INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

One hundred years after the discovery of the Shiga bacillus (later known as Shigella dysenteriae type 1) in Japan, shigellosis continues to be a major public health problem that kills hundreds of thousands of children in the developing world (18). The genus Shigella is now divided into four species or groups and at least 47 serotypes based on their biochemical and/or lipopolysaccharide (LPS) characteristics: S. dysenteriae (group A, 13 serotypes), S. flexneri (group B, 15 serotypes), S. boydii (group C, 18 serotypes), and S. sonnei (group D, 1 serotype). The World Health Organization and the Institute of Medicine (18, 42) consider the development of a vaccine against shigellosis a priority for developing countries. However, attainment of this goal has been hindered by the large number of serotypes, since it is thought that protective immunity is directed primarily against the Shigella O antigens and that protection is therefore serotype specific (6, 11, 13, 15, 17, 32). However, although Shigella spp. of any of the 47 serotypes are able to cause diarrhea and dysentery in humans, their prevalence is not evenly distributed. Of critical public health importance are S. sonnei, as the most prevalent Shigella spp. (with a unique serotype) in industrialized countries and of increasing prevalence in some Latin American countries (11); S. dysenteriae type 1, able to cause explosive pandemics resulting in high morbidity and mortality (1, 15, 23, 31); and S. flexneri, the most prevalent endemic group (comprising 15 serotypes) found in developing countries (7, 11, 20, 30). Therefore, although it may be impractical to construct a vaccine against all Shigella serotypes, a vaccine could be developed to protect against the most prevalent serotypes. Nonetheless, vaccine development still must address the 15 different serotypes of S. flexneri, which tend to be unevenly distributed in any given geographic area (7, 11, 20, 30). In this regard, while there are no significant cross-reactions among Shigella serotypes in groups A (S. dysenteriae, 13 serotypes) and C (S. boydii, 18 serotypes), there are major cross-reactions among 14 of the 15 serotypes included in Shigella group B (8). This is explained by the fact that the S. flexneri serotypes (with the exception of serotype 6) have some degree of antigenic relatedness attributable to a common repeating tetrasaccharide unit, -L-Rhap12--L-Rhap13--L-Rhap13-D-GlcpNAc1, to which -D-glucopyranosyl and O-acetyl groups are added, providing the basis for their "type" (i.e., I to VI) and "group" (i.e., 3,4, 6, and 7,8) antigenic factors (3, 8). Rabbit antisera raised against the specific type and group antigenic factors are routinely used by clinical microbiologists in agglutination reactions to identify the S. flexneri serotypes (8, 12). Van De Verg et al. (38) reported that challenge or immunization with S. flexneri 2a (type II, group 3,4) elicited cross-reacting antibodies with S. flexneri serotypes that bear the group factor 3,4 or the type factor II in humans, monkeys, and guinea pigs. Lindberg et al. (22) designed a strategy of cross-protection based on an attenuated S. flexneri Y strain which exclusively bears the common tetrasaccharide unit serologically identified as antigenic group factor 3,4. However, the fact that significant cross-protection between most S. flexneri serotypes has not been found is not surprising, given the antigenic variability conferred by their type and group factors (8, 12). For example, the addition of an O-acetyl group on the third rhamnose or of an -D-glucopyranosyl on the first rhamnose provides the antigenic group factors 6 and 7,8, respectively, which in many cases block the antigenicity of group factor 3,4. This is the case with S. flexneri serotypes 1b, 3a, 3c, and 4b (group factor 6) or serotypes 2b, 3a, 4c, 5b, and X (group factor 7,8) (Table 1). Therefore, based on the antigenic characteristics of the S. flexneri LPS presented above, we decided to investigate whether a mucosally administered attenuated vaccine consisting of a combination of serotypes 2a and 3a could protect against the rest of the S. flexneri serotypes. We selected S. flexneri serotypes 2a and 3a because together they bear the LPS group factors (3,4, 6, and 7,8) present in the S. flexneri group and two of the most prevalent type factors (II and III), providing the broadest spectrum with the minimum number of serotypes (Table 1). To test this strategy, we constructed attenuated strains of S. flexneri 2a and 3a based on the attenuation conferred by the previously published guaB-A and virG mutations (27) (Table 2). In addition, the recently described Shigella enterotoxin 1 (ShET1) (9, 10, 26) was genetically inactivated in the S. flexneri 2a vaccine candidate and Shigella enterotoxin 2 (ShET2) (25) was inactivated in both the S. flexneri 2a (strain CVD 1207) and 3a (strain CVD 1211) vaccine candidates (Table 2).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Bacterial strains and media. The strains used in this study are listed in Table 3. Wild-type Shigella strains were grown on Trypticase soy agar (TSA) (BBL Becton Dickinson, Cockeysville, Md.) with 0.01% Congo red dye (CR) (Sigma, St. Louis, Mo.) (TSA-CR). The same medium was supplemented with guanine (Sigma) (10 mg/liter) to grow the guaB-A strains, arsenite (6 µmol/liter) (Sigma), carbenicillin (50 µg/ml), or kanamycin (50 µg/ml) when appropriate.

