Binding of the low-density lipoprotein by streptococcal collagen-like protein Scl1 of Streptococcus pyogenes

Authors


*E-mail slukomski@hsc.wvu.edu; Tel. (+1) 304 2936405; Fax (+1) 304 2937328.

Summary

Several bacterial genera express proteins that contain collagen-like regions, which are associated with variable (V) non-collagenous regions. The streptococcal collagen-like proteins, Scl1 and Scl2, of group A Streptococcus (GAS) are members of this ‘prokaryotic collagen’ family, and they too contain an amino-terminal non-collagenous V region of unknown function. Here, we use recombinant rScl constructs, derived from several Scl1 and Scl2 variants, and affinity chromatography to identify Scl ligands present in human plasma. First, we show that Scl1, but not Scl2, proteins from different GAS serotypes bind the same ligand identified as apolipoprotein B (ApoB100), which is a major component of the low-density lipoprotein (LDL). Scl1 binding to purified ApoB100 and LDL is specific and concentration-dependent. Furthermore, the non-collagenous V region of the Scl1 protein is responsible for LDL/ApoB100 binding because only those rScls, constructed by domain swapping, which contain the V region from Scl1 proteins, were able to bind to ApoB100 and LDL ligands, and this binding was inhibited by antibodies directed against the Scl1-V region. Electron microscopy images of Scl1–LDL complexes showed that the globular V domain of Scl1 interacted with spherical particles of LDL. Importantly, live M28-type GAS cells absorbed plasma LDL on the cell surface and this binding depended on the surface expression of the Scl1.28, but not Scl2.28, protein. Phylogenetic analysis showed that the non-collagenous globular domains of Scl1 and Scl2 evolved independently to form separate lineages, which differ in amino acid sequence, and these differences may account for the variations in binding patterns of Scl1 and Scl2 proteins. Present studies provide insight into the structure-function relationship of the Scl proteins and also underline the importance of lipoprotein binding by GAS.

Introduction

The Gram-positive bacterium Streptococcus pyogenes (Group A Streptococcus; GAS) has long been recognized as a major pathogen of humans, causing a variety of diseases ranging from superficial infections of the throat and skin, to life-threatening invasive infections of deeper tissues, the blood stream, and multiple organs. This universality of the pathogen requires intimate interactions between GAS cells and a variety of host environments such as mucosal surfaces, skin and connective tissue, and blood and lymphatic systems. Interactions with plasma factors are essential for GAS survival not only in the blood stream but also during localized infections of the mucosa and soft tissues. The importance of these interactions is reflected by the fact that several plasma constituents are bound by various GAS-surface proteins including fibronectin, fibrinogen, immunoglobulins, complement inhibitors factor H and C4BP, a-macroglobulin and kininogens (Navarre and Schneewind, 1999).

Two related cell-surface streptococcal collagen-like proteins, Scl1 and Scl2 (also known as SclA and SclB, respectively), have recently been identified (Lukomski et al., 2000a; 2001; Rasmussen et al., 2000; Rasmussen and Bjorck, 2001; Whatmore, 2001). Both Scl proteins have conserved signal peptides and cell-wall/membrane (WM) regions. However, the extracellular portions of mature Scl proteins are highly variable and contain the arbitrarily designated N-terminal variable (V) regions and the adjacent collagen-like (CL) regions (Lukomskiet al., 2000a). The non-collagenous V regions differamong Scl1 and Scl2 variants in length and primary sequence, and, similarly, CL regions are of disparate lengths and vary in GXY-repeat content (Han et al., 2006). In addition, Scl1 proteins harbour the so-called linker (L) region containing different numbers of well-conserved direct repeats between the CL and WM regions. The scl1 and scl2 genes are located on the opposite sites of the GAS chromosome and are divergently oriented (Ferretti et al., 2001). They are widely distributed among GAS strains; however, the expression of the two scl genes is mediated by different regulatory mechanisms. The transcription of the scl1 gene is positively regulated by Mga, the positive transcriptional regulator of virulence genes in GAS (Rasmussen et al., 2000; Lukomski et al., 2001; Almengor and McIver, 2004), whereas the production of the full-length Scl2 protein is regulated at the translational level and depends on a number of short DNA sequence repeats, CAAAA, located within the scl2-signal sequence (Lukomski et al., 2001; Rasmussen and Bjorck, 2001; Whatmore, 2001). Furthermore, sequence data suggest that only some GAS isolates would simultaneously express both Scl proteins, and differences between representatives of various M types were observed.

The function of Scl proteins is not well known. Initial reports suggest that Scl proteins may contribute to GAS adherence to human cells (Lukomski et al., 2000a; Rasmussen and Bjorck, 2001) and that the adhesive properties of Scl proteins arise from their surface exposure as well as unique structure and organization. First, the CL regions of different Scl variants form stable CL triple helices (Xu et al., 2002; Han et al., 2006). Second, at least some Scl-CL regions are capable of interacting with the human collagen receptors, such as α2β1 integrins, and inducing intracellular signalling that results in protein phosphorylation of the focal adhesion pathway (Humtsoe et al., 2005). Third, different Scl variants share common ‘lollipop-like’ structural organization with stalks made of the CL domains and globular heads made of the non-collagenous V regions (Xu et al., 2002; Han et al., 2006). Thus, Scl proteins are both structural and functional collagen mimics.

The present study examines whether Scl proteins recognize and interact with plasma components. A series of purified recombinant Scl proteins (rScl), derived from several Scl1 and Scl2 variants, were used in search for plasma ligands by affinity chromatography. Apolipoprotein B (ApoB100)/low-density lipoprotein (LDL) complexes were identified as binding ligands for all Scl1- but not Scl2-derived constructs. These rScl proteins bind purified ApoB100 and LDL preparations in a concentration-dependent manner. Furthermore, the globular V domains of the Scl1 proteins seem to be responsible for ApoB100/LDL binding. The same binding pattern is observed with M28-type GAS cells expressing or lacking Scl1.28 protein. To our knowledge, this is first demonstration that the GAS surface protein, Scl1, interacts with human plasma component, an LDL.

Results

Recombinant rScl proteins bind human plasma component

Group A Streptococcus is a specialized human pathogen that interacts with the surrounding environment of the host through cell-surface components. Here, we tested the possibility that different variants of Scl1 and Scl2 proteins interact with plasma component(s). Several recombinant Scl proteins (rScl) derived either from Scl1 (rScl1) or Scl2 (rScl2) variants were produced in Escherichia coli (Table 1) (Han et al., 2006). These variants were selected for cloning because they originate from strains of epidemiologically important GAS-M types (M1, M12, M28) (Stevens and Kaplan, 2000) or contain sequence motifs of known biological function, such as potential integrin-binding sites GLPGER (Humtsoe et al., 2005) and RGD (Ruoslahti, 1996). All rScl constructs formed triple helices and were organized into lollipop-like structures as determined by circular dichroism spectroscopy and electron microscopy (EM) (Fig. 6B, Fig. S1) (Han et al., 2006).

Table 1.  Bacterial strains and plasmids used in the present study.
Strain or plasmidRelevant characteristicsSource or reference
  • a. 

    Resistance markers expressed upon induction with IPTG.

