A Chimeric Toxin Vaccine Protects against Primary and RecurrentClostridium difficile Infection

Haiying Wang,a Xingmin Sun,b Yongrong Zhang,c Shan Li,a,d Kevin Chen,d Lianfa Shi,d Weijia Nie,b Raj Kumar,e Saul Tzipori,b

Jufang Wang,a Tor Savidge,f and Hanping Fengd

School of Bioscience and Biotechnology, South China University of Technology, Guangzhou, Chinaa; Tufts Cummings School of Veterinary Medicine, North Grafton,Massachusetts, USAb; School of Bioengineering, East China University of Science and Technology, Shanghai, Chinac; Department of Microbial Pathogenesis, University ofMaryland Dental School, Baltimore, Maryland, USAd; Department of Basic Sciences, The Commonwealth Medical College, Scranton, Pennsylvania, USAe; and Departmentof Gastroenterology and Hepatology, The University of Texas Medical Branch, Galveston, Texas, USAf

The global emergence of Clostridium difficile infection (CDI) has contributed to the recent surge in severe antibiotic-associateddiarrhea and colonic inflammation. C. difficile produces two homologous glucosylating exotoxins, TcdA and TcdB, both ofwhich are pathogenic and require neutralization to prevent disease occurrence. However, because of their large size and complexmultifunctional domain structures, it has been a challenge to produce native recombinant toxins that may serve as vaccine can-didates. Here, we describe a novel chimeric toxin vaccine that retains major neutralizing epitopes from both toxins and conferscomplete protection against primary and recurrent CDI in mice. Using a nonpathogenic Bacillus megaterium expression system,we generated glucosyltransferase-deficient holotoxins and demonstrated their loss of toxicity. The atoxic holotoxins inducedpotent antitoxin neutralizing antibodies showing little cross-immunogenicity or protection between TcdA and TcdB. To facili-tate simultaneous protection against both toxins, we generated an active clostridial toxin chimera by switching the receptorbinding domain of TcdB with that of TcdA. The toxin chimera was fully cytotoxic and showed potent proinflammatory activities.This toxicity was essentially abolished in a glucosyltransferase-deficient toxin chimera, cTxAB. Parenteral immunization of miceor hamsters with cTxAB induced rapid and potent neutralizing antibodies against both toxins. Complete and long-lasting dis-ease protection was conferred by cTxAB vaccinations against both laboratory and hypervirulent C. difficile strains. Finally, pro-phylactic cTxAB vaccination prevented spore-induced disease relapse, which constitutes one of the most significant clinical is-sues in CDI. Thus, the rational design of recombinant chimeric toxins provides a novel approach for protecting individuals athigh risk of developing CDI.

Clostridium difficile is the most common cause of nosocomialantibiotic-associated diarrhea and is the etiologic agent of

pseudomembranous colitis (8, 27, 45). The drug-resistant bacte-rium causes a wide spectrum of disease symptoms, ranging frommild diarrhea and colitis to fulminant systemic disease and mor-tality (3, 5). With the recent emergence of hypervirulent strains,the incidence of C. difficile infection (CDI) has increased rapidlyworldwide, along with more severe forms of the disease, causinglengthy hospitalization, substantial morbidity, and mortality (37,42). Two exotoxins (TcdA and TcdB) are the major cause of thedisease (30, 41). Both toxins are large, single-chain proteins withsimilar multidomain structures that include an N terminus cata-lytic glucosyltransferase domain (GTD), an autoproteolytic cys-teine proteinase domain (CPD), a central translocation domain(TM), and a C-terminal, so-called receptor-binding domain(RBD) although its receptor binding function has yet to be con-firmed (22). The toxin glucosyltransferase inhibits host RhoGTPases in colonocytes, causing cytoskeletal disruption, barrierdysfunction, diarrhea, and colitis (58). Standard therapy dependson treatment with vancomycin or metronidazole, neither ofwhich is fully effective as up to 35% of patients develop recurringdisease within a few weeks (62).

Although fidaxomicin (38) and humanized monoclonal anti-toxin IgGs (39) have recently been shown to reduce CDI recur-rence in patients, it is generally considered that a vaccine approachtargeting the C. difficile toxins is a preferred preventative strategy(15–17, 56). In support of this notion, antibodies against bothtoxins are protective in hamsters (13, 28, 34), and serum anti-Tcd

antibodies in patients correlate with protection against symptom-atic disease and recurrence (31, 33). Vaccine candidates that targetthe C. difficile toxins include toxoids (1, 16, 18, 50, 51, 56) andrecombinant TcdA RBD fragments (4, 15, 43, 46–48, 59, 60). Tox-oids, generated by formalin inactivation of holotoxins, representthe major focus of vaccine development and have been tested inpatients (1, 19, 29, 51). While effective in animal CDI models (17),toxoids have yet to be proven as protective in patients and areassociated with several inherent shortcomings, including batch-to-batch variations and potential residual toxicity. In this study,attenuated recombinant toxins were generated that showed supe-rior efficacy over toxoids in protecting mice from C. difficile toxinchallenge but failed to confer adequate cross-protection. A toxinchimera constructed subsequently provided concurrent protec-tion against both toxins and C. difficile oral challenge, showingpotent efficacy as a new vaccine candidate in experimental CDImodels.

Received 29 February 2012 Returned for modification 13 April 2012Accepted 11 May 2012

Published ahead of print 21 May 2012

Editor: S. R. Blanke

Address correspondence to Hanping Feng, hfeng@umaryland.edu.

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/IAI.00215-12

2678 iai.asm.org Infection and Immunity p. 2678–2688 August 2012 Volume 80 Number 8

MATERIALS AND METHODSGeneration of mutant holotoxins, toxin fragment, and chimeras. Wehave previously cloned the full-length TcdA and TcdB genes into a shuttlevector pHis1522 (pHis-TcdA and pHis-TcdB, respectively) and expressedthe recombinant holotoxins in Bacillus megaterium (61). Based on thissystem, point mutations were introduced into conserved amino acids thatare responsible for the substrate uridine diphosphoglucose (UDP-Glc)binding in order to generate the glucosyltransferase (GT)-deficient holo-toxins (23). To generate GT-mutant holotoxin A, we first designed aunique restriction enzyme (BamHI) site between sequences encoding GTand CPD domains using overlapping PCR. The primer sets used were asfollows: pHis-F, 5=-TTTGTTTATCCACCGAACTAAG-3=; Bam-R, 5=-TCTTCAGAAAGGGATCCACCAG-3=; Bam-F, 5=-TGGTGGATCCCTTTCTGAAGAC-3=; and Bpu-R, 5=-ACTGCTCCAGTTTCCCAC-3=.

