Biochemical Journal

Research article

SCM, a novel M-like protein from Streptococcus canis, binds (mini)-plasminogen with high affinity and facilitates bacterial transmigration

Marcus Fulde, Manfred Rohde, Angela Hitzmann, Klaus T. Preissner, D. Patric Nitsche-Schmitz, Andreas Nerlich, Gursharan Singh Chhatwal, Simone Bergmann

Abstract

Streptococcus canis is an important zoonotic pathogen capable of causing serious invasive diseases in domestic animals and humans. In the present paper we report the binding of human plasminogen to S. canis and the recruitment of proteolytically active plasmin on its surface. The binding receptor for plasminogen was identified as a novel M-like protein designated SCM (S. canis M-like protein). SPR (surface plasmon resonance) analyses, radioactive dot-blot analyses and heterologous expression on the surface of Streptococcus gordonii confirmed the plasminogen-binding capability of SCM. The binding domain was located within the N-terminus of SCM, which specifically bound to the C-terminal part of plasminogen (mini-plasminogen) comprising kringle domain 5 and the catalytic domain. In the presence of urokinase, SCM mediated plasminogen activation on the bacterial surface that was inhibited by serine protease inhibitors and lysine amino acid analogues. Surface-bound plasmin effectively degraded purified fibrinogen as well as fibrin clots, resulting in the dissolution of fibrin thrombi. Electron microscopic illustration and time-lapse imaging demonstrated bacterial transmigration through fibrinous thrombi. The present study has led, for the first time, to the identification of SCM as a novel receptor for (mini)-plasminogen mediating the fibrinolytic activity of S. canis.

  • mini-plasminogen
  • M-like protein
  • transmigration
  • Streptococcus canis

INTRODUCTION

Streptococcus canis belongs to Group G streptococci and represents a common member of the resident microflora of the skin and mucosa of various animals [1]. It is considered an opportunistic zoonotic pathogen capable of causing serious invasive diseases, such as streptococcal toxic shock syndrome, necrotizing fasciitis, septicaemia and meningitis in domestic animals [2], as well as humans [35]. In spite of the emerging importance of S. canis as a health hazard, little is known about its molecular pathogenesis. Many Gram-positive pathogens which can cause invasive diseases in humans exploit the host proteins to invade and disseminate in tissues [6]. The exclusively human pathogenic bacterium Streptococcus pyogenes uses the host fibrinolytic system and interacts with PLG (plasminogen) [7]. PLG is the 92 kDa single-chained pro-enzyme of the broad-spectrum serine protease plasmin. It comprises a pre-activation peptide of ~8 kDa, five homologous disulfide-bonded triple-loop kringle structures (K1–K5; 65 kDa), and a serine-protease domain (25 kDa) [8]. PLG is cleaved by leucocyte elastase activity into mPLG (mini-PLG) comprising K5 and the catalytic domain (Val442–Asn790) [9]. Conversion of the single-chained pro-enzyme into active plasmin is mediated by proteolytic activation via the mammalian PLG activators, tPA (tissue-type PLG activator) and uPA (urokinase-type PLG activator) [10]. Plasmin is involved in intravascular fibrinolysis [11], degradation of ECM (extracellular matrix) material, and is relevant for cell invasion [12]. The capability of bacteria to bind and activate PLG to plasmin thereby represents a mechanism to invade and disseminate the host. One of the major PLG-binding components of S. pyogenes is the M-like protein PAM which mediates high-affinity binding to K2 [13,14].

The present paper describes the identification of a novel M-like protein from S. canis, designated SCM, which binds PLG. Detailed molecular characterization indicated a specific interaction of SCM with the C-terminal part of PLG comprising mPLG. This kind of interaction may represent a novel mechanism for PLG interactions different from PAM and other PLG-binding components of streptococci. Furthermore, PLG activation and the recruitment of plasmin on the bacterial surface of SCM-expressing S. canis led to the degradation and dissolution of fibrin thrombi, indicating a contribution of PLG binding to bacterial dissemination within the host.

EXPERIMENTAL

Bacterial strains, media and growth conditions

The strains of β-haemolytic Groups A, C and G streptococci used in the present study are listed in Table 1. Bacteria were routinely cultivated in TSB (tryptic soy broth) at 37 °C without shaking. Escherichia coli were grown in Luria–Bertani medium. If indicated, antibiotics were added at the following concentrations: ampicillin (100 μg·ml−1) and kanamycin (25 μg·ml−1).

View this table:
Table 1 Bacterial strains

Proteins and sera

Human PLG (Sigma), kringle domains 1–3 (K1–3, Sigma), and angiostatin (K1–K4, Hemochrom Diagnostica) were purchased commercially. Blood plasma from pig, goat, cat and dog was collected at the Veterinary School Hannover; K4 and mPLG were prepared as described previously [15]. PLG-specific antibodies and HRP (horseradish peroxidase)-conjugated anti-goat and anti-mouse secondary antibodies were purchased from Dako. Purified His-tagged SCM was used to raise polyclonal antibodies in BALB/C mice according to standard protocols. Pre-immune serum was collected before immunization. Institutional guidelines on procedures involving animals and their care are fully in compliance with German animal welfare acts.

Expression cloning and recombinant DNA techniques

If not stated otherwise, all enzymes were obtained from New England Biolabs. SCM was amplified from the S. canis G361 genome using the primer pair MF-1/MF-2 (5′-GGGGGATTCAACAGAGTTACTGAGGCCAGAGC-3′/5′-CCCGTCGACTGCTGTGAAGAATGGGTTGG-3′), incorporating a BamHI and a SalI restriction site respectively. Subsequent enzymatic restriction and ligation into the BamHI/SalI-digested pQE30 vector (Qiagen) resulted in the plasmid pQE30-SCM. The integrity of insert DNA was verified by sequence analysis using ABI Prism dye terminator cycle sequencing (PerkinElmer). The N-terminal part of SCM (SCM 25KD) was amplified with the primer pair MF-1/Sal_fog-S-region (5′-GGGGGATTCAACAGAGTTACTGAGGCCAGAGC-3′/5′-GCTGTCGACTTATTATGCTTTGTCGCTTGCTAATTGTTC-3′) and subcloned as a BamHI/SalI fragment into pQE30. Sequence analysis revealed SCM 25KD as the N-terminal part comprising amino acids 37–248.

