CipA of Acinetobacter baumannii Is a Novel Plasminogen Binding and Complement Inhibitory Protein

Abstract

Acinetobacter baumannii is an emerging opportunistic pathogen, responsible for up to 10% of gram-negative, nosocomial infections. The global increase of multidrug-resistant and pan-resistant Acinetobacter isolates presents clinicians with formidable challenges. To establish a persistent infection, A. baumannii must overcome the detrimental effects of complement as the first line of defense against invading microorganisms. However, the immune evasion principles underlying serum resistance in A. baumannii remain elusive. Here, we identified a novel plasminogen-binding protein, termed CipA. Bound plasminogen, upon conversion to active plasmin, degraded fibrinogen and complement C3b and contributed to serum resistance. Furthermore, CipA directly inhibited the alternative pathway of complement in vitro, irrespective of its ability to bind plasminogen. A CipA-deficient mutant was efficiently killed by human serum and showed a defect in the penetration of endothelial monolayers, demonstrating that CipA is a novel multifunctional protein that contributes to the pathogenesis of A. baumannii.

Acinetobacter baumannii is a gram-negative pathogen that is responsible for 2%–10% of hospital-acquired infections [1, 2]. In the United States, 63% of infections are caused by multidrug-resistant isolates [3], and pan-resistant A. baumannii strains are emerging [4, 5]. Multidrug resistance and its remarkable tolerance to desiccation allows A. baumannii to persist in healthcare settings housing critically ill patients [6, 7], particularly intensive care units [8]. A. baumannii is responsible for various infections, including ventilator-associated pneumonia and bacteremia, thereby contributing significantly to morbidity and mortality in susceptible individuals [9]. Numerous factors contribute to the virulence of A. baumannii [10]. However, not many of them are well understood, and novel treatment regimens are urgently needed.

As the first line of defense against invading microorganisms [11], complement is activated through 3 canonical pathways. The classical pathway is activated by immune complexes, the lectin pathway is activated by specific, pathogen-associated carbohydrates, and the alternative pathway is initiated through spontaneous hydrolysis of the central complement component, C3. Activation of complement results in generation of C3 convertases, which cleave C3, leading to deposition of C3b on the microbial surface. Opsonization with C3b leads to increased phagocytosis of invading pathogens [12]. Binding of C3b to surface-attached C3 convertases alters the substrate specificity of the convertase from C3 to C5. C5 convertases then cleave C5 into C5a and C5b, initiating the terminal pathway and leading to formation of the bacteriolytic terminal complement complex (TCC) [13].

The 92-kDa glycoprotein plasminogen (Plg) is present in human serum in micromolar concentrations. The proenzyme consists of 5 consecutive kringle domains (K1-K5) and the C-terminal protease domain (P) [14]. Plg is converted to the active serine protease plasmin through cleavage by activators such as urokinase-type Plg activator (uPA). Plasmin has a low substrate specificity and, in addition to its physiological substrate fibrinogen, is able to degrade constituents of the extracellular matrix (ECM) and complement components C3b and C5 [15]. Acquisition of Plg is an attractive strategy for pathogens, facilitating dissemination and allowing them to counter the deleterious effects of complement.

In this study, we identified a novel Plg-binding protein, which we have termed complement-inhibitor and Plg-binding protein of A. baumannii (CipA). We showed that CipA bound Plg, which, upon conversion to plasmin, cleaved both fibrinogen and C3b. CipA also inhibited the alternative pathway in vitro. A CipA-deficient mutant was efficiently killed by human serum and showed a marked defect in the penetration of endothelial cell monolayers.

MATERIALS AND METHODS

Bacterial Strains, Human Cell Line, and Culture Conditions

A. baumannii strains ATCC 19606 and ATCC 17978, as well as clinical isolates 11CS, 15CS, 17CS, 25CS, 27CS, and V754948 [16, 17], were grown at 37°C in lysogeny broth. The ΔcipA deletion mutant was generated using a previously described markerless mutagenesis approach [18].

Escherichia coli JM109 was grown at 37°C in yeast tryptone broth. Primary human umbilical vein endothelial cells (HUVEC) were purchased from PromoCell and cultured at 37°C and 5% CO₂ in Endothelial Cell Growth Medium 2 supplemented with 10% fetal calf serum.

