Strains of Salmonella, Escherichia coli APEC O78, and Shigella flexneri 2a were routinely grown in Luria-Bertani (LB; Théry et al., 2018) broth or on agar at 37°C. To indicate the reproducibility of isolated OMVs derived from omp mutants in different cultures, an acidic low-magnesium minimal medium (acidic MgM), which partially reproduced the intracellular environment, was also used to culture Salmonella for purification of OMVs (Yoon et al., 2009; Bai et al., 2014). The bacterial strains and plasmids used in this study are listed in Table 1. To construct the mutant strains, LB agar with 5% sucrose was employed for sacB gene-based counterselection (Kong et al., 2011). The growth of Δasd strains required 50 mg/ml diaminopimelic acid (DAP; Nakayama et al., 1988). All primers used in the study are listed in Table 2. E. coli was transformed by electroporation, and LB agar plates containing appropriate antibiotics were used to select transformants. We used the suicide plasmid method to construct mutant strains as previously described, and the bacterial strains used were derived from previous experiments (Liu et al., 2016a,b). The resulting PCR product had a terminal A base when using the LA Taq enzyme (TaKaRa, Otsu, Japan), and the amplicon was then inserted into the pYA4278 vector with a terminal T-overhang, generated by AhdI digestion. The same strategy was used to construct pQK014 for ompA gene deletion, pQK015 for ompC gene deletion, and pQK016 for ompD gene deletion. All mutations were introduced into Salmonella Typhimurium K084, resulting in a knocked out flagellin and truncated LPS to improve the cross-proteins of OMVs, generating the K021 mutant (Liu et al., 2016a,b).

Murine macrophage RAW 264.7 cells were cultured at 37°C with 5% CO₂ atmosphere in Dulbecco’s modified Eagle’s medium (Gibco BRL, United States) supplemented with 10% fetal bovine serum (HyClone, United States). All experimental protocols were approved by Nanchang University.

Native OMVs (without any detergent processing) were collected and isolated from culture supernatants of S. Typhimurium UK-1 and its mutants, as described previously (Muralinath et al., 2011). Briefly, 2 L bacterial culture supernatants in the logarithmic growth phase (OD₆₀₀ = 1) were collected by filtration through a 0.45-μm Steritop bottle-top filter unit (Millipore, Bedford, MA, United States). Then, the vesicles present in the collected supernatants were pelleted by centrifugation (40,000 × g, 4°C, 2 h) and resuspended in Dulbecco’s phosphate-buffered saline (DPBS; Mediatech, Manassas, VA, United States). Further, pelleted OMVs were purified using density gradient centrifugation (200,000 × g, 4°C, overnight) with a discontinuous OptiPrep density gradient medium [20, 25, 30, 35, 40, and 45% OptiPrep in 10 mM HEPES (pH 6.8) with 0.85% NaCl, from top to bottom; Sigma-Aldrich, St. Louis, MO, United States].

The concentration of OMVs was determined based on total protein concentration using a bicinchoninic acid assay (Thermo Pierce, Rockford, IL, United States). Each OMV sample (10 μg, based on the protein content) was analyzed by 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), and the gels were stained using the GelCode™ blue stain reagent (Thermo Pierce).

Particle size distribution and concentration of OMVs were determined by nanoparticle tracking analyses (NanoSight NS300, Malvern Instruments, United Kingdom). Purified OMV samples were diluted up to a concentration acceptable for analysis within the recommended range of 100–120 visible particles per frame. Triplicate videos of each sample were taken using the equipped 532-nm green laser, and a high-sensitivity video with camera level of 14 was captured and performed using NanoSight 3.0 or 3.1 software, with a threshold of between 3 and 5. Triplicate video statistics were averaged for each sample. Further, the number of OMVs per milliliter was normalized to colony-forming unit (CFU) per milliliter of each strain to determine the number of OMVs per CFU (White et al., 2018). Approaches for determining size and production met the MISEV 2018 recommendations (Théry et al., 2018).

