were resuspended in 2 mL His binding buffer (50 mM Tris–HCl, 300 mM NaCl, pH 8.0) and added to lysing matrix B (MP Biomedicals) for lysis on a FastPrep homogeniser (MP Biomedicals). Lysates were centrifuged twice at 17,000×g for 5 min to remove cell debris; then the supernatant was incubated with 2% Triton X-114 at 4°C, rotating, for 1 h or overnight. Triton X-114 was removed by heating at 37°C for 10 min then centrifugation at 17,000×g for 10 min. The upper phase was isolated and 6XHis-tagged proteins were purified using nickel affinity chromatography and stored in His elution buffer (50 mM Tris–HCl, 300 mM NaCl, 300 mM imidazole, pH 8.0).

2.6. Glycoconjugate production and purification for immunogenicity studies

Glycoconjugate and unglycosylated protein to be used in immunogenicity studies were produced in E. coli strain SDB1. SSOAg-ExoAComP glycoconjugate was produced using pBPSO (SSOAg), pEXT22_pglS (PglS) and pCH4 (ExoAComP) (38). The SSOAg-MdtAComP glycoconjugate was produced using pBPSO, pACT3_pglS (PglS) (36) and pEC415_MdtAComP (MdtAComP). Unglycosylated protein was produced using pCH4, pEC415_MdtAComP (proteins with ComP), pEXT20_ExoA or pEXT20_MdtA only (proteins without ComP).

2.6.1. Bacterial cell growth

All cultures were grown in batches of 2 L media in a 5 L shake flask. For production of SSOAg-ExoAComP glycoconjugate and all unglycosylated proteins, LB starter cultures were grown at 30°C for 16 h–18 h and used to inoculate 2 L LB at a 1:100 dilution. Cultures were incubated at 30°C, 150 rpm and induced with 0.5 mM IPTG at OD₅₉₅ 0.5–0.7, then incubated overnight. For the SSOAg-MdtAComP glycoconjugate, starter cultures in Terrific broth (Supplementary Table 4) were diluted 1:50 into 2 L Terrific broth. At OD₅₉₅ 0.6–0.8, cultures were induced with 0.5 mM IPTG, and the temperature was reduced from 30°C to 25°C. After 2 h, 0.2% L-arabinose was added; then cultures were incubated overnight. Cells were harvested at 7,500×g for 30 min at 4°C, and cell pellets were frozen at −20°C.

2.6.2. Protein extraction

For periplasmic extraction of SSOAg-ExoAComP, each 2 L cell pellet was resuspended in approximately 160 mL periplasmic extraction buffer (30 mM Tris–HCl, 20% sucrose, 1 mM EDTA, 1 mg/mL lysozyme, pH 8.0) and incubated at 4°C for 1 h on a rolling platform. Lysates were incubated for a further 30 min at room temperature with 10 units/mL of Benzonase nuclease and 2 mM MgCl₂ then centrifuged for 1 h, 31,000×g, 4°C. Supernatant was passed through a 0.2 µm filter then dialysed into His binding buffer overnight at 4°C using a Slide-A-Lyzer dialysis cassette (Thermo Fisher Scientific).

SSOAg-MdtAComP glycoconjugate and all unglycosylated proteins were extracted using whole-cell lysis. Cell pellets from 2 L cultures were resuspended in 100 mL–200 mL His binding buffer and lysed using a pre-chilled pressure cell homogeniser. Lysates were then incubated at room temperature on a rocking platform for 30 min with 1 mg/mL lysozyme and 10 units/mL of Benzonase nuclease then centrifuged for 1 h, 31,000×g, 4°C.

To remove lipid and lipid-linked glycan from the cell extracts, 2% Triton X-114 was added to chilled, filtered lysates and incubated at 4°C for 1 h or overnight on a rolling platform. Triton X-114 was removed by incubating at 37°C for 10 min then centrifugation at 3,900×g. The upper phase was isolated and centrifuged again at 3,900×g for 5 min; then the upper phase was saved for protein purification.

