|Home | About | Journals | Submit | Contact Us | Français|
The ability of genetically detoxified lipopolysaccharide (LPS) to stimulate adaptive immune responses is an ongoing area of investigation with significant consequences for the development of safe and effective bacterial vaccines and adjuvants. One approach to genetic detoxification is the deletion of genes whose products modify LPS. The msbB1 and msbB2 genes, which encode late acyltransferases, were deleted in the Shigella flexneri 2a human challenge strain 2457T to evaluate the virulence, inflammatory potential, and acquired immunity induced by strains producing underacylated lipid A. Consistent with a reduced endotoxic potential, S. flexneri 2a msbB mutants were attenuated in an acute mouse pulmonary challenge model. Attenuation correlated with decreases in the production of proinflammatory cytokines and in chemokine release without significant changes in lung histopathology. The levels of specific proinflammatory cytokines (interleukin-1β [IL-1β], macrophage inflammatory protein 1α [MIP-1α], and tumor necrosis factor alpha [TNF-α]) were also significantly reduced after infection of mouse macrophages with either single or double msbB mutants. Surprisingly, the msbB double mutant displayed defects in the ability to invade, replicate, and spread within epithelial cells. Complementation restored these phenotypes, but the exact nature of the defects was not determined. Acquired immunity and protective efficacy were also assayed in the mouse lung model, using a vaccination-challenge study. Both humoral and cellular responses were generally robust in msbB-immunized mice and afforded significant protection from lethal challenge. These data suggest that the loss of either msbB gene reduces the endotoxicity of Shigella LPS but does not coincide with a reduction in protective immune responses.
Shigellosis, or bacillary dysentery, is an acute colitis caused by Shigella flexneri, a gram-negative enteroinvasive bacterium that is transmitted to humans via the fecal-oral route. Shigella triggers its uptake into the M cells of the lower intestine, where they are taken up by the underlying antigen-presenting cells (macrophages and dendritic cells) (18). Shigella bacteria are released from macrophages after inducing cell death (12, 29) and invade the surrounding enterocytes, where they begin to multiply and spread to adjacent cells. Effector proteins secreted through a molecular-needle-like complex called the type III secretion system (TTSS) mediate the processes of macrophage cytotoxicity, enterocyte invasion, and modulation of the host cell immune response. The TTSS and associated effector proteins are encoded on a large virulence plasmid that is present in all invasive strains of Shigella (46). During replication and dissemination in host cells, components of the bacterial cell wall (lipopolysaccharide [LPS] and peptidoglycan) are released, inducing proinflammatory cytokines and chemokines which activate the innate immune response (reviewed in reference 31). Although the immune mechanisms of protection remain relatively undefined, previous infection with one serotype of Shigella confers a high level of protection from reinfection with a homologous serotype, suggesting that LPS is a key protective antigen (13).
Lipid A is a bioactive component of LPS and serves to anchor it in the outer membrane through hydrophobic interactions involving fatty acids, such as laurate and myristate. In Shigella, Escherichia coli, and Salmonella enterica serovar Typhimurium, the htrB and msbB gene products function as late fatty acyltransferases and are therefore responsible for the final steps of assembling hexa-acylated lipid A (5, 6). The results of several studies have indicated that inactivation of htrB leads to morphological changes, such as bulging and filamentation, as well as a conditional lethal phenotype at temperatures above 33°C (22). In contrast, deletion of the msbB gene in E. coli K-12 reduces lipid A acylation without causing observable phenotypes (39). Moreover, purified LPS from an E. coli msbB mutant induced significantly less tumor necrosis factor alpha (TNF-α) from adherent monocytes and had a reduced ability to stimulate E-selectin production by human endothelial cells (39). The characteristics of E. coli msbB mutants encouraged investigations into using msbB mutants of other bacteria to produce live bacterial vaccines or other therapeutics which contain less-toxic LPS (42). Since this initial report, msbB homologues have been characterized in many different pathogenic bacteria, including Shigella spp., Salmonella spp., Neisseria spp., Yersinia pestis, Klebsiella pneumoniae, and enterohemorrhagic E. coli (EHEC) (1, 4, 9, 23, 24, 33). All of these investigations have led to a general consensus that the deletion of late acyltransferases, such as msbB, results in pathogen attenuation due to underacylated or less-toxic lipid A. Interestingly, some of these reports also suggest that changes in lipid A acylation can affect virulence by influencing outer membrane function, including TTSS-mediated protein secretion and cell division (4, 28, 32, 40, 49).
Among enteric pathogens, Shigella and EHEC are unique because they contain a second functional paralog of the msbB gene (msbB2) (2, 46). The msbB2 gene in both Shigella and EHEC is situated within the shf-wabB-virK (ecf3)-msbB2 locus found on the virulence plasmids of each organism. In S. flexneri 2a, the msbB2 gene appears to be transcriptionally regulated by PhoPQ in response to magnesium concentration (14). This differential regulation may be a mechanism by which lipid A modifications contribute to Shigella pathogenesis, but this has not been demonstrated directly. Loss of either msbB gene in S. flexneri reduces the amount of hexa-acylated lipid A under standard growth conditions, indicating that neither paralog is functionally redundant (9, 14). Loss of both msbB genes in a different background (S. flexneri 5) results in mostly penta-acylated lipid A, which correlates with a dramatic reduction in both TNF-α production and histopathology in infected rabbit ileal loops. Infection of rabbit ileal loops with an msbB single mutant leads to a less dramatic but still significant level of attenuation compared to the virulence of wild-type strains (9).
