A Multi-epitope Peptide rOmp22 Encapsulated in Chitosan-PLGA Nanoparticles as Vaccine Candidate Against Acinetobacter baumannii Infection

Background: Development of vaccine is a promising and cost-effective strategy to prevent emerging multi-drug resistant (MDR) Acinetobacter baumannii infections. The purpose of this study was to prepare a multi-epitope peptide nanovaccine and evaluate its immunogenicity and protective effect in BALB/c mice. Results: The B-cell and T-cell epitopes of Omp22 from A. baumannii were predicted using bioinformatics method and identied by immunological experiments. Three dominant B-cell epitopes and two T-cell epitopes were linked in series and chemically synthesized to generate multi-epitope peptide rOmp22. Then, rOmp22 was encapsulated by chitosan (CS) and polylactic acid glycolic acid (PLGA) to prepare CS-PLGA-rOmp22 nanoparticles (NPs). CS-PLGA-rOmp22 NPs were small (mean size of 272.83 nm) with apparently spherical structural, positively charged (4.39 mV) and exhibited nontoxicity to A549 cells. We achieved a high encapsulation eciency (54.94%) and a continuous slow release pattern. Compared with non-encapsulated rOmp22, CS-PLGA-rOmp22 induced more rOmp22-specic IgG in serum and IFN-γ in splenocyte supernatant. Vaccination with CS-PLGA-rOmp22 decreased lung injury, suppressed bacterial burdens in the lung and blood, provided potent protection (57.14%-83.3%) against acute lethal intratracheal A. baumannii challenge in BALB/c mice. Conclusions: CS-PLGA-rOmp22 NPs could elicit specic IgG antibody, Th1 cellular immunity and protection against acute lethal intratracheal A. baumannii challenge. Our results indicate this nanovaccine is a disirable candidate to prevent A. baumannii infection.