2.6.3. Nickel affinity chromatography

Nickel affinity chromatography was performed using Ni-NTA resin in 10 mL polypropylene columns (Pierce). Resin was prepared by washing with 10 column volumes (CV) of dH₂O, 5 CV of His elution buffer and 10 CV of His binding buffer. Prepared resin was incubated with Triton X-114-treated lysates at 4°C for 1 h–2 h on a rolling platform before loading onto the column. The flowthrough fraction was captured and reserved; then resin was washed three times with 10 CV His wash buffer (50 mM Tris–HCl, 300 mM NaCl, 10 mM imidazole, pH 8.0). Resin was incubated for 2 min with 1 CV His elution buffer before elution; this was repeated a total of five times. For the SSOAg-ExoAComP glycoconjugate and ExoA and MdtA only, the reserved flowthrough fraction was passed over the re-equilibrated resin again, and the method was repeated one to two more times. Nickel affinity-purified samples were stored either at 4°C or with 20% glycerol at −80°C.

2.6.4. Anion exchange chromatography

Nickel affinity-purified samples were buffer exchanged into AEX start buffer using PD-10 desalting columns (Cytiva) or overnight dialysis using Slide-A-Lyzer dialysis cassettes. For the AEX start buffers, BisTris pH 6.0 was used for purification of SSOAg-ExoAComP glycoconjugate and unglycosylated ExoAComP or ExoA and Tris pH 8.6 for SSOAg-MdtAComP glycoconjugate and unglycosylated MdtAComP or MdtA. AEX was performed on an AKTA Go Protein Purification System (Cytiva), using a 1 mL Resource Q column (Cytiva). Glycoconjugate and unglycosylated protein were eluted from the column using AEX start buffer plus a NaCl gradient up to 500 mM. Fractions of interest, as confirmed by SDS-PAGE, Imperial stain, and western blotting, were pooled and transferred back into AEX start buffer using overnight dialysis, PD-10 desalting columns or dilution into start buffer. AEX was repeated three to four times for glycoconjugate purification and one to three times for protein only. Final fractions not undergoing further purification were transferred into endotoxin-free, sterile phosphate-buffered saline, pH 7.4, using PD-10 columns and quantified using a Qubit 3 then lyophilised.

2.7. SDS-PAGE and western blotting

Glycoconjugate and protein samples were mixed with 1 × LDS loading buffer (Thermo Fisher Scientific) and 1 mM–10 mM DTT then heated for 10 min at 70°C or 5 min at 95°C. 20 µL–40 µL was loaded onto a 4%–12% BisTris gel and run in MOPs buffer. For glycan-only samples, 5 mL cell pellets were resuspended in PBS to OD 1.5; then 500 µL removed and boiled for 10 min before mixing with LDS loading buffer as above. Gels were transferred to nitrocellulose membrane using an iBlot3 dry blotting system. Following transfer, SDS-PAGE gels were washed with dH₂O for 10 min, incubated with Imperial stain for 1 h then destained using multiple washes in dH₂O. Imperial-stained gels were imaged on a Licor Odyssey M. The same method was used for Imperial staining of those SDS-PAGE gels not also being used for western blotting.

Membranes were washed 3 × 5 min with PBS-Tween 0.1% then incubated for 45 min to 1 h with primary antibodies—mouse anti-S. sonnei monoclonal antibody (Abnova) at 1:2,000 and rabbit anti-6×Histag polyclonal antibody (Invitrogen) at 1:5,000 dilution. Membranes were washed as described then incubated for 30 min–45 min with secondary antibodies—IRDye680RD goat anti-mouse IgG and IRDye800CW Goat anti-rabbit IgG (LI-COR). Membranes were washed for a final time before imaging using the LI-COR Odyssey M.

For densitometry analysis of glycoconjugate production, western blot images were converted to grey scale and exported into ImageJ (v1.54g). A defined rectangle of the same size was placed around the glycoconjugate band(s) for each protein, and mean grey value was measured. An empty control lane was used to determine the background, and this was subtracted from each glycoconjugate reading. Statistical significance was determined from three biological replicates using Welch’s ANOVA with Dunnett’s T3 multiple comparisons test.

2.8. Animal study

All animal experiments were performed in accordance with UK Home Office regulations under project license PP2473276 following approval from the local University of Cambridge Animal Welfare and Ethical Review Body.

