Acinetobacter baumannii is an important nosocomial pathogen that displays high antibiotic resistance. It causes a variety of infections including pneumonias and sepsis which may result in disseminated intravascular coagulation. In this work, we identify and characterize a novel secreted, zinc-dependent, metallo-endopeptidase CpaA (coagulation targeting metallo-endopeptidase of Acinetobacter baumannii) which deregulates human blood coagulation in vitro and thus is likely to contribute to A. baumannii virulence. Three quarters of the clinical isolates tested (n = 16) had the cpaA gene; however, it was absent from two type strains, A. baumannii ATCC 17978 and A. baumannii ATCC 19606. The CpaA protein was purified from one clinical isolate and was able to cleave purified factor (F) V and fibrinogen and reduce the coagulation activity of FV in human plasma. CpaA-treated plasma showed reduced clotting activity in contact pathway-activated partial thromboplastin time (aPTT) assays, but increased clotting activity in tissue factor pathway prothrombin time (PT) assays. A significant portion of clinically relevant A. baumannii isolates secrete a protease which targets and deregulates the coagulation system.
Acinetobacter baumannii has become an important human pathogen due mainly to the emergence of multidrug resistant (MDR) and pan-drug resistant (PDR) strains, which routinely display resistance to every single class of antibiotic currently in clinical use (Gordon & Wareham, 2010). A recent CDC report attributed, over 12 000 healthcare-associated infections to the Acinetobacter genus, of which nearly 7000 were resistant to at least three classes of antibiotics, resulting in c. 500 deaths per annum in the United States alone (CDC, 2013). Acinetobacter baumannii is of particular concern among the critically ill; A. baumannii is responsible for 2% of healthcare-associated infections. However, Acinetobacter baumannii may account for up to 10% of intensive care unit (ICU) infections (Vincent et al., 2009). A. baumannii primarily causes pneumonias and bloodstream infections, although urinary tract and soft tissue infections are also reported (Joly-Guillou, 2005).
Research investigating how A. baumannii tolerates antimicrobial therapy is prevalent; however, relatively little is known regarding the virulence factors which may contribute to infection. Acinetobacter baumannii is capable of hemolytic activity and can lyse mammalian cells (Tayabali et al., 2012). A number of factors have been identified which contribute to A. baumannii growth in human serum, which is thought to be secondary to inactivation of the complement system, including OmpA (Choi et al., 2005), Ptk, EpsA (Russo et al., 2011), and PbpG (Russo et al., 2009). A. baumannii is known to produce outer membrane vesicles (Jin et al., 2011), and despite reports of the genus having no secreted protease activity (Bitrain et al., 2012), some clinical isolates (including strains from this study) have been shown to secrete active proteases which may contribute to virulence, such as the secreted serine protease PKF, which is associated with inhibition of biofilm formation and enhanced growth in human serum (King et al., 2013).
Acinetobacter baumannii bacteremia often originates in the respiratory tract and migrates to the bloodstream where it is associated with very high mortality rates due at least in part to septic shock or disseminated intravascular coagulation (DIC) (Leung et al., 2006). Patients suffering from A. baumannii sepsis have been shown to have abnormal coagulation-related proteins (Soares et al., 2009), which are critical to antimicrobial defense (Delvaeye & Conway, 2009; Massberg et al., 2010) and a target of numerous Gram-negative virulence factors (Imamura et al., 1995, 2008; Brunder et al., 1997; Suomalainen et al., 2007). To our knowledge, no studies have examined deregulation of coagulation as a means of A. baumannii virulence. Here, we purify, identify, and characterize CpaA (coagulation targeting metallo-endopeptidase of Acinetobacter baumannii), a novel reprolysin-like secreted zinc-dependent metallo-endopeptidase present in A. baumannii clinical isolates, but not ATCC 19606 or ATCC 17978 reference strains. We evaluate its prevalence among A. baumannii clinical isolates and identify several aspects of human coagulation which are affected by the protease.
