Identification of novel vaccine candidates against Acinetobacter baumannii using reverse vaccinology

Abstract

Acinetobacter baumannii (Ab) is a global emerging bacterium causing nosocomial infections such as pneumonia, meningitis, bacteremia and soft tissue infections especially in intensive care units. Since Ab is resistant to almost all conventional antibiotics, it is now one of the 6 top-priorities of the dangerous microorganisms listed by the Infectious Disease Society of America. The development of vaccine is one of the most promising and cost-effective strategies to prevent infections. In this study, we identified potential protective vaccine candidates using reverse vaccinology. We have analyzed 14 on-line available Ab genome sequences and found 2752 homologous core genes. Using information obtained from immuno-proteomic experiments, published proteomic information and the bioinformatics PSORTb v3.0 software to predict the location of extracellular and/or outer membrane proteins, 77 genes were identified and selected for further studies. After excluding those antigens have been used as vaccine candidates reported by the in silico search-engines of PubMed and Google Scholar, 13 proteins could potentially be vaccine candidates. We have selected and cloned the genes of 3 antigens that were further expressed and purified. These antigens were found to be highly immunogenic and conferred partial protection (60%) in a pneumonia animal model. The strategy described in the present study incorporates the advantages of reverse vaccinology, bioinformatics and immuno-proteomic platform technologies and is easy to perform to identify novel immunogens for multi-component vaccines development.

Keywords:

• Acinetobacter baumannii
• bioinformatics
• immuno-proteomic
• multi-drug resistance
• reverse vaccinology
• vaccine
• pneumonia

Introduction

Acinetobacter baumannii (Ab) is a gram-negative bacterium responsible for opportunistic nosocomial infections such as pneumonia, bacteremia, and meningitis.^1-6 Ab can stick to dry surfaces for a long time especially in hospital intensive care units through ventilator-associated machine and biotic surfaces. Ab accounts for >7,000 hospital-acquired infections (HAI) and over 500 deaths per year in the United States only.³ Since the numbers of multi-drug resistant (MDR) and pan-drug resistant (PDR) strains are increasing,^3-6 prevention and treatment of Ab infections have become a significant health care challenge and new strategies against Ab infections are urgently needed. In the absence of efficacious and safe antibiotics, vaccination represents the best strategy to combat MDR pathogens.^6,7 So far, several potential Ab antigens have been identified through active and passive immunizations of experimental animals. They include outer membrane vesicles (OMVs), outer membrane protein A (OmpA), auto-transporter (Ata), biofilm-associated protein (Bap), K1 capsular polysaccharide and Poly-N-acetyl-β-(1-6)-glucosamine (PNAG).^7-14 The sequence variability of these antigens and their absence in the circulating Ab strains may result poor cross-protective efficacy against HAI caused by new emerging Ab strains that may possibly mutate under selective immunological pressure and down-regulate the target antigens. In our hand, recombinant OmpA and a truncated fragment of Bap have been tested and found no protection in mouse pneumonia challenge studies; only inactivated Ab and/or PNAG-protein conjugate showed good protection in the mouse pneumonia challenge model (Chong et al., unpunished results). Therefore, the identification of highly conserved antigens expressed by the new emerging Ab strains during infections is still warranted.

With the ever increasing information on microbial genomes, reverse vaccinology (RV) has become a promising in silico approach to select vaccine targets.^6,15-17 RV integrates comparative analysis of genome sequences and web-based prediction tools for screening highly conserved outer-membrane proteins and soluble secreted proteins. Meningococcal group B vaccine (4CMenB) serves as an example of a successful RV approach.^6,16 Combined with other approaches such as in vitro proteomic analysis of OMVs secreted by bacteria, differential expression analysis of virulent strains, in vivo animal model challenge studies, and immuno-proteomics, it provides an important platform to identify novel protective antigens.^18-33 Furthermore, through these studies and analyses, we can also deeply understand the pathogenesis of Ab bacteria. In this study, we used an integrated approach combining reverse vaccinology, bioinformatics-based structural analysis and results obtained from published proteomics studies to screen and identify potential protective antigens against Ab infections. As a proof of concept, potential antigens were designed for in vitro/ in vivo vaccine evaluation in a mouse pneumonia challenge model.

