Assembly, Biochemical Characterization, Immunogenicity, Adjuvanticity, and Efficacy of Shigella Artificial Invaplex

Shigella species are bacteria that cause severe diarrheal disease worldwide primarily in young children. Treatment of shigellosis includes oral fluids and antibiotics, but the high burden of disease, increasing prevalence of antibiotic resistance, and long-term health consequences clearly warrant the development of an effective vaccine. One Shigella vaccine under development is termed the invasin complex or Invaplex and is designed to drive an immune response to specific antigens of the bacteria in an effort to protect an individual from infection. The work presented here describes the production and evaluation of a new generation of Invaplex. The improved vaccine stimulates the production of antibodies in immunized mice and guinea pigs and protects these animals from Shigella infection. The next step in the product’s development will be to test the safety and immune response induced in humans immunized with Invaplex.

the disease burden, increasing antibiotic resistance, and increasing appreciation of the postinfectious sequelae associated with shigellosis (6,7). Vaccine efforts are directed at the most prevalent serogroups, such as Shigella flexneri and Shigella sonnei, which together account for approximately 89% of shigellosis cases in developing regions (8).
Humans and higher primates infected with Shigella develop robust immune responses to several invasion plasmid antigens (Ipa), which are key effectors secreted by the type III secretion system (TTSS) involved in the ability of shigellae to invade the colonic mucosa. The Ipa proteins are highly conserved among all four Shigella species and share functional and structural homology with TTSS effector proteins of other Gram-negative bacteria, e.g., Salmonella and Yersinia (9). Although the prominent immune response to the Ipa proteins has been noted for decades (10,11), the role of the Ipa proteins in protective immunity in humans is not clear. In contrast, immunity to shigellosis is strongly associated with antibodies to the O antigen of lipopolysaccharide (LPS) and is serotype specific (12,13). Progressing from antigen identification to clinical trials and licensure has been difficult for the many Shigella vaccine approaches that are in various stages of development. To date, all vaccines in clinical trials are focused primarily on stimulating a response to the O antigen of the LPS in attempts to stimulate serotype-specific immunity and the potential of serogroup cross-reactivity (14).
The prototype Invaplex vaccine, native Invaplex (Invaplex NAT ), purified from virulent Shigella, contains the conserved invasin proteins IpaB and IpaC, serotype-specific LPS, and other proteins of undefined function and limited to no immunogenicity (15,16). Intranasal immunization of small animals with Invaplex NAT drives a pronounced immune response to LPS, IpaB, and IpaC that mimics that observed after natural infection with virulent Shigella. Protection with Invaplex NAT has been achieved in both the mouse and guinea pig models (15,17), prompting the clinical evaluation of the vaccine in two phase 1 safety and immunogenicity studies (18,19). The current good manufacturing practice (cGMP)-manufactured S. flexneri 2a Invaplex NAT vaccine was found to be safe, well tolerated, and immunogenic in humans. The efficacy results of a follow-on phase 2b study were inconclusive due to the product being less immunogenic, likely due to product instability upon prolonged storage and a lower-than-expected attack rate in placebo controls (M. S. Riddle and R. W. Kaminski, unpublished data).
Further purification of Invaplex NAT by size-exclusion chromatography (SEC) yielded a highly purified (HP) Invaplex NAT with a high molecular mass of Ͼ669 kDa. This high-molecular-mass complex (HMMC) consisted of only the principal antigens IpaB, IpaC, and LPS and showed enhanced immunogenicity and protection against infection (16). Using the molar ratios of the components present in HP Invaplex NAT , a synthetic Invaplex, termed artificial Invaplex or Invaplex AR , has been assembled with purified Shigella LPS and purified recombinant IpaB and IpaC proteins. Here, the first-generation Invaplex NAT vaccine is compared with the second-generation Invaplex AR vaccine in tests of biochemical composition, immunogenicity, adjuvanticity, and protective efficacy. The studies indicate that not only is the Invaplex AR a better-defined product biochemically, making it more compliant with FDA guidelines on chemistry, manufacturing, and controls, but it exhibits enhanced immunogenicity and protection compared to Invaplex NAT .

RESULTS
Assembly and characterization of artificial Invaplex. In order to evaluate and accurately compare the immunogenicities and protective efficacies of Invaplex AR and Invaplex NAT produced from wild-type shigellae, a sufficient amount of the Shigella Invaplex AR products to immunize groups of both mice and guinea pigs with ranging vaccine doses was required. The Invaplex AR was purified from the reaction mixture by ion-exchange chromatography. Two dominant peaks were observed at both 50% B (~0.5 M NaCl, 0.02 M Tris-HCl, pH 9.0) and 100% B. Spot blot analysis and total protein measurements showed the presence of IpaB, IpaC, and LPS within the 50% B elution peak but did not identify significant amounts of IpaB or IpaC within the 100% B peak. The 50% B peak elution fractions were combined and were defined as the Invaplex AR product. Additional experiments using S. flexneri 1a and Shigella dysenteriae 1 LPS also resulted in Invaplex AR products purified within the 50% B elution peak of anion chromatography (data not shown).
The yield efficiency of the Invaplex AR production was calculated by dividing the total protein yield of the product by the total protein amount of the reactants and then multiplying by 100 and ranged from 8 to 22% yield.
