|Home | About | Journals | Submit | Contact Us | Français|
Wounds are known to serve as portals of entry for group A Streptococcus (GAS). Subsequent tissue colonization is mediated by interactions between GAS surface proteins and host extracellular matrix components. We recently reported that the streptococcal collagen-like protein-1, Scl1, selectively binds the cellular form of fibronectin (cFn) and also contributes to GAS biofilm formation on abiotic surfaces. One structural feature of cFn, which is predominantly expressed in response to tissue injury, is the presence of a spliced variant containing extra domain A (EDA/EIIIA). We now report that GAS biofilm formation is mediated by the Scl1 interaction with EDA-containing cFn. Recombinant Scl1 proteins that bound cFn also bound recombinant EDA within the C-C′ loop region recognized by the α9β1 integrin. The extracellular 2-D matrix derived from human dermal fibroblasts supports GAS adherence and biofilm formation. Altogether, this work identifies and characterizes a novel molecular mechanism by which GAS utilizes Scl1 to specifically target an extracellular matrix component that is predominantly expressed at the site of injury in order to secure host tissue colonization.
Group A Streptococcus (GAS) is responsible each year for more than 730 million infections world-wide ranging from the clinically uncomplicated (pharyngitis, impetigo) to severe and invasive diseases (necrotizing fasciitis, toxic shock), and autoimmune complications (Carapetis et al., 2005; Tart et al., 2007). GAS causes over 100 million skin infections and is categorized as transient or contaminant flora of the skin (Tognetti et al., 2012). These infections are initiated via a portal of entry resulting from simple skin infringements, such as insect bites or cuts and scratches, as well as from more serious postsurgical wounds. For successful colonization within a wounded site, GAS utilizes surface adhesins to secure binding to host extracellular matrix (ECM) components (Chhatwal and Preissner, 2000), including collagen, fibronectin (Fn) and laminin.
The streptococcal collagen-like protein-1 (Scl1) is a ubiquitous surface adhesin, which is co-expressed with a range of known virulence factors that are regulated by the multiple virulence gene regulator of GAS (Mga) (Rasmussen et al., 2000; Lukomski et al., 2000b; Lukomski et al., 2001; Almengor et al., 2006). Scl1 is a homotrimeric protein protruding from the GAS surface that contains four structurally distinct regions. The outermost N-terminal variable (V) region is adjacent to a collagen-like (CL) region that consists of a varying number of GlyXaaYaa (GXY) repeats and adopts stable collagen-like triple helices (Xu et al., 2002; Mohs et al., 2007). At the C-terminus, Scl1 contains a linker (L) region which is a series of conserved, direct repeats adjoining the CL region to the cell wall/membrane (WM)-associated region. Functionally, Scl1 was shown to bind to host-cell integrin receptors (Humtsoe et al., 2005; Caswell et al., 2007; Caswell et al., 2008a) and plasma components (Han et al., 2006a; Gao et al., 2010; Påhlman et al., 2007). We recently reported that Scl1 selectively binds to cellular fibronectin (cFn), but not plasma fibronectin (pFn) (Caswell et al., 2010). Scl1 is also recognized to play a significant role in biofilm formation on abiotic surfaces (Oliver-Kozup et al., 2011).
The wound healing process is dependent on the interaction of cells with the surrounding ECM, which induces the formation of an ECM scaffold to promote cell migration and wound closure. A fundamental component within the ECM contributing to this process is cFn (Martin, 1997). The fibronectins form a large heterogeneous group of glycoproteins with isoforms resulting from alternative splicing events of pre-mRNA of a single gene (Pankov and Yamada, 2002). The soluble pFn is predominantly produced by hepatocytes, whereas cFn can be produced by diverse cell types and may generate up to 20 isoforms in humans. It contains varying amounts of two extra domains termed EDA (EIIIA) and EDB (EIIIB). EDA- and EDB-expressing cFn is highly regulated and plays a crucial role during embryogenesis and early development, and under non-diseased states is maintained at low levels within tissue of adult humans. The expression of cFn containing EDA (EDA/cFn) is upregulated during tissue repair and likely promotes wound cell function by the direct binding of α4β1, α4β7, and α9β1 integrins (Humphries et al., 2006; Liao et al., 2002; Singh et al., 2004; Sakai et al., 2001; Kohan et al., 2010; Ffrench-Constant et al., 1989; To and Midwood, 2011).
In this work, we characterized the Scl1-cFn interaction using recombinant Scl proteins (rScl) derived from epidemiologically distinct GAS strains and investigated the significance of this binding during GAS biofilm formation. We report that Scl1 protein selectively targets the EDA domain of cFn, which is expressed during tissue repair. We mapped the Scl1 binding site within the C-C′ loop region of EDA using peptide inhibition and antibody blocking assays, which overlaps with the EDA-α9β1-integrin-binding site. Furthermore, the Scl1-ECM binding contributes significantly to GAS biofilm formation, which can be inhibited by either the treatment with the synthetic C-C′ loop peptide or with the anti-EDA mAb, IST-9. In conclusion, this work identifies a novel cFn binding mechanism, which is conserved among pathogenically varying strains of GAS, and is mediated by Scl1 binding to the EDA/cFn variant that is expressed at the portal of pathogen entry. The Scl1-EDA/cFn binding may extend the time for GAS tissue colonization by impairing cFn-integrin binding and thereby delaying wound healing.
