Protective Efficacy Induced by Recombinant Clostridium difficile Toxin Fragments

Abstract

Clostridium difficile is a spore-forming bacterium that can reside in animals and humans. C. difficile infection causes a variety of clinical symptoms, ranging from diarrhea to fulminant colitis. Disease is mediated by TcdA and TcdB, two large enterotoxins released by C. difficile during colonization of the gut. In this study, we evaluated the ability of recombinant toxin fragments to induce neutralizing antibodies in mice. The protective efficacies of the most promising candidates were then evaluated in a hamster model of disease. While limited protection was observed with some combinations, coadministration of a cell binding domain fragment of TcdA (TcdA-B1) and the glucosyltransferase moiety of TcdB (TcdB-GT) induced systemic IgGs which neutralized both toxins and protected vaccinated animals from death following challenge with two strains of C. difficile. Further characterization revealed that despite high concentrations of toxin in the gut lumens of vaccinated animals during the acute phase of the disease, pathological damage was minimized. Assessment of gut contents revealed the presence of TcdA and TcdB antibodies, suggesting that systemic vaccination with this pair of recombinant polypeptides can limit the disease caused by toxin production during C. difficile infection.

INTRODUCTION

Clostridium difficile is a Gram-positive, sporulating, toxigenic bacterium which colonizes and infects both humans and animals (1, 2). Although asymptomatic carriage occurs in 4 to 20% of the healthy human population (3–8), susceptibility to C. difficile-associated infection (CDI) increases with age, hospitalization, immunodeficiency, and antibiotic treatment. Clinical symptoms range from mild to severe diarrhea with relapsing episodes. Complications include severe pseudomembranous colitis (PMC), toxic megacolon, and sepsis (9, 10).

CDI accounts for 15 to 39% of cases of antibiotic-associated diarrhea (11–13). The disease is particularly prevalent in health care facilities, where use of broad-spectrum antibiotic treatment severely disrupts the resident protective bowel flora, enabling C. difficile to colonize the gut and induce clinical symptoms through the production of two large exotoxins, TcdA and TcdB. TcdA and TcdB have molecular masses of 308 and 270 kDa, respectively (14, 15). Like all members of the large clostridial toxin family, they contain three distinct domains, namely, an N-terminal enzymatic domain consisting of glucosyltransferase (GT) and cysteine protease (CP) moieties, a central translocation (T) domain putatively governing their import into the host cell, and a C-terminal receptor binding domain (RBD) responsible for interaction with the cellular receptors (16).

In 2002, a hypervirulent C. difficile ribotype known as BI/NAP1/027 (or just ribotype 027) emerged in North America (17, 18) and Europe (19, 20). Ribotype 027 strains displayed increased severity and mortality and were frequently associated with large outbreaks (17–20). This change in clinical profile has been linked to alterations in resistance to fluoroquinolones (21), modifications in toxin production (22), sporulation rates (23), and the ability to infect a wider range of patients, including hospitalized children and pregnant women (24, 25).

Management of severe CDI is currently based on the use of vancomycin and metronidazole (26, 27). Recently, the novel antibiotic fixadomicin was reported to be more effective than vancomycin at preventing relapses (28).

Active or passive immunization can represent an alternative for CDI prevention and treatment. Individuals with high levels of serum IgG against TcdA and TcdB are protected from recurrent infection (4, 29–31), and administration of monoclonal antibodies against TcdA and TcdB has been shown to be effective at preventing recurrences (32).

Compared to passive immunization, the development of an effective prophylactic vaccine against C. difficile offers the opportunity to provide long-lasting protection against disease. Preparations of formaldehyde-inactivated toxoid from C. difficile culture supernatants have been shown to be well tolerated and able to induce seroconversion in clinical trials (33). However, long-lasting protection, particularly against TcdB, remains challenging (34). To eliminate the intrinsic risk of contamination or incomplete toxoid inactivation, recombinant polypeptides as potential vaccine candidates have been considered in several studies. In particular, TcdA and TcdB RBDs have been cloned and purified from a variety of hosts (35–39) and then evaluated for the ability to raise protective immunity. Fragments derived from the RBD of TcdA have been shown to induce both systemic and mucosal neutralizing antibodies in animal models (35–39).

Initial reports suggested that only anti-TcdA antibodies were necessary to provide complete protection against C. difficile-associated disease (4, 30). However, the role of TcdB in pathogenesis of CDI and the importance of anti-TcdB immunity in preventing disease and recurrences were recently reevaluated (31, 40, 41). Moreover, the increased frequency of TcdA-negative, TcdB-positive strains responsible for severe clinical symptoms (42–45) provides strong evidence that TcdB is a key factor in C. difficile disease and clearly suggests that neither toxin can be ignored in the development of an effective vaccine. The existence of TcdA-negative, TcdB-positive strains able to cause disease in humans therefore provided the rationale for including TcdB fragments in a vaccine formulation.

Recently, two chimeric recombinant vaccines against C. difficile were proposed. The first combined the RBDs of both toxins in a single polypeptide chain (46), while the second was derived from a full-length TcdB protein in which the original RBD was replaced by the corresponding portion of TcdA (47). Both constructs were protective in vivo, priming further research on the development of recombinant polypeptides as potential vaccine candidates.

