Acinetobacter baumannii Outer Membrane Vesicles Elicit a Potent Innate Immune Response via Membrane Proteins

Acinetobacter baumannii is increasingly becoming a major nosocomial pathogen. This opportunistic pathogen secretes outer membrane vesicles (OMVs) that interact with host cells. The aim of this study was to investigate the ability of A. baumannii OMVs to elicit a pro-inflammatory response in vitro and the immunopathology in response to A. baumannii OMVs in vivo. OMVs derived from A. baumannii ATCC 19606T induced expression of pro-inflammatory cytokine genes, interleukin (IL)-1β and IL-6, and chemokine genes, IL-8, macrophage inflammatory protein-1α, and monocyte chemoattractant protein-1, in epithelial cells in a dose-dependent manner. Disintegration of OMV membrane with ethylenediaminetetraacetic acid resulted in low expression of pro-inflammatory cytokine genes, as compared with the response to intact OMVs. In addition, proteinase K-treated A. baumannii OMVs did not induce significant increase in expression of pro-inflammatory cytokine genes above the basal level, suggesting that the surface-exposed membrane proteins in intact OMVs are responsible for pro-inflammatory response. Early inflammatory processes, such as vacuolization and detachment of epithelial cells and neutrophilic infiltration, were clearly observed in lungs of mice injected with A. baumannii OMVs. Our data demonstrate that OMVs produced by A. baumannii elicit a potent innate immune response, which may contribute to immunopathology of the infected host.


Introduction
Acinetobacter baumannii is a Gram-negative, lactose non-fermenting aerobic coccobacillus and an important opportunistic pathogen that causes various types of infections, including ventilatorassociated pneumonia, urinary tract infection, skin and wound infections, bacteremia, and meningitis [1]. This microorganism is regarded as a low virulence pathogen, but increasing evidence has highlighted the importance of A. baumannii as a nosocomial pathogen responsible for high morbidity and mortality of infected patients, especially in severely ill patients [1,2]. Clinical significance of A. baumannii has also increased due to its ability to develop antimicrobial resistance to currently available antimicrobial agents, which causes a serious therapeutic problem [2][3][4].
A number of virulence traits of A. baumannii, such as biofilm formation [5,6], adherence and invasion to host cells [7,8], serum resistance [9], and host cell death [10,11], have been characterized, however, much less is known regarding the immune responses to A. baumannii that are critical to disease development. An innate immune response against A. baumannii via sensing of lipopolysaccharide (LPS) through CD14 and Tolllike receptor (TLR) 4 effectively eliminated bacteria from the lungs in a mouse pneumonia model, whereas TLR2 signaling counteracted the robustness of innate immune responses [12,13]. Breij et al [14] recently reported an association of the outcome of A. baumannii-induced pneumonia with anti-inflammatory interleukin (IL)-10 and pro-inflammatory IL-12p40 and IL-23 cytokine levels in a mouse pneumonia model. However, little is known with regard to the interaction of A. baumanniiderived secretory products with host cells leading to the innate immune response.
Gram-negative pathogens secrete outer membrane vesicles (OMVs), which are recognized as delivery vehicles for bacterial effectors to host cells [15][16][17][18][19]. OMVs are spherical nanovesicles with an average diameter of 20 -200 nm and are composed of LPS, proteins, lipids, and DNA or RNA [16,17]. OMVs produced by Gram-negative pathogens transport diverse virulence factors to host cells simultaneously and allow interaction of pathogens with the host without close contact between bacteria and host cells [20]. In addition, OMVs contain adhesins, invasins, toxins, and pathogen-associated molecular patterns (PAMPs) and they can contribute to bacterial pathogenesis and immunopathology in the host. We previously demonstrated that A. baumannii OMVs contain multiple virulence factors, including outer membrane protein A (AbOmpA), proteases, phospholipases, superoxide dismutase, and catalase [21]. Of particular interest, A. baumannii OMVs interact with host cells and then deliver bacterial effectors to host cells via lipid rafts, resulting in cytotoxicity [22]. However, immune response to A. baumannii OMVs has not yet been characterized. The aim of this study was to investigate an innate immune response to A. baumannii OMVs in both in vitro cultured epithelial HEp-2 cells and an in vivo mouse model. We report here that A. baumannii OMVs are potent stimulators of inflammatory response both in vitro and in vivo.