In a previous study, we established that the Scl1 protein selectively binds the tissue extracellular matrix component, cFn (Fig. 1A) but not pFn (Caswell et al., 2010). Based on this finding, we hypothesized that the Scl1-cFn interaction is mediated via EDA, a domain that is not found in plasma fibronectin, but is thought to be an integral component to the wound healing process. To test this hypothesis, we evaluated by direct ELISA the binding of ECM ligands to immobilized rScl1 proteins P176 (Scl1.41), P144 (Scl1.1), and P161 (Scl1.28) that were previously demonstrated to bind cFn (Caswell et al., 2010). An immobilized rScl2 construct, P163 (Scl2.28), served as a negative control for cFn binding. The ECM ligands (1 μg/well) tested included: (i) purified cFn as a positive control, and confirmed to contain the EDA/cFn variant using the EDA-specific mAb IST-9, (ii) the rEDA, and (iii) the recombinant type III domain four (rIII4), which differs by primary sequence but is structurally homologous to EDA and is common to both cFn and pFn. As shown in Fig. 1B, rScl1 proteins that are positive for cFn binding are also positive for binding to rEDA. In contrast, binding of the rIII4 to rScl1 proteins was comparable to that of the negative control, P163, for all three ECM ligands. When increasing concentrations (0.02 – 4.0 μM) of ECM ligands cFn, rEDA and rIII4 were incubated with immobilized rScl proteins to assess binding specificity, concentration-dependent binding of cFn and rEDA to all three rScl1 constructs was observed, plateauing around the 3–4 μM concentration (Fig. 1C). In contrast, no binding by rIII4 to rScl1 constructs was observed over the entire concentration range nor was binding by any ECM ligand to the negative control, P163, detected (data not shown). We also observed that rScl1 proteins bind better to cFn than rEDA construct. One possible explanation is that the EDA conformation in cFn and rEDA preparations is not identical or that rScl1 binding alters cFn structure (Mao and Schwarzbauer, 2005), resulting in improved binding. The latter mechanism would not be surprising since it has been reported for Fn binding by other bacterial adhesins, including streptococcal (M, F1/SfbI) and staphylococcal (FnbpA) proteins, as well as the BBK32 protein of Borrelia burgdorferi (Marjenberg et al., 2011; Kim et al., 2004; House-Pompeo et al., 1996; Cue et al., 2001). Altogether, our data indicate that the Scl1-cFn interaction is mediated, at least in part, via the EDA domain, and that this interaction is specific and concentration-dependent.
The cFn/EDA spliced variant produced in response to tissue injury (Brown et al., 1993; Serini et al., 1998; Singh et al., 2004; Ffrench-Constant et al., 1989) includes a loop region between the C and C′ β-strands (Fig. 2A), which mediates binding to α4β1 and α9β1 integrins that actively participate in wound healing (Shinde et al., 2008). Since loop regions are often involved in protein-protein interactions, we hypothesized that the C-C′ loop region of EDA contains the Scl1 binding site. To test this, we examined the ability of an exogenous synthetic eleven amino-acid peptide (TYSSPEDGIHE), corresponding to the C-C′ loop region of EDA, to inhibit binding of ECM ligands to immobilized rScl1 proteins, P176, P144, and P161. In this assay, increasing concentrations (0.01 mM, 0.1 mM, and 1.0 mM) of the C-C′ loop peptide or a scrambled peptide control (SEDIHYTEGPS) were pre-incubated with immobilized rScl proteins prior to the addition of the ECM ligands, cFn and rEDA. As shown in Fig. 2B, binding of cFn or rEDA to all three rScl1 proteins was inhibited by the C-C′ loop peptide in a concentration-dependent manner up to 50% and 60% respectively, whereas no inhibition was observed with the scrambled peptide. These results mapped a Scl1 binding site to the EDA C-C′ loop region.
Previous studies reported that residues Asp41 and Gly42 (Fig. 2A) within the C-C′ loop region of EDA are involved in α9β1-integrin binding and that the monoclonal antibody IST-9 binds the C-C′ loop of EDA at adjacent residues Ile43 and His44, and blocks this integrin engagement (Liao et al., 1999; Shinde et al., 2008). We therefore used the anti-EDA mAb IST-9 to determine if Scl1 and α9β1 integrin have overlapping binding sites within the EDA C-C′ loop region (Fig. 2C). Both cFn and rEDA were pre-treated with increasing concentrations (0.1 μg, 1.0 μg, and 10 μg) of IST-9 before mixtures were added to and incubated in the wells with immobilized rScl1 proteins. With increasing concentrations of IST-9 mAb, we show a significant inhibition of binding by both cFn (10–45%) and rEDA (20–54%) to all three rScl1 constructs. These data mapped a Scl1 binding site within the C-C′ loop region of EDA, which is also recognized by the α9β1 integrin.
