The animal experimental procedures were approved by the Ethics Committee of Animal Care and Welfare, Institute of Medical Biology, CAMS (Permit Number: SYXK (dian) 2010–0007), in accordance with the animal ethics guidelines of the Chinese National Health and Medical Research Council (NHMRC) and the Office of Laboratory Animal Management of Yunnan Province, China. All efforts were made to minimize animal suffering.

All participants submitted a signed informed consent form to participate in the study. The protocol complied with the Helsinki Declaration and was approved by the Institutional Review Boards of the Institute of Medical Biology, Chinese Academy of Medical Sciences & Peking Union Medical College.

A. baumannii strains 1 to 7 (Ab1 to Ab7) were isolated from intensive care unit (ICU) patients hospitalized at the affiliated hospitals of Kunming Medical University (Kunming, China). Ab1, Ab4, Ab5, Ab6, and Ab7 were isolated from sputum, and Ab2 and Ab3 were isolated from peritoneal fluid drainage and blood, respectively. All strains are pan-drug resistant against a panel of 17 known antibiotics [20]. These strains were confirmed through DNA sequencing analysis of the intergenic spacer (ITS) between 16 S and 23 S rRNA genes based on the methods described by Chang et al. [21] The strains were grown in Luria-Bertani (LB) medium under the selection pressure of ampicillin and kanamycin antibiotics. Female ICR mice (6–8 weeks of age) were raised and maintained in the Central Animal Care Services of the Institute of Medical Biology, Chinese Academy of Medical Sciences (CAMS) & Peking Union Medical College (PUMC) (Kunming, China) under specific pathogen-free (SPF) conditions.

A single Ab1 colony was inoculated and cultured in LB medium at 37°C, overnight. The bacterial culture was centrifuged at 14,000×g for 15 min, and the supernatant was filtered through a 0.22-µm membrane (Millipore, Merck, Darmstadt, Germany). The filtered fraction was concentrated by ultra-filtration with a 500,000 nominal molecular weight cut off (500,000 NMWC) column (QuixStand Benchtop System, GE Healthcare, Piscataway, NJ, USA). The concentrate was ultra-centrifuged at 200,000×g for 2 h at 4°C. The vesicle pellets were resuspended in phosphate buffered saline (PBS; 0.02 mol/L phosphate buffer with 0.15 mol/L NaCl, pH 7.4) and then filtered through a 0.22-µm membrane [10]. The absence of viable bacteria in the OMV preparations was determined by spreading aliquots on agar plates to test for bacterial growth. OMVs were quantified using Bradford reagent according to the manufacturer's instructions.

The OMV sample was fixed with 2.5% cold glutaraldehyde in 0.2 M sodium cacodylate buffer (pH 7.4) for 2 h at 4°C and postfixed with 1% osmium tetroxide in 0.1 M sodium cacodylate buffer (pH 7.4) for 1 h at 4°C; then, the sample was observed and imaged using a transmission electron microscope (Hitachi, Tokyo, Japan) at 80 kV.

A preliminary study was performed using different vaccine doses (5, 25, 50 and 100 µg/ml) to immunize the mice, and the results showed that 50 µg/ml is most likely the optimal vaccination dose (data not shown). In this study, the vaccine was prepared by mixing the OMVs (isolated from Ab1) with an equal volume of Alum adjuvant (Thermo Scientific, Rockford, IL, USA) to a final concentration of 50 µg/mL OMVs and 2 mg/mL Alum. Then, each mouse was immunized intramuscularly with 200 µL of the vaccine three times, on days 0, 15, and 36. For studies with active immunization, an additional group of mice were injected with a mixture of PBS and adjuvant to serve as the control. For all active immunization studies, the challenge with homologous strains of A. baumannii occurred 11 days after the last immunization with OMVs (day 47). In addition, for the passive immunization studies, either 100 µL of naive serum or antiserum collected on day 52 from the vaccinated mice was administered intravenously 1 h prior to the bacterial challenge.