The final PCR product was digested with BsrGI and Bpu10I and usedto replace the corresponding sequence in pHis-TcdA. The resulting plas-mid was designated pH-TxA-b. Sequences encoding triple mutations(W101A, D287N, and W519A) in the GT were synthesized by Geneart(Germany) and cloned into pH-TxA-b through BsrGI/BamHI digestion.To generate the mutant holotoxin B construct, the sequence betweenBsrGI and NheI containing two point mutations (W102A and D288N)was synthesized and inserted into pHis-TcdB at the same restriction en-zyme sites, leading to a new plasmid, pH-aTcdB. The fragment with TcdB-RBD was generated by PCR amplification using the following primer set:5=-GGTTGCTGGATCCGCAAATAAGCTATCTTTTAACTTTAGTGATAAACAAGATGTACC-3= and 5=-CCATGCTGAGCTCGCTTCACTAATCACTAATTGAGCTGTATCAGGATCAAAATAATAC-3=. The fragmentwas expressed using pET32 vector (Novagen, NJ).

To generate the chimera TxB-Ar, a unique restriction endonuclease(RE) AgeI site was created at a position between the transmembrane do-main (TMD) and RBD without changing the amino acid sequence ofexpressed pHis-TcdB. Then, the gene encoding the RBD of TcdA wasamplified using the primers TxA-Ar-F (5=-AATTACCGGTTTTAACTTAGTAACTGGATGGC-3=) and TxA-Ar-R (5=-AATTGCATGCTGGTACCCTCCATATATCCCAGGGGCTTTTACTCC-3=), and the RBD sequenceof TcdB was replaced with that of TcdA through AgeI/KpnI digestion,generating a plasmid (pH-TxB-Ar). To generate the chimera cTxAB, theXhoI/Bpu10I fragment in pH-TxB-Ar was replaced by the fragment car-rying W102A and D288N mutations from pH-aTcdB. The resultant con-structs carrying full-length mutant toxin and chimera genes were used totransform B. megaterium, and mutant holotoxins and chimeras (Fig. 1A toD) were expressed and purified using methods described previously (61).

Glucosyltransferase activity of the toxins. GT activity of holotoxinsand mutant toxins was measured by their ability to glucosylate RhoGTPase Rac1 in a cell-free assay. Vero cell pellets were resuspended inglucosylation buffer (50 mM HEPES, pH 7.5, 100 mM KCl, 1 mM MnCl2,and 2 mM MgCl2) and lysed by passage through a syringe (25 gauge).After centrifugation (at 167,000 � g for 3 min), the supernatant was usedas a postnuclear cell lysate. To perform the glucosylation assay, the celllysates were incubated with TcdA, TcdB, or their mutants (5 �g/ml finalconcentration) at 37°C for 30 min. The reaction was terminated by heat-ing the sample at 100°C for 5 min in SDS sample buffer. To measure Rac1glucosylation, lysates were separated on a 12% SDS-PAGE gel and trans-ferred onto a nitrocellulose membrane. Antibodies that specifically recog-nize the nonglucosylated form of Rac1 (clone 102; BD Bioscience),anti-�-actin (clone AC-40; Sigma), and horseradish peroxidase (HRP)-conjugated anti-mouse IgG (Amersham Biosciences) were used for detec-tion using standard Western blotting with enhanced chemiluminescence.

CD spectroscopy of wild-type and mutant toxins. Circular dichroism(CD) spectra were recorded on an Aviv 62 spectropolarimeter in thewavelength range of 190 to 260 nm, with a bandwidth of 1.0 nm and scanstep of 0.5 nm using a 0.1-cm path length in a 1-cm quartz cell at 22°C.Protein concentrations were in the range of 50 to 200 �g/ml. In each caseat least five spectra were accumulated, smoothed, averaged, and correctedfor the contribution of solutes.

Cytopathic and cytotoxicity assay. Cytopathic and cytotoxic activi-ties of the toxins were assayed as described previously (20). CT-26 or Verocells seeded in 96-well plates were treated with wild-type or mutant toxins.To evaluate cytopathic effects of the toxins on cells, the morphologicalchanges of cells were observed using phase-contrast microscopy. MTT[3-(4,5-dimethylthiazol-2-yl)2 2,5-diphenyl tetrazolium bromide] assayswere also performed to measure the cytotoxic activities of the toxins. After72 h of incubation, 10 �l of MTT (5 mg/ml) was added to each well, andplates were incubated at 37°C for 4 h. The formazan was solubilized withacidic isopropanol (0.4 N HCl in absolute isopropanol), and absorbancewas measured at 570 nm using a 96-well enzyme-linked immunosorbentassay (ELISA) plate reader. Cell viability was expressed as the percentageof survival compared cells in untreated control wells. The experimentswere repeated three times, and triplicate wells were assessed for cytopathicchanges and cytotoxicity in each experiment.

Cytokine production by dendritic cells. To assess the proinflamma-tory activities of the chimeric toxin TcdB-Ar, mouse bone marrow-de-rived dendritic cells (BMDCs) were collected as described previously (11,12). Cells were either untreated or exposed to TxB-Ar (200 ng/ml) aloneor together with goat anti-TcdA and TcdB polysera (Techlab, Inc.) for 24h, and cell supernatants were harvested. The presence of the cytokine/chemokine interleukin-1� (IL-1�), IL-6, and CXCL1 (keratinocyte-de-rived chemokine [KC]) in the supernatants was measured by standardELISA following the manufacturer’s directions with murine cytokinequantification kits (Invitrogen and R&D Systems).

Mouse immunizations. C57BL/6, BALB/c, and CD1 mice (5 to 6weeks old) were purchased from Jackson Laboratory. All mice used in theexperiments were housed in groups of 5 animals per cage under the sameconditions. Food, water, bedding, and cages were autoclaved. All animalswere handled and cared for according to Institutional Animal Care andUse Committee guidelines and in accordance with the recommendationsin the Guide for the Care and Use of Laboratory Animals of the NationalInstitutes of Health. BALB/c or C57BL/6 mice were immunized intraperi-toneally (i.p.) or intramuscularly (i.m.) with 5 �g of purified mutanttoxins in phosphate-buffered saline (PBS) with alum as an adjuvant foreach injection. Control mice received PBS with alum. When aTcdA mixedwith aTcdB (5 �g each) or cTxAB was used as the immunogen, a total of10 �g of protein per injection was used, and mice were given three immu-nizations with an interval of 10 to 14 days. For passive immunization, 50�l of polysera collected from alpacas immunized with either aTcdA oraTcdB or with the mixed polysera (100 �l total) was injected i.p. into naïveC57BL/6 mice at 4 h after C. difficile challenge. Control mice were admin-istered the same amount of serum collected from animals prior to toxinimmunization (preserum).