Materials and methods
Bacterial strains, plasmids, and growth conditions
All bacterial strains and plasmids are listed in Table 1. Escherichia coli and A. baumannii strains were propagated in lysogeny broth (LB) at 37 °C with shaking (200 r.p.m.) unless stated otherwise. Antibiotic concentrations were 50 μg mL−1 Gentamicin (Gm) (for both A. baumannii and E. coli) (Bioshop, Burlington, ON, Canada) and 100 μg mL−1 Ampicillin (Amp) (Bioshop) (for E. coli).
Plasmid extraction was conducted according to Sambrook & Russell, 2001. PCR was performed using Phusion Taq polymerase (New England Biolabs, Pickering, ON, Canada) according to the manufacturer's instructions.
Protein and DNA sequencing
Protein and DNA sequencing was conducted at The Hospital for Sick Children Advanced Protein Technology Centre and The Centre for Applied Genomics, respectively (Toronto, ON, Canada). For identification and molecular weight estimation (using bottom-up approach), CpaA was excised from a Coomassie Brilliant Blue stained SDS-PAGE gel and digested in-gel according to the Advanced Protein Technology Centre's in-gel tryptic digest procedure followed by LCMS/MS on Thermo LTQ-Orbitrap XL according to the Advanced Protein Technology Centre's standard procedure.
Native CpaA purification
An overnight culture of A. baumannii AB031 was subcultured (1 : 100) and grown in 0.25x LB (to limit endogenous peptides) until the culture reached an A600 nm of 0.45. Cells were removed by centrifugation, and the resulting supernatant was filtered through a 0.22-μm filter. Cell-free culture supernatant was concentrated 40-fold by tangential flow filtration (Pall Scientific, Mississauga, ON, Canada) with a 30-kDa molecular weight cut-off (MWCO) filter and dialyzed exhaustively against 20 mM HEPES, 150 mM NaCl, pH 7.4 (HBS). The dialyzed material was passed over 12 mL (2 × 6 mL) of Q-Sepharose (GE Healthcare, Baie-D'Urfé, QC, Canada) to remove protein and nucleic acid contaminants (Supporting Information, Fig. S2a). The unbound material was concentrated 20-fold using centrifugal flow filtration (CFF) with a 10 kDa MWCO filter (Millipore, Etobicoke, ON, Canada) according to the manufacturer's instructions. The concentrated material was fractionated using a 30-mL Sephadex G-100 gel filtration column (Pharmacia, Uppsala, Sweden) and eluted in HBS. Fractions containing purified CpaA were pooled and concentrated 15-fold using CFF with a 10 kDa MWCO filter according to the manufacturer's instructions (Millipore) to yield a 1-mg mL−1 solution of purified CpaA. Purity was assessed by SDS-PAGE as described elsewhere (Laemmli, 1970).
Cell-free culture supernatant was concentrated 500-fold with a 10 kDa MWCO filter to highlight contaminants within the starting material (Fig. S2b), although this was not part of the purification procedure. SDS-PAGE with Coomassie Brilliant Blue staining shows all purification steps, including the purified CpaA protein (Fig. S2c).
CpaA was mixed with pooled human reference plasma (Precision Biologic, Dartmouth, NS, Canada) or purified human FV for 30 min at room temperature. The reaction mixture was diluted so that 0.25 μL of plasma or 0.165 ng purified FV was loaded per well. SDS-PAGE and Western blotting were performed as described elsewhere (Samis et al., 2007).
Factor V activity assay
This assay has been described in detail elsewhere (Tilley et al., 2012). In brief, equal volumes of CpaA-containing samples were mixed with normal human plasma for 30 min at room temperature and were diluted 20-fold with HBS. Samples were added to equal volumes of FV deficient plasma (Bloom et al., 1979) and thromboplastin (Trinity Biotech, Wicklow, Ireland). Finally, 25 mM CaCl2 was added (6.25 mM final) to initiate clotting.
