Vaccine 21 (2003) 3663–3674

Evaluation of the protective efficacy ofVibrio choleraeghost (VCG) candidate vaccines in rabbits

Francis O. Ekoa,∗, Tatiana Schukovskayab, E.Y. Lotzmanovab, V.V. Firstovab,N.V. Emalyanovab, S.N. Kluevab, A.L. Kravtzovb, L.F. Livanovab,

Vladimir V. Kutyrevb, Joseph U. Igietsemec, Werner Lubitzda Department of Microbiology, Biochemistry and Immunology, Morehouse School of Medicine, 720 Westview Dr., S.W. Atlanta, GA 30310, USA

b Russian Anti-Plague Research Institute, Microbe, Universitetskaya St. 46, Saratov 410005, Russiac National Center for Infectious Diseases, CDC, Atlanta, GA, USA

d Institute of Microbiology and Genetics, University of Vienna, Althanstraße 14, A-1090 Vienna, Austria

Received 8 January 2003; received in revised form 1 April 2003; accepted 12 May 2003

Abstract

An effectiveVibrio cholerae vaccine is needed to reduce the morbidity and mortality caused by this pathogen. Despite the availability ofcurrent oral vaccines with measurable efficacy, there is need for more effective vaccines with broad-spectrum efficacy in target populations.Recent studies have shown that bacterial ghosts, produced by the expression of cloned lysis geneE, possess adjuvant properties and areimmunogenic. In this study, ghosts were prepared fromV. cholerae O1 or O139 and evaluated as vaccines in the reversible intestinal tieadult rabbit diarrhea (RITARD) model. Rabbits were orally immunized with different doses ofV. cholerae ghost (VCG) formulations.The vaccine formulations elicited high levels of serum vibriocidal titers against indicator strains. The magnitude of the response wasmeasured as the geometric mean titer (GMT) increase for all rabbits in relation to prevaccination titers. The induction of cross protectionwas evidenced by the ability of serum from VCG-immunized rabbits to mediate complement-dependent killing of both the homologousand the heterologous strains. Immunized rabbits were protected against intraduodenal challenge 30 days after primary immunization.Protective immunity against challenge appeared to be dose dependent and was associated with marked inhibition of colonization. Theseresults indicate that VCGs represent a novel approach to cholera vaccine development and constitute an effective vaccine delivery vehicle.© 2003 Elsevier Science Ltd. All rights reserved.

Keywords: VCG-TCP; Vaccine; Protection

1. Introduction

Cholera is an acute diarrheal disease caused byVibriocholerae. It is still a major cause of morbidity and mor-tality in many parts of the world[1]. Although even acuteinfection can be treated by oral and/or intravenous rehydra-tion [2], treatment centers are still scarce in many endemicregions. The long-term control of cholera depends on theimprovement of hygiene but this is a distant goal for manydeveloping countries[3]. The availability of an effectivecholera vaccine would therefore, be a realistic means tocontrol the disease in endemic regions.

The current challenges in cholera vaccine design includedetermination of the relative value of live versus non-livingvaccine regimens and the need for heterotypic protection

∗ Corresponding author. Tel.:+1-404-752-1584; fax:+1-404-752-1179.E-mail address: feko@msm.edu (F.O. Eko).

across serogroups. Despite the potential of both live attenu-ated and inactivated vaccines, neither approach has yet de-livered the ideal cholera vaccine. Experience with parenteralwhole-cell cholera and toxoid vaccines has been disappoint-ing, with the most effective formulations providing onlytransient protection[4]. Presently, two types of oral choleravaccines have been approved for human use; a combinationof cholera toxin B-subunit and killed-whole-cells[5] anda genetically engineered, live attenuatedV. cholerae vac-cine, CVD 103-HgR[6]. Other live attenuatedV. choleraevaccines at various stages of development and evaluationinclude Peru-15[7] and strain 638[8].

The recent emergence of the O139 serogroup ofV.cholerae with epidemic potential poses a new problem incholera vaccine development. AlthoughV. cholerae O139shares many similar characteristics withV. cholerae O1 andproduces a diarrheal disease indistinguishable from choleradue to O1 strains, immunity to pre-existing O1 strains

0264-410X/$ – see front matter © 2003 Elsevier Science Ltd. All rights reserved.doi:10.1016/S0264-410X(03)00388-8

3664 F.O. Eko et al. / Vaccine 21 (2003) 3663–3674

offers no protection against infection byV. cholerae O139[9,10]. This further indicates a need to develop vaccines tocombat the latter. Taken together, the problems limiting thewidespread use of existing cholera vaccines, especially inregions where the disease is endemic make the developmentof a vaccine based on the use of inert and non-replicatingVibrio antigens a desirable approach to cholera vaccinedesign.

The genetic inactivation of pathogenic Gram-negativebacteria by the controlled expression of cloned bacte-riophage PhiX174 lysis geneE offers a promising newapproach in non-living vaccine technology[11,12]. Expres-sion of plasmid-encoded geneE leads to the formation of atransmembrane tunnel structure through the cell envelopeof Gram-negative bacteria that results in the loss of cyto-plasmic contents. The resulting bacterial ghosts share thefunctional and antigenic determinants of the envelope withtheir living counterparts[13]. RecombinantV. choleraeghosts (rVCG) are attractive for use as non-living vaccinesbecause they are non-toxic, possess intrinsic adjuvant prop-erties, maintain the structural and functional integrity ofexpressed antigens, and are excellent vehicles for deliveryof foreign or heterologous proteins and other antigens ofvaccine relevance to the primary antigen-presenting cells[11]. It has been demonstrated that the cell targeting andadjuvant properties of ghosts are superior to Alum andcomplete Freund’s adjuvant[12]. VCGs induce signifi-cant seroconversion toVibrio antigens in mice[14] andimmunization with toxin-coregulated pilus (TCP)-positiveVCG conferred protection against heterologous challengeinfection[15]. TCP have been shown to be critical for colo-nization by pathogenic strains of O1 and O139 serogroups,both in the infant mouse cholera model (IMCM)[16–20]and in human volunteers[21,22].

