|Home | About | Journals | Submit | Contact Us | Français|
We found that Acinetobacter baumannii contains a pgaABCD locus that encodes proteins that synthesize cell-associated poly-β-(1-6)-N-acetylglucosamine (PNAG). Both a mutant with an in-frame deletion of the pga locus (S1Δpga) and a transcomplemented strain (S1Δpga-c) of A. baumannii were constructed, and the PNAG production by these strains was compared using an immunoblot assay. Deleting the pga locus resulted in an A. baumannii strain without PNAG, and transcomplementation of the S1Δpga strain with the pgaABCD genes fully restored the wild-type PNAG phenotype. Heterologous expression of the A. baumannii pga locus in Escherichia coli led to synthesis of significant amounts of PNAG, while no polysaccharide was detected in E. coli cells harboring an empty vector. Nuclear magnetic resonance analysis of the extracellular polysaccharide material isolated from A. baumannii confirmed that it was PNAG, but notably only 60% of the glucosamine amino groups were acetylated. PCR analysis indicated that all 30 clinical A. baumannii isolates examined had the pga genes, and immunoblot assays indicated that 14 of the 30 strains strongly produced PNAG, 14 of the strains moderately to weakly produced PNAG, and 2 strains appeared to not produce PNAG. Deletion of the pga locus led to loss of the strong biofilm phenotype, which was restored by complementation. Confocal laser scanning microscopy studies combined with COMSTAT analysis demonstrated that the biovolume, mean thickness, and maximum thickness of 16-h and 48-h-old biofilms formed by wild-type and pga-complemented A. baumannii strains were significantly greater than the biovolume, mean thickness, and maximum thickness of 16-h and 48-h-old biofilms formed by the S1Δpga mutant strain. Biofilm-dependent production of PNAG could be an important virulence factor for this emerging pathogen that has few known virulence factors.
Acinetobacter baumannii is a nonfermentative, gram-negative bacillus found in many health care environments and is a very effective colonizer of humans in hospitals. Resistance to many classes of antibiotics is common among Acinetobacter spp., and the recovery of multi-drug-resistant (MDR) and pan-drug-resistant A. baumannii strains has been on the rise in the last two decades (9). A. baumannii infections tend to occur in immunosuppressed patients, often in intensive care units, in patients with an underlying illness, and in patients subjected to invasive procedures (8). A. baumannii is an increasingly common cause of ventilator-associated pneumonia, bacteremia, meningitis, and urinary tract infections (2), as well as infections of skin and soft tissues, the central nervous system, and bone (37).
Because of increasing interest in A. baumannii, progress is being made in identifying virulence determinants of this emerging pathogen, which to date include a novel pilus assembly system involved in biofilm formation (51), outer membrane protein Omp38 (4), a siderophore-mediated iron acquisition system, and an autoinducer synthase (34), among others. Analysis of the recently sequenced A. baumannii strain ATCC 17978 revealed the presence of 28 putative alien genetic islands, many of which carry genes predicted to be involved in virulence, including genes encoding drug resistance proteins, heavy metal resistance, type IV secretion systems, hemolysins/hemagglutinins, and cell wall biogenesis (47). Furthermore, using insertional mutagenesis, the virulence properties associated with six of these alien islands were confirmed in nonmammalian models of infection involving the worm Caenorhabditis elegans and the amoeba Dictyostelium discoideum (47).
A. baumannii has the ability to colonize both abiotic and medical devices (51) and form biofilms which display decreased susceptibility to antibiotics (52, 53). Biofilms are complex matrices that contain proteins, ions, nucleic acids, and polysaccharide polymers (33, 43, 48, 56). One of the important polysaccharides is poly-β-(1-6)-N-acetylglucosamine (PNAG), which has been well described as a major component of biofilms of both Staphylococcus epidermidis (28) and Staphylococcus aureus (29). In addition to its role in surface and cell-to-cell adherence (5, 28), PNAG is an important virulence factor (27, 40, 44, 45) and protects bacteria against innate host defenses (24, 54). In staphylococci the icaADBC operon encodes the proteins involved in the synthesis of PNAG (5, 14). Recently, other functionally and genetically related loci that encode proteins synthesizing a similar or identical hexosamine-rich exopolysaccharide have been described in the genomes of several other gram-negative bacteria, including Escherichia coli, Yersinia pestis, Yersinia enterocolitica, Bordetella pertussis, Bordetella parapertussis, Bordetella bronchiseptica, Burkholderia cepacia, Pseudomonas fluorescens, Actinobacillus pleuropneumoniae, and Aggregatibacter actinomycetemcomitans (7, 20, 21, 25, 36, 55), and production of PNAG in E. coli (55), A. pleuropneumoniae (20), and A. actinomycetemcomitans (21) has been confirmed. In addition, a polysaccharide similar to PNAG, termed Bps, has been shown to be synthesized by Bordetella spp. both under in vitro conditions and in a murine model of bacterial infection (36, 46).