The outer membrane complex (OMCs) and whole cell (WC) proteins were isolated from Ab1 as described previously [7], [8]. OMVs, OMCs, and WC samples were quantified using the Bradford reagent, loaded and separated on a 15% SDS-PAGE, and transferred to a polyvinylidene difluoride (PVDF) membrane (Pall Corporation, East Hills, NY, USA). The membrane was blocked with 5% milk powder and then incubated with sera collected from experimental mice at a dilution of 1∶3000. The sera collected on day 52 from the OMV-immunized or adjuvant only-treated mice were pooled for the assays. Horseradish peroxidase (HRP)-conjugated rabbit anti-mouse IgG (Life Technologies, Gaithersburg, MD, USA) was used as a secondary antibody. The blot was developed with enhanced chemiluminescence (ECL) substrate according to the manufacturer's instructions (Pierce, Rockford, IL, USA). The WC samples from seven separately collected clinical isolates were also analyzed with the antisera.

A sepsis model was established as described previously, with minor modifications [12]. Briefly, Ab1 under exponential growth conditions was collected and adjusted to the designated concentrations with PBS according to optical density 600 (OD₆₀₀) values based on a previously determined relation curve between optical density and colony forming units (CFUs). Inocula were prepared by mixing the bacterial suspension with 10% porcine mucin (w/v; Sigma-Aldrich, St Louis, MO, USA). Mice were injected intraperitoneally with 0.5 mL of the inocula, and the actual numbers of CFU reflecting injected bacterial loads were determined by plating sequential dilutions of the inocula on LB plates.

The pneumonia model was established as described previously, with minor modifications [22]. In brief, the mice were anesthetized with inhalation isoflurane (Baxter Healthcare Corporation, Glendale, CA, USA), and 50 µL of live Ab1 (10⁸ CFU/mL) in PBS was administered intranasally (i.n.). Control mice were administered PBS.

At the conclusion of the experiment, the mice were anesthetized and blood samples were obtained by cardiac puncture. Bronchoalveolar lavage fluid (BALF) was collected with 1 mL of ice cold PBS, flushing slowly three times. After centrifugation, the supernatants were kept for antibody and cytokine measurements. Cells were resuspended with PBS and counted using a hemocytometer, and differential analysis of inflammatory cells was performed on cytospin preparations (Cytospin, Shandon, Runcorn, UK) stained with HEMA 3 (Fisher Diagnostics, Pittsburg, PA, USA). The left lung was removed and immersed immediately in 10% neutral buffered formalin for histological analysis. The right lung and spleen were removed aseptically from the mice, weighed, added to 1 mL of ice-cold PBS, and homogenized using a Tissue Tearor (Biospec Products, Racine, WI, USA) to determine the bacterial burden.

The concentrations of cytokines (IL-1β, IL-13, IL-33, IFNγ, and TNFα) in BALF were detected using ELISA, which was performed using paired capture and biotinylated detection antibodies purchased from eBioscience, Inc. (San Diego, CA, USA) according to the supplier's instructions.

The bacterial burden was determined in BALF and lung and spleen homogenates by plating 10-fold dilutions on LB plates with double antibiotic resistance and incubating at 37°C overnight. The number of colony forming units was counted, and the results are expressed as the numbers of CFU/g tissue or the numbers of CFU/mL BALF.

Lung tissues were embedded in paraffin and serially sectioned (5 µm) sagittally. Specimens were stained with hematoxylin and eosin to examine peribronchial and perivascular accumulation of inflammatory cells.

At the designated time points during the experiment, blood samples were collected by sphenous sampling from the limb. At the end of experiment, the mice were anaesthetized, and blood samples were obtained by cardiac puncture. Measurements of OMVs-specific IgA and IgG antibodies were performed successively by coating microplates with previously prepared OMVs (2 µg/mL), adding serum or BALF samples, incubating with alkaline phosphatase-conjugated goat anti-mouse IgA or IgG (Life Technologies, Gaithersburg, MD, USA) and adding substrate tablets (Sigma-Aldrich, St Louis, MO, USA) to develop the reaction. The serum levels of OMV-specific IgG₁ and IgG_2a were also measured. Briefly, after sample incubation, either biotinylated goat anti-mouse IgG₁ or IgG_2a (Life Technologies, Gaithersburg, MD, USA) was used as a secondary antibody, and the reaction was developed using an avidin–phosphatase system.