In this study, VCG from strains of O1 or O139 serogroupwere produced under culture conditions conducive to, or un-favorable for, expression of TCP. Following immunizationwith TCP-positive or -negative VCG, the vibriocidal anti-body responses to both homologous and heterologous strainsof V. cholerae were assessed. The efficacy of each vaccineformulation was then determined by challenging immunizedrabbits by the reversible intestinal tie adult rabbit diarrhea(RITARD) model. The result revealed that regardless of theTCP status of the VCG preparations, all immunized rab-bits responded with significant increases in geometric meantiter (GMT) of vibriocidal antibodies. Furthermore, protec-

Table 1Characteristics ofV. cholerae strains used in this study

Strain designation Description Source Other featuresa

H1 O1 El Tor Ogawa Field isolate, 1985 toxR+, ctxA+, tcpA+P-3122 O1 El Tor Ogawa Cholera patient, 1970 toxR+, ctxA+, tcpA+AI-1838 O139 Bengal Field isolate, 1993 toxR+, ctxA+, tcpA+P-16064 O139 Bengal Cholera patient, 1993 toxR+, ctxA+, tcpA+

a Determined by PCR.

tive immunity against challenge was associated with markedinhibition of colonization.

2. Materials and methods

2.1. Animals

Locally supplied New Zealand White adult rabbits of bothsexes (weighing 2.0–2.5 kg at the onset of the experiments)were used in this study. They were maintained and allowedto acclimate in our animal care facility for at least 1 weekand were used only if they showed no evidence of diarrhealdisease during that time period.

2.2. Bacteria

V. cholerae O1 El Tor strain H1 and O139 strain AI-1838(Table 1) are laboratory strains used for production of VCG.Fully virulent V. cholerae O1 El Tor Ogawa, strain P-3122and O139 strain P-16064 were used to challenge immunizedanimals. Stock cultures were suspended in Brain Heart In-fusion (BHI) broth with 15% (w/v) glycerol and stored inworking samples at−70◦C. Before use, strains were grownin alkaline peptone broth (1% peptone, 0.5% NaCl, 0.25%Na2CO3, 0.01% KNO3, pH 8.4) at 37◦C for 6 h. A loopfulof this culture was then plated on alkaline nutrient agar andgrown at 37◦C for 16 h. To prepare strains for challenge,the S-form colonies were harvested and suspended in PBSto a density of 2× 109 cfu/ml.

2.3. Production of V. cholerae ghosts (VCG)expressing TCP

Lyophilized TCP-positive and -negative VCG were pro-duced essentially as described previously[15]. Briefly, V.cholerae strain H1 or AI-1838 each harboring the lysisplasmid pDKLO1 was induced for lysis after growth underculture conditions, which favor or repress the production oftoxin-coregulated pili. Bacteria were grown either in BHIbroth at 37◦C or in AKI medium[23] at 30◦C, conditionswhich repress or promote expression of TCP[24]. The AKImedium consisted of NaCl (0.5%), Bactopeptone (Difco,1.5%), yeast extract (Difco, 0.4%) and NaHCO3 (0.15%) atpH 7.0. The NaHCO3 was added to the autoclaved mediumfrom a freshly prepared, filter-sterilized stock (7.5 mg/ml)

F.O. Eko et al. / Vaccine 21 (2003) 3663–3674 3665

just before use. At the end of lysis, TCP-positive or -negativeghosts were harvested, washed and resuspended in PBS andthen lyophilized.

The efficiency of E-mediated killing ofV. choleraebacteria was estimated by plating samples of appropriatedilutions of freshly harvested and lyophilized VCG onBHI agar as previously described[25] and results werecompared with those from samples obtained prior to on-set of lysis. Results indicated a 100% killing efficiencyas no colony-forming units were found on plates withlyophilized VCG preparations at any dilution. LyophilizedVCG were weighed and the number of cfu/mg of VCG wascalculated.

2.4. Immunoblot analysis

Immunochemical detection of TCP on nitrocellulose pa-per was carried out as described previously[24]. Briefly,proteins were separated by SDS-PAGE and electrophoret-ically transferred to nitrocellulose in a semidry transblotcell (Bio-Rad). After blocking, the membrane was in-cubated with a highly absorbed rabbit anti-TCP serumkindly provided by Dr. Stephen R. Attridge, Departmentof Molecular Biosciences, The University of Adelaide,Australia. Bound antibody was detected by incubating theblot with HRP-conjugated goat anti-rabbit IgG, and de-veloped by immersion in 3-amino-9-ethylcarbazole (AEC)substrate reagent. The reaction was stopped by rinsing withTBS.

2.5. Immunization with VCG and collectionof blood samples

VCG preparations from the O1 El Tor strain H1 and O139strain AI-1838 were reconstituted with PBS. The result-ing suspensions were standardized on the basis of opticaldensity, such that 1 mg of lyophilized VCG correspondedto about 1× 109 to 2× 109 cfu/ml [14]. Two independentvaccination experiments were performed (Table 2). In ex-periment A, groups of rabbits (three or five per group) wereimmunized twice intragastrically (i.g.), 2 weeks apart, withdifferent doses (4 or 20 mg) of O1 VCG with or withoutTCP (VCG/O1-TCP or VCG/O1) in 5 ml of PBS. Anothergroup was immunized with a mixture consisting of 2 mgeach of VCG/O1-TCP and VCG/O139-TCP while onegroup each received, respectively, 4 mg of VCG/O1-TCP orVCG/O1 intramuscularly (i.m.). In addition, another groupreceived the whole-cell component (WC) of the non-livingB-subunit-whole-cell vaccine (B-WC) produced by the In-stitute Merieux, France in collaboration with the NationalBacteriological Laboratory of Sweden (SBL), which con-stituted the positive control. The vaccine was reconstitutedwith PBS and administered i.g. Groups that received PBS(5 ml) either i.g. or i.m. represented the negative controls(groups 7 and 10, respectively). In experiment B, groups ofrabbits were immunized with VCG/O139 with or without

Table 2Vaccine composition and immunization dose

Group No. ofrabbits

Vaccine formulation Dose Route

Experiment A1 5 VCG/O1-TCP 4 mg i.g.2 3 VCG/O1-TCP 20 mg i.g.3 5 VCG/O1 4 mg i.g.4 3 VCG/O1 20 mg i.g.5 5 VCG/O1-TCP

+ VCG/O139-TCP4 mg i.g.