Antibody titers and in vitro neutralizing assay. Antibody titers weremeasured using a standard ELISA against purified native or recombinantwild-type holotoxins. To assess in vitro neutralizing activities of serumsamples, we used mouse intestinal epithelial CT26 cells as these are sensi-tive to both TcdA and TcdB. The neutralizing titer is defined as the max-imum dilution of the samples that blocks cell rounding induced by toxinat a given concentration. This given concentration is four times the min-imum dose of the toxin that causes all CT26 cells to round after a 24-hexposure to the toxin, i.e., 5 and 0.25 ng/ml for TcdA and TcdB, respec-tively. To assess neutralizing epitopes, sera from aTcdB-immunized micewere preincubated with excessive amounts (50 ng/ml) of toxoid B orrecombinant TcdB-RBD fragment on ice for 30 min before being added toCT26 cells, and serum neutralizing titers were determined. Wells withoutserum were included as controls, and neither toxoid B nor the fragment at50 ng/ml inhibited TcdB cytotoxicity.

Measurement of antitoxin IgG isotypes. IgG1, IgG2a, IgG2b, IgG2c,and IgG3 anti-TcdB concentrations in the sera of aTcdB- or toxoid B-im-munized mice were determined by ELISA using biotinylated anti-mouseIgG subclass antibodies.

Primary and recurrent CDI models. C57BL/c mice were treated withan antibiotic cocktail followed by oral inoculation of C. difficile as de-

Chimeric Toxin Vaccine against CDI

August 2012 Volume 80 Number 8 iai.asm.org 2679

scribed previously (7). Ten days after the third immunization, mice weregiven 105 CFU of vegetative bacteria (laboratory VPI10463 strain) via oralgavage. To minimize oxygen exposure, the vegetative cell aliquots in air-tight vials were prepared in an anaerobic chamber, and the vials wereopened right before inoculation. To assess long-term immunity, micewere orally challenged with 106 CFU of vegetative bacteria 3 months afterthe third immunization. In some experiments, immunized mice werechallenged with 106 spores of UK1 (027/B1/NAP1 strain kindly providedby D. Gerding, VA Chicago Health Care System). To induce CDI relapse,surviving mice were given antibiotic cocktail treatment followed by oralgavage of C. difficile spores (106/mouse) 30 days after the primary infec-tion (55).

Hamster immunization and challenge. Hamsters were i.p. immu-nized with 10 �g of cTxAB in 100 �l of PBS with alum as an adjuvant threetimes with 10-day intervals. Control hamsters were immunized with anequal volume of PBS-alum. Serum samples were collected before and 9days after each immunization, and anti-TcdA and anti-TcdB IgG titerswere measured by standard ELISA. Five days after the third immuniza-tion, the hamsters were injected i.p. with clindamycin (30 mg/kg) andthen orally challenged with C. difficile UK1 spores (100 CFU/hamster) 5

days later. The animals were monitored daily for diarrhea and other signsof disease, and moribund animals were euthanized.

Statistical analysis. Data were analyzed by Kaplan-Meier survivalanalysis with a log rank test of significance, by analysis of variance(ANOVA), and by one-way or two-way ANOVA followed with Bonfer-roni posttests using the Prism statistical software program. Results areexpressed as means � standard errors of means.

RESULTSGeneration of attenuated C. difficile holotoxins. TcdA and TcdBare large clostridial toxins that possess complex structural confor-mations (22, 44). Formalin cross-linking of these toxins is likely toalter conformational epitopes, which can impact immunogenic-ity. To demonstrate this point, attenuated recombinant TcdA andTcdB were generated with point mutations in the highly con-served amino acids that regulate substrate binding in glucosyl-transferases (23), designated aTcdA and aTcdB, respectively (Fig.1A and B). The mutant holotoxins were essentially devoid of glu-cosyltransferase activity (Fig. 1E), cytotoxicity (Fig. 1F and G),

FIG 1 Generation of glucosyltransferase (GT)-mutant holotoxins. aTcdA (A) and aTcdB (B) represent mutant TcdA with triple mutations (W101A, D287N, andW519A) and TcdB with double mutations (W102A and D288N) in their respective GT domains. (C) TxB-Ar is TcdB with its RBD replaced with that of TcdA.(D) cTxAB is TxB-Ar with two mutations (W102A and D288N) in its glucosyltransferase domain. CPD, cysteine protease domain; TMD, transmembranedomain; RBD, receptor-binding domain; His6, six-histidine tag. (E) Vero cell lysates were exposed to wild-type or mutant toxin proteins for 30 min. Rac1glucosylation was analyzed by immunoblotting using monoclonal antibody (clone 102) that binds only to nonglucosylated Rac1. �-Actin was used as an equalloading control. CT26 cells in a 96-well plate were exposed to aTcdA or TcdA (F) or to aTcdB or TcdB (G) at different concentrations for 72 h. MTT assays wereperformed, and cell viability is expressed as the percentage of surviving cells compared to cells without toxin exposure. (H) BALB/c mice were i.p. challenged witheither 100 ng/mouse wild-type TcdA or TcdB or with 100 �g/mouse of mutant aTcdA or aTcdB. Mouse survival was monitored, and the data show theKaplan-Meier survival curves (n � 10; P � 0.001, between wild-type and mutant toxin groups).

Wang et al.

2680 iai.asm.org Infection and Immunity

and in vivo toxicity (Fig. 1H). In contrast to toxoids which dem-onstrate significant conformational changes compared with na-tive toxins, mutant holotoxins maintained a structure similar tothat of native toxins, as determined by circular dichroism spec-trometry (Fig. 2A to C). Thus, structurally intact glucosyltrans-ferase-deficient holotoxins are essentially nonpathogenic, makingthem eminently suitable vaccine candidates.

The mutant holotoxin is more immunogenic and protectiveagainst C. difficile toxin challenge than toxoid. Toxoids TcdAand TcdB, generated by formalin inactivation of holotoxins, rep-resent the major focus of vaccine development and have beentested in patients (1, 19, 29, 50, 51). Recent studies in patients (36)and in animal models (30, 41) have highlighted the relative im-portance of TcdB as one of the primary virulence factors in CDI.We therefore compared the immunogenicity of aTcdB with tox-oid TcdB (toxoid B). Both immunogens generated a Th2-typeresponse with serum antitoxin IgG1 and IgG2b in systemicallyimmunized mice (Fig. 3A). However, immunization with aTcdBinduced a stronger IgG response and significantly higher neutral-izing activity than immunization with toxoid B (Fig. 3B and C).Preincubation of polysera from aTcdB-immunized mice with tox-oid B did not significantly reduce their neutralizing activity (Fig.3C), indicating that the highly neutralizing epitopes were largelyabsent in toxoid B. As a consequence, aTcdB-immunized micewere fully protected against lethal wild-type TcdB challenge,whereas toxoid B-immunized mice all developed systemic tox-emia, with 70% lethality (Fig. 3D). Thus, aTcdB immunizationconferred protection against systemic toxin challenge that wassuperior to that of toxoid B.