Bacterial Strain
A. baumannii ATCC 19606 T was used for preparation of OMVs and infected cells. AbOmpA-deficient mutant KS37 strain was also used for preparation of OMVs [10]. A. baumannii ATCC 19606 T was provided by Lenie Dijkshoorn (Leiden University Medical Center, The Netherlands) and bacteria were grown in Luria-Bertani (LB) broth.

Cell Culture
Human laryngeal epithelial HEp-2 cells were obtained from Korean Cell Line Bank (Seoul, Korea) and were grown in Dulbecco's modified Eagle medium (HyClone) supplemented with 10% fetal bovine serum (HyClone), 2 mM L -glutamine, 1,000 U/ml penicillin G, and 50 mg/ml streptomycin at 37uC in 5% CO 2 . Confluent cells were harvested and seeded into wells of 96-well plates for the cell viability assay and 6-(4610 5 cells/well) or 12-well (5610 4 cells/well) plates for the cytokine gene assay. HEp-2 cells were treated with OMVs, live or formalin-fixed bacteria, or phosphate-buffered saline (PBS), and incubated until time of assay.

Purification of OMVs
OMVs produced by A. baumannii were prepared as previously described [22,23]. Briefly, A. baumannii ATCC 19606 T was grown in 500 ml of LB broth to reach late log phase at 37uC with shaking. Bacterial cells were removed by centrifugation at 6,000 6 g for 20 min at 4uC. The supernatants were filtered using a QuixStand Benchtop System (GE Healthcare) using a 0.2 mmsized hollow fiber membrane (GE Healthcare) and then concentrated using a QuixStand Benchtop System using a 100 kDa hollow fiber membrane (GE Healthcare). After ultrafiltration of OMVs, the samples were collected by ultracentrifugation at 150,000 6 g for 3 h at 4uC and resuspended in PBS. The protein concentration was determined using the modified BCA assay (Thermo Scientific). The purified OMVs were checked to sterility and stored at 280uC until used.

Treatment of OMVs with Proteinase K and Ethylenediaminetetraacetic Acid (EDTA)
Purified OMVs were treated with 0.1 mg/ml of proteinase K (Fermentas) for 1 h at 37uC for degradation of surface-exposed proteins in the OMVs and 0.1 M EDTA for 1 h at 37uC for disintegration of OMV membrane. OMV samples, including intact OMVs and proteinase K-and EDTA-treated OMVs, were separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and stained with Coomassie brilliant blue R-250 (Bio-Rad).

RNA Extraction and Quantitative Real-time Polymerase Chain Reaction (PCR)
HEp-2 cells were treated with A. baumannii OMVs (1-15 mg/ ml of media) or infected with live or formalin-fixed A. baumannii with multiplicity of infection (MOI) 1-300 for 24 h. Total cellular RNA was harvested using the Qiagen RNeasy kit according to the manufacturer's instructions. The RNA samples were treated with DNase (Qiagen) for removal of contaminating DNA. Harvested RNA was quantitated using a spectrophotometer (Bio-Rad). cDNA was generated by reverse transcription of 1 mg of total RNA using oligo dT primers and M-MLV reverse transcriptase in a total reaction volume of 20 ml (Fermentas). The reaction mixtures were incubated for 1 h at 37uC and the samples were stored at 220uC. For quantitative real-time PCR, primer sequences were designed using Primer Express Software . The amplification specificity was evaluated using melting curve analysis. Gene expression was normalized to GAPDH mRNA levels in each sample and fold change was determined using the DDCt method [24].

TEM Analysis
The purified OMV samples were applied to copper grids and stained with 2% uranyl acetate. The samples were visualized on a transmission electron microscope (Hitachi H-7500, Hitachi, Japan) operating at 120 kV.

Cell Viability
The cytotoxicity of HEp-2 cells treated with A. baumannii OMVs was measured using the Premix WST1 cell proliferation assay system (Takara, Japan) [10]. The cells were seeded at a concentration of 2.0 6 10 5 /ml in 96-well microplates. After treatment with different concentrations of OMVs, cellular cytotoxicity was measured at 450 nm 3 h after treatment with WST1.

