Development of Shigella conjugate vaccines targeting Shigella flexneri 2a and S. flexneri 3a using a simple platform-approach conjugation by squaric acid chemistry

Abstract

There is a need for vaccines effective against shigella infection in young children in resource-limited areas. Protective immunity against shigella infection targets the O-specific polysaccharide (OSP) component of lipopolysaccharide. Inducing immune responses to polysaccharides in young children can be problematic, but high level and durable responses can be induced by presenting polysaccharides conjugated to carrier proteins. An effective shigella vaccine will need to be multivalent, targeting the most common global species and serotypes such as Shigella flexneri 2a, S. flexneri 3a, S. flexneri 6, and S. sonnei. Here we report the development of shigella conjugate vaccines (SCV) targeting S. flexneri 2a (SCV-Sf2a) and 3a (SCV-Sf3a) using squaric acid chemistry to result in single point sun-burst type display of OSP from carrier protein rTTHc, a 52 kDa recombinant protein fragment of the heavy chain of tetanus toxoid. We confirmed structure and demonstrated that these conjugates were recognized by serotype-specific monoclonal antibodies and convalescent sera of humans recovering from shigellosis in Bangladesh, suggesting correct immunological display of OSP. We vaccinated mice and found induction of serotype-specific OSP and LPS IgG responses, as well as rTTHc-specific IgG responses. Vaccination induced serotype-specific bactericidal antibody responses against S. flexneri, and vaccinated animals were protected against keratoconjunctivitis (Sereny test) and intraperitoneal challenge with virulent S. flexneri 2a and 3a, respectively. Our results support further development of this platform conjugation technology in the development of shigella conjugate vaccines for use in resource-limited settings.

1. Introduction

Shigellosis affects both adults and children, with children living in resource-limited countries bearing the largest burden [1]. Shigella spp. infect intestinal epithelial cells and once internalized, spread to adjacent cells accompanied by a prominent intestinal inflammatory response that can clinically manifest as bacillary dysentery. Shigella infection is one of the top three global causes of moderate/severe diarrhea among children younger than five years of age, and is the leading cause in children less than two years of age [2,3]. Shigellosis can be caused by one of four bacterial species (Shigella flexneri, S. sonnei, S. boydii and S. dysenteriae). Globally, the majority of shigellosis is currently caused by S. flexneri and S. sonnei. Although there are numerous S. flexneri serotypes, S. flexneri 2a, 3a and 6 are the most common globally. It has been postulated that a vaccine that protects against S. flexneri 2a, 3a and 6, as well as S. sonnei could prevent > 80 % of global cases of shigellosis [4,5]. Serotype-specificity is determined by the O-specific polysaccharide (OSP) of the lipopolysaccharide (LPS), and IgG antibodies that target Shigella spp LPS are associated with protection against shigella infection in humans [6-12].

A number of shigella vaccines are in various stages of development, including ones based on conjugation, to develop immune responses to the OSP component of Shigella spp LPS [6,8,10-36]. We have previously described the use of an inexpensive conjugation vaccine technology based on squaric acid chemistry to directly link bacterial O-specific polysaccharides recovered from Vibrio cholerae to a recombinant fragment of tetanus toxin heavy chain (rTTHc) to produce a cholera conjugate vaccine (CCV) [37-41]. This technology attaches bacterial OSP to rTTHc via an active glucosamine present in the oligosaccharide core recovered from lipopolysaccharide (LPS), obviating the need for introduction of a linker, and resulting in a sun-burst type display of OSP similar to how OSP is displayed by bacteria [37,42,43]. This technology can load OSP onto rTTHc at controlled ratios, is reproducible, is scalable [37], and cGMP cholera conjugate vaccine using this approach has been produced, evaluated in pre-clinical and toxicologic analyses, and is currently being evaluated in Phase 1 studies in humans. Shigella flexneri also contains a single active amine in the oligosaccharide core of LPS [44]. Here, we describe the use of this conjugation platform technology to develop conjugate vaccines targeting S. flexneri 2a (Sf2a) and S. flexneri 3a (Sf3a) (Fig. 1).

Fig. 1.

Conjugation of purified Sf2a and Sf3a OSPs to rTTHc and BSA using squaric acid chemistry. The OSP repeating units (RUs) of both Sf2a and Sf3a are pentasaccharides [5,44,69] comprised of a common backbone tetrasaccharide (three rhamnose residues and one N-acetylglucosamine) with an α-D-Glcp branched out from different rhamnoses. The OSP is attached to an oligosaccharide core. Importantly, oligosaccharide core contains a single glucosamine that can be used to link polysaccharide to a protein carrier using squarate chemistry. No active amine is present in the RU in Shigella flexneri 2a or 3a. Such active-amine linkage allows single point attachment using native oligosaccharide core (obviating the need for introduction of a spacing linker), resulting in a “sun burst display” of OSP, which mimics polysaccharide display on S. flexneri bacteria. ND, not determined.