Results and Discussion

In silico screening of potential vaccine candidates

We used the 14 complete genome sequences currently available (levels: gapless chromosome) including 11 multi-drug resistance (MDR) strains from NCBI database for comparative analysis using MBGD program (available on http://mbgd.genome.ad.jp/).³⁴ Three genomes belong to international clade I (IC I), 8 genomes are IC II, 2 genomes are IC III and another one is unclassified indicating the wide coverage of Ab strains. In addition, these clones are characterized for MDR, the year of isolation, geographic location and by their accession number (Table 1). As shown in Figure 1, the phylogenetic lineage of A. baumannii strains analyzed in this study was inferred by the neighbor-joining method performed using the MEGA6 software and the previously published strategy.^35,36 Using the MBGD database, 2752 orthologous core genes were identified as novel potential vaccine antigens for these strains. The strain AYE which was used as a reference for annotating was associated with a human clinical isolate from the nationwide outbreak in France in 2001.^35,36 In order to predict the localization of proteins in the 14 available genomic sequences, PSORTb web tool was used to define outer membrane or extracellular proteins. This led to the identification of 565 non-redundant antigens as potential vaccine candidates as shown in Figure 2.

Table 1. Complete Ab genomes from NCBI used in this study

Strain	ORF	Year	Country	MDR	Accession no.
ATCC17978	3993	1951	USA	No	NC_009085
AB307-0294	3550	1994	USA	No	NC_011595
AYE	3913	2001	France	Yes	NC_010410
AB0057	3969	2004	USA	Yes	NC_011586
ACICU	3908	2005	Italy	Yes	NC_010611
MDR-ZJ06	3965	2006	China	Yes	NC_017171
D1279779	3396	2009	Australian	No	NC_020547
1656-2	3925	2011	Korea	Yes	NC_017162
TCDC-AB0715	4136	2011	Taiwan	Yes	NC_017387
MDR-TJ	3936	2012	China	Yes	NC_017847
TYTH-1	3682	2012	Taiwan	Yes	NC_018706
BJAB07104	3861	2013	China	Yes	NC_021726
BJAB0868	3819	2013	China	Yes	NC_021729
BJAB0715	3908	2013	China	Yes	NC_021733

Figure 1. Phylogenetic relationship of Acinetobacter baumannii. The phylogenetic lineage of fully sequenced Ab strains analyzed in this study was inferred by the neighbor-joining method performed in MEGA6 using the strategy previously published in.³⁵

Figure 2. Schematic summarizing the flow chart of Reverse Vaccinology used in this study. 1.Comparison of genomic sequences to obtain the orthologous core genes by the in silico web-based tools (MBGD). 2. 2.1 PSORTb 3.0 to predict the cellular localization of each protein. 2.2 Outer membrane and extracellular proteins identification by proteomics analyses/references. 3. Blasting of protein sequences and literature-review to determine if the proteins have ever been used as immunogens. 4. Expression and purification of the consensus sequence of candidate immunogens. 5. Utilization of the mouse pneumonia model to assess vaccine immunogenicity and protective ability against A. baumannii infection. 6. Development of a broadly protective A. baumannii vaccine formulation.