Previous work showed that Scl1 and Scl2 proteins are related but differ in their CL region length and type of triplet repeats used (Xu et al., 2002; Han et al., 2006b). Likewise, phylogenetic tree of the Scl1- and Scl2-V regions revealed that they form two separate branches but are evolutionary related (Han et al., 2006a). The V-region sequences also differ among Scl1 and Scl2 variants found in various GAS strains and they are typically M-type specific (Rasmussen and Björck, 2001; Whatmore, 2001; Lukomski et al., 2001; Rasmussen et al., 2000; Lukomski et al., 2000b). To date, majority of human ligands bind to Scl1-V but not Scl2-V regions and they display two main binding patterns e.g., some bind factor H (CFH) and factor H-related protein 1 (CFHR1)(Caswell et al., 2008b; Reuter et al., 2010), whereas other bind low-density lipoprotein (Han et al., 2006a) and ECM components, cFn and Lm (Caswell et al., 2010). To test whether cFn binding via EDA was conserved among Scl1 variants originated from GAS strains of various M types, we examined all previously tested rScl constructs (Caswell et al., 2008b) for binding to rEDA. In this assay, we used the rScl1 proteins, including P144, P161, P176, P186, P190, and P217 that were positive for binding to cFn and Lm, as well as the CFH/CFHR1-binding positive rScl1 proteins P179 and P216. In addition, the rScl2 proteins P163, P177, and P178 were used as ligand-binding negative controls. As shown in Fig. 3A, rScl1 binding to cFn corresponded with binding to rEDA, whereas no binding to pFn by any rScl1 or rScl2 proteins was observed. To test whether all rScl1 constructs bind EDA via the same mechanism, we examined binding of cFn and rEDA in the C-C′ peptide inhibition assay described above. The C-C′ loop peptide was able to compete with both cFn and rEDA binding to rScl1 proteins, decreasing binding by 50–70%, with no inhibition detected by the scrambled peptide (Fig. 3B). As expected, no additional binding inhibition was observed by the C-C″ loop peptide for cFn- and rEDA-binding negative rScl1 and rScl2 proteins. Peptide inhibition results were further strengthened by the IST-9 antibody blocking assay, which was also found to inhibit binding of cFn and rEDA 50% and beyond. The data provide evidence that cFn binding via the C-C′ loop region of EDA is conserved among pathogenically varying GAS strains expressing the ECM-binding Scl1 variants.
Scl1 proteins have been found to mediate GAS biofilm formation on an abiotic surface (Oliver-Kozup et al., 2011). To assess the importance of the Scl1-cFn interaction in GAS biofilm formation, cell biomass produced by three epidemiologically diverse M41-, M28-, and M1-type GAS strains was compared spectrophotometrically following crystal violet staining in an early adhesion (1 h) and mature biofilm (24 h) in untreated or cFn-treated wells (Fig. 4A). M41 wild-type (WT) and the available M41scl1-complemented (scl1-C) mutant strains form a more robust biofilm at 24 h (O.D.600 ~0.5 untreated vs. ~0.7 cFn-coated), as compared to M28 WT (~0.45 untreated v. ~0.6 cFn-coated) and M1 WT (~0.23 untreated; ~0.34 cFn-coated) strains. Importantly, in each case, more bacterial biomass was detected in cFn-treated wells than on abiotic surface during both the early adherence stage (1 h), as well as in mature biofilms (24 h). Finally, we show a statistically significant decrease in biofilm forming capacity by all Scl1-negative isogenic mutants as compared to their parent strains at both time points. Confocal laser scanning microscopy (CLSM) of 24 h biofilms confirmed crystal violet staining results (Fig. 4B). The M41 WT biofilm was the most robust with a thickness averaging 26 μm, while a nearly 50% decrease was observed for the M41scl1 mutant (15 μm). A similar pattern is shown for the M28 WT (22 μm) over the scl1 mutant (14 μm). The M1-type WT GAS (15μm) has a decreased capacity for biofilm formation with M1scl1 mutant (7 μm) poorly supporting a biofilm phenotype. When compared to GAS biofilms formed in untreated (abiotic) wells by the same strains (Oliver-Kozup et al., 2011), cFn-mediated biofilms were enhanced up to 40% in average thickness recorded for wild-type strains. Collectively, we can conclude from these studies that Scl1-mediated binding of GAS to cFn significantly enhances biofilm formation.
To develop a system that allows biofilm development on ECM more closely mimicking the ECM in tissue in vivo, normal human dermal fibroblasts (HDFa) were used to deposit extracellular matrix. To generate the fibroblast-derived ECM (fdECM), wells were denuded of HDFa cells and the presence of fdECM was visualized by Ponceau S staining (Fig. S1A). When the composition of fdECM coating was characterized by ELISA using antibodies that detect ECM proteins, fibronectin and collagens were identified as major constituents (Fig. S1B). Importantly, we detected the presence of the EDA/cFn variant within fdECM using the monoclonal antibody, IST-9. Thus, we confirmed that fdECM coating provides a more complex model of tissue matrix compared to single ECM component, which includes the EDA/cFn population recognized by Scl1, a streptococcal adhesin.