A BLASTP search using E. coli pgaABCD and homologous loci of other gram-negative bacteria with the NCBI A. baumannii nonredundant protein database sequences enabled us to identify a four-gene locus that shares a high degree of similarly with various genetic loci encoding PNAG biosynthetic proteins. Genetic and biochemical studies described in this work demonstrated that the A. baumannii pgaABCD locus encodes proteins for the synthesis of PNAG in this organism and that these genes are present in 30 clinical isolates, most of which produce detectable PNAG polysaccharide.
Bacterial strains used in this study are shown in Table Table1.1. Cells were grown in lysogeny broth (LB) or on LB agar plates. All strains were stored at −80°C in tryptic soy broth containing 10% glycerol. The antibiotic concentrations used for selection of A. baumannii and E. coli were as follows: kanamycin, 50 μg/ml; and ampicillin, 100 μg/ml.
The colony morphology of A. baumannii strains was studied on plates containing Congo red agar composed of brain hear infusion agar (Oxoid) supplemented with 5% sucrose (American Bioanalytical) and 0.8 mg/ml of Congo red (Sigma Chemical Co., St. Louis, MO) as previously described (13). On these plates PNAG-synthesizing cells produced red colonies, whereas PNAG-deficient cells produced white colonies.
DNA manipulations were carried out using standard procedures, as previously described (1, 42). Plasmid purification was performed using a QIAprep spin miniprep kit (Qiagen, Inc). Genomic DNA was prepared using a Wizard genomic DNA purification kit from Promega Corporation (Madison, WI). Primers were custom synthesized by Operon Technologies Inc. (Alameda, CA). Enzymes were purchased from New England Biolabs Inc. (Ipswich, MA) or Invitrogen Corp. (Carlsbad, CA) and used as specified by the manufacturer. Mutations were confirmed by DNA sequencing, which was carried out by the High Throughput Sequencing Service, Brigham and Women's Hospital, Boston, MA.
We constructed a vector to generate an in-frame deletion of the A. baumannii pga locus in the following manner. The 5′ region of pgaC and the 3′ region of pgaA were amplified by PCR using the primer pairs NdeI-F-pga/BmtI-R-pga and BmtI-F-pga/XhoI-R-pga, respectively (Table (Table2).2). The fragments were then digested with NdeI/BmtI and BmtI/XhoI, respectively, and gel purified, and they were subsequently ligated between the NdeI and XhoI sites of pSSK10, an R6K-dependent sacB-containing allele exchange vector that does not replicate in A. baumannii. The resulting plasmid, pΔpga, contains 3 kb of DNA flanking regions on both sides of the pgaC and pgaA genes. S1Δpga harboring an in-frame deletion of the pga locus was made by first transforming pΔpga into A. baumannii strain S1 by electroporation. Clones in which the plasmid was integrated into the chromosome were selected on LB agar plates supplemented with kanamycin. A. baumannii S1 merodiploids were grown for 2 days in LB without antibiotics and subsequently plated on LB agar supplemented with 10% sucrose at 25°C for 48 h to select for cells that had lost the plasmid after homologous recombination. Kanamycin-sensitive, sucrose-resistant colonies were screened by PCR using primers 2163F and 2159R to confirm deletion of pgaABC. Deletion of the pgaABC genes was confirmed by sequencing using primers Pga-1F and Pga-1R. The resulting strain contains the first 3 amino acids of the pgaA product in frame with the last 3 amino acids of the pgaC product.
Plasmid pBAD-Pga was constructed as follows. We first generated pBAD18kan-ori, a derivative of pBAD18Kan (12) that is able to replicate in A. baumannii. A 1,337-bp PCR product containing the replication origin of cryptic plasmid pWH1277 from Acinetobacter lwoffii DSM30013 (10) was generated using the primer pair BbvCI-Ori-F/BbvCI-Ori-R (Table (Table2).2). This fragment was then ligated into the BbvCI site of pBAD18kan. The pgaABCD locus was amplified by PCR with the Phusion high-fidelity enzyme (New England Biolabs Inc., Ipswich, MA) from the chromosomal DNA of A. baumannii S1 using primers Pga-comp-F and Pga-comp-R. The resulting 6,741-bp DNA fragment was cloned into pCR-XL-TOPO (Invitrogen) to generate plasmid pPga. The entire pgaABCD locus was then excised from pPga with XhoI and KpnI and ligated between the XhoI and KpnI sites of pBAD18kan-Ori. pBAD-Pga was then transformed into the A. baumannii S1 strain with the pga locus deleted by electroporation, generating strain S1Δpga-c.