In vitro identification of outer-membrane and secreted proteins

Outer-membrane vesicles (OMVs) secreted by many Gram-negative bacteria can mediate intercellular exchanges, of lipo-polysaccharides (LPS), cell-cell signals, proteins and DNA as well as the transfer of lethal virulence factors.³⁷ A. baumannii secretes OMVs which interact with host cells and then deliver bacterial effectors contributing to bacterial pathogenesis and immuno-pathology.¹² To enhance the power of reverse vaccinology, we also integrated functional genomics and proteomic analysis using bioinformatics tools to further refine the screening criteria for protective antigens. We (results to be published elsewhere) and others^17-18,25,28 used HPLC-MS/MS to analyze the proteomic profiles of OMVs and identified 65 potential antigens from A. baumannii ATCC17978 strains (Table S1). Meanwhile, we have also exploited proteomic information on outer membrane and/or secreted proteins predicted from 6 previous experimental in vivo studies.^{18,22,23,25-28} Taking together the information collected for homologous core genes, in vitro published proteomic information, and in silico predicted secretome profiles, 77 A. baumannii proteins were identified and selected as the most promising vaccine candidates (Fig. 2 & Table S2).

Selection of antigen targets

To minimize the number of vaccine candidates for immunogenicity studies, the selected 77 potential antigens were further analyzed through PubMed and Google Scholar searches to identify those previously used as vaccine candidates against A. baumannii infections. Our combined literature search led to the identification of 13 top vaccine targets based on sequence homologies, prevalence, cellular location and potential to be vaccine candidate in other pathogens (Table 2). These potential vaccine antigens are essentially present in all published A. baumannii strains and thus are excellent candidate immunogens to be included in an effective A. baumannii vaccine. According to the definition of Cluster of Orthologous Groups (COG) protein function,^36,37 these 13 proteins could fall into 5 different functional groups: 5 could belong to the “cell wall/membrane/envelope biogenesis” group; 4 are in the “posttranslational modification, protein turnover, chaperones” group; 2 are “energy production and conversion” proteins; one is involved in “lipid transport and metabolism;” and one in “nucleotide transport and metabolism” (Table 2).

Table 2. Most promising immunogens for an A. baumannii vaccine

Locus tag	Mw	Pi	Product and Antigenicity score^a	Ref.
Cell wall/membrane/envelope biogenesis
ABAYE0130	27.6	4.75	Putative exported protein (OmpK), 0.73	^38
ABAYE1048	29.6	9.84	Lipoprotein nlpD precursor, 0.64	^39
ABAYE1583	93.8	5.26	Outer membrane protein/protective antigen OMA87, 0.63	^40,41
ABAYE2931	22.5	9.30	Outer membrane protein/peptidoglycan-associated (lipo)protein, 064	^42
ABAYE3514	50.3	8.98	type I secretion outer membrane, TolC family protein, 0.63	^43
Posttranslational modification, protein turnover, chaperones
ABAYE3820	25.6	8.59	FKBP-type 22kD peptidyl-prolyl cis-trans isomerase (FKlB), 0.66	^44
ABAYE2564	20.8	9.03	alkyl hydroperoxide reductase subunit C22, 0.56	^45
ABAYE1454	18.5	5.38	Cyclophilin, 0.48	^46,47
ABAYE0619	24.2	5.16	ahpC/TSA family protein, peroxidase, 0.54	^48,49
Lipid transport and metabolism
ABAYE0711	50.6	5.12	putative long-chain fatty acid transport protein (Ompp1), 0.65	^50,51
Energy production and conversion
ABAYE0776	67.0	5.94	succinate dehydrogenase flavoprotein subunit, 0.54	^52
ABAYE0782	51.0	5.89	dihydrolipoamide dehydrogenase (Lpd), 0.49	^53
Nucleotide transport and metabolism
ABAYE3267	15.5	5.53	Nucleoside diphosphate kinase (NDK), 0.37	^54-56

^aAntigenicity scores were done by VaxiJen v2.0 server.

(http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen).