GAS strains used in this study express varying numbers of surface adhesins, including fibronectin-binding adhesins. The fdECM matrix contains a number of ECM components assembled into a higher-ordered structure more suitable for assessing whether Scl1-EDA/cFn binding significantly contributes to the adherence and biofilm formation by epidemiologically varying GAS strains. The structural complexity of the fdECM network was demonstrated by field emission election microscopy (FESEM) (Fig. 5A, row (i)). Using FESEM to image GAS adherence on a fdECM coating, wild-type GAS strains of M1- (ii), M28- (iii), and M41-type (iv) were observed to be targeted or preferentially bound to fdECM deposits, as compared to the surrounding abiotic surface at early time points in infection. At higher magnification, GAS chains are observed in close contact with fdECM fibrous structures. When GAS WT and scl1-inactivated mutant strains were compared based on crystal violet staining for bacterial biomass after 1 h (adherence) and 24 h (mature biofilm) on fdECM, or cFn-coated wells, GAS-fdECM binding produced comparable biofilm absorbance values to those of the cFn-treated wells (Fig. 5B). Scl1-negative isogenic mutants had decreased capacity for adherence and mature biofilm formation, as compared to their WT parent strains, on both ECM-treated wells. Our findings demonstrated that the Scl1 protein plays a significant role in GAS biofilm formation on a single cFn coating, as well as on a more complex fdECM, for all strains tested.
Experiments shown in Figs. 2–3 demonstrated that we can significantly inhibit rScl1 binding to both rEDA and cFn using the synthetic C-C′ loop peptide or IST-9 mAb. Here, we adapted both inhibition assays in order to evaluate the importance of the interactions between native Scl1 surface protein and EDA/cFn component of fdECM in total GAS adherence (Fig. 6). Graphs represent spectrophotometric measurements following crystal violet staining (OD600 values) obtained for each WT strain (M41, M28, and M1) on all three ECM coatings (cFn, rEDA, fdECM) in wells without inhibitors and in wells treated with either the C-C′ loop peptide (panel A) or IST-9 mAb (panel B). Below the graphs, we present these OD600-values data as percent (%) adherence inhibition, as compared with their corresponding untreated samples (100% adherence). For comparison, we also present % adherence inhibition, resulting from genetic inactivation, of their corresponding untreated scl1 mutants (OD600 values are shown in Fig. S2). First, the C-C′ loop or scrambled peptide (1 mM each) were added to GAS WT (Fig. 6A) or scl1 mutant strains (Fig. S2A). Next, peptide-GAS mixtures were added to ECM-coated wells and 1 h adherence was analyzed spectrophotometrically following crystal violet staining. The adherence of all WT strains was significantly inhibited on all three coatings, whereas adherence of scl1-inactivated mutants was unaffected by peptide treatment (Fig. S2). Inhibition levels of adherence obtained for peptide-treated strains type M41 and M28 were similar, ranging on assorted ECM coatings between 25 – 27% of their untreated samples. The remaining binding levels (as well as % inhibition values) were similar to the binding levels of untreated Scl1-negative mutant strains, ranging between 24 – 28%, which is depicted in the graph by horizontal lines over-imposed on bars representing untreated samples. Interestingly, a 25–33% C-C′ peptide inhibition range was also measured for the M1 WT strain; however, the binding-inhibition levels obtained for the M1-scl1 mutant on assorted ECM coatings were substantially higher (51–54%). No strains were inhibited by treatment with the scrambled peptide (data not shown).
In a second set of inhibition studies (Fig. 6B, Fig. S2B), ECM-coated wells were pre-treated with 10 μg of mAb IST-9 for 1 h followed by the addition of GAS strains. Again, 1 h adherence was analyzed spectrophotometrically following crystal violet staining. Similar to the peptide inhibition described above, we identified a statistically significant decrease in adherence by M41 and M28 WT strains treated with IST-9 mAb (24–30% range) and, again, the remaining binding levels of these IST-9-treated samples were similar to the binding levels of their untreated Scl1-negative mutants (25–32%). As previously shown, by comparison, the M1 scl1 mutant had substantially decreased adherence on all three ECM coatings (50–60% inhibition), whereas adherence inhibition of the M1 WT strain due to IST-9 treatment was lower (32–35%). In total, inhibition studies demonstrate that GAS adherence to ECM substrates, including complex fdECM network, can be abrogated with treatments that target the integrin-binding C-C′ loop region of the EDA/cFn.
The skin is an organ covering the human body that forms an effective barrier between the internal and external environments and protects against invading microbes (Holbrook and Smith, 2002). Group A Streptococcus (GAS) is, together with Staphylococcus aureus, a predominant pathogen of the skin and soft tissue (Tognetti et al., 2012; Bisno et al., 2005). GAS, as well as other bacteria, requires a portal of entry to initiate the infection. In the wounded site, GAS faces a new environment, which creates a unique opportunity for adhesion to the host’s cellular and extracellular components via surface adhesins designated MSCRAMMs for ‘microbial surface components recognizing adhesive matrix molecules (Patti et al., 1994). Among various MSCRAMMs expressed by GAS strains (Chhatwal and Preissner, 2000), Scl1 was recently shown to selectively bind cFn (Caswell et al., 2010). Here, we characterize Scl1-cFn interactions and investigate the role of this binding in GAS-ECM adherence and biofilm formation on single ECM coatings and on complex ECM structures deposited by human cells.
