Evaluation of the Immunogenicity and Biological Activity of the Citrobacter freundii Vi-CRM₁₉₇ Conjugate as a Vaccine for Salmonella enterica Serovar Typhi

Abstract

Typhoid fever remains a major health problem in developing countries. Young children are at high risk, and a vaccine effective for this age group is urgently needed. Purified capsular polysaccharide from Salmonella enterica serovar Typhi (Vi) is licensed as a vaccine, providing 50 to 70% protection in individuals older than 5 years. However, this vaccine is ineffective in infants. Vi conjugated to a carrier protein (i.e., an exoprotein A mutant from Pseudomonas aeruginosa [rEPA]) is highly immunogenic, provides long-term protection, and shows more than 90% protective efficacy in children 2 to 5 years old. Here, we describe an alternative glycoconjugate vaccine for S. Typhi, Vi-CRM₁₉₇, where Vi was obtained from Citrobacter freundii WR7011 and CRM₁₉₇, the mutant diphtheria toxin protein, was used as the carrier. We investigated the optimization of growth conditions for Vi production from C. freundii WR7011 and the immunogenicity of Vi-CRM₁₉₇ conjugates in mice. The optimal saccharide/protein ratio of the glycoconjugates was identified for the best antibody production. We also demonstrated the ability of this new vaccine to protect mice against challenge with Vi-positive Salmonella enterica serovar Typhimurium.

Salmonella enterica serovar Typhi (S. Typhi) is the causative agent of typhoid fever, a systemic disease which remains a major public health problem, predominantly in children in developing countries. Estimates of the global burden of typhoid fever range from 17 to 22 million cases per year with 216,000 to 600,000 associated annual deaths (2, 9). S. Typhi expresses a virulence (Vi) capsule encoded in the viaB locus of Salmonella pathogenicity island 7 (SPI7), which allows S. Typhi to modulate host responses during infection by evading innate immune surveillance (25, 26). Vi is also the main target for vaccines and a major protective antigen against typhoid fever. One of the two currently licensed typhoid fever vaccines is unconjugated Vi polysaccharide, available in more than 90 countries (41).

Vi consists of repeating (α1-4)-2-deoxy-2-N-acetyl galacturonic acid moieties and, similar to other polysaccharides consisting of repeating epitopes, is classified as thymus-independent type 2 (Ti-2) antigen (33). These types of polysaccharides provoke an immune response which is age related, with children under 2 years of age generally poor responders. Additionally, Ti-2 antigens lack affinity maturation of antibody response, generate limited immunologic memory and no booster effect, and produce predominantly immunoglobulin isotype IgM with limited class switching (1). The switch from IgM- to IgG-secreting B cells requires interaction with an activated antigen-specific CD4⁺ T helper lymphocyte, which can be engaged by the conjugation of the polysaccharide to a carrier protein. Vi, as a Ti-2 antigen, does not generate immunological memory, and antibodies induced by Vi are not able to be boosted by repeated vaccinations (16, 40).

In contrast, Vi conjugated to a carrier protein provides a valid solution against those major drawbacks. Vi conjugated to tetanus toxoid as a carrier protein (Peda Typh) and to the nontoxic recombinant exotoxin A of Pseudomonas aeruginosa (Vi-rEPA) has been tested in humans. Although no efficacy data are published for Peda Typh, which is licensed only in India (28), Vi-rEPA showed 89% efficacy over 46 months in a double-blind, randomized, and placebo-controlled trial in 2- to 5-year-olds (17). Vi-rEPA has also been reported to be safe when delivered at 2, 4, and 6 months of age. Additionally, a recent Cochrane review (6) reported a protective efficacy of 55% (95% confidence interval [CI], 30 to 70%) at 3 years with one dose of Vi-polysaccharide vaccine and 87% (95% CI, 56 to 96%) at 2.3 years with two doses of Vi-rEPA.

The immunogenicity of the polysaccharide component of a conjugate is affected by different factors, like size and structure of the polysaccharide component, the carrier protein, the conjugation chemistry, the ratio of saccharide to protein, and the eventual presence of free saccharide in the vaccine conjugate (4, 23, 24, 27, 34, 36-38). These factors need to be carefully optimized to produce an effective vaccine.

Here, we describe a new conjugate Vi vaccine, Vi-CRM₁₉₇, where Vi was obtained from Citrobacter freundii strain WR7011. C. freundii produces the same Vi capsular polysaccharide as S. Typhi (3), and being a biosafety level 1 (BSL-1) organism rather than a BSL-3 organism like S. Typhi, its use provides several advantages in terms of safety and manufacturing costs.

C. freundii WR7011 is a nitrosoguanidine-mutated strain derived from the parent strain WR7004. Unlike WR7004, WR7011 stably expresses Vi (21, 32). After optimization of WR7011 growth conditions, Vi was purified and conjugated to the well-characterized diphtheria toxin mutant CRM₁₉₇, which was used as the carrier protein. This 58.4-kDa protein is an approved carrier for licensed childhood conjugate vaccines and has already been shown to be safe and effective in numerous clinical trials (10, 29, 31, 39).

We report antibody responses elicited by Vi-CRM₁₉₇ conjugate at different doses, immunization schedules, Vi/CRM₁₉₇ weight/weight (wt/wt) ratios, and in the presence of unconjugated Vi. Additionally, protection results are presented from a murine immunization challenge model using a virulent Salmonella enterica serovar Typhimurium strain engineered to express the Vi capsule polysaccharide of S. Typhi (8).

MATERIALS AND METHODS

Source of Vi antigens and CRM₁₉₇.