ABAYE0130 is homologous with the OmpK of the fish pathogen Vibrio harveyi which has been shown to be effective as a fish vaccine candidate against V. harveyi infactions.³⁸ The Nlp gene of Yersinia pestisnlp D (similar to ABAYE1048) encoding for outer membrane lipoprotein could elicit a protective immunity against Y. pestis.³⁹ ABAYE1583 contains the conserved β-barrel assembly of outer membrane proteins and is similar to Treponema pallidum subspecies Tp92 protein and Pasteurella multocida Oma87 that have previously been identified as potential vaccine candidates.^40,41 The ABAYE2931 sequence shares homology with a peptidoglycan-associated lipoprotein related to Aliivibrio salmonicida Pal protein that had been used successfully as immunogen in an A. salmonicida fish vaccine.⁴² ABAYE3514 which has sequences homology to TolC protein of Samonella paratyphi and belongs to the type I secretion outer membrane channel component could be the vaccine target.⁴³ ABAYE3820 comprises of a rotamase domain that may plays an important role in pathological processes²⁴ and is similar to Haemophilus parasuis FKBP-type peptidyl-prolyl cis/trans isomerase proposed as a vaccine candidate for Glässer's disease in swine.⁴⁴ ABAYE2564 is similar to a putative alkyl hydroperoxide reductase subunit C that has been identified as a potential target for multivalent anthrax vaccines.⁴⁵ ABAYE1454 has sequence similarity with the cyclophilin proteins previously been shown to be efficacious vaccines against toxoplasmosis and Leishmania infantum infections.^46,47 ABAYE0619 is identified to contain the conserved peroxidase domain that is homology to the thiol-specific-antioxidant (TSA) of Leishmania major that has been used as a vaccine candidate against L. major.^48,49 ABAYE0711 has a protein sequence similar to that of the long-chain fatty acid transport protein, Ompp1 which has been considered as a potential vaccine target against Samonella paratyphi and Haemophilus influenza.^50,51 ABAYE0776 could be involved in energy product and conversion as a putative succinate dehydrogenase that has been considered as a potential vaccine candidate for Leptospira spp. in an immuno-proteomic study.⁵² ABAYE0782 contains the highly conserved putative dihydrolipoamide dehydrogenase (Lpd) gene which was depleted in Mycoplasma gallisepticum and tested as a live-attenuated vaccine strain.⁵³ It implies that Lpd protein is virulence factor and could be a possible subunit vaccine target. Finally, ABAYE3267 protein sequence is similar to a highly conserved protein, the nucleoside diphosphate kinase (NDK) that is present in a number of bacteria including Bacillus anthracis, Francisella tularensis and Campylobacter fetus and considered to be a potential virulence target and thus a vaccine candidate.^54-56

Using the MBGD platform,³⁴ the proteomic data of OMVs obtained from HPLC/MS/MS experiments and bioinformatics prediction of surface-exposed antigens, we have identified 77 antigens as potential vaccine candidates (Table S2). The final list was restricted to 13 putative antigens. Recently, Moriel et al. utilized a similar approach and identified 42 putative vaccine candidates.¹⁷ Only one of their potential antigen candidates was identical to ABAYE1048 in our 13 vaccine candidates list and other 6 antigens (ABAYE0360, ABAYE0990, ABAYE1048, ABAYE1129, ABAYE2921, ABAYE3478) can be found in our list of 77 vaccine candidates. Obviously, the analysis of different genomic databases using similar functional genomics and bioinformatics tools could lead to different results. The major difference is the study strategy that we searched for surface-exposed antigens with functional homology to proteins previously identified as vaccine candidates for other pathogens. In order to identify novel antigens; we excluded proteins which had been tested and found to be not protective in our hands. Our current approach may have failed to identify potential vaccine immunogens, such as biofilm associated protein (Bap, ABAYE0792),¹² trimeric autotranspoter (Ata),¹¹ and other unknown putative proteins that could play critical role in A. baumannii pathogenesis. In addition, a limitation in our approach is the fact that immuno-proteomic data could be influenced by different culture conditions such as growing the pathogen in clinically relevant media reflecting the infection environment.⁵⁸