Selection of virulent strains. To be tested for virulence, the wild-type strains (Table 3) were initially screened by CR uptake (24) and PCR amplification of virG by using a method and primers that were previously described (27, 29). A gentamicin protection assay in HeLa cells was performed with positive strains as described previously (27, 29), and intracellular organisms were recovered after 4 h of culture. HeLa cell-invasive organisms were passaged in guinea pig conjunctiva and recovered from a purulent keratoconjunctivitis (Sereny test) (34). In a second guinea pig passage, an infectivity dose of 10⁸ CFU was tested and Sereny-positive organisms were isolated and stored at 86°C as virulent stock cultures.

Inactivation of ShET1 in strain CVD 1205, yielding S. flexneri 2a strain CVD 1206. S. flexneri 2a strain CVD 1205 (guaB-A virG) (27) was derived from the wild-type S. flexneri 2a strain 2457T, which is known to be virulent based on experimental challenge studies in guinea pigs (28, 29) and adult volunteers (21). Details of the construction, characterization, and guinea pigs immunogenicity of strain CVD 1205 were published recently (27). The set1 operon is present in all Shigella flexneri serotype 2a strains but is rare in isolates of other Shigella serotypes (26). The construction of a set1 allele, with deletion of 85% of the subunit A of set1, and the consequent inactivation of the ShET1 enterotoxic activity were recently demonstrated by using Ussing chambers and in vivo perfusion experiments in rabbits (9). A suicide deletion cassette was constructed by cloning the set1 allele in pFM307 (27), yielding pFM804B. This deletion cassette was used to exchange the set1 allele for the proficient set1 allele in CVD 1205 by previously described methods (27). Clones in which set1 was successfully deleted were selected by the lack of DNA hybridization with a 52-bp probe (5'-CCTG GCCGGGCGGGCAAAACAACCCGTTATCTTTCATGGTCAGCTGACCG G-3') representing a deleted portion of the set1A gene. The deletion mutation in an arbitrarily selected clone was confirmed by PCR amplification of the truncated allele with the primers 5'-CGGGATCCCGGCCACCGGTTATGGCACCAATGAATACTGCGTTAT-3' and 5'-GCTCTAGAGCCTGGGCCCCCTGAACTGGACATACGACAAAACATC-3' and a protocol consisting of 94°C for 40 s, 60°C for 40 s, and 70°C for 4 min, for 30 cycles.

Inactivation of ShET2 in strain CVD 1206 and construction of S. flexneri 2a strain CVD 1207. ShET2 is a 62.8-kDa single-moiety protein encoded by the invasion plasmids of practically all Shigella serotypes (25). The sen allele was produced by PCR amplification and fusion of two 700-bp DNA segments that include the N and C termini of sen minus 300 bp corresponding to the putative active site in the N-terminal region of sen, as was done in previously described methods (27). The resulting sen allele was cloned into the suicide vector pFM307 (27). In addition, the proficient sen of S. flexneri 2a was cloned in pBluescript to serve as a positive control in corroborating the inactivation of ShET2 in sen. Supernatants from Escherichia coli DH5 (pBluescript::sen) were assessed in Ussing chambers as previously described (9, 10, 25).