Background
Acinetobacter baumannii is an opportunistic pathogen that predominantly causes pneumonia, bacteremia, meningitis, and urinary tract infections in patients. A. baumannii has recently emerged as one of the most important health problems due to its propensity to acquire multi-drug, extensive drug and even pan-drug resistance phenotypes at previously unforeseen rates [1][2][3]. The main antibiotic resistant mechanisms include alteration in the target sites, failure in the degradation-speci c enzymes, perfusion defects and modi cation in multidrug effusion pumps [1,4]. Besides, A. baumannii has a number of potential virulence factors, such as siderophore-mediated iron-acquistion system and bio lm formation, which could possibly affect clinical outcomes [5]. Global emergence of multi-drug resistant (MDR) and pan-drug resistant (PDR) A. baumannii have resulted in signi cantly increased mortality rates with limited or no options for therapeutic interventions [6,7].
Vaccination strategies are emerging as a viable option to prevent or treat MDR or PDR infections, but there is still no licensed vaccine against A. baumannii. Conventional vaccines developed from live attenuated or inactivated whole cells could induce strong humoral and cellular immunity, however, the clinical applications of such vaccines have been limited due to their complex compositions and potential safety concerns [8][9][10]. During the last decades, the research of vaccines against A. baumannii has primarily focused on various forms of recombinant antigens, including bio lm-associated protein Bap [11], auto-transporter (Ata) [12], outer membrane protein A (OmpA) [13], outer membrane protein assembly factor (BamA) [14], Poly-N-acetyl-β-(1-6)-glucosamine (PNAG) [15], and outer membrane protein 22 (Omp22) [16]. Animal studies showed that some single recombinant protein based vaccines provided only weak protection against A. baumannii infection or poor cross-protection against certain strains [17]. In addition, the formulation of such vaccines often leads to reactogenic and/or allergenic responses that are often not desired [18]. Thus, determining an antigen that has high immunogenicity and avoids the virulence of structural proteins is the key to prepare an A. baumannii vaccine.
Recently, the design of epitope-driven or peptide-based vaccines is becoming more attractive, because they are comparatively easier to produce and construct, lack any infectious potential and offer chemical stability [18][19][20]. There are many multi-epitope vaccine design studies involving various bacteria like Klebsiella pneumoniae [21], Shigella sonnei [19] and Meningitis-inducing Bacteria (Streptococcus pneumoniae, Neisseria meningitidis, and Haemophilus in uenzae Type b) [22]. Ren et al [23] rstly designed a multi-epitope assembly peptide (MEP) of A. baumannii and evaluated its immunogenicity and protective immunity in BALB/c mice. The results of that study indicated that the rMEP is a promising vaccine candidate for the control of infections caused by A. baumannii. However, to be optimally effective, peptide-based vaccines need to be administered with adjuvants. Many currently available adjuvants are toxic, not biodegradable and invariably invoke adverse reactions, including allergic responses and excessive in ammation. A nontoxic, biodegradable, biocompatible vaccine delivery system is urgently needed.
The nanotechnology-based approach is considered to be one of the most advantageous methods for the development of peptide-based vaccine [24]. Poly (lactic-co-glycolic) acid (PLGA) is a synthetic copolymer that has been approved by United States Food and Drug Administration (FDA) and European Medicine Agency (EMA) for various medical and pharmaceutical applications in humans [25]. PLGA NPs have been shown to be e cient for antigen delivery because of their effectiveness in enhancing immune responses, controlled release, high encapsulation e ciency and tissue bio-distribution, along with being biodegradable, non-toxic and small in size [26][27][28]. While PLGA NPs display many advantages in antigen delivery, in comparison with cationic bio/polymers, they can suffer poor encapsulation e ciency and instable when loading negatively charged molecules, such as protein or peptide antigen. Chitosan (CS) is a non-toxic and non-immunogenic naturally occurring linear amino poly-saccharide (poly 1,4-dayglucoamine), with an ability to enhance the penetration of large molecules across mucosal surfaces [29]. Such anionic PLGA NPs can be subjected to chitosan-surface coating, thus resulting in cationic chitosancoated PLGA (CS-PLGA) NPs, which hold promise as innovative formulations for targeted delivery [27]. Previous studies have demonstrated that CS-PLGA NPs are particularly effective for antigen delivery to APC, such as inducing antigen presentation to lymphocytes or modulating APC function [27].
In this study, we chose A. baumannii outer membrane protein 22 (Omp22), a highly conserved and highly immunogenic protein, as the candidate antigen. Previous study has found that immunization with recombinant Omp22 e ciently elicited high titers of speci c IgG, increased the survival rates of mice, and suppressed the bacterial burdens in the organs and peripheral blood [16]. However, Omp22 is not only a key protein involved in A. baumannii metabolic process, but also has certain toxicity [16]. Therefore, we used bioinformatics techniques and immunological methods to predict and identify optimal T-cell and Bcell epitopes on A. baumannii Omp22 protein. Subsequently, the identi ed dominant epitopes were connected in series by 6-aminocaproic acid and chemically synthesized to generate multi-epitope peptide rOmp22. Then, rOmp22 was encapsulated by CS-PLGA to prepare multi-epitope peptide nanovaccine (CS-PLGA-rOmp22). The physical-structural characterization, immunogenicity and protection e cacy of the vaccine were evaluated comprehensively in vitro and in vivo. This novel nanovaccine can retain the corresponding immunogenicity of Omp22, and avoid its harmfulness to the host, which should become a priority strategy against A. baumannii infection.