Six-week-old C57BL6J female mice were purchased (Charles River). Mice were housed in GM500 Individually Ventilated Cages—Biocontainment Systems (Tecniplast) and maintained on individual air handling units kept at between 19 °C and 23 °C and 45%–65% humidity; each cage received 60 air volume changes per hour. Mice were given access to food and water ad libitum and maintained on a 12-hour light/dark cycle. Aspen wood chip was used as the base bedding with nestlets and “fun tunnels” for environmental enrichment (Datesand). Mice were housed in groups of no more than five adults per cage and acclimatised for one week prior to immunisation. Welfare assessments were carried out daily as part of the morning check. This involved moving the cage forward on the rack runners and observing the mice in their micro-environment. Any abnormal signs of behaviour or physical clinical signs of concern were reported. “In hand” weekly health assessments were also performed, and every cage was removed from the rack, and each mouse given a full “head to toe” health check. All persons performing welfare checks on animals were trained and assessed as competent by qualified individuals.

Mice (n = 7 per group) were immunised subcutaneously in the flank with either glycoconjugates (SSOAg-ExoAComP or SSOAg-MdtAComP), carrier protein only (ExoAComP or MdtAComP) or PBS only. All glycoconjugates and carrier proteins were tested individually, not in combination. Mice were immunised with 10 µg of protein per dose in PBS and adjuvant Alhydrogel (1:1) (InvivoGen) on days 0, 14, and 28. On day 35, mice were terminally anaesthetised with a constant supply of 5% isoflurane in oxygen. Anaesthesia was induced by placing mice in an induction box and then transferring them to a nose cone to maintain a deep state of anaesthesia which was confirmed by absence of a toe-pinch withdrawal reflex. A terminal blood sample was collected by cardiac puncture using a 25-gauge needle and a 1 mL syringe. After removal of sufficient blood and exsanguination to ensure death, mice were immediately confirmed dead by cervical dislocation. Blood was collected into sterile 1.5 mL microcentrifuge tubes and left to clot for 30 min. The serum was removed by centrifugation at 21,000×g for 5 min and collected into a fresh tube and stored at −20°C until used in assays.

2.9. Anti-protein ELISAs

ELISAs were used to measure murine IgG responses to ExoA and MdtA. 96-well Maxisorp plates were coated with 0.2 µg/well of ExoA or MdtA, or PBS only. Following overnight, static, incubation at 4°C, plates were washed four times with PBS-Tween 0.05%, then 100 µL/well of blocking buffer (PBS-Tween 0.05%, 5% skimmed milk) was added, and plates were incubated at 500 rpm at room temperature for 1 h. Plates were washed three times with PBS-Tween 0.05%, then 100 µL/well serum was added, and plates were incubated at 500 rpm at room temperature for 1 h. Serum from each mouse was tested individually in technical duplicate and serially diluted in PBS-Tween 0.05% using a 6-point, 3-fold dilution series. Plates were washed four times with PBS-Tween 0.05%, then incubated with 100 µL/well of HRP-conjugated goat anti-mouse IgG at a 1:10,000 dilution at room temperature for 1 h. Plates were washed four times with PBS-Tween 0.05%, then incubated with 100 µL/well TMB substrate (Thermo Fisher Scientific). The reaction was stopped after 8 min using 100 µL/well of 1 M sulphuric acid, then OD₄₅₀ and OD₅₇₀ read using a plate reader.

Background subtractions were made using the OD₅₇₀ and no serum control readings. A coefficient of variation (CV) of <20% was considered acceptable for inter-plate and intra-plate variability. Inter-plate variability was measured using a pool of sera from mice immunised with ExoAComP or MdtAComP only and tested against their respective antigen on each plate. Intra-plate variability was determined using duplicate testing of individual serum samples. OD₄₅₀ values obtained from serial dilution of each serum sample were plotted on a 4-parameter logistic standard curve in GraphPad Prism and used to determine the reciprocal of the serum dilution that gave an OD of 0.5. The reciprocal determined for mice immunised with PBS only was below the lowest serum dilution tested (1/600); therefore, a serum dilution just outside this range (599) was used for visualisation and data analysis. Statistically significant differences in IgG responses against each protein were determined by the Kruskal–Wallis test with Dunn’s multiple comparison test.

For the anti-protein ELISAs measuring the IgG response to the carrier proteins with and without ComP, the above protocol was followed with minor amendments. Plates were coated with 0.5 µg/well of ExoAComP, ExoA, MdtAComP or MdtA, or PBS only, and probed with pooled sera from mice immunised with either ExoAComP or MdtAComP only at a 1:100,000 dilution. Sera were tested in technical triplicate, and OD₄₅₀ was reported after the background was subtracted using values from a no-antigen control.