In the present study, we selected three model GAS strains representing the intraspecies breadth. The M1-type strain represents the global M1T1 clone responsible for pharyngitis and invasive infections (Aziz and Kotb, 2008; Sumby et al., 2005), the M28-type strain has been historically associated with puerperal sepsis (Green et al., 2005a; Green et al., 2005b; O’Loughlin et al., 2007), and the M41-type strain has been predominantly linked to skin infections (Anthony et al., 1967; Dillon and Wannamaker, 1971; Dillon et al., 1974). Importantly, these strains vary in the number of fibronectin (Fn)-binding proteins they express in addition to Scl1 (Caswell et al., 2007). The M28-type strain harbors at least three genes encoding known Fn-binding proteins, including the serum opacity factor (sof) (Oehmcke et al., 2004; Katerov et al., 2000; Courtney et al., 1999), prtF1/sfbI (Hanski and Caparon, 1992) and prtF2/pfbp (Jaffe et al., 1996; Rocha and Fischetti, 1999). M41-type strain has the prtF2/pfbp gene, whereas the M1-type strain has none of them. Based on this knowledge, the contribution of each Scl1.1, Scl1.28, and Scl1.41 variant to the attachment and biofilm formation on single cFn and complex fdECM coatings by these strains was tested.
First, we investigated the selective recognition of and binding to cFn by rScl1 proteins. There is a large body of literature describing redundancy in Fn binding by GAS surface proteins (Henderson et al., 2011). They are often closely related among Gram-positive bacteria or contain related regions that are involved in Fn binding (Patti et al., 1994) and include GAS proteins M, F1/SfbI, SfbII, F2/PFBP, SOF, FBP54, and FbaA and B (Cue et al., 2001; Kreikemeyer et al., 2004; Jaffe et al., 1996; Courtney et al., 1999; Terao et al., 2001; Talay et al., 1994; Kreikemeyer et al., 1995; Terao et al., 2002). Unlike Scl1, these proteins bind pFn and subsequent studies mapped their binding sites within the N-terminal region spanning type I and type II repeats of Fn, which is identical in pFn and cFn (Courtney et al., 2003; Ozeri et al., 1998; Margarit et al., 2009; Ensenberger et al., 2001; Bingham et al., 2008). The cFn isoforms differ from pFn in that they include in varying proportions the extra domains, termed EDA/EIIIA and EDB/EIIIB (Pankov and Yamada, 2002). In the wound, platelets are activated, degranulate and release EDA-containing Fns into the apical aspect of the fibrin matrix in the region that is first encountered by microorganisms (Sakai et al., 2001). Within two days wound macrophages and fibroblasts produce cFns rich in the EDA segment and these cFns persist throughout the wound bed for at least a week following injury (Brown et al., 1993). Using peptide inhibition and antibody blocking experiments, we demonstrated that multiple rScl1 proteins bind cFn via the C-C′ loop region of EDA, indicating a common binding mechanism among pathogenically discrete strains. Thus, we have identified a novel Fn-binding mechanism unique to Scl1, which allows GAS to target the EDA/cFn spliced variant expressed in wounded tissue.
Within hours of injury, neutrophils populate the wound and thereafter macrophages arrive and epidermal keratinocytes commence migration over the provisional matrix (Martin, 1997). These cells express integrins that bind the EDA/cFn, including α9β1 (Taooka et al., 1999). The C-C′ loop region of EDA domain mediates binding to α9β1 andα4β1 integrins (Shinde and Van De Water, unpublished observations) and has been shown to participate in cell adhesion, cell spreading (Shinde et al., 2008) and matrix assembly (Bazigou et al., 2009). Aberrations in, or blocking of, the C-C′ loop region in EDA abrogates the EDA-integrin binding (Liao et al., 2002; Shinde et al., 2008). Our current binding-inhibition studies, as well as preliminary cell-attachment inhibition assays (not shown), demonstrated that rScl1 binds to the C-C′ functional loop region of EDA, which is also the integrin-binding site. This implies that during GAS infection, the Scl1 adhesin could abrogate human cell migration and adhesion, thus, impairing wound healing. Interestingly, a recent study showed that reepithelialization of a murine cutaneous wound was delayed by staphylococcal infections (Schierle et al., 2009). Considering the overall similarities between streptococcal and staphylococcal skin infections, it is tempting to speculate that GAS evolved analogous molecular mechanisms that augment tissue colonization by manipulating the local host’s environment at the site of infection.