Comparison of S. flexneri 2a Invaplex AR and Invaplex NAT by SDS-PAGE. Equal amounts, based on protein content, of the S. flexneri 2a Invaplex AR product and Invaplex NAT product were separated by SDS-PAGE followed by Western blotting or staining with either silver or Coomassie blue (Fig. 1). Western blotting and probing with the anti-IpaB specific monoclonal antibody (MAb) 2F1 and the anti-IpaC specific MAb 2G2 (20) identified IpaB and IpaC, respectively, in both Invaplex AR and Invaplex NAT (Fig. 1, left panel), although the antibody reactivity with the proteins in the Invaplex NAT product was weaker, suggesting that IpaB and IpaC are in lower concentrations within the total protein makeup of this product.
The IpaC protein band in the Invaplex AR product, which was produced using histidine-tagged IpaC (HTIpaC), migrated to the same position as the purified HTIpaC, while the IpaC band identified in the Invaplex NAT product migrated slightly lower. The higher apparent molecular mass of the HTIpaC band is attributed to the polyhistidine tag.
Coomassie blue staining (Fig. 1, center panel) of the Invaplex products clearly demonstrated the IpaB and IpaC proteins within the Invaplex AR product but did not detect these bands definitively within the Invaplex NAT product, instead showing multiple bands of various molecular weights that stained with various intensities. The Coomassie blue staining suggests that the proportion of the Ipa proteins in the total protein composition is lower within the less-defined Invaplex NAT product than in the Invaplex AR product, which contains only two proteins, IpaB and IpaC. The approximate antigen amounts in a 1-g dose of Invaplex AR are 370 ng IpaB, 600 ng IpaC, and 953 ng LPS.
The more sensitive silver staining technique detects both LPS and proteins. Many more polypeptide bands within the two Invaplex products were detected along with the S. flexneri 2a LPS "ladder" (Fig. 1, right panel). In addition to the full-sized IpaB and HTIpaC proteins, lower-molecular-weight bands were identified by silver staining within the purified IpaB and HTIpaC components and were likely degradation products of the antigens. The same lower-molecular-mass bands were also present in the Invaplex AR product. The LPS ladder was prominent within the Invaplex AR product compared to the Invaplex NAT product, which suggests a greater proportion of the S. flexneri 2a LPS component within the Invaplex AR product.
Comparison of native and artificial Invaplex products by SEC-HPLC and dynamic light scattering (DLS). The S. flexneri 2a Invaplex NAT and Invaplex AR products were separated by size-exclusion chromatography-high-pressure liquid chromatography (SEC-HPLC). Chromatograms were recorded at 215 nm in order to detect the Invaplex AR product as both the IpaB and IpaC proteins have a low proportion of aromatic amino acids (Fig. 2). The Invaplex AR product (17.5 g total protein loaded) revealed a single dominant symmetric peak with a retention time of 16.5 min (Fig. 2, red trace) which is resolved before thyroglobulin (669 kDa; 18.5 min) and is therefore interpreted as being larger than thyroglobulin. The Invaplex NAT product (70 g total protein loaded) resolved as several lower-molecular-mass peaks with retention times after thyroglobulin (Fig. 2, dashed blue trace) and one small peak at a retention time of 16.8 min that meets the definition of the previously reported high-molecular-mass complex (16). SEC-HPLC demonstrates that the production of an Invaplex AR product from individual components yields a slightly higher-molecular-mass product that is more homogeneous than Invaplex NAT prepared from wild-type Shigella water extract.
The S. flexneri 2a Invaplex NAT and Invaplex AR were also analyzed by dynamic light scattering (DLS) to determine their size distribution. Both Invaplex products displayed a polydisperse profile containing molecular species with different sizes. After controlling for the relative mass of the peaks, a major peak for each of the Invaplex products was identified (Fig. 3). The major peaks of the Invaplex products measured an average hydrodynamic radius of 18.6 Ϯ 3.9 nm for Invaplex NAT and 27 Ϯ 11.5 nm for Invaplex AR , although the majority of the Invaplex NAT product is smaller species of shorter hydrodynamic radii. The hydrodynamic radius of Invaplex AR is greater than the reported hydrodynamic radius of 8.5 nm for thyroglobulin and thus supports the results achieved using SEC-HPLC (21).
Immunogenicity and protective efficacy of native and artificial Invaplex vaccines. The mouse pulmonary challenge (22) and the guinea pig keratoconjunctivitis (23) models were used to determine whether intranasal immunization with S. flexneri 2a Invaplex AR stimulates a protective immune response that is effective against homologous challenge with wild-type S. flexneri 2a.
In mouse experiments utilizing a dose escalation study design, intranasal immunization with S. flexneri 2a Invaplex AR induced serum IgG and IgA antibodies directed to S. flexneri 2a LPS, IpaB, and IpaC ( Fig. 4) that were either significantly higher than or comparable to the antibody responses induced after immunization with similar dose amounts of S. flexneri 2a Invaplex NAT . In particular, the antibody responses directed to IpaB (IgA and IgG) and IpaC (IgG) were significantly higher after immunization with Invaplex AR than after immunization with Invaplex NAT at comparable dose amounts. Bronchial lavage fluid IgA responses showed increased IpaC titers in groups immunized with Invaplex NAT and Invaplex AR , with higher responses observed in the Invaplex AR groups. Furthermore, intranasal immunization of mice with Ն1 g of Invaplex NAT or Ն0.1 g of Invaplex AR resulted in significant protection against lethal challenge with S. flexneri 2a strain 2457T (Table 1). A delay in death was evident in all immunized mice. The day on which 27% (4 of 15) mice died occurred on day 5 in the saline control group, day 9 in the 0.1-g and 1.0-g Invaplex NAT groups, and day 11 in the 0.1-g Invaplex AR group. In all other groups, 4 or more mice did not die during the study. Moreover, animals immunized with 1 g (Fig. 5) or 2.5 g Invaplex AR lost significantly less weight (P Ͻ 0.02) after challenge than did mice immunized with Invaplex NAT . In contrast, mice immunized with 100% protective 5-g doses of either vaccine exhibited similar weight losses, each of which was significantly less than that of the saline control group (Invaplex AR , P ϭ 0.0005; Invaplex NAT , P ϭ 0.041). Mice immunized with the lowest dose (0.1 g) of either vaccine lost weight that was comparable to that observed in the saline control group.