Immunoblotting to detect PNAG was performed essentially as described previously (6, 23), with minor modifications. Cells were grown to stationary phase in LB and diluted in fresh LB to obtain an optical density at 650 nm of 1. Cells were pelleted, resuspended in 300 μl of 0.5 M EDTA (pH 8.0), boiled for 5 min, and centrifuged at 9,000 × g for 5 min. Supernatants were diluted 1:5 in Tris-buffered saline (TBS) and incubated with 100 μl of proteinase K (20 mg/ml; Qiagen) for 60 min at 65°C and then for 30 min at 80°C to inactivate the protease. Extracts were serially diluted in TBS, and 200 μl of each dilution was immobilized on a nitrocellulose filter attached to a slot blot vacuum manifold. The membrane was blocked with a solution of 1% bovine serum albumin (BSA) in TBS and then incubated for 30 min with a 1:1,000 dilution of goat antibody raised to the deacetylated PNAG (dPNAG) glycoform (~15% acetylation) conjugated to a diphtheria toxoid carrier and produced as described previously (27). Horseradish peroxidase-conjugated swine anti-goat immunoglobulin G (IgG) antibody (Southern Biotech, Birmingham, AL) diluted 1:10,000 in 1% skim milk in TBS and the enhanced chemiluminescence Western blotting system (Amersham, Piscataway, NJ) were used to detect the bound primary antibodies.
Confirmation that the extracts contained a β-1-6-linked glucosamine antigen was obtained by digesting the extracts with dispersin B (50 μg/ml in TBS, pH 7.5; Kane Biotech, Inc., Winnipeg, Canada) at 37°C for 1 h. Dispersin B has been identified as an endoglycosidase specific to β-1-6-linked poly-glucosamine molecules (39).
A. baumannii S1 and S1Δpga and the transcomplemented strain S1Δpga-c were grown on 13-mm round Thermanox coverslips (Electron Microscopy Sciences, Fort Washington, PA) for analysis by scanning electron microscopy (SEM). After overnight incubation at 37°C in LB under static conditions to promote formation of biofilms, the coverslips were carefully rinsed three times with phosphate-buffered saline (PBS), and the cells were fixed with 2.5% (vol/vol) glutaraldehyde (Electron Microscopy Sciences) in 0.1 M cacodylate buffer (Electron Microscopy Sciences) for 1 h. This was followed by a series of sequential ethanol dehydration steps (50%, 70%, 95%, and 100% ethanol; 10 min each) before samples were dried with 50% hexamethyldisilazane (Electron Microscopy Sciences) in 100% ethanol for 2 h and then with 100% hexamethyldisilazane twice for 30 min. Samples were air dried overnight and were sputter coated with a gold-palladium target in a Hummer V apparatus (Anatech, Alexandria, VA). The samples were examined with a 1450VP scanning electron microscope (Carl Zeiss, SMT, Inc., Thornwood, NY) with a LaB6 at 15 kV.
The bacterial strains were grown overnight in LB. Electron microscope grids (200-mesh Formvar-carbon-coated copper grids; Electron Microscopy Sciences) were placed on a 25-μl drop of a bacterial suspension for 15 min. The grids were removed and placed on a 10-μl drop of 1% BSA and 10% guinea pig serum in PBS (blocking buffer) for 30 min. After blocking, the grids were placed on a 25-μl drop of immune rabbit antiserum raised against dPNAG conjugated to diphtheria toxoid (diluted 1:25 in 1% BSA in PBS) for 15 min. Then the grids were washed three times for 5 min with PBS. The secondary antibody was then applied for 15 min. For all of the strains, the secondary antibody used was a 25-μl drop of 12-nm colloidal gold-labeled donkey anti-rabbit IgG(H+L) (Jackson ImmunoResearch) at a 1:10 dilution in 1% BSA in PBS. Each grid was then washed three times for 2 min in PBS and then twice in distilled water, and photographs were taken at a magnification of ×13,000.
The entire pga locus of A. baumannii S1 was amplified by PCR using primers Pga-comp-F and Pga-comp-R, which were designed based on the sequence of strain A. baumannii ATCC 17978. This PCR yielded a 6.7-kb amplicon that was subsequently sequenced by the High Throughput Sequencing Service, Brigham and Women's Hospital, Boston, MA.
Overnight cultures of A. baumannii were diluted 1:200 in 2 ml of fresh LB supplemented with 1% glucose and grown in 10-ml borosilicate glass culture tubes with vigorous shaking (270 rpm) for 5 h. To assay the attachment to glass, biofilms were first gently washed three times with 5 ml of PBS to remove loosely attached cells, dried, and stained for 15 min with a 0.2% crystal violet solution. Subsequently, the biofilms were rinsed with deionized water, dried, and photographed. Biomass was quantified by adding 2 ml of 100% ethanol to stained tubes containing 1 g of 1-mm glass beads. The tubes were then vortexed until the stained biomass was completely removed from the tube surface, and the optical density at 595 nm was determined (35).