Several other studies have demonstrated the feasibility of generating less-endotoxic LPS through deletion of the msbB gene (4, 23, 25, 40, 44, 49). Collectively, they demonstrate a direct correlation between loss of msbB and reduced innate immune responses; however, few have measured the adaptive immune responses in the context of live attenuated bacterial vaccines (25). To address the relationship between the inflammatory potential and adaptive immunity induced by Shigella msbB mutants, a series of S. flexneri 2a mutants were constructed which lack either one or both copies of the msbB gene. The studies described herein represent the first comprehensive immunological evaluation of S. flexneri msbB mutants and identify a strategy for incorporating these mutations into live attenuated S. flexneri vaccine strains.
Shigella strains were propagated at 37°C on tryptic soy agar (TSA) plates containing 0.2% galactose and 0.005% (wt/vol) Congo red (CR). The wild-type S. flexneri 2a strain 2457T (10) is part of the Walter Reed Army Institute of Research (WRAIR) collection of virulent strains. 2457T-CR is a S. flexneri 2a strain which has spontaneously lost the large invasion plasmid and was selected as a white colony on CR agar plates. Loss of the invasive phenotype for 2457T-CR was confirmed by colony immunoblot using IpaB (2F1) (27) antibodies and gentamicin protection assays (described below).
The S. flexneri 2a strains carrying single deletions in the msbB1 (WR10) and msbB2 (WR20) genes were made in 2457T as previously described (34). In an earlier report, WR10 and WR20 were referred to as 2457T(ΔmsbB1) and 2457T(ΔmsbB2), respectively. The WR30 strain used throughout all of our studies was constructed from WR20 by deleting the msbB1 gene. However, a second version of WR30 was constructed from WR10 and used to confirm phenotypes observed in the plaque assays. Thus, both WR10 and WR20 were independently used to generate strain WR30, with deletions of both the msbB1 and msbB2 gene. All gene deletions were performed by lambda red recombineering using protocols that we described previously (34). PCR from genomic DNA using two different primer sets was used to confirm each deletion.
A plasmid containing the msbB1 gene was constructed using pBR322 as a starting vector (New England BioLabs [NEB], Beverly, MA). The msbB1 open reading frame, including 320 bp upstream and 150 bp downstream, was amplified from Shigella genomic DNA by using PCR. Both PCR primers (5′ msbB1.BamHI, ATACGCGGATCCCCACGCGTATTTTAACGGTA, and 3′ msbB1.BamHI, ATACGCGGATCCGTGAAACGTGGCGACCGTAT) were designed to introduce a BamHI site, which was used to insert the PCR fragment into the BamHI site of pBR322, creating pmsbB1. The pmsbB1 plasmid was verified using multiple restriction enzyme digestions.
Invasion assays or gentamicin protection assays were done using confluent (>90%) HeLa cell monolayers (CCL-2; ATCC) cultured overnight in 8-well chamber slides (Lab-Tek Chamber Slide System) using minimal essential medium supplemented with 10% fetal bovine serum (FBS) and l-glutamine (cMEM). Cells were infected with log-phase bacterial cultures at a multiplicity of infection (MOI) of ~50 as determined by measurements of optical density at 600 nm (OD600), followed by plating on TSA plates. Infected cells were incubated in a humidified CO2 incubator at 37°C for 1.5 h, washed three times with Hank's balanced salt solution (HBSS), and incubated in cMEM containing 50 μg/ml of gentamicin for 2 h. Cells were washed with HBSS, fixed with 100% methanol, and stained with Giemsa stain. Infected cells were observed under ×180 magnification using a bright-field microscope. The ratio of infected cells (containing at least 1 bacterium) to noninfected cells was determined for each experiment. An average of 886 HeLa cells were counted for each strain per invasion assay. Additional invasion experiments performed in 24-well plates were designed to evaluate bacterial uptake and did not include a gentamicin incubation step (see Fig. S3 in the supplemental material). HeLa cells were infected using an MOI of 100, incubated for 45 min as described above, and then washed 10 times with HBSS. An aliquot of the last HBSS wash was plated to LB agar in order to enumerate the extracellular bacteria remaining after washing. Intracellular bacteria were recovered by lysing the HeLa cells using 0.1% Triton X-100. The number of intracellular bacteria was determined by subtracting the number of extracellular bacteria from the total number recovered from lysed cells. The data presented in Table Table33 and the supplemental material (see Fig. S3) represent the average levels of invasion from 3 to 4 independent experiments.
Plaque assays were carried out using BHK-21 (CCL-10; ATCC) cells as described previously (34). Plaquing efficiency was calculated in assays using MOIs that ranged from 5.0 to 5.0 × 10−3. MOIs were determined as described above. Average plaque sizes were determined using 10 plaques per strain (72 h postinfection) from two separate experiments, using a Finescale Comparator (Finescale, Inc., Orange, CA).