Selection of vaccine immunogens for a pilot study

Ideally, A. baumannii vaccine components should be potential protective immunogens also found in other pathogens and meet the following criteria: (i) expressed on the outer membrane and/or secreted as virulence factors that could interact with host cellular receptors; (ii) contain protein sequences and/or protective epitopes highly conserved among A. baumannii strains; and (iii) play an important role in pathogenesis during infection. The mono-valent vaccine may not meet all the above criteria, so multiple immunogens will very likely be necessary. Furthermore, under the selective immunological pressure generated by vaccination, these immunogens may possibly adapt and mutate to new strains and/or the prevalence of target antigens may be down-regulated. Therefore, 3 potential proteins OmpK, FKIB and Ompp1 corresponding to ABAYE0130, ABAYE3820 and ABAYE0711, respectively which were selected had the highest antigenicity scores 0.73, 0.66 and 0.65, respectively (Table 2). These antigens also are potential vaccine candidates in other bacterial pathogens^38,44,50,51 and should be evaluated for their protective ability in mouse immunogenicity studies. The protein sequence of each antigen from 14 published A. baumannii genomes was aligned by BlastP to obtain a consensus sequence, and the gene fragment of the identified consensus sequences was synthesized using the codon-optimization for E. coli expression.

Production of recombinant antigens

In a pilot study, 3 potential antigens (OmpK, FKIB, Ompp1) were produced and evaluated for their protective activities in a mouse pneumonia model.¹¹ Their respective genes were successfully expressed in E. coli as shown in Figure 3. After single-step purification using Ni-affinity chromatography, highly-purified OmpK, FKIB, and Ompp1 (>95% purity) were obtained and their purity was confirmed by SDS-PAGE (Fig. 3A) and western blot analysis (Fig. 3B). Trace amounts of degradation fragments were also detected as shown in the protein gel blot (Fig. 3B). These degradation products are likely the result of proteolytic digestion during the purification process. In any event, at least 20 mg of highly enriched antigens could be easily obtained from 1 liter of bacterial culture. Most of the E. coli endotoxin (LPS) was removed by passing the antigen preparations through an E membrane; and the residual LPS in the purified antigens was found to be below 30 EU per mL.

Figure 3. 12% SDS-PAGE and Western blot of purified recombinant proteins. The purification process was performed using immobilized-metal affinity chromatography. (A) Lane M-standard protein markers, Lane 1-Whole bacterial lysate after induction of OmpK (27.6 kDa) by IPTG, Lane 2-purified recombinant OmpK protein, Lane 3-Whole bacterial lysate after induction of Ompp1 (50.6 kDa) by IPTG, Lane 4-purified recombinant Ompp1 protein, Lane 5-Whole bacterial lysate after induction of FKIB (25.6 kDa) by IPTG, Lane 6-purified recombinant FKIB protein. (B) Western blot analysis of purified recombinant protein using mouse anti-His antibody.

Mouse immunogenicity studies

Blood samples from mice immunized with recombinant antigens were collected on day 42 and tested for antibody responses using antigen-specific ELISA. As shown in Figure 4A, individual recombinant antigens (OmpK, FKIB, Ompp1) formulated with CFA/IFA induced strong antigen-specific antibody responses (IgG titers > 10⁴). These results indicate that all 3 recombinant proteins are highly immunogenic since mice vaccinated with 2 × 3 μg of antigens produced very strong IgG antibodies with titer >10⁴ (data not shown).

Figure 4. Mouse immunogenicity studies of individual antigens. (A) Groups of C57BL/6 mice (n = 4) were intramuscularly immunized with either 3 or 10 μg of individual antigen (OmpK, FK1B and Ompp1) formulated with CFA/IFA adjuvant, or PBS on day 0, 14 and 21. At day 42, blood samples were collected and tested against each immunogen. Serum IgG antibody titers are reported. (B) Survival curves of C57BL/6 mice immunize with Ommp1. Groups of C57BL/6 mice (n = 5) were subcutaneously immunized with either 3 or 30 μg of Ommp1 formulated with CFA/IFA adjuvant, or PBS on day 0, 14 and 21. At day 42, the mice were challenged with 5 × 10⁷ CFU of freshly grown A. baumannii ATCC17978. The survival rates of mice were recorded.