Citrobacter freundii WR7011 was obtained from the U.S. Public Health Service and used as a source of Vi. The Vi polysaccharide was either purified from fermentation supernatant at the Novartis Vaccines Institute for Global Health (NVGH) or generously provided by S. Szu (Laboratory of Developmental and Molecular Immunity, National Institute of Child Health and Human Development, National Institutes of Health [NIH]). The carrier protein, CRM₁₉₇, was obtained from Novartis Vaccines and Diagnostics (NV&D).

Growth conditions for C. freundii WR7011.

C. freundii strain WR7011 was grown in baffled bottom shake flasks with vented caps (VWR) using both complex and chemically defined (CD) media. All cultures were grown at 37°C with constant agitation (200 rpm).

The following three complex media were evaluated: (i) Luria-Bertani (LB; 10 g/liter tryptone, 5 g/liter yeast extract, 10 g/liter NaCl); (ii) modified Luria-Bertani (Mod-LB; 5 g/liter yeast extract [Difco], 10 g/liter ultrafiltered yeast extract [PTK], 8 g/liter NaCl, 0.5% glycerol [not of animal origin]); and (iii) glutamine medium (Glut-medium; 10 g/liter glucose, 1 g/liter glutamine, 10 g/liter yeast extract, 0.07 M Na₂HPO₄, 0.03 M NaH₂PO₄, 0.01 M MgSO₄·7H₂O).

The following chemically defined medium was used with and without amino acid supplements: 13.3 g/liter KH₂PO₄, 4 g/liter (NH₄)₂HPO₄, 1.7 g/liter citric acid monohydrate, 0.05 M MgSO₄·7H₂O, 0.1 M thiamine hydrochloride, and 5 ml/liter PTM4 trace salts. PTM4 trace solution is composed of 2 g/liter CuSO₄·5H₂O, 0.08 g/liter sodium iodide, 3 g/liter MnSO₄·H₂O, 0.2 g/liter Na₂MoO₄·2H₂O, 0.02 g/liter boric acid, 0.5 g/liter CoCl₂·6H₂O, 0.5 g/liter CaSO₄·2H₂O, 7 g/liter ZnCl₂, and 22 g/liter FeSO₄·7H₂O. The salts were dissolved in water containing 1 ml/liter H₂SO₄ and then an additional 0.2 g/ml d-biotin in 2N NH₄OH was added. The carbon source used was either 0.5% glucose or glycerol.

CD medium with 0.5% glucose was also supplemented with different amino acid mixtures to evaluate the relevance of single or groups of amino acids for WR7011 growth. Amino acids were grouped according to Lederberg (16a) on the basis of the biosynthetic pathway leading to other amino acids (Table 1, groups 1 to 5) (15, 18). CD medium was initially supplemented with all amino acid groups but one. For example mix(−1) contained CD medium supplemented with amino acids from groups 2, 3, 4, and 5 but without amino acids from group 1; mix(−2) contained all amino acid groups but 2, etc. Growth of WR7011 in mix(−1), mix(−2), mix(−3), mix(−4), and mix(−5) media was recorded and compared to that of a positive control (CD medium supplemented with all amino acid groups) and a negative control (unsupplemented CD medium).

TABLE 1.

Grouping of amino acids used to supplement CD medium^a

Group	Amino acid	Amt of individual amino acids (mg/ml)	Mix: final OD600 (inoculum OD600 = 0.2)
G1	Lysine	30	Mix(−1): 0.4
	Arginine	10	Mix(−1): 0.4
	Methionine	25	Mix(−1): 0.4
	Cystine HCL	65	Mix(−1): 0.4
G2	Leucine	30	Mix(−2): 0.5
	Isoleucine	30	Mix(−2): 0.5
	Valine	150	Mix(−2): 0.5
	Asparagine	10	Mix(−2): 0.5
G3	Phenylalanine	60	Mix(−3): 0.6
	Tyrosine	5	Mix(−3): 0.6
	Tryptophan	25	Mix(−3): 0.6
	Glutamine	10	Mix(−3): 0.6
G4	Histidine	25	Mix(−4): 0.2
	Threonine	200	Mix(−4): 0.2
	Glutamate	10	Mix(−4): 0.2
	Proline	10	Mix(−4): 0.2
	Aspartate	100	Mix(−4): 0.2
G5	Alanine	10	Mix(−5): 0.4
	Glycine	10	Mix(−5): 0.4
	Serine	400	Mix(−5): 0.4

^aCD medium was supplemented with the following: mix(−1), G2, G3, G4, G5 amino acid mixes; mix(−2), G1, G3, G4, G5 amino acid mixes; mix(−3), G1, G2, G4, G5 amino acid mixes; mix(−4), G1, G2, G3, G5 amino acid mixes; mix(−5), G1, G1, G3, G4 amino acid mixes.

After this initial test to identify the amino acid group of interest (group 4; see Results), CD medium was supplemented with all amino acids except for those belonging to group 4 (for example, −His contained CD medium plus all amino acids but histidine; −Thr contained CD medium plus all amino acids but threonine, etc.). This test determined the most relevant amino acid for WR7011 growth. Finally, a confirmatory test was conducted by complementing CD medium with each of the amino acids belonging to group 4, one at a time. Thus, +His contained CD medium supplemented with histidine, +Thr contained CD medium plus threonine, etc.