Confocal laser scanning microscopy (CLSM) was used to evaluate the three-dimensional biofilm structure. Biofilms were grown on glass coverslips partially immersed in a 50-ml Falcon tube containing 10 ml of LB for S1 and S1Δpga and LB supplemented with 50 μg/ml of kanamycin for S1Δpga-c. Portions (50 μl) of overnight cultures were used to inoculate the cultures, and biofilms were allowed to form over a 16-h period. Biofilms were also grown for 48 h. In this case 24 h after inoculation the spent culture was replaced with fresh medium, and biofilms were allowed to form for another day. 4′,6-Diamidino-2-phenylindole (DAPI), which penetrates bacterial cell membranes to form a fluorescent complex with DNA, was used to visualize biofilms. After coverslips were rinsed three times with PBS, biofilm samples were incubated with DAPI (1 μg/ml in PBS) for 10 min. Excess DAPI was then removed by extensive rinsing with PBS before biofilms were imaged by CLSM as described below.
All confocal images were obtained at the Harvard Center for Neurodegeneration and Repair, Boston, MA. Biofilms were observed using a ×63 water immersion objective and a Zeiss 510 confocal microscope with a Coherent Chameleon pulse laser with the wavelength set at 760 nm and the emission filter set at 390 to 465 nm for the DAPI signal. For each biofilm five image stacks were taken at different locations throughout the biofilm, using 1-μm z-step increments. Experiments were done in duplicate. Images were digitally reconstructed with Image J software.
The biofilm structure was quantified using the confocal z stacks and the image analysis software package COMSTAT (Technical University of Denmark, Lyngby, Denmark) (15). In this study, three COMSTAT parameters were utilized to determine the differences between biofilms of A. baumannii S1, S1Δpga, and S1Δpga-c grown for 16 and 48 h. These parameters were biovolume, maximum biofilm thickness, and average biofilm thickness.
All of the restriction enzyme sequences were collected from the REBASE database (http://rebase.neb.com/rebase/rebase.html). Homology searches were performed using BLAST (http://www.ncbi.nih.gov/BLAST), and alignments of amino acid sequences were constructed using the ClustalW software (http://www.ch.embnet.org/software/ClustalW.html). The predicted open reading frame (ORF) nucleotide sequences were translated into amino acid sequences in reading frame 1 with the Transeq program from the EMBOSS suite of molecular biology applications (http://www.emboss.org/). Promoter prediction was done with BPROM (http://www.softberry.com/all.htm) (Softberry, Inc., Mt. Kisco, NY).
PNAG was prepared from a 6-liter culture of A. baumannii S1Δpga-c grown in LB containing 1% glucose inoculated with a single colony and incubated at 37°C for 72 h with continuous stirring. Bacterial cells were sedimented by centrifugation at 9,000 × g for 15 min and suspended in 100 ml of 20 mM Tris-HCl, 5 mM EDTA buffer (pH 8.0) containing lysozyme (500 mg), and the cell suspension was incubated at room temperature for 30 min. Then DNase I (25 mg) and RNase A (100 mg) were added, and the suspension was incubated at room temperature for 1 h and then at 37°C for 2 h. After cells were removed by centrifugation, the supernatant was precipitated with 3 volumes of ethanol, and the insoluble material was collected by centrifugation at 9,000 × g for 15 min. The ethanol-insoluble extract was suspended in water, dialyzed overnight against water, and freeze-dried.
The spectra of PNAG samples were recorded at 298 K in deuterium oxide using a Varian VNMRS-600 nuclear magnetic resonance (NMR) spectrometer. Monodeuterated water (HOD) (4.80 ppm) was used as the chemical shift reference in NMR experiments.
The statistical significance (P value) of biofilm formation on borosilicate glass tubes and data obtained using the COMSTAT software was determined by performing an unpaired, two-tailed t test using GraphPad Prism version 4.0 (GraphPad Software, San Diego, CA).
The complete genome sequence and annotation of A. baumannii ATCC 17978 has been deposited in the GenBank/EMBL/DDBJ database under accession no. CP000521. The nucleotide sequences and predicted amino acid sequences for the pgaABCD locus of A. baumannii strain S1 have been deposited in the NCBI GenBank database under the following accession numbers: pgaA, 1201582; pgaB, 1201614; pgaC, 1201625; and pgaD, 1201631.