Murine pneumonia challenge model studies

To evaluate the protective efficacy of each vaccine candidate, a murine pneumonia challenge model was established according to previous reports.^11,30,31 Because of the low virulence of A. baumannii in mice and porcine mucin was included in the inoculums. Mucin has been shown to decrease phagocytosis and increase bacterial virulence in the mouse challenge model. as shown in Figure S1. The results clearly indicate that A. baumannii can kill 100% of mice within 4 days at a challenge dose of 5 × 10⁷ cfu or within 2 days at a challenge dose of 1 × 10⁸ cfu.

When C57BL/6 mice (n = 5/group) vaccinated 3 times subcutaneously (s.c.) with the purified recombinant immunogens (3 or 30 μg/mouse) formulated with CFA/IFA were challenged with 5 × 10⁷ cfu of freshly grown A. baumannii (ATCC 17978 strain), all mice immunized with either OmpK or FKIP died within 6 days post challenge (data not shown). As shown in Figure 4B, only 50% of mice immunized with 30 μg dose of Ompp1 survived whereas 33% and 20% survival rates were observed in mice immunized with 3 μg dose of Ompp1 and PBS, respectively. When mice vaccinated 3 times s.c. with the mixture of recombinant antigens (3 × 3 or 10 × 3 μg of antigens mixture) formulated in CFA/IFA were challenged with 1 × 10⁸ cfu of freshly grown A. baumannii, the survival rate in the group vaccinated with 10 ug of immunogens was found to be 60%, whereas a 20% survival rate was observed for the group immunized with a 3 μg dose, that was not significantly different from the survival rate in the control group (Fig. 5).

Figure 5. Survival curves of C57BL/6 mice immunized with multivalent vaccines Groups of C57BL/6 mice (n = 5) were subcutaneously immunized with multivalent antigens or PBS on day 0, 14 and 21. At day 42, the mice were challenged with 1.34 × 10⁸ CFU of freshly grown A. baumannii ATCC17978. The survival rates of mice were recorded.

To mimic human disease, we had employed a murine pneumonia model to evaluate the immunogenicity and protective effects of potential vaccine candidates. Although in this pilot study individual purified recombinant antigens (OmpK, FKIB, Ompp1) with high antigenicity scores were shown to be highly immunogenic since mice vaccinated with 3 μg of antigens induced very strong specific IgG responses with titers >10⁴ (Fig. 4A), the current results suggest that OmpK and FKIP antigens were not synergistic and did not enhance the protective efficacy of Ompp1 in the mouse pneumonia challenge model. Therefore, in future it is worthy to evaluate more antigens from our top priority list.

In summary, this study provides a comprehensive analysis of A. baumannii genomes for vaccine development, and has identified one antigen, Ompp1 that could potentially be formulated in a novel protective vaccine against A. baumannii since pan-drug resistant A. baumannii are possibly to emerge in near future. The strategy described in the current study incorporates the advantages of reverse vaccinology, bioinformatics and immuno-proteomic platform technologies and is easy to perform to identify novel immunogen for multi-component vaccines.

Materials and Method

Ethics statement

All experiments were conducted in accordance with the guidelines of the Laboratory Animal Center of National Health Research Institutes (NHRI). Animal use protocols were reviewed and approved by the Institutional Animal Care and Use Committee of National Health Research Institutes (Approved protocol No. NHRI-IACUC-101041-A).