guaB-A virG set1

Strain CVD 1207 as a donor of a virulence plasmid containing virG, sen, and the arsenite resistance marker. The S. flexneri 3a wild-type virulent strain J17B was originally isolated by S. Formal in Tokyo, Japan, in the 1970s. This strain agglutinates with group B, type factor III, group factors 6 and 7,8 antisera but not with group factor 3,4 antiserum (Table 1), and it is susceptible to all the commonly used antibiotics that were tested (data not shown). The invasion plasmid pFN110 was transferred from strain CVD 1207 to strain J17B by slight modifications of previously published methods described by Sansonetti et al. (33). Briefly, (i) S. flexneri 2a strain CVD 1207 (guaB-A virG set1 sen) was electroporated with pF'_ts114lac::Tn5, and CVD 1207(pF'_ts114lac::Tn5) clones were selected with kanamycin (33); (ii) a suitable recipient was prepared by selecting a strain J17B clone that had spontaneously lost its invasion plasmid (as evidenced by lack of CR dye uptake and lack of hybridization with a virG probe [27, 29]) (J17Bavir) and electroporating it with pBluescript (Stratagene, La Jolla, Calif.) to give a temporary selection marker (ampicillin-carbenicillin); (iii) late-log-phase broth cultures of strain CVD 1207(pF'_ts114lac::Tn5) and J17Bavir (pBluescript) were mated on TSA medium for 4 h; (iv) clones of J17Bavir (pBluescript) that had acquired the ars-tagged virulence plasmid pFN110 [J17B(pFN110, pBluescript)] were selected on TSA medium containing arsenite and carbenicillin; (v) selected J17B(pFN110, pBluescript) clones were plated in replicate on TSA medium containing carbenicillin and arsenite and TSA medium supplemented with kanamycin to select for clones that were not transfected with pF'_ts114lac::Tn5; and (vi) after an overnight incubation at room temperature on Trypticase soy broth supplemented with guanine and 6 µM arsenite, carbenicillin-sensitive clones (that had spontaneously lost pBluescript) were selected by replica plating on medium containing arsenite or arsenite and carbenicillin. In addition, the presence of pFN110 in J17B(pFN110) was confirmed by PCR amplification of the sen::ars allele with the TaqPlus long PCR system (Stratagene, La Jolla, Calif.), primers 5'-GCTCTAGAGCAGATAATATTCAGCTTTTTATATTCTTCATAATTTCCAGA-3' and 5'-GCTCTAGAGCACCTAGGATGGTAAGTACAGAAAACTTCAAAAAAGTTAAG-3', and the cycling conditions 94°C for 30 s, 50°C for 30 s, and 68°C for 8 min, for 25 cycles.

Safety (Sereny) test. The guinea pig purulent keratoconjunctivitis test was used with slight modifications of the method described by Sereny (34). Briefly, 12 Hartley guinea pigs (3 animals per group) were randomized to be inoculated in their conjunctival sac with 10 µl of a suspension containing 10⁹ CFU of the S. flexneri wild-type strain 2457T or J17B or the vaccine strain CVD 1207 or CVD 1211. Follow-up, grading of inflammation, and statistical analysis were performed as described previously (27).

Immunizations and sample collections. Bacterial strains were cultured overnight at 37°C on TSA-CR supplemented with guanine, harvested on phosphate-buffered saline (PBS), and brought to the desired concentration (as measured by determining the optical density at 600 nm). In a preliminary study, S. flexneri 2a CVD 1207 and S. flexneri 3a CVD 1211 were individually evaluated as described previously (27). For each cross-protection study, 15 Hartley guinea pigs (weighing 300 g) were immunized intranasally (27, 28) with 10¹⁰ CFU each of CVD 1207 and CVD 1211, suspended in 100 µl of PBS; 10 guinea pigs received 2 × 10¹⁰ CFU of E. coli HS as placebo controls. Immunizations were performed on days 0 and 14 in animals previously anesthetized subcutaneously with ketamine HCl (40 mg/kg) (Fort Dodge Laboratories, Fort Dodge, Iowa) and xylazine (5 mg/kg) (Bayer, Shawnee Mission, Kans.). Tears were collected on days 0, 14, and 30 to 35 postimmunization as described previously (27, 28); sera were obtained on days 0 and 30 to 34 by anterior vena cava puncture (41) under intraperitoneal anesthesia with ketamine HCl and acepromazine maleate (1.2 mg/kg) (Ayerst Laboratories, Inc., New York, N.Y.).