Macrophage infection assays were done as previously described, with a few modifications (35). J774A.1 (TIB-67; ATCC) macrophages were grown in RPMI 1640 supplemented with 10% FBS and 2 mM l-glutamine (cRPMI) and used to seed a 96-well plate. After overnight growth, cells representing a monolayer were washed using fresh cRPMI and infected (in triplicate) with log-phase cultures of each strain, using MOIs of 2.0 and 0.2 as determined by plating and OD600 measurements. After the addition of bacteria, the 96-well plate was centrifuged (1,500 × g) and placed at 37°C, and culture supernatants were removed at different times (0, 120, 180, and 240 min) after infection. Wells that did not receive bacteria served as a negative control for spontaneous lysis. Lactate dehydrogenase (LDH) activity was measured from freshly harvested supernatants and used to estimate the percent cellular cytotoxicity according to the manufacturer's instructions (Promega, Madison, WI). The assay was performed three times, and the data (% cytotoxicity) were averaged and plotted as a function of time. Additional aliquots of culture supernatant from each macrophage infection assay were stored (−20°C) and later analyzed for the presence of three different proinflammatory cytokines (interleukin-1β [IL-1β], macrophage inflammatory protein 1α [MIP-1α], and TNF-α). Cytokine levels from culture supernatants were measured using a quantitative enzyme-linked immunosorbent assay (ELISA) according to the manufacturer's recommendations (R&D Systems, Minneapolis, MN). A preliminary kinetic analysis revealed that each cytokine reached a maximum level at different time points (data not shown). Therefore, the concentrations of individual cytokines were measured prior to saturation. The cytokine levels reported represent the averages from three independent experiments.
Shigella strains used in the Sereny test were assayed for invasiveness using an in vitro gentamicin protection assay as described above. The bacteria recovered were subcultured and prepared for ocular inoculation as previously described (35). For each strain tested, 4 male Hartley guinea pigs (~250 to 300 g; purchased from Charles River Laboratories) were sedated with a mixture of ketamine (10 to 35 mg/kg) and xylazine (2 to 5 mg/kg), and the conjunctival sac of the eye was inoculated with 25 μl of a bacterial suspension containing 1.25 × 107 to 2.0 × 107 CFU. Guinea pigs receiving 2457T were inoculated in one eye (a total of 4 eyes), while guinea pigs receiving WR10, WR20, or WR30 were inoculated in both eyes (a total of 8 eyes per strain). Using saline-inoculated eyes as a control, the level of inflammation for all animals was monitored for 5 days. The level of inflammation for each animal was rated numerically on day 3 using the following scale, modified slightly from the original publication: 0, no disease or mild conjunctivitis; 1, mild keratoconjunctivitis or late development and/or rapid clearing of disease; 2, keratoconjunctivitis with multifocal punctate corneal ulcerations and minimal ocular discharge; and 3, fully developed keratoconjunctivitis with purulence, copious ocular discharge, and corneal ulcerations affecting the entire cornea (16). Research was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to experiments involving animals and adhered to principles stated in the Guide for the Care and Use of Laboratory Animals (28a).
Bacterial suspensions used to inoculate mice were prepared from frozen stocks made from shaken broth cultures incubated at 37°C in LB. Bacteria were harvested during early log phase (OD600 of 0.2 to 0.3), suspended in M9 salts containing 15% glycerol, and quickly frozen in liquid nitrogen. Each stock (2457T, WR10, WR20, and WR30) was very similar with respect to CFU and percentage of CR-positive colonies. Accurate dosing was accomplished through dilution (in 0.9% saline) of each stock prior to inoculation. Six- to eight-week-old BALB/cJ (Jackson Laboratories, Bar Harbor, ME) female mice (n = 6 to 8 mice/group) were anesthetized intramuscularly with a 50-μl injection of ketamine HCl (90 mg/kg of body weight) and xylazine (10 mg/kg). Shigella strains were administered intranasally (2.5 × 107 or 5 × 107 CFU/mouse) in 30-μl volumes as small droplets to the external nares. Mice were evaluated for death twice daily for 3 days (72 h postinoculation). Mice that remained after 72 h were euthanized and processed for colonization and cytokine production assays and histopathology as described below. Challenge studies were performed twice for lethality and cytokine measurements and once for colonization assays and histology. Lung histopathology was performed on tissue that was harvested from the chest cavity, inflated with 10% neutral buffered formalin (NBF), and fixed in NBF for 4 days. The tissues were paraffin embedded, sectioned at 5 μm, mounted on a glass slide, and stained with hematoxylin and eosin (H&E). Quantification of the histological examination was based on criteria outlined elsewhere (3) and included the quantification (using a score of 0 to 4) of six different parameters. These parameters included the thickening of the interstitium, desquamation of the intra-alveolar epithelium, degree of mucopurulent exudate in the airways, number of neutrophils and mononuclear cells per field (×400 magnification), and degree of bronchus-associated lymphoid tissue activation. Bacterial colonization in lung tissue after challenge was determined by placing tissue samples in 5 ml of cold phosphate-buffered saline (PBS), pH 7.2, and homogenizing them to dislodge bacteria. Tenfold serial dilutions of the homogenate were plated on LB agar plates and incubated at 37°C. Cytokine production in lung tissue was assessed by washing the mouse lungs with 2 ml of lavage fluid containing PBS supplemented with a protease inhibitor cocktail for mammalian tissues (Sigma). The lung wash was centrifuged at 1,200 × g, and the supernatant analyzed by Luminex-based multiplex (22-plex) immunoassay to determine the concentrations of granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), IFN-γ, IL-10, IL-12, IL-13, IL-15, IL-17, IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-9, IP-10, KC, MCP-1α, MIP-1α, RANTES, and TNF-α, following the manufacturer's recommendations (LINCO Research, Inc., St. Charles, MO).