In this study, we explored the efficacy of a panel of rationally designed recombinant fragments derived from TcdA and TcdB as immunogens. Our results demonstrate that coadministration of the glucosyltransferase domain of TcdB with a carboxyl-terminal fragment of the TcdA RBD protects hamsters against lethal challenge with C. difficile and significantly reduces clinical signs of infection.

(This work was featured in poster presentations at the 4th International Clostridium difficile Symposium, Bled, Slovenia, 20 to 22 September 2012.)

MATERIALS AND METHODS

Cloning of recombinant fragments and generation of mutants.

All fragments reported in Fig. 1A, except TcdA-GT and TcdA-B1, were cloned into the pET15b+ vector (Novagen) by the polymerase incomplete primer extension (PIPE) method (48). In brief, sequences coding for each fragment were amplified by PCR from the C. difficile 630 genomic DNA, using the primers listed in Table S1 in the supplemental material. PCRs generated mixtures of incomplete extension products; by primer design, short overlapping sequences were introduced at the ends of these incomplete extension mixtures, which allowed complementary strands to anneal and produce hybrid vector-insert combinations. Escherichia coli HK100 cells (49) were then transformed with vector-insert hybrids. Single ampicillin-resistant colonies were selected and checked for the presence of the recombinant plasmid by PCR. Plasmids from positive clones were isolated and subcloned into competent E. coli BL21(DE3) cells (Novagen).

Fig 1.

(A) Design of recombinant TcdA and TcdB fragments of the enzymatic (ED) and binding (B) domains. Attempts to express fragments of the translocation domain (T) resulted in low yields and poor solubility, leading to the exclusion of such fragments from further analysis. (B) SDS-PAGE analysis of purified recombinant fragments. Apparent molecular masses are reported in kilodaltons.

TcdA-GT was expressed in Brevibacillus choshinensis, as described elsewhere (50), while TcdA-B1 (previously referred as 14CDTA) was expressed as described previously (35).

The PIPE method was employed to generate TcdA-GT (Y283A, D285A, and D287A), TcdA-CP (D589A, H655A, and C700A), TcdB-GT (D270A, R273A, Y284A, D286A, and D288A), and TcdB-CP (D587A, H653A, and C698A) mutants with abrogated enzymatic activity.

Expression and purification of recombinant fragments.

A single colony of E. coli BL21(DE3) cells expressing each recombinant fragment was inoculated into LB containing 100 μg/ml ampicillin and grown overnight (ON) at 37°C. The bacterial culture was diluted in fresh medium, and protein expression was induced by addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) to the culture during exponential growth phase. Cultures were then incubated for 4 h at 25°C. Expression of each recombinant fragment was determined by SDS-PAGE using a NuPAGE gel system (Invitrogen).

All recombinant polypeptides were purified by immobilized-metal ion affinity chromatography (IMAC), and buffer exchange was performed by use of a PD-10 desalting column (GE Healthcare) or by dialysis. Protein concentration was determined using the bicinchoninic acid (BCA) assay (Thermo Scientific). Protein purity was checked by SDS-PAGE using NuPAGE 4 to 12% Bis-Tris gels (Invitrogen) followed by Coomassie blue staining. Protein identity was confirmed by dot blotting using sera raised against inactivated TcdA and -B.

Toxin purification.

To provide native TcdA and TcdB toxins, C. difficile strain VPI10463 was inoculated onto BHIS (brain heart infusion supplemented with 5 mg/ml of yeast extract and 0.1% l-cysteine) plates. Colonies were recovered, inoculated into tryptone-yeast extract-mannitol (TYM) medium, and grown ON at 37°C under anaerobic conditions (Whitley MG 500 anaerobic system). Cultures were then diluted 1/100 in fresh medium and incubated for 5 days at 37°C under anaerobic conditions. Samples were centrifuged, and the supernatants were filtered through a 0.22-μm-pore-size filter and concentrated (6×) by tangential-flow filtration.

Concentrated supernatants were fractionated by precipitation with ammonium sulfate at 45% saturation for TcdA enrichment and at 65% saturation for TcdB enrichment. After stirring for 3 h at 0°C, the precipitate was concentrated by centrifugation at 9,000 × g for 30 min at 4°C. The pellet was resuspended in buffer A (50 mM Tris-HCl, pH 7.5, 50 mM NaCl) and then dialyzed at 4°C against two changes of buffer A. TcdA and TcdB were purified from the precipitated proteins by chromatography using two 5-ml HiTrap Q HP columns (GE Healthcare) connected in series. A linear gradient from 50 to 1,000 mM NaCl in buffer A was applied using 30 column volumes (CV) at 2 ml/min. TcdA and TcdB were eluted in distinct peaks around 200 and 500 mM NaCl, respectively, and were pooled separately. The pool containing TcdB was dialyzed against buffer B (20 mM piperazine-HCl, pH 5.0, 50 mM NaCl), and TcdB was further purified at pH 5.0 by chromatography on a 5-ml HiTrap Q HP column equilibrated in buffer B. A segmented gradient from 300 to 600 mM NaCl in buffer B was applied using 15 CV at 2 ml/min to elute nearly homogeneous TcdB. The pool containing TcdA was dialyzed against buffer A and applied to a second chromatography step on a 5-ml HiTrap Q HP column equilibrated in buffer A. A segmented gradient from 50 to 250 mM NaCl in buffer A was applied using 30 CV at 2 ml/min to elute nearly homogeneous TcdA. Fractions were collected and analyzed by SDS-PAGE. The absence of cross contamination in TcdA and TcdB preparations was verified by Western blotting with specific antibodies against each toxin. Purified toxins were dialyzed against 50 mM Tris-HCl, 500 mM NaCl, and 10% glycerol and then stored at −20°C until further use.