LPS extraction. Cultures of every wild-type serotype tested were obtained from frozen stock and grown overnight on TSA-CR, and their type and group factors were confirmed by agglutination with the corresponding specific antisera (12). Bacteria were suspended in Luria-Bertani broth and incubated at 37°C with shaking overnight. S. flexneri LPS from each wild-type strain (Table 3) was prepared by the method of Westphal and Jann (40) and further purified by the procedures of Thomashow and Rittenberg (36). Briefly, LPS was extracted from whole cells with hot phenol and the aqueous phase was collected, dialyzed, and treated successively for 1 h each with RNase A (100 µg/ml), DNase I (50 µg/ml plus 1 mM MgCl₂), and pronase (250 µg/ml). EDTA (5 mM) was added, and the phenol extraction was repeated. After dialysis, the aqueous material was centrifuged at 107,000 × g for 2 h and the sedimented LPS was suspended in water and lyophilized. Stock solutions of LPS of each serotype were prepared in gradient-pure water at 2 mg/ml prior to use.

ELISA. Immunoglobulin A (IgA) antibodies to each specific S. flexneri LPS were determined by enzyme-linked immunosorbent assay (ELISA) with rabbit anti-guinea pig IgA antibody (Bethyl Lab., Inc., Montgomery, Tex.) followed by phosphatase-conjugated goat anti-rabbit IgG antibody (Kirkegaard & Perry Laboratories, Gaithersburg, Md.). IgG antibodies were determined by ELISA with a phosphatase-conjugated goat anti-guinea pig IgG (Bethyl Lab., Inc.) as described previously (28, 29). ELISA titers were log transformed and compared by Student's t test as described previously (27-29).

Immunoblots. Stock LPS solutions were mixed with equal volumes of 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis sample lysis buffer and boiled for 10 min. Samples (30 µl containing 30 g) of each LPS preparation were electrophoresed in replicate sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels (15% polyacrylamide). One gel was oxidized with periodic acid as described by Tsai and Frasch (37) and silver stained (Bio-Rad, Hercules, Calif.) to confirm the presence of LPS ladders (data not shown); LPS from other gels were transferred to nitrocellulose. Tears and sera were pooled from 60 immunized animals from four different experiments and used at a concentration of 1:100 for Western immunoblotting. The presence of specific anti-LPS IgA in tears and IgG in serum was demonstrated by using the same commercial secondary antibodies as above, and the blots were developed with the phosphate substrate chromagen (Western Blue; Promega, Madison, Wis.).

Agglutination cross-reactions. Cultures of each wild-type serotype tested were grown at 37°C overnight on TSA-CR. Pools of sera from immunized animals were used neat or in 1:2 to 1:32 dilutions in PBS. Three or four colonies of each culture were mixed with a wooden toothpick with 15 µl of neat or diluted sera on a glass slide and rocked. The pooled serum dilution in which bacteria did not agglutinate after 2 min was considered negative.

Protective efficacy in guinea pigs. In each experiment, the guinea pigs immunized with the polyvalent vaccine or with E. coli HS were inoculated in one of their conjunctival sacs with 10⁸ CFU of one of the wild-type S. flexneri strains listed in Table 3. Guinea pigs were examined daily for 5 days, and inflammatory responses were graded with a severity score as described previously (28). Briefly, 0 = normal eye indistinguishable from contralateral noninoculated eye, 1 = lacrimation or eyelid edema, 2 = 1 plus mild conjunctival hyperemia, 3 = 2 plus slight exudate, and 4 = full purulent keratoconjunctivitis. The individuals examining the guinea pigs and scoring the results were blinded as to which strain(s) (vaccine or placebo) had been used for immunizations. The overall frequency of occurrence of inflammation of any severity (severity score, 1 to 4) in the vaccine and control groups was compared by Fisher's exact test. The statistical significance in peak severity scores was calculated by a nonparametric sum of ranks (Mann-Whitney test).