Strains
 GAS
  MGAS 6143scl1.28/scl2.28; wild-type strainThis work
  MGAS 6143 scl1scl1.28-inactivated isogenic mutant strainThis work
  MGAS 6143 scl2scl2.28-inactivated isogenic mutant strainThis work
  MGAS 6143 scl1/scl2scl1.28/scl2.28-inactivated double mutant strainThis work
 E. coli
  BL21Cloning strainStratagene
  DH5αCloning strainPromega
  XL1-BlueCloning strainStratagene
Plasmids
 pSL60-1 to -3Cmr (Spr);a spc1- to 3 cassettes in pBC KSLukomski et al. (2000b)
 pCR Script SK+Apr; E. coli cloning vectorStratagene
 pASK-IBA2Apr; E. coli cloning and expression vectorIBA-GmbH
 pVA838Cmr Emr; source of erm geneAB057644
 pSL164, pSL165, pSL166Apr (Emr);a erm1-3 cassettes in pCR Script SK+This work
 pFW11Emr or Spr; E. coli cloning vectorPodbielski et al. (1996)
 pFW14Cmr; source of cat genePodbielski et al. (1996)
 pSL146Spr; 3′-flanking region of scl1.28 in pFW11This work
 pSL147Spr; 5′-flanking region of scl1.28 in pSL146This work
 pSL151Spr; 3′-flanking region of scl2.28 in pFW11This work
 pSL152Spr; 5′-flanking region of scl2.28 in pSL151This work
 pSL168Cmr Spr; cat gene from pFW14 in pSL147This work
 pSL169Cmr Spr; cat gene from pFW14 in pSL152This work
 pSL170Cmr Emr; erm2 cassette from pSL165 in pSL168This work
 pSL171Cmr Spr; spc2 cassette from pSL60-2 in pSL169This work
 pSL144Apr; scl1.1 gene from MGAS5005 encoding P144Xu et al. (2002)
 pSL157Apr; scl1.1 CL region from MGAS5005 encoding P157Xu et al. (2002)
 pSL161Apr; scl1.28 gene from MGAS6274 encoding P161Han et al. (2006)
 pSL163Apr; scl2.28 gene from MGAS6274 encoding P163Xu et al. (2002)
 pSL176Apr; scl1.41 gene from MGAS6183 encoding P176Humtsoe et al. (2005)
 pSL177Apr; scl2.4 gene from MGAS321 encoding P177Han et al. (2006)
 pSL178Apr; scl2.77 gene from MGAS6191 encoding P178Han et al. (2006)
 pSL181Apr; chimeric construct containing P163CL and P176VHumtsoe et al. (2005)
 pSL183Apr; chimeric construct containing P177CL and P144VThis work
 pSL184Apr; chimeric construct containing P177V and P144CLThis work
 pSL186Apr; scl1.12 gene from MGAS6139 encoding P186Han et al. (2006)
Figure 6.

Visualization of the Scl1–LDL interaction by electron microscopy.
A. Field demonstrating negative-stained LDL seen as vesicles or spheres.
B. Rotary shadowed P176 with characteristic lollipop-like structural organization. The collagen-like domain of P176 is depicted by an arrow.
C. Representative example of an LDL vesicle seen in rotary shadowed samples with demarcated periphery decorated with ApoB100. Presumed βα1 globular domain of ApoB100 is depicted by open arrowheads.
D–I. Rotary shadowed complexes of P176 with LDL present in the mixture show associations of LDL spheres with rScl through globular heads of the lollipops. Collagenous tails of Scl lollipops are marked with arrows. Scale bars represent 100 nm.

The search for Scl-specific ligands present in human plasma was performed using affinity-chromatography columns containing Strep-Tactin Sepharose. Recombinant proteins were immobilized onto affinity columns through a C-terminal strep-tag II, which allowed each construct to be oriented in a similar manner to that on the GAS surface. The rScl1 proteins include P144, P161, P176 and P186, which are derived from serotype M1, M28, M41 and M12 GAS respectively. The rScl2 proteins, P163, P177 and P178 are from M28, M4 and M77 GAS respectively. Human plasma was passaged over rScl-bound columns or a control column, containing Strep-Tactin resin but no rScl protein (No rScl lane). The presumed complexes of rScl and bound ligand(s) were eluted from the columns, TCA-precipitated, and analysed by 12% SDS-PAGE (Fig. 1A).

Figure 1.

Interactions of recombinant Scl proteins with human-plasma proteins. Affinity-chromatography columns contained purified rScl proteins. 10 ml of human plasma was passaged over each column, including the control column with no rScl protein immobilized (No rScl lane). After washing, the complexes of rScl proteins and their ligands were eluted from the columns and TCA-precipitated.
A. Resolution of plasma proteins by 12% SDS-PAGE stained with Rapidstain. Bands corresponding to rScl polypeptides eluted from the columns are marked with solid (rScl1) and open triangles (rScl2). The predominant high-molecular-weight band present in all samples eluted from the columns containing rScl1 constructs was identified by RP-MS-MS analysis as apolipoprotein B (ApoB100) is marked by an arrow. Minor ∼33-kDa band was identified as ApoE (marked with a star). M; molecular mass markers (kDa).
B. Detection of plasma proteins and rScl constructs by dot immunoblotting. Protein samples shown in A were spotted onto a nitrocellulose membrane. Presence of ApoB100 was tested with anti-ApoB100 antibodies (upper row) and rScls were detected with a Strep-Tactin-HRP conjugate (bottom row). Immunoreactivity was visualized using chemiluminescence substrate.

Two general binding patterns were observed that correlated with Scl-protein type, but not with M-serotype. All samples eluted from the columns with immobilized rScl1 constructs contained a dominant high-molecular-weight band greater than 250 kDa. This band was not evident in eluates from columns containing rScl2 constructs. Furthermore, protein samples prepared from the control no-rScl column did not contain the high-molecular-weight band, indicating that binding was specific to immobilized rScls. In addition, the presence of a minor band of ∼33 kDa was also observed in rScl1 samples. Our results indicate that Scl1 and Scl2 proteins differentially interact with plasma component(s), whereas different Scl1 variants share common ligand-binding patterns.

Identification of the Scl1-protein ligand present in human plasma

To identify the major high-molecular-weight proteins that bound all rScl1 constructs tested, bands were excised from the gel, and the trypsin-digested fragments were analysed by reverse-phase chromatography coupled nano-electrospray ionization ion trap mass spectrometry (RP-MS-MS). The high-molecular-weight protein present in each sample was identified as apolipoprotein B-100, or ApoB100. The peptide fragments (representative example shows data for P144) exhibited a protein coverage of 47.5% by amino acid content (2168/4563) and 47.2% by mass (243381/515562) when compared with the known sequence of ApoB100 (Chen et al., 1986). Small amounts of the ApoE protein (∼33-kDa band) were also detected in rScl1-containing samples, and this observation is the subject of a separate investigation.

To validate RP-MS-MS data, the same samples eluted from the affinity columns were subjected to dot immunoblotting analysis (Fig. 1B). While all samples obtained from the columns containing immobilized rScl1 constructs reacted with anti-ApoB100-specific antibodies, samples obtained from the columns immobilized with rScl2 constructs did not show detectable amounts of ApoB100 protein. As expected, all samples contained eluted rScl polypeptides. Importantly, lack of immunoreactivity observed in the ‘No rScl’ control indicates that interactions were specific to immobilized rScl proteins and not to the affinity columns. Our results suggested that different Scl1 variants share the common biological feature of binding to ApoB100 protein or the ApoB100-containing plasma component, such as LDL.

Scl1 proteins bind to purified LDL and ApoB100

Human plasma contains several different lipoproteins, including LDL, which is the major transport particle of ApoB100 (Segrest et al., 2001). Thus, our results could indicate that Scl1 proteins bind LDL. To test this, rScl proteins were immobilized onto affinity-chromatography columns and tested for direct binding to purified LDL (Fig. 2A). SDS-PAGE analysis revealed the presence of LDL in all samples containing rScl1 constructs P144, P161, P176 and P186. However, LDL was absent in a sample containing the rScl2 construct, P163. LDL also did not bind to a Strep-Tactin column lacking rScl protein.

Figure 2.

Verification of rScl binding to human LDL and ApoB100. Purified preparations of LDL and ApoB100 were applied to affinity columns containing immobilized rScl1 constructs P144, P161, P176 and P186 (solid triangles). Column containing recombinant P163 protein (rScl2, open triangle) and a column without rScl protein, were used as controls. Protein complexes eluted from the columns were TCA-precipitated and analysed in 8% SDS-PAGE stained with RAPIDstain.
A. LDL binding to rScls.
B. ApoB100 binding to rScls. LDL and ApoB preparations were also used as markers.

Low-density lipoprotein binding might not necessarily involve ApoB100 directly. For example, the pH 6 antigen protein of Yersinia pestis binds to the lipid component of LDL rather than ApoB100 (Makoveichuk et al., 2003). Therefore, the affinity chromatography-binding assay was carried out using purified ApoB100 (Fig. 2B). Once again, all four rScl1 constructs bound purified ApoB100, but P163 did not. Furthermore, rScl1 binding was not accompanied by the degradation of LDL or ApoB100 because no change in the banding patterns was seen in stained SDS-PAGE (Fig. 2) or in Western blots (data not shown).