2. Materials and methods

2.1. Bacterial strains and reagents

The bacterial strains used in this analysis included two clinical isolates of S. flexneri 2a (Sf2a) Sf2a260214_1 and S. flexneri 3a (Sf3a) Sf3a050214_5 collected in a fully-approved study from patients presenting with diarrhea at the International Centre for Diarrhoeal Disease Research, in Dhaka Bangladesh (icddr,b). These strains were used for the generation of Sf2a and Sf3a LPS and OSP for use in immunologic assays, generation of the OSP used in the conjugations described in this work, and use in bactericidal assays. S. flexneri 2a 2457T and S. flexneri 3a J17B [27] were used in the Sereny and intraperitoneal challenge models, respectively. All strains were grown in Tryptic Soy broth (TSB, Becton Dickinson, Sparks MD) and Tryptic Soy Agar plates (BD, Sparks MD). Presence of the virulence plasmid was confirmed by Congo Red analysis [45].

2.2. Production and characterization of OSP, rTTHc, and OSP:rTThc

LPS was prepared from S. flexneri 2a Sf2a260214_1 and S. flexneri 3a Sf3a050214_5 as previously described [37,40,41,43,46]. Also as previously described, OSP was recovered from LPS using mild acid hydrolysis, chloroform extraction, and size exclusion chromatography [40,41,43]. Peak 2 material was further purified by ultrafiltration with a series of Millipore Amicon Ultra tubes (100 kDa, 30 kDa, and 3 kDa cut-offs). Ultrafiltrations were performed at 4 °C, 7,500 Relative Centrifugal Force (RCF) against ultra-pure water 8 times, per manufacturer’s instructions. The fraction that went through the 30 kDa membrane but was retained by the 3 kDa membrane was analyzed by ¹H and ¹³C NMR and used for conjugation.

Purified OSPs of Sf2a and Sf3a were converted to their respective squarate derivatives. Briefly, Sf2a or Sf3a OSP (1 equivalent) was dissolved in pH 7 phosphate buffer (0.5 M) to form a 4 mM solution. Before addition of the reagent, a fluorescamine assay [47] was performed to record the concentration of the free amine in the solution. 3,4-dimethoxy-3-cyclobutene-1,2-dione (20 equivalent) was added, and the mixture was gently stirred at room temperature for 18 h when the fluorescamine assay showed that the free amine concentration decreased by > 99 %. The solution was then transferred into an Amicon Ultra (3 k Da cutoff) tube and ultrafiltered against ultra-pure water (at 4 °C, 7,500 RCF) 8 times. The retentate was lyophilized to afford the OSP-squarate derivative as a white fluffy solid.

Recombinant rTTHc was prepared [40,41], and conjugation of OSP to rTTHc was performed using squaric acid chemistry as previously described [37,40,41,43,46] to produce OSP:rTTHc Sf2a and OSP:rTTHc Sf3a. Briefly, rTTHc (1 equivalent) and OSP-squarate (12 equivalents) were dissolved in 0.5 M pH 9.5 borate buffer to form a 5 mM solution with respect to the OSP-squarate. The clear solution formed was stirred at room temperature for 48 h. The reaction mixture was then ultrafiltered through a Millipore Amicon Ultra (30 kDa cutoff) tube against 10 mM ammonium carbonate solution (at 4 °C, 7,500 RCF) 8 times. The retentate, after lyophilization, yielded OSP:rTTHc conjugate as a white fluffy solid. To facilitate binding of saccharide to plastic used in immunologic assays, we also conjugated OSP to BSA (bovine serum albumin; Sigma) to produce Sf2a OSP:BSA and Sf3a OSP:BSA. Conjugates were resuspended in distilled water (1 mg/mL) prior to use.

2.3. Immuno-recognition of conjugates using anti-shigella monoclonal antibodies and human convalescent sera

We assessed immuno-recognition of OSP:rTTHc and OSP:BSA using mouse monoclonal antibodies and human convalescent sera. In brief, we coated 96-well microtiter plates (Nunc) with OSP: rTTHc, OSP:BSA or rTTHc (1 μg/ml). Following blocking and washing of plates, we assessed serotype-specificity by adding mouse anti-shigella specific monoclonal antibodies (Reagensia AB, Stockholm; MASF IgM B reacts with any Sf; IgM II, Y-5 react with Sf2a; IgG 6, 7/8 react with Sf3a). We detected the presence of antigen-specific antibodies using horseradish peroxidase-conjugated goat anti-mouse IgG/IgM antibody (diluted 1:5000 in 0.1 % BSA in phosphate buffered saline-Tween: PBS-T) (Jackson Immuno Research). After 1.5 h incubation at 37°C, we developed the plates with a 0.55 mg/ml solution of 2,2′ 0-azinobis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS; Sigma) with 0.03 % H₂O₂ (Sigma) and determined the optical density at 405 nm with a Vmax microplate kinetic reader (Molecular Devices Corp., Sunnyvale, CA).

To detect immune-recognition by human sera, we used 96-well microtiter plates coated and blocked as above and used pooled acute (day 2) and early convalescent (day 7) phase sera collected from humans in Bangladesh with culture-confirmed shigellosis, cholera or typhoid (diluted 1:500 shigella, 1:2500 cholera and typhoid in 0.1 % BSA in PBS-T). We detected the presence of antigen-specific antibodies using horseradish peroxidase conjugated to goat anti-human IgG antibody (diluted 1:5000 in 0.1 % BSA in PBS-T) (Jackson ImmunoResearch, West Grove, PA).