Heterologous expression and puri cation of recombinant Omp22
The recombinant Omp22 protein was synthesized and puri ed for further study. The Omp22-encoding gene was ampli ed by PCR, cloned into the plasmid pET-28a (+), and transformed into Escherichia coli (E. coli) BL21 (DE3). The positive clones were con rmed by restriction (Fig. 1a) and sequencing (Additional le 1: Fig. S1). The recombinant protein Omp22 was expressed in E. coli BL21 (DE3) and was puri ed with Ni-IDA resin. The puri ed protein was analyzed by SDS-PAGE, a speci c band was seen at 24 kDa, which was consistent with the expected size (Fig. 1b).
Epitope prediction and design of multi-epitope peptide rOmp22 According to bioinformatics analysis, such as secondary structure of epitope, surface accessibility, hydrophilicity, exibility and antigen index, four candidate B-cell epitope peptides (named Omp22 B1-Omp22 B4) and four candidate T-cell epitope peptides (named Omp22 T1-Omp22 T4) ( Table 1) were predicted and chemically synthesized. BALB/c mice were subcutaneously injected with 100 µg recombinant Omp22 (1 µg/µl in PBS) or an equal volume of PBS for three times with two-week interval. One week after the last immunization, blood and spleen were collected for further immunology study,the schedule of vaccination are shown in Fig. 1c. Three B-cell epitopes (amino sequences: NIPLSQARAQSVKNY;YATLDKVAQTL SVQLIMP) were recognized by serum from mice immunized with Omp22 whole-length protein (Fig. 1d). Splenocytes from Omp22 protein immunized mice could be stimulated by two synthetic T-cell epitopes (amino sequences: VPSSRIDAQGY; TFDTNKSNIKP) ( Fig. 1e). The identi ed optimal T-cell epitopes and B-cell epitopes were connected in series by 6-aminocaproic acid to design and generate a novel multi-epitope peptide (MEP) rOmp22 (Fig. 1f). MEP rOmp22 with 59 amino acid and molecular weight of 6535.48 Da was chemically synthesized and identi ed by liquid chromatography and mass spectrometry (Additional le 1: Fig. S2).

Physical-structural characterization of nanoparticles
The characteristics of CS-PLGA-PBS and CS-PLGA-rOmp22 nanoparticles were summarized in Table 2.
Transmission electron microscopy (TEM) techniques were employed to assess the morphology and size of the nanoparticles. By TEM analysis, CS-PLGA-PBS ( Fig. 2a) and CS-PLGA-rOmp22 ( Fig. 2b) NPs appeared to be apparently smooth with evenly double spherical structure. Particle size analysis showed that the diameter of CS-PLGA-PBS and CS-PLGA-rOmp22 NPs were at 295.78 ± 21.28 nm (Fig. 2c) and 272.83 ± 15.57 nm respectively (Fig. 2d), with nano-range size (100-300 nm) and moderate uniformity (PDI around 22.8%). As shown in Fig. 2e and 2f, the absolute value of the zeta potential for CS-PLGA-PBS and CS-PLGA-rOmp22 were about 3.70 and 4.39 mV, indicating a high repulsion between the nanoparticles, which kept the NPs in a state of dispersion instead of aggregation or clumping.
Encapsulation e ciency and release studies of CS-PLGA-rOmp22 nanoparticles We calculated the encapsulation e ciency (EE) by high performance liquid chromatography (HPLC) method. The results showed that EE was about 55% and loading capacity was about 0.94% ( Table 2).
The release of the rOmp22 peptide from CS-PLGA-rOmp22 NPs was measured by a sustained slow release over a 72 hours period. The release of rOmp22 peptide from CS-PLGA-rOmp22 was found to be 26.2%, 40.6%, 51.5% and 60% of the total encapsulated peptide on 12 h, 24 h, 48 h and 72 h, respectively (Fig. 3a). Cytotoxicity effect of rOmp22 and CS-PLGA nanoparticles on A549 cells Toxicity is of major concerns when using nanoparticles, even biodegradable polymers, in biomedical applications. We tested the toxicity of rOmp22 and nanoparticles against human lung adenocarcinoma epithelial cell line A549. Both dose-and time-dependent toxicity studies were conducted by A549 cells exposing to rOmp22 ranging from 1.25 µg/ml to 80 µg/ml and CS-PLGA nanoparticles ranging from 12.5 µg/ml to 800 µg/ml over a 6-, 24-and 48-h time period. At all examined time points, all concentrations of rOmp22 ( Fig. 3b) and nanoparticles (Fig. 3c, Fig. 3d) did not affect the viability of A549 cells, which demonstrated that they were non-toxic to A549 cells, and substantiated their safety and biocompatibility for studies in vivo.