6 3 WC-vaccine 1 mga i.g.7 5 PBS 5 ml i.g.8 5 VCG/O1-TCP 4 mg i.m.9 5 VCG/O1 4 mg i.m.

10 3 PBS 5 ml i.m.

Experiment B1 5 VCG/O139-TCP 4 mg i.g.2 3 VCG/O139-TCP 20 mg i.g.3 5 VCG/O139 4 mg i.g.4 3 VCG/O139 20 mg i.g.5 5 VCG/O1+ VCG/O139 4 mg i.g.6 4 PBS 5 ml i.g.7 5 VCG/O139-TCP 4 mg i.m.8 5 VCG/O139 4 mg i.m.9 3 PBS 5 ml i.m.

a A dose of this vaccine contained 1× 1011 heat- or formalin-killedV. cholerae cells representing three different strains.

TCP as described. An additional group was immunizedwith a mixture consisting of 2 mg each of VCG/O1 andVCG/O139.

Blood specimens were collected before the first immu-nization (day 0) and 2 weeks after each vaccine dose (days14 and 28).

2.6. Challenge of rabbits by the RITARD technique

About 2 weeks after the booster dose (day 30), rabbitswere challenged intraduodenally with 1× 109 cfu of fullyvirulent V. cholerae O1 (El Tor) or O139 strains by the re-movable intestinal tie adult rabbit diarrhea model describedpreviously [26] with modifications as described by Ahrenand Svennerholm[27]. This dose was previously determinedto cause severe diarrhea and death in 100% of non-immunerabbits.

2.7. Monitoring of infection and disease

For up to 5 days after the initial challenge, rabbits wereobserved for diarrhea and other symptoms of disease orfor death. Diarrhea was defined as the passage of at leasttwo loose stools or one watery stool per day. Rectal swabswere collected daily from each animal and plated on thio-sulfate citrate bile salts sucrose (TCBS) and/or blood agarfor identification of the challenge organism. Growth ofV. cholerae in these cultures was determined by charac-teristic colony morphology, hemagglutination pattern and

3666 F.O. Eko et al. / Vaccine 21 (2003) 3663–3674

agglutination with serotype-specific polyclonal antisera[28].

2.8. Determination of intestinal colonization

Animals that developed very severe diarrhea and pre-sented with signs of rapid deterioration, as well as thosealive after 120 h, were sacrificed by intravenous injectionof a lethal dose of pentobarbital. The distal ileum wasligated after the abdomen was opened and the intestinefreed from the mesentery. Segments of the small intes-tine (10 cm) were cut and transferred to sterile flasks.The segments were opened longitudinally, washed twicein sterile PBS, weighed and then homogenized in a Bio-homogenizer (BioSpec Products Inc., Bartlesville, OK,USA). Viable counts were determined by culturing seriallydiluted homogenates on nutrient agar for enumeration ofbacterial load. Isolated colonies were confirmed by slideagglutination with appropriate antiserum. The number ofbacteria recovered per gram of intestine was then deter-mined and expressed as the mean of the number of bacteria(±S.D.).

2.9. Vibriocidal assay

Antisera to VCG were assayed for vibriocidal antibodiesagainst homologous and heterologous strains ofV. choleraefollowing an assay protocol described previously[14,15,29]with modifications as described. Briefly, pooled sera werediluted 1:5 in PBS and treated with a 50% suspension offormalized sheep red blood cells for 1 h, to absorb het-erologous antibodies, following inactivation at 56◦C for30 min. The supernatant fluid was then serially diluted (1:5to 1:1280) in the wells of 96-well microtitre plates using0.1% peptone-saline supplemented with a 1× 10−3 dilutionof Kolmer solution (4% CaCl2·2H2O, 10% MgCl2·6H2O).All samples were tested in duplicates. An equal volumeof a chilled suspension comprising indicator bacteria (ca.4 × 105 ml−1) and complement (20% (v/v) guinea-pigserum) was then added. A standard serum was includedin each assay as a positive control. After incubation at37◦C for 60 min, BHI medium was added to the wellsand the plates incubated for another 4 h. Titers were de-termined by measuring the absorbance of the wells witha spectrophotometer (SLT 400 ATC ELISA reader; SLTLab Instruments, Austria) at 405 nm. The geometric meansof the vibriocidal titers were recorded as the reciprocal ofthe highest serum dilution that resulted in definite growthinhibition.

2.10. Statistical analysis

GMTs within an experimental group and between differ-ent experimental groups were compared by Student’st-testor χ2-test and the level of significance was judged atP <

0.05.

3. Results

3.1. Characterization of VCG

To study the morphological and structural integrity ofVCG, samples ofV. cholerae harboring the lysis plasmid,pDKLO1 were removed at the initiation of lysis induc-tion and at the end of lysis, and processed for electronmicroscopy. Electron micrographs ofV. cholerae cells andVCG are shown inFig. 1. The cytoplasmic content of unl-ysed cells can be detected as electron dense material intransmission electron micrographs (Fig. 1C). The charac-teristic comma-shapedVibrio morphology (Fig. 1A and C)remained intact in VCG (Fig. 1B and D).

To ascertain the TCP status of the H1 and AI-1838 VCGs,immunoblotting analysis was performed. Pilin monomerbands were only seen following growth in AKI medium at30◦C (Fig. 2), conditions known to favor the expressionof TCP [24]. A comparative analysis of the lipopolysac-charide and outer membrane protein profiles ofV. choleraeand VCG has previously been reported[15]. VCG possessthe normal array of cell surface antigens of live bacteria, inparticular those of greatest vaccine significance.