Attenuated holotoxins show limited cross-protectionagainst CDI. Because most pathogenic C. difficile isolates produce

TcdA and TcdB, which are highly homologous to each other (58),we examined cross-protection between the two toxins. Consistentwith a previous report (35), antibodies generated by aTcdB im-munization showed little cross-protection against TcdA and viceversa (Fig. 4A and B). aTcdB immunization protected mice fromlethal challenge of TcdB (Fig. 3C) but not of TcdA (Fig. 4C). Whenmice were immunized with both aTcdA and aTcdB, potent neu-tralizing antibodies were evident against both toxins (Fig. 4A andB), and animals were fully protected against lethal systemic chal-lenge with the two toxins (Fig. 4D). Moreover, immunization withaTcdA or aTcdB alone only partially protected mice from oralchallenge with C. difficile strain VPI10463 (Fig. 4E and F), whereassimultaneous administration of aTcdA and aTcdB induced com-plete protection against the lethality and diarrhea of CDI (Fig. 4Eand F). These data demonstrate a lack of cross-immunogenicitybetween the two toxins and a requirement for neutralizing anti-bodies against both toxins for full protection.

Generation of the active toxin chimera TxB-Ar and glucosyl-transferase mutant cTxAB. Because of the insufficient cross-pro-tection elicited by the individual antitoxins, we sought to generatea single immunogen capable of inducing potent neutralizing an-tibodies against both toxins. The RBD is the immunodominanttoxin region of TcdA and possesses potent adjuvant activity due toits lectin-like structural repeats (6, 57). In contrast, the neutraliz-ing epitopes in TcdB are primarily confined to the N terminussince preincubation of polysera from aTcdB-immunized micewith recombinant TcdB-RBD did not significantly reduce the se-rum neutralizing activity (Fig. 3C). Moreover, the N terminus inTcdB is more highly conserved between historical and hyperviru-lent strains than its RBD (32, 52). These observations prompted usto substitute the RBD of TcdB with that of TcdA, generating an

FIG 2 Circular dichroism (CD) structural analysis of wild-type and mutant toxins. For secondary structural analysis of wild-type TcdA and mutant aTcdA (A)and of TcdB, aTcdB, and toxoid B (B), far-UV CD spectra were recorded at 22°C. This structural analysis demonstrates that wild-type and GT-mutant toxins arestructurally similar since CD spectra are virtually superimposed, but a significant shift occurs in toxoid B. Secondary structural composition elements areillustrated in panel C.

Chimeric Toxin Vaccine against CDI

August 2012 Volume 80 Number 8 iai.asm.org 2681

active toxin chimera designated TxB-Ar (Fig. 1C). TxB-Ar re-tained potent glucosylating activity (Fig. 1E), cytotoxicity (Fig.5A), and proinflammatory activity (Fig. 5B) equal to that of thewild-type TcdA and TcdB (14). Thus, TxB-Ar functions as a trueclostridial glucosylating toxin and likely maintains a toxin-likeconformation similar to that of the native C. difficile toxins. Toestablish a glucosyltransferase-deficient chimera and retain thetoxin-like conformation, two point mutations (W101A andD288N) were introduced as described for aTcdB (Fig. 1B and 2Band C), and the resulting chimera was designated cTxAB(Fig. 1D). cTxAB displayed undetectable glucosyltransferase ac-

tivity in a cell-free glucosylation assay (Fig. 1E) and essentially lostits cytotoxicity (Fig. 5A).

cTxAB vaccination induces potent neutralizing antibodiesagainst both toxins and long-lasting protection in mice. Intra-peritoneal immunization of mice with cTxAB induced a potentsystemic antibody response against both toxins (Fig. 6A). A sig-nificant IgG response was evident after a single immunizationdose (Fig. 6A) that was not measured with aTcdB immunizationalone (Fig. 3B) and is likely due to the lack of the RBD adjuvantactivity of TcdA (6). After three immunizations, potent serumneutralizing activities with titers over 3,000 were evident againstboth TcdA and TcdB (Fig. 6B). Consequently, immunized micewere fully protected against lethal systemic challenge with eitherwild-type TcdA or TcdB (Fig. 6C).

Next, we evaluated the immunogen cTxAB in a mouse modelof CDI (7). After three parenteral immunizations and oral chal-lenge with C. difficile VPI10463, cTxAB-immunized animals failedto show any signs of disease, whereas all vehicle-immunized micedeveloped diarrhea and weight loss, and approximately 60% ofmice succumbed (Fig. 6D). cTxAB vaccination also establishedlong-term protection against CDI, lasting at least 3 months afterthe third immunization (Fig. 6E).

cTxAB vaccination protects animals from infection with anepidemic C. difficile strain. Because many different toxin iso-forms are evident in clinical isolates, we examined whether cTxABalso confers protection against the most clinically relevant strain,BI/NAP1/027. Immunized mice were challenged with spores ofthe UK1 clinical isolate BI/NAP1/027 (54). In vehicle-immunizedmice, significant disease symptoms (ruffled coat, lethargy, loss ofappetite, and severe diarrhea) were evident by day 2 postinfectionin all mice, and approximately 40% of mice succumbed by day 4(Fig. 7A). In contrast, cTxAB-immunized mice were fully pro-tected and showed no signs of disease at any stage (Fig. 7A). Wenext assessed the potential suitability of cTxAB for clinical intra-muscular immunization and demonstrated similar disease pro-tection against spore challenge (Fig. 7B). Furthermore, a singlecTxAB dose was sufficient to induce anti-Tcd antibody titers (Fig.6A) that were protective against CDI. Mice were immunized onceon the same day as antibiotic treatment was initiated and werethen challenged with spores 6 days later. A single cTxAB immuni-zation dose conferred significant protection against fulminantdisease (Fig. 7C) although surviving mice in both groups experi-enced similar diarrhea (data not shown) and weight loss (Fig. 7D).Finally, we examined the protection of cTxAB vaccination in atraditional hamster CDI model with oral UK1 spore challenge. Apotent antibody response against both toxins was induced thatconferred significant protection against UK1 spore-induced CDIin hamsters (Fig. 8A to C).