Proteins and Antibodies

Purified human Plg was purchased from Haematologic Technologies. Urokinase Plg activator was obtained from Merck Millipore. Fibrinogen and bovine serum albumin (BSA) were purchased from Sigma-Aldrich. Purified C3b and the polyclonal anti-C3 antiserum were obtained from Complement Technology. Polyclonal antisera raised against Plg (catalog no. R1598) and fibrinogen (catalog no. AP09017PU-N) were purchased from Acris Antibodies. The anti-hexahistidine antibody was purchased from GE Healthcare. Horseradish peroxidase (HRP)–conjugated immunoglobulins were obtained from Dako. Alexa Fluor 488–conjugated anti-mouse immunoglobulins were obtained from Life Technologies, and phycoerythrin-conjugated anti-goat immunoglobulins were purchased from R&D Systems.

Generation of Recombinant Proteins

To generate recombinant, hexahistidine-tagged CipA_His, the CipA-encoding gene of A. baumannii ATCC 19606 (HMPREF0010_01565; accession number ZP_05828182) was amplified by polymerase chain reaction (PCR), using primers AbCipA-BamHI and AbCipA-SalI (Supplementary Table 1), and cloned into pQE-30 Xa (Qiagen), yielding vector pQE-CipA 19606. CipA_His lacks the N-terminal signal sequence and encompasses amino acid residues 19–369. To generate C-terminally truncated CipA constructs, plasmid pQE-CipA 19606 was used in PCR with primer pQE-FP-30 in combination with CipA-351(-) or CipA-290(-). PCR products were cloned into pQE-30 Xa and sequenced to ensure no mutations had been introduced. In one instance, a spontaneous mutation resulted in the generation of a stop codon at position 153. This construct was also included in subsequent studies. To generate recombinant CipA proteins with various amino acid substitutions (lysine to alanine), PCR was performed with oligonucleotides designed for site-directed mutagenesis (Supplementary Table 1). Following incubation with DpnI, reactions were used to transform E. coli JM109 cells. Plasmid DNA was isolated and sequenced to ensure it contained the desired substitutions. The lipoprotein BBA70 from Borrelia burgdorferi was used as a positive control [19]. All recombinant proteins were produced in E. coli JM109 and purified by affinity chromatography as previously described [19].

Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis (SDS-PAGE), Western Blotting, and Far-western Blotting

SDS-PAGE, Western blotting, and far-Western blotting were all conducted as previously described [19].

Enzyme-Linked Immunosorbent Assay (ELISA)

Microtiter plates (MaxiSorp, Nunc) were coated with 100 µL of recombinant proteins (5 µg/mL) at 4°C overnight. For whole-cell ELISA, plates were coated with 100 µL of A. baumannii cell suspensions in phosphate-buffered saline (PBS) at 4°C overnight. Wells were blocked and incubated with 100 µL Plg (10 µg/mL) at room temperature for 1 hour. Where indicated, varying concentrations of tranexamic acid or sodium bromide (NaBr) were added. Wells were incubated with a polyclonal anti-Plg antiserum followed by HRP-conjugated anti-goat immunoglobulins at room temperature for 1 hour. Immune complexes were visualized using o-phenylenediamine (Sigma-Aldrich), and the absorbance was measured at 490 nm by using PowerWave HT (Bio-Tek Instruments).

Flow Cytometry

A total of 5 × 10⁸ bacterial cells were incubated in 1% BSA (w/v) containing 100 µg/mL Plg for 1 hour at room temperature. Cells were then incubated with a polyclonal anti-Plg antiserum, followed by phycoerythrin-conjugated anti-goat immunoglobulins. After fixation with PFA, cells were analyzed with a FACSCalibur flow cytometer (BD Biosciences). To determine surface exposure of CipA on A. baumannii, cells were incubated with a polyclonal anti-CipA antiserum, followed by Alexa Fluor 488–conjugated anti-mouse immunoglobulins.