3.2. Protection against infection and colonization

Rabbits were immunized with different doses of VCG for-mulations and challenged with liveV. cholerae O1 or O139.Most of the immunized rabbits were protected from symp-toms of disease and remained alive following challenge with109 Vibrio cells of either P-3122 (O1) or P-16064 (O139),whereas a similar challenge dose caused fatal diarrheal dis-ease in PBS-immunized control rabbits within 24 h. Threeof the five (60%) rabbits immunized with the vaccine formu-lation consisting of a mixture of VCG-O1 and VCG-O139got diarrhea following challenge with strain P-16064. Theserabbits also excretedVibrio as confirmed by positive culturesfrom rectal swabs. None of the postchallenge rectal swabsfrom rabbits immunized with the different VCG formula-tions at the 20 mg dose or WC-vaccine (positive control)were positive forV. cholerae, whereas 100% of similarlychallenged control rabbits excretedVibrio within 24–30 h(Table 3).

The capacity of the vaccine candidates to inhibit intestinalcolonization following challenge was studied by perform-ing viable counts of cultured serial dilutions of intestinalhomogenates. The isolation of high numbers ofV. choleraefrom cultured intestinal segments (∼108 cfu/g of intestine)of control rabbits but not from most of the VCG-immunizedrabbits showed that the challenge strains could colonize andmultiply in the rabbit intestine (Table 3). Immunization withthe WC-vaccine also inhibited intestinal colonization fol-lowing challenge with P-3122. Thus, immunization with thevarious vaccine formulations with or without TCP protectedrabbits against excretion of the challenge strain and pre-vented intestinal colonization.

F.O. Eko et al. / Vaccine 21 (2003) 3663–3674 3667

Fig. 1. Electron micrographs of E-lysedV. cholerae cells. (A) High resolution field emission scanning electron micrograph (FESEM) ofV. cholerae cellsjust before lysis. (B) High resolution FESEM ofV. cholerae after protein E-lysis, showing empty bacterial cells (VCG). Note the ‘deflated bag-shaped’ghost cells. HarvestedVibrio and VCGs were washed with PBS, fixed for 60 min at room temperature with 2.5% glutaraldehyde in 75 mM sodiumcacodylate 1 mM MgCl2 (pH 7.2) and stored in the cold. Cells were rinsed with the same buffer and postfixed in 1% osmium tetroxide for 30 min beforefurther processing. Cells for en bloc staining were incubated with 1% uranyl acetate in 20% acetone for 30 min and dehydrated with a graded seriesof acetone solutions. Following final dehydration with 100% acetone, cells were dried with liquid CO2, mounted with Tempfix on copper foils, andsputtered with gold–palladium using a Polaron high-resolution sputterer. All scanning electron micrographs were taken with a Hitachi S-800 field emissionscanning electron microscope. (C) Transmission electron micrograph of ultrathin sections ofV. cholerae cells before lysis. Cytoplasmic material is seenas electron dense dark material filling the cytoplasm. (D) Transmission electron micrograph of ultrathin sections ofV. cholerae cells after E-mediatedlysis. Notice the empty cells devoid of cytoplasmic contents. The inner membranes (im) and outer membranes (om) are clearly visible. For transmissionelectron micrographs,V. cholerae and VCGs were fixed and dehydrated as described. Cells were then infiltrated and embedded in Spur low-viscosityresin, polymerized, and cut with a diamond knife on a Reichert–Jung Ultracut E Ultramicrotome. Sections were mounted on uncoated copper grids andpoststained with 3% aqueous lead citrate (pH 13). Transmission electron micrographs were taken with an Elmiscope 101 (Siemens) electron microscope.

3668 F.O. Eko et al. / Vaccine 21 (2003) 3663–3674

Fig. 1. (Continued ).

F.O. Eko et al. / Vaccine 21 (2003) 3663–3674 3669

Fig. 2. TCP status ofV. cholerae O1 and O139 strains used for VCG production.V. cholerae cell lysates were subjected to SDS-PAGE, transferred tonitrocellulose and incubated with serum containing antibodies to TCP. Cell lysates from the O1 and O139Vibrio were prepared from bacteria grown inAKI medium at 30◦C (lanes: (1) H1 and (3) AI-1838) or BHI broth at 37◦C (lanes: (2) H1 and (4) AI-1838).

3.3. Characterization of the serum vibriocidal antibodyresponses induced by immunization with VCG alone andby ghosts expressing TCP

A vibriocidal antibody response is used as a predictor ofprotection in cholera[4]. This assay measures the killingof V. cholerae cells in the presence of immune sera andcomplement. The ability of pooled immune serum from thevarious groups of VCG-immunized and control rabbits toeffect complement-mediated killing ofV. cholerae cells was,therefore, assessed (Figs. 3–6). For the purpose of calcula-tions, sera with a titer≤5 were assigned a titer value= 0.

Pre-immunization serum exhibited no detectable vibri-ocidal antibody titers confirming that the inbred rabbitsused in this study had not been exposed toV. choleraeantigens or antigenically related organisms in the envi-ronment. Similarly, there were no detectable vibriocidalantibody titers from the serum of control rabbits that hadbeen immunized with PBS and were, therefore excludedfrom the figures (Figs. 3–5). Regardless of the TCP sta-tus of the VCG preparations used for immunization, allrabbits responded with significant increases in geometric

Table 3Protection against diarrhea and colonization in VCG-immunized rabbits following challenge withV. cholerae in the RITARD model

Vaccine Challenge strain No. responding/total no. of animals Vibrio colonization (no. of vibrios)

Diarrhea Excretion of vibrios >24 h Inoculated Recoveredb

VCG-O1/TCP (20 mg) P-3122 0/5 0/5 1.0× 109 0VCG-O1 (20 mg) P-3122 0/5 0/5 1.0× 109 0VCG-O139/TCP (20 mg) P-16064 0/5 0/5 1.0× 109 0VCG-O139 (20 mg) P-16064 0/5 0/5 1.0× 109 0VCG-O1/TCP+ VCG-O139/TCP (40 mg)a P-3122 0/5 0/5 1.0× 109 0VCG-O1 + VCG-O139 (40 mg)a P-16064 3/5 3/5 1.0× 109 6.2 × 104 (±0.90)WC-vaccine (10 mg) P-3122 0/3 0/3 1.0× 109 0PBS (5 ml) P-3122 5/5 5/5 1.0× 109 4.85 × 108 (±0.60)PBS (5 ml) P-16064 4/4 4/4 1.0× 109 1.82 × 108 (±0.71)

a These vaccine candidates consisted of a mixture of 2 mg each of VCG/O1 and VCG/O139 with or without TCP.b Each value represents the mean number (±S.D.) of Vibrio recovered per gram of intestinal wash or homogenate.