To better understand the immune responses and protective efficacy afforded after immunization with Invaplex NAT and Invaplex AR , the 50% protective dose (PD 50 ) and 50% immunizing dose (ImmD 50 ) were calculated using an adaptation of the Reed-Muench 50% lethal dose (LD 50 ) formula (24). The PD 50 for Invaplex NAT was determined to be 0.5 g, which was approximately 7-fold higher than the calculated PD 50 for Invaplex AR (Table 2). Using similar calculations, it was determined that the dose of Invaplex AR required to induce antibody responses in 50% of the immunized animals (ImmD 50 ) to Invaplex, LPS, IpaB, or IpaC was 2.7-to Ͼ21-fold lower than the dose of Invaplex NAT required to induce antigen-specific immune responses, demonstrating that Invaplex AR is a more potent immunogen than Invaplex NAT .
Using a similar dose escalation study design, groups of guinea pigs were intranasally immunized with S. flexneri 2a Invaplex NAT or Invaplex AR . Control groups were inoculated with saline. Blood collected on days 0, 28, and 40 was assayed by enzyme-linked immunosorbent assay (ELISA) for titers of antibodies directed to LPS, IpaB, and IpaC. Similar to the results in mice, immunization of guinea pigs with Invaplex AR promoted antibody responses directed to LPS, IpaB, and IpaC at similar or higher titers than those for animals immunized with Invaplex NAT (Fig. 6). Guinea pigs ocularly challenged with S. flexneri 2a strain 2457T were significantly protected after three immunizations with 25 g of Invaplex NAT (75% protection; P ϭ 0.0005), but immunization with lower doses

FIG 5
Mice immunized with 1 g Invaplex AR experience less weight loss and recover more rapidly than those immunized with Invaplex NAT . Mice immunized with S. flexneri 2a Invaplex AR (red) or Invaplex NAT (blue) were challenged intranasally with a lethal dose of S. flexneri 2a strain 2457T. Mice were weighed on the day of challenge and each day for 14 days postchallenge. The average weight for each group was calculated, and the difference was expressed as a percentage of the change in weight from the prechallenge values. The numbers of surviving animals/animals in groups immunized with Invaplex NAT and Invaplex AR are shown. While protection was significant for both products compared to saline controls, a significant difference in weight was observed for the 1-g Invaplex AR group compared to the 1-g Invaplex NAT group (*, P ϭ 0.0126).
of Invaplex NAT did not result in significant protection (Table 1). In contrast, groups immunized with 5 or 23 g of Invaplex AR were significantly protected against challenge (59% protection; P ϭ 0.0053). The Reed-Muench formula was again used to calculate the PD 50 and ImmD 50 for the guinea pig model, and similar trends were achieved as obtained for the mouse model in that a higher Invaplex NAT dose than Invaplex AR dose was necessary to provide protection and elicit a Shigella-specific immune response in 50% of the animals immunized with Invaplex NAT (Table 2).
Homologous and heterologous protection afforded after immunization with S. sonnei Invaplex AR . Naturally acquired immunity against shigellosis is primarily serotype specific, and therefore, a multivalent vaccine containing multiple O-antigen types will likely be required to achieve broadly protective efficacy. Consequently, a vaccine that is capable of inducing protective immunity against multiple serogroups or serotypes would be highly advantageous. Immunization of mice with S. sonnei Invaplex NAT can provide cross protection against challenge with S. flexneri 2a strain 2457T (25). The ability of Invaplex AR to induce protective immunity against heterologous species or serotypes was also tested in the mouse pulmonary lung model. Mice were immunized intranasally with 2.5 g of S. sonnei Invaplex AR and divided into two subgroups: one subgroup was challenged with S. sonnei strain Moseley, the homologous serotype to the vaccine, and the other subgroup was challenged with S. flexneri 2a strain 2457T. Controls for the study included animals immunized with saline and challenged with either S. sonnei strain Moseley or S. flexneri 2a strain 2457T (negative controls) and animals immunized with 5 g of either S. sonnei Invaplex NAT or S. flexneri 2a Invaplex NAT and challenged with the homologous serotype (positive controls). None of the animals immunized with saline survived the challenge (Fig. 7), and all lost Ͼ20% of their prechallenge weight prior to death. Mice immunized with the native Invaplex vaccines and challenged with the homologous serotype all survived challenge (100% protection). As expected, all animals immunized with S. sonnei Invaplex AR and challenged with S. sonnei strain Moseley survived challenge and exhibited Ͻ10% weight loss. Interestingly, all animals immunized with S. sonnei Invaplex AR and challenged with S. flexneri 2a strain 2457T survived challenge and followed a similar percent weight loss curve as animals immunized with S. flexneri 2a Invaplex NAT . Mice immunized with S. sonnei Invaplex AR and protected against S. flexneri 2a produced serum IgG antibodies (geometric mean titer [GMT]) to S. sonnei LPS (136), IpaB (5,760), and IpaC (1,091) but weak to undetectable levels against S. flexneri 2a LPS (GMT of 90). These results demonstrate the ability of the Shigella Invaplex AR to provide a level of protective immunity against a heterologous Shigella species challenge similar to that of the Invaplex NAT product (17,25).