Toxin inactivation.

After dialysis against phosphate-buffered saline (PBS), TcdA and -B were inactivated by treatment with formaldehyde under the following conditions: for TcdA, 0.58 μM TcdA, 56 mM lysine, and 10 mM formaldehyde in PBS; and for TcdB, 0.93 μM TcdB, 10 mM lysine, and 3.9 mM formaldehyde in PBS. After 120 h on a rotary shaker at 37°C, samples were dialyzed against 2 changes of a 500-fold volume of PBS twice for 24 h each. Samples were confirmed as being inactivated using the cell-based toxicity assay described below.

Mouse immunization.

CD1 outbred mice (Charles River, Italy) received three intraperitoneal (i.p.) injections with 10 μg of recombinant proteins in PBS, on days 1, 21, and 35. Two groups of eight animals each were immunized with each antigen or combination of antigens. The first group received the polypeptides adsorbed to aluminum hydroxide, while the second group received the antigens mixed with an oil-in-water emulsion of 4.3% squalene oil, 0.5% Tween 80, and 0.5% Span 85 detergents (MF59 adjuvant) (51). Control mice were immunized with adjuvant alone, and serum samples were collected on day 49. Prior to each experiment, the appropriate amount of aluminum hydroxide for complete antigen adsorption was determined. Aluminum hydroxide was titrated against increasing quantities of protein in PBS. After ON incubation at 4°C in a vertical rotator (PTR 25–360; Grant Instruments Ltd., Cambridge, United Kingdom), preparations were centrifuged and supernatants were analyzed for the presence of unbound antigen. For all the polypeptides, the minimum quantity of aluminum hydroxide required for full adsorption was 2 mg/ml, and this dosage was used for all mouse immunizations. The final osmolarity of each formulation was optimized to 0.300 ± 0.06 osmol/kg of body weight by addition of 2 M NaCl and 10 mM histidine, pH 6.5.

Enzyme-linked immunosorbent assay (ELISA).

Microtiter plates (Greiner Bio-One) were coated ON at 4°C with 1-μg/ml purified TcdA or TcdB. Wells were washed three times with PBS plus 0.1% Tween 20 (PBS-T) and blocked with 2.7% polyvinylpyrrolidone (PVP; Serva) for 2 h at 37°C. After three washes with PBS-T, plates were incubated with mouse sera diluted 1:1,000 for 2 h at 37°C, followed by incubation with alkaline phosphatase-conjugated anti-mouse antibodies diluted 1:2,000 in PBS-T plus 1% bovine serum albumin (BSA) for 90 min at 37°C. Samples were incubated with p-nitrophenyl phosphate (SigmaFast OPD; Sigma) at room temperature for 30 min, and the reaction was stopped with 4 N NaOH. Optical density was analyzed using a plate reader at a dual wavelength of 405/620 to 650 nm. Antibody titers were quantified via interpolation against a reference standard curve.

Cytotoxicity and neutralization assays.

IMR-90 human fibroblasts were obtained from the American Type Culture Collection (ATCC; Rockville, MD). Cells were grown to 80 to 90% confluence in 96-well plates in Eagle's minimum essential medium (EMEM) with 10% fetal calf serum. The minimal doses of TcdA and TcdB needed to cause 100% rounding in 24 h (1 CTU₁₀₀) were defined as 20 ng/ml for TcdA and 10 pg/ml for TcdB.

For the neutralization assay, 2-fold dilutions of mouse or hamster sera from 1:16 to 1:32,000 were preincubated with 1 CTU₁₀₀ of each toxin in cell medium for 90 min at 37°C. Serum and toxin mixtures were then added to the cells and incubated for 16 to 18 h before analysis. Preimmune sera and/or sera from mice treated with adjuvant alone were used as negative controls. The endpoint titer was defined as the reciprocal of the highest dilution able to inhibit cell rounding. To check the absence of cytotoxicity, recombinant TcdA-CP, TcdB-CP, TcdA-GT, and TcdB-GT were serially diluted from 40 μg/ml to 20 ng/ml and added to the cells for 24 h at 37°C; cells were observed after 16 to 18 h to evaluate morphological alterations. TcdA and TcdB (1 CTU₁₀₀ each) were used as positive controls.

Spore preparation.

C. difficile strains 630 and B1 were grown in BHI broth under anaerobic conditions for 7 to 10 days. Cultures were then pelleted by centrifugation and vortexed every 10 min for 1 h before centrifugation for 10 min. The pellet was then treated with 1% Sarkosyl in PBS for 1 h at room temperature and again pelleted by centrifugation, followed by incubation ON at 37°C with lysozyme (10 mg/ml) in 125 mM Tris-HCl buffer (pH 8.0). The sample was treated in a sonicating water bath (3 pulses of 3 min each; Branson model 1510 instrument) before centrifugation through a 50% sucrose gradient for 20 min. The pellet was incubated in 2 ml of PBS containing 200 mM EDTA, 300 ng/ml proteinase K, and 1% Sarkosyl for 30 min at 37°C before centrifugation through a 50% sucrose gradient for 20 min. The final pellet was then washed twice in sterile distilled water before finally being resuspended in 1 ml of sterile water. Spore preparations were stored at −80°C prior to use. Serial 10-fold dilutions of the spore preparations were inoculated onto BHI agar plates to determine the number of spores capable of germinating.