2.10. Anti-glycan ELISAs

Whole cell ELISAs were used to measure murine IgG responses to the SSOAg. A starter culture of E. coli W3110ΔmdtA harbouring either pBPSO, encoding the SSOAg locus, or empty vector (pBBR1MCS-3) was grown for 16 h–18 h and used to inoculate 20 mL LB at a 1:100 dilution. Cultures were incubated with shaking at 37°C until OD₅₉₅ reached 0.6, then centrifuged at 4,300×g for 10 min. Cell pellets were washed three times in 20 mL Carbonate buffer (15 mM Na₂CO₃ and 35 mM NaHCO₃), OD₅₉₅ was adjusted to 0.6, then diluted one in ten in Carbonate buffer and used to coat 96-well Maxisorp plates at 100 µL/well, which equated to approximately 3–4 × 10⁶ cells per well. Coated plates were air-dried in a microbiological safety cabinet, then fixed with 100 µL/well methanol and air-dried again before being stored for up to 1 week.

For the anti-glycan ELISA, coated plates were washed four times with PBS-Tween 0.05%, then 100 µL/well of blocking buffer (PBS-Tween 0.05%, 10% skimmed milk) was added, and plates were incubated at 500 rpm, room temperature, for 2 h. Plates were washed three times with PBS-Tween 0.05%, then 100 µL/well sera was added, and plates incubated at 500 rpm, room temperature, for 2 h. Pooled sera from each group were serially diluted in PBS-Tween 0.05% + 1% skimmed milk, using a 6-point, 3-fold dilution series. Plates were washed four times with PBS-Tween 0.05%, then incubated with 100 µL/well of HRP-conjugated goat anti-mouse IgG, at 1:10,000 dilution, room temperature for 1 h. Plates were washed four times with PBS-Tween 0.05%, then incubated with 100 µL/well TMB substrate. The reaction was stopped after 13 min using 100 µL/well of 1 M sulphuric acid, then OD₄₅₀ and OD₅₇₀ read using a plate reader.

Pooled sera were tested in technical triplicate, and background subtractions were made using the OD₅₇₀ and no sera control readings. A pool of sera from mice immunised with SSOAg-ExoAComP was tested against wells coated with W3110ΔmdtA + pBPSO on each plate to measure inter-plate variability, for which a CV of <20% was considered acceptable. A CV of <20% was also deemed acceptable for intra-plate variability, measured using triplicate testing of pooled sera samples. OD₄₅₀ values obtained from serial dilution of each serum sample were plotted on a 4-parameter logistic standard curve in GraphPad Prism and used to determine the reciprocal of the serum dilution that gave an OD of 0.2.

3. Results

3.1. Generation of Shigella-specific bioconjugates using PglS

The S. sonnei O-antigen (SSOAg) constitutes a two-component repeat unit of a reducing-end 2-acetamido-4-amino-2,4-dideoxy-D-fucose (D-FucNAc4N) linked to 2-amino-2-deoxy-L-altruronic acid (L-AltNAcA) (23, 42). This structure is distinct from those found in other Shigella and E. coli species and was acquired via horizontal gene transfer from P. shigelloides O17, which harbours an identical O-antigen (35). Using the nucleotide sequence from the P. shigelloides O-antigen biosynthesis locus (pBPSO), we recombinantly produced the SSOAg in E. coli SDB1, as confirmed by western blot using an S. sonnei-specific monoclonal antibody (Figure 2A).

Figure 2.