GAS biofilm is a relatively new concept (Lembke et al., 2006), which is increasingly being established with supporting data collected from clinical specimens (Akiyama et al., 2003) and through animal model research (Roberts et al., 2010; Connolly et al., 2011). Several GAS surface components have been reported to participate in biofilm formation (Cho and Caparon, 2005; Courtney et al., 2009; Manetti et al., 2007; Maddocks et al., 2011; Kimura et al., 2012) including the Scl1 protein (Oliver-Kozup et al., 2011). In the latter study, we reported that Scl1 plays a substantial role in GAS biofilm formation on an abiotic surface. We also proposed a model in which Scl1-ECM binding enhances biofilm formation by anchoring the growing biofilm structure. Here, we tested that model by assessing the significance of one interacting component, Scl1-cFn binding in the formation of GAS biofilm. Direct comparison of biofilms grown by the M1-, M28-, and M41-type strains showed a significantly increased biofilm biomass on cFn-coated surface compared to an inanimate surface. These differences were further confirmed microscopically since the average thickness of mature biofilms formed by each of these strains was higher on cFn coating by up to 40%. However, these results could also be explained by the non-Scl1-mediated interactions between cFn and other Fn-binding MSCRAMMs expressed by these strains. Therefore, a similar comparison was made between biofilms formed by the wild-type and isogenic scl1-inactivated mutants of each M-type strain on cFn coating. Despite the presence of additional Fn-binding proteins, especially in M28- and M41-type strains, that bind a different region of Fn, all Scl1-devoided mutants produced significantly lower biofilm biomass and lower biofilm thickness by nearly 50%, suggesting an important role for the Scl1-EDA interaction.
The complexity of natural ECM in tissue exceeds the single cFn coating we initially employed. Therefore, we developed a more complex fibroblast-derived ECM coating (fdECM), which was denuded of cells. Following eukaryotic cell removal, tissue culture wells contained a complex 2-D fibrillar network composed of several ECM components, including various collagens and laminin in addition to the EDA/cFn isoforms within fibronectin. A similar approach was used to study the attachment and biofilm formation of Klebsiella pneumoniae on coating derived from ECM deposited by bronchial epithelial cells (Jagnow and Clegg, 2003). This coating produced a rich biofilm phenotype that was more versatile as compared to a collagen coating or an abiotic surface. In our hands, complex fdECM supported wild-type GAS adherence and biofilm formation at the levels comparable with cFn coating. Scanning electron microscopy revealed targeted or preferential GAS binding to fdECM structures over the surrounding abiotic area. The scl1 isogenic mutants showed significantly decreased adherence and biofilm formation on fdECM similarly to that on cFn-treated wells. Furthermore, we were able to reduce the adherence of the wild-type strains to the levels shown by the scl1 mutants with the C-C′ loop peptide and mAb IST-9. The importance of Scl1-EDA/cFn binding in GAS adherence on fdECM was intriguing considering the fact that GAS cells are equipped with more than one Fn-binding protein and, likely, with more than one type of MSCRAMM targeting additional components of fdECM. Since inhibition was not complete, our data also point at the contribution and importance of other adhesins and alternative ECM-binding mechanisms to GAS adherence and host colonization.
In summary, we characterized Scl1-cFn interactions and mapped the Scl1 binding site within the functional C-C′ loop region of the EDA segment overlapping with the α9β1-integrin binding site. This observation opens new directions of investigations since α9β1 is expressed by a number of epithelial and inflammatory cells involved in tissue repair and wound healing (Singh et al., 2004; Singh et al., 2009; Taooka et al., 1999; Palmer et al., 1993). We studied the contribution of Scl1-ECM adhesion in GAS biofilm formation on a single cFn coating and on complex fdECM 2-D matrix. Based on these results, we propose a model of the Scl1-mediated tissue colonization by GAS (Fig. 7). In this mechanistic model, the opportunistic GAS pathogen exploits injured tissue as a portal of entry to establish skin infection. Within injured site, platelets, macrophages, and underlying injured fibroblasts deposit in the wound bed a provisional ECM matrix (proECM), which includes the EDA-enriched cFn to provide scaffolding for cell migration and proliferation via α9β1, α4β1 and α4β7 integrins, and promote structural remodeling and tissue repair (Humphries et al., 2006). Upon entry into the injured site, GAS utilizes the Scl1 protein to initiate binding to the EDA domain of cFn, which is mediated via the C-C′ loop region and secures adherence to ECM, facilitating microcolony formation. Scl1 acts in concert with other GAS adhesins that, depending on strain, may include Fn-binding proteins M, F1/SfbI, SfbII, F2/PFBP, SOF, FBP54, and FbaA and B (Cue et al., 2001; Kreikemeyer et al., 2004; Jaffe et al., 1996; Terao et al., 2001; Courtney et al., 1999; Schmidt et al., 1993; Talay et al., 1994; Kreikemeyer et al., 1995; Terao et al., 2002). The Scl1-EDA binding may also compete with EDA binding by the integrins expressed on migrating keratinocytes, thus, delaying wound reepithelialization and healing. Growing GAS microcolonies form a focused nidus of tissue infection until it is overcome by immune defenses or disperses to new superficial or deep tissue sites. Validation of this updated model is ongoing research.
The M41- (MGAS 6183), M28- (MGAS 6143), and M1-type (MGAS 5005) strains of group A Streptococcus were used. All GAS cultures were grown at 37°C in an atmosphere of 5% CO2–20% O2. For biofilm formation, GAS strains were grown overnight on brain-heart infusion agar (BHI) (BD Biosciences) and used to inoculate Todd-Hewitt broth (Difco) supplemented with 0.2% yeast extract (THY medium). Cultures were incubated until they reached logarithmic phase (O.D. 600 ~0.5) and used to inoculate treated or untreated tissue culture wells.