Generation of cpaA deletion mutant
cpaA gene function was disrupted using homologous recombination techniques described previously (Choi & Schweizer, 2005) with minor modifications. Briefly, the 5′ and 3′ regions of cpaA were amplified using primers A1/A2 (1–500 bp) and B1/B2 (1283–1782 bp), respectively (Table 2). The gentamicin-resistant cassette (aacC1) was amplified from the pUC18T-mini-Tn7T-Gm-LAC plasmid (Choi et al., 2005) using primers C1 and C2. The assembly of the 5′- and 3′-ends of the cpaA gene with the gentamicin-resistant cassette was carried out by Gibson assembly described elsewhere (Gibson et al., 2009). The cpaA disrupted fragment (2072 bp) was amplified using primers A1 and B2, gel extracted, and digested with BamHI (New England Biolabs; sites engineered in primers A1 and B2) and cloned into pUC18 to derive the plasmid pGM1. The presence of the cpaA/Gm fragment in pGM1 was confirmed by sequencing. To clone the cpaA fragment into the suicide plasmid pKNG101, pGM1 was digested with BamHI, and a 2026-bp fragment was purified and cloned into pKNG101 to derive the plasmid pK1. pK1 was introduced into AB031 via a tri-parental mating strategy with CC118λpir/pK1 as the donor and HB101/pRK2013 as the helper strain, using the technique described previously (Kumar et al., 2010). Transconjugants were selected on Gen50Amp32 to select for AB031 (and counter select E. coli) containing the incorporated aacC1 gene. The plasmid backbone was removed by counter selection on LB-agar supplemented with 10% sucrose and Gen50. Gene deletion was confirmed by amplifying the disrupted cpaA gene from the chromosome of the AB031: ΔcpaA strain using primers F1 and F2 and subsequent sequencing of the PCR product.
Amplifies the 1283–1782 bp fragment of cpaA with a 3′ BamHI site (underlined)
Amplification of aacC1 gene
Amplification of aacC1 gene
Amplifies the 599–1267 bp fragment of cpaA
Amplifies the 599–1267 bp fragment of cpaA
Acinetobacter 16S ribosomal RNA gene
Acinetobacter 16S ribosomal RNA gene
Sequencing primers for AB031: Δ cpaA
Sequencing primers for AB031: Δ cpaA
CpaA cleavage of fibrinogen
Cleavage of fibrinogen was assessed as described previously (Imamura et al., 2008). Cell-free secretions were harvested from AB031 or AB031: ΔcpaA as described above for CpaA purification (from AB031) and concentrated 50x using a 10 kDa MWCO filter. Equal volumes of concentrated cell-free secretions and bovine fibrinogen (Sigma, St. Louis, MO; 6 mg mL−1) were incubated for 16 h at 37 °C. The samples were then mixed with loading dye (Laemmli, 1970), and 10 μg of fibrinogen was loaded per well. SDS-PAGE was performed using 4% to 20% polyacrylamide gradient gels according to the manufacturer's instructions (Bio-Rad, Mississauga, ON, Canada). Protein bands were stained with Coomassie Brilliant Blue.
Activated partial thromboplastin time (aPTT) assay
aPTT times were determined using a microplate reader (Molecular Devices, Sunnyvale, CA). Purified CpaA in HBS (0.05 mg mL−1) or concentrated cell-free culture supernatants, which was prepared with 10 kDa centrifugal filters and titrated to have similar activity as 0.05 mg mL−1 purified CpaA (c. 50x concentrated), were mixed with equal volumes of normal human reference plasma for 30 min at room temperature. Then, an equal volume of aPTT reagent (bioMerieux, Durham, NC) was added (1 : 1 : 1 sample : plasma : aPTT reagent, final). This mixture was incubated for 5 min and then CaCl2 was added to a final concentration of 7.5 mM. Clot times were determined as described elsewhere (Tilley et al., 2012).