Glycosylation of Shigella-specific carrier proteins using the oligosaccharyltransferase (OST) PglS. (A) Recombinant production of the Shigella sonnei O-antigen (SSOAg) in E. coli using the O-antigen biosynthesis locus from Plesiomonas shigelloides (pBPSO). Cell lysates from either E. coli carrying pBPSO or empty vector (pBBR1MCS-3), P. shigelloides, or Shigella sonnei only were resolved by SDS-PAGE and detected via western blot using anti-SSOAg antibody (red). (B) ExoA, EmrK and MdtA were modified with a C-terminal ComP tag for glycosylation by the PglS OST and co-expressed in E. coli SDB1 with the S. sonnei O-antigen and PglS, or a no-OST control (pSta313). Glycosylated and unglycosylated carrier proteins were resolved by SDS-PAGE and detected via western blot using anti-6×His-tag antibody (green) and anti-SSOAg antibody (red). (C) Densitometry analysis of S. sonnei glycoconjugate production for the three carrier proteins. ImageJ software was used to quantify glycoconjugate production. Statistical significance was determined from three biological replicates using Welch’s ANOVA with Dunnett’s T3 multiple comparisons test, **p < 0.01, *p < 0.05. SSOAg, S. sonnei O-antigen.

OSTs can exhibit substrate specificity in the glycans they transfer, and the SSOAg harbours an atypical reducing end D-FucNAc4N, so we attempted to conjugate the glycan using OSTs from N-linked and O-linked glycosylation systems, namely C. jejuni PglB and Acinetobacter baylyi PglS, respectively (Figure 1) (11, 36, 38, 43). Western blot analysis using an anti-SSOAg antibody revealed glycoconjugate production in strains producing the SSOAg with PglS and its carrier protein ExoAComP, with no glycoconjugate production observed using PglB and its compatible carrier, ExoA (Supplementary Figure 2). Glycosylated ExoAComP migrated between 110 kDa and 120 kDa, compared to the unglycosylated ExoAComP at approximately 80 KDa. Glycosylated ExoAComP presented as a less defined band compared to the unglycosylated protein, possibly representing different sizes of glycoconjugate. As this carrier protein only harbours a single glycosylation site, this is likely a result of varying polymer lengths of the attached SSOAg. These results confirmed PglS as a viable OST for synthesis of S. sonnei-specific glycoconjugates.

We next investigated whether immunogenic Shigella-specific proteins could be used as novel glycoconjugate carrier proteins. Those selected, EmrK and MdtA, are both components of efflux pumps and conserved across S. sonnei and S. flexneri (44). Both were sero-reactive in a Shigella-specific protein array, with a significant increase in IgM, IgG, and IgA responses to both proteins when comparing acute and early convalescent sera from patients with confirmed Shigella diarrhoea (44). Both proteins were modified with a C-terminal ComP tag (Supplementary Figure 1), and when co-expressed with the SSOAg and PglS, produced S. sonnei-specific glycoconjugate (Figure 2B). Densitometry analysis of glycoconjugate production (Supplementary Figure 3) revealed that the highest glycoconjugate yield was obtained using ExoAComP as the carrier protein. Both ExoAComP and MdtAComP produced significantly more glycoconjugate than EmrKComP, based on total yield from a 40 mL culture (Figure 2C). Therefore, only ExoAComP and MdtAComP were taken forward for immunogenicity studies.

3.2. Purification of glycoconjugates for immunogenicity studies

SSOAg glycoconjugates were purified for immunogenicity studies using a combination of nickel affinity and anion exchange chromatography (AEX). Although an initial nickel affinity purification step was sufficient to capture S. sonnei glycoconjugate from the cell lysate, some contaminating proteins remained, in addition to a large proportion of unglycosylated protein. As glycosylation can alter the isoelectric point of a protein (45), AEX was used to further purify the glycoconjugates as this separates macromolecules based on charge.

Buffers with a pH higher than the predicted pI of each protein were used to confer a negative charge and enable protein binding to the positively charged AEX resin. Using Tris buffer pH 8.6, unglycosylated MdtAComP eluted from the column before the glycoconjugate, but this did not completely separate the two species; therefore, sequential rounds of AEX were employed to remove unglycosylated protein and yield pure glycoconjugate (Figure 3). For the production of SSOAg-ExoAComP, multiple rounds of AEX in BisTris buffer pH 6.0 were used, and the glycoconjugate eluted from the column before the unglycosylated material (Supplementary Figure 4).

Figure 3.

Sequential anion exchange chromatography (AEX) for purification of S. sonnei-specific glycoconjugates. Nickel affinity purified MdtAComP conjugated to the S. sonnei O-antigen (SSOAg) was further purified via AEX for separation of glycosylated and unglycosylated protein using an increasing NaCl gradient. Elution fractions from each round of AEX were resolved by SDS-PAGE and visualised by western blot (top panel) and Imperial staining (bottom panel). Fractions containing the most glycoconjugate were pooled and AEX repeated. Images (A–D) represent sequential rounds of AEX, from first to last, with each well containing a different elution fraction. Western blots were visualised using anti-6xHis-tag antibody (green) and anti-S. sonnei O-antigen (SSOAg) antibody (red/orange).