Six- to eight-week-old female BALB/c mice were purchased from Dashou Biotechnology Co., Ltd., and were allowed to adapt to their new environment for 1 week before the animal experiments were initiated. Lohmann chickens (1 days old) were purchased from Muxing Poultry Co., Ltd. (Chengdu, China) and were maintained under standard conditions until they were 10 days old. All experiments involving animals were approved by Animal Use Ethical Committee of Nanchang University and were conducted in compliance with the Animal Welfare Act and related regulations (Hansen et al., 2017). The principles stated in the Guide for the Care and Use of Laboratory Animals were followed (National Research Council, 2011).

Mice of approximately equal weights were divided into groups (n = 5 each) for intranasal (20 μg OMVs based on protein content in 10 μl DPBS per animal) and intraperitoneal immunizations (5 μg OMVs based on protein content in 100 μl DPBS per animal), which were performed two times without any adjuvant use. The negative control group was served the equivalent volume of DPBS. At 4 weeks, the doses were boosted. To determine the systemic and mucosal immunity induced by OMVs, blood samples of mice were collected via orbital sinus puncture, as well as vaginal secretions by repeatedly flushing the animals with 0.1 ml of PBS buffer at 2-week intervals after the first immunization. The blood serum and the supernatant of a vaginal wash were centrifuged and stored at −80°C for future use. To calculate the rate of protection, 5 weeks after the booster immunization, the mice were challenged with a lethal dose of S. Typhimurium S100 (1.1 × 10⁹ CFU, determined by continuously diluting and then counting on the LB plate) in 20 μl of buffered saline with 0.01% gelatin (BSG) via oral gavage. In another experiment, mice were immunized as above and then challenged with 1.3 × 10⁷ CFU of S. Choleraesuis (∼100-fold LD₅₀), 1.2 × 10⁷ CFU of S. Enteritidis (∼100-fold LD₅₀), 1.4 × 10⁸ CFU of APEC O78 (∼100-fold LD₅₀), and 1.1 × 10⁷ CFU of S. flexneri 2a (∼100-fold LD₅₀) in 20 μl of BSG 5 weeks after the booster immunization by oral gavage. Mice were monitored daily for weight, illness, and signs of obvious discomfort, distress, or pain. Mice that exhibited a weight loss of greater than 20% of their starting weight were euthanized via carbon dioxide followed by cervical dislocation. There were no expected deaths in this mice experiment.

To evaluate the protective efficacy of OMVs in chickens, 10-day-old Lohmann chickens were immunized intranasally (75 μg OMPs or OMVs in 100 μl DPBS per bird) at a 2-week interval as potential immune-protective administration, and 100 μl DPBS was used as a negative control. There were four groups of 20 chickens each (Wang et al., 2019). We marked the day of the first immunization as day 0 and the booster day as week 2. At weeks 1 and 4, 1–2 ml of blood was aseptically drawn from the chicken brachial vein after the first and second immunization. Then, we incubated the collected blood at room temperature for 30 min, removed the blood clot with sterile toothpicks, and centrifuged at 1,000 × g for 15 min. The supernatants were gently removed, and the prepared serum samples were stored at −80°C.