Generation of antibody-mediated immune responses
To measure the antibody-mediated immune responses at different vaccination regimens, BALB/c mice were subcutaneously injected with 40 µg rOmp22 or CS-PLGA-rOmp22 nanovaccine containing equal rOmp22. Mice injected with CS-PLGA-PBS or adjuvant were as control groups. Serum samples were collected from immunized mice after the third immunization ( Fig. 4a). We investigated the capacity of encapsulated rOmp22 to potentiate antibody immune responses in mice by quantifying rOmp22-speci c antibodies using ELISA. As shown in Fig. 4b, immunization with CS-PLGA-rOmp22 elicited signi cantly (P < 0.01) higher IgG compared to rOmp22 immunized mice. No antigen-speci c antibody was detected in the serum from the CS-PLGA-PBS control group or the adjuvant-treated group.
To further de ne the robustness of the rOmp22 speci c antibody, we next performed serial two-fold dilutions of sera to determine rOmp22 speci c antibody titers. As shown in Fig. 4c and Fig. 4d, sera from the CS-PLGA-rOmp22 mice had higher antibody titers and endpoint titers in comparison with those from non-encapsulated rOmp22 immunized mice and the control mice.
Splenocyte stimulation and measurements of cytokines IL-4 and IFN-γ Splenocytes were isolated from six mice in each group at day 7 after the third immunization and cultured with the stimulation of rOmp22. IL-4 and IFN-γ secretion in the culture supernatants were detected by ELISA method. The splenocytes from the CS-PLGA-rOmp22 immunized mice produced higher levels of IFN-γ than those from rOmp22-immunized group (P < 0.01) (Fig. 4e), whereas IL-4 secretion showed no difference between CS-PLGA-rOmp22 group and non-encapsulated rOmp22 group (Fig. 4f).

Flow cytometry analysis
Seven days after the last immunization, cells were harvested from the spleen and the draining lymph node. Flow cytometry analysis showed that nanovaccine CS-PLGA-rOmp22 induced signi cantly more CD 19 + B cells and less CD 3 + T cells in the lymph node after immunization compared with non-encapsulated rOmp22 ( Fig. 5a and 5c). There was no difference of CD 19 + cell levels in spleen among mice in different groups (Fig. 5b). The frequencies of CD 3 + cells were somewhat increased in the spleen of CS-PLGA-rOmp22 immunized group (Fig. 5d). No signi cant differences were observed in the frequencies of CD 11c + dendritic cells (DC) and CD 11b + CD 169 + macrophages (MAC) in spleen and draining lymph node

Immunizations reduced bacterial loads
Antibody and cell-mediated immunity are just predictors for immune responses. To determine whether CS-PLGA-rOmp22 vaccine confered protection, we assessed its effectiveness using an acute pneumonia model in BALB/c mice. Two weeks after the last boost immunization, all groups were challenged with lethal dose of A. baumannii ATCC19606 (2 × 10 8 CFU/mouse) and three clinical A. baumannii strains, CS-MDR-AB (1 × 10 9 CFU/mouse), CRAB (5 × 10 8 CFU/mouse) and PDR-AB (5 × 10 8 CFU/mouse) via the trachea. Drug sensitive test results of four A. baumannii strains were shown in Additional le 1 (Table  S1). Six mice were randomly selected from each group for the detection of bacterial load in the blood and lung tissue (Fig. 6a).
Blood samples collected from six mice in each group in 24 h post-challenge were serially diluted and plated on LB agar plates followed by incubation at 37℃ overnight. The number of CFUs was counted, and the log 10 CFU/ml was calculated and compared. Mice immunized with CS-PLGA-rOmp22 had lower bacterial loads in the blood than mice from the other three groups ( Fig. 6b and 6c).
After taken blood, six mice in each group were sacri ced, lungs were collected. The right lung lobes were prepared for CFU assessment. As shown in Fig. 6d-g, the lungs from CS-PLGA-rOmp22 groups showed signi cantly lower bacterial burden than those two control groups. Further, the bacterial burden was signi cantly reduced in the CS-PLGA-rOmp22 vaccinated group compared with the non-encapsulated rOmp22 group. These results indicated that immunization with CS-PLGA-rOmp22 could partially reduce the colonization of A. baumannii in the lung of mice.