mean titer of serum vibriocidal antibodies. The inductionof cross protection was evidenced by the ability of serumfrom rabbits immunized with either the O1 or O139 to me-diate complement-dependent killing of both homologousand heterologous strains. The vibriocidal titers were in mostcases similar against the homologous and the heterologousstrains, with or without TCP. The GMT of post-immuneserum of rabbits immunized with 4 mg of VCG (groups 3and 5, experiment A) ranged from 20 to 40 (Fig. 3). Thepresence of TCP resulted in a 10- and 12-fold increase inGMT against the homologous O1 strain on days 14 and 28,respectively. Also, while there was no difference in serumvibriocidal activity against the heterologous O139Vibrioin the presence or absence of TCP on day 14, there wasa 12-fold increase in activity on day 28. Rabbits that hadbeen vaccinated with 20 mg of VCG/O1 had significantlyhigher post-immunization vibriocidal reciprocal GMT (peakGMT = 140) than those vaccinated with the 4 mg dose(peak GMT= 40; P < 0.05; Figs. 3 and 4), suggesting adose-dependent response. On day 28, the GMT of this groupagainst O1 and O139Vibrio increased 76- and 64-fold, re-spectively, in the presence of TCP. For rabbits immunized

3670 F.O. Eko et al. / Vaccine 21 (2003) 3663–3674

Fig. 3. Serum vibriocidal antibody responses evoked in rabbits followingimmunization with 4 mg of VCG/O1 formulations. Rabbits were immu-nized and boosted 2 weeks later with the vaccine formulations. Serumsamples were collected on day 0 and 2 weeks after each vaccine dose.Pooled pre- (day 0) or post-immunization antiserum (days 14 and 28)serially diluted in 0.1% peptone-saline supplemented with a 1:1000 dilu-tion of Kolmer solution was tested for vibriocidal activity as described inSection 2. The data shown represent the geometric means of the endpointtiters for all rabbits in relation to prevaccination titers (day 0). Immu-nization of rabbits with PBS did not evoke measurable serum vibriocidalantibody titers and data were therefore excluded from the figures.

Fig. 4. Serum vibriocidal antibody responses evoked in rabbits follow-ing immunization with 20 mg of VCG/O1 formulations. Rabbits wereimmunized and serum samples were collected, pooled and analyzed asdescribed inFig. 3. Immunization of rabbits with PBS did not induceserum vibriocidal antibody titers (data not shown).

Fig. 5. Serum vibriocidal antibody responses evoked in rabbits followingimmunization with 4 mg of VCG/O139 formulations. Rabbits were im-munized and boosted 2 weeks later with the vaccine formulations. Serumsamples were collected on day 0 and 2 weeks after each vaccine dose.Pooled pre- (day 0) or post-immunization antiserum (days 14 and 28)serially diluted in 0.1% peptone-saline supplemented with a 1:1000 dilu-tion of Kolmer solution was tested for vibriocidal activity as described inSection 2. The data shown represent the geometric means of the endpointtiters for all rabbits in relation to prevaccination titers (day 0). Immu-nization of rabbits with PBS did not induce detectable levels of serumvibriocidal antibody titers (data not shown).

with 4 mg of VCG/O139, there was no consistent differencein the GM vibriocidal titers to the O1 and O139 serotypeson days 14 and 28 and the expression of TCP did not affectthe magnitude of the response to the O139Vibrio (Figs. 3and 5). The GMT of rabbits immunized with the controlWC-vaccine is shown inFig. 6. Titers were comparable tothose obtained on day 14 with rabbits immunized with 4 mgof VCG.

3.4. Induction of protective immunity followingimmunization with VCG vaccine formulations

The protective effect of immunization with the variousformulations of VCG/O1 against overt or fatal diarrheawas studied in the RITARD model. Immunized rabbits andPBS-vaccinated controls were challenged intraduodenallywith 2 × 109 organisms ofV. cholerae O1 P-3122 (bio-type El Tor, serotype Ogawa) or the O139 serogroup strainP-16064. The effect of immunization with VCG/O1 withor without TCP on the mean survival of rabbits followingintraduodenal challenge is shown inFig. 7. When micewere challenged with 2× 109 organisms ofV. cholerae O1P-3122, mice immunized with the VCG/O1 formulationssurvived for periods ranging from 60 to 120 h compared tothe PBS-immunized mice that survived for less than 24 h.Overall, the candidate vaccines, irrespective of their TCPstatus, showed significant protection against challenge. Thedifference in protection afforded by the presence of TCP in

F.O. Eko et al. / Vaccine 21 (2003) 3663–3674 3671

Fig. 6. Serum vibriocidal antibody responses evoked in rabbits followingimmunization with the WC-vaccine. Rabbits were immunized and serumsamples were collected and evaluated as described inFig. 5.

the vaccine candidates was more pronounced at the 20 mgdose; there was a significant inverse linear relationship be-tween the time to death and the administered dose (P <

0.05). Also, the protective effect of the vaccine candidatecomposed of 2 mg each of a mixture of TCP-expressingVCG of the O1 and O139 serogroups was found to besuperior to that of the 4 mg dose of either VCG/O1-TCPor VCG/O1 given orally or intramuscularly (Fig. 7). Thelevel of protection induced by the VCG formulations at the4 mg dose was comparable to that of the WC-vaccine; atthe 20 mg dose, however, protection afforded by the ghostformulations was significantly higher.