Protection of OMVs in chickens was evaluated in two different experiments. Experiment 1 was performed to determine the protection against wild type APEC O78 strain, and the immunized chickens were challenged via the air sac route with a lethal dose of the wild type APEC O78 strain (1.3 × 10⁸ CFU, ∼100-fold LD₅₀ in chickens), which were suspended in 20 μl of BSG, at 3 weeks after booster administration (week 5). After challenge, the chickens were monitored for 7 days, and the clinical signs scored as follows: none = 0, reluctance to walk = 1, coughing, head shaking, and unwillingness to eat = 2, gasping, or lying down of a long time, or both, accompanied by ruffled feathers = 3, and death = 4 (Grgić et al., 2008). Dead chickens were necropsied immediately on the day of death. At the end of the 7 days of observation period, the surviving chickens were sacrificed and necropsied. Macroscopic lesions in the heart and liver were recorded and scored separately as follows: heart and pericardium (normal = 0, turbid with excessive or cloudy fluid in the pericardial cavity or partial pericarditis = 1, marked pericarditis = 2, and severe pericarditis or death = 3) and liver (normal = 0, small amount of fibrinous exudate = 1, marked perihepatitis = 2, and severe perihepatitis or death = 3; Nagano et al., 2012). Experiment 2 was performed to determine the protection against S. Enteritidis, and the immunized chickens were inoculated orally with 5 ml of DPBS containing 1 × 10¹⁰ CFU of S. Enteritidis at 3 weeks after booster immunization. LD₅₀ of S. Enteritidis in chicken was not determined because this strain did not cause the death of chickens in our studies. Therefore, we monitored the fecal shedding of challenge strain to evaluate the protective efficacy against S. Enteritidis in chickens. Feces (5 g) were collected by rectal palpation and then were suspended in tetrathionate broth (TT; Oxiod, United Kingdom) and 10-fold serially-diluted suspensions were plated on brilliant green agar (Oxiod) supplemented with 1 μg/ml novobiocin and 50 μg/ml nalidixic acid (Oxiod). The colonies were counted after 24 h and five colonies were randomly selected for agglutination with anti-S. Enteritidis (ThermoFisher Scientific). The animal experiments were performed twice, and the data were combined for analysis (Figure 1).

Outer membrane proteins were isolated from Salmonella, APEC, and Shigella as previously described (Carlone et al., 1986). IgG and IgA responses of mice were determined using a quantitative ELISA method, and the titers of anti-OMP IgG in chicken serum samples were determined by ELISA as described below. Briefly, NUNC MaxiSorp™ 96-well plates (Thermo Scientific, Waltham, MA, United States) were coated with 2 μg of OMPs from various serotypes of Salmonella, APEC, and Shigella per well in 100 μl of sodium carbonate/bicarbonate coating buffer (pH 9.6). The plates were then incubated at 4°C overnight. To construct standard curves for each antibody isotype and make the experimental data more accurate, plates were coated in triplicate with 2-fold dilutions of appropriate purified mouse Ig isotype standard IgG and IgA (BD Biosciences, San Jose, CA, United States), starting at 0.5 μg/μl. PBS containing 0.1% Tween 20 (PBST) was used to wash the plate and 2% bovine serum albumin was used to block the plate at room temperature. Subsequently, a 100-μl volume of a suitably diluted sample was added to wells and incubated for 1 h at room temperature. After washing the wells, biotinylated goat anti-mouse IgA and IgG (Southern Biotechnology, Inc., Birmingham, AL, United States) were added. The streptavidin-alkaline phosphatase conjugate (Southern Biotechnology, Inc.) was used to develop the protein signals and the p-nitrophenyl phosphate substrate (Sigma-Aldrich) was used to detect the signal of each well. Absorbance was read at 405 nm using an automated ELISA plate reader (iMark; Bio-Rad, United States). The final Ig isotype concentration of the serum samples for quantitative ELISA was calculated using appropriate standard curves, and a log-log regression curve was calculated from at least four dilutions of the isotype standards and used for ELISA.

The opsonophagocytosis assay was performed in serum samples to determine the function of OMV-raised antibodies, as previously described (Campbell and Edwards, 2000). RAW 264.7 macrophages (5 × 10⁵ cells/well) were seeded in 24-well plates and incubated in RPMI complete medium without antibiotics (500 μl/well; Gibco, United States) for 2 h at 37°C, 5% CO₂ atmosphere. To perform opsonization with the antibody and complement system, a serum sample was diluted 1:100 and incubated with bacterial suspensions of 10⁵ CFU of APEC O78 and S. Enteritidis at 37°C for 10 min. To detect phagocytosis, we added the above opsonized bacteria to a macrophage monolayer in 24-well plates at a 1:10 (cells/bacteria) multiplicity of infection. The plates were incubated in an incubator at 37°C with 5% CO₂ atmosphere for 30 min to allow phagocytosis of the bacterial cells by macrophages. One hundred microgram of gentamicin per ml cultured for 72 h was used to kill unattached extracellular bacteria, and the plates were washed three times with PBS. The intracellular bacteria were released by immediately lysing the macrophages with 1% Triton X-100 at 37°C for 10 min. The lysates were consecutively diluted 10-fold and plated onto LB plates, which were incubated at 37°C overnight. The numbers of phagocytosed bacteria were calculated by enumerating the colonies on the plates.