The isogenic Scl1-negative mutants for all wild-type strains were constructed by allelic replacement as described previously. Briefly, the scl1 gene in M1-type strain was inactivated using the suicide plasmid, pSL134, harboring the scl1.1 allele with flanking regions and nonpolar spectinomycin resistance cassette, spc2, inserted in-frame to replace the scl1.1 coding region (Lukomski et al., 2000a). The scl1.28 and scl1.41 alleles were inactivated using the pSL170 suicide plasmid containing the erythromycin resistant scl1.28::erm2 construct (Han et al., 2006a; Caswell et al., 2007). Wild-type strains were electroporated followed by incubation in THY medium for 2.5 hours. Cultures were then plated onto BHI agar containing either 100μg ml−1 of spectinomycin or 3μg ml−1 of erythromycin. Drug-resistant transformants were initially screened by PCR for amplified single products, indicating double cross-over recombination, and were subsequently confirmed by DNA sequencing. For complementation of the scl1.41-inactivated mutant, the MGAS 6183 scl1-mutant cells were electroporated with the plasmid construct pSL230 encoding the Scl1.41 protein (Caswell et al., 2007). Scl1 expression or lack of it was confirmed by western immunoblotting.
Several scl1 and scl2 alleles encoding the extracellular protein portion were amplified by PCR and cloned into the pASK-IBA2 vector designed for periplasmic protein expression (Xu et al., 2002; Caswell et al., 2007; Han et al., 2006a; Humtsoe et al., 2005; Han et al., 2006b). Recombinant Scl1 and Scl2 (rScl) proteins used in these assays: P144/rScl1.1 (accession no. AF252861), P161/Scl1.28 (accession no. AY459361), P176/rScl1.41 (accession no. AY452037), P179/rScl1.6 (accession no. EU 127997) P186/rScl1.12 accession no. DQ309441), P190/rScl1.2 (accession no. EU127996), P216/rScl1.55 (accession no. EU127999), P217/rScl1.52 (accession no. EU127998), P163/rScl2.28 (accession no. AY069936), P177/rScl2.4 (accession no. DQ309442), P178/rScl2.77 (accession no. DQ309443). All recombinant Scl proteins contain short C-terminal tag, the Strep-tag II (WSHPQFEK), which has high binding affinity to Strep-Tactin-Sepharose for affinity chromatography purification (IBA-GmbH, Goettingen, Germany). E. coli DH5α and BL21 strains were used for cloning and protein expression, respectively.
The his-tagged recombinant EDA (rEDA) and III4 (rIII4) polypeptides were prepared as described (Liao et al., 1999). Briefly, EDA or III4 cDNA was amplified by PCR, cloned into plasmid pGEX-2T, expressed in E. coli, and purified using affinity chromatography nickel columns.
For binding studies, rScl proteins (0.5 μM) were immobilized onto Strep-Tactin-coated microplate wells for 1.5 h at room temperature and blocked with Tris-buffered saline (TBS) supplemented with 1% bovine serum albumin (BSA) overnight at 40C followed by incubation with the extracellular matrix (ECM) proteins cellular fibronectin (cFn), rEDA, and rIII4 or plasma fibronectin (pFn)(Sigma). The no rScl controls were performed in BSA-coated wells for each ligand and each antibody used. Final ODs were normalized by subtracting the BSA controls in each experimental setup.
All ECM ligands prepared at 0.25 μM concentration for binding inhibition studies or 1 μg of ECM ligand per well for protein screening experiments, were added to rScl-immobilized wells in triplicate and incubated at room temperature for 1 h. Bound ligands were detected with specific anti-cFn (Sigma) and anti-collagen I, -collagen II, -collagen IV primary antibodies and goat-anti-rabbit secondary antibodies conjugated to horseradish peroxidase (HRP)(BioRad). The HRP substrate, 2,20-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) was used and colorimetric reaction was recorded at O.D. 415 nm.
For peptide inhibition studies, synthetic 11-amino-acid peptide representing the C-C′ loop region of the EDA domain (TYSSPEDGIHE) or a scrambled loop peptide (SEDIHYTEGPS) were used (LifeTein, Inc.). rScl proteins (0.5 μM) were immobilized onto Strep-Tactin-coated wells as above and were either untreated or were pre-incubated with increasing concentrations (0.01–1.0 mM) of the synthetic C-C′ loop EDA peptide or scrambled peptide for 30 minutes to 1 h at room temperature. cFn or rEDA ligands were next added to the wells and incubated at room temperature for 1 h. Bound ECM ligands were detected with specific primary and HRP-conjugated secondary antibodies as above.
For antibody inhibition studies, the monoclonal antibody IST-9 was used, which recognizes the Ile43 and His44 epitope (Liao et al., 1999) within the C-C′ loop region of EDA. ECM ligands (0.25 μM) were pre-incubated with increasing concentrations (0–10 μg) of the IST-9 mAb for 30 minutes at room temperature. Mixtures were added to rScl1-immobilized wells and bound ECM ligands were detected as above with specific primary and HRP-conjugated secondary antibodies.