3.3. Anti-protein immune response

The immunogenicity of the SSOAg glycoconjugates was evaluated using C57BL/6j mice, immunised following a three-dose schedule of 10 µg protein per dose. Each glycosylated and unglycosylated protein was tested individually, alongside a PBS-only control group. All antigens were found to be well-tolerated.

Initial screening of the anti-carrier protein IgG response revealed that all serum samples, except those from PBS-immunised mice, recognised both ExoAComP and MdtAComP. As both proteins share the ComP tag, it was hypothesised that the observed cross-reactivity was attributable to anti-ComP antibodies. Production of ExoA and MdtA only, without ComP (Supplementary Figure 1B), eliminated this cross-reactivity, and an anti-protein IgG response was only detected when probed with sera from mice immunised with the cognate carrier protein (Supplementary Figure 5). As such, ExoA and MdtA only, without ComP tags, were used for characterisation of the anti-protein antibody response.

Immunisation with both glycosylated and unglycosylated carrier proteins elicited a protein-specific IgG response. A significantly increased anti-ExoA IgG response was observed in mice vaccinated with either glycosylated or unglycosylated ExoAComP, compared to the PBS control (Figure 4; Supplementary Figure 6). A comparable result was found for both MdtAComP vaccines against MdtA only. No significant difference in antibody response was observed between glycosylated and unglycosylated protein, although the mean IgG response to unglycosylated MdtAComP was higher than that for the glycosylated protein. Overall, these data indicate that glycosylation does not abolish the ability of either carrier to induce a protein-specific antibody response.

Figure 4.

Anti-carrier protein antibody response to ExoA and MdtA. ELISA quantification of IgG response to (A) ExoA and (B) MdtA in mice immunised with glycosylated or unglycosylated ExoAComP or MdtAComP. A 6-point, 3-fold dilution series of individual serum samples from each group were tested in duplicate against purified ExoA or MdtA and IgG response reported as the reciprocal serum dilution that gave an OD 0.5. The reciprocal for PBS-immunised mice was outside of the tested dilution series (>1/600) so a value of 599 was used for visualization and data analysis. Statistical significance was determined by Kruskal Wallis with Dunn's multiple comparison test, *p=<0.05, **p=<0.01, ***p=<0.001. SSOAg, S. sonnei O-antigen.

3.4. Anti-SSOAg immune response

To characterise the anti-glycan antibody response, serum samples were probed against whole E. coli cells producing the SSOAg. Given their similarity, both Shigella and E. coli encode copies of the MdtA protein, which share a high degree of amino acid sequence homology. Although MdtA is a putative periplasmic protein, it may be accessible on the cell surface; therefore, to avoid interference from anti-MdtA antibodies, mdtA was deleted from the E. coli strain W3110, using lambda red recombineering (Supplementary Figure 7). W3110ΔmdtA carrying either pBPSO (SSOAg) or pBBR1MCS-3 (empty vector) was then used to assess the anti-SSOAg IgG response. It was necessary to omit one serum sample from the PBS group as this was highly reactive against all E. coli cells, irrespective of glycan expression.

Serum IgG responses against the SSOAg were detected for both glycoconjugates (Figure 5). When tested against E. coli expressing the SSOAg, pooled sera from the SSOAg-ExoAComP immunisation group demonstrated an eight-fold increase in signal compared to ExoAComP only (Supplementary Figure 8). Similarly, the SSOAg-MdtAComP signal was approximately two-fold higher compared to MdtAComP only. Furthermore, the anti-SSOAg response for glycosylated ExoAComP was approximately three-fold higher than that for glycosylated MdtAComP. Collectively, we have demonstrated the first use of PglS to generate novel, immunogenic S. sonnei glycoconjugates, and the use of Shigella-derived proteins as alternative, species-specific glycan carriers.

Figure 5.