Pathological changes in mouse lung tissue
The left lung tissue was aseptically collected in 24 hours post-challenge and xed in 4% buffered formalin, stained with hematoxylin-eosin and observed under microscope. As shown in Fig. 7, the lung slices of the CS-PLGA-rOmp22 group showed less extensive of lung lesions, alveolar edema, lymphocyte in ltration and structural damage caused by in ammatory response compared with the rOmp22 vaccine group, CS-PLGA-PBS group and adjuvant control group. Moreover, in the CS-PLGA-rOmp22 vaccine groups, there were more normal structure and clearer alveoli than other three groups. The results of severity score showed that the severity of lung injury in CS-PLGA-rOmp22 groups was signi cantly lower than the other groups ( Fig. 7c-f), suggesting that mice immunized with CS-PLGA-rOmp22 vaccine showed decreased in ammatory response in lungs.
Survival rate, body weight changes and clinical score of mice post-challenge Six mice challenged with A. baumannii were randomly selected for recording survival, body weight changes and clinical score every day for seven days. As shown in Fig. 8a-d, all mice in CS-PLGA-PBS and adjuvant treated groups died 72 h post-challenge. After the lethal dose of ATCC19606, CS-MDR-AB, CRAB and PDR-AB strains challenged, the 7-day survival rates of mice immunized with CS-PLGA-rOmp22 were 83.33%, 71.43%, 66.67% and 57.14%, respectively, which were signi cantly higher than those immunized with rOmp22 (60%, 60%, 50% and 42.86%, respectively).
After challenge, the body weight ( Fig. 8e-h) and clinical symptom scores ( Fig. 8i-l) of each group decreased to the lowest after two to three days post infection. Then, the symptoms of mice in the CS-PLGA-rOmp22 group and rOmp22 vaccinated group gradually improved. The body weight of mice returned to that before challenge, and the symptoms disappeared in seven days post challenge. These results demonstrated that mice immunized with CS-PLGA-rOmp22 nanovaccine were better protected from lethal dose of A. baumannii infection, compared with mice vaccinated with non-encapsulated rOmp22.