Adaptive immune responses were assessed using a vaccination-challenge study in which groups (n = 20 mice/group) of BALB/cJ mice were administered Shigella strains intranasally (3 × 105 CFU/mouse) on days zero and 21 as described above. Each group of mice received 2457T, WR10, WR20, or WR30. Control mice were inoculated with 30 μl of 0.9% saline. Animals were bled by tail snip on days zero, 14, 28, and 42. Lung and intestinal wash fluids were collected on day 35 from 5 mice/group as previously described (21). Splenocytes from individual mice (n = 5/group) were aseptically removed and homogenized to create single-cell suspensions, contaminating red blood cells were lysed, and splenocytes were diluted to 1 × 106 cells/ml for use in proliferation assays (see below). Four weeks after the last immunization (day 49), the remaining animals (15/group) were challenged with a lethal dose (1.2 × 107 CFU) of 2457T. Mice were monitored for weight loss (data not shown) and death for 14 days. Antigen-specific antibody responses in sera and mucosal wash samples were assessed by ELISA as previously described (21). The coating concentrations of the various antigens plated at 50 μl/well were Invaplex 50 (1 μg/ml), S. flexneri 2a strain 2457T LPS (20 μg/ml), IpaB (1 μg/ml), and IpaC (4 μg/ml). The Shigella invasin complex (Invaplex) is isolated from water extracts of S. flexneri 2a (2457T) using column chromatography, and the main constituents are LPS, IpaB, and IpaC (43). Sample endpoint titers were determined for individual animals as the reciprocal maximum dilution at which the mean OD405 of duplicate samples was greater than twice the mean plus 10 standard deviations of the results for casein-only controls or 0.2, whichever was higher. S. flexneri 2a Invaplex-specific serum immunoglobulin G1 (IgG1) and IgG2a responses were assessed as a measure of Th1 and Th2 responses as previously described (21). Cellular proliferation after in vitro restimulation of murine splenocytes with S. flexneri 2a Invaplex 24 (5 μg/ml) or S. flexneri 2a LPS (1 μg/ml) was assessed using a nonradioactive method as previously described (21). Cell culture supernatants collected after 5 days of in vitro antigenic stimulation were assessed for Th1 and Th2 cytokine production using a Luminex-based multiplex bead assay according to the manufacturer's directions (BioSource, Camarillo, CA). The concentrations of each cytokine were determined from a standard curve run in parallel with unknowns.
All statistical comparisons were performed using Prism 4 for Macintosh (GraphPad, San Diego, CA). Invasion assay data were analyzed by comparing the mean invasion rate for three separate assays using a one-way analysis of variance (ANOVA) followed by Bonferroni's multiple-comparison posttest. Endpoint titers from ELISA experiments were log transformed, and a two-way ANOVA or a one-sample t test was performed to detect differences between groups. P values for all cytokine analysis were determined using one-way ANOVA followed by Bonferroni's multiple-comparison posttest. Survival rates after challenge were analyzed using the log-rank test. P values of <0.05 were considered significant. The median lethal dose (LD50) was estimated using a probit analysis based on the SAS system.
The S. flexneri msbB genes encode late acyltransferases shown to be essential for hexa-acylation of lipid A (9, 14). Deletion of these genes, either separately or in combination, produces Shigella strains that induce less TNF-α production from human monocytes and are significantly attenuated in a rabbit ileal loop model (9). A series of isogenic deletion mutants were constructed in the S. flexneri 2a strain 2457T in order to evaluate the role of msbB in a virulent human challenge strain. The mutants lacked the chromosomally encoded msbB1 (WR10), the plasmid-borne msbB2 (WR20), or both copies of the msbB gene (WR30) (Table (Table1).1). Routine propagation of WR30 in LB revealed that this strain grew slightly more slowly than the wild type when incubated at 37°C (see Fig. S1 in the supplemental material). This was unexpected, because previous investigations of S. flexneri msbB mutants did not report growth defects after deletion of both msbB genes (9, 14). However, the growth phenotype of WR30 is consistent with that of Salmonella, where deletion of msbB induced growth defects, including slow growth, filamentation, and sensitivity to MacConkey agar (28). Indeed, it was found that WR30 was sensitive to growth on MacConkey plates and was filamentous (~10 to 30% of total bacteria) during log-phase growth at 37°C (data not shown). These phenotypes were complemented after transformation of WR30 with a low-copy-number plasmid containing the msbB1 gene (pmsbB1), indicating that the defects were not related to secondary mutations introduced during strain construction (data not shown).
The mouse pulmonary model is useful for assessing the virulence and inflammatory potential of Shigella strains, including but not limited to the production of soluble cytokines, lung colonization, and acute lethality due to necrosuppurative bronchoalveolar pneumonia (3, 45). Each S. flexneri msbB mutant was assayed in this model, using a dose that is 100% lethal for wild-type Shigella 72 h after infection (Table (Table2).2). This same dose (5 × 107 CFU) only produced 37 to 50% lethality using either single or double msbB mutant strains. A similar trend was observed using a lower dose of bacteria (2.5 × 107 CFU), which caused 25% lethality using 2457T and 0 to 6% for the msbB derivatives. All Shigella strains colonized the lung tissue at similar levels, with only slight reductions for the msbB mutants (Table (Table22).
The inflammatory and immunostimulatory potential of each msbB mutant was evaluated by comparing cytokine levels in lung wash fluids of mice 72 h after inoculation with 2.5 × 107 CFU of S. flexneri 2a 2457T, WR10, WR20, or WR30. The cytokines evaluated included hallmarks of inflammation (IL-1α, IL-1β, IL-6, MIP-1α, TNF-α, G-CSF, and GM-CSF), cytokines involved in chemotaxis (IP-10, KC, MCP-1, IL-17, and RANTES), and immunomodulatory cytokines [IFN-γ, IL-10, IL-12(p70), IL-13, IL-15, IL-2, IL-4, IL-5, IL-7, and IL-9].