We used the predicted amino acid sequences encoded by the E. coli pgaABCD genes and homologous loci in other gram-negative bacteria to search for homologs of these loci in all of the A. baumannii genome sequences available in the NCBI database using the BLASTP program. We identified four ORFs that encode proteins with high levels of similarity to the E. coli proteins in all of the sequenced strains of A. baumannii in the database, including A. baumannii strain ATCC 17978. In this strain, this locus is located between coordinates 2,518,776 and 2,524,376 and contains the genes A1S_2162, A1S_2161, and A1S_2160. We designated these genes pgaA, pgaB, and pgaC, respectively.
A. baumannii pgaA encodes a predicted 812-amino-acid outer membrane protein that shares 26% identity with E. coli PgaA, 23% identity with Y. pestis HmsH, and 25% identity with B. pertussis BpsA. In E. coli, PgaA contains a predicted porin domain that facilitates PNAG translocation across the outer membrane, as well as a superhelical periplasmic domain that is thought to play a role in protein-protein interactions (18). In Y. pestis HmsH has been demonstrated to be an outer membrane protein (38).
pgaB encodes a predicted 510-amino-acid protein with a putative polysaccharide deacetylase domain. It shares 33%, 31%, and 37% identity with E. coli PgaB, Y. pestis HmsF, and B. pertussis BpsB, respectively. All these proteins have putative deacetylase activities. In Y. pestis, HmsF was isolated from the outer membrane by cellular fractionation (38). E. coli PgaB is predicted to be an outer membrane lipoprotein that, along with PgaA, is necessary for PNAG export (18).
pgaC is predicted to encode a 392-amino-acid N-glycosyltransferase that belongs to the glycosyltransferase 2 family. This family includes PgaC, BpsC, HmsR, and IcaA from E. coli, B. pertussis, Y. pestis, and S. aureus, respectively. A. baumannii PgaC shares 55% identity with E. coli PgaC, 47% identity with B. pertussis BpsC, and 55% identity with Y. pestis HmsR. In addition, the five amino acids that have been shown to be critical for the catalytic activity of glycosyltransferases are all conserved in PgaC (Asp112, Asp205, Gln241, Arg244, and Trp245).
In addition to the three annotated genes that belong to the pga locus in A. baumannii ATCC 17978, we identified a small ORF downstream from pgaC that was not annotated in the sequenced genomes. The protein encoded by this ORF is 150 amino acids long and has 32% identity with E. coli PgaD and Y. pestis HmsS. We therefore designated this ORF pgaD. We noted that pgaD is also present in other sequenced strains of A. baumannii, including ACICU, AYE, and AB0057. No functional domain was found in A. baumannii PgaD; however, this protein localizes to the cytoplasm and appears to assist the glycosyltransferase PgaC in the synthesis of PNAG in E. coli (18). Similarly, it has also been shown that in S. aureus IcaD, a PgaD homolog, is a cytoplasmic protein that augments the activity of IcaA, a PgaC homolog (10).
Considering the homology between the pgaABCD locus in A. baumannii and similar loci, such as pgaABCD in E. coli, hmsHFRS in Y. pestis, or bpsABCD in B. pertussis among others, we hypothesized that this locus is responsible for the synthesis of PNAG in A. baumannii.
We screened our A. baumannii collection for isolates that are susceptible to several antibiotics, including kanamycin, tetracycline, and gentamicin, and that would be amenable to genetic manipulation. From the drug-sensitive isolates we selected a clinical strain, S1, as it had a highly mucoid phenotype, which has been associated with PNAG production (32).
Sequence analysis of the pga locus in A. baumannii strain S1 revealed that the predicted proteins encoded by this locus shared 98%, 98%, 99%, and 98% identity with the A. baumannii ATCC 17978 PgaA, PgaB, PgaC, and PgaD proteins, respectively.
To determine if the pga locus is responsible for PNAG production in A. baumannii, we investigated production of this polysaccharide in strain S1Δpga, a derivative of strain S1 that contains a deletion of the pgaABC genes. We compared the PNAG levels by immunoblotting using specific antibodies to dPNAG in wild-type strain S1 and in S1Δpga. PNAG was detected in samples diluted 2,000- to 8,000-fold before they were blotted onto the membrane for the wild-type strain in order to avoid saturation of the signal (Fig. (Fig.1A,1A, first lane). In contrast, we could not detect PNAG-specific signals even when very high concentrations of the extracts of the ΔpgaS1 mutant were used (Fig. (Fig.1A,1A, third lane), confirming that this strain does not produce PNAG. In order to verify that the PNAG-negative phenotype was due to deletion of the pga locus, we complemented S1Δpga in trans with pBAD-Pga to produce strain S1Δpga-c. This plasmid carries the pgaABCD genes under control of their native promoters. Complementation fully restored PNAG production to levels comparable to those seen in the wild-type S1 strain (Fig. (Fig.1A,1A, fourth lane). Treatment of the extracts of the wild-type and S1Δpga-c strains with dispersin B, a PNAG-hydrolyzing enzyme, eliminated the reaction of the antibodies to PNAG with the extracts (Fig. (Fig.1A,1A, second and fifth lanes).