Recombinant proteins were obtained by chromatography under native conditions on Ni-NTA (Ni2+-nitrilotriacetate) resins according to the manufacturer's protocols (Qiagen).

Amplification of the SCM protein from different S. canis strains was performed using the primer pair all-canis_fwd/all-canis_rev (5′-TAGCGTTGGAACAGCATCAC-3′/5′-CTGCCTCTGCTTTTGCTTTC-3′).

CD spectroscopy

CD spectra were recorded with a Jasco-J-815 spectrometer using a quartz cell of 1 cm optical path-length, an instrument scanning speed of 100 nm/min and over a wavelength range of 195–260 nm at 20 °C. The reported CD spectrum of each sample represents the average of 10 scans taken at 20 °C. PBS was used as a blank. The helical content of the protein was calculated using the Jasco software.

Heterologous expression of SCM protein in Streptococcus gordonii strains

S. gordonii heterologously expressing SCM on its surface was generated using the mutagenesis system described by Oggioni and Pozzi [16]. Full-length SCM was amplified from the S. canis G361 genome as described above and cloned into the integration vector pSMB103. Transformation of the plasmid in competent S. gordonii GP1221 was performed as described by Talay et al. [17].

Binding experiments and competitive-inhibition assays

Purified PLG (Sigma) was radiolabelled with 125I using a standard chloramine T method [18], and binding experiments with 125I-PLG were performed as described previously [19]. Briefly, 2.5×108 bacteria were incubated with 20 nCi of iodinated PLG or iodinated K1–3. Bacteria were sedimented and bacteria-bound radioactivity was measured using a γ-counter (Packard). In competitive-inhibition assays, binding to viable streptococci was measured in the presence of increasing molar excesses of glu-PLG (Sigma), the lysine amino acid analogue 6-AHA (ϵ-amino caproic acid; Merck) or K1–3 (Sigma). For dot-blot experiments, purified SCM and SCM 25KD respectively, were spotted on to a nitrocellulose membrane (Bio-Rad). Blocking was performed in the presence of 5% (w/v) skimmed milk for 2 h at room temperature (22 °C). Hybridization with 125I-PLG was carried out for 4 h. Binding was detected using X-ray films (Kodak).

SPR (surface plasmon resonance)

The association and dissociation reactions of glu-PLG (Sigma) and mPLG to recombinant SCM and its N-terminal derivative respectively, were analysed in the BIAcore optical biosensor (BIAcore 2000 system) using CM 5 sensor chips. Covalent immobilization of PLG and mPLG was performed using a standard amine coupling procedure essentially as described previously [20]. Additionally, SCM was immobilized on the chip surface and PLG or mPLG respectively, were used as analytes. Binding analysis was performed in HBS BIAcore running buffer [10 mM Hepes, 150 mM NaCl, 1.4 mM EDTA and 0.05% Tween 20 (pH 7.4)] at 20 °C using a flow rate of 30 μl·min−1 in all experiments. The affinity surface was regenerated with 10 μl of 20 mM NaOH. Binding was assayed at least in duplicate using independently prepared sensor chips.

Analysis of BIAcore sensogram data

The interaction kinetics were analysed from raw data of the BIAcore sensograms suitable for analysis using the kinetic models included in the BIAevaluation software version 3.0. The experimental data were fitted globally by using the simple one-step bimolecular association reaction (1:1 Langmuir kinetic: A+B↔AB). For each evaluation, a minimum of four data sets were analysed.

Plasma absorption, Western blot and immunoblot analysis

Plasma proteins were eluted from the streptococcal surface using 10 mM glycine (pH 2.0). After neutralization with 2 M NaOH, proteins were subjected to SDS/PAGE (12% gels) according to the method described by Laemmli [21] and either stained with Coomassie Brilliant Blue or subsequently transferred on to a nitrocellulose membrane (Bio-Rad) using a semi-dry blotting system. In immunoblot analysis, PLG binding was detected by incubation of the membrane with monoclonal antibodies directed against human PLG (1:300, American Diagnostica), followed by a secondary antibody coupled to HRP. Peroxidase activity was detected by chemoluminescence using 100 mM Tris/HCl, 1.25 mM 3-aminopthalhydrazide, 225 μM p-coumaric acid and 0.01% H2O2 at pH 8.8 in water and exposure to chemoluminescence films (GE Healthcare).

For immunoblot analysis, human PLG and its fragments K1–3, K4, K1–4 and mPLG were spotted on to a nitrocellulose membrane at the amounts indicated. Membranes were blocked as described above and incubated with 100 μg of purified SCM overnight at 4 °C. Immunoblot analysis with anti-M-protein (1:200) antiserum was performed using HRP-conjugated secondary antibodies (1:3000) followed by incubation with the substrate solution containing 1 mg·ml−1 4-chloro-1-naphthol and 0.1% H2O2 in PBS.