The levels of IL-9, IL-4, IL-2, Il-10, GM-CSF, IL-17, and KC were not elevated in any of the groups inoculated with saline or bacteria (data not shown). The levels of IL-5, IFN-γ, IL-1α, IL-12, IL-6, IL-13, and TNF-α were elevated in groups inoculated with 2457T, WR10, WR20, or WR30 compared with the levels in saline-inoculated mice, but no significant differences were detected between groups inoculated with 2457T and the msbB mutants (Fig. (Fig.11 and data not shown). The concentrations of immunomodulatory cytokines in lung wash fluids were low for all cytokines measured and comparable between groups inoculated with saline, 2457T, or the msbB mutants, with the exception of IL-7 and IL-15, which were significantly decreased in mice inoculated with the msbB mutants compared to their levels in 2457T-inoculated mice (Fig. (Fig.11).
Of the inflammatory cytokines measured, significantly lower concentrations of MIP-1α, IL-1β, and G-CSF were found in the lungs of mice inoculated with msbB mutants compared to the concentrations in the lungs of mice inoculated with 2457T (Fig. (Fig.1),1), indicating a lower inflammatory potential of the msbB mutants at the 72-h time point. Inoculation of mice with 2457T induced high levels of the chemotactic cytokines IP-10, Rantes, and MCP-1, which were significantly reduced in mice inoculated with some msbB mutants (Fig. (Fig.1).1). Collectively, the results of the cytokine analysis demonstrate that the msbB mutants induced lower levels of proinflammatory and chemotactic cytokine production than inoculation with wild-type bacteria.
Histological analysis of infected mouse lung tissue was also conducted and scored using criteria outlined elsewhere (3). As expected, lung tissue in the wild-type Shigella inoculation group had interstitium thickening, intra-alveolar desquamation, and necrotic cellular debris, as well as increased levels of inflammatory (polymorphonuclear leukocytes) and mononuclear cells. However, there were no differences between groups of mice inoculated with 2457T or msbB mutant strains (data not shown).
The results described above suggested that Shigella strains lacking msbB have a reduced ability to elicit inflammatory responses in mouse lungs. To investigate the inflammatory potential of each strain in more detail, J774A.1 murine macrophages were infected with either wild-type or msbB mutant strains. Culture supernatants collected during the macrophage infections were analyzed for LDH (see Fig. S2 in the supplemental material), as well as for the presence of three proinflammatory cytokines (TNF-α, MIP-1α, and IL-1β), using a quantitative ELISA (Fig. (Fig.2).2). Cytokine production was similar for all strains when infections were done at an MOI of 2.0. However, infections using an MOI of 0.2 consistently resulted in a significant decrease in the amount of all three proinflammatory cytokines (Fig. (Fig.3).3). Complementation of the msbB double mutant using pmsbB1 [WR30(pmsbB1)] resulted in the restoration of proinflammatory cytokine production to levels that exceeded the levels with 2457T.
These data indicate that Shigella msbB mutants were fully cytotoxic to murine macrophages. However, the ability to elicit the release of proinflammatory cytokines was significantly reduced in Shigella strains lacking either one or both copies of the msbB gene.
Invasion of epithelial cells is a critical aspect of Shigella pathogenesis and was therefore evaluated for each S. flexneri msbB mutant. The efficiency of HeLa cell invasion was evaluated by determining the number of intracellular bacteria early after infection (45 min). Unlike the msbB single mutants and the wild type, WR30 was recovered from infected HeLa cells at reduced levels (see Fig. S3 in the supplemental material). Since defects in cellular invasion had not been previously reported for Shigella msbB mutants, the invasion defect was verified by counting the number of infected cells using a gentamicin protection assay (Table (Table3)3) (9, 14). Interestingly, microscopic examination of HeLa cells infected with WR30 revealed that the vast majority of infected cells had filamentous bacteria in the cytoplasm (data not shown). As with the other phenotypes, both the invasion defect and filamentation of WR30 were fully complemented with a plasmid expressing the msbB1 gene [WR30(pmsbB1)] (Table (Table3),3), indicating that at least one copy of msbB was required for wild-type levels of invasion in a 2457T background.
The plaque assay was used to further investigate the ability of each strain to invade, replicate, and spread within BHK cell monolayers, a process that results in areas of cell death or plaques after ~72 h (30). As expected, WR10 and WR20 formed plaques comparable to those formed by 2457T with respect to plaque size and plaquing efficiency. However, monolayers that were infected with WR30 did not contain visible plaques (Table (Table3).3). When a very high MOI was used (MOI of 5), only a few small areas of cell lysis were visible after infection with WR30. Interestingly, WR30 complemented with pmsbB1 displayed wild-type plaquing efficiency but produced plaques that were smaller in size. To address the possibility that additional mutations were introduced into WR30 during construction, the strain was independently derived a second time and the assay was repeated, producing the same result. these data suggest that complete complementation of all msbB phenotypes may not be achievable with pmsbB1 (data not shown).
A Sereny test, which is an in vivo assay for invasion and spread within epithelial cells, was used to evaluate the virulence of each S. flexneri msbB mutant strain. In this assay, bacteria are inoculated into the eye of a guinea pig, where they cause a purulent keratoconjunctivitis as a consequence of invasion, replication, and spread throughout the corneal epithelium. Consistent with its defects in cellular invasion and plaque-negative phenotype, WR30 was Sereny negative, while WR10 and WR20 induced a severe keratoconjunctivitis reaction (grade 3) that was indistinguishable from that induced by 2457T (Table (Table33).