Antibody response to S. sonnei O-antigen. ELISA quantification of IgG response to whole E. coli cells producing (A) S. sonnei O-antigen or (B) no O-antigen, in mice immunised with glycosylated or unglycosylated ExoAComP or MdtAComP. OD₄₅₀ is reported for the 6-point, 3-fold dilution series of pooled serum from each group, and data were fitted with a four-parameter logistic curve. Error bars represent standard deviation for three technical replicates. SSOAg, S. sonnei O-antigen.

4. Discussion

The application of bioconjugation for the synthesis of Shigella glycoconjugates offers flexibility in design and the potential for low-cost vaccine manufacturing, a current imperative given the burden of shigellosis in LMICs. Here, we demonstrate the first use of bioconjugation for the generation of dual-target S. sonnei-specific glycoconjugates.

The majority of glycoconjugate vaccines targeting S. sonnei and/or S. flexneri O-antigen currently in clinical trials use established protein carriers such as tetanus toxoid or ExoA (12, 13, 19, 46). Species-specific carrier proteins could circumvent issues of immunotolerance from repeated exposure to the same carrier proteins, provide a means of dual immune stimulation, and generate serotype-independent immunity using a common antigen (27, 28, 30, 47). Research into Shigella-specific protein vaccine candidates has primarily focused on the immunogenic, well-characterised and conserved Ipa proteins, which are part of the type III secretion system (48). Invaplex is a high-molecular-mass complex of IpaB and IpaC with one or more purified Shigella LPS (20). In a pulmonary lung model, mice immunised with Invaplex containing IpaB, IpaC and purified S. sonnei LPS were protected from challenge with both S. sonnei and S. flexneri 2a (21). Similar cross-serogroup protection was reported following immunisation of mice with S. flexneri 2a O-antigen chemically conjugated to IpaB (49), demonstrating the potential for protein-induced heterologous immunity. However, both chemical glycoconjugates and the Invaplex complexes require multiple expression and purification steps to yield the final product.

The generation of the novel Shigella glycoconjugates described here was enabled by two key developments, reported for the first time in this study. Firstly, we demonstrate biosynthesis of the S. sonnei O-antigen (SSOAg) in E. coli cells by adopting the locus from P. shigelloides O17, which encodes an identical O-antigen (35). Such efficient, heterologous production of the SSOAg could be used for the development of both chemical and biological conjugates. Secondly, the OST PglS was successfully employed to transfer SSOAg, which harbours an unusual reducing-end monosaccharide, FucNAc4N, to modified Shigella carrier proteins. The identification of PglB and characterisation of its relaxed glycan substrate specificity paved the way for bioconjugate vaccine development (10, 11, 50). However, as demonstrated here, reliance on wild-type PglB alone restricts antigen choice to only those glycans permissible for transfer by the OST. The identification of novel OSTs, such as PglS, and structural-guided mutagenesis of existing ones have expanded the glycoengineering toolbox and, consequently, the repertoire of glycans available for bioconjugation (38, 51, 52).

Protein arrays have been widely utilised in reverse vaccinology as a high-throughput means of identifying novel vaccine antigens for further characterisation and evaluation (53). This understanding could be exploited to produce next-generation bioconjugate vaccines. de Alwis et al. identified several novel Shigella immunoantigens by probing a subset of the Shigella proteome with acute and convalescent sera from patients with natural Shigella infection (44). This included both EmrK and MdtA, which could be glycosylated in this study when fused with a C-terminal ComP tag, the substrate of PglS (36). However, glycoconjugate yield varied between carrier proteins, despite their fusion to identical C-terminal tags. The structure of the carrier protein could potentially influence this, for example, if it prevents correct folding of ComP, which is required for PglS recognition and subsequent glycosylation (54), or partially blocks its access to the glycan attachment site.

Glycosylation of novel carrier proteins must avoid masking important protein epitopes with the glycan (28, 30). Bioconjugation offers an advantage over chemical conjugation for this process, as the site of glycosylation can be precisely determined by placement of the OST-recognition sequence. de Alwis et al. reported that immunisation of rabbits with purified MdtA elicited a robust, protein-specific IgG response (44). In our work, using C-terminal glycosylation, we confirm the immunogenicity of both glycosylated and unglycosylated MdtA in a mouse model, suggesting that both ComP fusion and glycosylation do not inhibit stimulation of an anti-MdtA IgG response. The mean IgG response against MdtAComP was higher compared to its glycosylated counterpart, although this difference was not significant. Notably, anti-ComP antibodies were also generated against both carriers, but this did not prevent the generation of an ExoA- or MdtA-specific response.