Proteolytic Degradation of Fibrinogen and C3b by Active Plasmin

Recombinant proteins (5 µg/mL) were immobilized on microtiter plates. After blocking and incubation with 10 µg/mL Plg for 1 hour at room temperature, 93.5 µL of a reaction mixture were added, containing 50 mM Tris/HCl (pH 7.5), 20 µg/mL fibrinogen or C3b, and 6.5 µL uPA (2.5 µg/mL). Following incubation at 37°C, reactions were terminated by addition of SDS-PAGE sample buffer. Samples were separated by SDS-PAGE and transferred to nitrocellulose membranes. Membranes were probed with polyclonal antisera raised against fibrinogen or C3.

Serum Susceptibility Assay

A total of 5 × 10⁷A. baumannii cells were suspended in 100% NHS or in PBS-diluted NHS (10%–90%). Following incubation at 37°C for 2 hours, bacteria were serially diluted in PBS and plated onto Mueller-Hinton agar (Oxoid) for counting of colony-forming units (CFU).

Hemolytic Assays

Hemolytic assays were performed as previously described [20]. Briefly, to investigate the alternative pathway (rabbit erythrocytes), CipA was preincubated with 7.5% NHS Mg–ethyleneglycoltetraacetic acid buffer for 15 minutes at 37°C. For the classical pathway (amboceptor-sensitized sheep erythrocytes), CipA was preincubated for 15 minutes with 1% NHS in GVB⁺⁺ buffer at 37°C. Erythrocytes were added, and reactions were incubated for 30 minutes at 37°C. Absorbance of supernatants at 414 nm was determined. For the terminal pathway, sheep erythrocytes were preincubated with 1.5 µg/mL C5b-6 for 10 minutes at room temperature. In parallel, recombinant CipA or controls were preincubated with purified complement components C7 (2 µg/mL), C8 (0.4 µg/mL), and C9 (2 µg/mL) for 5 minutes at room temperature. The preincubated complement components were then added to the C5b-6 treated erythrocytes, and reactions were incubated for 30 minutes at 37°C.

Transwell Assay

A total of 2.5 × 10⁵ HUVEC were seeded onto transwell inserts with a 1-µm pore size (ThinCerts, Greiner Bio-One) and cultivated until confluent. We then incubated 2.5 × 10⁷A. baumannii cells with 100 µg/mL Plg in PBS for 1 hour at room temperature, followed by incubation with 0.16 µg/mL uPA in PBS for 1 hour at room temperature. Cells were washed and resuspended in 300 µL of Endothelial Cell Growth Medium 2. HUVEC cells were infected with the pretreated A. baumannii cells (multiplicity of infection, 100) for 2 hours at 37°C. Culture medium from the wells was serially diluted and plated on Mueller-Hinton agar to count CFU.

Membrane Extraction

Membrane extraction of A. baumannii was performed according to a slightly modified protocol described by Radolf et al [21].

Ethics Statement

The study and respective documents were approved by the ethics committee at the University Hospital of Frankfurt (control number 492/13) and blood donors provided written informed consent.

RESULTS

CipA Is a Plasminogen-Binding Protein

To determine whether A. baumannii is able to bind human Plg, cells of strain ATCC 19606 and various clinical isolates were immobilized to microtiter plates, and binding of Plg was assayed via ELISA. Starting at 1 × 10⁷ cells, significant binding of Plg was observed (Figure 1A).

To identify Plg-binding proteins, far-Western blotting was performed using whole-cell lysates of various A. baumannii strains, yielding several distinct signals (Figure 1B). To exclude signals from cytoplasmic proteins, the analysis was repeated using membrane fractions of strain ATCC 19606. A clear signal was obtained in the hydrophobic fraction. The protein marked by an arrowhead in Figure 1B was excised from the gel and analyzed by mass spectrometry. The protein with an approximate molecular weight of 41 kDa, termed CipA, corresponded to ORF HMPREF0010_01565 of A. baumannii ATCC 19606. Western blot analysis, using a specific antibody, confirmed the presence of CipA in all investigated clinical A. baumannii isolates and a primary localization in the hydrophobic fraction (Figure 1C).

To determine surface localization of CipA, viable A. baumannii cells were incubated with an anti-CipA antibody and assayed by flow cytometry (Figure 1D). A mean (±SD) of approximately 44.5%±1.2% of A. baumannii cells stained positive for CipA.