Plasmin activity assay

Activation of plasmin was monitored in 96-well microtitre plates. For detection of secreted streptokinase activity, 150 μl of a 50 ml overnight culture was incubated with the same volume of canine or human plasma respectively, for 30 min at 37 °C. The chromogenic substrate D-valyl-leucyl-lysine-p-nitroanilide dihydrochloride (S-2251, Fluka) was added to a final concentration of 400 μM. Proteolytic cleavage was detected photometrically at 405 nm using an ELISA reader (Tecan Sunrise) following a time course of 120 min. For detection of cell-bound streptokinase activity, S. canis bacteria [109 cfu·ml−1 (cfu is colony-forming units)] were pre-incubated with 40 μg·ml−1 PLG in PBS for 15 min at 37 °C. After removal of unbound PLG by washing, urokinase (500 ng) was added. Plasmin activation was also measured in the presence of 2 units of the plasmin inhibitor aprotinin or the lysine analogue 6-AHA at concentrations of 2, 40 and 80 μM. PLG-incubated streptococci without activator and non-treated bacteria were used as controls. For the analysis of plasmin activation time of surface-bound compared with soluble PLG, SCM-expressing SGO bacteria were treated essentially as described above for radioactive-binding assays. Briefly, streptococci were incubated with 55 ng of PLG for 1 h at room temperature. Then, cells were washed with PBS and 50 ng of uPA was added. The amount of soluble PLG was calculated from binding experiments as 80% of the iodinated protein. Conversion into plasmin was determined as described above.

Degradation of fibrinogen

S. canis G361 bacteria were pre-incubated with PLG as described, and 5×107 cells were incubated with 4 μg of PLG-depleted human fibrinogen (Sigma) following a time series of 5, 10, 15, 20, 30 and 60 min at 37 °C. Degradation was monitored in the presence of uPA (500 ng) and the reaction was stopped with 50 μl of SDS-containing sample buffer. Bacteria were sedimented and samples were separated by SDS/PAGE followed by transfer of proteins on to PVDF (Immobilon-P, Millipore) using a semi-dry blotting system. The membranes were blocked as described above prior to incubation with rabbit antiserum directed against human fibrinogen (Dakopatts, 1:2000). Detection of fibrinogen peptides was carried out using HRP-conjugated anti-rabbit antibodies (1:3000) and chemiluminescence as described above.

Degradation of fibrin and transmigration of S. canis through fibrin

A fibrin matrix was produced on coverslips and transwell cell-culture inserts (polycarbonate membranes with a 3.0 μm pore size, Costar) by incubating 100 μl of 50 mg·ml−1 PLG-depleted human fibrinogen (Millipore) in PBS with 2 μl of 1.0 KU·ml−1 thrombin (from bovine plasma, MP Biomedicals) for 10 h at 37 °C. S. canis G361 pretreated with PLG as described above, was applied with a dose of 1×107 in 200 μl of PBS to the fibrin matrix. Surface-bound PLG was activated with uPA (500 ng). Bacterial transmigration from the upper to the lower chamber was quantified by plating serial dilutions of the lower chamber solution on blood agar plates. Experiments were carried out for up to 6 h and the samples were plated at intervals of 120 min. Data represent the means±S.D. of one experiment performed in triplicate. Experiments were repeated three times.

Time-lapse microscopy

Fibrin clots were generated in μ-Slides (μ-Slide 8-well, Ibidi) as described above. Bacteria (1×107) were added per well and samples were mounted on an inverted microscope (Axio Observer.Z1) equipped with a 25×/0.8 NA (numerical aperture) LCI Plan-Neofluar objective driven by AxioVision Software 4.7 (Zeiss). Time-lapse imaging was performed at 37 °C and frames were taken every 2 min. Video frames were processed for contrast and brightness using Fiji/ImageJ.

Electron microscopy

Samples were fixed in 5% formaldehyde and 2% glutaraldehyde in cacodylate buffer [0.1 M cacodylate, 0.01 M CaCl2, 0.01 M MgCl2 and 0.09 M sucrose (pH 6.9)] for 1 h on ice and washed with TE-buffer [20 mM Tris/HCl and 1 mM EDTA (pH 7.0)] before dehydrating in a graded series of acetone (10, 30, 50, 70, 90 and 100%) on ice for 15 min at each step. Samples were then subjected to critical-point drying with liquid CO2 (CPD 30, Bal-Tec). Dried samples were covered with a gold film by sputter coating (SCD 500, Bal-Tec) before examination in a field-emission scanning-electron microscope Zeiss DSM 982 Gemini using the Everhart Thornley SE detector and the SE-inlens detector at a 50:50 ratio with an acceleration voltage of 5 kV. Images were recorded on to an MO-disk, and contrast and brightness was adjusted applying Adobe Photoshop CS3.

Computational analysis

Prediction of Gram-positive signal sequences was carried out using the SignalP program (http://www.cbs.dtu.dk/services/SignalP). Prediction of fibrillar structure was performed in silico using the coiled–coil program (http://www.ch.embnet.org/software/COILS).

RESULTS

PLG binding to GCGS (Group C and G streptococci)

During the course of wound infection, streptococci encounter different components from host blood and tissue. Recruitment of human plasma components to bacterial S. canis isolates G1, G2, G14, G361 and Streptococcus equisimilis C90 has been analysed after plasma incubation. Coomassie Blue staining detected the presence of a 90 kDa protein after elution of surface-bound plasma proteins from G361, C90, and to a lesser extent, G1 (Figure 1A, black arrow). Immunoblot analysis with antibodies directed against PLG identified the respective protein as human PLG (Figure 1B). In addition, S. canis isolate G361 can also recruite PLG from plasma derived from different mammalian species including pig, goat, cat and dog (Figure 1C, black arrow). In radioactive-interaction studies, binding of iodinated human PLG was detected for S. canis isolates G1, G8, G13, G15 and G361 ranging from 16% binding for strain G8 up to 60% for S. canis G361 (Figure 1D). Comparative PLG-binding activity was determined for PAM-expressing S. pyogenes A158 (results not shown), whereas only marginal PLG binding has been observed for the isolates G2, G14 and G17 (Figure 1D). These results demonstrate a specific interaction of zoonotic S. canis isolates with human PLG and a non-species-specific recruitment of PLG from plasma on the cell surface of S. canis.