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Construction of vaccine strains. The suicide deletion cassette pFM804B was used to exchange the set1 allele for the proficient set1 allele in CVD 1205, yielding S. flexneri 2a strain CVD 1206 (guaB-A virG set1). The specific deletion in the set1 operon was confirmed by the lack of DNA hybridization with a 50-bp probe representing a deleted portion of the set1A gene and by the PCR amplification of set1 (Fig. 1). The deletion mutation of the gene encoding ShET1 (set1) and ShET2 (sen) was performed in S. flexneri 2a strain CVD 1205 (guaB-A virG) (27), yielding S. flexneri 2a strain CVD 1207 (guaB-A virG set1 sen). The lack of enterotoxic activity of strain CVD 1207 was confirmed by experiments in Ussing chambers (data not shown). The virG sen Shigella invasion plasmid pFN110 of CVD 1207 (pF'_ts114lac::Tn5) was inserted into S. flexneri 3a J17B by conjugation. Those J17B(pFN110) clones that were also transfected with pF'_ts114lac::Tn5 (50% occurrence) were identified by their resistance to kanamycin. Only J17B(pFN110) clones susceptible to kanamycin were selected. Figure 2 shows the PCR amplification of the sen::P_tac-ars allele in the invasion plasmid of J17B(pFN110) after conjugation. Consequently, the guaB-A deletion mutation was performed in strain J17B(pFN110), yielding S. flexneri 3a strain CVD 1211 (guaB-A virG sen). Strains CVD 1207 and CVD 1211 do not grow in minimum medium unless it is supplemented with guanine as described previously (27). In addition, these strains are resistant to at least 6 µM arsenite in the medium but no growth was obtained at this concentration of arsenite with any other Shigella strain (belonging to the four Shigella groups) tested.

View larger version (48K): [in this window] [in a new window]   FIG. 1.   Agarose gel (1% agarose) stained with ethidium bromide, showing 1-kbp ladder molecular weight markers (lane A), PCR amplification of set1 from S. flexneri 2a strain CVD 1206 (1.3 kbp) (lane B), and PCR amplification of wild-type set1 from S. flexneri 2a strain CVD 1205 (1.6 kbp) (lane C).

View larger version (70K): [in this window] [in a new window]   FIG. 2.   PCR amplification of sen::Ptac-ars from S. flexneri 3a strain CVD 1211 after conjugation with S. flexneri 2a strain CVD 1207 (pF'ts114lac::Tn5). Lanes: A, 1-kbp ladder molecular weight markers; B, sen::Ptac-ars (7.5 kbp) from strain CVD 1211; C, sen (2.2 kbp) from wild-type S. flexneri 3a strain J17B.

Sereny studies. No inflammatory response was observed in the conjunctivas of three guinea pigs that received 10⁹ CFU of S. flexneri 2a strain CVD 1207 or three guinea pigs that received 10⁹ CFU of S. flexneri 3a strain CVD 1211. Full purulent keratoconjunctivitis was observed in all controls that received the virulent strains 2457T and J17B.

Assessment of cross-reactivity elicited by an S. flexneri 2a/3a vaccine against the LPS of other S. flexneri serotypes. As determined by ELISA (Table 4), two doses of the polyvalent S. flexneri 2a/3a vaccine (CVD 1207/CVD 1211) elicited IgG antibodies in serum that strongly cross-reacted with S. flexneri LPS of serotypes 1a and 1b (Table 4). In contrast, the IgA antibodies elicited in tears by this combination vaccine showed an extensive cross-reaction that included S. flexneri serotypes 1a, 1b, 2b, 4b, 5b, and Y (Table 4). Interestingly, the cross-reaction against S. flexneri 2b by ELISA was modest even though this particular serotype shares antigenic type factor II with S. flexneri 2a and group factor 7,8 with S. flexneri 3a.