Prophylactic cTxAB vaccination protects mice from recur-rent CDI. CDI has become increasingly difficult to manage due, inpart, to the ineffectiveness of current antibiotic regimens whichare associated with high relapse rates (24). We tested the efficacy ofcTxAB immunization in preventing disease recurrence in a spore-induced mouse CDI recurrence model (55). The immunizationand challenge scheme is illustrated in Fig. 9A. In vehicle-immu-nized mice, CDI recurrence demonstrated a similar disease courseto the primary infection, with severe diarrhea and weight loss be-ing evident and 40% of animals becoming moribund (Fig. 9C). Incontrast, cTxAB immunization conferred complete protectionagainst primary and recurrent CDI (Fig. 9C), reflecting long-last-

FIG 3 Mouse antibody response and neutralizing titers after aTcdB or toxoidB immunization. Mice were immunized with aTcdB or toxoid B three times,and serum samples were collected. (A) The antitoxin IgG isotypes of the serumsamples were measured using standard ELISAs. OD405, optical density at 405nm. (B) TcdB-specific antibody titers after each immunization (im) withaTcdB or toxoid B (*, P � 0.05). (C) Neutralizing titers of sera from toxoidB-immunized mice (SeraToxoid) and from aTcdB-immunized mice(SeraaTcdB). In some groups, the sera were preincubated with 50 ng/ml oftoxoid B (SeraaTcdB � Toxoid) or recombinant fragment TcdB-RBD(SeraaTcdB � F4) before being added to the cells (**, P � 0.01; ns, not signifi-cant). (D) Kaplan-Meier survival curves of the data of immunized mice chal-lenged with a lethal dose (100 ng/mouse) of wild-type TcdB (aTcdB versustoxoid B, P � 0.001). All data are representative of at least three independentexperiments (error bars indicate standard errors of the means; the data wereanalyzed by Kaplan-Meier survival analysis or by ANOVA).

Wang et al.

2682 iai.asm.org Infection and Immunity

ing immunity after vaccination (Fig. 6E). To assess whethercTxAB immunization also protects against disease relapse in naïveanimals that recover from CDI, surviving mice were immunizedafter their recovery from the initial CDI as illustrated (Fig. 9B).cTxAB-immunized animals were completely protected from CDIrecurrence, whereas vehicle-immunized mice exhibited profounddisease (Fig. 9D).

DISCUSSION

A majority of pathogenic C. difficile isolates produce both TcdAand TcdB (58). Thus, neutralizing antibodies elicited against bothtoxins are ideally required to impart complete protection againstCDI. Because the two homologous toxins exhibit 66% overall

amino acid sequence similarity (58), it is important to understandcross-protective epitopes that may assist in designing immune-based interventions. It has long been known that immunizationwith individual C. difficile toxoids generates poor cross-protection(neutralizing titers of �20) (35). Peptide fragments containingfull or partial TcdA-RBD sequence, the immunodominant regionof the toxin, have been extensively evaluated as vaccine candidates(4, 15, 43, 46, 47, 59, 60) but generally have resulted in poor pro-tective responses (4, 60). A more recent report (43) has shown thatimmunization with a TcdA-RBD peptide fragment expressed inBacillus subtilis induced weak cross-protection against TcdB (neu-tralizing titers ranging from 10 to 100). The findings in our studyare consistent with these reports since immunization of mice with

FIG 4 Cross-protection of polysera from mice immunized with mutant toxins. Groups of mice were immunized with the indicated immunogens three times.(A and B) After three immunizations with the indicated immunogens, serum neutralizing titers against TcdA (A) and TcdB (B) were determined (*, P � 0.01compared to aTcdB in panel A or aTcdA in panel B). (C and D) Kaplan-Meier survival curves of aTcdB-immunized mice (C) or mice immunized with bothaTcdA and aTcdB (D) and challenged with a lethal dose of TcdA (panel C, n � 10; P � 0.535) or of TcdA and TcdB (panel D, n � 10; P � 0.001), respectively.(E and F) After oral challenge with the C. difficile VPI10463 strain, mouse survival (E) (aTcdA plus aTcdB versus PBS, P � 0.04) and the percentage of micedeveloping diarrhea (F) are illustrated. All data are representative of at least three independent experiments.

FIG 5 Cytotoxicity and proinflammatory activities of the chimeric toxins. (A) MTT survival assays of CT26 cells exposed to TxB-Ar or cTxAB. Cell viability isexpressed as the percentage of surviving cells compared to cells without toxin exposure. (B) Cytokine/chemokine secretion by BMDCs after TxB-Ar (200 ng/ml)exposure. P � 0.01, between TxB-Ar and the other groups. The data were analyzed by ANOVA followed by Bonferroni posttests. Error bars show � standarderrors of the means.

Chimeric Toxin Vaccine against CDI

August 2012 Volume 80 Number 8 iai.asm.org 2683

individual nonpathogenic holotoxins generated high neutralizingIgG titers against the respective toxin but demonstrated littlecross-neutralization (titers of �100) (Fig. 4A and B) and, moreimportantly, were not completely protective against experimentalCDI (Fig. 4E and F). Therefore, there is an urgent need to generatea fully cross-protective Tcd immunogen.

We report a novel clostridial toxin chimera, cTxAB, as a vac-cine candidate for CDI. In designing cTxAB, several aspects weretaken into consideration: (i) the immunogen should be nontoxicand safe; (ii) the native toxin structure should be maintained, thuspreserving conformational epitopes; (iii) the immunogen shouldretain the major immunodominant domains and neutralizingepitopes from both toxins; and (iv) the immunogen should beconserved, facilitating widespread protection against diverse clin-ical isolates. cTxAB maintains a native toxin-like conformationbut is essentially nontoxic in mice and hamsters. It is expressed inBacillus megaterium, a nonpathogenic and endotoxin-free pro-duction system suitable for clinical use. Additionally, cTxAB re-tains the major conserved neutralizing epitopes of TcdA and TcdBand thus induces potent neutralizing antibodies against both tox-ins that protect experimental animals against primary and recur-

rent CDI induced by both historic and epidemic strains of C. dif-ficile.

Because multiple C. difficile strains expressing different toxinisoforms are isolated from CDI patients, it is desirable for a vac-cine to confer broad protection across clinically relevant isolates.Recent reports have shown that TcdA is relatively well conservedbetween historical and epidemic strains, whereas TcdB shows ahigher degree of variability (32, 52). The RBD domain is the mostvariable region within TcdB, whereas the N terminus encompass-ing the GTD and CPD domains is more conserved between his-torical and epidemic strains (32). Consequently, we designedcTxAB to comprise the N terminus from TcdB and the C terminusRBD from TcdA, thus retaining conserved epitopes across toxinisoforms from different clinical strains. In support of this as asuccessful immunization strategy, cTxAB induced complete pro-tection not only against the historical VPI10463 stain but also, andmore importantly, against the most clinically relevant isolate, BI/NAP1/027, in mice. Whether this candidate vaccine indeed pro-vides global protection against all pathogenic C. difficile isolateswill ultimately need to be determined in clinical trials.