The opsonophagocytic killing assay was performed as previously described with modifications [12]. Murine macrophage RAW264.7 cells were cultured in DMEM/HIGH Glucose (HyClone, Logan, UT, USA) with 10% fetal bovine serum (FBS). The cells were activated with a 3-day exposures to 100 nM phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich, St Louis, MO, USA), harvested by scraping with cell scrapers (Thermo Scientific, Rockford, IL, USA), and added to the microplates at a 40∶1 ratio of macrophages (∼2×106 cells/mL) to Ab 1 to Ab 7 (∼5×104 CFU/mL). The pooled OMV antisera were diluted sequentially at 1∶1, 1∶10, 1∶100, 1∶1000, and 1∶10000 with culture medium. In addition to a pooled naive serum, a pooled irrelevant antigen-immunized serum was used as a control to confirm that there were no differences between the normal and the OMV-irrelevant immunization sera with respect to the killing effect. The irrelevant antigen is a recombinant fusion protein that consists of glutathione S-transferase (GST) and enhanced green florescence protein (EGFP), which was previously prepared in our lab using gene recombinant technology. The sera were subjected either to no heat or to heat treatment at 56°C for 30 min to inactivate the complement components. The test sera, bacteria, and macrophages were mixed and then incubated for 1 hour with gentle shaking, and the mixtures were diluted and plated for bacterial counting. The number of non-phagocytosed bacteria was determined, which reflects the number of bacteria phagocytosed via another pathway.

All statistical analyses were performed using Prism 5.0 (GraphPad Prism, GraphPad Software, Inc., San Diego, CA, USA). Survival was compared using the non-parametric log-rank test. The bacterial burdens were compared with an unpaired Student's t test. One-way ANOVA with Tukey's multiple comparison test was applied for the analyses of cytokine, IgG₁, and IgG_2a levels. Differences were considered significant if the p value was <0.05. All graphed values represent the mean, and the error bars represent the standard error.

OMVs were prepared from the supernatant of bacterial culture after centrifugation, observed for morphology, and imaged using a transmission electron microscope (Figure 1A). The size of the vesicles ranged from 20 to 100 nm, which was similar to that reported previously [10].

Each mouse was immunized intramuscularly three times on days 0, 15, and 36 with OMVs prepared from strain Ab1, and the challenge with the homologue strain A. baumannii occurred 11 days after the last immunization. To demonstrate the dynamic features of the IgG responses after OMV immunization (n = 6), mouse sera were collected on days 0, 14, 21, and 52, and ELISA was used to measure the levels of OMV-specific IgGs (Figure 1B). IgG levels were only slightly elevated 14 days after the primary immunization. However, the levels increased after the booster immunizations, and the titers reached 16,000 on day 14 (7 days after the first booster) and 64,000 on day 36 (21 days after the second booster). The control mice did not show any OMV-specific IgG responses.

Proteomic studies on OMVs from A. baumannii have indicated that OMVs contain proteins from the outer membrane, the inner membrane, the periplasm, and the cytoplasm [23], [24], whereas a recent study based on cryo-electron tomography showed that OMVs at the distal ends of early log-phase bacteria did not contain cytoplasmic proteins but only outer membrane and periplasmic proteins. However, inner-outer membrane vesicles are produced by A. baumannii in the stationary phase, and they also contain cytoplasmic proteins [25]. To establish the dominant antigen components in the OMV preparation in the current study and determine their possible cellular distribution, OMVs, OMCs, and WC proteins were used for western blot analyses with the antisera collected from vaccinated mice or adjuvant only-treated mice on day 52 (Figure 1C). There were no reactive bands when the sera from adjuvant only-treated mice were used (Figure 1C, right panel). Antisera from the vaccinated mice showed that there were 8 major specifically reactive bands observed in the OMVs (Figure 1C, left panel). Bands 3, 4, 5, and 7 were observed in all samples, indicating that they are membrane proteins because they were clearly detected in both OMC and OMV samples. Band 3, with a molecular weight of approximately 38 kDa, has been reported to be OmpA and used as a vaccine candidate [12], [14], and it was the most dominant component in all three samples. Bands 1, 2, and 8 were only observed in the OMVs, strongly indicating that they were highly immunogenic, relatively enriched in OMVs, and most likely periplasmic proteins, considering that no similar bands appeared in OMCs. The other bands appeared only in WC or/and OMCs and were more likely to be non-membrane proteins because of their low levels in OMVs. In summary, the Western blot results showed that vaccination with OMVs elicited antibodies against multiple bacterial antigens, the majority of which are most likely located on the bacterial outer membrane.