Mouse Inflammation Model
Female Balb/c mice (eight weeks old) were maintained under specific pathogen free conditions. For induction of inflammatory response to A. baumannii OMVs in the skin, mice received intradermal injection of OMVs (200 mg of OMVs suspended in 100 ml of PBS). For assessment of inflammatory response to A. baumannii OMVs in the lungs, mice were anesthetized with Avertin (Sigma) and OMVs (200 mg of OMVs suspended in 100 ml of PBS) were administered intratracheally [11]. The control mice were injected with 100 ml of PBS (pH 7.4) in both experiments. Mice were sacrificed 24 h after OMV challenge. Skin and lung tissues were stained with hematoxilin and eosin (H & E). The animal experimental procedures were approved by the Animal Care Committee of Kyungpook National University (KNU2012-5).

A. baumannii OMVs Elicit a Pro-inflammatory Response in Epithelial Cells
OMVs were purified from the culture supernatant of A. baumannii ATCC 19606 T and TEM analysis was performed. A. baumannii OMVs were spherical bilayered nanovesicles and maintained membrane integrity (Fig. 1). Because OMVs from A. baumannii ATCC 19606 T induced cytotoxicity in macrophage U937 cells [22], the cytotoxic activity of the purified OMVs was determined in HEp-2 cells. Cultured HEp-2 cells were treated with various concentrations (1-50 mg/ml) of A. baumannii OMVs for 24 h and cellular damage was assessed using inverted microscopy and WST1 cell proliferation assay. Cytotoxicity of HEp-2 cells was not observed in response to #15 mg/ml of A. baumannii OMVs, however, $20 mg/ml of OMVs induced cytotoxicity such as cellular shrinkage, rounding of cells, and cell detachment from the bottom. Results of the WST1 assay showed that $20 mg/ml of A. baumannii OMVs also induced cytotoxicity of HEp-2 cells (data not shown). Next, in order to determine whether A. baumannii OMVs could trigger a pro-inflammatory response, HEp-2 cells were treated with sublethal doses of A. baumannii OMVs (1-15 mg/ml) and quantitative real-time PCR was performed for analysis of expression of pro-inflammatory cytokine genes, including IL-1b, IL-6, IL-8, MIP-1a, and MCP-1. A. baumannii OMVs stimulated significant transcription of all proinflammatory cytokine genes tested (Fig 2). Expression of proinflammatory cytokine genes, except IL-1b, was clearly dosedependent and showed a sharp increase in response to treatment with 10 mg/ml of A. baumannii OMVs.

Comparison of Pro-inflammatory Cytokine Response in HEp-2 cells Treated with A. baumannii and its Derived OMVs
In order to determine the pro-inflammatory response to A. baumannii, HEp-2 cells were treated with live or formalin-fixed A.
baumannii with MOI 1-300 for 24 h and expression of IL-6 gene was measured. Expression of IL-6 gene was not increased in either live A. baumannii with MOI 10 (data not shown) or formalin-fixed bacteria with MOI up to 300 (Fig 3A). Treatment with live A. baumannii with MOI 100 and 300 resulted in stimulation of IL-6 gene expression in HEp-2 cells. To compare expression of proinflammatory cytokine genes in host cells respond to A. baumannii infection and OMV treatment, HEp-2 cells were infected with A. baumannii with MOI 300 or treated with 5 or 15 mg/ml of OMVs. Expression level of pro-inflammatory cytokine genes in HEp-2 cells infected with live bacteria was similar to that of cells treated with 5 mg/ml of OMVs (Fig 3B). However, 15 mg/ml of A. baumannii OMVs elicited profoundly greater pro-inflammatory cytokine response than that observed in response to live A. baumannii with MOI 300. Surface-exposed Membrane Proteins in A. baumannii OMVs are Responsible for Pro-Inflammatory Cytokine Response To determine which OMV components are responsible for proinflammatory cytokine response, HEp-2 cells were treated with proteinase K-treated A. baumannii OMVs and gene expression of pro-inflammatory cytokines was measured. Treatment with proteinase K resulted in alteration of the protein profile of OMVs ( Fig 4A). All proteins in the OMVs were not degraded, but many high molecular weight bands disappeared, suggesting degradation of membrane proteins and protection of luminal proteins. Upregulation of pro-inflammatory cytokine genes was not observed in response to proteinase K-treated A. baumannii OMVs (Fig. 4B). Next, in order to determine whether lysed OMVs stimulated proinflammatory response like that of intact OMVs, A. baumannii OMVs were pre-treated with EDTA in order to disintegrate the OMV membrane, resulting in lysis of OMVs, and HEp-2 cells were treated with lysed OMVs for 24 h. Treatment with EDTA did not result in alteration of the protein profile of OMVs (Fig 4A). The lysed A. baumannii OMVs induced up-regulation of IL-1b, IL-6, IL-8, and MIP-1a genes, but not induce MCP-1 gene (Fig. 4B). Intact A. baumannii OMVs elicited greater pro-inflammatory cytokine gene expression than the response to the lysed A. baumannii OMVs (Fig 4B). To determine whether AbOmpA in A. baumannii OMVs was responsible for pro-inflammatory cytokine response, HEp-2 cells were treated with OMVs purified from A. baumannii ATCC 19606 T and its isogenic AbOmpA-deficient mutant KS37 strain for 24 h. Both OMVs from wild-type and AbOmpA mutant strains up-regulated gene expression of proinflammatory cytokines; however, there was no significant difference in expression of pro-inflammatory cytokine genes between OMVs from wild-type and AbOmpA mutant strains (Fig 5).