To investigate the basis for the invasion-related defects, all three msbB mutant strains were assayed for TTSS-mediated protein secretion. Log-phase broth cultures were grown in the presence or absence of the TTSS activator Congo red. The addition of Congo red to msbB mutant strains induced secretion profiles that were identical to those of wild-type cultures (see Fig. S4B in the supplemental material). Under noninducing conditions, very few proteins were secreted into culture supernatants, indicating that TTSS protein secretion was regulated normally and the presence of effector proteins was not due to bacterial lysis. The identity and relative quantity of both IpaB and IpaC were verified via Western blot with monoclonal antibodies 2F1 and 2G2 (27), respectively (data not shown). Similar results were obtained for strains grown on agar plates, where TTSS secretion of IpaB was detected via colony blot (see Fig. S4A in the supplemental material).
Overall, the results indicate significant defects in epithelial cell invasion for the msbB double mutant (WR30). Although the basis for these defects is currently unknown, they are clearly related to the absence of a functional msbB gene. The virulence assays collectively indicate that S. flexneri strains lacking msbB are attenuated in both cell culture and animal models of infection. However, the nature and magnitude of the attenuation are different depending on the model system used (i.e., mouse lung versus Sereny test).
The mouse lung model was used to assess the adaptive immune responses induced by each Shigella msbB mutant strain. Mice were intranasally inoculated on days zero and 21 with ~3.0 × 105 CFU of 2457T, WR10, WR20, or WR30 and then challenged with a lethal dose (1 × 107) 4 weeks after the last inoculation. Blood samples collected on days zero, 28, and 42 were analyzed by ELISA for serum IgG and IgA responses against S. flexneri 2a LPS, IpaB, and IpaC. Antigen-specific responses were not detected in samples collected before immunization (day zero) or from saline-immunized animals. In contrast, LPS-specific IgG and IgA responses were detected on days 28 and 42 in all immunization groups regardless of the strain used for vaccination (Fig. (Fig.3).3). Comparable levels of anti-LPS serum IgG endpoint titers were elicited among all vaccinated groups on day 42, with endpoint titers ranging from 1,024 to 4,096. LPS-specific IgA responses showed a similar trend, with comparable mean endpoint titers among all groups on day 28, one week after the last immunization. While the anti-LPS IgA responses remained relatively high on day 42 in 2457T- and WR20-immunized animals, significantly lower endpoint titers were seen in WR10- and WR30-inoculated animals. Interestingly, whereas the LPS-specific IgG responses were comparable among the different groups, there was a significant reduction in the anti-IpaB IgG response in mice immunized with WR20 or WR30 compared to the response to immunization with wild-type Shigella or WR10 (Fig. (Fig.3C).3C). Minimal levels of IpaC-specific serum IgG were detected in all groups (Fig. (Fig.3D3D).
Mucosal responses induced by the S. flexneri msbB mutants were assessed by determining anti-LPS and anti-Invaplex IgA titers in intestinal and lung wash fluids collected two weeks (day 35) after the last immunization. Invaplex is a native complex isolated from water extracts of wild-type Shigella (43) and consists of IpaB, IpaC, and LPS, major antigens recognized after natural infection with Shigella and providing enhanced sensitivity with ELISAs. Lung and intestinal wash fluids collected from saline-immunized animals did not contain detectable levels of antigen-specific IgA, as expected (Fig. (Fig.4).4). Intestinal wash fluids contained comparable levels of antibodies for both LPS and Invaplex across all groups (Fig. 4A and B). The responses in lung wash fluids were more complex and variable (Fig. 4C and D). LPS-specific lung IgA responses in mice immunized with 2457T, WR20, and WR30 were of similar magnitudes. However, WR10-immunized mice did not have detectable levels of anti-LPS IgA in the lung (Fig. (Fig.4C).4C). The levels of anti-Invaplex IgA were comparable among all groups except for WR30-immunized mice, which had significantly lower levels.
In addition to antibody responses, the induction of antigen-specific cell-mediated responses was also determined two weeks after the last immunization (day 35). Splenocytes excised from individual mice were restimulated in vitro using S. flexneri 2a LPS or Invaplex (Table (Table4).4). Stimulation of cells from saline-inoculated mice did not induce cellular proliferation, with stimulation indices (SIs) below 2.0. Cells isolated from all animals responded to the lymphocyte mitogen concanavalin A at similar levels (range, 12.9 to 16.6). In contrast, cells from mice immunized with 2457T, WR10, WR20, and WR30 responded to both LPS and Invaplex, indicating the induction of Shigella-specific cellular immune responses (Table (Table4).4). The proliferative responses in WR30-infected mice tended to be more moderate (SI of 5.8 after LPS stimulation and 6.9 after Invaplex stimulation) than the proliferation levels induced by the wild-type strain (SI of 8.4 after LPS and 10.6 after Invaplex) or the single msbB mutants.
Analyzing serum IgG subclass responses and determining the secretion of cytokines after in vitro antigen stimulation characterized the phenotype of the immune response. Serum samples collected on day 42 from individual mice were analyzed for Invaplex-specific IgG subclass responses. The antigen-specific serum IgG subclass profile elicited by all immunized groups was consistent with a mixed Th1/Th2 response (Fig. (Fig.5),5), with detectable levels of IgG1 (indicative of Th2 responses) and IgG2a. The ratios of anti-Invaplex serum IgG1 endpoint titers to serum IgG2a titers were ≥4 in groups inoculated with 2457T, WR10, and 30, whereas the ratios calculated for WR20 were ≤2, indicating a more balanced Th1/Th2 response in WR20-inoculated animals. The phenotype of the immune response was also assessed by measuring cytokines secreted from splenocytes restimulated with S. flexneri 2a LPS (data not shown) or Invaplex (Table (Table5)5) in vitro. The results support a mixed Th1 and Th2 response, with similar levels of IL-12, IFN-γ, IL-5, and IL-10 secreted from splenocytes of mice immunized with 2457T, WR10, WR20, and WR30. Interestingly, cells from mice inoculated with WR10 had the highest levels of IFN-γ secretion after in vitro stimulation with Invaplex.