Binding of LDL and ApoB100 to rScl proteins was further tested by an enzyme-linked immunosorbent assay (ELISA) (Fig. 3). rScl1 and rScl2 proteins were immobilized into Strep-Tactin-coated microplate wells and exposed to purified LDL or ApoB100 preparations. Ligand binding to rScls was detected with anti-LDL or anti-ApoB100 antibodies respectively. All four rScl1 constructs tested (P144, P161, P176 and P186) were able to bind to LDL and ApoB100 to various degrees. P144 and P176 seemed to bind both ligands to a greater degree than did P161 and P186. The rScl2 constructs (P163, P177 and P178) were negative for binding. In addition, proteolytic fragmentation of ApoB100 with trypsin led to a partial decrease in the interaction (data not shown). These results support the Scl1-ApoB100/LDL binding initially detected by affinity chromatography and suggest that GAS binds to plasma lipoproteins through interactions of the surface protein Scl1 and the LDL component ApoB100.

Figure 3.

Binding between rScl proteins and LDL or ApoB100 detected by ELISA. rScl1 (P144, P161, P176 and P186) and rScl2 (P163, P177 and P178) proteins were immobilized into Strep-Tactin-coated microplate wells at a concentration of 0.1 μM. LDL (0.5 nM) or ApoB100 (1 nM) were added to the rScl-immobilized wells. Bound ligands were reacted with the corresponding anti-LDL or anti-ApoB100 antibodies and detected with the appropriate secondary antibodies conjugated to HRP. Colour reaction was developed with an HRP substrate and absorbance was recorded at 415 nm. Mean absorbance values ± SD from triplicate wells are shown after subtracting the OD values obtained for the control LDL and ApoB100 wells with no rScl; the OD values recorded in the wells immobilized with rScl2 constructs were, after subtraction, of a slightly negative values which is not shown on the graph.

Interactions of LDL and ApoB100 with Scl1 protein

Initially, the amounts of LDL and ApoB100 bound to rScl constructs were estimated by ELISA. Strep-Tactin-coated microplate wells were immobilized with 0.1 μM of rScls and incubated with increasing concentrations of both LDL and ApoB100 ligands (up to 5 nM). The binding of LDL and ApoB100 to rScls was concentration-dependent and varied among constructs (data not shown).

We further examined binding specificity of selected rScl proteins to LDL using surface plasmon resonance (SPR). First, LDL was captured by polyclonal anti-LDL antibodies, which had been immobilized onto a flow cell of a CM5 chip. Non-specific IgG antibodies were also immobilized onto a flow cell as a background control. Different concentrations of LDL were injected and their binding was specific to anti-LDL antibodies. We were able to saturate the anti-LDL antibodies with a 0.1 mg ml−1 concentration of LDL, resulting in an RU of ∼300. Increasing concentrations of rScl proteins, P161, P163, or P144, were then passed over the surface containing captured LDL (Fig. 4). Responses to the reference anti-LDL antibody flow cell, which have no LDL captured, were subtracted from the data. As we expected, P163 did not bind to LDL. P144 showed better binding to LDL than did P161. The association rate (kon) and dissociation rate (koff) for the P144 and LDL interaction were 2.1 × 105 Ms−1 and 1.5 × 10−2 s−1, respectively, resulting in a KD of 69.4 nM. P161 binds to LDL with kon of 3.1 × 105 Ms−1 and koff of 8.25 × 10−3 s−1, resulting in a KD of 26.6 nM. Taken together (Figs 1–4), we demonstrated that rScl1 proteins derived from different Scl1 variants tested here interact with LDL through direct binding to ApoB100.

Figure 4.

SPR analysis of rScl proteins binding to LDL. LDL was captured by the anti-LDL antibodies at the saturation concentration of 0.1 mg ml−1. Non-specific binding of LDL to the control unrelated IgG antibodies was subtracted from LDL binding to anti-LDL antibodies. Increasing concentrations of rScl proteins (1, 10, 30, 60 and 100 nM) were tested for ability to bind to captured LDL. All sensorgrams were corrected for non-specific background detected in reference cell containing only anti-LDL antibodies.

Low-density lipoprotein/ApoB100 binding involves the V region of Scl1 proteins

As all Scl1-based rScl constructs (M1/P144, M28/P161, M12/P176 and M41/P176) contain the V, CL and L regions of Scl1 proteins, it was important to determine which region is responsible for conveying the binding specificity to LDL. Several lines of evidence support that the V region of Scl1 proteins is the binding domain for the LDL/ApoB100 ligand (Fig. 5). First, samples eluted from the column with immobilized P157, which is a truncated derivative of P144 containing the CL and L regions but not the V region, lost the ability of LDL binding, suggesting that the CL-L portion of Scl1 is not required for the interaction. Next, chimeric rScl constructs were generated by domain swapping. Two different LDL binding-positive rScl1 constructs, P144 and P176, and two LDL binding-negative rScl2 constructs, P163 (M28) and P177 (M4), were used to construct chimeric molecules. P183 contains the V region of P144 and the CL region of P177. Conversely, P184 is composed of the V region of P177 and CL region of P144. P181 is made up of the V region of P176 and the CL region of P163. These chimeric constructs were tested for lipoprotein binding using affinity chromatography with human plasma (Panel A) and by ELISA with purified LDL and ApoB100 (Panel B). Using both methods, LDL/ApoB100 binding was detected only in samples with rScl constructs containing a V region originating from Scl1 proteins (P181 and P183). Thus, we conclude that LDL binding by Scl1 proteins involves direct interactions between the Scl1 globular V domain and the ApoB100 protein of LDL.

Figure 5.

Mapping of Scl1 domain involved in LDL/ApoB100 binding.
A. Affinity chromatography-binding assay. rScl constructs (solid triangles) containing various combinations of the V and CL regions were immobilized to affinity columns. Protein samples eluted from the columns were analysed by 8% SDS-PAGE.
B. LDL/ApoB100 binding by ELISA. A series of chimeric rScl proteins (P181, P183 and P184) were immobilized onto Strep-Tactin-coated microplate wells and LDL or ApoB100 were added to the wells. Bound ligands were reacted with the corresponding anti-LDL or anti-ApoB100 antibodies and detected with the appropriate secondary antibodies conjugated to HRP. The reaction was developed with an HRP substrate and absorbance was recorded at 415 nm. Mean absorbance values ± SD from triplicate wells are shown after subtracting the OD values obtained for the control LDL and ApoB100 wells with no rScl.

Visualization of LDL-P176 interactions using EM

Analytical and structural studies showed LDL as a quasi-spherical particle of about 20 nm in diameter with a size range between 18 and 25 nm, which is attributed to changes in lipid content (Segrest et al., 2001). Our electron microscopic data of the negative-stained LDL preparation revealed spheric vesicles about 22 nm in diameter (Fig. 6A). However, neither ApoB100 nor rScl structures were seen using this technique. Similar data were reported previously indicating that ApoB100 could not be resolved on the surface of negative-stained LDL, despite its great size of around 530 kDa [it is a single-chain protein composed of 4536 residues (Chen et al., 1986)]. Several techniques were used to render ApoB100 model structures, including electron cryomicroscopy, immunoelectron microscopy and low-resolution crystallography (Chatterton et al., 1995; Orlova et al., 1999; Lunin et al., 2001). These studies identified a ‘knob-shaped’ electron dense region on the surface of the LDL particle corresponding to the N-terminal globular βα1 domain of ApoB100. Here, we used rotary shadowed LDL samples to study Scl-LDL interactions.