First, the effectiveness of the vaccine was confirmed in a sepsis model. To establish the sepsis model, mice (n = 12/group) were injected intraperitoneally with 0.5 mL of different doses of A. baumannii: 1×10³ CFU/mL; 1×10⁴ CFU/mL; 1×10⁵ CFU/mL; 1×10⁶ CFU/mL; and 1×10⁷ CFU/mL. The survival rate was calculated continuously over the next 7 days (Figure 2A). Correspondingly, the survival rates in the experimental groups were 100, 75, 41.67, 16.67, and 0%, respectively, and all mouse deaths occurred within 24 h. The bacterial dose of 5.5×10⁵ CFU/mouse was used as a challenge dose 11 days after the last immunization to assess the efficacy of the active and passive immunization approaches. The results showed that the mice vaccinated with OMVs (n = 15) had a higher survival rate than the adjuvant control mice (n = 14), with 73.3% survival versus 7.14% (p<0.001, Figure 2B). In addition, the mice that were administered passive treatment with antisera were completely protected from A. baumannii infection, whereas only ∼20% of the mice survived in the control group after being administered naive serum (p<0.01, non-parametric log-rank test, n = 12, Figure 2C). As a brief summary, both active and passive immunization with OMVs protected the mice from a challenge with the A. baumannii strain from which the OMVs were derived.

Mice were intranasally administered 50 µL of live Ab1 (10⁸ CFU/mL). No deaths occurred in the following 2 weeks. The kinetics of the bacterial burden in major tissues were investigated to determine bacterial clearance to establish an optimal time point for other measurements. The bacterial burden in BALF, lung, and spleen was detected continuously in the mice challenged with the A. baumannii homologs (n = 12) and in the control mice administered PBS instead of A. baumannii (n = 12; Figure 3A). The bacterial loads decreased from day 3 and persisted for at least 5 days in the BALF and 9 days in the lung and spleen. To determine the protective effects of active and passive immunization strategies, the bacterial burden was analyzed on days 1, 3, and 5 after the A. baumannii challenge. With regard to active immunization, the results showed that the OMV vaccine (n = 12) significantly reduced the bacterial loads in the BALF, lung, and spleen compared with the adjuvant controls (n = 12). The mean of the CFU values on days 1, 3, and 5 were as follows: BALF − 5815, 7460, and 1730.6, respectively; lung − 1,346,000, 1,485,000, and 520,000, respectively; and spleen − 303,000, 471,000, and 67,900, respectively. The protective effect lasted for at least 5 days (Figure 3B). In addition, the mice that were administered the antisera one hour prior to the bacterial challenge showed a significantly reduced bacterial burden on day 1. However, there was a trend toward a reduced bacterial burden on day 3, and no difference was observed on day 5 between the immunized mice (n = 12) and the control mice (with naive serum, n = 12). The CFU values on day 1 were 5603, 1,260,000, and 462,000 in the BALF, lung, and spleen, respectively. A significant reduction in the bacterial burden by passive immunization was observed only on day 1 (Figure 3C).

In the pneumonia model (n = 12/group), the accumulation of immune cells dominated by neutrophils was significantly increased in the BALF after bacterial challenge, which continued until day 14 (Figure 4A). Immunization with OMVs, including both active and passive approaches, significantly decreased inflammatory cell accumulation in the BALF, which corresponded to a reduction in the bacterial burden. Active immunization presented a significant reduction in inflammatory cell accumulation on all the investigated days after the A. baumannii challenge (n = 12/group; Figure 4B). However, passive immunization showed a significant difference between the vaccinated mice (n = 12) and the control mice (with naive serum, n = 12) on days 1 and 3 only. There was only a trend toward reduction on day 5 (Figure 4C). We also examined the histopathological changes in the lungs. The A. baumannii challenge caused significant bronchial, peribronchial, and perivascular infiltration of inflammatory cells. Both active and passive immunization significantly reduced inflammatory cell infiltration in the lungs of the immunized mice compared with the control mice (3 days after the A. baumannii challenge; Figure 4 D).