Discussion
Multidrug-resistant A. baumannii is a rapidly emerging pathogen causing infections with high mortality rates due to inadequate medical treatment [1,2,6]. New ways to prevent and treat such infections are of a critical medical need. Despite decades of effort in the development of A. baumannii vaccine, there is still no effective vaccine against this pathogen.
Recently, there is increasing interest in the development of vaccines which use only minimal components from pathogens. Such vaccines are based on recombinant proteins or even minimal fragments carrying immunological information from this protein, called peptide epitopes [24,30]. Epitopes or antigenic determinants are the minimal immunogenic part of any particular antigen, which are capable of inducing speci c immune responses [24]. In this research, we chose A. baumannii Omp22, a highly conserved and highly immunogenic protein, as the candidate antigen. Bioinformatics techniques and immunological methods were used to predict and identify optimal T-cell and B-cell epitopes on Omp22 protein. Then, the identi ed dominant epitopes were connected in series by 6-aminocaproic acid and chemically synthesized to generate a novel multi-epitope peptide rOmp22, which preserved the antigenic epitopes and avoided toxic structure fragments of Omp22.
Peptide-based subunit vaccine holds great potential to be a safer and more e cient alternative to traditional vaccination strategies. However, nontoxic and effective delivery systems are needed to protect their respective immunogens from rapid degradation and to potentiate immunological responses [31]. A novel type of composite microspheres, CS-PLGA, was demonstrated to possess advantages of improving the stability of encapsulated proteins and increasing the subsequent release. In this view, we encapsulated a multi-epitope peptide rOmp22 in CS-PLGA NPs with the high encapsulation e ciency rate and slow release pattern. To the best of our knowledge, this study was the rst to undertake the encapsulation of the multi-epitope peptide rOmp22 in CS-PLGA NPs, physical-structural characterization studies, and its immunogenicity and protection e cacy. This is also the rst time that the CS-PLGA NPs delivery system has been used for epitopes vaccine research in bacteria.
The morphological characteristics of nanoparticles can affect the release of biomaterials from the nanoparticle. A leaky or porous structure enhances the release of biomaterials by diffusion, whereas a smooth surface reduces the burst release [32]. The physical and structural characteristics of CS-PLGA-rOmp22 revealed that it had a homogeneous morphology with a smooth spherical shape. CS-PLGA-rOmp22 was also uniform in particle size distribution and fairly dispersed without aggregation. The zeta potential value is one of the most important particle characterizations because it can affect both particle stability and particle adhesion [33]. As shown in Fig. 2e and 2f, CS-PLGA NPs exhibited positive charge, which could adsorb more peptides, increase encapsulation e ciency and improve stability.
The release pattern of a peptide in nanoparticles is important to the development of a nanovaccine, since it can in uence the immune response and the immunization regimen, such as the peptide concentration and the frequency of immunization. In the present study, the CS-PLGA-rOmp22 release pro les were biphasic, characterized by a burst of peptide followed by a sustained release. The burst release at the rst day could induce strong immune response. The sustained slow release of the peptide was an attractive property for a vaccine candidate as this might reduce the number of immunizations as well as enhancing the presentation of the peptide to APCs. Cell toxicity studies showed that CS-PLGA NPs was not toxic to cells at concentration as high as 800 µg/ml, thus indicating the safety of this delivery system as reported by several researchers [27,28].
The present study showed that subcutaneous administration of CS-PLGA-rOmp22 nanovaccine induced systemic antibody responses. As shown in Fig. 4b-4d, high titers of antigen-speci c antibody IgG were detected in the serum of mice immunized with CS-PLGA-rOmp22 after boosting twice. Meanwhile, CS-PLGA-rOmp22 vaccination induced much higher levels of antigen-speci c IFN-γ and IL-4 secretion in the spleen cell culture supernatants in CS-PLGA-rOmp22 and non-encapsulated rOmp22 groups compared to those in control mice. The CS-PLGA-rOmp22 vaccination induced much higher levels of antigen-speci c IFN-γ production from splenocytes than rOmp22-immunized group (P < 0.01), whereas IL-4 secretion showed no difference between CS-PLGA-rOmp22 group and non-encapsulated rOmp22 group ( Fig. 4e and  4f). These results indicated that humoral and cellular immune responses, especially Th1-type immune response had been induced and created a full protection.
The challenge experiments proved that the mice immunized with CS-PLGA-rOmp22 acquired potent protection against the infection of A. baumannii ATCC 19606 and three clinical A. baumannii strains. The bacterial load in the blood of the mice inoculated with CS-PLGA-rOmp22 after infection was much lower than that of non-encapsulated rOmp22 group and the control groups. Almost no pathological change was observed in the lung tissue of the mice immunized by CS-PLGA-rOmp22. These results indicated that the high titer of antigen-speci c antibody contributed to the potent protection in the mice immunized with CS-PLGA-rOmp22.

Conclusions
In summary, this work reported a novel multi-epitope peptide nanovaccine against A. baumannii. CS-PLGA-rOmp22 NPs could elicit speci c IgG antibody, Th1 cellular immunity and protection against acute lethal intratracheal A. baumannii challenge in BALB/c mice. Our results indicate this nanovaccine is a desirable candidate to prevent A. baumannii infection.   Table S1) according to clinical and laboratory standards institute (CLSI) M100. The E. coli BL21 (DE3) and the plasmid pET28a (+) used in the study were purchased from Novagen company (Beijing, China) and kept in our laboratory. For all experiments, unless otherwise stated, bacteria were grown on Luria-Bertani (LB: 10 g/L tryptone, 5 g/L yeast extract, 10 g/L sodium chloride) agar plates or in LB broth at 37℃.