Figure 1 PLG recruitment to S. canis and S. equisimilis after incubation with human plasma or purified radiolabelled PLG

(A) Coomassie Blue stain of plasma proteins eluted from the streptococcal surface after plasma incubation. The arrow indicates the 90 kDa PLG protein. (B) Immunoblot analysis with PLG-specific antibodies after incubation of different S. canis (G1, G2, G14, G361) and S. equisimilis (C90) strains with human plasma. (C) Coomassie Blue stain of PLG recruitment to the human S. canis isolate G361 after incubation with plasma derived from pig (swine), goat (caprine), cat (feline), dog (canine) and human. (D) Binding of 125I-labelled PLG to streptococci. PLG binding was performed in triplicate and was calculated as a percentage of employed iodinated PLG. In (A, B and C) the molecular mass (MW) in kDa is indicated on the left-hand side.

Identification of SCM as an M-like protein of S. canis mediating PLG binding

M- and M-like proteins of human Group A, C and G streptococci have been identified as PLG receptors [22]. Based on a DNA sequence which has been supposed to encode an M-like protein of S. canis (GenBank® accession number FJ594772) and an ongoing genome sequencing project, the respective gene was amplified from the genome of strain G361 and termed SCM. The amino acid sequence of the purified His-tagged protein elucidates a N-terminal signal sequence with a predicted cleavage site between alanine and glutamic acid residues (VKAEH) and a LPxTG motif for sortase-mediated covalent cell-wall attachment at the C-terminus (Figure 2A). Using the prediction program COILS, a probability for coiled–coil formation of more than 90% was calculated for amino acids 109–360 of full-length SCM. In further studies, CD spectroscopy revealed a typical curve progression for α-helical proteins with two minima at 208 and 222 nm respectively. The helical content was estimated as 72.4%. Furthermore, a 222/208 nm ratio of 1.13 was calculated (Figure 2B). Since this ratio is ≥1 for coiled–coil proteins [23,24], these data suggest a dimeric form of SCM under physiological conditions. In addition, a typical feature of coiled–coil proteins is a repeating heptad of the generalized sequence a-b-c-d-e-f-g, where positions a and d are occupied by hydrophobic residues [25]. An alignment with other M- and M-like proteins elucidated the presence of the hydrophobic amino acids in positions a and d also within the SCM sequence (Figure 2C). Furthermore, sequence alterations occur with deletion of seven amino acids supporting the coiled–coil structure of SCM.

Figure 2 Identification of the PLG-binding protein as an M-like protein of S. canis

(A) Amino acid sequence of the S. canis M-like protein SCM. The signal sequence (VKAEH, not in bold), LPxTG-motif (in italic) and the last amino acids of the 25 kDa N-terminal part of SCM (SCM 25KD, underlined) are indicated. (B) CD spectrum showing molar ellipticity plotted against wavelength for native SCM. The value for the 222/208 nm ratio is shown. (C) Alignment of the SCM with selected α-helical, coiled–coil proteins, including SpaSc of S. canis (ACM47242.1), SpaZ of S. equi sub. zooepidemicus (ACG63105.1), Spa of S. pyogenes (AAL98520.1), SzM of S. equi sub. zooepidemicus (CAW97903.1), Mrp4 of S. pyogenes (AAB33261.1), M18.1 of S. pyogenes (AAB03086.1), PAM of S. pyogenes (CAA80222.1) and M6 of S. pyogenes (AAT87854.1). The generalized heptad a-b-c-d-e-f-g sequence is shown above the alignment; respective amino acids are indicated in grey. (D) Illustration of PCR analysis of different S. canis isolates with oligonucleotides specific for the S. canis scm gene. Positive (+) and negative (−) PLG binding of the respective strains is shown below the gel. Accession numbers were retrieved from GenBank®.

In order to detect scm in the genome of different S. canis isolates, a specific PCR with oligonucleotides based on the N-terminal signal sequence and the conserved C-terminal LPxTG motif has been established (Figure 2D). Sequence analysis revealed that the scm fragment is not ubiquitously present in all S. canis isolates (Figure 2D). The scm gene could be detected in the genomes of strains G1, G8, G13, G15 and G361. Within these strains, the presence of scm correlates with a high PLG-binding capacity (Figure 1D). In contrast, no amplification of scm was observed for the S. canis strains G2, G14 and G17 which correlates with only a weak PLG-binding capacity (Figure 2D). A total number of 19 strains have been tested, and all strains showing strong PLG binding were also positive for scm (Figure 2D). These results indicated that SCM is necessary for PLG binding in S. canis.

Biochemical analyses of PLG binding to SCM

SCM was heterologously expressed in the non-PLG binding S. gordonii. In contrast with the wild-type strain (SGO WT), the SCM-expressing S. gordonii strain (SGO SCM) showed strong binding to iodinated human PLG (Figure 3A). Results of dot-blot overlay analyses confirmed recombinant purified SCM as a PLG-binding protein of S. canis. Furthermore, the PLG-binding domain of SCM could be narrowed down to the first 214 amino acids of the mature protein representing SCM 25KD (Figure 3B). SPR analysis was performed using PLG as the immobilized ligand and purified SCM and SCM 25KD as analytes at a series of concentrations (0.125–4 μM). Sensogram data of SCM binding to immobilized PLG revealed a specific and concentration-dependent binding interaction (Figure 3C). Evaluation of the binding data determined dissociation constants of 2.69×10−8 M for the PLG interaction with SCM (Figure 3C) and 2.07×10−8 M for the N-terminal part of SCM (Figure 3D). Dissociation constants within the same range (5.10×10−7 M) could also be confirmed in kinetic analysis with SCM as immobilized ligand and PLG as analyte in solution (Figure 3E). A summary of the BIAcore results representing the association rate constant (ka), the dissociation rate constant (kd), the affinity constant (KD) and the χ2 values are shown in Table 2.