C. freundii WR7011 fermentation.

For Vi production, C. freundii WR7011 was grown in a 7-liter bioreactor (EZ-Control, Applikon), containing 4 liters of either Mod-LB or Glut-medium. The bacteria were grown overnight from glycerol stocks in a shake flask containing the same medium to be used for fermentation. Cultures were diluted to obtain an inoculum with an optical density at 600 nm (OD₆₀₀) of between 0.06 and 0.1 (UV-visible spectrophotometer Biomate 3UV/VIS). The culture pH was controlled (AppliSens pH electrode, Applikon) at 7.2 by the automatic addition of 14% NH₄OH. The temperature was maintained at 37°C with a heating blanket and closed-loop water circulation. The air sparging flow rate was set at 1 liter/min of air per liter of culture volume, and the dissolved oxygen (DO) was controlled at ≥30% saturation (calibrated polarographic electrode, oxygen sensor low drift; Applikon). DO was achieved by automatic variation of the agitation rate in the range of 500 to 1,000 rpm (stirrer motor P100; Applikon). When needed, glucose concentrations were measured by the glucose oxidase method using a glucose analyzer (GM8 Micro-Stat analyzer; Analox Instruments).

Approximately 10 h postinoculation, when the cultures reached stationary phase, the fermentation broth was harvested by centrifugation (8,000 × g for 1 h) followed by sterilization via filtration with a 0.2-μm-pore-size filter (Stericup; Millipore). Vi expression by C. freundii WR7011 was checked using anti-Vi polyclonal rabbit antiserum (anti-Salmonella Vi; BioStat Sifin) by slide agglutination. The Vi concentration in culture medium was measured by high-performance anion-exchange chromatography, coupled with pulsed amperometric detection (HPAEC-PAD) (19).

Purification of Vi and synthesis and purification of Vi-CRM₁₉₇ conjugates.

Vi was purified by a process optimized at NVGH that combined precipitation, resolubilization, and filtration steps without the use of phenol (F. Micoli et al., unpublished data). Briefly, the Vi polysaccharide was precipitated from the culture supernatant by the addition of cetyl-trimethyl ammonium bromide, resuspension of the pellet in ethanol, and filtration. The Vi solution in ethanol was then precipitated with sodium chloride (NaCl), and the pellet was resuspended in an aqueous solution of NaCl and filtered. Compared with Vi obtained from NIH, Vi purified with this new protocol presented higher O acetylation levels.

Conjugates were synthesized according to a method based on that already reported by Kossaczka et al. (14) and as detailed by Micoli et al. (19). Briefly, CRM₁₉₇ was derivatized with adipic acid dihydrazine. Then, Vi was activated with 1-ethyl-3(3-dimethylaminopropyl) carbodiimide hydrochloride and linked to the derivatized protein. All conjugates were prepared using the same conditions but with different ratios of Vi/CRM₁₉₇ (wt/wt). Conjugation reaction mixtures containing 1:1, 2:1, and 10:1 (wt/wt) nominal amounts of Vi (both from NVGH and NIH) per CRM₁₉₇ were used to generate conjugates for immunogenicity studies on the influence of different Vi/CRM₁₉₇ ratios. A conjugation reaction mixture containing a 0.5:1 (wt/wt) amount of Vi (NIH) was used to generate the conjugate for dose escalation and challenge studies. The effective Vi/CRM₁₉₇ ratios of the conjugates, obtained after conjugation reaction, are reported in Table 2, as measured by HPAEC-PAD and the Micro BCA protein assay kit.

TABLE 2.

Characteristics of Vi-CRM₁₉₇ conjugates used in immunogenicity studies

Conjugate	Vi manufacturer	% Vi O acetylation by nuclear magnetic resonance	Vi content by HPAEC-PAD (μg/ml)	Protein content by Micro BCA (μg/ml)	Vi/protein ratio experimentally determined (nominal ratio before conjugation)	Immunogenicity studies
Vi (NVGH)-CRM197	NVGH	>90	11.7	13.8	0.9 (1:1)	Influence of different Vi/CRM197 ratios; impact of unconjugated Vi
	NVGH	>90	28.2	13.8	2.1 (2:1)	Influence of different Vi/CRM197 ratios; impact of unconjugated Vi
	NVGH	>90	61.4	6.1	10.1 (10:1)	Influence of different Vi/CRM197 ratios; impact of unconjugated Vi
Vi (NIH)-CRM197	NIH	68	99.6	137.6	0.7 (0.5:1)	Dose escalation; bacterial challenge
	NIH	68	14.3	16.7	0.9 (1:1)	Influence of different Vi/CRM197 ratios and impact of unconjugated Vi
	NIH	68	22.4	11.9	1.9 (2:1)	Influence of different Vi/CRM197 ratios and impact of unconjugated Vi
	NIH	68	32.9	5	6.6 (10:1)	Influence of different Vi/CRM197 ratios and impact of unconjugated Vi

Conjugates were purified by gel filtration chromatography on Sephacryl S1000, and only the first half of the eluting Vi-CRM₁₉₇ peak was collected and used. Under these conditions, the Vi (NVGH) used in this study was well separated from the Vi (NVGH)-CRM₁₉₇ conjugates (data not shown). The Vi conjugates used are described in Table 2.

Analytical methods for Vi and Vi-CRM₁₉₇ characterization.

Complete details of characterization methods are provided by Micoli et al. (19). Vi concentration was estimated by HPAEC-PAD, a methodology developed to release and quantify the saccharide content in Vi preparations, and gave results in agreement with the currently used acridine orange dye binding assay (35). This method allowed for quantification of Vi in complex medium and provided a sensitivity (≥1 μg/ml) higher than other tests, similar to acridine orange quantification. Protein concentration was measured by the Micro BCA kit (Thermo Scientific), using bovine serum albumin (BSA) as the standard reference (30). All purified conjugates were characterized by SDS-PAGE (3 to 8% Tris-acetate gels stained with Coomassie) and by analytical high-performance liquid chromatography-size exclusion chromatography (HPLC-SEC). ¹H nuclear magnetic resonance (NMR) spectra of de-O-acetylated Vi samples in 200 mM NaOD at room temperature were recorded in order to estimate O acetylation levels (12).