The morphology of the wild-type, S1Δpga, and S1Δpga-c strains was investigated using plates supplemented with Congo red (CRP), a dye known to bind to bacterial polysaccharides (50). As shown in Fig. Fig.1B,1B, wild-type A. baumannii colonies were red on Congo red plates, whereas the S1Δpga strain formed white colonies. Complementation of the Δpga mutation restored the colony color to that of the parental S1 strain.
To confirm that the A. baumannii pgaABCD genes are sufficient for PNAG production, we analyzed whether PNAG could be synthesized by E. coli carrying the A. baumannii pga genes. There was no detectable PNAG production in E. coli Top10 cells harboring cloning plasmid pCR-XL-TOPO (Fig. (Fig.1C,1C, first lane). However, when the A. baumannii pgaABCD genes were inserted into this plasmid to create plasmid pPga, production of PNAG was readily detectable (Fig. (Fig.1C,1C, third lane). Extracts of E. coli(pPga) cells did not exhibit an immunologic reaction with antibody to dPNAG after dispersin B treatment (Fig. (Fig.1C,1C, fourth lane). These results show that the A. baumannii pgaABCD genes are sufficient for PNAG production, as determined by detecting polysaccharide synthesis in a heterologous host.
Additional phenotypic analysis of the A. baumannii S1, S1Δpga, and S1Δpga-c strains was carried out using SEM analysis of cells grown overnight as biofilms on Thermonox coverslips. The SEM images revealed that wild-type S1 cells are interconnected by an extracellular mesh of stringy material (Fig. 2a and b). This mesh was completely absent in the biofilm formed by the S1Δpga cells (Fig. 2c and d). Complementation of the S1Δpga mutant with an intact pga locus restored the intercellular network of fibers seen in the wild-type strain biofilms (Fig. 2e and f). These results are in agreement with the macroscopic observations that both the wild-type A. baumannii S1 and complemented S1Δpga-c strains have an obvious mucoid phenotype that is absent in the S1Δpga mutant.
Further confirmation that the pgaABCD encodes synthesis of PNAG in A. baumannii was obtained by immunoelectron microscopy using goat antibodies specific to dPNAG and gold-conjugated secondary antibodies. As shown in Fig. Fig.3,3, while heavy surface immunogold labeling was detected on the surface of the wild-type strain A. baumannii S1 (Fig. 3a and b), very few gold particles were visible on the surface of the S1Δpga strain (Fig. 3c and d). Finally, analysis of the complemented A. baumannii S1Δpga-c strain revealed a dense gold-stained pattern similar to that observed for the A. baumannii S1 wild-type strain (Fig. 3e and f).
In order to investigate the role of the pgaABCD locus in biofilm formation by A. baumannii, we first compared the biofilm-forming abilities of A. baumannii wild-type strain S1, the S1Δpga mutant, and the S1Δpga-c strain in polystyrene tissue culture wells using LB supplemented with 1% glucose at either 37 or 30°C under static conditions. Although PNAG synthesis is associated with biofilm production in a number of bacterial species (11, 14, 21, 55), we did not observe significant differences in the patterns of biofilm formation between wild-type strain S1 and strain S1Δpga under static growth conditions.
We also tested biofilm formation under more dynamic conditions in borosilicate glass tubes. Our results (Fig. (Fig.4A4A and and4B)4B) show that under very vigorous shaking conditions A. baumannii wild-type strain S1 forms strong doughnut-shaped rings of adherent cells, which are significantly less visible in the S1Δpga mutant cultures. Complementation of the biofilm-negative S1Δpga strain with an intact pgaABCD locus restored the biofilm-positive phenotype of the A. baumannii S1 strain.
CLSM was used to examine the morphological features of biofilms formed by A. baumannii strains S1, S1Δpga, and S1Δpga-c after 16 h and 48 h of growth. Biofilms were grown on glass coverslips and viewed with a ×63 water immersion objective that allowed the samples to remain hydrated and observed in situ. As Fig. Fig.55 (top panels) shows, after 16 h of growth A. baumannii wild-type strain S1 formed a multilayer three-dimensional biofilm at the liquid-air interface of the glass coverslip. Conversely, A. baumannii S1Δpga adhered poorly to a coverslip and only partially covered the glass surface with a monolayer of cells. Complementation of S1Δpga with an intact pga locus fully restored the biofilm architecture to that of the wild-type S1 strain. CLMS analysis of 2-day-old biofilms (Fig. (Fig.5,5, lower panels) revealed that while wild-type strain S1 biofilms exhibited the same three-dimensional structure that was observed at 16 h, A. baumannii S1Δpga grew as a flat monolayer of cells covering the entire glass surface but did not develop any pillar-like structures typical of biofilms. A. baumannii S1Δpga-c also exhibited the three-dimensional structure after 2 days of growth.