Low-density lipoprotein vesicles seen in rotary shadowed samples appear to be collapsed and have a demarcated periphery which is decorated by a studded structure most likely corresponding to the N-terminal globular ba1 domain of ApoB100 (Fig. 6C; open arrowheads). The rScl protein P176 (example shown in Fig. 6B) was combined with LDL in solution allowing for binding to occur and the complexes were rotary shadowed and visualized by EM (Fig. 6D–I). The globular head of P176, corresponding to the V region of the protein (Xu et al., 2002), can be seen binding to the LDL sphere, while the collagenous tail of P176 (depicted by arrows) is seen to be pointed away from the area in which P176 and LDL are interacting. Furthermore, the ‘stud’ structure does not appear to be involved in the binding of rScl1 proteins to LDL. Hence, EM data agree and further support a binding model involving the globular V region of Scl1 protein.

Binding inhibition assay

An antibody (anti-Pep42) raised against a synthetic peptide corresponding to amino acids 42–56 of the P144-V region (Lukomski et al., 2000a) was tested to specifically block P144-LDL binding. In an ELISA, preincubation with anti-Pep42 antibody significantly inhibited the binding of both LDL and ApoB100 to P144 (Fig. 7), as compared with controls, which lacked anti-Pep42 antibody treatment. Binding inhibition by anti-Pep42 antibody further supports LDL binding by Scl1, which involves interactions between the Scl1-V region and the ApoB100 component of LDL. Interestingly, during group A streptococcal infections, both humans and experimental animals seroconvert to Scl (Akesson et al., 2004; Lukomska et al., 2005), and a subset of these anti-Scl antibodies are directed against the V region. Thus, our data suggest that an anti-Scl1 humoral response, elicited by patients with GAS infections, may inhibit interactions between the pathogen and the plasma lipoprotein LDL.

Figure 7.

Inhibition of LDL and ApoB binding by anti-P144 antibody. P144 was immobilized onto Strep-Tactin-coated microplate wells and incubated with increasing concentrations of anti-Pep42 antibody directed against Scl1-V region. LDL or ApoB100 were added to the wells and LDL or ApoB100 binding was detected with specific antibodies as described previously using HRP-conjugated secondary antibodies followed by development of the peroxidase reaction. Absorbance was measured at 415 nm. Differences in LDL and ApoB100 binding were evaluated for statistical significance by Student's t-test (**P < 0.01).

Low-density lipoprotein binds to Scl1 protein expressed on GAS surface

To study LDL binding by GAS cells, isogenic mutant strains were obtained by allelic replacement employing the previously constructed non-polar spectinomycin (spc) cassette (Lukomski et al., 2000b) and a second non-polar erythromycin (erm) resistance cassette constructed here for multiple-gene inactivation. Both spc and erm cassettes were used to generate a series of isogenic mutants of M28-type MGAS 6143 strain, harbouring and simultaneously expressing both scl genes (Lukomski et al., 2001). Single scl1 and scl2 or both scl genes were inactivated (Fig. 8). Polymerase chain reaction (PCR) analysis of the chromosomal DNA verified that the single scl1 mutant of 6143 carried the wild-type copy of the scl2 gene. Similarly, the 6143 scl2 single mutant carried the wild-type copy of the scl1 gene. As expected, the 6143 scl1scl2 double mutant had both genes mutated. Northern blot analysis of the total RNA isolated during exponential growth confirmed the genotypes of the mutants. DNA probes corresponded to the regions of the scl genes that were replaced with the resistance cassettes. While the wild-type strain expressed both scl genes, no transcripts were detected in RNA samples obtained from the double mutant. As expected, single scl1 and scl2 mutants transcribed the non-mutated scl genes. All RNA samples yielded control recA transcripts. Immunoblot analysis with sera specific for either the Scl1.28 or Scl2.28 was used to study the expression of the Scl proteins by the isogenic strains during exponential growth. Both Scl proteins were detected in the sample of the wild-type strain MGAS 6143 and no Scl proteins were detected in a sample obtained from the double mutant. The 6143 scl1 mutant was devoid of Scl1.28 production but still expressed the Scl2.28 protein. Similarly, the 6143 scl2 mutant produced the Scl1.28 but not the Scl2.28 protein.

Figure 8.

Inactivation of scl1 and scl2 genes in M28-type MGAS6143. Single scl1 and scl2 mutants and a double scl1/scl2 mutant were generated by allelic replacement using non-polar promoterless erm and spc cassettes respectively. Isogenic strains were characterized by PCR analysis with scl1- and scl2-specific primers (upper left panel). Lack of scl1 and scl2 transcripts in corresponding mutants are shown by Northern blot analysis (bottom left panel). Isogenic mutants are devoid of the production of the corresponding Scl proteins as determined by Western immunoblotting using specific anti Scl1.28 and anti-Scl2.28 antibodies (right panel).

Three assays were used to study the LDL binding by live GAS cells to test whether the in vitro binding, observed using recombinant rScl constructs, had physiological relevance when naturally expressed Scl1 protein is displayed on the cell surface. MGAS 6143/M28 was selected for the binding experiments to assess specific LDL binding by the Scl1.28 but not Scl2.28 proteins (Fig. 9). In an ELISA-based assay, wild-type and mutant cells were immobilized onto microplate wells, and LDL was added and tested for binding (Fig. 9A, upper panel). Strains possessing Scl1 protein, such as wild-type and scl2-isogenic mutant, were able to bind LDL. In contrast, LDL binding was negligible by the strains devoid of Scl1 protein e.g. scl1-isogenic mutant and scl1/scl2 double mutant.

Figure 9.

Scl1-dependent binding of LDL by M28-type GAS cells. Isogenic GAS strains used corresponded to MGAS 6143 wild-type (WT), scl1-mutant (Scl1 Mut), scl2-mutant (Scl2 Mut), or scl1/scl2-double mutant (Scl1-Scl2 Mut). GAS cultures were grown until they reached logarithmic phase (OD600 = 0.5–0.6). Cells were harvested, washed and resuspended in PBS.
A. Upper panel: LDL binding to GAS cells immobilized in microplate wells. Bound LDL is detected with anti-LDL-specific antibodies followed by HRP-conjugated secondary antibodies and the HRP-substrate reaction was measured spectrophotometrically at 415 nm. Mean absorbance values ± SD from triplicate wells are shown after subtracting the OD values obtained for the control wells with no LDL. Bottom panel: LDL binding to GAS cells in suspension. Streptococcal cells were mixed with 100 ml of human plasma. Following incubation, GAS cells with adsorbed proteins were pelleted, washed, and surface-bound proteins were then dissociated from the cells at low pH and TCA-precipitated. Detection of ApoB100 was carried out by immunoblotting using anti-ApoB100 antibodies and chemiluminescent detection method.
B. Scl1-mediated binding of FITC-labelled LDL by GAS. Wild type and Scl1-deficient mutant cells were incubated with FITC-labelled LDL. LDL binding was viewed under fluorescent microscope at 100 × magnification. Presence of GAS cells in all samples was verified by staining with crystal violet.

Scl1-specific absorption of plasma LDL to GAS cells was also determined in liquid phase (Fig. 9A, bottom panel). Wild-type and mutant cells of MGAS 6143 were incubated with human plasma and unbound ligands were removed by washing and centrifugation. The amount of LDL absorbed from the plasma to the surface of the GAS cells was detected after SDS-PAGE separation followed by Western blot analysis with a polyclonal antibody to LDL. Again, LDL-specific immunoreactivity was detected only in samples containing GAS strains expressing Scl1 protein on the cell surface.

Finally, Scl1-mediated binding of LDL to GAS cells was visualized using fluorescence microscopy. FITC-labelled LDL was used with GAS wild type and scl1 mutant cells, and binding was determined in the liquid phase as above. A positive fluorescence signal was observed only with the wild-type cells, expressing surface-exposed Scl1 protein, but not with mutant cells (Fig. 9B). Both samples contained GAS cells as evidenced by crystal violet staining. Background fluorescence observed in mutant cells was due to the presence of free FITC reacting with GAS cells during the binding step in the assay. Altogether, using three binding assays, we determined that (i) GAS expressing Scl1 protein on cell surface is capable of LDL binding, and (ii) this binding is specific to Scl1 but not Scl2 protein.