Prothrombin time (PT) assay
PT times were determined using a microplate reader (Molecular Devices, Sunnyvale, CA). Purified CpaA in HBS (0.05 mg mL−1) or concentrated cell-free culture supernatants, which was prepared with 10 kDa centrifugal filters and titrated to have similar activity as 0.05 mg mL−1 purified CpaA (c. 50x concentrated), were mixed with equal volumes of normal human reference plasma. An equal volume of thromboplastin (Trinity Biotech, Wicklow, Ireland) was then added (1 : 1 : 2 sample : plasma : PT reagent, final), and clot times were determined as described elsewhere (Tilley et al., 2012).
Protein concentration was determined by the Bradford method (Bradford, 1976) and/or the bicinchoninic acid method (Fisher, Toronto, ON, Canada) according to the manufacturer's instructions, using bovine serum albumin as a reference standard.
All experiments were performed a minimum of three times independently, in triplicate. Statistical comparisons were made using sigma plot 11.0 (San Jose, CA, USA) using Student's t-test, Mann–Whitney rank sum test, or anova with Tukey's post hoc analysis where appropriate.
Results and discussion
The success and dissemination of MDR/PDR A. baumannii strains in nosocomial settings necessitate a better understanding of the virulence mechanisms utilized by this organism. Mechanisms of antibiotic resistance have captivated the research surrounding this organism; however, the mechanisms surrounding virulence remain largely unidentified or uncharacterized. Here, we identify a novel virulence protease present in several A. baumannii clinical isolates.
Identification of CpaA
Numerous bacterial secreted proteases are known to interact with host systems to promote infection. We isolated the secreted protease CpaA from the cell-free secretions of an A. baumannii clinical isolate and determined the identity and amino acid sequence by tandem mass spectrometry (seven unique peptides, nine spectra, identified 78/593 amino acids; 13% coverage; GenBank: KJ461713). Unidentified regions were deduced by nucleotide sequencing (Sanger sequencing; 5x coverage), identifying the predicted conserved active site (Fig. S1). The amino acid sequence had limited homology to other characterized proteins. CpaA has a predicted isoelectric point of 8.46, and a predicted extinction co-efficient of 104 360 M−1 cm−1. The protein contains 7% acidic, 8% basic, 43% neutral, and 42% hydrophobic amino acids, with a molecular weight of 64 307.3 Da after secretion, according to mass spectrometry. A secretion signal (a.a. 1–22) was identified toward the N-terminal region, while reprolysin class III M12B- (a.a. 418–529) and pappalysin-1-like (a.a. 459–529) domains were identified overlapping the C-terminal metzincin conserved proteolytic active site (a.a. 519–529) with the characteristic HExxHxxGxxH sequence; however, large sections of the polypeptide lacked homology with other proteins sequenced to date (Fig. 1). Members of the metzincin protease family are present across all kingdoms of life, sharing a common scaffold and a zinc-dependent active site environment (Gomis-Ruth, 2009). CpaA has limited homology to any other characterized peptides in regions other than the active site; it was therefore important to determine whether the protease is active, and whether it targets host-derived substrates as a potential virulence mechanism.
CpaA targets the common pathway of coagulation
The reprolysin M12B-like domain of CpaA is of interest because this subgroup of metzincins has potent effects on coagulation, particularly those proteases present in snake venoms (Bjarnason & Fox, 1995). Additionally, these proteases are associated with numerous human disorders, including asthma, cardiac hypertrophy, rheumatoid arthritis, endotoxic shock, inflammation, and bleeding disorders (Gomis-Ruth, 2009). Given the broad role of this group of enzymes in human health and disease and the high prevalence of secreted virulence proteases among bacterial pathogens, we decided to investigate CpaA as a potential virulence factor.
CpaA was purified from the clinical isolate AB031 to homogeneity of ≥ 95% (Laemmli, 1970) (Fig. 2a).