Figure 3 Biochemical characterization of PLG binding of S. canis isolates and SCM protein

(A) PLG-binding analysis of S. gordonii heterologously expressing SCM (SGO-SCM) was analysed with iodinated PLG. (B) Dot-blot overlay of recombinant S. canis M-like protein SCM and its N-terminal fragment SCM 25KD. SCM and SCM 25KD were immobilized in serial dilutions on nitrocellulose (0.25 μg, 0.125 μg, 0.063 μg and 0.031 μg) followed by incubation with iodinated human PLG. (CE) Sensograms of SPR kinetics of PLG binding to SCM (C) and to the N-terminal fragment of SCM 25KD (D). PLG was immobilized on Biacore CM5 sensorchips and both SCM and SCM 25KD were used as analytes. (E) Binding kinetic analysis after immobilization of full-length SCM and with PLG as the analyte.

View this table:
Table 2 Kinetic parameters of the Biacore sensogram

Characterization of SCM-binding domain in human PLG

Kringle domains of PLG are well-known binding sites for a variety of bacterial PLG receptors containing exposed lysine residues [26,27]. In order to characterize the region of interaction between SCM and the PLG molecule in detail, different inhibition studies were performed in radioactive-binding experiments. A dose-dependent inhibition of PLG binding to the surface of S. canis G361 was observed in the presence of increasing amounts of 6-AHA (Figure 4A). Addition of 300 mM 6-AHA led to a residual binding capacity of 25.4%. A dose-dependent inhibition of PLG binding from 66.5% with 62.5 nM inhibitor to 19.2% with 2 μM of non-labelled PLG underlined the specificity of the interaction (Figure 4B). In contrast, no difference in PLG-binding capacity was observed in the presence of 2 μM non-labelled K1–3 (Figure 4B). PAM-expressing S. pyogenes strain A158 bound to K1–3 with a capacity of 58.0% (Figure 4D), which was inhibited by the addition of 2 μM non-labelled PLG and 2 μM K1–3 (Figure 4C). These results were supported by the fact that neither S. canis strain G361 (7.6%) nor the SCM-expressing S. gordonii strain SGO SCM (2.8%) were able to bind to iodinated K1–3 (Figure 4D). For mapping the binding site on PLG, dot-blot experiments using different fragments of the entire PLG molecule and recombinant SCM were performed. Figure 4(E) shows a specific dose-dependent binding signal of SCM only to full-length PLG and to the C-terminal elastase-digested product mPLG, whereas no interaction could be detected using the fragments K1–3, K1–4 and K4 respectively. SPR analyses of binding kinetics with immobilized mPLG and SCM as the analyte revealed a high dissociation constant of 1.38×10−8 M, which is within the same high-affinity range as determined for the kinetics between SCM and PLG (Figure 4F and Table 2).

Figure 4 Binding analysis and inhibition studies of PLG and different PLG derivatives to different streptococci

(A) Inhibition of lysine-dependent PLG binding to S. canis G361 in the presence of 0.001 mM, 0.01 mM, 0.1 mM, 10 mM, 100 mM and 1000 mM of the lysine analogue 6-AHA. (B and C) Inhibition of PLG binding of S. canis G361 (B) and S. pyogenes isolate A158 (C) by non-labelled PLG and non-labelled K1–3 respectively. Inhibitors were employed at concentrations of 0.5 μM, 1.0 μM, 1.5 μM and 2.0 μM. (D) Binding analysis of iodinated human K1–3 to S. canis, S. gordonii and S. pyogenes. (E) Dot-blot analysis of SCM binding to PLG and PLG derivatives: K1–3, angiostatin (representing K1–4), K4 and mPLG (representing K5 and the catalytic domain). PLG and its fragments were immobilized on nitrocellulose at 3.0 μg, 1.5 μg, 0.75 μg, 0.38 μg and 0.19 μg. Binding of SCM was detected with SCM-specific antibodies and HRP-conjugated secondary antibodies. Non-specific cross-reactivity of antibodies was not detected (results not shown). (F) Kinetic analysis of SCM binding to immobilized mPLG using SPR.

Activation of PLG and recruitment of plasmin on the cell surface of S. canis

Activation of PLG to proteolytically active plasmin was determined by photometrical measurement of the plasmin-specific chromogenic substrate S-2251 (Figure 5A). No plasmin activity was detected in the supernatant of the zoonotic isolate G361 with both human or canine plasma (Figures 5A and 5B). In contrast, the S. equisimilis strain C90 recruited PLG from human plasma (shown in Figure 1B) and converted human PLG, but not canine-plasma-derived PLG, into plasmin (Figures 5A and 5B). These results indicate that no endogenous PLG activator, such as streptokinase, is expressed in S. canis, and also emphasizes the species-specific activity of streptokinase derived from S. equisimilis [28]. Interestingly, human PLG bound to the surface of S. canis G361 was activated in a time-dependent manner in the presence of the eukaryotic activator urokinase (Figure 5C), whereas no plasmin activity was detected in the absence of PLG or of PLG activators (Figure 5C). The serine protease inhibitor aprotinin inhibits PLG activation on the bacterial surface, indicating the serine-protease specificity of plasmin-mediated substrate conversion (Figure 5C). Plasmin activity was also significantly reduced in the presence of 2, 40 and 80 mM 6-AHA in a dose-dependent manner (Figure 5D). In addition to the results obtained from binding analyses, the inhibitory activity of 6-AHA for PLG activation confirmed that the lysine-binding sites within PLG kringle domains are involved in the interaction of SCM with PLG. Interestingly, PLG bound to SCM-expressing SGO bacteria underwent a faster conversion into plasmin by urokinase as compared with soluble PLG (Figure 5E). These results indicate that binding of PLG to SCM on the bacterial surface accelerates the rate of plasmin formation.