Structural and phylogenetic analysis of the Scl1- and Scl2-V regions

Amino acid sequences of the V regions of 21 distinct Scl1 and 14 Scl2 variants were analysed (Fig. 10 and Fig. S2). Multiple sequence alignments (MSAs) were generated separately for Scl1 and Scl2, converted into Hidden Markov Models, and compared with the database of known protein families and with each other using HHsearch (see Experimental procedures). This analysis detected sequence similarity between the V regions of Scl1 and Scl2 (P-value of 0.000025). Scl1 and Scl2 are much more closely related to each other than to any other protein family in the database. The combined MSA of the V regions of both Scl1 and Scl2 families was used to infer a phylogenetic tree, which provides insight into relationships between their sequences. The phylogenetic analysis reveals that the V regions of the Scl1 and Scl2 families form separate branches with no evidence for recombination in the history of the V region. Because a putative ancestral sequence is unknown, we assume that the position of the root is between the Scl1 and Scl2 branches.

Figure 10.

Structural and phylogenetic analyses of the Scl1- and Scl2-V regions. Amino acid sequences of 21 distinct Scl1-V and 14 Scl2-V regions were studied.
A. Schematic representation (not to scale) of the secondary structure prediction in the Scl1.1-V region (complete sequence alignment is shown as Fig. S2). V region is likely to form two amphipatic a-helices (AH1 and AH2 shown in grey fields), whereas the central, hyper-variable region is predicted to form a loop (HVL). Numbers correspond to amino acid residues.
B. The phylogenetic tree of the V region of Scl1 and Scl2. The branch lengths are scaled to the distance computed using the Gonnet250 matrix. The bootstrap values (statistical support for the given bifurcation) are shown for all nodes.

The fold-recognition analysis carried out via the GeneSilico metaserver (Kurowski and Bujnicki, 2003) revealed no significant similarity to any protein of known three-dimensional structure. However, secondary structure predictions suggested that the V region in both families forms two α-helices. Furthermore, the pattern of intervening polar and hydrophobic residues suggests that these helices may have an amphipathic character. Interestingly, the separate analysis of both Scl1 and Scl2 families using the PCOILS server revealed that these helical regions have high tendency to form coiled-coil structures (probability > 80%). All these results suggest that the V region in both Scl families may form similar structures, however, presently it is difficult to predict its detailed conformation, as the coiled-coils may pair with each other in different ways, especially given the trimeric structure of the Scl protein. Nonetheless, the extreme variability of polar amino acids on the predicted surface-exposed side of amphipathic helices, as well in the loop between the helices, suggests that the conserved structural scaffold of the V region will be decorated by completely different side chains. In particular, the N-terminal part of the Scl1-V region is rich in positively charged amino acids (average theoretical pI = 7.90), while in the Scl2 this region is strongly acidic (pI = 4.97). This analysis supports a divergence between the Scl1- and Scl2-V regions observed in phylogenetic tree, and predicts that the V regions of Scl proteins from both families may have very different preferences for interactions with other molecules.

Discussion

Bacteria communicate with the environment of the host through surface molecules, and these interactions can be used for attachment, sensing and host-pathogen cross-talk, as well as for enzymatic activities. Bacterial adhesins which are responsible for specific interactions with host molecules are also referred to by the acronym MSCRAMMs, for ‘microbial surface components recognizing adhesive matrix molecules’ (Patti et al., 1994). MSCRAMMs molecules typically have multidomain organization and frequently recognize more than one ligand; these binding domains are often conserved and shared by adhesins of different pathogens (Fischetti, 2000). In this regard, the streptococcal collagen-like proteins, Scl1 and Scl2, are unique. They do not share ligand-binding domains with known adhesins and do not have sequence similarity in their extracellular portions to previously characterized surface proteins, except for the CL regions of newly identified bacterial CL proteins. Proteins containing these CL regions, potentially capable of forming the collagen triple helix, have been identified in silico in more than 100 prokaryotic proteins (Rasmussen et al., 2003). They were found in a variety of organisms that are pathogenic to humans and animals, including species from such genera as Streptococcus, Bacillus and Clostridium.

Sequence comparisons using highly sensitive bioinformatic tools reveal that the V regions of Scl1 and Scl2 form separate branches but are evolutionarily related, i.e. that they evolved from the ancestral copy of the same Scl gene/protein. Structure prediction methods used in this study suggest that the V regions of both families form similar structures comprising two α-helices, a result which supports an earlier structural prediction for this region (Rasmussen et al., 2000). Furthermore, these two helices may form a coiled-coil structure, although the arrangement of the helices in the Scl trimer remains to be investigated. Nonetheless, the analysis of polar amino acids predicted to be exposed on the surface of the V region reveals considerable differences in composition and, in particular, charge of residues. The V region of Scl1 is positively charged, while the V region of Scl2 exhibits negative charge, suggesting that they became adapted to bind different ligands. These results are in agreement with the experimental data reported in this article.

Neither sequence similarity-based searches nor folding-based approaches allowed the prediction of Scl-V region ligands. Therefore, we applied affinity chromatography using immobilized recombinant rScl proteins to identify potential ligands in human plasma. A striking difference has been observed between binding patterns of the rScl1 and rScl2 constructs. A predominant rScl1-specific ligand was identified as apolipoprotein B (ApoB100), which is a protein component of the major serum lipoprotein, the low-density lipoprotein or LDL, as well its metabolic precursor, VLDL (Segrest et al., 2001). LDL is responsible for the transport and removal of dietary lipid (Yu and Cooper, 2001), but it can also be viewed as a part of our innate immune system that participates in sequestering and removing of bacterial endotoxins, such as LPS (Levels et al., 2005), and some exotoxins that include staphylococcal α- and σ-toxins (Birkbeck and Whitelaw, 1980; Bhakdi et al., 1983), and streptolysins O and S (Ginsburg, 1970; Badin and Denne, 1978). Moreover, interaction with plasma LDL is important for the establishment of some bacterial (Makoveichuk et al., 2003) and viral infections (Hofer et al., 1994).

Several lines of evidence suggest that Scl1–LDL interactions occur between the globular V domain of Scl1 and ApoB protein. First, direct Scl1-ApoB100 binding was verified by two methods and was shown to be concentration dependent. Second, LDL binding was decreased by trypsin proteolysis and by antibodies against the Scl1-V region, and third, by EM analysis. ApoB is one of the largest known proteins, which remains associated with the lipoprotein particle throughout its metabolism. The ‘ribbon and bow’ model (Chatterton et al., 1995) predicts that ApoB100 wraps twice across both hemispheres of the LDL particle, and different functional domains of the ApoB100 were identified. The N-terminal βα1 domain creates a ‘lipid pocket’ for the assembly of the lipoprotein particle (Segrest et al., 1999), and is seen as a knob-shaped structure on LDL viewed under EM. Our rotary-shadowed LDL particles also contained a large ‘stud’ on their surface. Interestingly, the Scl1 globular domain does not interact with this structural element but seem to interact with the other hemisphere of LDL. The C-terminal β2 strand of ApoB100 stretches into that hemisphere and comprises the LDL receptor (LDLR)-binding domain (Jeon and Blacklow, 2005). Present studies are underway to define whether or not Scl1 binds to the same region of ApoB100 as does the LDLR.