To confirm the CLSM observations of biofilm structure, the COMSTAT image analysis software was used to evaluate three variables of biofilm architecture: biovolume, average thickness, and maximum thickness. The biovolume was calculated by normalizing the volume of the biofilm (μm3) by the surface area of the field of view (μm2), which gave the biovolume parameter biofilm volume per unit of surface area (μm3/μm2). The average thickness is based on determinations from the entire view field. The maximum biofilm thickness reflects the maximum distance from the substratum that the biofilm reaches. As shown in Table Table3,3, COMSTAT analysis revealed important differences in the biofilm architecture between the three strains. Biofilms formed by both the wild-type S1 and complemented S1Δpga-c strains had significantly greater biomass, average thickness, and maximum thickness than biofilms formed by the S1Δpga strain after both 16 and 48 h of growth (P < 0.001).
Another interesting observation was that while two of the parameters investigated by COMSTAT analysis, biovolume and average thickness, increased for A. baumannii S1 and S1Δpga biofilms from 16 h to 48 h, moderate decreases were observed for the complemented S1Δpga-c strain biofilms.
Overall, the use of COMSTAT analysis to quantify the biofilms observed in the CLSM image stacks confirmed our previous observation that PNAG production is critical for the three-dimensional architecture of biofilms, as demonstrated by the thick and voluminous biofilms produced by the wild-type and complemented S1 strains compared to the relatively flat, uniform mat of bacterial cells produced by the PNAG-deficient S1Δpga strain.
In order to obtain insight into the distribution of the pgaABCD gene cluster among A. baumannii clinical isolates, we screened 30 MDR A. baumannii isolates (a kind gift from Tse Hsien Koh) by performing PCR using primers based on A. baumannii sequenced strain ATCC 17978. We could PCR amplify a product that was the same size as the pgaABCD locus from all 30 strains tested (data not shown). We then tested the 30 strains to determine if they were capable of producing cell-associated PNAG under laboratory growth conditions, using A. baumannii strains S1 and S1Δpga as positive and negative controls in an immunoblot assay and antibodies specific to PNAG. Figure Figure66 shows that 14 strains produced high levels of PNAG, 14 strains produced low to moderate levels of PNAG, and 2 strains did not synthesize detectable PNAG. This variable synthesis of PNAG might reflect differences in gene regulation and/or mutations in the pga loci among these clinical isolates.
PNAG was extracted from transcomplemented A. baumannii S1Δpga-c cells using enzymatic treatment with lysozyme. Polysaccharide extracts were analyzed by 1H NMR and 1H-1H COSY NMR. Assignment of each nonexchangeable proton was made using COSY, and the chemical shifts are shown in Fig. Fig.7.7. The chemical shift values of the polysaccharide samples are in complete agreement with those obtained previously by us and other groups for PNAG isolated from S. aureus, E. coli, and A. pleuropneumoniae (20, 29, 55). Our NMR analysis revealed that PNAG contains 60% N-acetylglucosamine residues and 40% glucosamine residues, as determined by comparing the integration of the area under the curves for the GlcNAc-H1 peak with that of the GlcNH2-H1 peak (or with that of the GlcNH2-H2 peak). The ratio of the integrals showed that ~60% of the glucosamine residues contained N-acetyl groups.
We identified a cluster of four genes, pgaABCD, in A. baumannii that share homology with the pga locus in E. coli and other related genetic loci encoding enzymes for synthesis of PNAG, including bps in B. pertussis and the hms locus in Y. pestis (16, 22, 25, 36, 55). In-frame deletion of the pgaABC genes in A. baumannii resulted in an A. baumannii mutant defective for PNAG production. This PNAG-negative phenotype was fully reversed when the S1Δpga mutant strain was complemented in trans with the intact pgaABCD genes. Furthermore, heterologous expression of PNAG in E. coli was obtained when the pga locus from A. baumannii was transformed into an otherwise PNAG-negative E. coli strain. These results clearly imply that the pgaABCD genes encode enzymes that synthesize PNAG.