View larger version (66K): [in this window] [in a new window]   FIG. 3.   Cross-reactions against S. flexneri LPS after immunization with the attenuated S. flexneri 2a/3a (CVD 1207/CVD 1211) vaccine. (A) Immunoblotting with pooled guinea pig sera showing IgG cross-reactions. (B) Immunoblotting with pooled guinea pig tears showing IgA cross-reactions. The designation of each S. flexneri serotype is indicated above and below each lane.

Protection elicited by S. flexneri 2a strain CVD 1207 and S. flexneri 3a strain CVD 1211 against challenge with their homologous serotypes. In preliminary experiments, we tested the ability of the individual vaccine strains to protect against their homologous serotypes. S. flexneri 2a strain CVD 1207 conferred 85% protective efficacy to guinea pigs against keratoconjunctivitis produced by wild-type strain 2457T (P = 0.05). Likewise, S. flexneri 3a strain CVD 1211 conferred 75% protection against conjunctival challenge with wild-type strain J17B (P = 0.01).

Cross-protection among S. flexneri serotypes elicited by the S. flexneri 2a/3a vaccine. The protection conferred by the combination of S. flexneri 2a and S. flexneri 3a vaccine strains against other S. flexneri serotypes is shown in Table 5. There was considerable variation in the degree of cross-protection conferred by this vaccine against the heterologous serotypes. The highest cross-protection (92.5%) was observed against S. flexneri 2b, which, as discussed above, shares with the combination vaccine its type and group factors (Table 5). Interestingly, this high degree of protection was achieved despite a comparatively low cross-reacting serum IgG response against its LPS (Table 4). At the other extreme was the very low protection (protective efficacy, 20% [not significant]) achieved against S. flexneri type 4b strain 3143-94. However, despite the low "total" protection obtained against this serotype, the severity of the inflammatory response that was blindly recorded in vaccinees was significantly milder than that in the placebo controls (Table 5). In contrast, the attack rate and the degree of inflammation in guinea pigs that received the combination vaccine or the placebo control and were challenged with S. flexneri serotype 6 were basically equivalent.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

There are major cross-reactions among the multiple serotypes included in Shigella group B (S. flexneri) (8), and data from experimental observations (35, 39) and clinical trials (5) support the belief that cross-protection may be conferred by some of the members of this group against other serotypes of the same group. Weil and Farsetta (39) reported 50 years ago that mice immunized with S. flexneri III-Z (later known as S. flexneri 3a [type III; group 6, 7,8]) conferred cross-protection against challenge with wild-type S. flexneri I to III (later known as S. flexneri 1b [type I; group 6]) but much less protection against challenge with S. flexneri II-W and VI-Boyd 88 (later known as S. flexneri 2a [type II; group 3,4] and S. flexneri 6, respectively). Twenty-five years later, Sereny et al. (35) reported that immunization of guinea pigs with S. flexneri 4b (type IV; group 6) protected them against challenge with the homologous wild-type S. flexneri 4b and the heterologous S. flexneri 3a (type III; group 6, 7,8). Although the association was not made at the time, in these two instances the strains used to immunize the animals and the heterologous strains to which they elicited protection possessed the antigenic group factor 6. Other reports are more difficult to analyze because of the lack of identification of the specific subserotypes involved in the studies. An example is the study by Cooper et al. (5), in which serum from children immunized with S. flexneri types II, III, and VII (later known as S. flexneri serotypes 2, 3, and X, respectively) protected mice against the homologous serotypes and against the heterologous S. flexneri I (later known as S. flexneri serotype 1), but not against serotype VI (later known as S. flexneri serotype 6). As mentioned above, S. flexneri 6 has a different LPS structure from the rest of the group B serotypes. More recently, Lindberg et al. (22) constructed a vaccine candidate (strain SFL114) based on an attenuated strain of S. flexneri Y (group factor 3,4) which protected monkeys against challenge with wild-type strains of the serotypes S. flexneri Y (homologous), S. flexneri 1b (type I; group 6), and S. flexneri 2a (type II; group 3,4) (19). Hartman et al. (16) confirmed the capacity of the same S. flexneri Y vaccine to elicit protection in guinea pigs against the wild-type homologous serotype. However, immunized animals were not protected against challenge with wild-type S. flexneri 2a. We do not have a plausible explanation for the discrepancy observed with the two animal models. It is known that monkeys often acquire natural infections with Shigella spp. (14) and subsequently may be more prone than guinea pigs to produce cross-reactive immune responses (14, 38).