Mice intranasally immunized with 2457T, WR10, WR20, WR30, or saline were challenged with a lethal dose of 2457T (1 × 107 CFU) four weeks after their last immunization. Mortality was assessed on a daily basis for 14 days (Fig. (Fig.6).6). All strains induced significant protection compared to mortality in saline-immunized mice, with WR20 demonstrating the lowest level of protection (P < 0.012). Interestingly, WR10 demonstrates the highest level of protection and is significantly (P < 0.008) more efficacious than WR20.
Numerous studies have demonstrated a clear link between underacylated lipid A and reduced endotoxic potential. Although the molecular basis for this effect is not completely understood, studies suggest that it is related to a reduction in Toll-like receptor 4 (TLR4)-dependent activation via interaction with the endotoxin-MD-2 complex (7, 19). Since the acylation pattern of lipid A can influence its biological activity, translational research into biosynthetic lipid A modifications has been pursued as a potential way to create novel LPS species more suitable for inclusion into human biologics, such as vaccines and adjuvants. A step toward this goal involves the construction and testing of genetic mutations in late acyltransferases, such as msbB, which produce bacterial strains with underacylated LPS and reduced ability to stimulate innate immune responses. However, in the context of live bacterial vaccines, it is also important to consider how reduced endotoxicity influences the nature and magnitude of acquired immunity against protective bacterial antigens.
The mouse pulmonary model of shigellosis was chosen to evaluate the role of S. flexneri msbB because both innate and acquired immunity can be measured and correlated with protective efficacy. In this model, virulent Shigella bacteria instilled by the intranasal route invade the mouse pulmonary epithelium and elicit an acute, cytokine-driven inflammatory response that can rapidly progress to suppurative pneumonia (45). Proinflammatory cytokine secretion after primary infection is most likely due to recognition of bacterial products (i.e., LPS) by key receptors (TLR4 and CD14), which are rapidly upregulated on the bronchial epithelium and alveolar macrophages in the presence of LPS (36). Studies with a plasmid-cured Shigella strain or a strain lacking a key invasion-related protein (IpaB) suggest that cellular invasion and macrophage cell death are important for full virulence in this model (41, 50). In contrast, mutants deficient in their ability to move within a cell [ΔvirG(icsA)], which are highly attenuated in other animal models (guinea pigs and nonhuman primates) and humans (reviewed in reference 48), are virulent in mouse lungs, capable of inducing a cytokine-driven pathology that leads to significant lethality (41).
The results from the 22-plex bead-based assay demonstrate that high doses (2.5 × 107 CFU) of wild-type Shigella induce the production of 10 different chemokines and cytokines 72 h after intranasal challenge. Consistent with a reduced endotoxic potential, all three msbB deletion strains induced lower levels of IL-1β, IL-6, KC, MCP-1, IL-17, and MIP-1α in mouse lungs than were induced in lungs infected with wild-type 2457T. The basis for reduced virulence (i.e., lethality) in mouse lungs appears to be directly related to the production of less-endotoxic LPS. Evidence for this correlation comes from the finding that msbB single mutants induce lower levels of cytokine production and less mortality in the absence of any associated phenotypes. Interestingly, reduced lethality could not be correlated with significant histopathological differences in lung tissue between the different treatment groups. That is, pneumonia and hemorrhage with necrosuppurative bronchiolitis do not necessarily portend a lethal outcome (45). Moreover, the pathology is consistent with the persistent colonization of the lung tissue (Table (Table2)2) and the ability to induce cell death in murine macrophages (3). The lung pathology may also reflect the finding that lung lavage fluids from all infected groups contained comparable levels of TNF-α, a key cytokine responsible for causing lesions in experimentally infected animals (9). The TNF-α results do not coincide with a previous finding using S. flexneri 5 msbB mutants in rabbit ileal loops (9). This most likely reflects the promiscuous nature of the murine TLR4, which can fail to effectively discriminate between penta-acylated and hexa-acylated LPS (15), especially at high MOIs (20). Indeed, reduced TNF-α secretion for each S. flexneri msbB mutant was only detected when relatively low MOIs were used to infect murine macrophages. Under these conditions, both single and double msbB mutants induced significantly less TNF-α, MIP-1α, and IL-1β secretion in culture supernatants. Overall, these findings demonstrate that loss of msbB in S. flexneri 2a results in reduced endotoxic responses in mouse lungs and cultured macrophages.
The msbB gene was originally identified in E. coli (K-12) as a high-copy-number suppressor of the conditionally lethal htrB mutation (6, 39). Although no growth phenotypes were attributed to deletion of msbB in K-12, subsequent mutation of msbB in some pathogenic gram-negative bacteria has been shown to induce growth defects, such as filamentation, sensitivity to MacConkey agar, and slow growth (28, 40). Deletion of both msbB genes in the S. flexneri 2a human challenge strain 2457T (WR30) results in a slow-growth phenotype that is associated with a filamentous bacterial morphology. WR30 also displays a temperature-dependent sensitivity to growth on MacConkey agar plates (data not shown). All of these growth defects could be recovered by expressing MsbB from a plasmid, indicating that they were not due to secondary mutations elsewhere in the genome.