C. difficile infection causes a wide spectrum of clinical symp-

FIG 6 Chimera cTxAB immunization induces potent neutralizing antibodies against both toxins and protective immunity against CDI. (A and B) Serumanti-TcdA and anti-TcdB IgG titers (A) or neutralizing titers (B) are shown. (C) Mice immunized with PBS (solid lines) or cTxAB (dashed lines) were dividedinto two groups and challenged with lethal doses of TcdA or TcdB, respectively (P � 0.01 between PBS and cTxAB-immunized groups). (D and E) Mice werechallenged with C. difficile VPI10463 vegetative cells 10 days (D) or 3 months (E) after the third immunization with PBS or cTxAB. Mouse mortality (P � 0.05),weight loss, and frequency of diarrhea are illustrated. With the exception of the experiment shown in panel E, all experiments were performed at least three times,and representative results are shown (n � 10 in each experiment; *, P � 0.05).

Wang et al.

2684 iai.asm.org Infection and Immunity

toms, ranging from mild diarrhea to fulminant disease and mor-tality (27, 45). Unlike the hamster model of CDI where animalsconsistently develop a fulminant and lethal disease course (40),the recently established mouse CDI model develops a wider range

of disease symptoms that more accurately resemble major aspectsof the human disease (7). In this study, we chose to challenge micewith doses of C. difficile that recapitulate the full clinical spectrumof disease symptoms ranging from a self-limiting diarrhea to ful-

FIG 7 cTxAB immunization protects mice against infection with an epidemic C. difficile strain. (A and B) After three i.p. (A) or i.m. (B) immunizations, micewere challenged with C. difficile UK1 spores (106/mouse). Mouse mortality (P � 0.05), weight loss, and frequency of diarrhea are illustrated (dotted line, cTxAB;solid line, PBS; dashed line, antibiotic cocktail treatment without spore challenge). Symbols for the weight loss graphs are the same in both panels. (C and D) Micewere immunized with cTxAB once on day �6 when the antibiotic treatment was initiated and then challenged with UK1 spores on day 0. Mouse mortality (C)(P � 0.05) and weight loss (D) are illustrated. With the exception of the experiment shown in panel B, all experiments were performed at least three times, andthe representative results are shown (n � 10 in each experiment; *, P � 0.05).

FIG 8 Protective response of cTxAB vaccination in hamsters. (A) Serum IgG titers after each immunization with cTxAB (10 �g per i.p. injection). (B)Kaplan-Meier survival curves of PBS-immunized (n � 15) or cTxAB-immunized (n � 10) hamsters orally challenged with C. difficile spores (P � 0.0007). (C)Percentage of hamsters in the PBS or cTxAB group that developed diarrhea. All hamsters in the PBS group developed severe diarrhea, whereas only one hamsterin the cTxAB group developed diarrhea. The statistics were analyzed by Kaplan-Meier survival analysis or ANOVA.

Chimeric Toxin Vaccine against CDI

August 2012 Volume 80 Number 8 iai.asm.org 2685

minant systemic disease (53). We also conducted confirmatoryexperiments in hamsters in accordance with previous studies thatmainly use mortality as a single disease parameter for evaluation(28, 43). In our opinion, while mortality is an important param-eter, it is more useful to fully evaluate the efficacy of vaccinesagainst a spectrum of complex disease symptoms, including dis-ease recurrence as is evident in CDI. To illustrate this point in ourstudy, aTcdA-immunized mice were fully protected against mor-tality after C. difficile challenge (Fig. 4E). This result, although inagreement with several previous findings (28, 43), should not beinterpreted as a demonstration of cross-protection against TcdBsince aTcdA immunization was still associated with significantintestinal disease in these animals (Fig. 4F). In yet another exam-ple, although a single cTxAB immunization dose conferred signif-icant protection against mortality induced by CDI (Fig. 7C), it didnot altogether prevent diarrhea or weight loss (Fig. 7D and datanot shown). Although repeated immunization is important inconferring complete disease protection in mice, prophylactic im-munization may remain an attractive preventative strategy in pa-tients who have previously been exposed to the infection. Thus, itremains to be determined in a clinical trial setting whether a single

vaccination dose is sufficient to provide prophylactic protectionagainst CDI.

Disease recurrence is one of the most significant clinical com-plications associated with CDI, occurring in 25 to 35% of patientswith primary infection and rising up to 50% in subsequent infec-tive episodes (24). Using a mouse model of CDI relapse that wehave recently described (55), we assessed the efficacy of the pro-tective immunity of cTxAB both before and after the initial infec-tive episode. In all cases, cTxAB vaccination conferred completeprotection against disease recurrence. Thus, a regimen of vaccina-tion after the initial episode of CDI may be especially relevant in aclinical setting because such patients are at greater risk of devel-oping recurrent CDI.

It has long been known that systemic antitoxin antibodies areprotective against CDI (18, 26), but the underlying mechanismsfor mucosal protection are not fully understood. Our data dem-onstrated that the parenteral cTxAB vaccination induced full dis-ease protection, which may be mediated at both systemic andmucosal levels. Systemic cTxAB immunization induced potentserum neutralizing antibodies, and mice were completely pro-tected against lethal systemic toxin challenge (Fig. 6C). We have

FIG 9 cTxAB vaccination protects mice against recurrent CDI. (A and B) Immunization and challenge schemes for CDI relapse models. (C and D) After thesecond spore challenge illustrated in panel A (C) or panel B (D), mouse mortality, weight loss, and diarrhea were monitored. Symbols for the weight loss graphsare the same in both panels. All experiments were performed at least three times with similar results (n � 8 to 10 in each experiment) Survival figures show pooleddata from two experiments and data were analyzed by Kaplan-Meier survival analysis (P is �0.05 in each case. Weight loss was analyzed by ANOVA (*, P � 0.05;error bars indicate standard errors of the means).

Wang et al.

2686 iai.asm.org Infection and Immunity

previously reported that toxins are released into the bloodstreamof piglets experimentally infected with C. difficile (21, 54), andtoxemia is associated with systemic and fulminant CDI in bothpiglets and mice (53). Although toxemia has not yet been defini-tively demonstrated in CDI patients, preliminary reports indicatethat toxins may cause systemic complications and fulminant dis-ease (2, 9, 10, 25, 49). Systemic IgG neutralizing antibodies aretherefore likely to play an important role in disease prevention byinactivating toxins released from lumen into submucosa and cir-culation. As for the protection at the mucosal surface, we failed toidentify a detectable amount of antitoxin neutralizing IgA in theintestinal lavage specimen from cTxAB-immunized mice (datanow shown). Whether these antitoxin IgG antibodies also medi-ated the protection of intestinal mucosal surfaces is under inves-tigation.