Prediction and Identi cation of T-cell epitopes and B-cell epitopes
The physicochemical properties of A. baumannii Omp22 protein was analyzed and all possible dominant B-cell and T-cell epitopes were predicted using the immuno-informatics approach. The B-cell epitopes were predicted using bioinformatics software OptimumAntigen™ Design Tool (GenScript, China). Four candidate B-cell epitopes were predicted according to their secondary structure, surface accessibility, hydrophilicity, exibility, and antigenic index. When predicting T-cell epitopes, The IEDB (Immune Epitope Database Analysis Resource) (https://tools.iedb.org/mhci/) was used and four T-cell epitopes were stored based on their scores [34,35]. The predicted B-cell and T-cell epitope peptides were chemically synthesized by Jiangsu GenScript Biotechnology Co. Ltd.
Twelve female BALB/c mice (6 to 8 week) were randomly divided into two groups, Omp22 immunized group and phosphate-buffered saline (PBS) control group. Mice were subcutaneously injected with 100 µg recombinant Omp22 (1 µg/µl in PBS) or an equal volume of PBS for three times with two-week interval. In the Omp22 vaccine group, Freund's adjuvant was added to enhance the immune effect. One week after the last immunization, serum of each mouse was collected to detect B-cell epitope speci c antigen by indirect ELISA. Splenocytes were isolated from vaccinated mice and adjusted to a concentration of 1 × 10 6 cells/ml, and 200 µl of the cell suspension was added to each well of a 96-well plate and stimulated with 20 µg/ml of candidate T-cell epitopes. After incubated for 72 h, supernatants were collected and levels of gamma interferon (IFN-γ) were measured using mouse IFN-γ ELISA kits.
Based on the above analysis, three B-cell epitopes (amino sequences: NIPLSQARAQSVKNY; YATLDKVAQTL and SVQLIMP) and two T-cell epitopes (amino sequences: VPSSRIDAQGY; TFDTNKSNIKP) from Omp22 were selected to design the multi-epitope protein.

Design and synthesis of rOmp22
Three optimal B-cell epitopes and two T-cell epitopes were connected in series by 6-aminocaproic acid. A multi-epitope peptide rOmp22 of 59aa, with a molecular weight of 6536. An equivalent volume of PBS as used for rOmp22 was similarly encapsulated in CS-PLGA to obtain CS-PLGA-PBS to serve as a negative control. All lyophilized nanoparticles were stored at -80℃ until used.

Encapsulation e ciency and peptide loading level
The encapsulation e ciency of rOmp22 in CS-PLGA was measured using HPLC (Agilent, USA) by quantitating rOmp22 in supernatant after ultracentrifugation. The rOmp22 encapsulation e ciency (EE) and the peptide loading capacity (LC) were calculated using the following formulas: Where A is the total amount of rOmp22, B is the free amount of rOmp22, and C is the CS-PLGA-rOmp22 weight. These measurements were performed three times.
Particle size and zeta potential The particle size, polydispersity index (PDI) and zeta potential were measured by particle size analyzer (Anton Paar, Graz, Austria). CS-PLGA-rOmp22 or CS-PLGA-PBS was suspended in ltered distilled water, sonicated, and placed in a cuvette to measure size and zeta potential. Each sample was measured three times and reported as the mean of triplicates for size (diameter in nanometers) and zeta potential (millivolt). These experiments were conducted at least three times.

Transmission electron microscopy (TEM)
The morphology of CS-PLGA-rOmp22 and CS-PLGA-PBS were observed using high-resolution TEM (Hitachi, HT7700 Exalens). One drop of the complex was deposited on the copper grid (carbon-coated copper grid, 200 mesh). After adding phosphotungstic acid, the grids were dried for 10 min prior to TEM analysis.
In vitro peptide release The release of the rOmp22 peptide from CS-PLGA was determined following the method of Bouissou et al [37]. Antibody titer measurement with ELISA Multi-epitope peptide rOmp22 was rst diluted to an optimal concentration (10 µg/ml) to coat a 96-well plate. The resulting solution was then added into each well (100 µl per well) and incubated for 12-18 h at