Additional confirmation that A. baumannii synthesizes PNAG was obtained by showing that the enzyme dispersin B destroyed the reactivity of cellular extracts of wild-type A. baumannii and the complemented S1Δpga-c strain, as well as the immune reactivity of extracts of E. coli containing the cloned A. baumannii pga genes. Dispersin B is a β-(1-6) hexosaminidase that has been shown to specifically cleave the β-(1-6) linkage of glucosamine and depolymerize PNAG (19). Full confirmation that the pga locus in A. baumannii encodes proteins for PNAG synthesis came from isolating polysaccharide material from the transcomplemented S1Δpga-c strain that, as determined by NMR analysis, was β-1-6-poly-N-acetylglucosamine.
PNAG isolated from A. baumannii strain S1Δpga-c had a 40:60 ratio of GlcNH2 to GlcNAc. This level of deacetylation is substantially higher than the levels of deacetylation of PNAG isolated from S. aureus strain MN8m previously determined by us (3 to 8% deacetylation) or the levels reported by other workers for PNAG-like molecules, such as polyacetylglucosamine in E. coli (3% deacetylation) or the major component (polysaccharide I) of S. epidermidis polysaccharide intercellular adhesin, in which 15 to 20% of the amino groups were deacetylated (28, 29, 55). While our knowledge of the real acetylation levels in all of the multiple PNAG-producing species is limited at this point, apart from data for four individual strains belonging to four different bacterial genera, differences in the regulation of the pga-like loci and especially the pgaB-encoded deacetylase could explain the acetylation levels observed for PNAG isolated from A. baumannii S1Δpga-c.
The ability of many pathogens to adhere to human tissues and medical devices and produce biofilms is a major virulence factor that correlates with increases in antibiotic resistance, reduced phagocytosis, and overall persistence of the bacterial population. Of all of the different molecules identified as biofilm components, PNAG is perhaps one of the most important and widely conserved factors produced by different bacterial species. Although in the present work we did not find significant differences in biofilm formation rates between wild-type strain S1 and its PNAG-negative counterpart under static conditions, when cultures were grown in glass tubes with vigorous shaking, very striking differences in biofilm formation were observed. We speculate that in a more dynamic environment with higher shear forces, PNAG is more essential for maintaining the integrity of A. baumannii biofilms.
The finding that A. baumannii is another human pathogen that synthesizes PNAG further broadens the spectrum of the potential effectiveness of a PNAG-based vaccine. Indeed, previous work carried out by our group has shown the effectiveness in animal studies of PNAG as a candidate vaccine against two important human pathogens, S. aureus (30-32) and E. coli (3). The recent global expansion of MDR and pan-resistant clones of A. baumannii has often resulted in situations where there are very few effective antibiotics that can be used (26), suggesting that active or passive immunotherapy targeting PNAG might be an option for countering the growing problem of infections with MDR A. baumannii. One general requirement for any potentially useful vaccine candidate is that its target antigen needs to be widely expressed in human clinical isolates. In this work we showed that both the occurrence of the pga locus and the synthesis of the surface-associated PNAG appear to be common in A. baumannii clinical isolates. All 30 strains that we tested had a chromosomal locus that could be amplified by PCR using pga-specific primers that yielded a fragment that was the same size as the pga locus of the sequenced A. baumannii ATCC 17978 strain. Approximately one half of the clinical isolates synthesized significant amounts of PNAG, and the other half produced smaller amounts of this surface antigen. Efforts are already under way to further investigate the potential of PNAG as a candidate vaccine against A. baumannii.
Overall, our genetic, immunologic, and chemical analysis provided conclusive evidence that many isolates of A. baumannii can synthesize PNAG. As this surface polysaccharide is a known virulence factor in various staphylococcal infections (27, 40, 41, 44, 45) and is a target for active and passive vaccination approaches (29-32), PNAG production probably has a role in the pathogenesis of A. baumannii infection. If additional studies also determine that PNAG is a good vaccine target in A. baumannii, there is some possibility that efforts under way to develop PNAG-specific immunotherapies might also be useful for A. baumannii infections. This could ultimately provide an effective tool for prevention or therapy of infections caused by this emerging pathogen that already presents a significant challenge for modern antibiotic chemotherapy for serious infections.
We thank Silvia A. Pineiro for providing the suicide vector pSSK10 and Tse Hsien Koh for providing a collection of MDR clinical isolates of A. baumannii. We also acknowledge the contributions to this work of Dennis Benedetti and Jennifer O'Malley and of Lai Ding (Optical Imaging Core Facility, Harvard Medical School) for his help with the CLSM studies.
This work was supported by NIH grants AI 046706 (G.B.P.) and AI 073586 (T.M.-L.).
G.B.P. and T.M.-L. declare financial interests in development of PNAG-based vaccines and immunotherapies via licensing and consulting income, and G.B.P. has equity interest in a company developing PNAG-specific passive immunotherapies.
Published ahead of print on 24 July 2009.