A. baumannii OMVs Induce Inflammatory Response in vivo
In order to determine whether A. baumannii OMVs could induce an inflammatory response in vivo, mice received intradermal injection of A. baumannii OMVs in the back and the inflammatory response was determined. As shown in Fig 6A, massive neutrophilic infiltration was observed in the skin. Because the respiratory tract was the most common site of A. baumannii infection, we determined the ability of A. baumannii OMVs to elicit an inflammatory response in the lungs. A. baumannii OMVs were administered intratracheally and lungs were removed from mice. A. baumannii OMVs also elicited pro-inflammatory cytokine genes, including IL-1b, IL-6, IL-8, MIP-1a, and MCP-1, in the lungs    (Fig. 6B). In addition, early inflammatory processes, including congestion, hemorrhage, vacuolization and detachment of bronchiolar epithelial cells, and neutrophilic infiltration, were clearly observed in lungs of mice injected with A. baumannii OMVs (Fig. 6C).

Discussion
Several reports have described the production, proteomic analysis, functional roles in host cells, and vaccine trial of A. baumannii OMVs [21,22,25,26], however, little is known about the innate immune response to A. baumannii OMVs associated with immunopathology. Thus, we investigated the innate immune response to A. baumannii OMVs both in vitro and in vivo. Our data demonstrate that A. baumannii OMVs elicit a pro-inflammatory response via surface-exposed membrane proteins in vitro and trigger a potent inflammatory response in vivo.
We have previously shown that A. baumannii ATCC 19606 T and ATCC 17978 produce OMVs during both in vitro culture and in vivo mouse infection [22]. Koning et al. [27] recently reported that A. baumannii 19606 T produces morphologically different types of OMVs during the various stages of bacterial culture. Regularshaped and small-sized vesicles are produced by log phase bacteria, whereas deformed and large-sized vesicles are produced by stationary phase bacteria. OMVs derived from different stages of bacterial culture may show compositional difference. In this study, purified OMVs from A. baumannii ATCC 19606 T exhibited a regular shape and sized with 40-70 nm (Fig 1), suggesting that A.
baumannii OMVs obtained in this study originated from log phase bacteria. Because OMVs from A. baumannii 19606 T were proven to contain cytotoxic proteins such as AbOmpA and the interaction of A. baumannii OMVs with host cells induced host cell damage [22], we determined the cytotoxicity of A. baumannii OMVs in HEp-2  cells. In the previous study, $50 mg/ml of A. baumannii OMVs induced cytotoxicity in macrophage U937 cells. However, in this study, $20 mg/ml of A. baumannii OMVs induced cytotoxicity of HEp-2 cells. Discrepancy in amounts of A. baumannii OMVs for induction of cell death may be due to the different bacterial culture stages for OMV purification, which may result in compositional difference of the purified OMVs.
An innate immune response is accompanied by bacterial colonization and infection. OMVs produced by Gram-negative bacteria contain various PAMPs, such as LPS, outer membrane porins, flagellins, peptidoglycans, and DNA [16,19]. These immune activating ligands in OMVs interact with and are internalized by neighboring epithelial cells and immune cells [18,19,28]. The potency of OMVs in triggering an innate immune response was evident by several pathogens, such as Pseudomonas aeruginosa [20,29] and Salmonella enterica serovar Typhimurium [30]. OMVs stimulate expression of major histocompatibility complex, production of pro-inflammatory cytokines and chemokines, and synthesis of nitric oxide in professional antigen presenting cells and epithelial cells, leading to inflammatory response in the hosts. OMVs from A. baumannii ATCC 19606 T have also been reported to carry a variety of PAMPs, such as porins, other outer membrane proteins, and LPS, like OMVs from other Gramnegative bacteria [22,25]. In this study, the ability of