Immunization studies.

To evaluate the antigenicity of Vi and Vi-CRM₁₉₇, CD1 female mice of approximately 5 weeks of age were used. Vaccinations were given by subcutaneous injection (200 μl) at 2-week intervals. In all cases, “dose” refers to the amount of Vi injected, not the mass of the conjugate.

Dose response studies.

Two dose escalation studies were performed with six mice per group immunized with increasing doses of Vi-CRM₁₉₇ conjugate (from 0.125 to 16 μg of Vi/injection). Mice were given either one or two injections, and blood was collected at days 0 (preimmune serum), 14, 28, and 42. The Vi for these Vi-CRM₁₉₇ conjugates was obtained from NIH.

Polysaccharide-to-protein ratio study.

Six mice per group were immunized twice at 14-day intervals with increasing amounts of Vi-CRM₁₉₇ (from 0.125 μg to 8 μg Vi/injection) prepared at 1:1, 2:1, and 10:1 (weight/weight) polysaccharide-to-protein ratios. Vi conjugates used in this study were Vi-CRM₁₉₇ with Vi either purified at NVGH or obtained from NIH. Another group of mice was given unconjugated Vi (NVGH) polysaccharide. Blood was obtained at days 0 (preimmune), 14, 28, 42, and 56.

Impact of unconjugated Vi.

Vi-CRM₁₉₇ doses containing 1 μg of Vi were spiked with increasing amounts of unconjugated “free” Vi (from 0 to 50% of total Vi content; i.e., 0 to 1 μg/dose). Ten mice per group were given two injections, and blood was obtained at days 0 (preimmune), 14, 28, 42, and 56.

ELISA method.

Anti-Vi and anti-CRM₁₉₇ antibody levels in mouse sera were determined by enzyme-linked immunosorbent assay (ELISA). The wells of 96-well ELISA plates (Maxisorp; Nunc) were coated with 100 μl of either 1 μg/ml Vi or 2 μg/ml CRM₁₉₇ in carbonate buffer (0.05 M, pH 9.6) and left overnight at 4°C. Vi used for coating was purified from C. freundii WR7011 obtained from NIH; CRM₁₉₇ was supplied by NV&D. The following morning, the plates were blocked for 1 h at room temperature (RT) with 200 μl/well of 5% fat-free milk in phosphate-buffered saline (PBS) containing 0.05% Tween 20 (PBST). After being washed with PBST (ELISA washer ELx405, BioTek), 100 μl/well of mouse serum (at least 1:200 diluted in PBST containing 0.1% BSA) was incubated for 2 h at RT. After three more washes, 100 μl/well alkaline phosphatase-conjugated goat anti-mouse IgG secondary antibody (Sigma A3438, diluted 1:10,000 in PBST, 0.1% BSA) was incubated for 1 h at RT. Alkaline phosphatase substrate (p-nitrophenyl phosphate; Sigma) dissolved in diethanolamine buffer (1 M, pH 9.8) was added after another wash and incubated for 1 h at RT. Color development was read at 405 and 490 nm using an ELISA reader (ELx800; BioTek). Absorbance values for antibody ELISA unit determination were obtained by subtracting OD₄₉₀ values from OD₄₀₅ values. ELISA units are expressed relative to a mouse anti-Vi or anti-CRM₁₉₇ antibody standard serum curve, with the best four-parameter fit determined by a modified Hill plot. One ELISA unit is defined as the reciprocal of the dilution of the standard serum that gives an absorbance value equal to 1 in this assay. Each mouse serum was run in triplicate.

Preparation of bacterial inocula for mouse challenge study.

This study was performed in conjunction with the Wellcome Trust Sanger Institute (Cambridge, United Kingdom). Salmonella enterica serovar Typhimurium C5.507 was derived from S. Typhimurium C5 that has been modified to express Vi polysaccharide (8). S. Typhimurium C5.507 and Vi-negative S. Typhimurium SGB1 strains were grown overnight under static conditions at 37°C in 10 ml of low-salt (5 g per liter or 0.085 M) LB medium to optimize Vi expression. The following morning, 100 μl of bacterial suspension was transferred to 10 ml LB medium and cultured to stationary phase. The cultures were centrifuged and resuspended in PBS to give 10⁴ CFU per dose for infection. The presence or absence of Vi on C5.507/SGB1 was checked using anti-Vi polyclonal rabbit antisera (Remel Europe, Dartford, United Kingdom) by antibody agglutination and fluorescence microscopy.

Mouse immunization and bacterial challenge.

Four groups of 25 BALB/c female mice, 6 to 8 weeks old, were immunized intranasally on days 0 and 8 and subcutaneously on day 22 with 25 μl containing 10 μg of Vi-CRM₁₉₇ with or without 1 μg Escherichia coli heat labile toxin (LT) as the adjuvant, and control groups were immunized with LT alone or PBS. Mice were bled 1 week after each immunization. Five mice from each group were challenged intraperitoneally (IP) on day 61 with 10⁴ CFU of Vi-positive S. Typhimurium (strain C5.507). Five more mice from each group were challenged on day 104 with 10⁴ CFU of C5.507, and another five per group were challenged on the same day with 10⁴ CFU of Vi-negative S. Typhimurium (strain SGB1). The remaining mice were challenged on day 111 with 10⁴ CFU of C5.507. One day after challenge (either at days 62, 105, or 112), mice were sacrificed, cardiac bleeds were performed, and spleens and liver were aseptically removed and homogenized in 5 ml of sterile distilled water in a Colworth stomacher. Viable bacterial counts were determined by plating the homogenate on LB agar and counting the visible colonies.