Fig. 7. Protection against experimental cholera after immunization withVCG/O1 formulations and challenge with virulentV. cholerae O1 strain.Rabbits were immunized twice, orally or i.m. 2 weeks apart and challengedwith V. cholerae O1 (Ogawa) strain P-3122 by the RITARD technique30 days after primary immunization (i.e. about 2 weeks after the boosterdose). The survival rates were monitored at 6 h intervals for 120 h. Rabbitsthat survived after 5 days were euthanized by an overdose of Nembutal.Rabbits immunized with PBS and challenged with P-3122 did not survivefor more than 24 h.

Fig. 8. Protection against experimental cholera after immunization withVCG/O139 formulations and challenge with virulentV. cholerae O139strain P-16064. Rabbits were immunized, challenged and monitored asdescribed inFig. 7. Rabbits immunized with PBS and challenged withP-16064 did not survive for more than 24 h.

Rabbits immunized with the VCG/O139 candidate vac-cines were also assessed for protection following challengewith V. cholerae O139 strain P-16064. The PBS-immunizedcontrol rabbits survived for only 24 h following challengeand were, therefore, not protected (Fig. 8). The level ofprotection induced by the various vaccine formulations wasdose dependent and was not significantly affected by thepresence or absence of TCP. In contrast to the O1 challenge,the vaccine candidate composed of 2 mg each of a mixtureof VCG of the O1 and O139 serogroups was not as effec-tive as the 4 mg dose (either orally or intramuscularly) inprotecting against an O139 challenge.

4. Discussion

Adequate supplies of uncontaminated food and wa-ter together with proper sanitary measures would controlcholera. However, this is not feasible in many affected ar-eas. An effective and economical cholera vaccine, whichwould mitigate the spread of cholera and stabilize areas ofendemicity, would be an alternative means for controllingthe disease. The parenterally administered cholera vaccine,made up of phenol-killedV. cholerae organisms[30], pro-vides approximately 50% protection for only a short dura-tion. Thus, there is need for new oral cholera vaccines thatconfer reliable and long-term protection in all age groups,especially in endemic settings. The inactivated vaccineformulation, consisting of a combination of cholera toxinB-subunit and killed-whole-cells has proven to be effectivein conferring sustained protection in adults although youngchildren, the most important target population for a choleravaccine, were not well protected by this vaccine[31]. Asingle dose of the live attenuatedV. cholerae vaccine CVD103-HgR demonstrated good protection againstV. cholerae

3672 F.O. Eko et al. / Vaccine 21 (2003) 3663–3674

O1 challenge in immunologically naive human volunteers[6,32] although it was less effective in a large-scale fieldtrial in an endemic setting[33]. Also, since immunity tothe pre-existing O1 strains offers no protection against in-fection byV. cholerae O139, a need for vaccines to combatthe latter is clearly indicated. The development of a choleravaccine must take into account the target population andthe current biotype and serotype of the endemic or pan-demic organism. The bacterial ghost platform technology isa novel vaccine delivery system endowed with intrinsic ad-juvant properties as well as carrier and targeting functions[34]. These ghosts can be regarded as naturally engineeredliposomes, being composed of two membranes separatedby a distinct space in which the rigid peptidoglycan corsetand membrane-derived oligosaccharides are located[35].

The objective of this study was to investigate the im-munogenicity and effectiveness of VCG as oral vaccine can-didates against challenge with the O1 and O139 strains ofV.cholerae in the RITARD cholera model. The effect of TCPexpression on the immunogenicity and ability of VCG-TCPto confer heterotypic protection against both O1 and O139was also evaluated. Initial vaccine evaluation was based onthe measurement of serum vibriocidal antibodies because oftheir correlation with protection in cholera[36]. Using thismodel, we demonstrated that immunization with the differ-ent VCG preparations induced significant increases in serumvibriocidal antibodies. The magnitude of response wasmeasured as the GMT increases for all post-immunizationsera in relation to prevaccination titers. The induction ofcross protection was evidenced by the ability of serum fromVCG-immunized rabbits to mediate complement-dependentkilling of homologous and heterologous strains. Allpost-immunization sera showed strong bactericidal activityagainst both homologous and heterologousVibrio (GMT rise≥20-fold). This contrasts with a previous report in whichonly marginal responses were recorded against the heterolo-gous indicator strain, in the absence of TCP[15]. The reasonfor this difference is not clear but may be related to the assaydiluent and the lower bacterial concentration used in thisstudy. The critical impact of assay diluent on the efficiencyof complement activation and the reduction in lytic titers re-sulting from high concentrations of indicator bacteria haverecently been reported[9,37]. In general, the presence ofTCP in the vaccine formulations increased the vibriocidalactivity of immune serum from VCG/O1-immunized rab-bits, especially on day 28. This effect was not so obvious inthe case of the O139 vaccine, at the 4 mg dose. The reasonsfor this observation are unclear. However, TCP expressionby live attenuated, oralV. cholerae O1 vaccine strains hasbeen reported to increase the vibriocidal antibody response[21] possibly through some colonization mechanism[38].

Previous reports have shown that toxin-coregulated pili(TCP) induce protection against both the O1 and O139serogroups in infant mice[20]. We have recently shown thatantibodies raised against VCG expressing TCP are immuno-genic and confer passive immunity to challenge against both

homologous and heterologous strains ofV. cholerae in the in-fant mouse cholera model (IMCM)[14,15]. Although thesepili are only weakly immunogenic in man[38], it has beensuggested that their inclusion in vaccine formulations mayfurther enhance their protective efficacy by promoting mu-cosal immunity[9,16,21,39]. Our results demonstrate thatVCG candidate vaccines, intended for human use, can beevaluated in the RITARD cholera model. The data showthat immunization with the different VCG formulations sig-nificantly protected rabbits against challenge with liveV.cholerae organisms. Protective immunity against challengeappeared to be dose dependent and was associated withmarked inhibition of colonization. Furthermore, the impactof TCP on the protective efficacy of the O1 vaccine candidatewas most pronounced at the 20 mg dose. The results alsoshowed that all rabbits that received the vaccine formulationrepresented by the experimental group A5 were protectedfrom symptoms of disease and remained alive, whereas threerabbits in group B5 developed clinical signs of diarrhea andone died (Tables 2 and 3). These rabbits excretedVibrioorganisms following challenge but the numbers of bacteriaisolated from washed intestinal homogenates were signifi-cantly lower than those obtained from the PBS-immunizedcontrol groups. The difference between the levels of pro-tection afforded by both vaccine regimens may be partiallyexplained by the presence of TCP in the one and its ab-sence in the other. Although TCP are immunogenic in rab-bits when either purified pili or piliatedVibrio [40] or VCGexpressing TCP[15] are administered intravenously or sub-cutaneously, it is yet unknown if TCP is required for colo-nization of rabbit intestine. While a repertoire of antibodies(including anti-LPS, anti-OMP, anti-TCP) is known to me-diate the inhibition of intestinal colonization in mice andhumans[4], the specific factors that induce the expressionof these colonization–inhibition antibodies in rabbits havenot been completely delineated. However, the observationthat an intestinal infection-immunization of rabbits with arough mutant ofV. cholerae conferred significant protectiveanti-colonization immunity against challenge with fully vir-ulent smooth cholera bacteria in the absence of any anti-LPSantibody response[41] suggests that antigens, other thanLPS, play a role in protective immunity associated with in-hibition of colonization in rabbits.