Adjuvanticity of S. flexneri 2a Invaplex AR . In addition to inducing Shigella-specific immune responses, Invaplex NAT also functions as an adjuvant, capable of enhancing the magnitude of the immune response directed to coadministered protein antigens (26) or antigens encoded on plasmid DNA (27). To determine if adjuvanticity was retained with Invaplex AR , groups of mice were intranasally immunized with either ovalbumin (OVA) alone or in combination with S. flexneri 2a Invaplex NAT , Invaplex AR , or cholera toxin (CT), a known mucosal adjuvant. Comparable levels of OVA-specific serum IgG (Fig. 8) and IgA titers (data not shown) were detected on days 28 and 42 after immunization with OVA mixed with Invaplex NAT , Invaplex AR , and cholera toxin. All responses were significantly higher (P Ͻ 0.001) than the OVA-specific immune response induced after immunization with OVA alone. In addition to enhancing the OVA-specific immune response, Shigella-specific immunity was induced in the groups immunized with OVA combined with Invaplex NAT or Invaplex AR (Fig. 8), indicating that Invaplex AR could be used in a combination vaccine strategy as both an adjuvant and an immunogen. Antigen-specific antibodies in lung washes collected 2 weeks after the third immunization (day 42) were also assessed by ELISA to investigate the mucosal immune responses induced after immunization (Fig. 8). Similar levels of OVA-specific lung IgA were induced after immunization with OVA mixed with Invaplex AR , Invaplex NAT , or CT, and these levels were significantly higher (P Ͻ 0.001) than those induced after immunization with OVA alone (Fig. 8). Immunization with OVA mixed with Invaplex AR induced similar levels of LPS-specific (data not shown) and Invaplex NAT -specific IgA in lung washes as those after immunization with OVA mixed with Invaplex NAT and higher levels than those induced after immunization with OVA alone or OVA mixed with CT (Fig. 8).

DISCUSSION
Analysis of the immune response generated after natural infection may guide the rational design of a Shigella vaccine capable of stimulating a protective immune response comparable to that induced postinfection. Several studies in humans (13,28,29) have demonstrated that individuals orally infected with Shigella are resistant to subsequent infections with homologous or closely related Shigella serotypes but not heterologous serotypes. In nonhuman primates (10), infected animals producing serum antibodies to homologous LPS and the Ipa proteins are resistant to homologous challenge but not challenge with a heterologous serotype. The results of these studies combined with epidemiological data (30) suggest that serotype-based immunity is prevalent after natural infection. Shigella infection induces a potent immune response to LPS with the O-specific polysaccharide (O-SP) generating serotype-specific immunity in both humans and nonhuman primates (10,21). In addition, a pronounced antibody response to the protein effectors of the TTSS, specifically IpaB, IpaC, and IpaD, which is characterized by cross-reactivity with all serotypes and species of Shigella, is induced (10-12, 20, 31, 32). Other protein antigens such as VirG (10,12), OmpA (33), or pan-Shigella surface protein 1 (PSSP-1) (34) may also stimulate cross-reactive immune responses. However, in some cases chromosomally encoded protein antigens, such as major outer membrane proteins, are broadly conserved among enteric Gram-negative bacteria and it is often difficult to identify specific infection-induced signals over the abundance of background cross-reactivity due to the structural similarity.
Vaccines that stimulate an immune response that mimics the natural response in magnitude, specificity, and functionality hold promise for inducing protective immunity in humans. Candidate Shigella vaccines with this capability include live attenuated Shigella (32,35), inactivated whole cells (36)(37)(38), and the Generalized Modules for Membrane Antigens (GMMA) product, although the last is deficient in the invasins (39). Vaccines focused primarily on only O-SP or only conserved proteins are promising but may not stimulate the comprehensive, fully protective immune response observed after infection. The Shigella Invaplex vaccine has been developed to mimic both the predominant serotype-specific aspect and pan-Shigella (protein-specific) immune responses induced after natural infection by presenting the LPS, IpaB, and IpaC antigens in a manner or context that induces a functional and effective immune response.
The first-generation Shigella Invaplex vaccines were isolated from water extracts of wild-type, virulent organisms (15) and have been given the nomenclature of "native" Invaplex. The prototype Invaplex NAT vaccines were evaluated in a series of studies that demonstrated the immunogenicity and efficacy of the product in preclinical studies (15, FIG 8 OVA-specific and Invaplex NAT -specific serum IgG endpoint titers in mice after intranasal immunization with OVA alone or OVA combined with Invaplex AR , Invaplex NAT , or cholera toxin (CT). Groups of mice were intranasally immunized on days 0, 14, and 28 with OVA or OVA combined with either S. flexneri 2a Invaplex AR or Invaplex NAT or CT. Blood collected on days 0, 28, and 42 was analyzed by ELISA for anti-OVA and anti-Invaplex NAT serum IgG endpoint titers. Lung washes collected on day 42 were analyzed by ELISA for anti-OVA and anti-Invaplex NAT IgA endpoint titers. Data represent the mean (n ϭ 5 mice/group) endpoint titer Ϯ 1 standard error of the mean. Significant differences between group mean log-transformed titers were determined by two-way analysis of variance (P Ͻ 0.05). 17), the feasibility of cGMP manufacture, and the safety and immunogenicity of the product in a series of phase 1 clinical trials (18,19). The Invaplex NAT product was safe and immunogenic when delivered intranasally to over 100 human volunteers. Although promising in early clinical trials, potential instability upon long-term storage, resulting in lower than previously achieved immunogenicity, combined with a lower than expected attack rate, hampered efforts to demonstrate proof-of-concept efficacy in a phase 2b trial (R. W. Kaminski and M. S. Riddle, unpublished results). The composition of Invaplex NAT includes the target major Shigella antigens (LPS, IpaB, and IpaC) and many other proteins, including IpaD, elongation factor G (EF-G), and DnaK, along with many other proteins with limited to no immunogenicity (see list of proteins in Table S1 in the supplemental material). The lack of product homogeneity and definition may have been associated with the instability and lower immune response induced by Invaplex NAT products stored for a prolonged period of time and suggested that a better understanding of the active components in Invaplex NAT was warranted. For this reason, efforts were undertaken to improve the product's homogeneity, manufacturing yield, and immunogenicity, and these efforts have led to the development of a secondgeneration Invaplex product.