Statistical analysis.

For the immunization studies, the day 28 sample was considered the definitive bleed for comparison of antibody responses among groups, and the corresponding data were used for statistical analysis. The Spearman rank correlation was used for the two dose response studies (single or double immunization) and for the study addressing the impact of unconjugated Vi. The Kruskal-Wallis analysis of variance (ANOVA) was used to compare groups in the polysaccharide/protein ratio study. Student's t test was used on log-transformed data to compare Vi (NVGH)-CRM₁₉₇ and Vi (NIH)-CRM₁₉₇ conjugates at different ratios and to evaluate data from the challenge study.

RESULTS

Growth conditions for C. freundii WR7011.

C. freundii WR7011 grew well in all complex media, but no growth was observed in the chemically defined medium. Supplementation of CD medium with amino acid mixes resulted in weak growth in shaker flasks (OD₆₀₀ of ∼0.5 after 5 h growth starting from an OD₆₀₀ of 0.2) (Fig. 1). To evaluate the relevance of individual or groups of amino acids, different amino acids were divided into five groups (Table 1) and CD medium was initially supplemented with all but one of the amino acid groups, from mix(−1) to mix(−5). From this experiment, growth was not supported in mix(−4); thus, WR7011 needed one or more amino acids present in group 4 (Table 1) to grow in defined medium.

FIG. 1.

Growth of WR7011 is proline dependent. Shake flasks were inoculated at OD₆₀₀ of 0.2 and incubated at 37°C with constant agitation (200 rpm) for 24 h. Growth was monitored by OD₆₀₀. CD medium was supplemented with histidine, threonine, glutamate, proline, or aspartate. Open symbols correspond to medium supplemented with all amino acids but the one specified, whereas closed symbols correspond to medium supplemented with only the specified amino acid.

A second experiment was designed to identify the most relevant amino acid(s) of group 4. When the medium was deficient in proline (−Pro group), growth of WR7011 was completely impaired (Fig. 1); the exclusion of the other amino acids present in group 4 (histidine, threonine, glutamate, and aspartate) had little impact. The auxotrophic requirement for proline was confirmed by growth of WR7011 in CD medium supplemented with each of the amino acids in group 4, where the highest OD₆₀₀ value was obtained in the proline-containing medium (Fig. 1). Thus, proline was identified as the single most important amino acid required to support the growth of WR7011.

Fermentation of C. freundii WR7011.

The best growth of WR7011 in shaker flasks was obtained using Mod-LB and Glut-medium (OD₆₀₀ of 4.4); therefore, these media were used for fermentations (Table 3). Different carbon sources (e.g., glycerol, glucose, shift of carbon source), nutrient feeds (e.g., yeast extract, glutamine), and fermentation strategies (e.g., batch versus fed batch) were evaluated. The maximum OD₆₀₀ value reached in the bioreactor was 7 to 9 at 10 h postinoculation.

TABLE 3.

Growth of WR7011 under different (5-liter) bioreactor conditions^a

Medium	OD600 (inoculum)	OD600 (final)	Carbon source	Feedb
Mod-LB	0.07	4	Gly	Gly and Glu
	0.01	6	Gly
	0.06	4.5	Glu	Glu and YE
	0.02	8.2	Glu	YE
Glut-medium	0.09	8	Glu
	0.07	7	Glu	Glutamine
	0.13	9	Glu	Glu and glutamine

^aDO, pH, and temperature were controlled similarly in all fermentation runs.

^bGly, glycerol; Glu, glucose; YE, yeast extract.

Glut-medium with glutamine and glucose feed supported the highest growth (OD₆₀₀ of 9; duplication time of 1 h for the first 5 h, entering stationary phase after about 8 h) (Table 3). Higher OD₆₀₀ readings resulted in greater amounts of total Vi released into the culture supernatant. Vi (approximately 4 μg/ml) was detectable by HPAEC-PAD at an OD₆₀₀ of 4, and the level progressively increased, reaching about 100 μg/ml at approximately an OD₆₀₀ of 9. In contrast, when yeast extract was used as feed, a very low Vi yield (about 4 μg/ml) was obtained.

Vi conjugate synthesis and characterization.

At the end of each fermentation, Vi was purified from culture medium, obtaining an overall yield >50%, compared to the Vi amount obtained in the fermentation supernatant. The polysaccharide was thoroughly characterized for appearance, purity, identity, and integrity before performing conjugation. Vi purified at NVGH contained higher acetylation levels than Vi obtained from NIH (>90% compared to 68%).

Conjugation reaction mixtures containing 1:1, 2:1, and 10:1 (wt/wt) nominal amounts of Vi (both NVGH and NIH) per CRM₁₉₇ were used to generate conjugates for immunogenicity studies on the influence of different Vi/CRM₁₉₇ ratios. A conjugation reaction mixture containing 0.5:1 (wt/wt) amount of Vi (NIH) was used to generate the conjugate for dose escalation and challenge studies. Conjugates were purified by gel filtration chromatography on Sephacryl S1000 gel, separating the conjugate by both free Vi and unconjugated CRM₁₉₇. Experimentally determined Vi/CRM₁₉₇ ratios of the purified used conjugates are reported in Table 2, as measured by HPAEC-PAD and the Micro BCA kit.

Dose response studies.