In a bid to determine the impact, if any, of the route ofadministration of our vaccine formulation, groups of rabbitswere also immunized via the intramuscular route. Signifi-cantly, there was no statistical difference between the pro-tective efficacy of the 4 mg dose vaccines delivered eitherorally or intramuscularly, with or without TCP (P < 0.05).Intramuscular immunization has recently been reported toinduce a high level of protective local mucosal immune re-sponse against a mucosal challenge infection[42]. While theultimate test of protective efficacy afforded by these vaccineformulations would be a field trial in a cholera endemic area,the promising findings obtained with particularly the vac-cine formulation consisting of 2 mg each of TCP-expressing

F.O. Eko et al. / Vaccine 21 (2003) 3663–3674 3673

VCG from the O1 and O139 serogroups, show that it is pos-sible to immunize against cholera due to both serogroupswith a single regimen. Thus, it is also possible that 2 mg ofVCG expressing TCP will separately be sufficient to induceprotective efficacy approximating that afforded by the 4 mgdose. However, the dose and immunization schedule androute of administration of a VCG vaccine in humans is yetto be determined; the amount of VCG antigen is likely todepend on the route and method of delivery.

It has long been established that both antibacterial andantitoxic immunity exist, with antibacterial immunity be-ing the most important component[4,43,44]. Results froma large-scale field trial in Bangladesh also demonstratedthat an oral vaccine consisting of killedV. cholerae or-ganisms could confer substantial protective immunity[44].Results from our study are consistent with previous re-ports, which show that antibacterial antibodies alone areprotective. It is noteworthy that immunizations were car-ried out with lyophilized ghosts resuspended in salinewithout the addition of adjuvants, stabilizers or other sub-stances. Thus, the use of lyophilized ghosts may eliminatethe need for refrigeration or other special cold chain stor-age requirements, which would otherwise pose a problemin developing countries. The substantial level of protec-tion demonstrated in these studies, by rabbits receivinga dose ofV. cholerae that caused diarrhea in 100% ofPBS-immunized controls, suggests that these vaccines mayalso provide protection for humans. These results indicatethat VCGs represent a novel approach to cholera vaccinedevelopment and constitute an effective vaccine deliveryvehicle.

Acknowledgements

This work was supported by grants P12209MOB,S07208MOB and InSan I0594-V-4397 (WL) from theAustrian FWF and Fraunhofer-Gesellschaft, Munichand in part by PHS grants AI41231, GM08248 andU50/CCU304522-11 (FOE) from the NIH and the Cen-ters for Disease Control and Prevention, CDC. We thankDr. Stephen R. Attridge for kindly providing the rabbitanti-serum to H1-TCP.

References

[1] Bern C, et al. The magnitude of the global problem of diarrhoealdisease: a ten-year update. Bull WHO 1992;70:705–14.

[2] Carpenter C. The treatment of cholera: clinical science on the bedside.J Infect Dis 1992;166:2–14.

[3] Mekalanos J, et al. Live cholera vaccines: perspectives on theirconstruction and safety. Bull Inst Pasteur 1995;93:255–62.

[4] Kaper J, Morris Jr J, Levine M. Cholera. Clin Microbiol Rev1995;8(1):48–86.

[5] Holmgren J, et al. The oral B subunit/inactivated whole cell vaccine.In: Woodrow G, Levine M, editors. New generation vaccines. NewYork: Marcel Dekker; 1990. p. 289–304.

[6] Levine M, et al. Safety, immunogenicity, and efficacy of recombinantlive oral cholera vaccines, CVD 103 and CVD 103-HgR. Lancet1988;ii:467–70.

[7] Cohen M, et al. Randomized, controlled human challenge study ofthe safety, immunogenicity, and protective efficacy of a single doseof Peru-15, a live attenuated oral cholera vaccine. Infect Immun2002;70(4):1965–70.

[8] Perez J, et al. Passive protection of serum from volunteers inoculatedwith attenuated strain 638 ofVibrio cholerae O1 in animal models.Vaccine 2001;19:376–84.

[9] Attridge S, et al. Susceptibility ofVibrio cholerae O139 toantibody-dependent, complement-mediated bacteriolysis. Clin DiagLab Immunol 2000;7(3):444–50.

[10] Albert M, et al. Lack of cross-protection against diarrhea due toVibrio cholerae O139 (Bengal strain) after oral immunization ofrabbits with V. cholerae O1 vaccine strain CVD103-HgR. J InfectDis 1994;169:230–1.

[11] Szostak MP, et al. Bacterial ghosts: non-living candidate vaccines.J Biotechnol 1996;44(1–3):161–70.

[12] Eko FO, et al. New strategies for combination vaccines based onthe extended recombinant bacterial ghost system. Vaccine 1999;17:1643–9.

[13] Witte A, et al. Dynamics of PhiX174 protein E mediated lysis ofEscherichia coli. Arch Microbiol 1992;157:381–8.

[14] Eko FO, et al. Immunogenicity ofVibrio cholerae ghosts followingintra-peritoneal immunization of mice. Vaccine 1994;12:1330–4.