Research conducted on a highly purified Invaplex NAT product clearly demonstrated that LPS, IpaB, and IpaC were combined in an HMMC that was essential for the product's immunogenicity and protective efficacy (16). Furthermore, the molar ratios of the IpaB and IpaC proteins and the protein-to-lipopolysaccharide ratios in the HMMC provided a blueprint for the design and construction of second-generation Invaplex AR . The methodology used to assemble LPS, IpaB, and IpaC into a synthetic or artificial high-molecular-mass complex has inherent flexibility that allows the three constituents to be assembled in ratios that optimize immunogenicity, protective efficacy, and maintenance of component solubility. The resultant Invaplex AR product is much more defined than Invaplex NAT and can be scaled up more readily than the process used to manufacture Invaplex NAT .
The customization of Invaplex AR is evidenced in the ease of switching from one Shigella serotype/species to another serotype by simply substituting the LPS used during the assembly process. Research-grade lots of Invaplex AR have been manufactured for S. flexneri 2a and S. sonnei as well as S. flexneri 3a, S. flexneri 1a, and S. dysenteriae 1 (data not shown), in which the same IpaB and IpaC lots were used with the various Shigella LPSs. Indeed, the LPS component was not only key as an antigen but was also essential for the solubility of the Invaplex AR product as, in experiments in which only IpaB and IpaC were mixed, the proteins precipitated out of solution. The presence of LPS prevented the precipitation of IpaC (data not shown), similar to the use of a detergent (40). The insolubility of purified IpaC has hampered subunit vaccine efforts including this critical virulence protein (40). IpaB and IpaC contain substantial regions of hydrophobicity (40)(41)(42), and both IpaC and IpaB have been reported to insert themselves into red blood cell and mammalian cell membranes (40,(43)(44)(45). The amphiphilic nature of LPS likely provides a favorable environment to facilitate the hydrophobic collapse or folding of the Ipa proteins in a soluble and potentially active state in a manner similar to the folding and membrane insertion of Escherichia coli OmpA protein (46) and E. coli OmpT protein (47,48). Once the Invaplex AR is formed, the complex is stable without evidence of precipitation or insolubility of any of the components or the complex. It is unknown if the complex represents a structure naturally found in Shigella, but as an LPS/IpaB/IpaC structure is also found in Invaplex NAT and is of similar size and composition and capable of inducing uptake in mammalian cells, we speculate that the Invaplex AR represents the active complex or HMMC found in Invaplex NAT . The slightly earlier SEC-HPLC retention time for the Invaplex AR product may be the result of the increased mass of the product compared to Invaplex NAT in part due to the use of a His-tagged IpaC product in the manufacture of the secondgeneration vaccine. Investigations to better define and characterize the structure of Invaplex AR are under way. The amount of lipopolysaccharide required in the assembly process to achieve levels of LPS-specific immunity which correlate with protection across multiple Shigella serotypes is also under active investigation. Studies optimizing the LPS content for new Shigella serotypes will permit the manufacture of a multivalent Shigella Invaplex AR product targeting the most predominant serotypes.
The immunogenicity and protective efficacy of the Invaplex NAT and Invaplex AR products were compared directly in the mouse and guinea pig models. Both models have been used extensively to evaluate potential Shigella vaccines, with the mouse model used primarily for early-stage vaccine candidates and the guinea pig conjunctivitis model applied to more advanced vaccine candidates immediately prior to initial studies in humans (19,22,23) as it depends largely upon a mucosal immune response for protection (36). In both models, Invaplex AR was comparable or superior to Invaplex NAT in the magnitude of the immune response directed to Shigella LPS, IpaB, and IpaC as well as affording significant protection against a lethal pulmonary or keratoconjunctivitis challenge. Neither product required an adjuvant to induce a protective immune response. Furthermore, the total protein dosage of Invaplex AR was 2-to 10-fold lower than that of Invaplex NAT for both immunogenicity and protective efficacy in both animal models. Solid protection (less weight loss and increased number of survivors) was achieved with 1 g of Invaplex AR . This dose amount contains 370 ng IpaB, 600 ng IpaC, and 953 ng LPS. Other reports (49) using IpaB as an antigen indicate that more than 2.5 g of IpaB is required for protection in mice, suggesting that less IpaB is required when presented in the context of Invaplex AR . The improved potency of Invaplex AR will not only conserve product but may open the possibility of including adjuvants to the formulation that may further enhance the immunogenicity of the product.