Mice were immunized once (Fig. 2 A and B) or twice (Fig. 2C and D) with increasing doses of Vi (NIH)-CRM₁₉₇ conjugate. Anti-Vi and anti-CRM₁₉₇ antibodies (total IgG) were measured by ELISA. Anti-Vi antibodies from mice receiving a single or a double immunization of Vi-CRM₁₉₇ were detected at the lowest dose tested (0.125 μg) and showed a significant relationship between dose and antibody response following either vaccination regimen (Spearman rank correlation for day 28 samples following one immunization, ρ = 0.5 and P = 0.004; two immunizations, ρ = 0.6 and P = 0.0006). The highest anti-Vi antibody levels were detected in day 28 samples and did not increase further. In contrast, anti-CRM₁₉₇ antibodies increased at each successive time point in both studies (Fig. 2B and D). Anti-CRM₁₉₇ antibody responses were also dose dependent in both studies (Spearman rank correlation for day 28 samples following one immunization, ρ = 0.8 and P = 2.1 × 10⁻⁷; two immunizations, ρ = 0.7 and P = 0.00001).

FIG. 2.

Anti-Vi and anti-CRM₁₉₇ antibodies in a dose escalation study. Data are presented as scatter plots of individual mouse ELISA units, and bars represent the geometric means of each group. CD1 female mice (n = 6 per group) were immunized subcutaneously with PBS or increasing doses of Vi-CRM₁₉₇. Panels A and B show results for mice receiving a single immunization on day 0; panels C and D show results for mice vaccinated twice, on days 0 and 14. Mice were bled prior to the first vaccination and on days 14, 28, and 42. Sera were analyzed by ELISA for anti-Vi (A and C)- and anti-CRM₁₉₇ (B and D)-specific IgGs. Vi-CRM₁₉₇ was prepared using Vi from NIH with a polysaccharide-to-protein ratio of 0.7.

Influence of different Vi/CRM₁₉₇ ratios on antibody responses.

To evaluate the influence of the polysaccharide/carrier ratio, mice were immunized twice with increasing doses of Vi conjugate (from 0.125 μg to 8 μg/injection) prepared at 1:1, 2:1, and 10:1 Vi/CRM₁₉₇ ratios (wt/wt). In this study, Vi (NVGH)-CRM₁₉₇ was also compared with Vi (NIH)-CRM₁₉₇ at 1 μg/dose.

Conjugates containing similar amounts of Vi (NVGH) and CRM₁₉₇ by weight (actual postconjugation ratios of 0.9 and 2.1) induced higher IgG antibody responses than conjugate with postconjugation ratios of 10.1, as shown in Fig. 3A. A significant difference was observed only at a 0.125-μg dose of Vi (NVGH)-CRM₁₉₇, with the 0.9 and 2.1 ratio conjugates being more immunogenic than the 10.1 ratio conjugate (Kruskal-Wallis ANOVA: chi-squared value of 10.9, post hoc analysis P value of 0.004).

FIG. 3.

Anti-Vi and anti-CRM₁₉₇ antibody levels in the sera of mice after two immunizations with Vi-CRM₁₉₇ conjugates at different ratios. CD1 female mice (n = 6 per group) were immunized subcutaneously on days 0 and 14 with Vi (NVGH)-CRM₁₉₇ conjugates with 0.9, 2.1, and 10.1 polysaccharide-to-protein ratios. Mice were bled prior to vaccination and on days 14, 28, 42, and 56 (final bleed). (A) Anti-Vi ELISA units; (B) Anti-CRM₁₉₇ ELISA units.

Anti-CRM₁₉₇ antibody levels also declined with increasing Vi/CRM₁₉₇ ratios (Fig. 3B), with significant differences detected for all Vi (NVGH)-CRM₁₉₇ doses at ratios of 0.9 and 2.1 [and ratios of 0.9 and 1.9 of Vi (NIH)-CRM₁₉₇], compared with a 10.1 conjugate ratio [and 6.6 ratio of Vi (NIH)-CRM₁₉₇] (Kruskal-Wallis ANOVA: chi-squared range, 9.8 to 12.5; post hoc analysis P value range of 0.002 to 0.007).

Similarly, when delivered at a dose of 1 μg, Vi (NIH)-CRM₁₉₇ conjugates (Fig. 4) prepared at 0.9, 1.9, and 6.6 postconjugation ratios induced high IgG antibody responses that were not statistically different. Comparison of immunogenicity of Vi (NVGH)-CRM₁₉₇ and Vi (NIH)-CRM₁₉₇, prepared at different Vi/CRM₁₉₇ ratios and delivered at a 1-μg dose, showed no significant difference for either anti-Vi (Fig. 4) or anti-CRM₁₉₇ (see Fig. S1 in the supplemental material). As previously reported by others (5, 7), unconjugated Vi elicited very weak IgG antibody responses even at a dose of 8 μg/injection (Fig. 4).

FIG. 4.

Anti-Vi antibody levels in the sera of mice after immunization with Vi (NVGH)-CRM₁₉₇ and Vi (NIH)-CRM₁₉₇. CD1 female mice (n = 6 per group) were immunized subcutaneously on days 0 and 14 with 1 μg/dose of Vi (NVGH)-CRM₁₉₇ (polysaccharide-to-protein ratios of 0.9, 2.1, and 10.1) and Vi (NIH)-CRM₁₉₇ (polysaccharide-to-protein ratios of 0.9, 1.9, and 6.6). Mice were bled prior to vaccination and on days 14, 28, 42, and 56 (final bleed). Control groups: mice immunized with unconjugated Vi (8 μg/injection) or with PBS.

Influence of unconjugated Vi on antibody response.

To determine the influence of unconjugated Vi on the generation of antibody response following Vi (NVGH)-CRM₁₉₇ conjugate vaccination, mice were immunized twice with a fixed immunization dose of 1 μg conjugate spiked with increasing amounts of unconjugated “free” Vi. No significant difference in anti-Vi (Fig. 5) or anti-CRM₁₉₇ (data not shown) antibody responses was detected by Spearman rank correlation (P > 0.05).

FIG. 5.

Unconjugated Vi does not interfere with Vi-CRM₁₉₇ antigenicity. CD1 female mice (n = 10 per group) were immunized subcutaneously on days 0 and 14 with 1 μg/dose Vi (NVGH)-CRM₁₉₇ conjugate prepared at a 10.1 polysaccharide-to-protein ratio spiked with increasing amounts of unconjugated Vi (from 0 μg to 1 μg free Vi). Mice were bled prior to vaccination and on days 14, 28, 42, and 56.

Protection elicited by Vi-CRM₁₉₇ immunization on bacterial challenge.

Mice were immunized intranasally on days 0 and 8 followed by subcutaneous immunization on day 22 with 10 μg of Vi (NIH)-CRM₁₉₇, with or without the addition of 1 μg LT, as previously described by Hale et al. (8). Bacterial challenge was performed by IP administration with 10⁴ CFU of virulent Vi-positive S. Typhimurium (strain C5.507) on days 61, 104, and 111 and with 10⁴ CFU of virulent Vi-negative S. Typhimurium (strain SGB1) on day 104 to evaluate the specificity of immune protection. One day after challenge, mice were sacrificed and viable bacterial counts determined in spleens and livers.

The data collected, irrespective of challenge day, indicated that only mice immunized with Vi-CRM₁₉₇ were able to control the growth of the S. Typhimurium Vi-positive strain, in comparison to control mice (Fig. 6, data shown are for day 105, both C5.507 and SGB1 challenge). The addition of LT to Vi-CRM₁₉₇ reduced bacterial colonization, with significant differences in bacterial counts seen in the spleens of the Vi-CRM₁₉₇ plus LT group compared with bacterial counts in the spleens of the PBS group on day 62 (P = 0.027) (see Fig. S2 in the supplemental material) and in liver samples on day 105 (P = 0.037) (Fig. 6C). On day 112, significantly fewer Vi-positive S. Typhimurium bacterial CFU were counted in both spleens and livers of mice after either Vi-CRM₁₉₇ or Vi-CRM₁₉₇ plus LT immunization compared to that in the PBS group (P values of <0.03) (see Fig. S2 in the supplemental material). S. Typhimurium Vi-negative bacteria infected all groups of mice, regardless of the recipient's vaccination or immune status. The protective effect seen was Vi specific, as no reduction in bacterial counts was recorded with Vi-negative S. Typhimurium (Fig. 6B and D).

FIG. 6.

Protection elicited by Vi-CRM₁₉₇ immunization following bacterial challenge. One hundred BALB/c female mice were divided into four groups and immunized on days 0, 8 (intranasally), and 22 (subcutaneously). Mice were immunized as follows: group 1, 10 μg Vi-CRM₁₉₇ plus 1 μg LT; group 2, 10 μg Vi-CRM₁₉₇; group 3, 1 μg LT; and group 4, PBS. Mice were challenged on days 61, 104, or 111 with Vi-positive S. Typhimurium (strain C5.507) and day 104 with Vi-negative S. Typhimurium (strain SGB1). Spleen and liver bacterial CFU of mice challenged with C5.507 were counted on days 62 (see Fig. S2 in the supplemental material), 105 (A and C), or 112 (see Fig. S2 in the supplemental material). Spleen/liver bacterial CFU of mice challenged with SGB1 were counted on day 105 (B and D). Vi-CRM₁₉₇ was prepared using Vi from NIH with a polysaccharide-to-protein ratio of 0.7.

Total anti-Vi IgG levels, measured in sera of mice bled at day 112 (1 day after challenge with Vi-positive S. Typhimurium), were detected in immunized mice and not in control groups as expected. Analysis of anti-Vi IgG subclasses indicated that IgG1 subclass contributed to the majority of total IgG responses. No significant IgG2 levels above the background were detected (P > 0.05) (see Fig. S3 in the supplemental material). To confirm that strain C5.507 recovered from immunized mice still expressed Vi antigen, colonies were screened by anti-Vi antibody immunoblotting. All bacteria recovered from tissues of mice challenged with Vi-positive S. Typhimurium expressed Vi (data not shown).

DISCUSSION

The Vi-CRM₁₉₇ vaccine described here is one of several new Vi conjugate vaccines with the potential of preventing typhoid fever, especially in young children. S. Typhi infection is a serious public health problem, and the increasing transmission of multidrug-resistant strains has led to untreatable typhoid fever cases (13, 20, 22, 41, 43). Although the oral attenuated Ty21a and the Vi polysaccharide vaccines are currently licensed, they have not been implemented as a routine public health measure in most countries where typhoid fever is endemic. This is despite low vaccine costs (a dose of Vi polysaccharide vaccine is $0.57 [42]), compared with the elevated costs of medical treatment for typhoid fever (World Health Organization [WHO] estimates that the total cost per case of typhoid fever requiring hospitalization ranged from $129 to $820 in India and approximately $334 in the rest of Asia [43]). A major drawback of the licensed vaccines, which may have also impacted their introduction, is their requirement for repeated booster vaccination, modest efficacy (∼70%), and ineffectiveness in young children and infants, who are highly susceptible to the disease.

Conjugation of the protective antigen Vi to a carrier protein is an effective method to convert a thymus-independent antigen into a thymus-dependent antigen, with the consequence of eliciting an immune response in infants as well as developing high levels of long-lasting antibody. CRM₁₉₇ is a well-characterized diphtheria toxin mutant with an excellent safety and effectiveness profile, as shown by its use in licensed childhood conjugate vaccines. Thus, CRM₁₉₇ was selected by NVGH as the preferred carrier protein and used for conjugation to Vi polysaccharide.

The source of Vi used in these studies was obtained from C. freundii WR7011 and is identical to Vi from S. Typhi (3). Fermentation and handling of a biosafety level 1 (BSL-1) organism, like C. freundii, instead of S. Typhi, a BSL-3 organism, poses several advantages in terms of safety and manufacturing costs. The strain chosen, WR7011, has been mutagenized to constitutively express high Vi capsule levels, but the chemical mutagenesis may have altered other metabolic pathways, making it more difficult to grow to high cell density either at a small scale or in bioreactors.

Although Citrobacter isolates, including the WR7011 parent strain (WR7004), generally grow in defined media (8) (our unpublished observation), WR7011 cannot, probably as a result of alterations in metabolic pathways. Growth in defined synthetic media is preferred in biopharmaceutical production processes to reduce costs, ensure reproducible growth, and minimize regulatory concerns; therefore, it is an important asset when choosing an isolate as a potential producing source. Additionally, the quality of material originating from animals is hard to control and could easily have an impact on process performance.

We first observed that growth of WR7011 in CD medium was sustained at low levels by the addition of various amino acids to the medium. After grouping individual amino acids and verifying growth of WR7001, it was possible to identify proline as essential for the growth of the bacterium in such medium. This is probably only one of the several metabolic pathways that may have been damaged through chemical mutagenesis. Although we were unable to reach ODs higher than 9 even in fermentation conditions, Vi production per OD₆₀₀ unit was satisfactory at 12.5 μg/ml/OD₆₀₀ and was substantially higher than that reported by Jang et al. (11) from an optimized fermentation process of S. Typhi Ty2 that resulted in 6.1 μg/ml/OD₆₀₀. The Vi yield obtained from a 1,000-liter fermentor of C. freundii WR7011 followed by a purification process with >50% recovery would correspond to approximately 2 million 25-μg doses. However, further optimization of the Citrobacter fermentation to allow higher final cell densities would make this process commercially more attractive.

Vi obtained from WR7011 was finally conjugated to CRM₁₉₇. Immunogenicity studies and bacterial challenge experiments in mice were performed to evaluate the usefulness of Vi-CRM₁₉₇ as a new vaccine candidate against S. Typhi. Mice immunized once or twice, even with low doses (0.125 μg/dose) of Vi-CRM₁₉₇, responded by generating antibodies to both Vi and CRM₁₉₇. A correlation between Vi dose and specific antibody response was found with both immunization regimens.

Few studies have investigated the influence of the saccharide-to-protein ratio on immunogenicity of the conjugates, although this has a significant impact on manufacturing costs, safety, and immunogenicity. Our data showed that a ratio of Vi to CRM₁₉₇ of 10.1 at low doses (0.125 μg) was suboptimal in antibody production compared to that of conjugates with a Vi-to-CRM₁₉₇ ratio of 0.9 or 2.1. These data support the development of an equal weight ratio of Vi to CRM₁₉₇.

The presence of free Vi in the conjugate preparation is unlikely to impact the immune response. This is in contrast to other polysaccharide conjugate vaccines; in the case of Streptococcus pneumoniae capsular polysaccharide-tetanus toxoid conjugate, the presence of more than 10% free polysaccharide significantly decreased the immunogenicity of the conjugate and led to a persistent state of unresponsiveness (27). To test the possible impact of free Vi, we used mixtures of Vi and Vi-CRM₁₉₇, with a 10.1 ratio of Vi to carrier, because under these conditions the suboptimal antibody response to Vi may lead to a more stringent test. However, we did not observe any interference of free Vi on the level of anti-Vi antibodies even with a proportion of free Vi/Vi-CRM₁₉₇ ratio of 1:1. This finding corroborates previous data from our group where conjugate preparations deliberately produced to contain free Vi were immunologically indistinguishable from purified Vi-CRM₁₉₇ containing only conjugated antigen (19). Additionally, immunization with 8 μg of free Vi failed to induce a response above the background. Thus, there is no evidence that the presence of free Vi is likely to be a problem in Vi-CRM₁₉₇ conjugate vaccines.

To assess whether the immune response generated after Vi-CRM₁₉₇ immunization would be sufficient to protect mice upon bacterial challenge, we utilized a mouse infection model developed to test immunity to Vi. Mice immunized with Vi-CRM₁₉₇ were protected against Vi-positive S. Typhimurium colonization. The immune protection was Vi specific, as no reduction in bacterial counts was recorded in the organs of animals challenged with Vi-negative S. Typhimurium. There are two features of this model that are particularly important. First, this S. Typhimurium has been transfected with the Salmonella pathogenicity island 7 (SPI7) from S. Typhi, thus providing further evidence that the anti-Citrobacter Vi response is relevant for S. Typhi. Second, the BALB/c mice used in these challenge experiments constitute a hypersusceptible model of infection due to impaired macrophage function, and thus this constitutes a stringent challenge (8).

In summary, conjugates based on a novel source of Vi and a commercially available carrier, CRM₁₉₇, are highly immunogenic in the absence of adjuvant in mice. The antibodies produced protected mice against challenge with a Vi-positive S. Typhimurium. The antibody response was similar over a range of Vi-to-CRM₁₉₇ ratios and was not affected by the presence of free Vi. These findings indicate that Citrobacter as a source of Vi and a Vi-CRM₁₉₇ conjugate may be an industrially attractive route to an affordable conjugate vaccine for typhoid fever vaccines, especially in young children in low-income countries.