Lipoprotein binding seems to participate in GAS pathogenesis. All class II GAS strains express serum opacity factor (SOF) (Bessen et al., 1989). Interestingly, both sof and scl1 transcription is controlled by Mga in a similar fashion, which is different from the control mechanism characteristic for the genes located within the mga-emm-scpA region (Almengor et al., 2006). SOF binds to the apolipoprotein AI component of the high-density lipoprotein (HDL) (Saravani and Martin, 1990) and produce opacification reaction (Ward and Rudd, 1938). Recent studies demonstrated that SOF binds and displaces both ApoAI and ApoAII from HDL, causing the formation of lipid droplets and making the serum opaque (Courtney et al., 2006). Here, we report that recombinant rScl1 constructs, derived from the streptococcal cell-surface protein, Scl1, specifically bind to the LDL through interactions with ApoB100. Importantly, experiments involving live GAS cells demonstrate that Scl1.28 protein expressed on a cell surface mediated LDL binding to M28-type GAS. This binding was absent in mutant cells that lacked surface-exposed Scl1 protein. Absorption of plasma LDL to GAS surface represents a new type of ‘stealth strategy’ used by this pathogen. There are multiple human receptors that are involved in lipoprotein uptake including LDLR and LDLR-related proteins (LRPs) (Jeon and Blacklow, 2005), and several classes of scavenger receptors (Murphy et al., 2005). These receptors are expressed on hepatocytes, mast cells, endothelial and smooth muscle cells, and macrophages (Abraham and Malaviya, 1997; Yu and Cooper, 2001). In addition to lipoprotein metabolism, they are important in host defence that involves innate immunity and inflammatory responses (Suzuki et al., 1997; Shimaoka et al., 2001). Recent studies have shown that residual macrophages play an important role in the early stages of GAS infections and are responsible for both phagocytic killing and proinflammatory stimuli (Goldmann et al., 2004; Thulin et al., 2006). Depending on the location of the Scl1-binding site within ApoB100, LDL-Scl1 interactions could prevent GAS recognition by the macrophages. Alternatively, binding by Scl1 could augment GAS adherence to the macrophages and alter their proinflammatory behaviour. Thus, different human plasma lipoproteins are primary targets for SOF and Scl1, and ongoing studies will show how lipoprotein binding by both streptococcal proteins contributes to the pathogenesis of GAS infections.

Bacterial adhesive structures include pili and curli, as well as non-fimbrial adhesins, which are monomeric or oligomeric proteins anchored to the cell wall (Hultgren et al., 1993). EM after rotary shadowing revealed that Scl proteins are organized into ‘lollipop-like’ structures with stalks made of their collagenous domains and globular heads made of the non-collagenous V regions (Xu et al., 2002). This structural organization of the Scl proteins seems particularly suited for ligand binding, where the CL region projects the globular V region away from the cell surface, facilitating interactions between the V regions and their ligands. Until this study, no V region-specific ligand had been identified. Interestingly, similar structural organization, composed of collagenous and non-collagenous domains, has also been reported for another member of the ‘prokaryotic collagen’ family, bacillus collagen-like protein, BclA, which is a major surface protein found in the exosporium of the Bacillus anthracis spores (Sylvestre et al., 2002). However, the adhesive properties or other biological functions of BclA remain unknown. Furthermore, lollipop-like structures are formed by the cell-surface protein YadA of Yersinia enterocolitica, a member of afimbrial adhesin family (Hoiczyk et al., 2000). Although YadA lacks a CL region, it is also homotrimeric and its globular domain is attached to a coiled-coil rod domain. The globular domain of YadA has a characteristic fold of a parallel β-roll, and YadA-like domains are found in proteins of many bacterial species (Nummelin et al., 2004) that confer cell attachment, binding to extracellular matrix, serum resistance and autoaggregation (Flugel et al., 1994; Hoiczyk et al., 2000; Mintz, 2004; Zhang et al., 2004; Biedzka-Sarek et al., 2005). Earlier studies showed that Scl proteins were involved in GAS adherence to human cells (Lukomski et al., 2000a; Rasmussen and Bjorck, 2001). Hence, Scl proteins may represent an analogous class of afimbrial adhesins that form homotrimeric lollipop-like structures on the cell surface through the CL structural element.

In summary, streptococcal collagen-like proteins, Scl, have a collagenous domain which was thought to play a structural role for the proper assembly, folding and cell-surface exposure of the non-collagenous globular V domain. Lack of V-specific ligands identified until now, hampered our understanding of the structure-function relationship of the Scl proteins. Furthermore, in silico searches identified numerous prokaryotic proteins expressed by the pathogenic microorganisms containing the CL structural elements that are linked to various non-collagenous domains of yet unknown functions. Therefore, identification of the natural ligands specific for those non-collagenous domains is important for our understanding of the evolution of prokaryotic CL proteins. This study is also the first report of LDL binding by the Scl1 protein and adds to the importance of lipoprotein binding by GAS.

Experimental procedures

Bacterial cultures

Bacterial strains and plasmids used in this study are characterized in Table 1. GAS strains were grown in Todd-Hewitt broth (Difco Laboratories) supplemented with 0.2% yeast extract (THY medium). Brain-heart infusion (BHI) agar (Difco Laboratories) and triptic soy agar (TSA) containing 5% sheep blood (Becton-Dickinson) were used as solid media. GAS cultures were incubated in 5% CO2-20% atmospheric O2 at 37°C. For antibiotic selection erythromycin (3 μg ml−1), spectinomycin (150 μg ml−1) and chloramphenicol (10 μg ml−1) were added to the medium.

Escherichia coli strains were grown at 37°C in Luria–Bertani (LB) broth, Miller (Difco Laboratories), and LB agar was used as a solid medium. Erythromycin (300 μg ml−1), spectinomycin (100 μg ml−1) and ampicillin (100 μg ml−1) were used for selection.

Recombinant proteins

Recombinant streptococcal collagen-like proteins (rScl) were produced in E. coli using the Strep-tag II expression and purification system (IBA-GmbH) as described previously (Xu et al., 2002; Han et al., 2006). Briefly, DNA fragments encoding the extracellular portions of different Scl1 and Scl2 variants were PCR-amplified using high-fidelity Vent DNA Polymerase (New England BioLabs) and cloned into the plasmid vector pASK-IBA2, designed for periplasmic expression. E. coli strains harbouring plasmid constructs were grown in the presence of ampicillin and protein expression was induced during the exponential growth with anhydrotetracycline for 3 h. Recombinant proteins were extracted from the periplasm by incubation in a high-sucrose buffer and were purified by affinity chromatography with Strep-Tactin Sepharose. All rScl constructs include the Scl's N-terminal non-collagenous V region and central CL region. In addition, all Scl1-derived rScls also contain the so-called L region. Seven different rScl proteins (Table 1) were obtained which were either Scl1-based (rScl1: P144, P161, P176 and P186), or were Scl2-derived (rScl2: P163, P177 and P178). In addition, chimeric rScl constructs were generated through domain swapping, and expressed and purified as above. All rScls also contain a short affinity strep-tag II (WSHPQFEK) at the C-terminus, which binds to Strep-Tactin-Sepharose.

Generation and characterization of GAS isogenic mutants

Construction of non-polar erm cassettes.  The promoterless erythromycin resistance cassettes, erm1-3, were developed for a non-polar gene inactivation in GAS, and were constructed based on similar spc1-3 cassettes (Lukomski et al., 2000b). Each cassette contains a promoterless erythromycin resistance marker encoded by the erm gene originating from the plasmid pVA838. The 5′ synthetic region flanking the erm1-3 cassettes contains three stop codons, each in a different reading frame. The 3′ synthetic region following the erm-gene stop codon contains a consensus ribosome-binding site (GGAGG) and the ATG start codon placed in each erm cassette in a different reading frame. In order to avoid a polarity effect, the erm cassette is inserted in such a way that the 3′-ATG start codon is in frame with downstream sequence. Thus, one of the erm1-3 cassettes will be suitable at any restriction site available in the target gene.

The erm gene was amplified with primers containing flanking regulatory sequences described above. Three sets of primers were used to amplify three cassettes erm1-3 using Vent DNA Polymerase. One forward primer (erm1.F, GGACCCGGGTGACTAAATAGTGAGAAGGAGTGATTACATGAACAA) and three different reverse primers (erm1.R, CCTCCCGGGCATGTGATTTTCCTCCTTTTTATTTCCTCCCGTTAAATAATAG; erm2.R, CCTCCCGGGCCATGTGATTTTCCTCCTTTTTATTTCCTCCCGTTAAATAATAG; and erm3.R, CCTCCCGGGTCCATGTGATTTTCCTCCTTTTTATTTCCTCCCGTTAAATAATAG) were used. The resulting PCR products, which are flanked by SmaI restriction sites, were cloned into the E. coli vector pCR Script SK+ to produce plasmids pSL164, pSL165 and pSL166, containing the cassettes, erm1, erm2 and erm3 respectively (Table 1). To rule out the presence of any spurious mutations, each erm cassette was confirmed by DNA sequencing. All three erm cassettes confer erythromycin resistance to E. coli in the presence of isopropyl-β-d-thiogalactopyranoside.