A modified serum bactericidal activity (SBA) assay was performed to evaluate the bactericidal ability of antibodies in serum samples in vitro. The serum was diluted 10-fold with LB medium. We cultured the wild type APEC O78 strain and S. Enteritidis to the logarithmic phase (OD₆₀₀ = 0.7) and then suspended the cells in PBS at a concentration of 10⁶ CFU/ml. Ninety microliter of the diluted sera was mixed with 10 μl of the bacterial suspensions in the wells of 96-well plates and then incubated at 37°C with shaking at 100 rpm for 1 h. The mixture was consecutively diluted 10-fold and inoculated on plates at 37°C overnight, followed by counting the viable colonies. The survival rates of bacteria in different serum samples were calculated by dividing the CFU counts after serum treatment by the original CFU counts. To make the results more accurate, we independently tested all serum samples three times.

All statistical analyses were performed using the GraphPad Prism 5 software (GraphPad Software, San Diego, CA, United States). One- or two-way analysis of variance was performed followed by Tukey’s post hoc test to determine the significance of differences between the mean values of the experimental and control groups. Values are presented as mean ± standard deviation (SD). Differences in the survival rates among all groups were analyzed using the log-rank sum test. p < 0.05 was indicated statistically significant.

Outer membrane vesicles purified from strains K084, K015, K016, K017, K018, K019, K020, and K021 (all mutations were identified by PCR and gene sequencing, Supplementary Figure 1) were evaluated using protein profile assays. The major OMPs, including OmpA, OmpC, and OmpD, are marked in Figure 2A on the right, which were consistent with previous study (Liu et al., 2016a). Except the corresponding bands of the OmpC, OmpD, and OmpA proteins, the bands of other proteins in the protein profiles were basically the same. The results were consistent with the expected deficiency of certain proteins. Further, cryo-EM data showed that OMVs purified from the ΔompAΔompCΔompD mutant strains were spherical, and no other contaminant was visible, which indicated that deletion of major OMPs did not affect the structure of OMV particles (Supplementary Figure 2). We also determined the particle size of OMVs and enriched fractions during density gradient purification process using nanoparticle tracking analysis (Théry et al., 2018). The results showed that the majority of the produced OMVs derived from omp mutant strains and parental strain had a diameter of 70–90 nm, and the highest concentration of particles was in consecutive fractions 25 and 30% OptiPrep (Supplementary Figures 3, 4), which indicated that the omp knockout mutants did not affect the particle size of OMVs secreted by the bacteria. Further, we observed a statistically significant increase in OMVs release from ΔompA, ΔompAΔompC, ΔompAΔompD, and ΔompAΔompCΔompD compared to that in the parental strain (Figure 2B). The results based on protein quantification of OMVs also confirmed this observation (Figure 2C). Moreover, the results of OMVs cultured in acidic minimal medium (MgM, pH 5.0) showed similar trends as those in LB medium (Supplementary Figure 5). Compared with that of the parental strain, the omp knockout mutants showed certain effects on the OMV yields both in LB and acidic MgM growth conditions, among which the ompA mutant strain had a significantly higher yield than did the parental strain, suggesting that ΔompA might be beneficial for OMV production.