In summary, attenuated chimeric C. difficile toxins capable ofinducing potent neutralizing antitoxins and protection againstCDI can be manufactured inexpensively using a reproducible andsafe, endotoxin-free expression system in B. megaterium and mayprovide a novel prophylactic and therapeutic approach to combatprimary and recurrent CDI.

ACKNOWLEDGMENTS

These studies were supported by grants NO1-30050, R01AI088748,R01DK084509, K01DK076549, R01AI10094001, 1UL1RR029876-01, andR21-DK078032-01 and by the John S. Dunn Gulf Coast Consortium forChemical Genomics Robert A. Welch Collaborative Grant Program.

We thank Abraham L. Sonenshein and Charles B. Shoemaker (TuftsUniversity) for critical review of the manuscript.

We report that we have no conflicts of interest.

REFERENCES1. Aboudola S, et al. 2003. Clostridium difficile vaccine and serum immu-

noglobulin G antibody response to toxin A. Infect. Immun. 71:1608 –1610.

2. Abougergi MS, Broor A, Cui W, Jaar BG. 2010. Intravenous immuno-globulin for the treatment of severe Clostridium difficile colitis: an obser-vational study and review of the literature. J. Hosp. Med. 5:E1–E9.

3. Bartlett JG. 2002. Clinical practice. Antibiotic-associated diarrhea. N.Engl. J. Med. 346:334 –339.

4. Belyi IF, Varfolomeeva NA. 2003. Construction of a fusion protein car-rying antigenic determinants of enteric clostridial toxins. FEMS Micro-biol. Lett. 225:325–329.

5. Borriello SP. 1998. Pathogenesis of Clostridium difficile infection. J. An-timicrob. Chemother. 41(Suppl C):13–19.

6. Castagliuolo I, et al. 2004. Clostridium difficile toxin A carboxyl-terminus peptide lacking ADP-ribosyltransferase activity acts as a mucosaladjuvant. Infect. Immun. 72:2827–2836.

7. Chen X, et al. 2008. A mouse model of Clostridium difficile-associateddisease. Gastroenterology 135:1984 –1992.

8. Cloud J, Kelly CP. 2007. Update on Clostridium difficile associated dis-ease. Curr. Opin. Gastroenterol. 23:4 –9.

9. Cunney RJ, Magee C, McNamara E, Smyth EG, Walshe J. 1998. Clos-tridium difficile colitis associated with chronic renal failure. Nephrol. Dial.Transplant. 13:2842–2846.

10. Dobson G, Hickey C, Trinder J. 2003. Clostridium difficile colitis causingtoxic megacolon, severe sepsis and multiple organ dysfunction syndrome.Intensive Care Med. 29:1030.

11. Feng H, Zeng Y, Whitesell L, Katsanis E. 2001. Stressed apoptotic tumorcells express heat shock proteins and elicit tumor-specific immunity.Blood 97:3505–3512.

12. Feng H, et al. 2005. Listeria-infected myeloid dendritic cells produceIFN-beta, priming T cell activation. J. Immunol. 175:421– 432.

13. Fernie DS, Thomson RO, Batty I, Walker PD. 1983. Active and passiveimmunization to protect against antibiotic associated caecitis in hamsters.Dev. Biol. Stand. 53:325–332.

14. Flegel WA, et al. 1991. Cytokine response by human monocytes to Clos-tridium difficile toxin A and toxin B. Infect. Immun. 59:3659 –3666.

15. Gardiner DF, Rosenberg T, Zaharatos J, Franco D, Ho DD. 2009. ADNA vaccine targeting the receptor-binding domain of Clostridium diffi-cile toxin A. Vaccine 27:3598 –3604.

16. Ghose C, et al. 2007. Transcutaneous immunization with Clostridiumdifficile toxoid A induces systemic and mucosal immune responses andtoxin A-neutralizing antibodies in mice. Infect. Immun. 75:2826 –2832.

17. Giannasca PJ, Warny M. 2004. Active and passive immunization againstClostridium difficile diarrhea and colitis. Vaccine 22:848 – 856.

18. Giannasca PJ, et al. 1999. Serum antitoxin antibodies mediate systemicand mucosal protection from Clostridium difficile disease in hamsters.Infect. Immun. 67:527–538.

19. Greenberg RN, Marbury TC, Foglia G, Warny M. 2012. Phase I dosefinding studies of an adjuvanted Clostridium difficile toxoid vaccine. Vac-cine 30:2245–2249.

20. He X, et al. 2009. Antibody-enhanced, Fc gamma receptor-mediatedendocytosis of Clostridium difficile toxin A. Infect. Immun. 77:2294 –2303.

21. He X, et al. 2009. An ultrasensitive rapid immunocytotoxicity assay fordetecting Clostridium difficile toxins. J. Microbiol. Methods 78:97–100.

22. Jank T, Aktories K. 2008. Structure and mode of action of clostridialglucosylating toxins: the ABCD model. Trends Microbiol. 16:222–229.

23. Jank T, Giesemann T, Aktories K. 2007. Clostridium difficile glucosyl-transferase toxin B-essential amino acids for substrate binding. J. Biol.Chem. 282:35222–35231.

24. Johnson S. 2009. Recurrent Clostridium difficile infection: causality andtherapeutic approaches. Int. J. Antimicrob. Agents 33(Suppl 1):S33–S36.

25. Johnson S, et al. 2001. Fatal pseudomembranous colitis associated with avariant Clostridium difficile strain not detected by toxin A immunoassay.Ann. Intern. Med. 135:434 – 438.

26. Katchar K, et al. 2007. Association between IgG2 and IgG3 subclassresponses to toxin A and recurrent Clostridium difficile-associated disease.Clin. Gastroenterol. Hepatol. 5:707–713.

27. Kelly CP, LaMont JT. 2008. Clostridium difficile–more difficult than ever.N. Engl. J. Med. 359:1932–1940.

28. Kim PH, Iaconis JP, Rolfe RD. 1987. Immunization of adult hamstersagainst Clostridium difficile-associated ileocecitis and transfer of protec-tion to infant hamsters. Infect. Immun. 55:2984 –2992.

29. Kotloff KL, et al. 2001. Safety and immunogenicity of increasing doses ofa Clostridium difficile toxoid vaccine administered to healthy adults. In-fect. Immun. 69:988 –995.

30. Kuehne SA, et al. 2010. The role of toxin A and toxin B in Clostridiumdifficile infection. Nature 467:711–713.

31. Kyne L, Warny M, Qamar A, Kelly CP. 2001. Association betweenantibody response to toxin A and protection against recurrent Clostridiumdifficile diarrhoea. Lancet 357:189 –193.