[15] Eko FO, et al. Characterization and immunogenicity ofVibriocholerae ghosts expressing toxin-coregulated pili. J Biotechnol2000;83:115–23.

[16] Taylor R, et al. Use of phoA gene fusions to identify a piluscolonization factor coordinately regulated with cholera toxin. ProcNatl Acad Sci USA 1987;84:2833–7.

[17] Attridge SR, Voss E, Manning PA, et al. The role of toxin-coregulatedpili in the pathogenesis ofVibrio cholerae O1 El Tor. Microb Pathog1993;15:421–31.

[18] Attridge SR, et al. Relative significance of Mannose-sensitivehemagglutinin and toxin-coregulated pili in colonization of infantmice by Vibrio cholerae El Tor. Infect Immun 1996;64:3369–73.

[19] Rhine J, Taylor R. TcpA pilin sequences and colonizationrequirements for O1 and O139Vibrio cholerae. Mol Microbiol1994;13:1013–20.

[20] Voss E, Manning P, Attridge S. The toxin-coregulated pilus is acolonization factor and protective antigen ofVibrio cholerae El Tor.Microb Pathog 1996;20:141–53.

[21] Herrington D, et al. Toxin, toxin-coregulated pili, and the toxRregulon are essential forVibrio cholerae pathogenesis in humans. JExp Med 1988;168:1487–92.

[22] Tacket C, et al. Investigation of the roles of toxin-coregulated pili andmannose-sensitive hemagglutinin pili in the pathogenesis ofVibriocholerae O139 infection. Infect Immun 1998;66:692–5.

[23] Iwanaga M, Yamamoto KT. New medium for the production ofcholera toxin byVibrio cholerae O1 biotype El Tor. J Clin Microbiol1985;22:405–8.

[24] Voss E, Attridge S. In vitro production of toxin-coregulated pili byVibrio cholerae El Tor. Microb Pathog 1993;15:255–68.

[25] Eko FO. Production ofVibrio cholerae ghosts (VCG) by expressionof a cloned phage lysis gene: potential for vaccine development.Vaccine 1994;12:1231–7.

[26] Spira W, Sack R, Froehlich J. Simple adult rabbit model forVibriocholerae and enterotoxigenicEscherichia coli diarrhea. Infect Immun1981;32(2):739–47.

[27] Ahren C, Svennerholm A-M. Experimental enterotoxin-inducedEscherichia coli diarrhea and protection induced by previousinfection with bacteria of the same adhesin or enterotoxin type. InfectImmun 1985;50(1):255–61.

[28] Eko F, Udo S, Antia-Obong O. Epidemiology and spectrum of vibriodiarrheas in the lower Cross River basin of Nigeria. Centr Eur JPubl Hlth 1994;2(1):37–41.

3674 F.O. Eko et al. / Vaccine 21 (2003) 3663–3674

[29] Benenson A, Saad A, Mosely W. Serological studies in cholera. 2.The vibriocidal antibody response of cholera patients determined bya microtechnique. Bull WHO 1968;38:277–85.

[30] Feeley J, Gangarosa E. Field trials of cholera vaccine, in choleraand related diarrheas. In: Ouchterlony O, Holmgren J, editors.Proceedings of the 43rd Nobel Symposium. Basel: S. Karger; 1980.p. 204–10.

[31] Clemens J, et al. Field trial of oral cholera vaccines in Bangladesh:results from three year follow-up. Lancet 1990;335:270–3.

[32] Tacket C, et al. Randomized double-blind, placebo-controlled,multicentered trial of the efficacy of a single dose of live oralcholera vaccine CVD103-HgR in preventing cholera followingchallenge following challenge withVibrio cholerae O1 El TorInaba three months after vaccination. Infect Immun 1999;67:6341–5.

[33] Richie E, et al. Efficacy trial of single-dose live oral cholera vaccineCVD103-HgR in North Jakarta, Indonesia, a cholera-endemic area.Vaccine 2000;18:2399–410.

[34] Lubitz W. Bacterial ghosts as carrier and targeting systems. ExpertOpin Biol Ther 2001;1(5):765–71.

[35] Jechlinger W, et al. Bacterial ghosts as carrier and targeting systemsfor antigen delivery. In: Dietrich G, Goebel W, editors. Vaccinedelivery strategies. Wymondham: Horizon Scientific Press; 2002.p. 163–184.

[36] Levine M, Tacket C. Recombinant live cholera vaccines. In:Wachsmuth I, Blake P, Olsvik O, editors.Vibrio cholerae and cholera.

Molecular to global perspectives. Washington: ASM Press; 1994.p. 395–413.

[37] Attridge S, et al. Sensitive microplate assay for detection ofbactericidal antibodies toVibrio cholerae O139. Clin Diag LabImmunol 2002;2:383–7.

[38] Hall R, et al. Immunogenicity ofVibrio cholerae O1 toxin-coregulated pili in experimental and clinical cholera. Infect Immun1991;59:2508–12.

[39] Sun D, Mekalanos J, Taylor R. Antibodies directed againstthe toxin-coregulated pilus isolated fromVibrio cholerae provideprotection in the infant mouse experimental cholera model. J InfectDis 1990;161:1231–6.

[40] Hall R, et al. Morphological studies on fimbriae expressed byVibriocholerae O1. Microb Pathog 1988;4:257–65.

[41] Jonson G, Svennerholm A-M, Holmgren J.Vibrio cholerae expressescell surface antigens during intestinal infection which are notexpressed during in vitro culture. Infect Immun 1989;57(6):1809–15.

[42] Igietseme JU, Murdin A. Induction of protective immunity againstChlamydia trachomatis genital infection by a vaccine based on majorouter membrane protein-lipophilic immune response-stimulatingcomplexes. Infect Immun 2000;68(12):6798–806.

[43] Levine M, et al. Volunteer studies of deletion mutants ofVibriocholerae O1 prepared by recombinant techniques. Infect Immun1988;56:161–7.

[44] Clemens J, et al. Field trial of oral cholera vaccines in Bangladesh.Lancet 1986;ii:124–7.