To verify the immunogenicity of OMP-deficient OMVs, mice were intranasally and intraperitoneally injected with OMVs purified from the Δomp mutants and the parental strain, and systemic and mucosal antibody levels were measured in the mice. When the mice were intraperitoneally injected, a small bump appeared in their abdominal cavity but disappeared in less than 12 h. All OMVs elicited lower levels of IgG response via the intranasal route than that via the intraperitoneal route at 4 weeks post-immunization, whereas those from the ΔompCΔompD mutant caused a relatively higher production of serum IgG than did the other groups, except the parental OMVs, regardless of the administration route. After booster immunization, a relatively higher IgG level was observed in the group immunized with parental OMVs compared with the other groups; by contrast, the production of IgG from the ΔompCΔompD OMV group was higher than those from the other groups of Δomp mutants (Figures 3A,C). We also determined the levels of anti-OMP IgA in vaginal secretions from the mice as an indicator of mucosal response. After boosting, we could observe a sharp increase in the levels of mucosal IgA, with mice injected with OMVs from the ΔompCΔompD mutant having higher levels than the mice immunized with the other OMVs from Δomp mutants, which was consistent with the effect of IgG level (Figure 3B). Measurements of the IgA responses induced via the intraperitoneal route showed that the values of all groups were lower than 0.5 μg/ml, and no significant difference was observed. At 5 weeks after boost immunization, we infected the mice with 1 × 10⁹ CFU (approximately 2,000 × LD₅₀) of wild type S. Typhimurium strain S100 to monitor the protective efficiency and recorded survival rates. Immunization with OMVs from the parental strain, ΔompCΔompD, and ΔompAΔompCΔompD mutants induced effective protection against the challenge. All the mice in the control (PBS) group succumbed within 10 days after infection with wild type S. Typhimurium strain S100 (Table 3).

Our previous studies have reported that truncation of the bacterial outer membrane may continuously expose conserved antigenic proteins to improve the cross-immune effect of OMVs (Liu et al., 2016a,b). Therefore, we evaluated the cross-immune responses of OMVs derived from different OMP mutants. In short, we determined the concentrations of IgG in the mice immunized with OMP-deficient OMVs and in those immunized with OMPs isolated from S. Enteritidis, S. Choleraesuis, APEC O78, and Shigella by a quantitative ELISA to evaluate the effects of cross-immunity. The result for the two routes of immunization showed that the IgG concentrations were roughly the same, as both routes resulted in high IgG levels against OMPs derived from S. Choleraesuis, S. Enteritidis, APEC O78, and Shigella. Regardless of the immunization route, analysis and integration of the data showed that the anti-OMP_SC IgG concentrations in the ΔompAΔompCΔompD and ΔompA groups were slightly higher or similar to those in the parental OMV group via intranasal or intraperitoneal immunization, respectively (Figure 4A). Regarding S. Enteritidis, the result of cross-immunity in the ΔompAΔompC group showed that the level was close to that in the group of parental OMVs; only the ΔompAΔompCΔompD group performed better than did the parental OMVs group (Figure 4B). OMVs from the ΔompA, ΔompAΔompC, ΔompAΔompD, and ΔompAΔompCΔompD mutants elicited remarkedly higher levels of serum IgG than did the parental OMVs against OMPs from APEC O78 (Figure 4C). A significant increase in IgG levels against Shigella OMPs was observed in the sera from the mice immunized with OMVs from the ΔompA and ΔompAΔompCΔompD mutants compared with those in the sera from the mice immunized with OMVs from the parental strain (Figure 4D). Both intranasal and intraperitoneal injections resulted in a concentration of IgG that was half that of the homologous group but still showed a good protective efficacy.

To assess cross-protective productivity, mice from different groups were immunized with OMVs from the ΔompA, ΔompAΔompC, ΔompAΔompD, and ΔompAΔompCΔompD mutants via the intraperitoneal and intranasal routes, with the PBS group used as a control. Similar to the experiments on cross-reactivity, mutant strains with a better efficacy were selected to explore their potential as cross-protective vaccines. In the case of Salmonella with different serotypes, the groups immunized with OMVs derived from various OMP mutants all showed good cross-protection, regardless of the route of immunization, compared with that in the PBS group (Figure 5A,B). In particular, the group immunized with ΔompAΔompCΔompD OMVs via the intraperitoneal route showed better or similar protection than did the parental group (Figures 5E,F). Even upon challenge with APEC O78 (Figures 5C,G) and Shigella (Figures 5D,H), ΔompAΔompCΔompD OMVs still provided an established protection efficiency compared with OMVs from the parental strain and other Δomps mutant strain. Irrespective of the wild type strain used as a challenge and the immunization route, OMVs from the ΔompAΔompCΔompD mutant strain remained highly protective, and the survival rate of mice remained above 50%. In the case of the homologous bacterial infection, the survival rate even exceeded 80%.