Characterization of the CipA-Plg Interaction

To analyze binding of Plg, recombinant CipA and the Plg-binding lipoprotein BBA70 of B. burgdorferi (positive control) were blotted to a nitrocellulose membrane and incubated with Plg. Far-Western blot analysis using a Plg-specific antiserum revealed that CipA bound Plg (Figure 2A). Binding was confirmed by ELISA (Figure 2B) and occurred in a dose-dependent fashion (Figure 2C). Nonlinear regression allowed the approximation of the dissociation constant for the CipA-Plg interaction, with K_d = 36 nM (±6 nM).

As lysine residues play an important role in the interaction of bacterial proteins with Plg [22], we investigated, whether lysine residues are involved in the CipA-Plg interaction. Addition of the lysine analog tranexamic acid significantly reduced binding of Plg (Figure 2D). Lysine residues are positively charged at neutral pH and increasing ionic strength should thus affect binding of Plg by CipA. Interestingly, increasing ionic strength with NaBr had no significant effect on binding of Plg by CipA (Figure 2E), even at concentrations of 1 M NaBr indicating that ionic interactions are dispensable for Plg-CipA interactions.

Localization of the Plg-Interacting Region Within CipA

To identify the Plg-interacting regions of CipA, a number of truncated CipA proteins were created (Figure 3A) and their ability to bind Plg was assayed (Figure 3B and 3C). Deletion of 18 amino acids at the C-terminus of CipA (CipA_19-351), encompassing 4 lysine residues, was sufficient to significantly reduce Plg binding. Next, we employed site-directed mutagenesis, to identify the lysine residues required for Plg binding. The modified proteins CipA_K365A-K369A, CipA_K368A/K369A, and CipA_K369A no longer bound Plg (Figure 3C). Plg binding to the modified proteins was also assayed by ELISA, and again CipA_K365A-K369A, CipA_K368A/K369A, and CipA_K369A bound significantly less Plg than wild-type CipA (Figure 3D).

To determine which kringle domains interact with CipA, Plg fragments comprising various kringle domains were assayed for their ability to bind CipA. Fragment K1-5-P bound to CipA, as did fragment K4-5-P. By contrast, neither fragment K5-P nor fragment K3-5-P were able to bind to CipA (Figure 3E), suggesting that binding occurs via lysine-binding sites located within kringle 1 and possibly kringle 4.

CipA-Bound Plasmin Degrades Fibrinogen and C3b

Plg is converted to plasmin via proteolytic cleavage by activators such as uPA. To assess whether CipA-bound Plg is accessible to uPA, we performed a fibrinogen degradation assay. CipA was immobilized and incubated with Plg, and a reaction mixture containing uPA and fibrinogen was added. Reactions were incubated at 37°C, and aliquots were taken at the time intervals indicated in Figure 4A. For CipA, complete degradation of the fibrinogen α-, β-, and γ-chain could be observed after just 60 minutes. A similar result was seen for BBA70. Some background degradation was also observed with BSA and BBA66, serving as negative controls, as well as in control reactions omitting Plg and including the lysine analog tranexamic acid. Further control experiments (Supplementary Figure 1) showed that fibrinogen remained stable for 120 minutes when incubated at 37°C, but these controls also show degradation when fibrinogen was incubated with uPA in the absence of plasminogen.

Next, a C3b degradation assay was performed to determine whether CipA-bound plasmin cleaves C3b. The appearance of C3b degradation products with molecular weights of 27 kDa, 37 kDa and 43 kDa indicates that CipA-bound plasmin was able to cleave this central complement component (Figure 4B). Similar results were obtained for BBA70. Some background degradation was also observed in control reactions where Plg was added together with tranexamic acid or when Plg or uPA were omitted. Additionally, some degradation was observed with BSA, although there is a clear difference between BSA and CipA/BBA70, especially at earlier time points of 1 or 2 hours. Additional control experiments indicated that C3b remained stable at 37°C for up to 24 hours and showed that degradation products generated by plasmin are distinct from the cleavage fragments generated by complement regulators factor I and factor H (Supplementary Figure 2).