Hamster immunization and C. difficile challenge.

Experiments on female Golden Syrian hamsters (Harlan Olac, United Kingdom) were performed in strict accordance with the requirements of the Animals (Scientific Procedures) Act 1986. Prior approval for these procedures was granted by the University of Glasgow Ethical Review Panel and by the UK Home Office (project license 60/4218).

Three weeks prior to the start of vaccination, telemetry chips (Vitalview Emitter) were inserted by laparotomy into the body cavities of all animals. These chips allowed continuous monitoring of the body temperature and activity of each animal. Hamsters were immunized i.p. at days 0, 14, 28, and 42 with 50 or 20 μg of each antigen in PBS mixed at a 1:1 volume ratio with the MF59 adjuvant, diluted to a final volume of 200 μl/dose with PBS. Control hamsters were immunized with adjuvant alone or PBS. Three weeks after the last immunization, animals received 30 mg/kg of clindamycin phosphate orogastrically and were challenged the following day with 1,000 spores of C. difficile strain B1 or 630 administered by oral gavage (52). Animals were monitored for symptoms of infection, including onset and duration of loose stools (wet tail); hamsters showing a drop in temperature of more than two degrees (35°C) or losing more than 10% of their body mass were culled (52). Animals surviving for more than 2 weeks postchallenge were culled to provide an endpoint to the experiment. Sera were collected before culling.

To exclude the interference of infection with the antitoxin immune response, the neutralization activity of serum collected before challenge from a hamster vaccinated with the combination of TcdA-B2, TcdA-GT, TcdB-B3, and TcdB-GT (TcdA-B2+TcdA-GT+TcdB-B3+TcdB-GT) was also analyzed.

Measurement of toxins in gut samples.

Production of C. difficile toxins was detected in vitro by using filtered cecum contents taken postmortem as described previously (53). Briefly, monolayers of Vero cells (kidney epithelial cells) or HT29 cells were washed with preheated sterile PBS before addition of serially diluted filtered gut content in supplemented EMEM and then were incubated for 18 h at 37°C with 5% CO₂. Cells were washed with PBS, fixed in 1% formalin for 10 min, and then washed again. Fixed adherent cells were stained with Giemsa stain for 30 min, washed before addition of 0.1% SDS, and left to stand for 1 h. The optical density at 600 nm was taken using an EL_x808 Ultra microplate reader (Bio-Tek Instruments) and compared with those of noninfected hamster cecum and colon gut contents as a negative control. If the toxin dilution was able to cause cell toxicity (cell rounding), loss of cell adherence was observed, resulting in reduced staining and a reduced optical density of the wells. Results were expressed as log reciprocal titers (53). Toxin levels were analyzed using the nonparametric two-tailed Mann-Whitney test.

Dot blot analysis.

Twofold dilutions of native TcdA and TcdB, ranging from 100 down to 3.12 ng, were spotted on a nitrocellulose membrane. After blocking in 10% milk in PBS with 0.05% Tween 20, membranes were incubated with hamster gut washes (see above) diluted 1:100 and then incubated with horseradish peroxidase (HRP)-conjugated secondary anti-hamster antibodies. Spots were visualized with a Super Signal West Pico chemiluminescent substrate kit (Pierce).

Histology.

Cecum samples were prepared for simple histology as previously described (52). Cecum pathology was scored in a blinded fashion, grading neutrophil accumulation (0, no neutrophil accumulation; 1, local acute neutrophil accumulation; 2, extensive submucosal neutrophil accumulation; and 3, transmural neutrophilic infiltrate), hemorrhagic congestion (0, normal tissue; 1, engorged mucosal capillaries; 2, submucosal congestion with unclotted blood; and 3, transmural congestion with unclotted blood), hyperplasia (0, no epithelial hyperplasia; 1, 2-fold increase in thickness; 2, 3-fold increase in thickness; and 3, 4-fold or greater increase in thickness), and percentage of epithelial barrier involvement (0, no damage; 1, less than 10% of mucosal barrier involved; 2, less than 50% of mucosal barrier involved; and 3, more than 50% of mucosal barrier involved). Results are expressed as the mean pathology score per strain for each criterion and are reported in Fig. S5 in the supplemental material.

RESULTS

Design and expression of recombinant fragments.

The presence of epitopes inducing neutralizing antibodies was explored along the TcdA and TcdB sequences by expressing the recombinant fragments summarized in Fig. 1A.

Nucleic acid sequences of TcdA and TcdB from C. difficile strain 630 were used as templates for subcloning. As suggested by the tripartite organization of both toxins, the glucosyltransferase (GT) and cysteine protease (CP) regions were subcloned. Moreover, the RBDs were subdivided into several fragments.