Allelic-replacement mutagenesis.  Two suicide plasmids, pSL170 and pSL171, were constructed by a similar strategy in order to inactivate the scl1 and scl2 genes respectively (Table 1). DNA fragments, largely harbouring the 3′- and the 5′-sequences flanking the chromosomal scl1.28 and scl2.28 alleles of MGAS 6143, were sequentially cloned into the multiple cloning sites, MCSI and MCSII, of the E. coli vector pFW11 (Podbielski et al., 1996). A chloramphenicol resistance cat gene was also inserted into these plasmids outside the scl sequences to be used as an additional marker during GAS mutagenesis for differentiation between a single (CmR phenotype) and double (CmS phenotype) cross-over events. The erm2 and spc2 cassettes were then cloned using available restriction sites located within the corresponding 5′- and the 3′-sequences of the scl1.28 and scl2.28, respectively, resulting in plasmid constructs pSL170 and pSL171.

To produce the electrocompetent cells, GAS was grown in THY medium overnight, and cells were harvested by centrifugation. Following two washes with equal volumes of ice-cold PBS, cells were washed once in a 1/10th and then resuspended in a 1/200th of an original volume of the 20% glycerol. Aliquots (50 μl) of the competent cells were stored at −80°C, and these electrocompetent cells, stored for as long as 1–2 years, were successfully used in mutagenesis experiments. Electroporation was performed using an Electro Cell Manipulator 630 (BTX). An electroporation mixture contained 40 μl of H2O, 1–2 μg of purified plasmid DNA, and 50 μl of competent cells. This mixture was placed in a 1 mm gap electroporation cuvette and subjected to 1.8 kV electric shock with a resistance of 400Ω and 25-mF capacitance.

Immediately following electroporation with pSL170 and pSL171, 1 ml of THY medium was added to each electroporation mixture, and the cells were transferred into Eppendorf tubes and allowed to incubate in 5% CO2-20% O2 atmosphere at 37°C for 2.5 h without agitation. Subsequently, 200 μl aliquots were plated on BHI agar, supplemented with either erythromycin (pSL170) or spectinomycin (pSL171). GAS colonies grown on each medium were tested for the chloramphenicol resistance or sensitivity and only the latter colonies were subjected to further analyses by PCR, DNA sequencing, Northern blot and Western immunoblotting.

Northern blot analysis.  Total RNA was prepared from GAS cultures (10 ml) grown in THY medium to mid-log phase (OD600 = ∼0.5), as described previously (Lukomski et al., 2000b). Briefly, cell pellets were suspended in 500 μl of TE buffer [10 mM Tris (pH 7.0), 1 mM EDTA] and treated at 37°C for 5 min with 6 μl of mutanolysin (1 mg ml−1), 60 μl of lysozyme (10 mg ml−1) and 25 μl of the RNase inhibitor aurintricarboxylic acid (Sigma). GAS cells were lysed with 60 μl of 20% sodium dodecyl sulphate and 600 μl of acid-phenol-chloroform (Ambion) at 65°C for 5 min. Following extraction with acid-phenol-chloroform, total RNA was precipitated with ethanol.

For Northern blot analyses, 15-μg samples of total RNA were heat-denatured and separated in a 0.8% formaldehyde gel, blotted onto a nylon membrane, and UV cross-linked. DNA probes specific for the scl1.28, scl2.28 (Lukomski et al., 2001) and control recA genes (Tao et al., 1995) were used in transcript detection. RNA transfer, hybridization (42°C) and post-hybridization washes were performed with NorthernMax reagents (Ambion). Biotinylated molecular weight markers (Ambion) were used to evaluate transcript sizes.

Western blot analysis.  The production of the Scl1.28 and Scl2.28 proteins by the wild-type GAS and isogenic mutants was tested by immunoblotting using rabbit antibodies specific for their V regions (Lukomski et al., 2001). Cell-wall protein fractions were obtained from the exponential-phase cultures after treatment with mutanolysin and lysozyme. Horseradish peroxidase-conjugated goat anti-rabbit IgG (Bio-Rad) was used as a secondary antibody, and detection was performed with chemiluminescence reagent (SuperSignal West Pico; Pierce).

Interaction of rScl proteins with ApoB100/LDL complex

Affinity chromatography-binding assay.  Affinity-chromatography columns were packed with 0.5 ml of the Strep-Tactin-Sepharose resin (50% suspension) (IBA-GmbH) and equilibrated with buffer W (100 mM Tris-HCl pH 8.0, 1 mM EDTA, 150 mM NaCl). One ml of 5 μM purified rScl protein was applied onto the column to allow binding through C-terminal affinity tags. After washing with 8 column volumes (8 CV) of buffer W, a 10 ml sample of human plasma was applied to each column. Plasma obtained from healthy volunteers in accordance with WVU institutional regulations was passaged over rScl-bound or control non-rScl bound columns. In some experiments, 1 ml of 0.1 mg ml−1 human LDL (Sigma) or 0.1 mg ml−1 ApoB100 (Sigma) was applied to the columns instead of plasma. Columns were washed with 8 CV of buffer W. Complexes of rScl proteins and their ligands were eluted in 0.5 ml fractions with 4 CV of buffer E (100 mM Tris-HCl pH 8.0, 1.0 mM EDTA, 150 mM NaCl, 25 mM desthobiotin). Total proteins present in each sample were precipitated with 10% trichloroacetic acid (TCA) for 1 h at room temperature. Following centrifugation (12 000g, 5 min), the pellets were dissolved in 1 M Tris-HCl pH 8.0 buffer and assayed by SDS-PAGE stained with Rapidstain (Geno Technology).

Enzyme-linked immunosorbent assay.  A 100 μl vol of 0.1 μM rScl proteins was immobilized on Strep-Tactin-coated microplate (IBA-GmbH) wells at room temperature for 1 h. Samples containing 100 μl of 0.5 nM LDL or 1 nM ApoB100 were added into the wells and incubated at room temperature for 1.5 h. After three washes with TBS-0.05% Tween 20, 100 μl of goat polyclonal anti-LDL antibody (Sigma) or goat polyclonal anti-ApoB100 antibody (Chemicon) was added to the corresponding wells and incubated for 1.5 h. HRP-conjugated donkey anti-goat (Santa Cruz Biotechnology) was used as secondary antibody and the reaction was developed with ABTS substrate (Bio-Rad). Absorbance was recorded at 415 nm after 15 min of colour development.

An ELISA was also used to assay Scl1 binding to trypsin-treated LDL. Proteolysis using a molar LDL : trypsin ratio of 100 : 1 was carried out for 15 h. The reaction was stopped with phenylmethylsulphonyl fluoride, and partial ApoB100 proteolysis was verified by SDS-PAGE and Western blotting using goat polyclonal anti-ApoB100 antibody. LDL binding was assayed by ELISA, as described above.

Surface plasmon resonance analysis.  SPR was used to determine binding specificity between rScl proteins and LDL. Measurements were carried out at an ambient temperature using the BIAcore 3000 system (Biacore AB). First, polyclonal anti-LDL-specific antibodies and non-specific control IgG antibody were coupled to an activated CM5 sensor chip, resulting in a similar mobilization rate of 5725 and 4792 response units (RU) respectively. Different concentrations of LDL in 25 mM HEPES buffer, pH 8.0, were passed over the immobilized antibodies at 5 μl ml−1 flow rate for 2 min and captured by only anti-LDL antibody. Binding of LDL to anti-LDL antibody reached saturation at an LDL concentration of 0.1 mg ml−1, resulting in an additional RU of ∼300.

Increasing concentrations (1–100 nM) of rScl proteins in 25 mM HEPES buffer, pH 8.0, were passed over the LDL captured by anti-LDL antibody at 20 μl ml−1 for 2 min. Binding of rScl proteins to reference flow cell, which contains 5648 RU of anti-LDL antibody without LDL, was also measured and subtracted from RU of flow cell containing antibody with captured LDL. The BIAevaluation software was used to overlay SPR sensorgrams from different samples and to determine the association rate (kon), dissociation rate (koff) and KD with a 1:1 binding model.