Reported herein is the construction of two novel vaccine candidates with a combination of deletion mutations in metabolic (guaA and guaB) and virulence (virG, sen, and set1) genes that give striking attenuating characteristics (Table 2). However, despite their marked attenuation, both strains were demonstrated to be immunogenic and protective in our guinea pig animal model. In addition, during construction of these strains, we inserted a nonantibiotic selection marker in the middle of the sen allele to allow facile transfer of the virulence plasmid containing the deletion mutation in virG and sen to other vaccine strains (i.e., S. flexneri serotype 6, S. dysenteriae serotype 1) and the identification of the vaccine strain in the field. The ability to readily transfer the arsenite resistance-tagged virG sen virulence plasmid in S. flexneri 2a strain CVD 1207 (pF'_ts114lac::Tn5) was demonstrated in the construction of S. flexneri 3a strain CVD 1211. Sansonetti et al. (33) had previously used the F' factor encoded in pF'_ts114lac::Tn5 to transfer an invasion plasmid (pWR110) from a virulent Shigella strain into a plasmidless avirulent one. In this report, we have demonstrated that the same technique can be applied to transfer an invasion plasmid with specific attenuating mutations to facilitate the construction of Shigella vaccines.

The attenuated S. flexneri 2a strain CVD 1207 and S. flexneri 3a CVD 1211 were constructed because, in addition to belonging to serotypes that are very prevalent in developing countries, together they bear the immunodominant antigenic group factors of the S. flexneri group. The results presented herein demonstrate that broad cross-protection, albeit not complete, is achieved by a vaccine consisting of a combination of these two serotypes. As may be expected, given the antigenic diversity of the S. flexneri serotypes, the degree of cross-protection varied. Thus, a high degree of cross-protection may be achieved if the antigenic type factor as well as the group factors are covered by the combination vaccine. This was observed when the CVD 1207-plus-CVD 1211-immunized guinea pigs were challenged with a wild-type virulent strain of the S. flexneri 2b (type II; group 7,8) or Y (group 3,4) serotypes (Table 5). Likewise, a similar outcome may occur against virulent strains of S. flexneri serotypes 3b (type III; group 3,4, 6) and 3c (type III; group 6). However, we cannot rule out the possibility that a low cross-protection rate will be observed against certain serotypes, such as S. flexneri 4b (type IV; group 6) (20% vaccine efficacy in this study). Nevertheless, even in those cases, the vaccine may confer a significant degree of protection against the severity of disease (Table 5).

The vaccination strategy against the S. flexneri group presented in this paper may simplify the construction of a broad-spectrum vaccine against shigellosis. Ideally, it will be desirable for this vaccine to be flexible so that it can meet the specific needs of the geographic area to be targeted. For example, one may envision an attenuated vaccine for oral administration containing S. flexneri serotypes 2a and 3a (covering most of the S. flexneri group) and 6 (which does not cross-react with the other S. flexneri serotypes), to which can be added S. sonnei and/or S. dysenteriae type 1 depending on the geographic area for which these are intended. Alternatively, another Shigella serotype(s) may be added if it is thought to be prevalent and not covered by this vaccine (i.e., an S. dysenteriae 2-13 or S. boydii 1-18 serotype, which are usually of low prevalence) or that inadequate cross-protection is achieved against it (i.e., S. flexneri 4b). Thus, hypothetically, attenuated strains of five or six serotypes contained in an oral vaccine formulation could protect against the great majority of the causes of shigellosis in the world.