Fragment design was assisted by computer modeling. The TcdA and TcdB binding domains both consist of a succession of short and long sequence repeats (54). The X-ray structure of the TcdA C-terminal fragment (residues 2208 to 2710) revealed that the repetitive sequence folds into a repetitive solenoid-like structure (54), and this provided the rationale to model the entire cell binding domains of TcdA (TcdA-B) and TcdB (TcdB-B) by using the available atomic coordinates as a template. To preserve the original fold, the TcdA-B and TcdB-B portions were then designed by avoiding interruptions of the structural repeats predicted by computer modeling (see Fig. S1 in the supplemental material). TcdA-B was subdivided into three different fragments that were progressively shortened at the N terminus (called TcdA-B1, TcdA-B2, and TcdA-B3). One additional fragment, called TcdA-B4, corresponded to a region near the N terminus of TcdA-B3. TcdB-B was subdivided into fragments TcdB-B1, TcdB-B2, and TcdB-B3, which overlapped at the carboxyl terminus. Recombinant GT and CP portions of the N-terminal domains were mutated in their catalytic active sites to remove the enzymatic activity. Abrogation of toxicity was assessed by an in vitro assay on human IMR-90 cells (see Fig. S2). Electrophoretic profiles of the proteins used for immunization were evaluated by Coomassie staining of an SDS-PAGE gel (Fig. 1B).

Immunogenicity.

Systemic antibody responses to toxin fragments were analyzed by measuring antigen-specific IgG titers in the sera of immunized mice (see Fig. S3 in the supplemental material). RBD fragments, with the exception of TcdA-B4, were highly immunogenic. Lower titers were detected against the enzymatic portions, although a significant IgG response was induced by TcdB-GT. IgG titers were comparable and were independent of the adjuvant formulation. Sera against TcdA fragments showed very low IgG titers against TcdB and vice versa, indicating that the immune response against each toxin fragment was poorly cross-reactive.

In vitro toxin neutralization.

To verify the functionality of the antibodies, the capacity to neutralize the cytotoxicity of TcdA and TcdB was investigated in vitro (Table 1). All antibodies against the RBD fragments neutralized TcdA, except those elicited by TcdA-B4. Remarkably, antibodies directed to TcdA-GT had neutralizing activity despite the low overall IgG titer (as measured by ELISA against TcdA). Compared with that of TcdA, TcdB cytotoxicity was inhibited by higher serum concentrations, with neutralizing titers elicited only by the longer TcdB-B3 fragments and TcdB-GT. Overall, no correlation was observed between ELISA titers and the toxin neutralization capability of single fragments; moreover, none of the fragments were able to neutralize both toxins. Mice were therefore immunized with combinations of fragments in an attempt to achieve concurrent neutralization of TcdA and TcdB. Such an approach resulted in the production of sera able to neutralize both toxins in vitro. An example of such cross-neutralization is reported in Fig. 2.

Table 1.

Neutralization titers of sera from mice immunized with toxin A and B fragments^a

Antigen(s)	Anti-TcdA neutralization titer		Anti-TcdB neutralization titer
	Al(OH)3	MF59	Al(OH)3	MF59
TcdA-B1	2,102 ± 555	2,667 ± 666	16 ± 0	16 ± 0
TcdA-B2	11,200 ± 1,960	6,400 ± 979	0	0
TcdA-B3	4,400 ± 976	2,800 ± 489	0	0
TcdA-B4	0	ND	0	ND
TcdA-CP	0	0	0	0
TcdA-GT	1,333 ± 333	208 ± 48	0	0
TcdB-B1	0	0	0	0
TcdB-B2	0	0	0	0
TcdB-B3	0	0	320 ± 64	213 ± 42
TcdB-CP	0	0	0	0
TcdB-GT	0	0	298 ± 112	224 ± 32
TcdA-B1+TcdB-B3	8,000 ± 0	5,500 ± 1,500	256 ± 0	298 ± 112
TcdA-B1+TcdB-GT	7,000 ± 1,000	4,667 ± 1,764	160 ± 32	171 ± 43
TcdA-B2+TcdB-B3	4,000 ± 1,095	4,667 ± 1,764	256 ± 0	256 ± 0
TcdA-B2+TcdB-GT	7,333 ± 1,909	8,000 ± 2,828	170 ± 43	256 ± 0
TcdA-B3+TcdB-B3	4,667 ± 1,764	4,667 ± 1,764	170 ± 42	128 ± 0
TcdA-B3+TcdB-GT	5,333 ± 1,333	2,000 ± 1,000	149 ± 56	170 ± 43
TcdA-B1/TcdB-GT chimerab	3,500 ± 500	4,000 ± 0	0	0

^aThe titer was defined as the reciprocal of the highest dilution able to inhibit 100% rounding in IMR-90 human fibroblasts treated with 1 CTU₁₀₀ of toxin A or B. Values represent the geometric means for 3 to 5 independent experiments ± standard errors (SE). Combinations selected for testing in the hamster animal model are indicated in bold. ND, not determined.

^bTcdA-B1 and TcdB-GT were also combined in a single chimeric polypeptide (TcdA-B1/TcdB-GT).

Fig 2.

In vitro neutralization assay on IMR-90 human fibroblasts. Images show the morphology of control cells (A) and cells incubated with 1 CTU₁₀₀ of TcdA (C) or TcdB (D). (E and F) Cell rounding was inhibited by serum against TcdA-B1+TcdB-GT. (E) TcdA (1 CTU₁₀₀) plus 1/8,000-diluted serum; (F) TcdB (1 CTU₁₀₀) plus 1/256-diluted serum. The negative control was the corresponding preimmune serum in the presence of 1 CTU₁₀₀ of TcdA (B) or TcdB (not shown). The complete panel of images showing the titration of neutralizing activities is reported in Fig. S4 in the supplemental material.