Bacterial strains, growth conditions and OMVs preparation

Genomes of 14 A. baumannii strains were analyzed (Table 1). A. baumannii strain ATCC17978 was grown in LB broth medium. Initial OD_600nm was set to 0.05 and bacteria were incubated at 37°C with shaking (220 rpm) until OD reached 0.4. OMVs were prepared as previously described.^8,9 Briefly, cells were centrifuged at 6,000xg for 10 minutes at 4°C and supernatant was filtered using a 0.22 um (low protein binding) PES filter (Millipore). The flow-through was concentrated using 100 kDa MWCO Amicon Ultracentrifugal concentrators (Millipore, BlossomBio, Taiwan). The precipitate containing OMVs was collected by centrifuging at 100,000xg for 1 hour at 4°C and washed 3 times with PBS. The protein concentration of supernatant was measured by spectro-photometry using a NanoDrop Technologies spectrophotometer (Wilmington, DE). The resulting pellet was submitted to proteomic analysis and the supernatant to secretomic analysis.

Proteome and secretome analyses using HPLC-MS/MS

OMVs preparations as described above were resuspended in ice-cold ethanol and centrifuged at 100,000xg for 1 hour at 4°C. The pellet was dried using a SpeedVac at low drying rate for 20 minutes. Dried pellet was resuspended in 50 μL of resuspension buffer (50 mM ammonium bicarbonate, 3 M urea, 5 mM DTT) and 5 μL of freshly prepared 50 mM ammonium bicarbonate containing 250 mM iodoacetamide were added; and the mixture was incubated for 30 minutes at room temperature in the dark. After incubation, 100 μL of 50 mM ammonium bicarbonate were added to reduce the concentration of urea to 1 M. Digestion was performed using 1 μg/μL trypsin overnight at 37°C. Protein digests were analyzed by online capillary LC-MS/MS (Waters SYNAPT™ HDMS™ System) and peptide fingerprints were analyzed using MassLynx 4.0 software and A. baumannii protein database (NCBI).

Bioinformatics tools

The orthologous core genes were determined by using Microbial Genome Database (MBGD) platform³⁴ which allowed us to do comparative analysis of completely sequenced microbial genomes. PSORTb 3.0.2 was performed to predict the sub-cellular localization of the proteins. The phylogenetic relationship of sequenced A. baumannii strains was inferred with a bootstrapped neighbor-joining analysis using the Molecular Evolutionary Genetics Analysis version 6.0 (MEGA6) based on the nucleotide sequences of 6 of the 7 housekeeping genes in the multi-locus sequence typing (MLST) of A. baumannii.^35,36

Cloning and expression of the recombinant antigens: OmpK, Ompp1and FKIB

Consensus protein sequences of OmpK, Ompp1, and FKIB were established using BlastP. Corresponding nucleotide sequences were codon optimized for expression in E. coli, and chemically synthesized (GeneArt, Life technology, Taiwan). The DNA fragments were designed to code for a 6xHis tag attached to the C-terminal of the recombinant protein and cloned into a pET-29a expression vector (Vovagen, Darmstadt, Germany) The resulting plasmids were transformed into the expression host strain E. coli BL21 (DE3) cells (GeneMark, Taiwan).Two-liter cultures were grown in Luria-Bertani (LB) broth containing 30 μg/mL of kanamycin to an optical density (OD) of 0.6 at 600 nm. Protein expression was induced using 1 mM isopropyl β-d-thiogalactopyranoside (IPTG) at 37°C for 3 h. Cells were harvested by centrifugation at 6,000 rpm for 30 minute and cell pellets were frozen and stored at −20°C.