3. Immunogenicity

3.1. Vaccination and collection of samples

To assess immunogenicity of the SCV-Sf2a and SCV-Sf3a OSP: rTTHc conjugate vaccines following immunization, we performed two serial vaccination analyses in animals. We first intramuscularly immunized cohorts (n = 15) of 3–5 week old female Balb/c mice with 20 μg of saccharide/dose. To assess the kinetics of immune responses, mice were vaccinated on days 0, 21, and 42, and then boosted on day 70 and 98. Control mice were vaccinated with PBS alone at all time points. We collected blood samples via tail bleeding on days 0, 21, 28, 42, 49, 56, 77, and 105. Samples were collected, processed, aliquoted, and stored as previously described [37,40,41]. These initial samples were used to assess serum IgG, IgM, and IgA antibody responses as well as adsorption and serum bactericidal responses. Following confirmation of immunogenicity, we then immunized second cohorts of mice with SCV Sf2a OSP:rTTHc and SCV Sf3a OSP:rTTHc or PBS, confirmed immunogenicity and then used these cohorts in direct virulent organism challenge models (described below). The second cohort of mice vaccinated against Sf2a received 10 μg of saccharide/dose on day 0, 21, 42, was boosted on day 75, and was challenged by Sereny conjunctival challenge on day 90. The second cohort of mice vaccinated against Sf3a received 10 μg of saccharide/dose on day 0, 21, 42, and was challenged on day 60 via intraperitoneal inoculation of virulent bacteria.

3.2. Antigen-specific antibody responses in serum

OSP, LPS and TT-specific IgG, IgM and IgA responses in serum of vaccinated mice were measured using standard enzyme-linked immunosorbent assay (ELISA) protocols, as previously described [37,40,41]. Briefly, to assess anti-OSP, anti-LPS and anti-TT antibody responses, ELISA plates were coated with OSP:BSA (100 ng/well), LPS (250 ng/well) or rTTHc (100 ng/well) in 50 mM carbonate buffer. Sera (OSP, TT diluted 1:25 in 0.1 % BSA in PBS-T; LPS 1:100 IgG, 1:1000 IgM in 0.1 % BSA in PBS-T) were added at 100 μl per well and the presence of antigen-specific antibodies was detected using horseradish peroxidase-conjugated goat anti-mouse IgG, IgM or IgA antibody (diluted 1:1000 in 0.1 % BSA in PBS-T) (Southern Biotech, AL). Plates were developed and read as described above. We defined a responder as having an increase in antigen-specific antibody responses compared to day 0 that was greater than the maximal increase observed at any time point in control animals.

3.3. Adsorption studies

To assess cross-reactivity among antibody responses elicited for SCV Sf2a and SCV Sf3a, we performed cross-adsorption analyses. Briefly, we adsorbed day 105 serum from mice immunized with SCV Sf2a OSP:rTTHc with Sf3a OSP:BSA and assessed the impact on IgG responses using plates coated with Sf2a OSP:BSA or Sf3a OSP:BSA. We also performed the reciprocal analysis: we adsorbed day 105 sera of mice vaccinated with SCV Sf3a OSP:rTTHc with Sf2a OSP:BSA and assessed the impact on IgG responses using plates coated with Sf3a OSP:BSA or Sf2a OSP:BSA. Adsorption was performed by mixing post-vaccination sera with 100 ng of OSP:BSA overnight at 4 °C in PBS solution. The sera were then applied to plates coated with the target OSP:BSA and developed as described above.

3.4. Serum bactericidal responses

We assessed serum bactericidal antibody titers against S. flexneri 2a Sf2a260214_1 and S. flexneri 3a Sf3a050214_5 in a micro-assay. We inactivated endogenous complement activity of mouse serum by heating it for 30 min at 56° C. We then added 50 μl aliquots of two-fold serial dilutions of heat-inactivated sera in 0.15 M saline (1:12.5 to 1:12,800) to wells of sterile 96-well tissue culture plates containing 50 μl/well of S. flexneri 2a strain Sf2a260214_1 or S. flexneri 3a strain Sf3a050214_5 (OD 0.1) in 0.15 M saline and 25 % guinea pig complement (EMD Biosciences). Plates were then incubated for 1 hr at 37°C. 150 μl of brain heart infusion broth (Becton Dickinson) was then added to each well, and plates were incubated overnight (16–18 hrs) at 25° C, when optical density at 595 nm was assessed. We calculated the bactericidal titer as the dilution of serum associated with 75 % reading in optical density compared with that of wells containing no serum. We defined a responder as having ≥ 4-fold increase in bactericidal reciprocal end-dilution titer compared to baseline.

3.5. Mouse challenge models

To assess protection directly in vaccinated animals, we used the second cohorts of mice vaccinated with 10 μg/dose of polysaccharide of SCV-Sf2a OSP:rTTHc (day 0, 21, 42; boosted day 75; challenged day 90) or PBS, or 10 μg/dose of polysaccharide SCV-Sf3a OSP:rTTHc (day 0, 21, 42; challenged day 60) or PBS. We performed a Sereny assay on mice vaccinated with SCV-Sf2a OSP:TTHc or PBS, using 10⁷ CFU of virulent S. flexneri 2a strain 2457T [48,49]. Briefly, we administered 10 μl containing Sf2a 2457T into one conjunctival sac of vaccinated anesthetized mice, closed the eyes, and gently massaged for 1 min. Mice were then allowed to wake up and were returned to their cages and regularly monitored [48]. The degree of conjunctivitis was scored and reported at 48 hrs (1 – no disease; 2-mild keratoconjunctivitis; 3- keratoconjunctivitis with some purulence; 4 fully developed conjunctivitis with copious purulence). We also directly assessed protection in mice vaccinated with SCV-Sf3a OSP:TTHc or PBS, using 10⁶ CFU of virulent S. flexneri 3a strain J17B [50,51]. In this assay, vaccinated mice were injected intra-peritoneally with 100 μl containing J17B and monitored for up to 48 h.