These results were consistent with the significant decrease in the inflammatory cell accumulation in the BALF, indicating that pulmonary inflammation caused by the A. baumannii infection was significantly suppressed by the immunization with OMVs.

The intranasal challenge with A. baumannii caused a cytokine storm in the BALF. The levels of IL-13, IL-33, IFNγ, TNFα, and IL-1β were measured on days 1, 2, 3, 7, and 14 (n = 12/group). All cytokines increased dramatically in the infected mice compared with the controls on all investigated days, except for day 14, when the cytokine levels returned to baseline. The increase in IFNγ, TNFα, and IL-1β in this context has been previously reported [7]. Interestingly, the Th2 cytokines IL-13 and IL-33 were also found to be elevated. Cytokine levels decreased gradually from day 3, reaching baseline levels on day 14 (Figure 5A). Both active and passive immunization (n = 12/group) significantly lowered the levels of cytokines in the vaccinated mice compared with the control mice 3 days after the A. baumannii challenge (Figure 5B&C). Corresponding to the bacterial burden and inflammatory cell infiltration, the cytokine accumulation in BALF was suppressed more significantly by active immunization than by passive immunization between the adjuvant control and OMV vaccine.

The levels of antigen-specific IgA and IgG were measured in the serum and BALF samples using ELISA (n = 12/group). The samples were collected on day 52 (day 5 after the challenge), which was the end of this experiment. The results showed that OMV-specific IgG was clearly detectable in BALF after passive intravenous immunization with the antisera from vaccinated mice (Figure 6A), and active immunization with OMVs showed higher IgG levels compared with passive immunization (Figure 6B). Interestingly, following the intranasal A. baumannii challenge, high levels of OMV-specific IgAs were dramatically induced in both the serum and BALF in those mice that were intramuscularly immunized with the OMVs (Figure 6C&D). The control mice had no detectable OMVs-specific IgA or IgG antibodies.

To determine the Th1 or Th2 response profile to the OMV immunization and A. baumannii infection, the OMV-specific IgG₁ and IgG_2a levels were determined using ELISA. A 1∶500 dilution for the serum samples was used for both the IgG₁ and IgG_2a measurements. The results showed that the OMV immunization elicited strong IgG_2a responses in addition to the induction of IgG₁, and subsequent pulmonary infection produced a response profile of the antibody subtypes similar to that induced by the OMV immunization (Figure 6E&F). The IgG₁/IgG_2a ratios indicated a balanced response of the IgG subtypes evoked by either OMV immunization or A. baumannii infection (Figure 6G).

The pooled antisera mediated a distinct phagocytic response to the A. baumannii homologs in RAW264.7 macrophages in a dose-dependent manner compared with the controls (Figure 7A). Complement components were also involved in the opsonic phagocytosis because the complement-inactivated serum showed a reduced killing effect against bacteria (Figure 7B). Two control sera were used in these experiments, and the results showed that there were no differences in the plating CFUs between control 1 (naive serum) and control 2 (irrelevant antigen GST/GFP-immunized serum with Alum as adjuvant). The results presented here are from one of three separate experiments that showed highly similar results.

To investigate whether the antisera had the potential to protect against non-homologue strain challenges, seven clonally distinct clinical isolates were used for the killing test. The whole cell proteins of the seven isolates were analyzed with SDS-PAGE, and Western blot assays were performed with the antisera from the vaccinated mice (Figure 7C). The results showed that the antisera reacted to the whole cell preparations of the seven isolates in almost identical patterns, with similar specifically reactive bands. The subsequent opsonophagocytic assay showed that the pooled antisera from the homolog strains for vaccine preparation remarkably reduced the survival of six of the seven isolates, with no apparent effect on one isolate (Figure 7D). The results presented here are from one of three separate experiments that showed highly similar results.