Given the relatively limited homology of CpaA to other proteases, we first sought to determine whether the protease targets the coagulation system like other reprolysin proteases. The coagulation system is a critical component of the innate immune response (Delvaeye & Conway, 2009), is vital to antimicrobial defense (Massberg et al., 2010), and is a target of several bacterial virulence factors. CpaA was proteolytically active toward the common pathway of coagulation. FV is a critical component of the prothrombinase enzyme complex necessary for effective thrombin generation and fibrin clot formation (Mann et al., 1982). FV plays an important role in antimicrobial host defense. Depletion of FV in either the plasma or platelet compartment of the host during bacterial infection causes a markedly increased mortality rate in mouse models (Sun et al., 2009), and FV is a target of bacterial virulence factors (Brunder et al., 1997). CpaA was capable of cleaving FV in human plasma in a dose-dependent manner when treated with 50 ng–5 μg of purified CpaA. The proteolysis generated a detectable 150 kDa cleavage fragment after treatment with 0.5 or 5 μg of CpaA (Fig. 2b). Cleavage of FV was associated with inhibition of FV clotting activity in human plasma. FV clotting activity was inhibited by 50 ng–5 μg of purified CpaA, reducing activity by 5.58 ± 0.34–60.6 ± 0.8% [mean ± standard error of the mean (SEM)], respectively (Fig. 2c). Cleavage of FV did not absolutely correlate with inhibition of clotting activity, particularly at 500 ng, where almost complete cleavage of FV is associated with a 15.4 ± 0.28% (mean ± SEM) reduction in FV clotting activity. CpaA cleavage of FV was direct and did not require other host proteases. Concentrated culture supernatant from AB031, but not from AB031: ΔcpaA, was capable of cleaving purified FV (Fig. 2d). Proteolytic cleavage and inactivation of FV is a strategy used by other pathogens and would be expected to promote dissemination and enhance permeability within the vasculature, promoting infection (Loof et al., 2011).
CpaA was also capable of deregulating the common pathway of coagulation by proteolysis of fibrinogen. It was proteolytically active against the Aα chain of fibrinogen, but not the Bβ or γ chains (Fig. 2e). The Aα chain is targeted by ASP from Aeromonas sobria (Imamura et al., 2008), and fibrinogen is targeted by a number of virulence factors from Porphyromonas gingivalis (Imamura et al., 1995). Degradation of fibrin is essential to Yersinia pestis pathogenicity; Pla deletion mutants which are unable to activate the host fibrinolytic system display a million-fold increase in the median lethal dose in mice (Sodeinde et al., 1992). Taken together, these data demonstrate that CpaA is capable of targeting the common pathway of coagulation through cleavage and inactivation of FV in addition to cleaving the Aα chain of fibrinogen. Disruption of the common pathway is a common virulence mechanism employed by bacterial pathogens to promote bleeding and dissemination during infection.
CpaA diminishes contact-activated clotting but enhances tissue factor-activated clotting
To further evaluate the role of CpaA in deregulating the coagulation system, we next sought to evaluate its ability to deregulate the contact activation (intrinsic) and tissue factor (extrinsic) pathways of coagulation. Evidence has begun to accumulate which demonstrates that the contact activation pathway has coevolved as part of the innate immune system. Activation of the contact system on the surface of bacterial pathogens occurs in a manner which is remarkably similar to that of the complement system (Opal & Esmon, 2003) and high-molecular-weight kininogen (HK) processing after activation on the membranes of Streptococcus pyogenes, Staphylococcus aureus, and Salmonella results in the generation of antimicrobial peptides. Decreased levels of contact coagulation factors in patients suffering from severe bacterial infection are correlated with increased mortality rates (van Deuren et al., 2000).
Purified CpaA-treated human plasma displayed a significantly increased activated partial thromboplastin clotting time (aPTT) [109.9 ± 12.0 s (mean ± SEM)] when compared to buffer-treated controls [64.9 ± 0.1 s (mean ± SEM)] (P = 0.029; Fig. 3a), indicating inactivation or inhibition of one or more contact pathway proteins. Concentrated cell-free culture supernatant was also capable of prolonging the time required for aPTT clot formation. Concentrated cell-free secretions from the CpaA expressing clinical isolate AB031 were 110.9 ± 9.7 s (mean ± SEM), significantly longer than AB031:Δ cpaA 64.6 ± 5.3 s (mean ± SEM) or the buffer-treated control 64.5 ± 0.9 s (mean ± SEM; P < 0.001; Fig. 3b). Taken together, these data indicate that CpaA is capable of interfering with contact-activated clot formation in human plasma, a process which is required for effective innate immune defense during A. baumannii infection.