Based on the results of the previous experiments in the mouse model and the current status of S. Enteritidis as a major food-borne pathogen, we further studied the immunity of mutant vaccine to S. Enteritidis in a chicken model. To verify the cross-immunogenicity of OMVs from the ΔompAΔompCΔompD mutant, which could induce high levels of cross-protection and high survival rates after the challenge in mice, chickens were intranasally immunized with purified OMVs from strain K021 and the parental strain (K084), as well as with OMPs isolated from S. Enteritidis, as potential immune-protective injections. At weeks 1 and 4, the anti-OMP_SE IgG was at a high level in the ΔompAΔompCΔompD OMVs group, which was consistent with that of SE OMPs group (Figure 6A). In the period between week 1 and week 4, the bacterial survival rates declined, and at week 4, levels of ΔompAΔompCΔompD OMVs group were significantly decreased compared with the SE OMP group, which indicated that booster immunization with OMVs from the ΔompAΔompCΔompD mutant could significantly enhance SBA and display a long-term bacterial killing or growth suppression effect (Figure 6B). Compared with that of the PBS- and SE OMP-immunized groups, a significantly higher capability for opsonophagocytosis was shown at all time points (weeks 1 and 4) by the parental OMV- and ΔompAΔompCΔompD OMV-immunized groups (Figure 6C). Immunized chickens in the different groups were challenged via the air sac route with 1 × 10¹⁰ CFU of SE at week 5 to evaluate the cross-protective efficiency. As shown in Figure 6D, a significant level of reduction in shedding was observed in the ΔompAΔompCΔompD OMVs-immunized group of chickens, and birds were negative by day 17 post-challenge. Moreover, the ΔompAΔompCΔompD OMVs-immunized group could shed higher levels of the challenge strain compared with that the parental OMVs immunized group and SE OMP-immunized group (Figure 6D).

Further, we also determined the cross-immunogenicity of ΔompAΔompCΔompD OMVs to APEC O78 in chickens and evaluated their cross-protective efficacy. At the first week, the production of anti-OMP_O78 IgG in all experimental groups exhibited levels significantly higher than that in the PBS group. At week 4, the titers of anti-OMP_O78 IgG in the ΔompAΔompCΔompD OMVs group showed a similar trend as in the O78 OMP group, and were significantly higher than those in the parental group (Figure 7A). In addition, SBA assay showed that the bacterial survival rate of all experimental groups was significantly lower than that of PBS group at the first week; all experimental groups maintained this trend at the 4th week, although the ΔompAΔompCΔompD OMVs group showed significantly lower bacterial survival rate than O78 OMP group (Figure 7B). Further, the bacterial uptake rate in the ΔompAΔompCΔompD OMVs-immunized group was significantly higher than that in the O78 OMP group (Figure 7C). The post-challenge survival rate in the ΔompAΔompCΔompD OMVs group approached 60%, higher than other groups, even in the O78 OMP group (Figure 7D). Moreover, immunization with ΔompAΔompCΔompD OMVs provided significant reductions in clinical score compared to the parental and PBS groups. Similarly, liver lesion score was lesser in the ΔompAΔompCΔompD OMVs immunized chickens than in the control and PBS groups (Figure 7E). Although there was no significant difference in the heart lesion score among the experimental groups, slightly lower score was observed in the ΔompAΔompCΔompD OMV-immunized group compared with parental and O78 OMP groups. The relatively decreased clinical scores and liver lesion scores demonstrated that immunization of ΔompAΔompCΔompD OMVs could lower the severity of perihepatitis induced by APEC O78 in the birds (Nagano et al., 2012). Taken together, these data clearly demonstrated the feasibility of using ΔompAΔompCΔompD OMVs as a broad-spectrum vaccine.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.