Recent epidemiological data (8) suggest that to achieve broad protection, a Shigella vaccine would need to provide coverage against S. sonnei, S. flexneri 2a, S. flexneri 3a, and potentially S. flexneri 6, with the idea that all LPS serotypes and serogroups, defined as group (3)4, group 6, and group 7(8), would be represented and provide expanded coverage (14). The Invaplex AR vaccine approach is amenable to multivalency by admixing successful monovalent vaccines. Bivalent formulations of Invaplex NAT have successfully demonstrated the feasibility of this approach (17). In addition, as indicated in our study, the ability of monovalent Invaplex AR vaccines to provide cross protection against multiple Shigella serotypes could be leveraged to broaden coverage. The LPS component of Invaplex AR provides the serotype and serogroup specificity of the vaccine. A quadrivalent Invaplex AR vaccine directed at the 4 prevalent serotypes would theoretically provide immunity against Ն80% of the Shigella isolates that cause global shigellosis (8). Furthermore, as the Ipa proteins are highly conserved across all Shigella serotypes, they add an attractive advantage of Invaplex AR for use in a pan-Shigella vaccine. Studies in mice using IpaB and IpaD (31) have demonstrated that a proteinonly-based vaccine is capable of protecting against challenge with other Shigella serotypes in the murine pulmonary model. Using a live attenuated S. flexneri 2a vaccine overexpressing TTSS antigens, broad protection has been observed in guinea pigs (32). However, other studies in guinea pigs immunized with only purified Ipa proteins have not demonstrated similar protective effects; when LPS was delivered in combination with IpaB and IpaC, the vaccine formulation generally resulted in significant protection (R. W. Kaminski and E. V. Oaks, unpublished data).
Another attribute of Invaplex NAT is the ability of the complex to function as a mucosal adjuvant. Studies conducted with protein antigens (ovalbumin, protective antigen from Bacillus anthracis, CFA/I antigen from enterotoxigenic E. coli, and FlaA from Campylobacter jejuni) have demonstrated that Invaplex NAT increases the magnitude of the humoral and cellular immune response directed to the coadministered antigen (26). Furthermore, Invaplex NAT also enhanced the immune response directed to vaccine antigens encoded on DNA plasmids (27). The adjuvanticity of Invaplex has been partially attributed to its inherent ability to induce cellular uptake of coadministered antigens as well as providing the necessary Toll-like receptor 4 (TLR-4) agonist that stimulates robust immune responses. The ability to induce cellular uptake is likely due to the biological activity associated with the Ipa proteins, which as TTSS effectors are actively involved in host cell invasion by Shigella spp. (9). Invaplex AR retains similar biological functions and can induce cellular uptake in normally nonphagocytic epithelial and fibroblast cell lines (50). The slightly larger size of Invaplex AR than Invaplex NAT may improve its interactions with antigen-presenting cells (APCs) and enhance its immunogenicity and adjuvanticity. The retention of the biological activity combined with inclusion of LPS allows Invaplex AR to also function as a mucosal adjuvant when codelivered with heterologous antigens such as ovalbumin (27). The adjuvanticity of Invaplex AR is very appealing when considering the development of a combination vaccine, such as an antidiarrheal vaccine to protect against enterotoxigenic Escherichia coli (ETEC) and Shigella infection.
Collectively, the comparisons between the first-generation Invaplex NAT vaccine and the second-generation Invaplex AR vaccine have demonstrated Invaplex AR to have comparable and enhanced properties in relation to Invaplex NAT . The Invaplex AR products have improved biochemical consistency and potency and offer the increased formulation flexibility necessary for a multivalent Shigella vaccine approach. Research to transition the recombinant IpaC protein to a non-histidine-tagged version and improved levels of recombinant protein expression, as well as efforts to increase the production yield efficiency of the mixing process, is actively under way in order to make a consistent product that is suitable for cGMP manufacturing and transition to clinical evaluations to evaluate the safety and immunogenicity of the second-generation product.

MATERIALS AND METHODS
Bacterial strains and growth. Shigella flexneri 2a strain BS103 and Shigella sonnei strain Moseley, both grown in brain heart infusion (BHI) broth, served as the source of purified LPS, while Invaplex NAT was isolated from virulent S. flexneri 2a strain 2457T grown in a modified animal-product-free formulation based on antibiotic medium number 3 (Becton, Dickinson) as previously described (18).
Recombinant clones expressing the Ipa of interest were kindly provided by Bill Picking of the University of Kansas (51)(52)(53). Escherichia coli strain BL21(DE3)pLysS harboring the IpaB pCAYC/HTIpgC pET15b or HTIpaC pET15b plasmid construct was grown in LB broth (Lennox) with appropriate antibiotics. Both recombinant strains were induced at an optical density at 600 nm (OD 600 ) of 0.5 with isopropyl-␤-D-thiogalactopyranoside (IPTG), followed by continued incubation at 37°C for 3 h, and then harvested for further processing (see below).
Purification of IpaB and HTIpaC (IpaC). Clarified supernatants of sonicated cells containing IpaB/His-tagged IpgC (HTIpgC) complexes were applied to a 20-ml HisPrep FF 16/10 nickel Sepharose (GE Healthcare) column in binding buffer (20 mM Tris-HCl, 500 mM NaCl, 5 mM imidazole, pH 7.9) (51). Bound IpaB/HTIpgC complexes were eluted with 20 mM Tris-HCl, 500 mM NaCl, 50 mM EDTA, pH 7.4. Fractions containing both IpaB and IpgC were pooled and dialyzed followed by addition of N-octyl-oligooxyethylene (OPOE; 1% final concentration; Alexis Biochemicals) (51) and incubated at room temperature with gentle swirling for 30 min. The OPOE-treated sample was applied to a 25-ml XK 16/10 nickel Sepharose HP (GE Healthcare) column. Fractions were collected during the loading and initial wash buffers. These void volume fractions, containing predominantly IpaB, were pooled, the OPOE concentration was made 1%, and the pools were incubated at room temperature for 30 min and then reapplied to the column according to the steps above to remove residual HTIpgC. The final purified IpaB product was stored at Ϫ80°C.