Next, we sought to evaluate the effects of CpaA on the tissue factor pathway using the PT assay. Tissue factor pathway activation occurs when the typically sequestered tissue factor is exposed to the blood. Buffer-treated plasma clotted after 21.9 ± 0.1 s (mean ± SEM), whereas CpaA-treated plasma clotted significantly faster 19.2 ± 0.6 s (mean ± SEM; P < 0.001; Fig. 3c). Concentrated cell-free secretions from the CpaA expressing clinical isolate AB031 also reduced the time required for clot formation (19.9 ± 0.5 s; mean ± SEM) when compared to buffer controls (21.9 ± 0.8 s; mean ± SEM) or AB031: Δ cpaA (21.8 ± 0.6 s; mean ± SEM; P = 0.003). This is a curious result, as FV inactivation would be expected to increase PT clot times (Bates & Weitz, 2005). Activation of the tissue factor pathway may promote the generation of a fibrin ‘shield’ at the site of infection, a process which is known to limit antibiotic penetration and promote antibiotic resistance (Bergeron et al., 1993).
CpaA interacts with the coagulation cascade in a complex manner, acting both as an activator and an inhibitor of coagulation. The antimicrobial-associated contact activation pathway is inhibited, while the tissue factor pathway is enhanced. Activation of the tissue factor pathway may inhibit the effectiveness of antimicrobial therapy and promote resistance, while inhibition of the contact pathway may abrogate the coagulation-related immune response. Several bacterial virulence proteases are capable of activating and inhibiting coagulation, including the Pla protease from Y. pestis (Sodeinde et al., 1988). To our knowledge, interaction of A. baumannii proteins with the human coagulation system has not been reported.
Prevalence of CpaA among clinical relevant A. baumannii
Finally, we sought to evaluate the prevalence of CpaA among A. baumannii strains of clinical importance. blast searches revealed multiple predicted proteins of similar amino acid composition (65–99%) within the Acinetobacter genus, including A. baumannii which had several predicted proteins with amino acid compositions which varied between 98 and 99% of CpaA. We designed nucleotide primers for cpaA and tested 16 A. baumannii strains in our collection which have been isolated from Canadian hospitals (Fernando et al., 2013). The presence of cpaA was assessed using primers designed to amplify nucleotides 599–1247, and cell-free secretions were assessed for proteolytic activity against FV (Fig. 4). Primers designed to amplify nucleotides 63–160 and 1065–1167 of cpaA were also used along with SDS-PAGE and silver staining of the cell-free secretions for identification of secreted proteins (data not shown). PCR amplicons were sequenced to confirm identity (data not shown). ATCC 17978 and 19606 genomes were not predicted to contain cpaA which correlated with our data. In total, 12/16 (75%) of A. baumannii clinical isolates generated amplicons from primers designed to amplify cpaA, and all but one of these (11/12) had proteolytic activity against FV generating an apparent 150 kDa cleavage fragment showing a very tight correlation between the detection of the gene by PCR and the proteolytic activity. AB012 generated a cpaA amplicon, and a protein of the appropriate molecular weight was present in the cell-free portion of the culture; however, no cleavage of FV was observed. It remains unclear whether these positive strains harbor CpaA specifically or close homologs. Nevertheless, this clearly indicates that a significant portion of clinically relevant A. baumannnii isolates secrete a protease which targets and deregulates the coagulation system.
This research was supported by grants from the Applied Research and Innovation Center, Centennial College (S01-04 and S01-12) (D.T.), Natural Science and Engineering Research Council (A.K.), and University of Ontario Institute of Technology (J.A.S.).
D.T., J.A.S. and A.K. conceived and designed the experiments, interpreted the results, and wrote the manuscript. R.L. and S.W. helped write the manuscript. D.T., R.L., and S.W. performed the experiments.