Figure 5 Plasmin activation mechanism of S. canis strain G361

(A and B) Determination of secreted plasmin activity by measurement of chromogenic plasmin substrate S-2251 at 405 nm in a time series including 0 min, 30 min, 90 min and 120 min after incubation of S. canis G1, G2, G361 and S. equisimilis C90 with human plasma (A) and canine plasma (B). (C) Plasmin activity was measured after incubation of S. canis bacteria with purified human PLG in the presence of urokinase. Plasmin activity was determined over a time period up to 120 min and also in the presence of the serine protease inhibitor aprotinin. (D) Dose-dependent inhibition of plasmin activation by the addition of 2.0 mM, 40 mM and 80 mM of the lysine analogue 6-AHA. (E) Time-dependent conversion of PLG bound to the surface of SCM-expressing SGO and soluble PLG. No plasmin activation was detected without PLG or urokinase respectively (results not shown).

Binding of SCM protein to PLG promotes degradation of fibrinogen and fibrin and transmigration of streptococci through microthrombi

Fibrinogen is the major component of fibrin clots generated in response to vessel injury and inflammatory processes. Immunoblot analysis of fibrinogen degradation detected a time-dependent degradation of purified human fibrinogen by PLG-coated S. canis bacteria in the presence of urokinase (Figure 6A). Degradation products of fibrinogen fragments were produced after 5 min of incubation with bacteria, and fragmentation was enhanced following a time series up to 60 min. In contrast, no degradation was detected by PLG-coated S. canis bacteria without any activator (Figure 6A, lane 2). In order to analyse fibrinolytic activity by PLG-coated bacteria, fibrin clots were generated on glass coverslips by thrombin incubation of PLG-depleted fibrinogen. Photographic visualization of the fibrin clots demonstrated that the white and turbid microthrombi were totally dissolved into a clear solution by PLG-coated S. canis G361 in the presence of urokinase within 4 h of incubation (Figure 6B). No dissolution of the fibrin clots could be observed after incubation with bacteria or PLG-coated bacteria without activator (Figure 6B), indicating that the fibrin thrombi represent a tight, but protease-sensitive, barrier. Plasmin-mediated transmigration activity of S. canis bacteria through fibrin clots generated on membranes of transwell cell-culture inserts was determined by plating of diluted aliquots from the lower well (Figure 6B, table). Transmigration of 3.7 (±0.7)×106 cfu·ml−1 S. canis bacteria was detected after 4 h and 8.3 (±0.8)×106 cfu·ml−1 bacteria after 6 h of incubation of thrombi with plasmin-coated bacteria. Dissolution of fibrin bundles by plasmin-coated S. canis bacteria was also visualized by time-lapse microscopy. The time-dependent fibrin degradation is shown in Figure 6(C) illustrating the dynamics of plasmin-mediated fibrin degradation.

Figure 6 Plasmin-mediated degradation of fibrinogen and fibrin matrices by S. canis G361

(A) Immunoblot analysis of degradation of human fibrinogen by PLG-encoated S. canis G361 bacteria. Degradation was analysed in the presence of urokinase after the time points indicated (0 min, 5 min, 10 min, 20 min, 30 min and 60 min) with fibrinogen-specific antibodies. Fibrinogen was not degraded after incubation of PLG-coated S. canis for 0 min and for 60 min with fibrinogen without activator (Ctrl 0, Ctrl 60). (B) Illustration of PLG-mediated degradation of fibrin generated on a porous transwell membrane by S. canis G361 in the presence or absence of urokinase. Plasmin-mediated transmigration of S. canis 361 through fibrin clots was quantified by determination of bacterial cfu·ml−1 after different incubation periods (0 h, 2 h, 4 h and 6 h) by plating of serial dilutions from the lower well. (C) Time-lapse imaging of fibrin degradation by plasmin-encoated S. canis G361 bacteria. S. canis G361 bacteria (arrows) are visible within the dissolved fibrin solution.

Electron microscopic visualization of fibrin degradation by S. canis and SCM-expressing S. gordonii

Electron microscopic studies illustrated the dissolution of the fibrin bundles of S. canis G361 (Figures 7A–7H). Fibrin fibres of a generated fibrinogen thrombus were microscopically visualized as ordered cable-like structures of up to 0.35 μm in diameter (Figures 7A and 7B). Without plasmin-mediated fibrinogen degradation, the cell chains of S. canis bacteria showed strong adherence to the fibrin surface layer (Figure 7C) and also to the non-dissolved fibrin fibrils (Figures 7D and 7E). The fibrin bundle structure remained unchanged after incubation with PLG-coated bacteria in the absence of a PLG activator (Figure 7E). Fibrin fibres were substantially degraded after incubation of thrombi with PLG-pretreated bacteria in the presence of urokinase (Figures 7F–7H). Higher magnifications showed that the bacteria-bound proteolytic activity resulted in an enhanced dissolution of the fibrin matrix forming a halo of dissolved fibrin around the bacteria and led the bacteria to sink into the underlying fibrin layer (Figures 7G and 7H). Plasmin-mediated fibrin degradation was also analysed by electron microscopy after incubation of the microthrombi with S. gordonii bacteria expressing SCM (Figures 7I–7P). No fibrin degradation was detected by S. gordonii bacteria (Figure 7I) or PLG-pre-incubated S. gordonii (Figure 7J) which lacks SCM on its surface (SGO WT). Moreover, no dissolution of the fibrin bundles was observed by PLG-pre-incubated S. gordonii (SGO WT) bacteria in the presence of the PLG activator urokinase (Figures 7K and 7L). Similar to S. canis, S. gordonii bacteria expressing SCM on the surface (SGO SCM) showed no fibrin degradation activity, but strong adherence to the fibrin matrix (Figures 7M and 7N). Moreover, in conjunction with S. canis, dissolution of the ordered fibrin structures was only detected by plasmin-coated S. gordonii expressing SCM (Figures 7O and 7P). In conclusion, the degradation analyses indicate that plasmin-mediated dissolution of soluble fibrinogen and fibrinous thrombi represents a key function of the S. canis M-like protein for streptococcal transmigration and dissolution of the fibrin matrix.