His-tagged IpaC (HTIpaC) was released from induced recombinant bacteria by suspension in binding buffer containing 6 M urea (52). The treated cells were sonicated and centrifuged, and the supernatant was stored frozen at Ϫ80°C prior to application to a 60-ml XK 26/10 nickel Sepharose HP (GE Healthcare) column previously washed and equilibrated in binding buffer containing 6 M urea (53). After washing, HTIpaC was eluted from the resin with elution buffer (20 mM Tris-HCl, 0.5 M NaCl, 1 M imidazole, 6 M urea, pH 7.9). The final sample was dialyzed against 20 mM Tris-HCl, 0.5 M NaCl, 6 M urea, pH 7.9, and the resulting purified HTIpaC was stored at Ϫ80°C. LPS purification. LPS was extracted from Shigella flexneri 2a strain BS103 and S. sonnei strain Moseley by the procedure of Westphal and Jann (54). Purified LPS was lyophilized and analyzed by SDS-PAGE (silver stained) and by ELISA with serotype-specific monoclonal antibodies. Residual protein was accessed by the Bradford total protein assay (55). Protein levels were less than 0.15% by weight.
Purification of Invaplex NAT . Native Invaplex (Invaplex NAT ) was purified by anion-exchange chromatography of water extracts of virulent shigellae as previously described (15). The final Invaplex NAT product contained LPS, IpaB, IpaC, and many other proteins in low concentrations (see Table S1 in the supplemental material).
Production of artificial Invaplex (Invaplex AR ). Using the IpaC:IpaB molar ratio previously described (16), HTIpaC (8 M) was mixed with IpaB (1 M) and diluted in a buffer containing 20 mM Tris-HCl, 500 mM NaCl, 5 M urea, pH 7.9. Reaction mixtures that contained a final urea concentration less than 5 M were prone to precipitation as evidenced by marked turbidity of the solution. Reaction volumes of 5 to 25 ml were used. The IpaB-HTIpaC mixture was slowly added to a vessel containing dry Shigella LPS (S. flexneri 2a strain BS103 LPS or S. sonnei strain Moseley LPS) at a ratio of 0.56 (mass/mass) of the total protein content of the final reaction mixture. After gentle swirling to solubilize the LPS, the vessel containing the "reaction mixture" was placed at 37°C and incubated with shaking at 200 rpm for 2 h. The reaction mixture was diluted with 4 volumes of 20 mM Tris-HCl, pH 7.9, at 37°C and applied to a 5-ml HiTrap Q HP column (GE Healthcare). The Invaplex product was eluted using the Invaplex NAT chromatography purification strategy (15). A step gradient of 0%, 24%, 50%, and 100% buffer B (1 M NaCl in 20 mM Tris-HCl, pH 9.0) was used to elute fractions. Invaplex NAT elutes in the 24% and 50% fractions. Elution fractions were analyzed for total protein by the bicinchoninic acid (BCA) assay (Pierce), and the presence of IpaB, IpaC, and LPS was analyzed by spot blotting. Fractions containing all three antigens were consistent with the Invaplex product and eluted in the 50% B step gradient (i.e., 500 mM NaCl). The resulting Invaplex AR product was further characterized by SDS-PAGE, quantitative ELISA, and Limulus amebocyte lysate (LAL) assay.
Electrophoresis, Western blotting assays, and ELISA. Polyacrylamide gel electrophoresis with Coomassie blue and silver staining and Western blotting assays were performed as previously described (10,15). ELISAs were used to measure levels of antibodies to S. flexneri 2a LPS, IpaB, IpaC, and ovalbumin (OVA; Sigma-Aldrich). Primary antibodies were diluted in 2% casein and were incubated with the antigen for 2 h. After washing in phosphate-buffered saline (PBS)-Tween 20, plates were probed with anti-IgG or -IgA conjugated with alkaline phosphatase (mouse studies) or probed with anti-IgG conjugated with alkaline phosphatase or anti-IgA conjugated with horseradish peroxidase (guinea pig studies). The substrate used for assays utilizing alkaline phosphatase was p-nitrophenylphosphate (1 mg/ml in diethanolamine), and the optical density was measured at 405 nm. The substrate used for assays utilizing horseradish peroxidase was 3,3=,5,5=-tetramethylbenzidine (TMB), the 30-min reaction was stopped with 1 N HCl, and optical density was measured at 450 nm.
Positive, negative, and background controls were assayed on each plate, and strict acceptability criteria (acceptability ranges) were employed to reduce assay variability. Geometric mean titers (GMTs) were determined for each group using the individual titers within that group at a specific time point. To calculate GMT, titers less than the limit of detection (LOD) were assigned a value of 1/2 LOD. A 4-fold increase in the GMT over baseline samples was considered seroconversion.

SEC-HPLC.
A TSK-Gel G5000PW XL 7.8-mm by 30-cm column (Tosoh Bioscience) with a 10-m particle size and an exclusion limit of 1 ϫ 10 7 Da was calibrated using blue dextran (2 MDa), thyroglobulin (669 kDa), catalase (232 kDa), ovalbumin (43 kDa), and RNase A (13.7 kDa) (all from GE Healthcare) in 0.02 M Tris-HCl, 0.5 M NaCl, pH 9.0 (Invaplex AR buffer), connected to a Shimadzu 10ADvp HPLC system. S. flexneri 2a Invaplex AR (17.5 g) and Invaplex NAT (70 g) were applied in separate runs under the same conditions at a flow rate of 0.5 ml/min and 400-lb/in 2 maximum pressure. Chromatographic traces were recorded both at 280 nm using an SPD-10Avp UV detector and at 215 nm using the SPD-M10Avp photodiode array.