Purification of the recombinant antigens

Recombinant proteins were purified by nickel-chelated immobilized metal affinity chromatography (IMAC). Briefly, each E. coli pellet was suspended in chilled lysis buffer (50 mM Tris-HCl, 500 mM NaCl, pH 9.0) and the cell suspension was disrupted using a French Press (Constant System, Daventry, UK) at 27 Kpsi. The cell lysate was clarified by centrifugation (32,000 rpm at 4ºC, 45 min) to collect insoluble material. The inclusion body was solubilized with buffer containing 6M urea, 300 mM NaCl and 50 mM sodium phosphate, pH 7.2. After centrifugation at 32,000 rpm at 4ºC for 45 min, the supernatant was collected and filtered through a 0.22 μm MILLEX®GP filtration membrane (Millipore Corp., Ireland) and loaded onto a 10 mL IMAC Sepharose^TM High Performance resin (GE Healthcare Bio-Sciences Corp., Sweden) equilibrated with column buffer (50 mM Sodium phosphate pH 7.4, 500 mM NaCl, 5 mM imidazole) and washed 5 × 10 mL of column buffer, then eluted with a 250–500 mM linear imidazole gradient in the column buffer. Fractions containing recombinant proteins (OmpK, Ompp1 and FKIB) were analyzed by SDS-PAGE. Single protein band fractions were pooled and dialyzed against phosphate buffer saline (PBS) and the protein concentrations were determined by bicinchoninic acid (BCA) assay (Thermo Scientific, Pierce, USA). Residual endotoxin content was determined using the Limulus amoebocyte lysate (LAL) assay (Associates of Cape Cod, Inc.., Cape Cod, MA). The final recombinant protein products were frozen and stored at −20°C for further studies.

Mouse immunogenicity studies

Groups of C57BL/6 mice (n = 4) were intramuscularly immunized with either 3 or 10 μg of individual antigen (OmpK, FKIB and Ompp1) formulated with CFA/IFA adjuvant, or PBS on day 0, 14 and 21. On day 42, blood samples were collected and tested against each immunogen. IgG antibody titers were determined using antigen-specific enzyme-linked immunosorbent assay (ELISA).

Murine pneumonia challenge model

Vaccination and murine pneumonia challenge model were performed as previously described.^11,30,31 Briefly, mice (n = 5/group) were immunized under isoflurane inhalation anesthesia with recombinant antigens (OmpK, FKIB and Ompp1 at 3 × 3 ug or 3 × 10 μg/mouse) at 2-week intervals on days 0, 14, and 28. The initial vaccination was performed subcutaneously (s.c.) with individual recombinant protein emulsified with complete Freund's adjuvant (CFA) (Sigma), whereas for the next 2 immunizations, immunogens were formulated with incomplete Freund's adjuvant (IFA) (Sigma). Mice in the control group received adjuvant in PBS (n = 5). Blood samples were collected before injections on day 0 and 40 for ELISA to detect antigen-specific immune responses. All the mice were challenged intra-tracheally (IT) on day 42 with a lethal dose (5 × 10⁷ cfu) of freshly grown ATCC 17978 strain, and mice were monitored over 7 days.

Antigen-specific ELISA

ELISA plate wells were coated with 100 ng of either OmpK, or FKIB or Ompp1 overnight, and blocked with 5% nonfat dry milk (w/v) in PBS. Mouse antisera were 2-fold serially diluted with PBS containing 1% BSA (Calbiochem, Darmstadt, Germany), and added to the wells. After incubation at room temperature (RT) for 2 hours, plates were washed 3 times with PBST (PBS containing 0.05% Tween 20), and anti-IgG isotypes (Invitrogen, Carlsbad, CA.) or anti-IgA (Invitrogen, Carlsbad, CA.) HRP-conjugated IgG (KPL, Gaithersburg, MD) specific antibodies diluted in PBS containing 1% BSA were added to the wells and incubated at RT for 1 hour. After washing with 3 × PBST, the plates were treated with TMB peroxidase substrate (KPL) at room temperature in the dark for 20 min. To determine anti-OmpK, anti-FKlB and anti-Ompp1 antibody titers, OD_450nm absorbance was measured using a spectrophotometer.

Statistical analysis

Statistical comparison of survival time per each experimental group was assessed by Kaplan-Meier (Log-Rank) test for dose-response of the survival curve. The difference was considered statistically significant when p-values were less than 0.05 (p < 0.05).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.

Acknowledgements

The authors thank Professor Michel Klein and Miss Lori Chong for their comments on the manuscript.

Funding

This work was supported by the grants from Ministry of Science and Technology (NSC 101-2321-B-400-020-MY2) and National Health Research Institutes, Taiwan, ROC.