WR30 is less invasive in HeLa cells and does not form plaques in BHK cell monolayers. Both defects directly correlate with a Sereny-negative phenotype. In Salmonella and Neisseria gonorrhoeae msbB mutants, defects in cellular invasion were attributed to either reduced secretion of Sip proteins or increased susceptibility to innate mechanisms of intracellular killing (11, 49). WR30 appears to have a functioning TTSS, as indicated by the controlled secretion of the Ipa proteins (see Fig. S4 in the supplemental material). Moreover, WR30 efficiently kills macrophages, a process that requires a TTSS that can secrete functional translocators (IpaB and -C) (37, 38). The phenotype for WR30 in epithelial cells may be a pleiotropic effect resulting from its filamentous morphology, even though log-phase cultures contain similar numbers of viable bacteria. The data presented in Table Table33 and Fig. S3 in the supplemental material indicate that WR30 induces uptake into HeLa cells at a reduced efficiency and then fails to replicate normally once inside the cytoplasm. The defects in cellular invasion were complemented with MsbB1 expressed from a low-copy-number plasmid, suggesting that WR30 has not acquired secondary mutations. However, these observations are not consistent with the results of a previously published study (9) where no in vitro growth or invasion-related defects were observed with an msbB double mutant. The conflicting results may reflect the fact that late acyltransferases (like msbB) can be strain-specific virulence factors in Shigella, similar to what has been found for both Neisseria and Salmonella species. Another possible explanation could be related to the fact that msbB mutant backgrounds (in Salmonella) are reported to have higher rates of spontaneous suppressor mutations (28). Extragenic suppressors picked up during strain construction could suppress growth phenotypes without affecting the pattern of lipid A acylation (28).
The mouse pulmonary vaccination-challenge model has been used to evaluate a number of Shigella vaccine candidates, including live attenuated vaccine strains (26). Mice immunized with two sublethal doses of Shigella develop systemic (IgG and IgA) and mucosal (secretory IgA) antibodies to bacterial antigens and are protected (56 to 79%) against lethal challenge (45). Protection from challenge in this model appears to be antibody mediated, although the results of other studies suggest an essential role for IFN-γ in resistance to primary infection (50, 51). Immunization using two sublethal doses of each msbB derivative or wild-type Shigella induced equally significant serum antibody (IgA and IgG) responses to LPS. However, this was not true for anti-IpaB serum responses, which were significantly reduced in animals inoculated with WR20 and WR30 even though these strains clearly expressed wild-type levels of IpaB (see Fig. S4A and B in the supplemental material). Anti-IpaC responses were below the level of detection in the sera of all vaccinated groups, even in animals immunized with 2457T. The lack of a detectable IpaC response could be a result of the immunizing dose being relatively low compared to the dose in previous studies where only a modest anti-IpaC response was detected (45).
Significant differences were observed in the magnitude of the mucosal immune responses in the lungs of mice in some groups (WR10 and WR30), whereas the intestinal wash fluids showed comparable responses across all strains. While not as pronounced, the level of cell-mediated immunity was also reduced in animals inoculated with WR30 compared to the level in mice inoculated with 2457T. Despite these differences, there were no variations noted in the phenotype of the response, with all strains inducing comparable levels of Invaplex-specific IgG1 and IgG2a, indicative of a mixed Th1/2 response. When mice immunized with 2457T or its msbB derivatives were subsequently challenged with a lethal dose of 2457T, the adaptive immune responses evoked in all 4 groups of mice correlated with significant protection from challenge compared to that in saline-immunized mice. WR10 induced the highest level of protection and is significantly (P < 0.01) more efficacious than WR20. Splenocytes from WR10-immunized mice also induced the highest level of IFN-γ after in vitro stimulation with Invaplex. IFN-γ plays a critical role in promoting the clearance of intracellular shigellae (50). However, protection from lethal challenge after immunization with wild-type S. flexneri 2a (2457T) can range from 56 to 79% (45), making it difficult to interpret the increase in protection observed with WR10.
The data presented herein are consistent with previously published work that shows that deletion of msbB in Shigella significantly reduces innate inflammatory responses compared to those induced by wild-type strains (9). Most importantly, the reduction does not coincide with a consistent decrease in acquired serum or mucosal immunity or protective capacity. Live attenuated vaccine strains, like SC602 (S. flexneri 2a), WRSS1 (S. sonnei), and WRSd1 (S. dysenteriae 1), have the loss of the virG(icsA) gene (8, 17, 47) as their primary attenuating feature. Although these vaccines are highly attenuated and immunogenic, they induce mild fever and diarrhea in an unacceptable percentage of human volunteers (reviewed in reference 48). Incorporation of an msbB deletion into existing virG(icsA)-based vaccine strains will likely reduce or eliminate these febrile reactions and increase the safety of these low-dose vaccine candidates.
We thank Edwin Oaks for the IpaB and IpaC monoclonal antibodies, Shigella Invaplex, and the mouse challenge stocks used in these studies. We also thank Edwin Oaks, Robert Bowden, and Akamol E. Suvarnapunya for critically reading the manuscript. Finally, we thank Richard Helm for completing the analysis of LPS and Suramya Fonseka for cellular invasion assays.
The content of this publication does not necessarily reflect the views or policies of the U.S. Department of the Army or the U.S. Department of Defense, nor does the mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.
Editor: A. Camilli
Published ahead of print on 2 November 2009.
†Supplemental material for this article may be found at http://iai.asm.org/.