CipA Inhibits the Alternative Pathway

Since CipA-bound plasmin was able to cleave the key complement component C3b, we sought to investigate whether CipA is able to protect erythrocytes from complement-mediated lysis in a hemolytic assay. The inhibitory effect of CipA on the classical pathway, alternative pathway, and the terminal pathway was assessed by preincubating increasing amounts of CipA with NHS before erythrocytes were added. Following incubation, erythrocyte lysis was determined by measuring the release of hemoglobin. In addition, the effect of CipA on the terminal pathway was studied using purified complement components. While CipA neither inhibited the classical pathway nor the terminal pathway, a strong inhibition of the alternative pathway was observed (Figure 5). To determine whether alternative pathway inhibition correlated with binding of Plg, CipA_K365A-K369A was also analyzed. While this protein was unable to bind Plg (Figure 3C and 3D), it retained its inhibitory activity (Figure 5).

Inactivation of the CipA-Encoding Gene Renders A. baumannii Serum Sensitive

Using a marker-free mutagenesis approach, the CipA-encoding gene in A. baumannii ATCC 19606 was inactivated, resulting in generation of mutant ΔcipA. Western blot analysis using an anti-CipA antibody (1:100) verified that the mutant did not produce CipA (Figure 6A). Compared with the wild type, the ΔcipA mutant showed a significant defect in Plg binding (Figure 6B). When assaying Plg binding to viable cells, only a mean (±SD) of 7.3%±1.5% of mutant cells bound Plg, compared with 48.1%±2.2% of wild-type cells (Figure 6C).

To investigate the impact of CipA deletion on serum resistance, we performed serum-susceptibility tests. Bacteria were incubated with increasing concentrations of human serum, and survival was determined. Compared with the wild type, the ΔcipA mutant displayed a marked defect in serum resistance at concentrations ranging from 20% to 50% (Figure 6D).

Loss of Plasminogen Binding Results in Defect in Penetration of Endothelial Monolayers

Active plasmin can degrade numerous constituents of the ECM, including fibronectin [23], which plays an important role in adhesion of A. baumannii to human cells [24]. Hypothesizing that A. baumannii–bound plasmin would promote transmigration and thus dissemination of the bacteria, we performed a transwell assay. Wild-type A. baumannii cells or the ΔcipA mutant were incubated with Plg and treated with uPA. HUVEC monolayers were then infected, and CFU counts of bacteria that penetrated the monolayer were determined. As shown in Figure 6E, cells of the wild-type, but not of the ΔcipA mutant penetrated the endothelial monolayer. Unlike the wild-type, the ΔcipA mutant did not bind Plg or plasmin under the experimental conditions (Figure 6F).

DISCUSSION

To establish an infection and persist in the human host, A. baumannii must overcome innate immunity, in particular the bactericidal effects of complement. It has recently been shown that A. baumannii resists complement-mediated killing [25], which in turn results in very high bacterial blood densities [26].

Here, we identified CipA as a novel, surface-exposed Plg and complement-inhibitory protein, which was detected in A. baumannii ATCC 19606, as well as clinical isolates. Using flow cytometry, only half of the A. baumannii cells stained positive for CipA, suggesting that CipA is expressed stochastically under the experimental conditions used. Expression of the CipA-encoding gene is, however, upregulated 7.8-fold during growth in human serum [27]. CipA bound Plg dose dependently, with an apparent dissociation constant of 36 nM (±6 nM), which is comparable to other bacterial Plg binding proteins [28, 29]. Binding of Plg to CipA is mediated by lysine residues, and independent of ionic strength, similar to the elongation factor Tuf of Leptospira interrogans [30] or ErpP of B. burgdorferi [31].