Figure 7 Visualization of fibrin degradation by plasmin-coated S. canis G361 and S. gordonii bacteria analysed by field-emission scanning-electron microscopy

(A and B) Microscopic images illustrate a dense network of microthrombi formed by bundled fibrin fibres (control). (CE) Bacterial attachment to the fibrin matrix was shown by non-treated S. canis G361 (C and D), as well as by PLG-pre-incubated bacteria (E, PLG). (FH) The fibrin bundles were degraded by proteolytically active S. canis bacteria in the presence of urokinase (PLG, uPA). (I and J) Only minor bacterial attachment to fibrin thrombi was detected by S. gordonii lacking the SCM protein on the surface (I; SGO WT, white arrows) and PLG-pre-incubated SGO WT (J; SGO WT PLG, white arrows). (KM) No degradation of fibrin bundles was observed with plasmin-encoated S. gordonii (K and L; PLG, uPA) or SCM-expressing S. gordonii (SGO SCM) without PLG incubation (M; PLG). (N) PLG pre-incubation did not mediate microthrombi dissolution by SGO-SCM. (O and P) Remarkable degradation of fibrin bundles was only detected by S. gordonii expressing the SCM protein on its surface (SGO SCM) supplemented with PLG and urokinase (PLG, uPA).

DISCUSSION

S. canis is well-known as an emerging infective agent responsible for a variety of invasive diseases in domestic animals [2], as well as humans [35]. Despite its zoonotic potential, the knowledge regarding virulence mechanisms required for establishing an S. canis infection remained mostly elusive. The subversion of host-derived proteolytic activity for invasive and degradative dissemination within the host tissue belongs to one of the most important pathogenicity mechanisms described for pathogenic bacteria so far [6]. Several bacterial PLG-binding proteins have been identified up to now, but the pathophysiological role of PLG binding to zoonotic streptococci such as S. canis has not been investigated in detail. The present study revealed the presence of a novel PLG receptor (SCM) on the surface of S. canis. Detailed biochemical analysis characterized SCM as a high-affinity ligand of the elastase-digested PLG fragment mPLG. Despite the lack of endogenous streptokinase activity, surface-bound PLG is convertible into plasmin by host-derived PLG activators, which enables S. canis to degrade ECM matrix molecules and dissolve aggregated fibrin thrombi. The use of plasmin might therefore represent an important virulence trait for establishing an S. canis infection.

S. canis is mostly isolated from animals, in particular, from dogs, cats and cows [1]. A close genetic relationship and a similar clinical phenotype to other Group A, C and G streptococci, lead to inclusion of S. canis in the pyogenic group of streptococci [29]. It has been described for the species S. pyogenes [GAS (Group A streptococci)] and Streptococcus dysgalactiae sub. equisimilis (GCGS) that their ability to bind and activate PLG has an outstanding role in pathogenicity [30,31]. Receptors for PLG binding on the streptococcal surface are M- and M-like proteins, such as PAM, GCS3 and MLG72 [13,31]. M-proteins form fibrillar structures on the surface of streptococci which have been characterized extensively as important virulence factors in the pathogenesis of GAS and GCGS infections [32,33]. Usually, emm genes are detectable by specific oligonucleotides. S. canis is characterized as emm-negative in the established PCR-test described by the CDC (Centers for Disease Control) for emm typing. One exception has been published by Ahmad et al. [34]. Recently, Yang et al. [35] identified a protein of S. canis (SPASc) showing similarity to the protective antigen SpaZ of S. equi sub. zooepidemicus, SPA of S. pyogenes and SeM, the M-like protein of S. equi. An orthologue of SPASc was recombinantly expressed as an His-tag fusion protein. CD spectroscopy demonstrated an α-helical structure and a high probability for dimerization under physiological conditions. The formation of a coiled–coil structure is supported by the fact that an alignment of the amino acid sequence with known M- and M-like proteins revealed the generalized heptad sequence characteristic for α-helical coiled–coil proteins. In summary, structural properties, binding of ECM proteins and sequence homologies characterize SCM as an M-like protein of S. canis.

Dot-blot overlay assays showed a concentration-dependent binding of SCM to iodinated PLG. The quality of SCM-mediated PLG binding was supported by the determination of a dissociation constant within the nanomolar range by SPR studies. Furthermore, strong PLG-binding activity of heterologously SCM-expressing S. gordonii bacteria also confirmed the high relevance of SCM for the PLG interaction of S. canis.

