Streptococcus pneumoniae Endopeptidase O (PepO) Is a Multifunctional Plasminogen- and Fibronectin-binding Protein, Facilitating Evasion of Innate Immunity and Invasion of Host Cells*

Background: Pneumococci have developed multiple strategies to infect the host. Results: PepO is a ubiquitously expressed pneumococcal protein that interacts with host proteins and facilitates host cell invasion and evasion of innate immunity. Conclusion: PepO is a plasminogen- and fibronectin-binding pneumococcal invasin. Significance: Understanding the mechanism of pneumococcal interaction with host aids designing better therapeutical strategies and gaining control over the pathogen. Streptococcus pneumoniae infections remain a major cause of morbidity and mortality worldwide. Therefore a detailed understanding and characterization of the mechanism of host cell colonization and dissemination is critical to gain control over this versatile pathogen. Here we identified a novel 72-kDa pneumococcal protein endopeptidase O (PepO), as a plasminogen- and fibronectin-binding protein. Using a collection of clinical isolates, representing different serotypes, we found PepO to be ubiquitously present both at the gene and protein level. In addition, PepO protein was secreted in a growth phase-dependent manner to the culture supernatants of the pneumococcal isolates. Recombinant PepO bound human plasminogen and fibronectin in a dose-dependent manner and plasminogen did not compete with fibronectin for binding PepO. PepO bound plasminogen via lysine residues and the interaction was influenced by ionic strength. Moreover, upon activation of PepO-bound plasminogen by urokinase-type plasminogen activator, generated plasmin cleaved complement protein C3b thus assisting in complement control. Furthermore, direct binding assays demonstrated the interaction of PepO with epithelial and endothelial cells that in turn blocked pneumococcal adherence. Moreover, a pepO-mutant strain showed impaired adherence to and invasion of host cells compared with their isogenic wild-type strains. Taken together, the results demonstrated that PepO is a ubiquitously expressed plasminogen- and fibronectin-binding protein, which plays role in pneumococcal invasion of host cells and aids in immune evasion.

The Gram-positive bacterium Streptococcus pneumoniae (the pneumococcus) is a commensal that asymptomatically colonizes the upper respiratory tract. However, alterations within host-pathogen homeostasis result in the ability of pneumococci to gain access to the normally sterile parts of the airways and cause infections. These include mild local infections such as otitis media and sinusitis or life threatening invasive diseases, including lobar pneumonia, sepsis, and meningitis (1). The pneumococcus is the prime cause of community-acquired pneumonia in adults and accounts for two-thirds of all cases of bacteraemic pneumonia (2). In addition, pneumococcal septicemia is a major cause of infant mortality in developing countries, among children under the age of 5 years (3).
Pneumococci utilize multiple mechanisms for colonization of the respiratory tract, transcytosis through host cells, dissemination into the bloodstream, and in evasion of the host immune attack (4 -6). Surface proteins like choline-binding proteins, LPXTG motif containing proteins and lipoproteins (7)(8)(9), have been described as ligands for cellular receptors and binding molecules interacting with various components of the extracellular matrix (ECM) 3 or serum proteins of the host. For example, pneumococcal surface protein C (PspC) is a multifunctional choline-binding protein and a major virulence factor * This work was supported in part by Swedish Research Council Grant K2012-of pneumococci that interacts directly and in a human-specific manner with the polymeric immunoglobulin receptor, and mediates pneumococcal adherence and invasion of host cells (10 -12). In addition, lipoproteins such as ABC metal permease PsaA (13), a pathogenicity-island encoded protein PsrP (14,15) and pili, expressed by a subclass of pneumococci (16,17) have been suggested to have adhesive functions. Furthermore, a number of pneumococcal proteins interacts with fibronectin or plasminogen, a strategy commonly employed by many invading pathogens to colonize or disseminate within the host (18,19). These include pneumococcal adherence and virulence factor (Pav) A, PavB, plasmin and fibronectin-binding protein (Pfb) A as well as PfbB that have been identified as plasminogen-and fibronectin-binding proteins (20 -24).
Fibronectin is a high molecular weight glycoprotein present as an insoluble component in the ECM, on the cell surface, and in the basement membrane or as a soluble component in plasma and other body fluids (25). It is highly conserved among vertebrates and plays a pivotal role in cell adhesion, growth, differentiation, and migration and is an important factor in wound healing and embryonic development (26). Plasminogen, a single chain glycoprotein, is a proenzyme of the serine protease plasmin and plays an important role in fibrinolysis (27), homeostasis (28), and in degradation of ECM (29). In addition to the above mentioned proteins, pneumococci can bind plasminogen by surface displayed enolase and glyceraldehyde-3phosphate dehydrogenase (GAPDH) (30,31). Additionally, interactions with vitronectin, thrombospondin-1, and complement inhibitor Factor H facilitate bacterial adherence to and invasion of host cells (32)(33)(34)(35).
Despite the knowledge and understanding of various virulence factors, the use of antibiotics and the availability of vaccines, the mortality rate due to pneumococcal infection remains alarming. Consequently there is an urgent need to develop new and improved therapies and alternative pneumococcal vaccines to combat pneumococcal diseases. The most promising approach is to develop vaccines based on pneumococcal proteins that contribute to virulence and are common to all serotypes. Therefore, the identification of the essential factor(s) facilitating colonization and subsequent dissemination and the elucidation of the molecular mechanism of pneumococcal-host interactions are important.
While screening the pneumococcal genome for novel proteins that might contribute to virulence, we identified a 72-kDa protein endopeptidase O (PepO), a predicted metallo-endopeptidase that shares homology with M13 peptidase family. M13 peptidase family includes mammalian endopeptidases such as neutral endopeptidase, involved in regulation of many physiological and pathological processes including various aspects of immune response, and endothelin converting enzyme 1 responsible for processing endothelin-1 into its biologically active form. In this study we identify PepO as a new ubiquitously expressed plasminogen-and fibronectin-binding protein. We found that the PepO protein interacts with human epithelial and endothelial cells and inactivation of the pepO gene in S. pneumoniae significantly reduced the bacterial ability to invade these cells.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-All pneumococcal strains used in this study, except for NCTC10319 and D39, were isolated from blood of patients suffering from sepsis in the Southwest county of Skåne, Sweden (Table 1). For in vitro experiments pneumococci were cultured on blood agar plates at 37°C and 5% CO 2 , or in Todd-Hewitt broth (Oxoid) supplemented with 0.5% yeast extract (THY) to a density of 5 ϫ 10 8 cfu/ml (approximately A 600 of 0.5). Escherichia coli strains DH5␣ and BL21(DE3) were cultivated on Luria-Bertani agar or broth.
Generation of PepO-mutant and Preparation of Recombinant Protein-Isogenic mutants that do not express PepO were constructed for the encapsulated D39 strain and for S. pneumoniae NCTC10319, which is a low encapsulated strain and thus suitable for cell culture infection experiments as described earlier (32,36). PepO-deficient mutants of D39 and NCTC10319 were generated by replacement of the pepO sequence with the spectinomycin gene cassette. Briefly, the full-length pepO gene was amplified by PCR from the chromosomal DNA of S. pneumoniae D39 with primers 5-CCATGGCACGTTATCAAGAT-GATTT-3 and 5-CTCGAGCCAAATAATCACGCGCTC-3, which incorporated flanking NcoI and XhoI (underlined) restriction sites. The amplified DNA was cloned in pJET1.2 (Fermentas) and later into pET28a for protein expression (Novagen). For generation of a pepO-mutant, the inserted pepO fragment was digested using EcoRV and the spectinomycin gene cassette was blunt-end ligated with the plasmid. The integrity of the antibiotic gene cassette was verified by PCR. The transformation of pneumococci was performed as described previously using competence-stimulating peptide-1 (NordicBiosite) and cultivated in the presence of spectinomycin (50 g/ml) (37). Transformation of E. coli strains with plasmid DNA was carried out with CaCl 2 -treated competent cells according to a standard protocol.
Proteins and Antibodies-Pneumococcal PepO with an N-terminal His 6 tag was expressed from the pET28a vector in E. coli BL21(DE3) (Stratagene) and purified using a nickel-nitrilotriacetic acid column (GE Healthcare) according to the manufacturer's instruction. Polyclonal antiserum against purified PepO was raised in rabbits by routine immunogenic procedures using His 6 -tagged PepO as antigen (Agrisera, Sweden). Purification of rabbit anti-PepO antiserum was performed by affinity chromatography using protein G-Sepharose columns (GE Healthcare). Human Glu-plasminogen, human fibronectin, and sheep anti-human plasminogen antibodies (Abs) were purchased from Hematologic Technologies, whereas rabbit anti-human fibronectin Abs and peroxidase-conjugated swine anti-rabbit IgG and rabbit anti-sheep IgG Abs were purchased from DakoCytomation. Laminin from Engelbreth-Holm-Swarm murine sarcoma basement membrane was purchased from Sigma, whereas C3b was from Complement Technology. Proteins were labeled with 125 I using the chloramine T method. Normal human serum was prepared from freshly drawn blood obtained from 6 healthy volunteers with informed consent and permission of the ethical board of Lund University. The pooled blood was allowed to clot for 30 min at room temperature (RT) and then incubated for 1 h on ice. After two centrifugations, the serum fraction was frozen in aliquots and stored at Ϫ80°C.
Western Blot Analysis-Purified proteins, bacterial lysates, or culture supernatants were subjected to SDS-PAGE, transferred to a polyvinylidene difluoride membrane and submitted to Western blotting using a rabbit polyclonal anti-PepO Abs and a secondary peroxidase conjugated Abs. The polyclonal anti-PepO IgG was used at a dilution of 1:1,000 and the swine antirabbit IgG peroxidase conjugate (DakoCytomation) was used at 1:2,000 dilutions and the activity was detected using diaminobenzidine (Sigma) and H 2 O 2 .
Direct Binding Assays-Microtiter plates (Maxisorb; Nunc) were coated with 50 l of PepO, plasminogen, or fibronectin at a concentration of 5 g/ml in 75 mM sodium carbonate buffer (pH 9.6) overnight at 4°C. Wells coated with 1% BSA (Applichem) were used as control. Blocking was performed with 250 l of blocking solution (50 mM Tris-HCl (pH 8), 150 mM NaCl, 0.1% Tween 20, 3% fish gelatin (Norland)) for 2 h at RT. PepO, plasminogen, or fibronectin were diluted in binding buffer (50 mM HEPES (pH 7.4), 150 mM NaCl, 2 mM CaCl 2 , and 50 g/ml BSA) and incubated with the immobilized plasminogen or fibronectin for PepO, and with immobilized PepO for plasminogen or fibronectin binding, at the concentrations indicated in the figure legends. For investigation of the effect of ionic strength and the presence of lysine analog ⑀-ACA (⑀-aminocaproic acid) (Sigma) on plasminogen-PepO interactions, the binding buffer was supplemented with NaCl to a final concentration ranging from 0 to 800 mM or with ⑀-ACA to a final concentration ranging from 0 to 40 mM. The plates were washed with 50 mM Tris-HCl (pH 8), 150 mM NaCl, and 0.1% Tween 20 between each step. Bound PepO, plasminogen, and fibronectin were detected with a rabbit anti-PepO, sheep antihuman plasminogen, and rabbit anti-human fibronectin polyclonal Abs, respectively, followed by a swine anti-rabbit peroxidase conjugate or rabbit anti-sheep peroxidase-conjugated Abs (DakoCytomation). The plates were developed with ophenylenediamine (DakoCytomation) substrate and H 2 O 2 , and the absorbance at 490 nm was measured (Varian Cary 50 MPR Microplate Reader).
Binding from Serum-Microtiter plates were coated with 50 l of PepO at a concentration of 5 g/ml in 75 mM sodium carbonate buffer (pH 9.6) overnight at 4°C. Wells coated with 1% porcine gelatin (Sigma) were used as controls. After incubation with blocking solution for 2 h at RT, human serum diluted in binding buffer was added to the wells at the concentrations indicated in figures. After incubation for 1.5 h at RT, bound plasminogen was detected using sheep anti-human plasminogen Abs followed by a rabbit anti-sheep peroxidase-conjugated Abs. The plates were developed with o-phenylenediamine substrate and H 2 O 2 and the absorbance at 490 nm was measured. Biacore Analysis-To determine the kinetics of PepO binding to fibronectin or plasminogen, surface plasmon resonance (SPR) analysis was performed using Biacore 2000. Fibronectin or plasminogen were diluted to 50 g/ml in 10 mM sodium acetate (pH 4.0) and immobilized on the surface of a CM5 sensor chip to reach 2466 and 2800 response units, respectively. All experiments were performed at a continuous flow rate of 30 l/min using Biacore buffer (150 mM NaCl, 10 mM HEPES, 0.002% Tween 20, pH 7.4). The analyte, PepO, was injected in a concentration gradient as indicated in the figures followed by two consecutive injections of 2 M NaCl, 100 mM HCl, and 0.05% SDS for regeneration. The obtained sensorgrams were analyzed using Bio-evaluation software 3.0 using a 1:1 Langmuir binding model of interaction with drifting baseline.
Plasminogen Activation Assay-A PepO (5 g/well)-coated plate was incubated with plasminogen (1 g) in binding buffer for 1.5 h at RT. After extensive washing, 10 units/well of uPA (urokinase-type plasminogen activator) (Sigma) was added and activity of the newly generated plasmin was assayed with chromogenic substrate S-2251 (H-D-Valyl-L-leucyl-L-lysine-p-nitroaniline dihydrochloride) (Chromogenix). The plate was incubated at 37°C, and cleavage of the chromogenic substrate was followed for the time periods indicated in the figure by measuring the absorbance at 405 nm.
C3b, Fibrinogen, and Extracellular Matrix Protein Degradation-Microtiter plates were coated with 50 l of PepO (5 g/well) overnight in PBS at 4°C. Blocking was performed with 250 l of blocking solution for 2 h at room temperature. Plasminogen (5 g/well) diluted in binding buffer was incubated with the immobilized PepO for 1.5 h at RT. After washing, fibrinogen (5 g/well), 125 I-labeleld C3b, fibronectin, or laminin (100 kcpm) together with uPA (10 units/well) were added and incubated at 37°C for the indicated time points. Fibronectin and laminin degradation was followed for 18 h and stopped by addition of reducing SDS-PAGE sample buffer and boiling at 95°C for 5 min. Thereafter the samples were separated by SDS-PAGE. Degradation of fibrinogen was evaluated by subjecting the sample to Western blot analysis using rabbit antiserum to human fibrinogen, followed by swine anti-rabbit IgG peroxidase conjugate. For C3b degradation, positive and negative control reactions were prepared with Factor H and with or without Factor I, respectively. Degradation products of C3b, fibronectin, and laminin were visualized using a Phospho-Imager (Fuji).
Pneumococcal Binding to Immobilized Fibronectin-Microtiter plates were coated with 50 l of fibronectin at a concentration of 5 g/ml in PBS overnight at 4°C. Blocking was performed with 250 l of blocking solution for 2 h at RT. Labeling of the bacteria was performed by incubating 2 ϫ 10 9 cfu of pneumococci with fluorescein isothiocyanate (FITC) (1 mg/ml) in 75 mM sodium carbonate buffer (pH 9.6) for 1 h at 37°C. After extensive washing, the FITC-labeled pneumococci (1 ϫ 10 8 cfu/ml) were added to the wells and incubated for 1 h at 37°C for binding. Fluorescence was measured at 485/535 nm (excitation/emission) using a Victor 2 1420 multilabel counter (Wallac). Measurements were done after three washing steps with 50 mM Tris-HCl (pH 8), 150 mM NaCl, and 0.1% Tween 20.
Bacteria ELISA-Pneumococcal strains D39, NCTC10319, and their corresponding isogenic pepO-mutant strains were cultured overnight on blood agar plates, washed in PBS, and adjusted to 10 8 cfu/ml. Microtiter plates were coated with 50 l of bacterial suspension and incubated overnight at 37°C. After incubation with blocking solution for 2 h at RT, plasminogen diluted in binding buffer was added to the wells at concentrations indicated in the figures. After incubation for 1.5 h at RT, bound plasminogen was detected using sheep anti-human plas-minogen Abs followed by a rabbit anti-sheep peroxidase-conjugated Abs. The plates were developed with o-phenylenediamine substrate and H 2 O 2 and absorbance at 490 nm was measured.
Cell Lines and Culture Conditions-Epithelial cells were cultivated as described previously (32). Briefly, human A549 cells (lung alveolar epithelial cells, type II pneumocytes, ATCC catalogue number CCL 185) were cultured in Dulbecco's modified Eagle's medium (DMEM) (PAA Laboratories) supplemented with 10% of heat-inactivated FCS (Invitrogen), 2 mM glutamine (PAA Laboratories), penicillin G (100 units/ml) and streptomycin (0.1 mg/ml) (both from HyClone) at 37°C under a 5% CO 2 atmosphere. Human umbilical vein endothelial cells (HUVECs) were obtained from Invitrogen and cultivated in M200 media supplemented with low serum growth supplement (Invitrogen). The cells were used for experiments between passages 1 and 5.
Flow Cytometric Analysis of PepO Binding to Epithelial Cells-Binding of recombinant PepO to A549 epithelial cells and HUVECs was measured using flow cytometry. A549 cells were seeded at a density of 1 ϫ 10 5 cells per well, whereas HUVEC were seeded at 5 ϫ 10 4 cells per well in plain medium on 6-well tissue culture plates (Nunc) and cultivated for 48 h. Confluent monolayers were washed thoroughly and incubated with recombinant PepO protein at the concentration mentioned in figures in culture medium for 3 h at 4°C. Following the incubation, the cells were thoroughly washed with cold PBS ϩ 0.05% FCS and PepO bound to the epithelial cells was stained by incubating with polyclonal rabbit anti-PepO Abs at a dilution of 1:1000 in PBS ϩ 0.05% FCS for 1 h on ice. Thereafter the cells were carefully washed and bound Abs were detected with an Alexa Fluor 488 (green)-labeled goat anti-rabbit Ig (Invitrogen) for 1 h on ice. Finally after washing, the cells were fixed using 1% paraformaldehyde (Sigma) and scrapped from the plate. The flow cytometry analysis was performed using CyFlow space (Partec) to detect the binding of PepO. Cells were detected using log-forward and log-side scatter dot-plot, and a gating region was set to exclude debris and larger aggregates of cells. Fluorescence from 15,000 cells/events were analyzed using logscale amplifications. The geometric mean fluorescence intensity was recorded as a measure for binding activity.
Fluorescence Microscopy-Recombinant PepO attached to host cells were stained using rabbit polyclonal anti-PepO Abs in combination with a secondary goat anti-rabbit IgG coupled with Alexa Fluor 488 (green) (Invitrogen). Post-incubation of cells with PepO, nonspecific binding sites were blocked with 10% FCS and before incubating the cells with PepO Abs (1:100) the cell layer was thoroughly washed with PBS. Bound Abs were detected with an Alexa Fluor 488-labeled goat anti-rabbit Ig conjugate (Invitrogen). The glass coverslips were embedded "upside down" in mounting media (DakoCytomation), sealed with nail polish, and stored at 4°C. A confocal laser scanning microscope (Zeiss LSM 510 META) and the appropriate software was used for image acquisition.
Pneumococcal Host Cell Adherence and Invasion Assay-A549 cells were seeded at a density of 5 ϫ 10 4 cells per well in plain medium either on 24-well tissue culture plates (Nunc) or on glass coverslips (diameter, 12 mm) when assayed by immu-  nofluorescence and cultivated for 48 h. HUVEC cells were seeded at a density of 3 ϫ 10 4 cells per well. Confluent monolayers were washed thoroughly and infected with pneumococci in DMEM supplemented with 1% FCS (infection medium) at 37°C using a multiplicity of infection of 25. The infection assays were carried out for 3 h in a total volume of 500 l after adding the bacteria. Post-infection cells were washed three times with infection media to remove unbound bacteria. The total number of adherent and intracellular recovered bacteria was monitored after detachment and lysis of cells with saponin (Sigma) (1% w/v) and plating the bacteria on blood agar plates. The number of viable intracellular bacteria was quantified by employing the antibiotic protection assay as described (32). Briefly, host cells were infected with pneumococci for 3 h at 37°C under 5% CO 2 . Then, the infected cells were washed and the host cells were incubated for 1 h with infection medium containing 100 g/ml of gentamicin (Sigma) and 100 units/ml of penicillin G (Sigma) at 37°C and 5% CO 2 to kill extracellular and nonadherent pneumococci. Invasive and viable pneumococci were recovered from the intracellular compartments of the host cells by saponin-mediated host cell lysis (1.0% w/v) and the total number of invasive pneumococci was monitored after plating sample aliquots on blood agar plates, followed by colony formation and enumeration. In inhibition experiments, infection assays were carried out in the presence of recombinant PepO protein. Each experiment was repeated at least three times, and results were expressed as mean Ϯ S.D. Whole Blood Killing Assay-Pneumococcal strain D39 and its isogenic pepO-mutant strain were grown, washed, and resuspended in PBS. Diluted cultures (10 l) containing 10 3 cfu were mixed with 250 l of refludan (50 g/ml) (Pharmion) treated fresh human blood and the mixture was rotated for 3 h at 37°C. Viable bacterial counts were determined by plating samples onto blood agar.

PepO Is a Ubiquitously Expressed Pneumococcal Protein-To
determine the presence of the gene encoding for PepO, genomic DNA was extracted from several clinical isolates of pneumococci and subjected to PCR using the pepO gene-specific primers. The results showed the presence of a 1,893-bp long pepO gene in all the tested clinical isolates belonging to different serotypes (Table 1, Fig. 1A). Thereafter we tested whether the pepO gene was expressed. The pneumococcal cell lysates were separated by SDS-PAGE and transferred onto a polyvinylidene difluoride membrane. The presence of PepO protein was detected using polyclonal rabbit anti-PepO Abs. The results showed the presence of a protein band corresponding to the 72-kDa band of PepO in all the tested serotypes (Fig.  1B, upper and lower panels). In addition, flow cytometry analysis showed that anti-PepO antiserum bound to the surface of the S. pneumoniae NCTC10319 strain (Fig. 1C). In addition we checked if the pepO gene was conserved among various pneumococcal strains and other bacterial species. The data clearly indicated that PepO is highly conserved not only within pneumococci but also among other streptococcus species ( Table 2).
Secretion of PepO by S. pneumoniae-Pneumococci express many nonclassical cell surface proteins, such as the glycolytic enzymes ␣-enolase and GAPDH, which lack both the LPXTG motif as well as the signal peptide but are surface exposed and secreted by the bacteria (30,31). Therefore, we tested whether PepO could also be secreted by pneumococci. Collections of pneumococcal isolates, representing different capsule serotype (Table 1) were cultured in THY medium until mid-logarithmic phase and the culture supernatants were collected. The supernatants were subjected to SDS-PAGE and analyzed by Western blotting using polyclonal anti-PepO Abs. PepO was detected in the culture supernatant of all pneumococcal isolates tested ( Fig.  2A). To analyze growth phase-dependent expression of PepO, S. pneumoniae strain NCTC10319 (type 35A) was cultured until stationary phase and culture supernatants were collected at regular time intervals. Results of immunoblot analysis demonstrated that PepO was secreted by S. pneumoniae throughout its growth cycle (Fig. 2, B and C). Furthermore, we could estimate that a 1-ml early mid-logarithmic phase culture of S. pneumoniae strains NCTC10319 and D39 contains about 0.088 Ϯ 0.05 and 0.025 Ϯ 0.02 g of PepO protein, respectively. Taken together, our data indicated that PepO is a ubiquitously expressed secreted pneumococcal protein.
PepO Binds Plasminogen-Microtiter plates coated with recombinant PepO were incubated with increasing concentrations of plasminogen and binding was detected using specific Abs. BSA was used as a negative control. A dose-dependent binding of plasminogen to immobilized PepO was observed (Fig. 3A). Similarly in a reverse setting, a dose-dependent binding of PepO to plasminogen was observed (Fig. 3B). Furthermore, SPR was used to analyze the interaction between PepO and plasminogen (Fig. 3C). Once again a concentration-dependent binding of PepO to immobilized plasminogen was detected (K D ϭ 92 nM). Representative sensorgrams for the interaction are shown and the calculated kinetic values are presented in Table 3. Taken together these data indicated that PepO binds plasminogen with high affinity.
To determine whether interaction between plasminogen and PepO is hydrophobic or ionic in character, the binding assay was conducted in the presence of varying NaCl concentrations. Binding of plasminogen to PepO decreased with increasing NaCl concentrations (Fig. 3D). At a NaCl concentration of 800 mM, plasminogen binding to PepO was reduced by 40% as compared with binding at the physiological concentration of 150 mM NaCl. Furthermore, to analyze whether lysine residues are critical for binding of plasminogen to PepO, the effect of ⑀-ACA, a lysine analog, was investigated. In the presence of increasing amounts of ⑀-ACA as a lysine analog, binding of plasminogen was significantly reduced (Fig. 3E). Thus, binding of plasminogen to S. pneumoniae PepO is weakly influenced by ionic strength and it is mediated by lysine residues.
To corroborate the above results we analyzed the recruitment of plasminogen via PepO from human serum. Microtiter plates coated with PepO were incubated with several dilutions of human serum in binding buffer and binding of plasminogen was detected with specific Abs. A dose-dependent increase in binding of plasminogen, proportional to the serum concentration was obtained (Fig. 3F).
PepO Binds Fibronectin-To test binding of PepO to fibronectin, microtiter plates coated with PepO were incubated with increasing concentrations of human plasma fibronectin followed by detection using specific Abs. A dose-dependent binding of soluble fibronectin to immobilized PepO was observed (Fig. 4A). Additionally, the binding of pneumococcal PepO to immobilized fibronectin was also detected in reverse settings (Fig. 4B). The interaction of fibronectin with immobilized PepO was more readily detectable compared with the binding of PepO to immobilized fibronectin. In addition, the interaction between PepO and fibronectin was analyzed by SPR (Fig. 3C). A concentration-dependent, high-affinity binding of PepO to immobilized fibronectin was detected (K D ϭ 62 nM; Table 3). Representative sensorgrams for the interaction are shown. Taken together, the data confirms that PepO interacts with fibronectin.
Plasminogen Does Not Compete with Fibronectin for Binding to PepO-To address the question whether plasminogen and fibronectin bind simultaneously to PepO or if they compete with each other for the binding sites on PepO, a constant concentration of plasminogen (5 g/ml) was added together with increasing concentrations of fibronectin to microtiter plates coated with PepO. The bound plasminogen and fibronectin were detected using specific Abs. Both plasminogen and fibronectin bound to immobilized PepO and the binding of plasminogen was not affected by the presence of increasing B and C, growth phase-dependent secretion of PepO by S. pneumoniae strain NCTC10319 serotype 35A. Bacteria were cultured in THY media and the culture supernatant was collected at regular intervals (B) and separated by 10% SDS-PAGE (C). After blotting, the PepO protein was detected using rabbit anti-PepO polyclonal Abs. THY media containing no bacteria was used as control and the molecular weight protein marker was used as a reference.
concentrations of fibronectin (Fig. 5A). However, in the reverse setting, the presence of a constant concentration of fibronectin (5 g/ml) together with increasing concentrations of plasmin-ogen, binding of fibronectin was slightly affected (Fig. 5B). At a plasminogen concentration of 20 g/ml, the fibronectin (5 g/ml) binding to PepO was reduced by 37% as compared with FIGURE 3. Plasminogen binding to PepO. Microtiter plates were coated with either PepO (5 g/ml) (A) or plasminogen (5 g/ml) (B) and increasing amounts of plasminogen or PepO was added. Binding was detected using specific polyclonal Abs. BSA was used as negative control. Statistical significance was calculated using two-way analysis of variance and Bonferroni post test. C, binding of PepO to immobilized plasminogen as analyzed by surface plasmon resonance. Increasing concentrations of PepO (22-695 nM) were injected onto plasminogen-coated CM5 sensor chip. The amount of PepO associating with the plasminogen was measured in response units. Representative sensorgrams are presented. D, microtiter plates were coated with PepO and the effect of different concentrations of NaCl on binding of plasminogen (5 g/ml) to PepO was analyzed. Amount of plasminogen bound in the absence of NaCl was set at 100%. Specific polyclonal Abs detected bound plasminogen. BSA was used as negative control. One-way analysis of variance and Dunnett's post test were performed to calculate the statistical difference compared with the binding at 150 mM NaCl. E, the inhibitory effect of the lysine analog ⑀-ACA for binding of plasminogen to PepO was evaluated. Specific polyclonal Abs detected bound plasminogen. The signal obtained in the absence of ⑀-ACA was set to 100% and one-way analysis of variance and Dunnett's post test was performed to calculate statistical difference compared with the binding in the absence of ⑀-ACA. F, binding of plasminogen to PepO from human serum was assessed. Microtiter plates were coated with PepO (5 g/ml) and increasing amounts of serum was added. Bound plasminogen was detected using specific polyclonal Abs. Gelatin was used as negative control. Statistical significance was calculated using two-way analysis of variance test. The data represents the mean Ϯ S.D. of three independent experiments performed in duplicates. ns, not significant; *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. binding in the absence of plasminogen. Taken together, the data suggested that the two proteins do not compete with each other for binding to PepO indicating that plasminogen and fibronectin can bind simultaneously and to nonoverlapping binding sites on PepO. However, the presence of excess of plasminogen may compete with fibronectin for binding to PepO.
Plasminogen Bound to PepO Is Converted to Functionally Active Plasmin That Cleaves the Complement Protein C3b and ECM Components-Plasminogen interacts with complement proteins C3 and C5 and upon activation to plasmin, it inactivates complement at the level of C3 and C5 thereby regulating the complement cascade (38). In accordance, activated plasminogen bound to Haemophilus influenzae Protein E has been shown to cleave C3b thereby facilitating complement evasion (39). We therefore tested whether plasminogen bound to PepO could be readily activated by its activator uPA and converted to the active protease plasmin. Thus, PepO-bound plasminogen was incubated with uPA and the generated plasmin activity was evaluated by incubating with the chromogenic substrate S-2251. A time-dependent cleavage of the substrate by the generated plasmin was observed (Fig. 6A). Interestingly, no cleavage of the substrate was observed when the plasminogen was incubated with PepO without addition of uPA or in the presence of only uPA, demonstrating the specific protease activity of the generated plasmin.
In addition, we tested the ability of plasminogen bound to PepO to cleave its natural substrate fibrinogen. To this end, plasminogen bound to immobilized PepO was incubated with uPA and substrate fibrinogen. At the indicated time points, sample aliquots were taken out and separated by SDS-PAGE and Western blotting was performed to detect fibrinogen fragments. A time-dependent cleavage of fibrinogen by the generated plasmin was observed (Fig. 6B).
Furthermore, we assayed whether PepO-bound plasminogen can also affect complement. For this, plasminogen bound to immobilized PepO was incubated with uPA and 125 I-labeled C3b and following the incubation, samples were separated by SDS-PAGE and C3b cleavage products were analyzed. A timedependent cleavage of C3b by activated plasminogen (plasmin) was observed (Fig. 6C). Following the incubation two major bands of about 40 and 30 kDa were obtained. Importantly, no C3b cleavage was observed upon incubation with uPA (Fig. 6C), suggesting that the PepO-bound plasminogen cleaved C3b when activated with uPA.  In addition to its ability to cleave fibrin clot and C3b, protease plasmin also acts on ECM components. Therefore, we tested whether plasminogen bound to PepO upon activation could also cleave ECM components such as fibronectin and laminin. Compared with the cleavage of fibronectin by protease plasmin (Fig. 6D), which was readily observed, laminin cleavage was present but weak (Fig. 6E) after 18 h of incubation. Taken together, plasminogen bound to PepO is readily activated to the active protease plasmin.
Role of PepO in Pneumococcal Binding to Fibronectin and Plasminogen-To elucidate the role of PepO in pneumococcal binding to fibronectin, pepO-mutants of pneumococci were generated by insertional inactivation. The expression of PepO in wild-type strain D39 and NCTC10319 and their respective isogenic mutants (⌬pepO) were analyzed in the whole cell lysates (Fig. 7A) and culture supernatants (Fig. 7B). The immunoblots revealed a successful disruption of PepO in the mutants (Fig. 7, A and B). Phosphoglycerate kinase was used as a control (Fig. 7, A and B) (40). A binding assay with FITC-labeled pepOmutants in strain D39 and NCTC10319 background indicated that although PepO of the D39 strain contributes significantly to pneumococcal binding to immobilized fibronectin no inhibition was observed with the NCTC10319 pepO-mutant (Fig.   7C). Compared with their respective wild-type strains, the pepO-mutants in D39 background showed 27.2 Ϯ 6.9% binding, whereas the NCTC10319 pepO-mutant showed a 74.6 Ϯ 26.4% binding to immobilized fibronectin (Fig. 7C). The data suggested a strain-dependent interaction of PepO with fibronectin. Similarly, binding of plasminogen to the wild-type and corresponding pepO-mutant strain was analyzed by ELISA. A dose-dependent binding of plasminogen to the wild-type D39 and NCTC10319 strain was observed. Moreover, a significant reduction in plasminogen binding to the pepO-mutant in both D39 and NCTC10319 background was observed at a plasminogen concentration of 10 g/ml. Although a minor reduction in plasminogen binding was observed at other concentrations, it was not significant.
Recombinant PepO Protein Interacts with Host Epithelial Cells-We found that recombinant PepO protein bound to epithelial cells. As shown in Fig. 8A, flow cytometric analysis revealed a dose-dependent binding of recombinant PepO with A549 lung epithelial cells. To corroborate these data, immunofluorescence staining of PepO bound to the host cells was performed and confocal images were taken. Fig. 8B demonstrates the binding of PepO (Alexa-488 stained, green) with A549 cells.
PepO Mediates Pneumococcal Adherence and Invasion of Host Epithelial Cells-To analyze the role of PepO in pneumococcal adherence and invasion of host cells, cell culture infection experiments were performed. The A549 human lung epithelial cells were infected for 3 h with wild-type NCTC10319 or D39 strains and their respective isogenic pepO-mutant (⌬pepO). Although, wild-type D39 bacteria showed significant levels of adherence and invasion into these host cells, PepOdeficient bacteria showed a drastic reduction in adherence and invasion (Fig. 8, C and D). Surprisingly, disruption of the gene encoding PepO in NCTC10319 background had no effect on the adhesive capacity of the bacteria compared with its wildtype strain (Fig. 8C), whereas compared with its wild-type NCTC10319 strain, the pepO-mutant showed a reduction of 30% in the number of recovered intracellular pneumococci (Fig.  8D). Due to the presence of a thick capsule in S. pneumoniae D39, the number of adhered and recovered D39 (28 Ϯ 4 ϫ 10 2 CFU/well) was relatively low compared with the NCTC10319 strain (12.57 Ϯ 1.96 ϫ 10 4 cfu/well), nevertheless, a clear effect was observed for the D39 strain. Because PepO directly interacts with host cells, the ability of recombinant PepO in blocking pneumococcal adherence was investigated. The A549 epithelial cells pre-treated with PepO protein were infected with NCTC10319 and D39 bacteria. The ability of the pneumococci to adhere to the epithelial cells was reduced by 48 and 44% for NCTC10319 and D39 background, respectively (Fig. 8E). Taken together, the data suggested that PepO, in a strain-dependent manner, facilitates pneumococcal attachment and internalization of host epithelial cell.
Pneumococcal Adherence and Invasion of Endothelial Cells Is Facilitated by PepO-Our data demonstrated that PepO interacts with host epithelial cells and facilitates pneumococcal strain-dependent adherence and invasion of host cells. Therefore, to investigate if a similar mechanism is also relevant for endothelial cells, first, binding of recombinant PepO protein to HUVECs was assayed. Similar to A549 epithelial cells (Fig. 8A), flow cytometric analysis revealed a dose-dependent binding of the PepO protein to HUVECs (Fig. 9A). To assess the role of PepO in pneumococcal adherence and invasion, HUVECs were infected for 3 h with S. pneumoniae D39 strain and its isogenic pepO-mutant. Although wild-type bacteria showed significant levels of adherence to and invasion into endothelial cells, PepOdeficient bacteria demonstrated a significant reduction of 97 and 99% in adherence and invasion, respectively (Fig. 9, B and  C). In addition, we explored the ability of recombinant PepO protein to block pneumococcal adherence to HUVECs. The pre-treatment of HUVECs with recombinant PepO resulted in 43% reduction in the adherence of D39 strain (Fig. 9B). In conclusion, our data from the cell culture infection assays indicated that PepO modulates pneumococcal attachment and internalization by host cells.
Influence of PepO on Bactericidal Activity of Human Blood-Bactericidal assay was performed to investigate the function of PepO during systemic infections. Approximately 10 3 cfu of S. pneumoniae D39 and its isogenic pepO-mutant were incubated separately with 250 l of human whole blood, and viability was determined by plating and enumeration of the obtained colonies. The results indicate significant killing of both strains in whole blood after 3 h of incubation. However, compared with the ϳ4% survival of the wild-type D39 strain, the pepO-mutant strain showed only 0.4% survival efficiency (Fig. 9D). The results indicate the important role of PepO in pneumococcal infections.

DISCUSSION
In this study we have identified a 72-kDa pneumococcal protein PepO as a plasminogen and fibronectin-binding protein that facilitates pneumococcal invasion of host epithelial cells. PepO is a predicted metallo-endopeptidase that shares homology with the peptidase family M13 proteins. These M13 peptidases are present in a wide range of organisms including mammals and bacteria (41). The M13 family includes neprilysin (neutral endopeptidase), endothelin-converting enzyme I, erythrocyte surface antigen KELL, phosphate-regulating gene with homologies to endopeptidase on the X chromosome, soluble secreted endopeptidase, and damage-induced neuronal endopeptidase (42,43). In mammals, these peptidases have been implicated in processes such as blood pressure regulation, nervous control of respiration, regulation of the function of neuropeptides in the central nervous system, and cardiovascu- FIGURE 6. Plasminogen bound to PepO is functionally active. A, PepO (5 g/ml) was immobilized on a microtiter plate. After blocking it was incubated with plasminogen (Plg, 1 g/well) in the absence or presence of the activator uPA together with the chromogenic substrate S-2251. Measuring the absorbance at 405 nm assessed conversion of the substrate by the generated plasmin. The mean values of three independent experiments and the S.D. values are indicated. B, degradation of the natural substrate fibrinogen (Fbg) by plasmin(ogen) bound to immobilized PepO. Plasminogen (5 g/well) was bound to immobilized PepO (5 g/well) and after washing, uPA (10 units/well) together with fibrinogen (5 g/well) were added. At the indicated time points, sample aliquots were removed, separated by SDS-PAGE, transferred to a membrane, and degradation of fibrinogen was assayed by Western blotting using a rabbit fibrinogen antiserum and a peroxidase-conjugated secondary Abs. Fibrinogen incubated in the absence or presence of uPA alone were used as control. Representative data from three independent experiments are shown. PepO-bound plasminogen cleaves complement protein C3b (C), ECM components fibronectin (Fbn) (D), or laminin (E). Microtiter plates coated with PepO (5 g/well) were incubated with plasminogen (5 g/well) followed by addition of uPA (10 units/well) and 125 I-C3b, fibronectin, or laminin (100 kcpm) and incubation at 37°C. Samples were taken at the indicated time points for C3b degradation and after an 18-h incubation for fibronectin and laminin and then separated by SDS-PAGE. For C3b degradation, the positive control (ϩve Ctrl) contained Factor H mixed with Factor I, 125 I-labeled C3b, whereas in the negative control (Ϫve Ctrl) Factor I was omitted. An arrow marks the cleavage products of C3b. Representative data from three independent experiments are shown. lar development. These are typically type II membrane-anchored enzymes, known to activate or inactivate oligopeptide (pro)-hormones. In bacteria they have been implicated in milk protein cleavage (44).
In S. pneumoniae, PepO was found in all the tested strains at both the gene and protein levels, suggesting it to be conserved among the strains. Moreover, within various strains of pneumococci PepO shows 99 -100% identity at the gene level. Importantly, the expression of PepO was independent of the serotype indicating its ubiquitous expression. Although the gene sequence of PepO lacks both membrane spanning domains, such as the LPXTG motif region, a characteristic feature shared by many surface-exposed proteins, and the typical signal sequences, apparently PepO was found on the cell surface and in culture supernatants, suggesting PepO is a secreted protein. However, the exact mechanism of PepO secretion or presence into the culture supernatant is still not clear. Interestingly, in pneumococci a number of proteins lacking any signal peptide required for secretion and peptidoglycan-anchoring motifs are transported to and associated with bacterial surface structures. These include the glycolytic enzyme such as ␣-enolase and GAPDH (30,31) or choline-binding domain containing proteins such as PspC, which binds the phosphorylcholine on the pneumococcal cell wall (45).
In this study we found that PepO binds plasminogen in a dose-dependent manner influenced by ionic strength. In addition, a dose-dependent inhibition with the lysine analog ⑀-ACA suggests that lysine residues are relevant for interaction between plasminogen and PepO. This type of interaction is similar to that observed for pneumococcal enolase and plasminogen (46). Similarly interactions of plasminogen with other pathogen proteins such as H. influenzae Protein E, Pseudomonas aeruginosa Tuf, Candida albicans Pra1 and Gmp1, and Borrelia burgdorferi CRASP depends on lysine residues and is ionic strength dependent (39,(47)(48)(49)(50). The specificity of the interaction between PepO and plasminogen was demonstrated in our SPR studies and binding of plasminogen from human serum. Moreover, the pulldown using biotin-labeled plasminogen and co-immunoprecipitation studies using unlabeled plasminogen further confirmed the PepO-plasminogen interactions (data not shown). In addition, we demonstrated that plasminogen bound to PepO was readily activated by uPA to plasmin that not only cleaved its natural substrate fibrinogen but also ECM components such as fibronectin and laminin. In addition it also cleaved C3b, thereby mediating complement control. A similar mechanism has been reported for H. influenzae Protein E-bound plasminogen (39). Interestingly, PepO alone did not cleave plasminogen (data not shown). PepO being a secreted protein may play a role, which can be different from the other cell-surface bound plasminogen-binding proteins. A similar role has been suggested for Skizzle, a secreted protein of group B Streptococcus that binds and enhances urokinase-catalyzed activation of plasminogen (51). Taken together, our data suggests PepO to be an additional component that contributes toward the pneu- FIGURE 7. Effect of PepO deficiency on pneumococcal binding to fibronectin and plasminogen. Immunoblot analysis of PepO production in the wild-type (WT) strains D39 and NCTC10319 or their respective isogenic pepO-mutants (⌬pepO) using whole bacterial lysate (A) or culture supernatant (B). The presence of PepO was detected using rabbit anti-PepO polyclonal Abs. Pneumococcal phosphoglycerate kinase (PGK) was used as a loading control. C, binding of pneumococci to immobilized fibronectin. Microtiter plates were coated with fibronectin (5 g/ml) and binding of FITC-labeled S. pneumoniae D39, NCTC10319 wild-type, or their respective isogenic pepO-mutants was assessed. Binding was measured at 485/535 (excitation/emission). The data represents the mean Ϯ S.D. of three independent experiments performed in duplicates. Statistical significance was calculated using two-way analysis of variance and Bonferroni post-test. ***, p Ͻ 0.001. D, binding of plasminogen to pneumococci. Microtiter plates were coated with 50 l of 10 8 cfu/ml of S. pneumoniae D39, NCTC10319 wild-type, or their respective isogenic pepO-mutants strains and binding of plasminogen (0, 2.5, 5 and 10 g/ml) was assessed. Bound plasminogen was detected using specific polyclonal Abs. The data represent the mean Ϯ S.D. of three independent experiments performed in duplicates. Statistical significance was calculated using one-way analysis of variance and Tukey post test. *, p Ͻ 0.05; ***, p Ͻ 0.001. mococcal pathogenesis. Interestingly, PepO did not bind laminin, collagen type II, or vitronectin (not shown).
At par with other Gram-positive bacteria such as Streptococcus pyogenes and Staphylococcus aureus, pneumococci also express a number of fibronectin-binding proteins such as PfbB, PavB, PavA, and PfbA (20 -24). Our data indicated that PepO binds fibronectin, in addition to plasminogen. Both plasminogen and fibronectin are abundant plasma proteins and it seemed highly possible that they may compete with each other for binding to PepO. However, our results demonstrate that binding of plasminogen is not outcompeted by fibronectin indicating that plasminogen and fibronectin interact with different motifs on pneumococcal PepO. Apparently, the presence of excess plasminogen seems to compete with fibronectin, suggesting that the binding sites may be present in close proximity resulting in this minor competition.
Our data demonstrate PepO as a secreted protein, which is also present on the bacterial surface and plays an important Host cells were incubated with increasing amounts of recombinant PepO, after washing the cells, binding was detected with rabbit anti-PepO Abs followed by Alexa 488-conjugated secondary Abs. Binding of protein was quantified by flow cytometry (A) or detected by immunofluorescence microscopy (B). Representative flow cytometry data from three independent experiments are shown. Bar represents 10 m. C, adherence of pneumococci NCTC10319 and D39 and their respective isogenic pepO-mutants was determined by counting the cfu per well obtained from sample aliquot plated onto the blood agar plate after 3 h of infection. D, invasion and intracellular survival of pneumococci were determined by the antibiotic protection assay. E, recombinant PepO protein inhibits pneumococcal adherence to A549 cells. Epithelial cells were incubated with recombinant PepO for 30 min prior to infections. The total number of bacteria associated with host cells was determined after removing unbound extracellular bacteria and plating the cells on blood agar plates. The data represents the mean Ϯ S.D. of three independent experiments performed in duplicates. Statistical significance was calculated using Student's t test. ns, not significant; **, p Ͻ 0.01; ***, p Ͻ 0.001. role in pneumococcal survival in whole blood. Although PepO appeared to significantly contribute toward the binding of whole bacteria to fibronectin, the ability of PepO to bind host proteins depends on the pneumococcal strain. This was evident by the fact that only the pepO-mutant of the highly encapsulated D39 serotype-2 strain showed a significant reduction in binding to fibronectin compared with the wild-type. Interestingly, the adhesive capacity of the pepOmutant of NCTC10319 type 35A strain was not abrogated compared with its wild-type strain. This could be due to masking of the PepO effect, by other known and yet unknown surface-exposed fibronectin-binding proteins in NCTC10319 background, which has a relatively thin capsule compared with the D39 strain. Moreover, a similar strainspecific effect was demonstrated for PavB, a plasminogenand fibronectin-binding protein in pneumococci (21). Additionally these fibronectin-binding proteins have been shown to function as adhesins facilitating pneumococcal adherence and invasion of host cells (21,22,24). In accordance, our data indicates that PepO facilitates pneumococcal adhesion and invasion of host cells. A similar effect was observed for both epithelial and endothelial cells suggesting it to be cell-type independent phenomenon. Interestingly, similar to the binding to fibronectin, the role of PepO in bacterial adherence and invasion was also strain dependent, as the inhibition was observed only for the pepO-mutant in D39 background. On the contrary, the PepO-deficient NCTC10319 strain showed only 30% reduction in invasion, without affecting the total adherence compared with the wild-type strain. Interestingly, a similar role has been associated to the endopeptidase O from Porphyromonas gingivalis that has been demonstrated to play a role in invasion of host epithelial cells but not adherence (52).
In addition, our direct binding studies with soluble PepO protein and host cells showed a dose-dependent binding, suggesting that PepO may interact directly with a cellular receptor(s). Consistent with this, the presence of soluble PepO in blocking experiments significantly inhibited pneumococcal adherence to host cells. The data thus demonstrate that the PepO interaction with host cells facilitates bacterial colonization and subsequent internalization.
In conclusion, ubiquitously expressed PepO has the potential to be a new virulence factor facilitating in the pneumococcal infections. Besides having a plasminogen-and fibronectinbinding activity, PepO is a novel protein facilitating pneumococcal adherence and internalization of host epithelial and endothelial cells. Moreover, the ability to acquire plasminogen and allow complement control at the C3 level by activated plasmin represents an immune and complement escape mechanism.