3.6. Ethics statement

Animal work met all governmental and institutional requirements, guidelines, and policies. This work was approved by the Massachusetts General Hospital Subcommittee on Research Animal Care (SRAC). The work adheres to the USDA Animal Welfare Act, PHS Policy on Humane Care and Use of Laboratory Animals, and the “ILAR Guide for the Care and Use of Laboratory Animals”. Sera from humans recovering from shigellosis, cholera and typhoid were collected from patients at the International Centre for Diarrhoeal Disease Research in Dhaka, Bangladesh (icddr,b). This study was approved by the Ethical Review and Research Review Committee of the icddr,b and the Institutional Review Board of Massachusetts General Hospital. Written informed consent was obtained from guardians of child participants (<18 years), and adult participants (≥18 years) provided their own consent.

3.7. Statistics and graphs

We compared data from different groups using Mann-Whitney U tests. Within each group comparisons of data from different time points were carried out using Wilcoxon Signed-Rank tests. Kaplan-Meier and log rank analysis were performed to compare survival curves in the challenge study. All reported P values were two-tailed, with a cutoff of P < 0.05 considered a threshold for statistical significance. We performed statistical analyses using GraphPad Prism 4 (GraphPad Software, Inc.).

4. Results

4.1. Production and characterization of OSP and conjugates

OSP was recovered from LPS following mild hydrolysis followed by column chromatography (Hiprep 16/60 Sephacryl S-200 column, Cytiva); Sf2a and Sf3a OSP both eluted as three peaks (Supplemental Fig. 1). Following additional purification steps using the Millipore Amicon Ultra tubes (100 kDa, 30 kDa, and 3 kDa cut-offs), for use in conjugations and analyses we focused on peak 2 containing polysaccharides between 3 kDa and 30 kDa in size. Recovery rates of the desired OSP fraction from peak 2 material were 53 % for Sf2a and 69 % for Sf3a of column purified fractions. ¹H and ¹³C NMR of Sf2a OSP and Sf3a OSP confirmed identity of the fractions (Fig. 2). OSP-squarate derivatives were produced and conjugated to rTTHc or BSA as described above. Sf2a OSP: rTTHc was produced from 6.0 mg of Sf2a OSP-squarate conjugated with 2.0 mg of rTTHc to produce 4.0 mg of Sf2a OSP:rTTHc conjugate (yield 98 %, conjugation efficiency 34 %). Sf3a OSP:rTTHc was produced from 3.87 mg of Sf3a OSP-squarate conjugated with 1.29 mg of rTTHc to produce 2.4 mg of Sf3a OSP:rTTHc conjugate (yield 97 %, conjugation efficiency 30 %). Sf2a OSP:BSA was produced from 1.76 mg of Sf2a OSP-squarate conjugated with 0.75 mg of BSA to produce 1.37 mg of Sf2a OSP:BSA conjugate (yield 95 %, conjugation efficiency 37 %); and Sf3a OSP:BSA was produced from 1.76 mg of Sf3a OSP-squarate conjugated with 0.75 mg of BSA to produce 1.1 mg of Sf3a OSP:BSA conjugate (yield 85 %, conjugation efficiency 26 %).

Fig. 2. (A) ¹H NMR of purified Sf2a OSP and Sf3a OSP antigens used in conjugation; (B) ¹³C NMR of same.

The signals of OSP predominate; however, some signature peaks, such as the two peaks in ¹³C NMR between 53 and 54 ppm representing C-2 of the two amino sugars (GlcNAc and GlcN) in core demonstrate retention of inner and outer core. Both ¹H NMR and ¹³C NMR experiments were measured in D₂O at 25 °C in Shigemi NMR tube, at 600 MHz (150 MHz for ¹³C NMR) on a Bruker Avance Spectrometer with a TCI cryoprobe. Samples were prepared at ~10 mg/mL concentration.

The average molecular weight of the conjugates was determined by SELDI-TOF MS (Fig. 3). Once the average molecular weight of the conjugate was known, the efficiency (the percentage of antigen effectively attached to carrier protein) and the yield (reaction yield based on carrier protein recovery) of the conjugation was calculated. SELDI-TOF MS also provided information about the size of antigens. In the case of Sf3a, SELDI-TOF profiles showed peaks and valleys at lower loading stages of the conjugation (Fig. 3B and 3D). Peaks at 75,437 Da, 88,894 Da, and 100,021 Da in Fig. 3B represent 2, 3, and 4 loading; peaks at 78,598 Da, 90,052 Da, 103,217 Da in Fig. 3D represent 1, 2, and 3 loading, respectively. Therefore, the average molecular weight of the Sf3a OSP-core antigen was determined to be ~12 kDa. Based on an estimated average molecular weight of ~100 kDa for the Sf3a OSP: rTTHc, the average molar loading would be approximately 4.0 OSPs to 1 rTTHc. The oligosaccharide core of Shigella spp approximates 1.8 kDa [44], and the pentasaccharide repeating unit of Sf2a and 3a approximates 0.820 kDa. Knowing the size of the Sf3a OSP of ~12 kDa, and the molecular weight of a single pentasaccharide repeating unit to be ~821 Da (with no O-acetylation), the average repeating unit length is estimated at 12–13 for Sf3a OSP.

Fig. 3. SELDI-TOF MS results of Sf2a and Sf3a OSP conjugates.

(A) Sf2a OSP:rTTHc, (B) Sf3a OSP:rTTHc, (C) Sf2a OSP:BSA, (D) Sf3a OSP:BSA. rTTHc 52 kDa; BSA 66 kDa. The average mass of OSP for Sf3a (12 kDa) is calculated based on the mass difference of two neighboring peaks in the SELDI-TOF-MS (Surface-Enhanced Laser Desorption Ionization-Time of Flight-Mass Spectrometry) figures. The average molar ratio (loading) of OSP to rTTHc is determined by the intensity of different peaks. ~4 is determined to be the loading of Sf3a OSP:rTTHc. No specific peaks are identifiable in Sf2a conjugates, suggesting a wider range of saccharide length distribution. (The average RU length and molar loading of OSP for Sf2a conjugates is estimated by integration of ¹H NMR peaks as described in the main text).

OSP-core antigen is a mixture of OSP-core molecules with different numbers of OSP repeating units. The wider the distribution of different molecular weight material in the mixture, the broader the peaks that will show on SELDI-TOF. Compared to the Sf3a conjugates with observable peaks and valleys on SELDI-TOF MS analysis, both Sf2a OSP conjugates showed a broad curve (Fig. 3A and 3C) indicating that the Sf2a OSP has a wider distribution of OSP-core molecules with different number of OSP repeating units compared to Sf3a OSP. To estimate an average repeating unit length for Sf2a OSP, we compared integration ratio of the combined methyl groups of rhamnoses in the OSP to the H-1 of the alpha-galactose in the core oligosaccharide and estimated an average of approximately 19 repeating units for Sf2a OSP. This result would suggest an OSP to rTTHc molar loading in the Sf2a conjugate of approximately 3.1:1.

4.2. Average saccharide to protein ratio (w/w) in conjugates

Based on a molecular weight of rTTHc of 52 kDa and BSA of 66.43 kDa, and average weight of the conjugates, the saccharide to protein w/w ratios would be: Sf2a OSP:rTTHc, based on average m. w. of 106 kDa, the sugar/protein ratio (w/w) would be 1.04/1; Sf3a OSP:rTTHc, based on average m.w. of 100 kDa, the sugar/protein ratio (w/w) would be 0.92/1; Sf2a OSP:BSA, based on average m.w. of 128 kDa, the sugar/protein ratio (w/w) would be 0.93/1; For Sf3a OSP:BSA, based on average m.w. of 115 kDa, the sugar/protein ratio (w/w) would be 0.73/1.

4.3. Immuno-reactivity of OSP-conjugates with serotype-specific monoclonal antibodies and convalescent human sera from patients recovering from culture-confirmed shigellosis in Bangladesh

Once we confirmed OSP identity and conjugation, we next moved to assess whether the OSPs in the conjugates were displayed in immunologically relevant manners. To do this, we first assessed immuno-recognition of OSP:rTTHc and OSP:BSA using mouse monoclonal antibodies targeting specific S. flexneri epitopes. As shown in Supplemental Fig. 2, both Sf2a and Sf3a OSP:rTTHc and OSP:BSA were correctly recognized. To assess potential clinical relevancy, we then assessed whether these conjugates were displaying OSP in a manner recognized by patients recovering from shigellosis in Bangladesh (Fig. 4). Both Sf2a and Sf3a OSP:rTTHc and OSP:BSA were correctly recognized by human convalescent sera from shigellosis patients in Bangladesh, and not by human convalescent sera from patients recovering from cholera or typhoid fever. These data suggest that the conjugates not only displayed serotype-specific OSP moieties, but displayed them in a clinically relevant manner.

Fig. 4. Immuno-recognition of OSP components of (A) Sf2a and Sf3a OSP:BSA and (B) Sf2a and Sf3a OSP:rTTHc by early convalescent phase sera (Day 7) compared to acute phase (Day 2) of humans in Bangladesh recovering from shigellosis caused by Sf2a or Sf3a.

Note no increased convalescent phase recognition by serum of patients with cholera caused by V. cholerae O1 or typhoid fever in Dhaka, Bangladesh. *denotes a statistically significant increase (p ≤ 0.05) on day 7 compared to day 2; ** p ≤ 0.01; *** p ≤ 0.001.

4.4. Immunogenicity of SCV Sf2a OSP:rTTHc and Sf3a OSP:rTTHc in vaccinated animals: Immune responses, serotype-specificity, and bactericidal responses

Mice immunized with Sf2a OSP:rTTHc developed OSP-specific IgG (Fig. 5A; p < 0.001) and IgM (Fig. 5B; p < 0.001) targeting both the homologous serotype (example, vaccinated with Sf2a OSP and immunoreactivity with Sf2a OSP), and also against the heterologous serotype (example, vaccinated with Sf2a OSP and immunoreactivity with Sf3a OSP). Similarly, mice immunized with Sf3a OSP: rTTHc developed OSP-specific IgG (Fig. 6A; p < 0.001) and IgM (Fig. 6B; p < 0.001) targeting both the homologous and heterologous serotype. Responder frequency following vaccination was highest when considering homologous serotypes (Figs. 5, 6; and Supplemental Table 1). Immune responses were detected following a single immunization (p < 0.05) and increased further through day 55. Booster vaccinations on day 70 and 98 did not boost cohort mean responses, although there was an increase in responder frequencies. LPS-specific IgG and IgM responses were detected following vaccination with Sf2a OSP:rTTHc or Sf3a OSP:rTTHc (Supplemental Figs. 3 and 4). Immunization with SCVs Sf2a OSP: rTTHc or Sf3a OSP:rTTHc also induced rTTHc-specific IgG responses (Fig. 7A; P < 0.001), and a slight increase in rTTHc-specific IgM responses on day 21 (Fig. 7B). There was no induction of serum IgA OSP or rTTHc responses (Supplemental Fig. 5).

Fig. 5.

Serum IgG (A) and IgM (B) responses at different time points against Sf2a O-specific polysaccharide (OSP) in various cohorts of mice following intramuscular (IM) immunization with Shigella conjugate vaccine-Sf2a OSP:rTTHc (SCV-Sf2a), SCV-Sf3a, or PBS. Dots represent responses in individual mice. Mean and standard error of the mean are reported for each group. * denotes a statistically significant difference (p ≤ 0.05) from baseline (day 0) response; ** p ≤ 0.01; *** p ≤ 0.001. Responder frequencies are also listed. Statistically significant differences among compared cohorts are also represented. See Supplemental material for responder frequency comparisons.

Fig. 6.

Serum IgG (A) and IgM (B) responses at different time points against Sf3a O-specific polysaccharide (OSP) in various cohorts of mice following intramuscular (IM) immunization with Shigella conjugate vaccine-Sf2a OSP:rTTHc (SCV-Sf2a), SCV-Sf3a, or PBS. Dots represent responses in individual mice. Mean and standard error of the mean are reported for each group. * denotes a statistically significant difference (p ≤ 0.05) from baseline (day 0) response; ** p ≤ 0.01; *** p ≤ 0.001. Responder frequencies are also listed. Statistically significant differences among compared cohorts are also represented. See Supplemental Material for responder frequency comparisons.

Fig. 7.

Serum IgG (A) and IgM (B) responses at different time points against rTTHc in various cohorts of mice following intramuscular (IM) immunization with Shigella conjugate vaccine-Sf2a OSP:rTTHc (SCV-Sf2a), SCV-Sf3a, or PBS. Dots represent responses in individual mice. Mean and standard error of the mean are reported for each group. * denotes a statistically significant difference (p ≤ 0.05) from baseline (day 0) response; ** p ≤ 0.01; *** p ≤ 0.001. Responder frequencies are also listed. See Supplemental material for responder frequency comparisons.

To judge serotype-specificity of anti-S. flexneri antibody responses induced by vaccination, we performed adsorption studies, demonstrating that while some antibody could be adsorbed away by absorption with the heterologous OSP, appreciable serotype-specific immuno-recognition remained (Supplemental Fig. 6). Despite the induction of some serotype-cross reactive antibodies as assessed by ELISA, vaccinated animals only developed bactericidal responses to homologous strains (i.e., mice vaccinated against Sf2a developed bactericidal antibodies to Sf2a, and not Sf3a; and mice vaccinated against Sf3a developed bactericidal antibodies to Sf3a, and not Sf2a (Fig. 8).

Fig. 8. Bactericidal responses in vaccinated cohorts of mice.

Dots represent responses in individual mice on day 0 (d0) and day 105. Mean and standard error of the mean are reported for each group. * denotes a statistically significant difference (p ≤ 0.05) from baseline (day 0) response; *** p ≤ 0.001. Responder frequency for ≥4-fold increase in bactericidal reciprocal end-dilution titer compared to baseline.

4.5. Protection against challenge with virulent S. flexneri 2a 2457T and S.flexneri 3a J17B in mice vaccinated with SCVs Sf2a OSP:rTTHc or Sf3a OSP:rTTHc, respectively

To assess protection afforded by vaccination, we used two challenge models. Using mice vaccinated with SCV Sf2a OSP:rTTHc, we confirmed protection in a virulent keratoconjunctival Sereny assay with strain S. flexneri 2a 2457T (Fig. 9A; p < 0.01). Using mice vaccinated with SCV Sf3a OSP:rTTHc, we next confirmed protection in a virulent intra-peritoneal inoculation model using S. flexneri 3a strain J17B (Fig. 9B; p < 0.001). In this assay, by 48 h after challenge, all mice vaccinated with SCV Sf3a OSP:rTTHc were alive and healthy, while all mice vaccinated with PBS were dead (100 % vaccine efficacy; p < 0.001).

Fig. 9. Assessment of protection afforded by vaccination with SCV-Sf2a or Sf3a OSP:rTTHc using two different virulent strain challenge models and distinct vaccine cohorts.

(A) Mice (N = 10) vaccinated with SCV-Sf2a OSP:rTTHc are protected in Sereny assay against wild type Shigella flexneri 2a (Strain 2457 T; 10x7 CFU) intra-conjunctival challenge. (B) Mice (N = 10) vaccinated with SCV-Sf3a OSP:rTTHc are protected against death following intraperitoneal inoculation of wild type Shigella flexneri 3a (Strain J17B; 10x6 CFU). Survival curves are compared by log rank testing.

5. Discussion

Here, we describe development of conjugate vaccines against S. flexneri 2a and S. flexneri 3a using a platform technology. These vaccines display OSP in an immunologically relevant manner, are functional as judged by serotype-specific bactericidal activity, and are protective in virulent challenge models. Studies performed in humans over 20 years ago in endemic regions noted that serum antibody responses and antibody secreting cell (ASC) responses to LPS and invasion plasmid antigens (Ipas) occur following shigellosis [6,52-54], and an association was noted between serum antibody responses and protection against shigella, especially LPS/OSP responses [6-8]. Such antibody responses have been associated with protection against challenge in both animal and human volunteer studies [22,55-57].

Shigella spp. are acid-resistant, human-restricted pathogens that lack a meaningful environmental reservoir. They are usually transmitted person-to-person via fecal-oral contamination, and infectious dose 50s (ID50s) are as low as 10–100 organisms [58]. These attributes make the spread of shigella and outbreaks of shigellosis particularly common in crowded unsanitary conditions such as in informal settlement areas (slums), refugee camps, and among displaced persons. Although providing safe water and adequate sanitation are ultimate and laudable long-term goals, these goals will not be achievable for those most at risk of shigellosis in the short or medium term, which underscores the need for vaccines. Indeed, the growth of informal settlement areas lacking sanitary infrastructure, and the emergence of highly and extensively drug-resistant Shigella species [59-61] have resulted in the realization that effective low-cost vaccines against shigellosis are urgently needed.

A number of shigella vaccines are being developed, including parenteral conjugate vaccines targeting the OSP of Shigella spp (including expression from bioengineered E. coli of a bioconjugate containing Sf2a OSP attached to Pseudomonas aeruginosa exotoxin A; a synthetic glycan based conjugate targeting Sf2a; and an outer membrane vesicle approach [generalized modules for membrane antigens-GMMA] approach targeting S. sonnei) [6,13,15-20,31,32, 34,36,62,63], oral live attenuated Shigella spp vaccines [21-26], inactive oral whole-cell killed vaccines [27,28], and parenteral adjuvanted subunit/multiunit vaccines [17,29,30,64]. All approaches pursue a multivalent approach, usually targeting S. flexneri 2a, 3a, 6 and S. sonnei. Controlled human infection models (CHIMs) are also being developed to inform future vaccine efficacy studies [65-68].

LPS consists of lipid A, oligosaccharide core, and O-specific polysaccharide. The latter defines serotype specificity. Shigella flexneri OSP structures have been defined and do not contain active amines [5,44,69]. Shigella flexneri oligosaccharide cores are related to E. coli cores [44], and importantly contain a single active amine as part of a glucosamine in the inner core [44]. This allowed us to use this amine to create a single-point attachment of OSP-core to a carrier protein via squaric acid chemistry and represents a platform approach since a number of Gram negative pathogens have a single active amine in the core oligosaccharide. Indeed, the technology used in the shigella conjugate vaccines described in our current work builds upon our previous experience with a cholera conjugate vaccine [37-41]. The current approach rests on mild hydrolysis of LPS with recovery of an OSP attached to the oligosaccharide core. We confirmed the characteristic peaks of the repeating saccharides via ¹H and ¹³C NMR analyses of Sf2a and Sf3a OSPs. The purified OSP-core was then directly attached to carrier protein using squaric acid chemistry engaging a single free amine group in the glucosamine present in the residual oligosaccharide core. Such an approach minimizes derivatization steps as it does not require synthesis and introduction of a linker and can employ OSP-core produced in biofermenters [37,42,43]; such attributes can significantly lower production costs. Molar or quantity loading ratios of OSP to carrier protein can be targeted [37,40], and this technology results in non-cross-linked conjugates that are easy to characterize and that display OSP in sun-burst, single-point attachment mode that mirrors OSP display on source bacteria [37,41-43]. Sunburst display of Shigella OSP has been shown to be more immunogenic than lattice displayed OSP [70]. Our biochemical analyses, immune-recognition analyses using human convalescent phase sera, and virulent strain challenge data suggest that our technique results in an immunologically relevant product. The rTTHc that we used in these vaccines has been previously described by us [37,40,41]. It is a recombinant 52 kDa fragment of tetanus toxoid heavy chain whose expression in E. coli has been optimized, including via codon optimization [37]. Optimized recovery, purification, and characterization of rTTHc has also been established [37].