In general, neutralizing titers indicated that the components performed with comparable efficiencies when used individually or in combination (Table 1).

TcdA-B1 and TcdB-GT fragments were also combined in a single chimeric polypeptide (TcdA-B1/TcdB-GT). This protein induced antibodies with efficient neutralizing activity against TcdA but not against TcdB (Table 1), suggesting that the neutralizing epitopes of TcdB-GT were compromised in the chimeric protein. Overall, this screening indicated that while a limited portion of the TcdA RBD was sufficient to neutralize TcdA, the entire TcdB domain was necessary to induce functional antibodies against the cognate toxin. It also highlighted the ability of TcdB-GT to elicit neutralizing antibodies against TcdB. The fragment combinations TcdA-B1+TcdB-B3, TcdA-B1+TcdB-GT, and TcdA-B2+TcdB-GT emerged as the most promising and were subsequently tested for efficacy in the hamster model of disease.

Protective efficacy of toxin fragments.

The protective efficacies of fragments in a hamster model of infection are summarized in Table 2. Hamsters were vaccinated i.p. with four doses of antigens. For these studies, we decided to include MF59 as an adjuvant because of its well-established ability to induce effective long-term protection in the elderly, which is the estimated target population for a C. difficile vaccine (55, 56).

Table 2.

Protection against C. difficile disease in hamsters vaccinated with recombinant TcdA and TcdB fragments with MF59 adjuvant^a

Challenge strain and vaccine antigen(s)	% Survival (no. of survivors/total no. of animals)	Duration of diarrhea	Anti-TcdA neutralization titer	Anti-TcdB neutralization titer
Strain 630
TcdA-B1+TcdB-B3	100 (6/6)	5 h 55 min ± 3 h 25 min	4,667 ± 1,764	558 ± 121
TcdA-B1+TcdB-GT	100 (6/6)	None	8,000 ± 0	512 ± 0
Tcd-B1/TcdB-GT chimera	60 (3/5)	16 h 57 min ± 7 h 39 min	1,667 ± 333 (32 ± 0)	512 ± 0 (16 ± 0)
Adjuvant alone	0 (0/7)	4 h 5 min ± 0 h 44 minc
Naive animals	0 (0/7)	3 h 37 min ± 0 h 33 minc
Strain B1
TcdA-B1	0 (0/6)	NDc	ND	ND
TcdB-B3	0 (0/6)	NDc	0	426 ± 85
TcdA-B1+TcdB-B3	83 (5/6)	24 h 46 min ± 7 h 43 min	2,000 ± 0 (32 ± 0)	512 ± 0 (16 ± 0)
TcdA-B1+TcdB-GT	100 (6/6)	7 h 53 min ± 2 h 34 min	1,200 ± 5,060	341 ± 85
TcdA-B2+TcdB-GT	100 (6/6)	12 h 19 min ± 3 h 4 min	8,000 ± 0	256 ± 0
TcdA-B1+TcdB-B3+TcdB-GT	83 (5/6)	8 h 8 min ± 4 h 59 min	1,000 ± 0 (32 ± 0)	2,000 ± 0 (32 ± 0)
TcdA-B2+TcdA-GT+TcdB-B3+TcdB-GT	83 (5/6)	6 h 1 min ± 1 h 28 min	6,667 ± 1,333 (512 ± 0)	512 ± 0 (0)
Adjuvant alone	0 (0/10)	3 h 2 min ± 0 h 22 minc
Naive animals	0 (0/10)	2 h 34 min ± 0 h 17 minc
Strain B1 with reduced antigen dose (20 μg/dose)
TcdA-B1+TcdB-GT	100 (8/8)	7 h 25 min ± 2 h 6 min	10,000 ± 1,852	112 ± 10
TcdA-B2+TcdB-GT	37 (3/8)	9 h 39 min ± 4 h 44 min	2,133 ± 5,333 (8,000 ± 0)	128 ± 0 (0)
TcdA-B2+TcdA-GT+TcdB-B3+TcdB-GTb	86 (6/7)	15 h 15 min ± 2 h 3 min	8,000 ± 0 (512 ± 0)	256 ± 0 (0)

^aNeutralization titers were determined with pooled sera from protected animals at the experimental endpoint. Values represent geometric means for 3 to 5 independent experiments ± SE. Titers in parentheses refer to single or pooled sera from unprotected animals. ND, not determined.

^bSample sera from vaccinated animals were collected before the challenge and showed neutralization titers comparable to those measured at the experimental endpoint (data not shown).

^cDiarrhea was observed until the body temperature dropped below 35°C.

Three weeks after the last vaccination, animals were treated with clindamycin to disrupt their intestinal commensal flora and were challenged with C. difficile spores given by oral gavage. Vaccination efficacy was tested using strains 630 and B1, which show significant differences in severity of disease within this model (52). Survival and clinical signs were evaluated for 14 days postchallenge to provide an indication of vaccine efficacy. Control groups included naive animals and animals treated with adjuvant only, which were not protected following challenge with both strains of C. difficile.