32. Lanis JM, Barua S, Ballard JD. 2010. Variations in TcdB activity and thehypervirulence of emerging strains of Clostridium difficile. PLoS Pathog.6:e1001061. doi:10.1371/journal.ppat.1001061.

33. Leav BA, et al. 2010. Serum anti-toxin B antibody correlates with protec-tion from recurrent Clostridium difficile infection (CDI). Vaccine 28:965–969.

34. Libby JM, Jortner BS, Wilkins TD. 1982. Effects of the two toxins ofClostridium difficile in antibiotic-associated cecitis in hamsters. Infect. Im-mun. 36:822– 829.

35. Libby JM, Wilkins TD. 1982. Production of antitoxins to two toxins ofClostridium difficile and immunological comparison of the toxins bycross-neutralization studies. Infect. Immun. 35:374 –376.

36. Loo VG, et al. 2011. Host and pathogen factors for Clostridium difficileinfection and colonization. N. Engl. J. Med. 365:1693–1703.

37. Loo VG, et al. 2005. A predominantly clonal multi-institutional outbreakof Clostridium difficile-associated diarrhea with high morbidity and mor-tality. N. Engl. J. Med. 353:2442–2449.

38. Louie TJ, et al. 2011. Fidaxomicin versus vancomycin for Clostridiumdifficile infection. N. Engl. J. Med. 364:422– 431.

39. Lowy I, et al. 2010. Treatment with monoclonal antibodies against Clos-tridium difficile toxins. N. Engl. J. Med. 362:197–205.

40. Lyerly DM, Saum KE, MacDonald DK, Wilkins TD. 1985. Effects ofClostridium difficile toxins given intragastrically to animals. Infect. Im-mun. 47:349 –352.

41. Lyras D, et al. 2009. Toxin B is essential for virulence of Clostridiumdifficile. Nature 458:1176 –1179.

Chimeric Toxin Vaccine against CDI

August 2012 Volume 80 Number 8 iai.asm.org 2687

42. McDonald LC, et al. 2005. An epidemic, toxin gene-variant strain ofClostridium difficile. N. Engl. J. Med. 353:2433–2441.

43. Permpoonpattana P, et al. 2011. Immunization with Bacillus sporesexpressing toxin A peptide repeats protects against infection with Clostrid-ium difficile strains producing toxins A and B. Infect. Immun. 79:2295–2302.

44. Pruitt RN, Chambers MG, Ng KK, Ohi MD, Lacy DB. 2010. Structuralorganization of the functional domains of Clostridium difficile toxins Aand B. Proc. Natl. Acad. Sci. U. S. A. 107:13467–13472.

45. Rupnik M, Wilcox MH, Gerding DN. 2009. Clostridium difficile infec-tion: new developments in epidemiology and pathogenesis. Nat. Rev. Mi-crobiol. 7:526 –536.

46. Ryan ET, et al. 1997. Protective immunity against Clostridium difficiletoxin A induced by oral immunization with a live, attenuated Vibrio chol-erae vector strain. Infect. Immun. 65:2941–2949.

47. Sauerborn M, Leukel P, von Eichel-Streiber C. 1997. The C-terminalligand-binding domain of Clostridium difficile toxin A (TcdA) abrogatesTcdA-specific binding to cells and prevents mouse lethality. FEMS Micro-biol. Lett. 155:45–54.

48. Seregin SS, Aldhamen YA, Rastall DP, Godbehere S, Amalfitano A.2012. Adenovirus-based vaccination against Clostridium difficile toxin Aallows for rapid humoral immunity and complete protection from toxinA lethal challenge in mice. Vaccine 30:1492–1501.

49. Siarakas S, Damas E, Murrell WG. 1995. Is cardiorespiratory failureinduced by bacterial toxins the cause of sudden infant death syndrome?Studies with an animal model (the rabbit). Toxicon 33:635– 649.

50. Siddiqui F, et al. 2012. Vaccination with parenteral toxoid B protectshamsters against lethal challenge with toxin A-negative, toxin B-positiveclostridium difficile but does not prevent colonization. J. Infect. Dis. 205:128 –133.

51. Sougioultzis S, et al. 2005. Clostridium difficile toxoid vaccine in recurrentC. difficile-associated diarrhea. Gastroenterology 128:764 –770.

52. Stabler RA, et al. 2009. Comparative genome and phenotypic analysis ofClostridium difficile 027 strains provides insight into the evolution of a

hypervirulent bacterium. Genome Biol. 10:R102. doi:10.1186/gb-2009-10-9-r102.

53. Steele J, et al. 2012. Systemic dissemination of Clostridium difficile toxinsA and B is associated with severe, fatal disease in animal models. J. Infect.Dis. 205:384 –391.

54. Steele J, Feng H, Parry N, Tzipori S. 2010. Piglet models of acute orchronic Clostridium difficile illness. J. Infect. Dis. 201:428 – 434.

55. Sun X, et al. 2011. Mouse relapse model of Clostridium difficile infection.Infect. Immun. 79:2856 –2864.

56. Torres JF, Lyerly DM, Hill JE, Monath TP. 1995. Evaluation of formalin-inactivated Clostridium difficile vaccines administered by parenteral andmucosal routes of immunization in hamsters. Infect. Immun. 63:4619 –4627.

57. von Eichel-Streiber C, Zec-Pirnat I, Grabnar M, Rupnik M. 1999. Anonsense mutation abrogates production of a functional enterotoxin A inClostridium difficile toxinotype VIII strains of serogroups F and X. FEMSMicrobiol. Lett. 178:163–168.

58. Voth DE, Ballard JD. 2005. Clostridium difficile toxins: mechanism ofaction and role in disease. Clin. Microbiol. Rev. 18:247–263.

59. Ward SJ, Douce G, Dougan G, Wren BW. 1999. Local and systemicneutralizing antibody responses induced by intranasal immunization withthe nontoxic binding domain of toxin A from Clostridium difficile. Infect.Immun. 67:5124 –5132.

60. Ward SJ, Douce G, Figueiredo D, Dougan G, Wren BW. 1999. Immu-nogenicity of a Salmonella typhimurium aroA aroD vaccine expressing anontoxic domain of Clostridium difficile toxin A. Infect. Immun. 67:2145–2152.

61. Yang G, et al. 2008. Expression of recombinant Clostridium difficile toxinA and B in Bacillus megaterium. BMC Microbiol. 8:192. doi:10.1186/1471–2180-8-192.

62. Zar F, Bakkanagari S, Moorthi KMLST, Davis M. 2007. A comparison ofvancomycin and metronidazole for the treatment of Clostridium difficile-associated diarrhea, stratified by disease severity. Clin. Infect. Dis. 45:302–307.

Wang et al.

2688 iai.asm.org Infection and Immunity