The PAM-binding site has been identified within the kringle domain 2 of PLG [36,37]. Kringle domains have been shown to participate in high-affinity-binding interactions with streptokinase and also mediate interaction of PLG with PLG activators, plasmin substrates, fibrin and cell-surface receptors [38,39]. In contrast with PAM, binding analysis revealed that SCM shows no interaction with the first four kringle domains of PLG. Moreover, to the best of our knowledge, SCM is the first bacterial PLG-binding protein interacting in a lysine-dependent manner with the C-terminal part of PLG, termed mPLG. SPR studies confirmed the interaction of SCM with mPLG reaching KD values within the nanomolar range. This kind of interaction has only been suggested for the PLG interaction with Helicobacter pylori [40], Trypanosoma cruzi epimastigotes and streptococci from bovine origin [41,42]. Several reports describe an interaction of PLG-binding proteins with the lysine-binding sites present in the N-terminal kringle domains of PLG. Since we demonstrate an interaction of SCM with the C-terminal part of PLG, composed of kringle domain 5 and the catalytic domain, the different interaction sites may also contribute to a different mechanism of host–pathogen interaction. Detection of mPLG was associated with low levels of PLG and high levels of elastase-a1 protease inhibitor complex in patients suffering from septicaemia [43,44] and is also able to produce clot lysis after activation with tPA in PLG-depleted plasma [9]. Therefore the interaction of pathogenic S. canis with mPLG may enable the streptococci to recruit plasmin activity after binding of PLG or mPLG present under different patho-physiological conditions. The inhibition of PLG binding by the lysine analogue 6-AHA is often interpreted as identification of lysine-binding sites of kringle domains as major interaction sites [13,45,46]. PLG binding of SCM, as well as conversion of PLG into proteolytic active plasmin on the S. canis surface, was inhibited in a concentration-dependent manner by 6-AHA. These results would support an interaction of SCM with the lysine-binding site present in K5 of mPLG. In contrast with SCM, Group A streptococcal PAM has been shown to bind to the K2 of PLG, which exhibits a low-affinity lysine-binding site [36]. X-ray analysis of the PAM–K2 protein complex revealed that a ‘pseudo-lysine’-like binding is formed in PAM by positively and negatively charged amino acids within the PLG-binding region of PAM [37]. This kind of pseudo-lysine interaction has also been proposed for binding of lysine analogues to K1 and K5 [37]. The detailed characterization of specific amino acids involved in the interaction of SCM with mPLG is currently in progress and will elucidate the role of lysine residues in binding of SCM to mPLG.

Immobilization of PLG on physiological fibrin surfaces or on the cell surface of bacteria has been shown to accelerate conversion into proteolytically active plasmin by eukaryotic activators such as tPA or uPA respectively [47]. Plasmin activation measurements confirmed an increase in activation of PLG bound to SCM-expressing S. gordonii bacteria in contrast with significantly slower plasmin activation in the presence of soluble PLG. Accelerated activation was also shown for PLG which was bound to surface-displayed eukaryotic enolase [26]. In contrast with GAS and GCGS, S. canis does not express an endogenous PLG activator [31]. Therefore a host-derived PLG activator is required for activation of cell-surface-bound PLG in an aprotinin-sensitive manner. Blood coagulation resulting in fibrinogen-enriched thrombi has been suggested to capture invasive pathogenic microorganisms in order to prevent spreading of bacteria into the bloodstream and into deeper tissue sites [36,48]. Electron-microscopic illustration and time-lapse visualization clearly demonstrate the proteolytic degradation activity recruited on the surface of S. canis or SCM-expressing S. gordonii bacteria. Generated fibrin thrombi were completely dissolved by plasmin-encoated bacteria. Therefore bacteria-bound proteolytic activity enables the bacterial release from the coagulative capture within blood thrombi and presents an effective evasion mechanism for invasive pathogenic bacteria. Moreover it has been shown that bacterial-bound plasmin activity is associated with pathological phenotype. In patients with Gram-negative septicaemia, high levels of fibrin-degradation products have been detected [49], and also interaction of Group A streptococcal streptokinase with the host fibrinolytic system facilitates bacterial access to the vasculature [50]. Interestingly, neonatal septicaemia and meningitis, which is a clinical feature of S. canis as well as E. coli infections, has also been associated with tPA-catalysed activation of PLG bound to E. coli fimbriae [2,3,51,52].

The recruitment of PLG to the surface of the invasive zoonotic pathogen S. canis might, therefore, represent an important virulence mechanism leading to effective evasion of proteolytically active bacteria from thrombus capture within the vasculature of the host and to bacterial dissemination into deeper tissue sites. The elucidation of the interaction mechanism between SCM and mPLG offers new insights into understanding the interaction of M-like proteins with the human fibrinolytic system and enables the development of new intervention therapies.

AUTHOR CONTRIBUTION

Marcus Fulde performed experimental analyses and contributed to the preparation of the manuscript. Electron microscopic visualization was performed by Manfred Rohde. Angela Hitzmann and D. Patric Nitsche-Schmitz contributed to experimental analyses. Klaus Preissner provided essential reagents and contributed to the editing of the paper prior to submission. Gursharan Singh Chhatwal provided financial support, contributed to preparation of the manuscript and discussion. Simone Bergmann contributed to experimental assay design and preparation of the manuscript.

FUNDING

This work was supported by the European Community's Seventh Framework Programme [grant number HEALTH-F3-2009-223111].

Acknowledgments

We are grateful to Ina Schleicher, Franziska Voigt, Agnes Zimmer and Nadine Nachtigall for technical assistance. We thank Uwe Schubert (Justus-Liebig-University, Giessen, Germany) for technical support and Dr Oliver Goldmann for helpful discussion. We also like to thank Professor J.F. Presocott (Department of Pathobiology, University of Guelph, Ontario, Canada), Professor B.W. Beall (Centers for Disease Control and Prevention, Atlanta, GA, U.S.A.) and Dr M. van der Linden (National Reference Centre for Streptococci, Aachen, Germany) for providing bacterial strains.

Abbreviations: 6-AHA, ϵ-amino caproic acid; cfu, colony-forming unit; ECM, extracellular matrix; GAS, Group A streptococci; GCGS, Group C and G streptococci; HRP, horseradish peroxidase; PLG, plasminogen; mPLG, mini-plasminogen; SCM, Streptococcus canis M-like protein; SPR, surface plasmon resonance; tPA, tissue-type plasminogen activator; uPA, urokinase-type plasminogen activator

References

View Abstract