The relevance of the in vitro neutralization data to protection was confirmed in vivo, as hamsters immunized with the individual TcdA-B1 or TcdB-B3 fragment succumbed to infection after the challenge, indicating that immunity to both toxins was necessary for protection. However, when hamsters were immunized with both TcdA-B1 and TcdB-B3, protection appeared to correlate with disease severity for the different strains: full protection was observed after challenge with 630, a strain known to cause less severe pathology, while the same combination ensured an 83% survival rate against strain B1. In contrast, vaccination with TcdA-B1+TcdB-GT conferred 100% protection against both strains, even when the vaccine dosage was reduced to 20 μg/dose. Combinations of three and four toxin fragments, including both binding and glucosyltransferase domains, did not increase the level of protection induced by TcdA-B1+TcdB-GT (Table 2). Besides preventing fatal infection, vaccination significantly reduced the severity of disease. All control animals infected with strain 630 or B1 showed the typical acute response with occurrence of diarrhea followed by death, while surviving animals developed mild diarrhea or no signs of disease followed by full recovery. Remarkably, after challenge with strain 630, total inhibition of diarrhea was observed in animals vaccinated with TcdA-B1+TcdB-GT.

Vaccinated and surviving animals showed high levels of toxin neutralizing antibodies in the sera 14 days after infection. Conversely, sera from hamsters which did not survive the challenge showed very low or no neutralization titers, suggesting that toxin neutralization correlated closely with protection (Table 2). The presence of toxin-specific IgGs in the intestinal lumens of animals vaccinated with TcdA-B1+TcdB-GT was also investigated. Although the response to TcdA was higher than that to TcdB in the acute phase of infection, rising amounts of anti-TcdB antibodies were detectable at the endpoint (Fig. 3).

Fig 3.

IgG antibodies against toxin A (A) and toxin B (B) in cecum samples from hamsters vaccinated with TcdA-B1+TcdB-GT. Twofold dilutions of TcdA and TcdB (100 ng to 3 ng) were serially spotted onto a nitrocellulose membrane. Dot blots were then performed with filtered cecum samples taken from vaccinated animals in the acute phase of infection (48 h postchallenge) (hamsters 1 and 2) and at the experimental endpoint (14 days postchallenge) (hamsters 3 to 8). Control animals were treated with adjuvant only and infected under the same experimental conditions (hamsters 9 and 10).

Postvaccination analysis.

To further evaluate the effects of vaccination with TcdA-B1+TcdB-GT, toxin levels produced during infection and gut histology were examined both in the acute phase of infection (48 h postchallenge) and at recovery (14 days postchallenge). Equivalent levels of TcdA and TcdB were detectable in both control and vaccinated hamsters at 48 h postchallenge (Fig. 4). However, while severe gut inflammation accompanied by epithelial necrosis and polymorphonuclear leukocyte (PMN) influx was observed in unvaccinated animals (green and black arrows in Fig. 5B), the tissue from vaccinated hamsters showed less epithelial damage and a limited amount of PMN infiltrate (green and black arrows in Fig. 5C). Hyperplasia, associated with the appearance of mucin-producing cells, and crypt-to-tip length increases were observed (red arrows in Fig. 5B and C), particularly in the lower colon. Protected animals showed significantly lower levels of toxin within the intestinal lumen 14 days after infection, despite the presence of large numbers of C. difficile colonies associated with the intestinal tissue (data not shown). The gut epithelia appeared to revert to normality, with an absence of polymorph influx (Fig. 5D). Interestingly, while no hyperplasia of the cecum was evident, it persisted in the terminal colon (red arrows in Fig. 5D).

Fig 4.

Toxin A and B levels in hamsters vaccinated with the TcdA-B1+TcdB-GT combination. Values indicate the fold dilutions required to eliminate cell rounding. Filtered cecum samples were taken from vaccinated animals in the acute phase of infection (■; 48 h postchallenge) and at the experimental endpoint (▲; 14 days postchallenge). Toxin levels in the cecum samples of animals at day 14 were significantly lower (**, P = 0.0014 for toxin A and P = 0.0015 for toxin B) than those measured in control animals (acute disease) or vaccinated animals at 48 h. Control animals (●) were treated with adjuvant only and infected under the same experimental conditions. Toxin levels were evaluated at the acute end stage of the infection.

Fig 5.

Histopathology of hamster gut tissues following immunization with TcdA-B1+TcdB-GT and challenge with C. difficile B1. Samples were taken from control hamsters (A), unvaccinated hamsters during the acute phase of infection (B), and vaccinated hamsters at 48 h (C) and 14 days (D) postchallenge. Arrows indicate an influx of PMNs (black), hyperplasia (red), and epithelial disruption (green). Histopathological scores are reported in Fig. S5 in the supplemental material.

Overall, postvaccination analyses indicated that the presence of antibodies neutralizing both toxins strongly limited gut epithelial damage and mediated recovery from disease.

DISCUSSION

There is well-established evidence that protection against severe CDI is mediated by systemic antibodies to TcdA and TcdB (10, 34). In this study, we evaluated the ability of recombinant toxin fragments to induce robust immunity against lethal challenge with C. difficile. The use of recombinant proteins is an attractive strategy for vaccine design, as antigen fragments can be engineered to meet all the quality standards required during large-scale production, excluding issues of incomplete inactivation and destruction of conformational epitopes associated with the use of formaldehyde-detoxified toxoid.

Initial screening in vitro revealed that immunization of mice with TcdB fragments induced relatively low neutralization titers. The limited activity of the anti-TcdB antibodies could be explained in part by the difference in cytotoxicity of TcdB, which has been reported to be 1,000 times more potent in vitro than TcdA (14). A second striking difference between the toxins was the localization of the protective epitopes. While the RBD domain of TcdA was clearly predominant in inducing neutralizing antibodies to this toxin, both TcdB-B3 and TcdB-GT induced antibodies in mice that were comparably efficient at neutralizing TcdB. This observation was subsequently confirmed in hamster experiments. All animals immunized with the TcdA-B1+TcdB-GT and TcdA-B1+TcdB-B3 combinations survived challenge with C. difficile strain 630. Unlike the control animals, hamsters immunized with TcdA-B1+TcdB-B3 were protected from death, although most suffered a single episode of self-limiting diarrhea. In contrast, animals immunized with TcdA-B1+TcdB-GT and challenged with strain 630 did not develop diarrhea, suggesting that this combination provided enhanced protection. The hypothesis was confirmed by the observation that only vaccination with TcdA-B1+TcdB-GT fully protected all animals after challenge with the B1 strain, with the vaccine remaining fully protective even when the formulation was reduced to 20 μg/dose. Remarkably, the addition of other components to the TcdA-B1+TcdB-GT combination had a detrimental effect on animal survival. This was particularly evident for formulations with lower antigen concentrations, where the addition of TcdA-GT and TcdB-B3 was unable to eliminate clinical signs or enhance the neutralization titers of immune sera. TcdA-B1+TcdB-GT was therefore the minimal combination necessary and sufficient to ensure the survival of all vaccinated hamsters and induce neutralizing antibody titers able to prevent the onset of disease. This evidence opens new perspectives on the use of recombinant antigens to vaccinate against CDI. It is widely accepted that C. difficile vaccines should be based on the generation of antibodies interfering with the initial binding and internalization of TcdA and TcdB within the host cells. For this reason, an immune response toward protective epitopes of both toxins is of fundamental importance.

Historically, the first evidence that antitoxin immunity can protect hamsters against lethal challenge was obtained by immunizing animals with formalin-inactivated toxoid from culture filtrates (57–59). Significantly, Torres and colleagues compared the effectiveness of toxoid preparations at different antigen doses and immunization routes (60). They showed that a combination of mucosal and systemic immunization induced protection against both the diarrheal symptoms of infection and lethality in hamsters. The combined use of mucosal and parenteral administration of both toxoids was further investigated by Giannasca et al., who used more-purified toxoid preparations and replaced the i.p. route with the clinically acceptable intramuscular injection route (61). Full protection was achieved only in animals vaccinated with a combination of rectal immunization with E. coli heat-labile toxin adjuvant and intramuscular injection of alum-adjuvanted toxoids. The absence of antitoxin antibodies in feces suggested that circulating antibodies were responsible for protection of vaccinated hamsters. Subsequent studies investigated the efficacy of several recombinant toxin fragments by systemic vaccination. Vaccine candidates included the entire RBD from TcdA (62), smaller TcdA internal fragments (35), and, more recently, chimeric proteins that incorporate regions from both toxins into a single polypeptide chain (46, 47). Such fusion proteins have been tested with alumimun hydroxide at doses ranging from 10 μg (47) up to 100 μg (46). Full protection from death has been observed only at higher antigen doses after challenge with the less virulent strain 630.

Overall, the TcdA-B1+TcdB-GT combination reported in the present study appears to be a promising vaccine candidate, as four doses of 20 μg were able to protect hamsters against severe infection with strain B1. Whether a vaccine containing such a combination could potentially be optimized by reducing the total number of injections is an aspect that requires further investigation.

Our results clearly indicate that the protective epitopes of TcdB are not exclusively localized in the RBD and emphasize the importance of carrying out an initial screening to identify the most promising vaccine candidates. These findings align with previous studies showing that TcdB neutralizing epitopes are located within the N-terminal GT domain (47). The importance of the TcdB-GT domain was also revealed by epitope mapping of humanized monoclonal antibodies for passive immunotherapy (32).

The absence of a clear correlation between immunogenicity and protective efficacy suggests a biased immune response toward immune-dominant nonneutralizing epitopes. From a vaccine perspective, the dissection of the toxin polypeptides into recombinant fragments provides the double advantage of reducing antigen size and enhancing the population of neutralizing antibodies.

Our analysis in vivo revealed that antibody-mediated toxin neutralization is effective at the level of the epithelial barrier. Indeed, toxin levels in the gut lumens of vaccinated animals recovering from diarrhea 48 h after challenge and those found in control animals were equivalent. Therefore, the reduction in diarrheal episodes observed in protected animals could be associated with early toxin production, which causes limited damage to the vasculature. The resulting lesion allows permeation of the immune serum containing neutralizing antibodies. The protective effect of such infiltration was appreciable from inspection of gut tissues of vaccinated animals, as initial damage and neutrophil infiltration were limited and were followed by tissue repair.

It remains to be determined whether the higher protection induced in hamsters by combinations containing TcdB-GT is due to the ability of specific antibodies to limit cell binding and uptake of TcdB or whether the direct neutralization of the enzymatic domain is able to protect against the effects of still-uncharacterized extracellular enzymatic activity of the toxin. Whichever is the case, TcdA-B1+TcdB-GT appears to generate significant protection, providing new hope for reducing the impact of CDI.