Selection and design of new glycoconjugate carrier proteins requires multiple considerations such as immunogenicity, toxicity, manufacturability and regulatory approval. Additionally, as demonstrated here, it is necessary to characterise each novel carrier individually, as although the same C-terminal ComP tag was used for glycosylation of each protein, there was still variation in glycoconjugate yield and immune response. This could be further investigated through testing multiple configurations of each carrier protein, both in terms of number and position of glycosylation site(s). For example, glycoconjugate yield can be improved through inclusion of additional glycosylation sites, often achieved using multiple terminal “glycotags” (39). As was found in this study, this approach can be challenging when using large glycosylation tags such as the 114-amino-acid ComP, but shorter ComP sequons have been identified that can still be glycosylated and could be more amenable to multiple copies (54). Indeed, Knoot et al. recently demonstrated enhanced glycosylation of ExoA through incorporation of up to six 23-amino-acid ComP sites, and demonstrated how position of these sites within the protein sequence can impact glycosylation (55). Furthermore, directed evolution of PglS generated a mutant capable of glycosylating two serine residues within a single ComP sequence, again providing opportunities for increased polysaccharide loading of carrier proteins (56).

Critically, both glycoconjugates induced production of an anti-SSOAg IgG response, a key characteristic of effective Shigella glycoconjugate vaccines. A review of evidence obtained from both observational and vaccine efficacy studies found that serum IgG antibodies against Shigella LPS are associated with protection from shigellosis (17). The anti-O-antigen response was approximately two-fold higher for SSOAg-ExoA compared to SSOAg-MdtA, but the reason for this is unknown. Although both glycoconjugates were highly glycosylated with minimal unglycosylated protein present, vaccine dose was based on protein concentration; therefore, there may have been variation in the amount of glycan in each vaccine. Reglinski et al. investigated the use of novel carriers in double-hit bioconjugate vaccines against Streptococcus pneumoniae serotype 4, and also found heterogeneity in the anti-capsular antibody response between the different carriers (32).

The detection of antibody binding to a chosen antigen is a key first step in the development of a novel vaccine candidate. One limitation of this study is the lack of full characterisation of the immune response to determine whether the novel glycoconjugates developed can confer heterologous protection across Shigella serogroups. Determining protective efficacy of a vaccine using challenge studies has previously been limited for Shigella due to the lack of suitable small animal models, but these are improving (57). Instead, serum bactericidal assays (SBAs) have been used to assess antibody functionality (58). Although it was not possible to perform SBAs in this study due to lack of sufficient sera, IgG binding was confirmed for both the SSOAg and the two protein carriers. Previous studies have demonstrated the bactericidal activity of antibodies against Shigella O-antigens (59, 60). Furthermore, de Alwis et al. reported that antibodies raised in rabbits against MdtA were bactericidal against S. flexneri (44). This demonstrates the potential for these glycoconjugates to raise bactericidal antibodies and warrants further investigation.

In summary, we demonstrate the heterologous production of a prototype, two-pronged Shigella bioconjugate by combining a conserved Shigella-specific protein and S. sonnei glycan antigen. This generated a glycan- and protein-specific antibody response in a mouse immunisation model, thus demonstrating the flexibility and utility of bioconjugation for the generation of such dual-target glycoconjugates.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This work was supported by the Wellcome Trust (WT223838/Z/21/Z) and a BactiVac Pump-Priming Project Award.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

Ethics statement

The animal study was approved by the University of Cambridge Animal Welfare and Ethical Review Body. The study was conducted in accordance with the local legislation and institutional requirements.

Author contributions

CH: Writing – review & editing, Writing – original draft, Investigation, Project administration, Data curation, Formal analysis, Conceptualization, Visualization, Methodology. TF: Data curation, Writing – review & editing, Investigation. SC: Writing – review & editing, Investigation, Methodology. KH: Investigation, Writing – review & editing. EK: Methodology, Investigation, Writing – review & editing. SB: Funding acquisition, Writing – review & editing, Conceptualization. BW: Writing – review & editing, Conceptualization, Funding acquisition.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2026.1907725/full#supplementary-material