A. baumannii OMVs to elicit a pro-inflammatory response was determined. Our data clearly demonstrate that A. baumannii OMVs are potent stimulators of pro-inflammatory cytokines, including IL-1b, IL-6, IL-8, MIP-1a, and MCP-1, in epithelial cells (Fig 2). We determined gene expression of pro-inflammatory cytokines in response to live A. baumannii infection. Expression level of proinflammatory cytokine genes in HEp-2 cells infected with A. baumannii with MOI 300 was comparable to that of cells treated with 5 mg/ml of OMVs (Fig 3B). Although OMV concentrations for treatment of cells are a relatively high to obtain it in vitro culture condition, OMV production is associated with bacteria stress condition and is increased under harsh conditions [25,31]. These results suggest that OMVs produced by A. baumannii can induce pro-inflammatory response during in vivo infection.
OMVs derived from Gram-negative pathogens play a role as protective transport vehicles, delivering bacterial effector molecules such as toxins, enzymes, and DNA to host cells [19]. A. baumannii OMVs bind to the cytoplasmic membrane of host cells and deliver bacterial effectors to host cells [22]. In this study, intact OMVs elicited a pro-inflammatory response in a dose-dependent manner, whereas gene expression of pro-inflammatory cytokines in response to lysed OMVs did not reach that of intact OMVs (Fig 4B). This result suggests that OMV-mediated delivery of bacterial effectors is critical to induction of pro-inflammatory response. In addition, pro-inflammatory response to proteinase K-treated OMVs did not induce expression of cytokine genes, although the luminal proteins were conserved in the OMVs, confirming the role of surface-exposed membrane proteins in triggering a pro-inflammatory response. AbOmpA, an abundant protein in A. baumannii OMVs, is known as a specific virulence factor [10]. AbOmpA contributes directly or indirectly to multiple aspects of A. baumannii pathogenesis through biofilm formation [32], serum resistance [9], host cell cytotoxicity [10,11], and adherence and invasion of host cells [8]. These results may suggest that AbOmpA plays a role in pro-inflammatory response. However, our results showed that AbOmpA packaged in A. baumannii OMVs did not exert on the expression of proinflammatory cytokine genes in HEp-2 cells (Fig 5). Furthermore, recombinant AbOmpA (rAbOmpA) did not induce any proinflammatory cytokine in HEp-2 cells, although rAbOmpA induced gene expression of TLR 2 and inducible nitric oxide synthase [33]. Future studies will focus on determining which membrane proteins are critical to the observed pro-inflammatory responses.
We have previously shown that A. baumannii induced an inflammatory response such as recruitment of inflammatory cells and exudates in the lungs of neutropenic mice. Neutrophil recruitment and activation are important for host defense to systemic A. baumannii infection [34]. Our data showed that A. baumannii OMVs recruited neutrophils in skin of mice. After confirming inflammatory properties of A. baumannii OMVs, we determined pulmonary inflammation in mice administered with A. baumannii OMVs. A. baumannii OMVs induced early inflammatory response in the lungs, but inflammatory response in the lungs was weak, as compared to skin lesions injected with A. baumannii OMVs. Our data highlight the potential inflammatory consequences of OMVs produced by A. baumannii during colonization or infection. Future studies should be conducted in order to determine whether the innate immune response to A. baumannii OMVs can stimulate clearance of bacteria or enhance pathogenic potential of bacteria.
In conclusion, the data presented here demonstrate that A. baumannii OMVs are potent stimulators of innate immune response and that membrane proteins in OMVs are critical for induction of an innate immune response. Epithelial response to A. baumannii OMVs may explain in part the innate immune response during colonization or early infection of A. baumannii.