Site-directed mutagenesis identified the lysine residue at position 369 as an essential determinant for binding of Plg. C-terminal lysine residues are often essential for interactions with Plg as shown for elongation factor Tuf of A. baumannii [32], enolase of Streptococcus pneumoniae [33] or BBA70 of B. burgdorferi [19]. Furthermore, we showed that kringle domain 1, and to a lesser degree kringle domain 4, participate in the Plg-CipA interaction. Qualitatively, the interaction of CipA with K1-5-P seemed stronger than K4-5-P, presumably owing to the higher affinity of the lysine binding site in kringle 1 (17 µM, compared with 36 µM in kringle 4) [34]. Interestingly, while fragments K1-5-P and K4-5-P bound CipA, fragment K3-5-P did not. It is conceivable, that the binding site in kringle 4 is not accessible in fragment K3-5-P and only becomes accessible in fragment K4-5-P following truncation of kringle 3, similar to a conformational change in lys-Plg [34]. In that case, binding of Plg to CipA could be mediated solely by the high-affinity binding site in kringle 1.

Bacteria can become trapped in fibrin networks, limiting their dissemination [35]. Degradation of fibrin, thus may help A. baumannii to escape entrapment and promote dissemination, as has been demonstrated for S. canis [36]. A. baumannii secretes the CpaA protease, which is able to cleave fibrinogen and deregulate coagulation [37]. However, CpaA is notably absent from ATCC 19606. Acquisition of Plg through CipA would endow A. baumannii strains lacking CpaA with a similar proteolytic activity.

Several pathogens are able to acquire Plg, including S. pneumoniae, Staphylococcus aureus, Haemophilus influenzae, and B. burgdorferi [22, 38, 39] and many use the proteolytic activity to cleave complement [40, 41]. Upon alternative pathway activation, susceptible A. baumannii isolates show large amounts of C3 deposited on the surface while deposition is limited in serum-resistant isolates [25]. Based on Figure 4B, at least some C3b was degraded by CipA-bound plasmin, though this would seem to play a subordinate role, considering that binding of Plg did not correlate with inactivation of the alternative pathway by CipA. A. baumannii secretes the serine protease PKF, which specifically inhibits the alternative pathway, although the mechanism remains unclear [42]. Binding of Plg through CipA could supplement endogenous proteases or replace them in strains lacking PKF. Plg is present in human serum in micromolar concentrations, and the CipA-encoding gene is upregulated 7.8-fold in serum-resistant A. baumannii grown in human serum [27]. It might be more economical for A. baumannii to co-opt Plg as opposed to producing and secreting endogenous proteases. However, in extravascular fluids where Plg concentrations are lower, endogenous proteases might take over.

Of importance, here we identified for the first time a complement inhibitory molecule of A. baumannii, which specifically inhibits alternative pathway activation, although the mechanism underlying downregulation of the alternative pathway by CipA remains to be elucidated.

Inactivation of the CipA-encoding gene resulted in a marked defect in Plg-binding and rendered cells highly susceptible to complement, which underlines the contribution of CipA to the survival of A. baumannii in the presence of active complement.

In transmigration assays, A. baumannii cells lacking CipA were unable to penetrate an endothelial monolayer, implicating the necessity for CipA. Efficient dissemination requires that A. baumannii degrade ECM constituents and basement membranes. Plasmin is able to degrade a number of ECM proteins. Acquisition of Plg may thus supplement endogenous proteases and aid A. baumannii in penetrating cell layers and disseminate more efficiently, as is the case for B. burgdorferi and L. interrogans [43, 44]. Figure 7 summarizes how CipA could serve as a multifunctional molecule facilitating the immune escape of A. baumannii.

Taken together, we identified CipA as a novel virulence factor, contributing to the survival of A. baumannii in human serum. The characterization of CipA might pave the way for the development of new therapeutics against multidrug resistant A. baumannii.

Notes

Acknowledgments. We thank Jüri Habicht, and Axel Teegler, for skillful and excellent technical assistance; Dr Gottfried Wilharm (Robert Koch Institute, Werningerode, Germany), for A. baumannii strains; Dr Alf Theisen (Central Research Facility, University Hospital of Frankfurt, Germany), for providing rabbit erythrocytes; and Prof Dr Michael Kirschfink (Institute of Immunology, University Hospital of Heidelberg, Germany), who kindly provided sheep erythrocytes. This work forms part of the doctoral thesis of A. K.

Disclaimer. The funder had no role in study design, data collection, and reproduction to publish, or preparation of the manuscript.

Financial support. This work was supported by departmental funds from the Institute of Medical Microbiology and Infection Control, University Hospital of Frankfurt (to P. K.).

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes