Identification of a novel plasmin(ogen)-binding motif in surface displayed α-enolase of Streptococcus pneumoniae

Authors


E-mail S.Hammerschmidt@mail.uni-wuerzburg.de; Tel. (+49) 931 2153; Fax (+49) 931 2578.
Present address: Center for Infectious Diseases, University Würzburg, Röntgenring 11, 97070 Würzburg, Germany.

Summary

The interaction of Streptococcus pneumoniae with human plasmin(ogen) represents a mechanism to enhance bacterial virulence by capturing surface-associated proteolytic activity in the infected host. Plasminogen binds to surface displayed pneumococcal α-enolase (Eno) and is subsequently activated to the serine protease plasmin by host-derived tissue plasminogen activator (tPA) or urokinase (uPA). The C-terminal lysyl residues of Eno at position 433 and 434 were identified as a binding site for the kringle motifs of plasmin(ogen) which contain lysine binding sites. In this report we have identified a novel internal plamin(ogen)-binding site of Eno by investigating the protein–protein interaction. Plasmin(ogen)-binding activity of C-terminal mutated Eno proteins used in binding assays as well as surface plasmon resonance studies suggested that an additional binding motif of Eno is involved in the Eno-plasmin(ogen) complex formation. The analysis of spot synthesized synthetic peptides representing Eno sequences identified a peptide of nine amino acids located between amino acids 248–256 as the minimal second binding epitope mediating binding of plasminogen to Eno. Binding of radiolabelled plasminogen to viable pneumococci was competitively inhibited by a synthetic peptide FYDKERKVYD representing the novel internal plasmin(ogen)-binding motif of Eno. In contrast, a synthetic peptide with amino acid substitutions at critical positions in the internal binding motif identified by systematic mutational analysis did not inhibit binding of plasminogen to pneumococci. Pneumococcal mutants expressing α-enolase with amino acid substitutions in the internal binding motif showed a substantially reduced plasminogen-binding activity. The virulence of these mutants was also attenuated in a mouse model of intranasal infection indicating the significance of the novel plasminogen-binding motif in the pathogenesis of pneumococcal diseases.

Introduction

Streptococcus pneumoniae can colonize the mucosal surfaces of the human respiratory tract without the outcome of clinical symptoms (Austrian, 1986), but is also known as a common aetiologic agent of upper and lower respiratory diseases. Pneumococci cause relatively harmless non-invasive diseases such as otitis media, sinusitis and are the most common cause of community acquired pneumonia. However, pneumococci can also cause life-threatening invasive diseases such as bacteraemia and meningitis with high morbidity and mortality especially in the high risk groups (Musher, 1992). A prerequisite for the ability to colonize the mucosal surfaces and subsequently to invade the epithelial cells is the expression of bacterial adhesins that interact with host receptors. These interactions facilitate invasion, transmigration and dissemination of the pathogen. In addition to the pneumococcal factors that mediate direct adherence to human cells, such as the phosphorylcholine (Cundell et al., 1995) and the SpsA protein (also designated CbpA and PspC) (Hammerschmidt et al., 1997; 2000; Rosenow et al., 1997; Brooks-Walter et al., 1999; Zhang et al., 2000), bacterial adhesins have been identified which bind to components of the extracellular matrix (ECM). Recently, PavA was identified as a pneumococcal adhesin for fibronectin and as being essential for virulence (Holmes et al., 2001). The surface displayed α-enolase has been shown to be a major plasmin(ogen)-binding protein of S. pneumoniae (Bergmann et al., 2001).

Plasminogen is a single chain glycoprotein of ≈ 92 kDa which is present in human plasma at approximately 2 µM (Pollanen et al., 1991). The active form of plasminogen, the serine protease plasmin, is composed of an amino-terminal peptide followed by five triple-disulphide-bonded kringle domains containing lysine-binding sites (LBS) and an catalytic domain with serine protease activity (Miyashita et al., 1988). Plasmin plays a key role in both extrinsic and intrinsic fibrinolysis and serves as an essential component to maintain homeostasis and vascular potency (Plow et al., 1991; 1995; Saksela and Rifkin, 1988). Degradation by plasmin occurs within a wide variety of substrates including extracellular matrix components like fibronectin, laminin and vitronectin (Gonzales-Gronow et al., 1991; Duval-Jobe and Parmely, 1994) and induces enhanced activation of other matrix-degrading proteinases, such as collagenases (Wong et al., 1992). Binding of plasminogen and its subsequent activation by the tissue-type plasminogen activator tPA has been demonstrated to promote migration of pneumococci through reconstituted basement membranes and represents therefore an important determinant of virulence (Eberhard et al., 1999).

In previous reports binding of plasmin(ogen) to bacterial glycolytic enzymes have also been reported for the α-enolase SEN (streptococcal surface enolase) and the glycerinealdehyde-3-phosphate dehydrogenase SDH/Plr of Streptococcus pyogenes (Lottenberg et al., 1992a; Pancholi and Fischetti, 1992; Winram and Lottenberg, 1998; Pancholi and Fischetti, 1998). Prokaryotic α-enolases have been shown to be highly conserved glycolytic enzymes that elicit various functions and binding capacities in several biological and pathophysiological processes (Pancholi, 2001).

Despite the absence of a signal sequence and typical motifs required for membrane anchoring the attachment of these glycolytic enzymes to streptococcal and pneumococcal cell surfaces has been demonstrated and implications in plasmin(ogen) acquisition have been clearly shown (Pancholi and Fischetti, 1998; Bergmann et al., 2001). Furthermore, immunoelectron microscopic studies indicated the ability of the pneumococcal α-enolase, designated Eno, to reassociate to the bacterial cell surface resulting in enhanced plasmin(ogen)-binding capacity (Bergmann et al., 2001).

Plasmin(ogen)-binding to different eukaryotic receptors including eukaryotic enolases (Miles et al., 1991; Redlitz et al., 1995; Winram and Lottenberg, 1998) and to α-enolases of S. pyogenes and S. pneumoniae is mediated by C-terminal lysyl residues (Pancholi and Fischetti, 1998; Bergmann et al., 2001). However, binding of plasmin(ogen) to SEN, which shows a 93% sequence similarity to pneumococcal Eno, is mediated by more than one binding site as shown by Scatchard analysis (Pancholi and Fischetti, 1998).

In the present study a novel internal binding motif involved in the Eno-plasminogen complex formation was identified by using surface plasmon resonance and spot-synthesized synthetic peptides. A synthetic decapeptide representing the internal plasmin(ogen) binding motif of Eno confirmed in competitive inhibition experiments its function as a plasmin(ogen)-binding motif. Amino acid substitutions in the novel binding site FYDKERKVY substantially reduced plasminogen-binding activity of pneumococcal mutants and significantly reduced their virulence in a mouse model.

Results

Binding of plasminogen to membrane spotted and reassociated Eno proteins

In previous work it was shown that recombinant pneumococcal α-enolases with modified C-terminal lysyl residues exhibit under reducing conditions a substantially reduced plasminogen-binding capacity (Bergmann et al., 2001). In order to investigate the recruitment of plasminogen under native conditions and therefore more related to in vivo conditions, wild-type Eno protein and C-terminal modified Eno proteins EnoKL., EnoLL. and Enodel were immobilized on a PVDF-membrane and binding of plasminogen was investigated. In the modified α-enolases EnoKL and EnoLL, respectively, the ultimate or both the penultimate and ultimate lysyl residues were substituted by a leucine. In Enodel both lysyl residues were deleted. Blot-overlay assays with radiolabelled plasminogen showed binding of 125I-plasminogen to both C-terminal deleted Enodel protein as well as to EnoKL and EnoLLFig. 1). In addition, the characteristic feature of pneumococcal α-enolase to reassociate to the cell surface and, as a consequence, to increase plasmin(ogen)-binding activity (Bergmann et al., 2001), was used to examine plasminogen recruitment of C-terminal modified Eno proteins. Pneumococcal strain R6x showed, after preincubation with increasing amounts of the C-terminal truncated Enodel and modified Eno proteins EnoKL. and EnoLL., strongly enhanced plasminogen-binding activity (data not shown). These binding assays indicated that under native conditions wild-type Eno as well as C-terminal modified Eno proteins are able to promote the acquisition of plasminogen irrespective of the absence of C-terminal lysyl residues. These data indicated the presence of an additional plasminogen-binding epitope in the pneumococcal α−enolases and that this second binding site contributes to the formation of Eno–plasminogen complexes.

Figure 1.

Binding of radiolabelled 125I-plasminogen to immobilized wild-type Eno and mutated Eno proteins. Proteins were purified under native conditions and different amounts of the His-tagged Eno proteins using serial dilutions (1:2) were bound on a PVDF membrane using a Bio-Dot SF microfiltration apparatus (BIORAD). Binding of plasminogen was detected to all C-terminal modified Enodel, EnoLL., EnoKL proteins, but not to Enoint and Enoint/del containing amino acid substitutions in the internal plasminogen-binding motif. Binding to the highest concentration of Eno, Enodel, EnoLL. and EnoKL., respectively, was defined as 100%. Densitometric analysis indicated a concentration dependent decrease of plasminogen binding to Eno, Enodel, EnoLL. and EnoKL. With respect to the lowest concentrations immobilized on the membrane binding was reduced to 15.0%, 10.6%, 12.3%, and 22.3% for wild-type α-enolase, Enodel, EnoKL., and EnoLL. respectively. Densitometric quantification of specific signals was performed using the Phoretix 2D Advanced software.

Association and dissociation kinetics of the plasmin(ogen) interaction to immobilized eno proteins

The kinetics of formation and dissociation of the Eno, Enodel, EnoKL, EnoLL-plasmin(ogen)/LBS 1 complexes were analysed using surface plasmon resonance (SPR). The LBS 1 contains the first three kringle of plasminogen. First, Eno wild-type protein and mutated Eno proteins were covalently bound on the surface of an activated sensor chip. The kinetics of complex formation with soluble plasminogen, plasmin and LBS 1 was measured. All used analytes bound to Eno wild-type protein and also to C-terminally mutated Enodel, EnoKL., and EnoLL. The observed rates of formation of the complexes were all dependent on the concentration of the analyte applied Fig. 2). A qualitative assessment of the sensorgrams revealed significant differences in the interaction of glu-plasminogen, plasmin, and LBS 1 binding with Eno (data not shown), whereas only minor differences between the association reactions of wild-type Eno and mutated Eno proteins with the analytes was observed (Fig. 2).

Figure 2.

Surface plasmon resonance measurements of the Eno–plasminogen interactions. Eno and C-terminal Enodel, EnoLL., EnoKL. were coated on a BIAcore™ CM5 sensor chip and plasminogen was used as an analyte. Changes in plasmon resonance are shown as relative response units. Sensorgrams show the binding kinetics and concentration dependence of rates of binding of plasminogen to Eno and modified Eno proteins. Plasminogen was used in concentrations of 250 nM, 125 nM, 62.5 nM, 31.25 nM, 15.625 nM, 7.8 nM and 1.9 nM. The blank run was subtracted from each sensorgram. plg, start of injection of plasminogen; w, stop of injection and start of its dissociation.

Binding experiments using C-terminal mutated Eno proteins and radiolabelled plasminogen indicated the presence of more than one binding site in Eno for plasminogen. Therefore, data were fitted according to the heterogeneous ligand model. Evaluation of the kinetics data using this model resulted in different association rates ka1 and ka2 and dissociation rates kd1 and kd2. These results therefore suggested that binding of plasminogen to wild-type Eno is associated with a fast and slow binding of the analyte. In contrast, binding of plasminogen to Enodel and EnoLL. with C-terminal amino acid substitutions resulted in association rates which were both in the same range of magnitude Table 1). The binding kinetics verified results of the radioactive binding experiments with respect to the binding of plasminogen to two different binding sites in the ligand Eno.

Table 1. . Surface plasmon resonance measurement of the interaction of human plasminogen with α-enolase and modified α-enolases.
 ka1[M−1s−1]
(x 104)
ka2[M−1s−1]
(x 104)
kd1[s−1]
(x 10−4)
kd2[s−1]
(x 10−4)
KD1[M]
(x 10−9)
KD2
[M]
(x 10−9)
χ2
  1. The binding data of the interaction of plasminogen with the wild-type Eno protein and C-terminal modified EnoKL, EnoLL and Enodel were fitted globally to the heterogeneous ligand binding model included in the BIAevaluation 3.0 software. Listed are the association and dissociation rate constants (ka1, ka2, kd1 and kd2) and equilibrium constants (KD1 and KD2). The values of χ2-statistical test and the distribution of the residuals of the fitted sensogram data were used to determine whether a particular reaction scheme was a good mathematical representation of the plasmin(ogen) binding sensorgram data. The heterogeneous ligand binding model that displayed low χ2-values and that did not show any great variations of magnitude within the examined range of plasminogen-concentrations, was judged to be a good approximation of the binding interaction.

Eno-plasminogen0.8715.50.005133  0.5586.221.3
EnoKL.-plasminogen1.0613.00.002142230 0.01 3.38
EnoLL.-plasminogen6.09 6.020.48-  0.47 78.778.123
Enodel-plasminogen5.04 5.094.97  5.04 999937.3

Simultaneous evaluations of association and dissociation using the kinetic models included in the BIAevalution software, however, did not result in good mathematical fits. In order to calculate the association and dissociation separately, the observed rates of complex formation (kobs) were evaluated. The kobs were dependent on the concentration of plasmin(ogen) and LBS 1 applied. The results revealed comparable observed association rate constants kobs for the formation of the different Eno proteins and plasmin(ogen) Table 2).

Table 2. . Kinetic parameters obtained for the observed association rate constants kobs and dissociation rate constants obtained for the multiple component dissociation reaction model (Ai + Bi AiBi).
 associationdissociation
kobs[M−1s−1] (x 10−2)χ2kd1[s−1]
(x 10−3)
kd2[s−1]
(x 10−3)
χ2
  1. The association rate constants kobs were determined according to the simple 1 : 1 Langmuir binding model. Dissociation rate constants kd1 and kd2 for two parallel dissociation reactions were estimated according to R = R1e–kd1(tt°) + (R0-R1)e–kd2(t–t°). This reaction scheme represents a two or more component dissociation kinetic model (Andronicos et al., 2000).

Eno-plasminogen2.548.932.742.713
EnoKL.-plasminogen2.340.317.63.55 3
EnoLL.-plasminogen2.410.438.12.51 1
Enodel-plasminogen2.850.39.62.65 1

After formation of the complexes the dissociation was succeeded. Recent studies investigating the human α-enolase–plasminogen interaction have suggested that a conformational change to glu-plasminogen upon binding has taken place so that dissociation is associated with two steps (Andronicos et al., 2000). Therefore, a multiple component dissociation reaction model (Ai + Bibsl00076AiBi) was used to fit the data. By applying the adequate multiple component dissociation reaction model, which is independent of the association phases, strong fits of the Eno-plasmin(ogen) interaction were obtained (Table 2).

Identification of plasmin(ogen) binding to spot synthesized peptides of Eno

In order to map the plasmin(ogen)-binding sites, the complete Eno sequence was divided into 141 overlapping peptides, each consisting of 15 amino acids, with an offset of three amino acids. The synthetic peptides were assayed for their capability to bind human plasminogen. The results showed binding to the ultimate peptide containing the C-terminal lysyl residues and additional strong binding to a stretch of further five peptides Fig. 3A and B). To precisely map the sequence of the identified second binding domain, a second spot membrane with an offset of one amino acid and a peptide length of four to 15 amino acids per spot was prepared. The complete sequence used for narrowing down the internal binding epitope spanned amino acid 232 to amino acid 266. This strategy resulted finally in 200 different peptides. The synthetic peptides contained, at least partially, the sequence that was formerly assayed positive for plasmin(ogen) binding (see Supplementary material). The results identified a nine-amino-acid-peptide with the sequence FYDKERKVY as the minimal internal binding motif of Eno (Fig. 3C). This sequence is located in Eno between amino acids 248 and 256. In addition, array spotted synthetic peptides with individual amino acid substitutions at every position of the internal binding motif FYDKERKVYD were examined for plasminogen-binding activity in order to identify critical amino acids in the motif. For this systematic mutational analysis, every single amino acid was individually substituted by each of the 20 amino acids. The results demonstrated in particular a critical role of the acidic amino acids, aspartic acid (position 3), glutamic acid (position 5), and also a role of the lysyl residues (position 4 and 7) for plasminogen-binding activity, when non-conserved amino acid substitutions were performed (data not shown).

Figure 3.

Determination of the minimal plasminogen-binding motif of Eno.
A. Spot membrane of the 434 amino acid sequence of Eno divided into 141 overlapping peptides, of 15 amino acids each, with an offset of three amino acids, analysed with human plasminogen. Twenty-five peptides were immobilized in one line. Binding of human plasminogen was detected to spot number 79, 80, 81, 82, and 83. Reactivity of other spots was due to non-specific binding of anti-plasminogen antibody and secondary antibody used (see Experimental procedures).
B. Sequences of spots (78’85) and binding reactivity with plasminogen.
C. Spot membrane with 198 overlapping peptides, with peptide length from four amino acids up to 15 amino acids residues per spot analysed with plasminogen. The amino acid sequence used to construct the spot-membrane corresponds to amino acids 232–267 (VPGKDVDCASSEFYDKERKVYDYTKFEGEGAA) of pneumococcal a-enolase. The peptide FYDKERKVY (spot no. 76) located between amino acid 248 and amino acid 256 was identified as the minimal internal binding motif for plasminogen. The length of the peptides spotted on the membrane is indicated by numbers and arrows in the figure. The sequences of the peptides spotted on the membrane in Fig. 3A and C are provided as Supplementary material and peptides with specific plasminogen activity are printed in bold.

Competitive inhibition of human plasminogen binding to pneumococci by synthetic peptides

A synthetic peptide with the sequence FYDKERKVYD which has been shown to bind plasminogen in the spot-membrane analysis was used to competitively inhibit binding of human 125I-plasminogen to pneumococcal cells in order to verify the pivotal role of the internal binding site identified by epitope mapping. The peptide was used in increasing amounts in plasminogen-binding studies. The results demonstrated the ability of the soluble peptide to competitively inhibit binding of radiolabelled human plasminogen to pneumococci Fig. 4A). On the basis of the results of the systematic mutational analysis, synthetic peptides with substitutions at critical amino acid residues of the internal plasminogen-binding motif was used in a competitive inhibition assay. In contrast to the wild-type peptide, a peptide FYAKERKVYD with individual amino acid substitution at position 3 (Asp’Ala) and a synthetic peptide FYDLGRLVYD with multiple amino acid substitutions at position 4, 5, and 7, respectively (Lys’Leu; Glu’Gly; Lys’Leu), were not able to inhibit binding of radiolabelled plasminogen to pneumococci (Fig. 4B). The studies demonstrated that a soluble peptide representing the minimal plasminogen-binding motif is capable of recruiting soluble plasminogen. The results therefore confirmed the spot-membrane analysis and the important function of the novel binding site as plasmin(ogen)-binding epitope.

Figure 4.

Competitive inhibition of 125I-labelled plasminogen binding to pneumococcal strain R6x by synthetic peptides.
A. The synthetic peptide FYDKERKVYD which showed strong plasminogen binding in the spot-membrane analysis was used to competitively inhibit binding of plasminogen to pneumococci. The binding assay was performed with radiolabelled 125I-plasminogen in the presence of increasing concentrations of unlabelled peptide. The results are expressed as a percentage of total radioactivity bound to the bacteria.
B. Synthetic peptides FYAKERKVYD and FYDLGRLVYD with substitutions of critical amino acids (underlined) were also used in this assay and are shown here for a representative experiment using 50 µg of the peptides. In contrast to the wild-type peptide, peptides with amino acid substitutions were not able to inhibit binding of radiolabelled plasminogen to pneumococci. Binding of 125I-plasminogen to pneumococci in the absence of a peptide was defined as 100%. Each value is a mean of triplicates.

Effect of site-directed mutagenesis of the internal binding motif of Eno on human plasminogen-binding activity of pneumococci

In order to investigate the effect of amino acid substitutions in the proposed internal plasminogen-binding motif FYDKERKVY of the pneumococcal α-enolase, genetic mutants were constructed. The codons of eno encoding C-terminal lysyl residues of the major plasmin(ogen)-binding protein of pneumococci at amino acid position 433 and 434 were deleted by insertion duplication mutagenesis. As a control a pseudomutant was constructed which contained, like the designated enodel mutant, an integration of the pJDC9 vector at the 3′-end of eno, but expressed the wild-type protein (Enowt). Amino acid substitutions in the proposed internal plasminogen-binding motif FYDKERKVY of the pneumococcal α-enolase were performed to obtain a mutant expressing Enoint with substitutions in the internal binding motif and a double mutant. The double mutant expressed an α-enolase designated Enoint/del with a deletion of the C-terminal lysyl residues as well as amino acid substitutions in the internal plasminogen-binding motif at positions 251 (Lys’Leu), 252 (Glu’Gly), and 254 (Lys’Leu) in Eno. These amino acid substitutions are identical to amino acid substitutions at position 4, 5, and 7 (Lys’Leu; Glu’Gly; Lys’Leu), respectively, of the synthetic peptide FYDLGRLVYD which was not able to inhibit binding of plasminogen to pneumococci in the competitive inhibition assay. The type 2 parental strain, the pseudomutant (pJDC9::eno) and the pneumococcal mutants expressing Enodel, Enoint, and Enoint/del were labelled with fluorescein isothiocyanate and the plasminogen-binding activity of the eno mutants was determined. Binding to human immobilized plasminogen was a function of the concentration of both of the pneumococci and the plasminogen (data not shown). The adhesion of the parental strain to plasminogen was defined as 100%. Binding of the pseudomutant (98.0% ± 0.3) and also binding of the pneumococcal mutant expressing Enodel (103% ± 0.4) was similar to the parental type 2 pneumococcal strain used. In contrast, binding of the enoint mutant expressing an α-enolase with amino acid substitutions in the proposed internal plasminogen-binding motif to human plasminogen was affected. The pneumococcal enoint mutant showed decreased binding (44.2% ± 0.8) of more than 50% to immobilized plasminogen. The binding of the enoint mutant to human plasminogen was also reduced (48.1% ± 0.5) to the same extend as double mutant enoint/delFig. 5). These results verified the novel internal plasminogen binding motif of the pneumococcal α-enolase and indicated its crucial role in Eno–plasminogen complex formation.

Figure 5.

Effect of amino acid substitutions in the internal binding motif of α-enolase on human plasminogen-binding activity of pneumococci. Binding of S. pneumoniae type 2 and corresponding eno mutants expressing wild-type Enowt (pseudomutant), Enodel, Enoint, and Enodel/int, respectively, to immobilized human plasminogen was tested. Plasminogen (2.0 µg) was immobilized in wells of a 96-well microtitre plate, blocked with 1% BSA/PBS and incubated for 1 h at 37°C with fluorescein isothiocyante-labelled pneumococci. Values are from a representative experiment and are expressed as percentage of the mean of triplicates of adherence to plasminogen. Binding of S. pneumoniae type 2 to human plasminogen was defined as 100%.

Virulence of pneumococcal eno mutants

Groups of six BALB/c mice were each inoculated intranasally with ∼108 cfu of mouse virulent wild-type S. pneumoniae D39 (serotype 2) and corresponding eno mutants. The median survival time of the eno mutants enodel (P = 0.065), enoint (P = 0.015) and the double mutant enoint/del (P = 0.002) were significantly higher than for those groups of mice inoculated with the parental pneumococcus D39. The survival rate of double mutant enoint/del (P < 0.0001) and of the enoint mutant (P < 0.0217) was significantly higher than that of groups inoculated with wild-type D39. The mouse intranasal challenge studies indicated that defined amino acid substitutions in the novel plasmin(ogen)-binding motif of Eno affect the virulence of S. pneumoniae and contribute to the pathogenesis of diseases Fig. 6).

Figure 6.

Virulence of S. pneumoniae D39 (wild type) and isogenic mutants expressing Enodel (pJDC9::enodel), Enoint (pJDC9::enoint) and Enoint/del (pJDC9::enoint/del). Survival time of mice was recorded after intranasal challenge with approximately 108 CFU of the strains and are shown here for a representative experiment. A time to death> 96 h indicates survival (mice surviving at 96 h did not die up to 10 days post infection). Each dot represents one mouse.

Discussion

Dissemination of bacteria in the host is promoted by acquisition of plasminogen and its subsequent conversion to active plasmin (Lottenberg et al., 1994; Lottenberg, 1997). Streptococcus pneumoniae binds plasminogen via the surface displayed α-enolase without affecting its proteinase acitivity. Previous results suggested that plasmin(ogen) binding to pneumococcal α-enolase is mediated by C-terminal lysyl residues. The role of C-terminal lysines was supported by the fact that binding of plasminogen to pneumococci was blocked by the lysine-binding sites I (LBS 1), which contains the first three kringles of plasminogen, and that amino acid substitutions of the lysine residues at the C-terminus led to a reduction of plasminogen-binding activity under reducing conditions (Bergmann et al., 2001).

This study provided evidence that a novel internal plasmin(ogen)-binding site in Eno contributes to the interaction of the pneumococcal α-enolase and human plasmin(ogen). Pneumococcal eno mutant expressing Enodel protein with a deletion of both the penultimate and ultimate lysyl residues, did not show any alteration in plasminogen-binding activity. This effect could be explained by the fact that additional bacterial ligands on the pneumococcal surface are able to mediate binding of human plasminogen (Bergmann et al., 2001) or that an additional binding site of Eno contributes to the Eno–plasminogen interaction. The plasminogen-binding activity of native Eno and C-terminal modified Enodel, EnoLL, and EnoKL as well as their ability to mediate a dose-dependent binding of plasminogen to pneumococci after reassociation to the bacterial cell surface, strengthened the idea that, in addition to the C-terminal lysyl residues, a further plasminogen-binding site is present in Eno. Moreover, these data are consistent with the observation that the plasmin(ogen)-binding protein SEN, the α-enolase of S. pyogenes, exhibits more than one interaction site for plasminogen and plasmin as indicated by Scatchard analysis (Pancholi and Fischetti, 1998). Binding of plasmin(ogen) to SEN, which showed an identity of 93% to the pneumococcal α-enolase, differs therefore from those of the reported eukaryotic plasminogen binding α-enolases (Redlitz et al., 1995). In view of these reports and our findings, the Eno-plasmin(ogen) interaction was analysed by surface plasmon resonance using immobilized Eno, EnoLL, EnoKL and Enodel proteins. The kinetic analysis of the formation and dissociation of the Eno-plasminogen complexes showed binding of plasmin(ogen) and LBS 1 to Eno, EnoLL, EnoKL and also Enodel. The kinetics were not effected by mass transfer limitations (data not shown). The two equilibrium constants, which were obtained by applying the heterogeneous ligand model, were in the lower nanomolar range (KD1 = 0.55 nM and KD2 = 86.2 nM) as described also for SEN (Pancholi and Fischetti, 1998). Because the dissociation of plasminogen from enolase is mediated by at least two reactions (Andronicos et al., 2000), the multiple component dissociation reaction model was used and confirmed the presence of a second plasmin(ogen)-binding motif in wild-type Eno protein of S. pneumoniae. In contrast, the kinetic data of Enodel and EnoLL, which contained a mutated plasmin(ogen)-binding site at the C-terminus, suggested the presence of only one binding site in these modified Eno proteins for plasminogen.

In order to precisely map the second plasminogen-binding site in Eno, peptides synthesized as array of spots were used. By using plasminogen as an antigen a nonameric peptide FYDKERKVY between amino acid 248 and 256 was identified as the minimal plasminogen-binding motif in Eno. Computer-aided primary and secondary structural analysis (data not shown) (Garnier et al., 1978; Janin and Wodak., 1978; Emini et al., 1985) indicated that the amino acids 248–256 of Eno are with highest probability exposed to the enolase surface and therefore accessible for the interaction with plasminogen. This peptide, without amino acid substitutions, is also present in SEN of S. pyogenes and in the α-enolase of S. intermedius (Pancholi, 2001). In the α-enolases of S. thermophilus (accession no. Q8VVB4 or Q9XDS7) and S. agalactiae (Glaser et al., 2002; Tettelin et al., 2002) the lysine residues at position four of the peptide are substituted by an alanine. Both enolases possess two lysine residues at the C-terminus which is highly homologous to the C-terminus of the pneumococcal enolase. The enolase of S. agalactiae was also identifed as a major outer membrane protein (Hughes et al., 2002). Comparison with sequences of other enolases deposited in the databases showed homologies of less than 71% among bacterial enolases and homologies less than 50% among other species (data not shown) with respect to the internal plasminogen-binding site in Eno of S. pneumoniae. The validation of FYDKERKVY (positions 248–256 in Eno) as the internal plasmin(ogen)-binding motif of Eno was achieved by using synthetic peptides and pneumococcal eno mutants which exhibit critical amino acid substitutions in the internal-plasmin(ogen)-binding motif of Eno. The identified nonameric plasminogen-binding peptide of Eno did not show any homology to the internal plasmin(ogen)-binding motif in PAM, which lacks C-terminal lysines (Wistedt et al., 1995) and also no homology to plasmin(ogen)-binding sites in fibrin(ogen) (Christensen, 1984; Voskuilen et al., 1987). However, because of the strong sequence identity of the α-enolase SEN of S. pyogenes to the pneumococcal α-enolase, we suggest that this motif is probably also the assumed second binding site for plasmin(ogen) in SEN. The growth of the pneumococcal eno mutants and the enzymatic activity of the mutated α-enolases Enodel, Enoint and Enoint/del was not affected (data not shown). Interestingly, the plasminogen-binding activity was already affected for the pneumococcal mutant expressing Enoint. These data indicated that the internal plasminogen-binding motif participates in the interaction of pneumococci and plasminogen. As a result of plasminogen binding bacteria subvert the fibrinolytic activity of plasmin to their own advantage for tissue invasion (Pancholi and Fischetti, 1998). The impact of the novel binding site on pneumococcal pathogenesis was demonstrated in an intranasal mouse infection model. Pneumococcal eno mutants expressing enolase with amino acid substitutions in the internal plasmin(ogen)-binding motif were substantially less virulent than the wild-type strain D39 indicating the significance of the novel binding site.

The contradictory results of plasmin(ogen)-binding acitivity of Enodel under physiological conditions and its loss of binding activity under reducing conditions implies that the interaction of the α-enolase with plasminogen, which is mediated by the internal binding motif, is dependent on the structure in Eno. Therefore we assume that this structure is sensitive to denaturation or reduction. This study precisely localized the internal plasmin(ogen)-binding domain of Eno of S. pneumoniae by employing different approaches. The binding affinity of physiological plasminogen-binding ligands like fibrin(ogen) and lysine analogues, e.g. EACA differs between the different kringle domains of plasminogen (Novokhatny et al., 1989; Matsuka et al., 1990). The binding of plasminogen to eukaryotic cells is mediated through an interaction with the LBS 1 (Miles et al., 1988). The kringle 2, which has only weak affinity to lysine analogues is predominantly involved in binding to PAM of S. pyogenes (Wistedt et al., 1998). Crystal structure analysis of the α-enolase in complex with the kringle domains will elucidate the precise role of the amino acids in the internal plasminogen-binding motif and will also indicate the kringle domain which mainly participates in this protein–protein cross-talk.

Experimental procedures

Bacterial strains, culture conditions and protein purification

Streptococcus pneumoniae serotype 2 strain ATCC 11733, mouse virulent serotype 2 strain D39 and its unencapsulated derivate R6x (Tiraby and Fox, 1973) were used for binding assays and mutational analysis. Pneumococcal strains were cultured at 37°C in Todd-Hewitt broth (Oxoid) supplemented with 0.5% yeast extract (THY) to mid-log phase or grown on blood agar plates (Becton Dickinson). Escherichia coli M15[pREP4] (Qiagen) was used as the host strain for recombinant pQE expression plasmids and cultured at 37°C on Luria–Bertani (LB) agar or grown on LB-agar containing 100 µgl−1 ampicillin and 50 µgl−1 kanamycin. Expression and purification of the His-tagged fusion proteins was performed as described previously (Bergmann et al., 2001).

Human plasminogen and plasmin

Human plasminogen and lysine binding site 1 (LBS 1) of plasminogen was purchased commercially (Sigma). LBS 1 contains the first three kringle of plasminogen. Human plasmin was generated from plasminogen by incubation with urokinase (20 IU ml−1, Sigma) for 1 h at 37°C in HBS-buffer (50 mM Hepes-NaOH pH 7.4, containing 1 mM MgCl2, 0.15 mM CaCl2, 0.1% gelatin) containing 40 mM lysine as described by Lottenberg et al. (1987). Conversion of glu-plasminogen to plasmin was monitored under denaturating conditions on 10% SDS-polyacrylamide gels.

Recombinant DNA techniques and site-directed mutagenesis

Transformation of E. coli with recombinant plasmids was achieved by electroporation (Calvin and Hanawalt, 1988). Plasmid DNA was isolated using Qiagen Plasmid Kit (Qiagen), and PCR products were purified using the PCR purification kit (Qiagen). T4 DNA ligase and restriction enzymes were purchased from New England Biolabs and used according to the manufacturer's instructions. Expression cloning of eno into pQE30 vector (QIAGEN), resulting in plasmid pQSB5 as well as site-directed mutagenesis of the eno gene were performed previously by inverse PCR. The PCR was done using DNA primers containing HindIII restriction enzyme sites for cloning and modifications in the reverse primer to introduce the appropriate base substitutions at the 3′ end of eno. This strategy finally resulted in substitution of the terminal lysine at position 434 and in substitution of both the penultimate and ultimate lysyl residues at positions 433–434 respectively. The lysine residues were replaced by leucines and the modified Eno proteins were designated EnoKL and EnoLL respectively (Bergmann et al., 2001). Deletion of the codons for the C-terminal lysyl residues at position 433 and 434 was also conducted by inverse PCR using oligonucleotides SB 49 (5′-AAGCTTAATTAGCTGAGCTTGGAC-3′) and SB 45 (5′-AAGCTTTTAAAGGTTGTAGAAT GATTTCAATC-3′) and plasmid pQSB5 as DNA template. This resulted in Enodel which is expressed by plasmid pQSB13. Amino acid substitutions in the internal plasmin(ogen) binding epitope of Eno was performed using the QuikChange™ site-directed mutagenesis kit (Stratagene) according to the protocol of the manufacturer. Briefly, recombinant mutated Eno proteins containing CTT (code for leucine), GGT (code for glycine), and CTT (code for leucine) substitutions at positions 251 (Lys’Leu), 252 (Glu’Gly), and 254 (Lys’Leu) of α-enolase were generated using one pair of polyacrylamide gel-electrophoresis-purified complementary oligonucleotide primers SB 58 (5′-GTGCTTCATCAGAATTCTACGATCTTGGTCGTCTT GTT TACGACTACACTAAATTTGAAG-3′) and SB 59 (5′-CTTCAAATTTAGTGTAGT CGTAAACAAGACGACCAA GATCGTAGAATTCTGATGAAGCAC-3′). Plasmids pQSB5 and pQSB13, respectively, were used as templates for site-directed mutagenesis to obtain mutated α-enolases Enoint and Enoint/del, both containing amino acid substitutions in the internal plasminogen-binding site. In Enoint/del the C-terminal lysyl residues are also deleted.

Construction of pneumococcal mutants

Pneumococcal mutants with altered C-terminal sequences and/or amino acid substitutions in the internal plasmin(ogen) binding motif were constructed by insertion duplication as described previously (Morrison et al., 1984 Insert DNA-sequences of pQSB5, and pQSB13 were digested with BamHI and HindIII, subcloned into similarly digested integration vector pJDC9 resulting in pJDC9::eno and pJDC9::enodel. Plasmids were transformed by electroporation into E. coli DH5α. Transformants were selected on LB-agar containing 250 µg ml−1 erythromycin, and plasmids were examined for pneumococcal insert DNA by sequencing. S. pneumoniae mutants of the wild-type strains ATCC 11733 (type2), D39 (type 2) and R6x expressing mutated Eno proteins were obtained by transformation with plasmid DNA from selected recombinant E. coli using the competence stimulating peptide to enhance the transformation efficiency (Håvarstein et al., 1995 Integration of the vector, resulting in expression of wild-type Eno and mutated α-enolase Enodel, respectively, was confirmed by Southern blot analysis, PCR, and subsequent sequence analysis. Pneumococci transformed with pJDC9::eno and therefore expressing wild-type α-enolase were designated pseudomutants and used in the experiments as a control strain to exclude downstream effects of the mutagenesis. In order to obtain pneumococcal mutants expressing Enoint and Enoint/del proteins, site-directed mutagensis was performed as described above. For this, the plasmids pJDC9::eno and pJDC9::enodel, respectively, as DNA templates and primers SB58 and SB59 were employed. The site-specific mutations in eno were verified by Southern blot analysis, PCR and subsequent Sau3A restriction analysis and by sequence analysis using the ABI PRISM dye terminator cycle sequencing (Perkin-Elmer).

Preparation of spot synthesized peptides on membranes

The complete amino acid sequence of Eno was divided into 141 peptides consisting of 15 amino acids each, with an offset of three amino acids. The peptides were synthesized as array of spots on aminopegylated cellulose membrane (AIMS Scientific Products GmbH, Germany) (Frank and Overwin, 1996). On a second spot membrane, a 36-amino-acid region of the identified plasminogen binding domain (corresponding to amino acid 232–267 of Eno) was divided into 198 overlapping synthetic peptides with an offset of one amino acid. The membrane consisted of 12 series each with lengths varying from four amino acids up to 15 amino acids per spot. Each series spanned parts of the 27-amino-acid region. This was recognized to be positive for plasminogen binding in the first spot membrane described above (see Supplementary material). In order to identifiy critical amino acids in the identified internal binding motif FYDKERKVYD a third membrane was constructed for systematic mutational analysis. This spot-membrane contained 200 immobilized decapeptides with individual amino acid substitutions at each position of the original peptide sequence FYDKERKVYD. Each amino acid of the motif was substituted individually with each of the 20 different amino acids. For assaying plasminogen binding, membranes were treated as described previously (Hammerschmidt et al., 2000). Binding of plasminogen was performed using 10 µg ml−1 plasminogen (Sigma) in conjunction with goat anti-plasminogen antibody (1:200) and peroxidase labelled anti-goat antibodies (1:1000). Binding was detected by staining with a substrate solution containing 1 mg ml−1 4-chloro-1-naphthol and 0.1% H2O2 in PBS.

Blot-overlay analysis and competitive inhibition assay

Binding of radioiodinated ligands were performed as described previously (Bergmann et al., 2001). Briefly, human plasminogen (Sigma) was radiolabelled with 125I by a standard chloramin T method (Chhatwal et al., 1987). In blot-overlay assays the membrane spotted Eno proteins were blocked with 10% skim milk (Oxoid) in 10% PBS before the binding reaction with 40 ng (approximately 200000 c.p.m.) 125I-plasminogen was carried out. Binding was detected by exposure of the membrane to X-ray film. In competitive inhibition experiments binding of 20 ng 125I-labelled plasminogen to 4 × 108 pneumococci was determined in the presence of increasing concentrations of unlabelled synthetic peptides. These peptides represented the internal plasmin(ogen) binding motif FYDKERKVYD or peptides with amino acid substitutions in the proposed binding motif (FYAKERKVYD and FYDLGRLVYD) and were used in concentrations from 0.01 ng up to 200 µg in the inhibition assay.

Assay for binding of pneumococci to immobilized human plasminogen

To study the binding of S. pneumoniae and corresponding eno mutants to plasminogen, fluorescein isothiocyanate (FITC)-labelled pneumococci were added to immobilized plasminogen. Two micrograms or 0.5 µg of human plasminogen was coated per well of a 96-well microtitre plate (polystyrene surface) at 4°C overnight. Bacteria were grown to mid-log phase and after washing with 0.1 M sodium carbonate buffer pH 9.2, 109 bacteria were FITC-labelled in 500 µl of a sodium carbonate buffer FITC solution (1 mg ml−1) for 1 h. Extensively washed FITC-labelled bacteria were added to washed and BSA (1% in PBS) blocked (2–3 h) plasminogen containing in the wells and incubated for 1 h for binding. Bacteria were used in serial dilutions starting with approximately 2.5 × 107 FITC-labelled bacteria per well. Fluorescence was measured at 485 nm/538 nm (excitation/emission) using a Titertek Fluoroskan II before and after a washing procedure which eliminates unbound bacteria. The values of the bound bacteria were determined and binding of the parental S. pneumoniae strain used in the assay was defined as 100%.

Surface plasmon resonance

The association and dissociation reactions of glu-plasminogen (Sigma), plasmin and LBS 1 to recombinant and mutated Eno proteins were analysed on the BIAcore optical biosensor (BIAcore 2000 system) using CM 5 sensor chips. Covalent immobilization of Eno proteins was performed essentially as described (Nice et al., 1996). Briefly, Eno wild-type protein and modified Eno proteins (each 1 mg ml−1 in 20 mM sodiumacetate, pH 4.0) were coupled at 10 µl min−1 onto N-hydroxysuccinimide (NHS, 0.05 M)/N-ethyl-N′-(diethylaminopropyl)carbodiimide (EDC, 0.2 M)-activated sensor chips (70 µl, 10 µl min−1). Binding of analytes was performed in HBS BIAcore running buffer (10 mM Hepes, 150 mM NaCl, 1.4 mM EDTA, 0.05% Tween-20, pH 7.4) at 20°C using a flow rate of 10 µl min−1 in all experiments. The affinity surface was regenerated between subsequent sample injections of analytes with two times 10 µl of 10 mM NaOH and 20 mM NaOH. Binding was assayed at least in duplicate using independently prepared sensor chips.

Analysis of BIAcore sensorgram data

The interaction kinetics of plasmin(ogen) and LBS 1 binding to immobilized Eno proteins was analysed from raw data of the BIAcore sensorgrams 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 + Bbsl00076AB). The two-state reaction/conformational change model (A + Bbsl00076ABbsl00076A*B) was applied taking into consideration the possibility of conformational changes in the course of binding interaction. To elucidate whether plasmin(ogen) binds independently to two ligand binding sites, data were also evaluated according to the model of heterogeneous ligand model/parallel reaction (A + B1bsl00076AB1; A + B2bsl00076AB2). Also the bivalent analyte model (A + Bbsl00076AB; AB + Bbsl00076AB2), describing the binding of a bivalent analyte to a monovalent ligand, was applied. Furthermore, a multiple component dissociation reaction model (Ai + Bibsl00076AiBi) which was set out as a suitable kinetic model to evaluate the dissociation rate of human α-enolase and human plasminogen was used to fit the experimental data. This model, which takes into consideration that two different dissociation rates contribute to the dissociation of plasminogen, does not require the concentration of the analyte plasminogen (Andronicos et al., 2000). All results recorded in this report were within the typical dynamic ranges of BIAevaluation 3.0 software. For each evaluation, a minimum of six data sets corresponding to plasminogen binding reactions at concentrations between 0.19 nM and 500 nM were analysed.

Intranasal mouse infection model of virulence

In order to prepare bacteria for inoculation into the nostrils of mice, cultures of mouse virulent S. pneumoniae strain D39 (serotype 2) were grown in THY medium containing 10% fetal calf serum to a density of approximately 108 cells ml−1. Groups of six BALB/c mice, anaesthetized by inhalation of metoxyfurane (Metofane: Janssen-Cilag), were intranasally infected in at least two independent experiments with 20 µl of a suspension containing 108 wild-type S. pneumoniae D39 and corresponding eno mutants respectively. The groups were monitored for up to 10 days. Differences in median survival time between groups were analysed by the Mann–Whitney U-test (two tailed). Differences in the overall survival rate between groups were analysed by the Fisher's exact test.

Acknowledgements

The authors are grateful to W. Tegge for peptide synthesis and S. Daenicke for technical assistance. The authors are also grateful to P. Matzander for critical reading of the manuscript and to J. Gerber for statistical analysis. This work was partially supported by the Deutsche Forschungsgemeinschaft (Sonderforschungsbereich 587, Teilprojekt A6) and BMBF (CAPNETZ). Part of this work was carried out as a partial fulfilment of the diploma thesis of D.W.

Supplementary material

The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/mmi/mmi3557/mmi3557sm.htm

Fig. S3A. Peptide array of the 434 amino acid sequence of the α-enolase of Streptococcus pneumoniae used in the initial screening to map the internal plasminogen-binding site. The amino sequence of the α-enolase was divided into 141 overlapping synthetic peptides, of 15 amino acids each, with an offset of 3 amino acids. Twenty-five peptides were immobilized in one line. Binding of human plasminogen was detected to spot numbers 79, 80, 81, 82 and 83.

Fig. S3C. Sequences of synthetic peptides which were used in a peptide array to narrow down the internal plasminogen-binding site within α-enolase of Streptococcus pneumoniae. One hundred and ninety-eight overlapping peptides, with peptide length from four amino acids up to 15 amino acids residues per spot, were analysed for binding of human plasminogen. The amino acid sequence selected for the spot-membrane corresponds to amino acids 232–267 (VPGKDV FIGFDCASSEFYDKERKVYDYTKFEGEGAA) of pneumococcal α-enolase. The peptide FYDKERKVY (spot no. 76) located between amino acid 248 and amino acid 256 was identified as the minimal internal binding motif for plasminogen.

Ancillary