DLS. Dynamic light scattering (DLS) experiments were performed using a Malvern Zetasizer V system using the Zetasizer software version 7.11 for the analysis of data. Fifteen microliters of each sample (Invaplex NAT at 1.8 mg/ml in 20 mM Tris-HCl, 250 mM NaCl, pH 9.0, and Invaplex AR at 234 g/ml in 20 mM Tris-HCl, 500 mM NaCl, pH 9.0) was transferred to a 2-l-window Hellma Analytics quartz cuvette. The sample chamber was cooled to 4°C, the laser power and attunement were automatically adjusted for each run so that they were consistent for each scan, and the results for each run are the averages from 11 to 13 individual scans. The final size of the product is reported as the hydrodynamic radius and is the average from 5 runs with 1 standard deviation. The reported results represent the HMMC of Invaplex NAT compared to the predominant species present in the Invaplex AR sample as determined by mass.
Quantitative IpaB, IpaC, and LPS assays. The concentration of IpaB and IpaC in HP Invaplex fractions was determined using a modified ELISA procedure as previously described (16). The amount of S. flexneri 2a LPS in each Invaplex product was determined using the kinetic colorimetric EndoSafe-PTS device and EndoSafe cartridges from 0.1 to 10 endotoxin units (EU)/ml (Charles River, Inc.).
Immunogenicity and protective efficacy of Invaplex in animal models. The ability of Invaplex NAT and Invaplex AR to promote an immune response in BALB/cByJ mice (Jackson Laboratories) was tested in groups of 15 mice. Mice were immunized intranasally with either S. flexneri 2a Invaplex AR (5, 2.5, 1, and 0.1 g) or Invaplex NAT (10, 5, 2.5, 1, and 0.1 g) or saline on days 0, 14, and 28. A total antigen volume of 25 l was delivered in small drops applied to the external nares with a micropipette. Mice were bled on days 0, 28, 42, and 63. Three weeks after the final immunization (day 49), mice were challenged intranasally with a lethal dose of S. flexneri 2a strain 2457T (1.6 ϫ 10 7 CFU/30 l) as described previously (15,22). Prior to intranasal immunization or challenge, mice were anesthetized with a mixture of ketamine hydrochloride (40 mg/kg of body weight) and xylazine (12 mg/kg). After challenge, mice were monitored daily for weight loss and death for 14 days.
The ability of S. sonnei Invaplex AR to protect against a homologous challenge with S. sonnei strain Moseley and a heterologous challenge with S. flexneri 2a strain 2457T was assessed by immunizing BALB/cByJ mice intranasally on days 0, 14, and 28 and challenging them on day 49 using a similar study design as outlined above. Groups of mice were immunized with S. flexneri 2a Invaplex AR (2.5 g) or S. sonnei Invaplex AR (2.5 g). Control groups were immunized with saline (negative control), and the positive-control groups were immunized with S. flexneri 2a Invaplex NAT (5 g) or S. sonnei Invaplex NAT (5 g). On day 49, mice were challenged with either the homologous challenge organism (1.6 ϫ 10 7 CFU S. flexneri 2a in 30 l) or the heterologous challenge organism (8 ϫ 10 6 CFU S. sonnei in 30 l).
Guinea pigs, Hartley strain (6 per group), were immunized intranasally with 25 (23, for Invaplex AR ), 5, 1, 0.1, or 0.01 g/dose of S. flexneri 2a Invaplex NAT or Invaplex AR or with saline. Prior to intranasal immunization, guinea pigs were anesthetized with ketamine-xylazine. The antigen was delivered with a micropipette, 50 l per nostril. Guinea pigs were immunized on days 0, 14, and 28 and were bled on days 0, 28, 40, and 63. Four weeks after the third immunization, guinea pigs were challenged intraocularly with 3 ϫ 10 8 CFU of S. flexneri 2a strain 2457T in 25 l and observed daily for 5 days for the occurrence of keratoconjunctivitis (23). Research was conducted under Walter Reed Army Institute of Research (WRAIR) IACUC-approved protocols 11-BRD-27 (mouse) and IB03-10 (guinea pig) in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals and adhered to principles stated in the Guide for the Care and Use of Laboratory Animals (56).
PD 50 and ImmD 50 . The dose of Invaplex necessary to achieve 50% protection (PD 50 ) and the dose of Invaplex necessary to seroconvert 50% of immunized animals (ImmD 50 ) were calculated for the mouse and guinea pig models using an adaptation of the Reed-Muench LD 50 formula (24). To determine the PD 50 in the mouse model, animals surviving challenge were scored as protected. In the guinea pig model, an eye was scored "negative" if the eye score was 0/1 and positive if the eye score was 2/3 on day 3 postinfection. To calculate the ImmD 50 , a 4-fold or higher increase in the antigen-specific serum IgG endpoint titer over baseline was scored as positive in both animal models.
Adjuvant effect of S. flexneri 2a Invaplex AR . The ability of Invaplex AR to function as a mucosal adjuvant was determined in mice using OVA as a model protein antigen. BALB/cByJ mice (5 mice/group) were intranasally immunized on days 0, 14, and 28 with OVA (5 g) alone or OVA (5 g) combined with either Invaplex AR (2.5 g), Invaplex NAT (5 g), or cholera toxin (CT; 5 g). Blood was collected on days 0, 28, and 42. Lung washes were collected on day 42. Serum endpoint titers on days 0, 28, and 42 with specificity for Invaplex NAT and OVA were determined by ELISA. Shigella antigen-specific antibodies in mucosal washes were also assessed.