O-acetylation and glucosylation are recognized structural variables in Shigella spp and could affect immunogenicity [71,72]. The mild acid hydrolysis involved in OSP generation, and the alkaline process involved in conjugation in our protocol could be expected to result in de-O-acetylated final product. Despite this, SCV Sf2a OSP:rTTHc and Sf3a OSP:rTTHc were able to be recognized by serotype-specific monoclonal antibodies and convalescent phase human sera, induce LPS-specific immune responses (with preserved O-acetylation patterns) in vaccinated animals, were able to induce serotype-specific bactericidal functional antibody responses, and were able to provide protection in virulent challenge assays in vaccinated animals. Previous groups have also found that non-O-acetylated repeating units of Sf2a OSP are potent immunogens, and have noted a lack of impact of O-acetylation on immune responses and cross-reactivity [73,74]. A non-O-acetylated Sf2a conjugate vaccine is being evaluated in humans [32]. Less is known on the impact of O-acetylation variation of mediating protection against Sf3a [75].

We found the highest immune-recognition of our OSP-conjugates was by convalescent phase sera from humans recently infected with the homologous serotype organism; however, we also detected cross-immuno-recognition of S. flexneri 2a with 3a, and vice versa. In light of the homology of these serotypes, common saccharide backbone, and the use of sera from humans infected with Shigella spp in an informal settlement area of Bangladesh highly endemic for shigellosis, these results are not surprising. Our adsorption studies of sera from vaccinated animals confirmed that, although some cross-reactivity could be adsorbed away, significant serotype-specific immune responses had been induced and were retained. In future work, we will evaluate the degree if any of cross-protection afforded by vaccination in animal models using heterologous strain challenge (example: assess protection against virulent challenge with Sf2a following vaccination with Sf3a OSP:rTTHc and vice versa). The ability of antibodies from patients with shigellosis to activate complement and to kill bacteria in cell free in vitro testing has been described [76-79]. Serum bactericidal activity (SBA) has been found to correlate with protection against shigellosis [80]. These responses may target not only OSP, but also other Shigella antigens [80]. Our current vaccines induced bactericidal activity, with a higher responder frequency for SCV Sf3a than Sf2a; interestingly, bactericidal responses were serotype-specific, despite our detection of cross-reactivity in ELISA-based assays. This supports continued development of multivalent vaccine approaches.

Our vaccines induced prominent serum IgG OSP and LPS responses, and less prominent IgM OSP responses. Although we did not perform a formal dose ranging study or evaluate the impact of immunoadjuvants, we found equivalent induction of OSP-specific responses following vaccination with 10 μg/dose and 20 μg/dose of Sf2a and Sf3a OSP:rTTHc in our challenge and immunogenicity sub-studies, and excellent protection following vaccination with the lower dose. We have also previously shown comparable immune responses following vaccination dosing of 10, 20, and 50 μg/dose and increased immune responses by including ALUM using a cholera conjugate vaccine made by the same technology [37,40]. In our current study, we did not directly assess mucosal responses, but we did not detect induction of serum IgA responses. Shigella spp. cause an inflammatory enteropathy with direct invasion of intestinal epithelia cells and can result in frank bacillary dysentery with fever and muco-bloody diarrhea. Parenteral vaccination and IgG responses can protect against the tissue invasive phase of enteric pathogens, including Shigella spp and invasive salmonellosis and typhoid fever [81,82]. Such protection may be particularly enhanced among populations with exudation of serum antibody at intestinal surfaces due to a high burden of intestinal tropical/environmental enteropathy and lack of intestinal epithelial integrity [83]; such conditions are common among populations that bear the largest global burden of shigellosis.

Importantly, our vaccines were protective in the challenge models. Shigella infection is a human-restricted process so no animal model is perfect. Non-human primates perhaps are most pertinent [84-86], but are expensive and cumbersome. Sereny assays in guinea pigs, rabbits and mice have also been used [48,87], especially to assess protection at mucosal surfaces; however, initial evaluation often occurs in mice due to their ease of use and availability of immunologic reagents, and mouse challenge assays include intra-pulmonary, keratoconjunctival (Sereny) and intraperitoneal challenge models [48,51,88]. In our study, we performed two separate direct challenge assays in vaccinated mice, and confirmed protection via Sereny and intra-peritoneal challenge, the latter resulting in 100 % protection.

In conclusion, here we report the development of shigella conjugate vaccines that target S. flexneri 2a and S. flexneri 3a, two of the most common global causes of shigellosis. The platform technology that we used can also be used to produce conjugate vaccines that target S. flexneri 6 and S. sonnei. Production and evaluation of such vaccines is ongoing. A cholera conjugate vaccine using this platform approach is entering Phase 1 evaluation in humans [37]. We envision development of a low-cost, scalable, manufacturable, multivalent vaccine able to protect humans from shigellosis caused by S. flexneri 2a, 3a, 6 and S. sonnei. Due to common epitopes, such a quadrivalent vaccine is predicted to provide protection beyond those species and serotypes included in the vaccine [5], and the platform approach means that additional or alternate serotypes of species could be included in the future, if indicated. Populations that bear the largest global burden of shigellosis often lack access to safe water and adequate sanitation and reside in densely-packed informal settlement areas in resource-limited settings. Such communities are often also at risk of other intestinal infections of high consequence, including cholera. Our platform approach, therefore, could also be used to develop a multivalent vaccine against several enteric pathogens including shigellosis and cholera, with target incorporation into the Expanded Programme of Immunizations (EPI) of children residing in such high-risk communities [37]. If successful, such an approach could be quite impactful.

Data availability

Data will be made available on request.