Establishment of pneumonia models
A. baumannii ATCC19606 strain and 3 clinical A. baumannii strains (CS-MDR-AB, CRAB and PDR-AB) were grown in LB broth to the late-logarithmic phase at 37℃/150 rpm. Cells were harvested by centrifugation at 4000 g for 10 min, washed and resuspended in PBS and mixed with porcine mucin (Sigma-Aldrich) to a nal concentration of 5% mucin. Desired CFU/ml was obtained by appropriate dilutions and the nal concentration was quanti ed by plating serial dilutions onto LB agar plates. The mice were anaesthetized with intraperitoneal (i. p.) injection of pentobarbital sodium, placed in a supine position and their trachea were exposed surgically. Lethal doses of ATCC19606 (2 × 10 8 CFU) and three clinical A. baumannii strains, CS-MDR-AB (1 × 10 9 CFU), CRAB (5 × 10 8 CFU) or PDR-AB (5 × 10 8 CFU) in a total volume of 100 µl was intra-tracheally to mice to induce acute pneumonia. The incised area was sealed with sterile surgical sutures. The mice were monitored for 7 days, body weight, clinical score and survival number from each group were recorded every day.
Bacterial load assessment in blood and lung Blood samples were collected from six mice in each group at 24 h post-challenge. To determine the bacterial loads in blood, samples were serially diluted and plated on blood agar plates. After taking blood, the mice were killed, lungs were removed aseptically, weighed, and homogenized. Serial dilutions of tissue homogenates were plated onto blood agar plate. Bacterial CFUs were enumerated after 24 hours of incubation at 37℃.

Histopathological examination
Lungs were removed under aseptic conditions and xed in 4% formalin. Histopathological examination of the section after embedding in para n was observed under microscope after staining with haematoxylineosin (HE). Lung injury was estimated by the percentage of the lesion area in the total lung area using an ImagePro macro.

Statistical analyses
Statistical analyses were performed using the Statistical Package of Social Sciences (SPSS, version 23.0; SPSS Inc., Chicago, IL) and Graphpad Prism (version 6.0; Graphpad software Inc., La Jolla, CA). All data were expressed as Mean ± SD. The one-way analysis of variance (ANOVA) was used for multiple comparisons, followed by Bonferroni's post hoc test. Survival data were compared using the log-rank test. A P value of < 0.05 was considered signi cant.     immunization. i and j Scatterplots present the major histocompatibility complex class II (MHC II) expression on CD 11c+ dendritic cells in the lymph node (i) and spleen (j). k and l Scatterplots present the major histocompatibility complex class II (MHC II) expression on CD 11b+ CD 169+ cells in the lymph node (k) and spleen (l). Statistics was performed using a one-way ANOVA with Tukey's post hoc correction (n=4). In all studies, *P < 0.05, **P < 0.01, ***P < 0.001. MFI, mean uorescence intensity.

Figure 6
Bacterial loads in the blood and lung tissues of mice. a Immunization and A. baumannii challenge process. Two weeks after the last immunization, BALB/c mice were intratracheally infected with lethal dose of A. baumannii ATCC19606 (2×108 CFU/mouse) and three clinical A. baumannii strains, CS-MDR-AB (1×109 CFU/mouse), CRAB (5×108 CFU/mouse) and PDR-AB (5×108 CFU/mouse). b and c Bacterial loads in the blood of mice in 24 hours post-challenge. Blood samples were serially diluted and plated on LB agar plates followed by incubation at 37℃ overnight. The number of colony forming units (CFU) was counted, and the log10 CFU/ml blood was calculated and compared. d-g Bacterial loads in lungs. Lungs were removed at 24 h after intra-tracheal challenge with A. baumannii. Bars indicate mean ± SD (n=6). * P < 0.05, ** P < 0.01, *** P < 0.001.  , 50 μm). c-f Semi-quantitative analysis of the in ammation area in the lung tissue (n=6). Histograms show the mean percentage of lesion area within the total lung. Data are presented as the means± SD (n=6). * P < 0.05, ** P < 0.01, *** P < 0.001.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.