Binding-inhibition ELISA.  Inhibition of P144 binding to LDL and ApoB100 was examined using polyclonal antibody, anti-Pep42. Anti-Pep42 was raised against the synthetic peptide TTMTSSQRESKIKEI, which corresponds to amino acid residues 42–56 (Pep42) of the P144-V region, and was subsequently affinity-purified (Lukomski et al., 2000a). P144 was immobilized to Strep-Tactin-coated microplate wells as described above, and then allowed to react with increasing concentrations (0–100 μg ml−1) of anti-Pep42 antibody at room temperature for 2 h. After washing with TBST, LDL- and ApoB100-binding assays were performed as described above.

Whole cell binding assays.  LDL binding by the M28-type strain MGAS6143 and isogenic mutants, was tested using GAS cells which were either immobilized in microplate wells or were in suspension (Kotarsky et al., 1998). GAS were grown in a THY medium until cultures reached OD600 = 0.5–0.6 (i.e. logarithmic-phase of growth). Streptococcal cells were harvested by centrifugation, washed twice with PBSA (0.15 M NaCl, 0.03 M phosphate, 0.02% sodium azide, pH 7.2), and resuspended in an equal volume of PBSA. GAS cell suspensions were added to microplate wells (MaxiSorp; Nalge Nunc International) and incubated at room temperature for 1.5 h. Wells were washed and blocked overnight with 200 μl of 1% BSA in TBST (Tris-buffered saline with Tween 20) overnight at 4°C. LDL binding was performed as described above in the rScl binding ELISA.

For an LDL binding to GAS cells in suspension, 100 μl of human plasma was added to 1 ml of the cells. Following a 1 h incubation at room temperature, bacterial cells were pelleted and washed three times with PBS containing 0.05% Tween-20 (PBST). After the final centrifugation, proteins bound to the bacteria were dissociated from the cells by incubation with 200 μl of 0.1 M glycine-HCl solution pH 2 for 15 min. Bacterial cells were then removed by centrifugation and the proteins in the supernatant were precipitated with 10% TCA, and analysed by SDS-PAGE and immunoblotting. Electroblotting was carried out with a constant current of 700 mA for 3 h to transfer the high molecular weight proteins, such as ApoB100. Immunodetection of ApoB100 was performed with a goat anti-ApoB antibody followed by HRP-conjugated donkey anti-goat secondary antibody, as described above.

FITC-labelled LDL was prepared by mixing 50 μl of 5 mg ml−1 LDL with 300 μl of PBS and 20 μl of 10 mg ml−1 FITC (in acetone). The mixture was incubated at 37°C for 15 min. Excess FITC was removed using Micro Bio-Spin 6 Chromatography Column (Bio-Rad) and by subsequent dialysis against PBS for 48 h with three PBS changes. Ten microlitres of FITC-labelled LDL were added into 1 ml of GAS-cell suspensions and incubated at room temperature for 30 min. After washing with PBS three times, cells were re-suspended in 1 ml of PBS and 4 μl samples were spread onto glass slides. Observations were taken using Axiostar Plus fluorescent microscope (Zeiss) equipped with AxioCam MRc5 digital camera and analysed by AxioVision Rel. 4.4 software. These same mixtures of FITC-LDL/GAS cells were stained with crystal violet and observed under light microscope.

Mass spectrometry

Following one-dimensional SDS-PAGE, protein bands of interest were excised from the gel and placed into low-affinity binding microcentrifuge tubes. Gel destaining was carried out by extraction with 350 μl of a 1:1 ammonium bicarbonate (ABC):methanol solution for 30 min, 350 μl of 100 mM ABC for 30 min, and 350 μl of acetonitrile (ACN) for 10 min. The gel pieces were dried under a vacuum for 5 min and then rehydrated and digested by the addition of 50 μl of 2 μg ml−1 trypsin in 25 mM ABC and overnight incubation at 37°C. Peptides were extracted by the addition of 100 μl of 25 mM ABC and sonication for 15 min. The peptides were concentrated and desalted using a ZipTip C18 according to the manufacturer's recommendation (Millipore, Bedford, MA). Protein identification was determined by reverse-phase chromatography coupled nano-electrospray ionization ion trap mass spectrometry (RP-MS-MS) using a LCQ-DecaXP plus mass spectrometer (ThermoFinnigan). The peptides were dried in a Speed Vac (ThermoSavant) and reconstituted into 2 μl of 5% ACN, 0.1% formic acid. Peptides were applied to a C18 column with a helium pressure cell and eluted inline using a gradient of 5–50% ACN, 0.1% TFA. Proteins were identified by searching the SWISSPROT database using SEQUEST software (BioWorks 3.0, ThermoElectron). Peptides with cross-correlation (X-corr) values of 2.0 and 2.5 for +2 and +3 charge states, respectively, and delta CN scores of greater than 0.1 were used for protein identification.

Electron microscopy

Preparations of the recombinant P176 and commercial LDL were diluted in 0.1 M ammonium bicarbonate to a final concentration of 1 μM. A mixture of P176-LDL (1:1; v : v) or LDL alone were then dialysed against 0.1 M ammonium bicarbonate at 4°C overnight, and analysed by EM. Samples of LDL were prepared by negative staining and rotary shadowing, while P176-LDL complexes were visualized by rotary shadowing, only. For negative staining, 10 μl of 1 μM LDL was placed on freshly glow-discharged carbon-coated grids. After incubation for 60 s, the sample was removed with filter paper, and 10 μl of 4% phosphotungstic acid, pH 7.4 was immediately added to the grid surface. Following 60-second incubation, the PTA was removed with filter paper, and the grid was allowed to air dry before examination by EM.

For the rotary shadowing, samples were prepared using a ‘sandwich technique’ instead of atomizing through an airbrush. The 0.1 μM protein samples (either P176-LDL or LDL alone) were mixed with glycerol to a final glycerol concentration of 50% (v : v). From this mixture, 4 μl was placed in the crevice formed when cleaving a 6 mm diameter disc of mica. The two halves of the mica disc were then put back together, so that the sample can be spread over both surfaces. After 5 min, the halves were separated and rotary-shadowed with carbon/platinum as described previously (Xu et al., 2002). Photomicrographs were taken using Philips 410 electron microscope.

Sequence analyses and structure prediction

Multiple sequence alignments of Scl1 and Scl2 protein sequences were generated using MUSCLE (Edgar, 2004), followed by manual adjustments based on the secondary structure prediction obtained by a series of methods via the GeneSilico metaserver (Kurowski and Bujnicki, 2003). Comparison of MSAs, represented as Hidden Markov Models, was performed using HHSearch (Soding, 2005). Based on the combined MSA of the V region comprising both Scl1 and Scl2 sequences we inferred the phylogenetic tree using the Neighbour-Joining method implemented in the MEGA 3 package (Kumar et al., 2004). Tertiary structure prediction was attempted using fold-recognition methods via the GeneSilico metaserver (Kurowski and Bujnicki, 2003) (see the website http://genesilico.pl/meta/ for links to all methods used). Prediction of coiled-coil regions was carried out using the PCOILS server (http://toolkit.tuebingen.mpg.de/index.php?view=coils) (Lupas et al., 1991).

Statistical analysis

Determination of statistical significance was performed using the two-paired Student's t-test. Results are expressed as mean of triplicate measurements ± SD from at least two independent experiments.

Acknowledgements

We thank S. F. Tufa and Z. Zhao for technical assistance, and J. Olson and L. Salati for the critical reading of the manuscript. The ongoing support of Fran Rubin is greatly appreciated. This work was supported by National Institutes of Health Grant AI50666 and by a West Virginia University Health Science Center Internal Grant, Office of Research and Graduate Education (to S. Lukomski). J. Bujnicki and M. Pawlowski were supported by the Polish Ministry of Education and Science Grant PBZ-KBN-088/PO4/2003. Mass spectrometry analyses were performed at the WVU Health Sciences Center Proteomics Core Facility supported by a COBRE grant from the National Institutes of Health National Center for Research Resources (5P20RR016440-05).

Ancillary