Untreated or pre-treated tissue culture polystyrene 24-well plates were used to assess biofilm formation. ECM-coated wells were pre-treated with cellular fibronectin, rEDA or 1% BSA in TBS overnight at 4°C and then blocked for 2 h with 1% BSA in TBS at room temperature. Exponential-phase GAS cultures (0.5 ml) were seeded into untreated or treated wells and incubated at 37°C in an atmosphere of 5% CO2–20% O2. Medium was removed at 1 h or 24 h and wells were washed with PBS followed by the addition of 0.5 ml of 1% (v/v) solution of crystal violet reagent (Becton Dickinson) in PBS, and incubation at room temperature for 30 minutes. Biomass staining was solubilized with 0.5 ml of ethanol and spectrophotometric readings were recorded at O.D. 600 nm for each sample at each time point. Statistical analysis is reported based on three independent experimental repeats (n = 3 ± SD) with triplicate technical replicates in each experiment. Statistical significance was denoted as *P ≤ 0.05 and **P ≤ 0.001.
For inhibition of biofilm formation both synthetic C-C′ loop peptide and IST-9 mAb were used. GAS WT and mutant strains were grown to logarithmic phase and pre-incubated with 1 mM synthetic C-C′ loop or scrambled peptide. 0.5 ml of the mixture was then added to ECM-treated wells in triplicate and incubated at 37°C in an atmosphere of 5% CO2–20% O2. Adherence inhibition at 1 h was analyzed spectrophotometrically following crystal violet staining. For antibody inhibition assay, ECM-treated wells were pre-incubated with mAb IST-9 (10 μg) for 1 h at room temperature. Wells were then washed with PBS to remove unbound antibody. 0.5 ml of logarithmic phase GAS wild-type and mutant strains were added to triplicate wells and incubated at 37 °C in an atmosphere of 5% CO2–20% O2. Bacterial biomass was measured spectrophotometrically after 1 h following crystal violet staining. . Statistical analysis is reported based on three independent experimental repeats (n = 3 ± SD) with triplicate technical replicates in each experiment. Statistical significance was denoted as *P ≤ 0.05 and **P ≤ 0.001.
To visualize GAS cells by CLSM, GAS wild-type and scl1-inactivated mutants were transformed with plasmid pSB027 (Cramer et al., 2003; Caswell et al., 2010; Oliver-Kozup et al., 2011) to express green fluorescent protein (GFP). GAS cells grown to O.D. 600 nm ~0.5 were inoculated onto untreated or cFn-coated 15-mm glass cover slips placed into 24-well tissue culture plate wells and incubated for 24 h. Biofilms were washed gently with PBS and fixed with 3% paraformaldehyde in PBS at room temperature for at least 2 h. Cover slips were mounted to slides using Prolong Gold (Invitrogen). Fluorescent images were acquired using the confocal microscope (LSM 510 Carl Zeiss) equipped with a 63x/1.40 Oil Plan-Apochromatic objective. Z-stack profiles were further deconvoluted stepwise by AutoQuant software at 0.36 μm and were visualized as 3-dimensional images using NIS – Elements software. Additional visualization was performed using the Zeiss trinocular EPI fluorescence microscope equipped with four achromatic objectives. The Zeiss AxioCam Mrc5 camera and Zeiss AxioVision 4.8 software was used for image acquisition.
GAS biofilm samples were prepared and fixed as above. The samples were post-fixed in osmium tetroxide, dehydrated in an ethanol series, and dried using hexamethyldisalizane. Samples were mounted onto aluminum stubs, sputter-coated with gold/palladium, and imaged on a Hitachi S-4800 field emission scanning electron microscope as previously reported (Oliver-Kozup et al., 2011).
Adult human dermal fibroblasts (HDFa) were grown in Medium 106 (Invitrogen) supplemented with low serum growth supplement to confluency at 37°C in an atmosphere of 5% CO2. For fdECM production, HDFa were grownon sterile 15-mm glass cover slips inserted into wells. Cells were removed by treatment with 2 mM ethylene glycol tetraacetic acid, leaving naturally deposited network of extracellular matrix. Samples were washed gently with PBS and fixed with 3% paraformaldehyde for at least 2 h or overnight, and analyzed microscopically following Ponceau S staining or by FESEM. ELISA was performed to determine the presence of extracellular matrix components: total cellular fibronectin, EDA-containing cFn, collagens type I, II, and IV, and laminin. Components present were detected with specific primary antibodies: anti-Fn (Sigma),anti-EDA/cFN (IST-9: Santa Cruz Biotechnology) anti-collagens I, II, IV (Chemicon/Millipore), and anti-laminin (Sigma). Appropriate secondary antibodies conjugated to horseradish peroxidase (BioRad) were used and developed with ABTS substrate.
Statistical analysis was performed using the two-tailed paired Student’s t-test and significance was denoted at a level of *P ≤ 0.05 or **P ≤ 0.001. Error bars represent standard deviations with analyses based on three independent experimental repeats (n=3), each performed in triplicate technical replicates.
We thank Joan Olson and Fred Minnear for the critical reading of the manuscript. The support of Xin-Xing Gu is greatly appreciated. This work was supported in part by National Institutes of Health Grant AI50666 (to S. Lukomski) and GM056442 (to L. Van De Water). Authors also wish to acknowledge the WVU Research Office’s Program to Stimulate Competitive Research (PSCoR), as well as the Research Funding Development Grant (RFDG) (to S. Lukomski). H. Oliver-Kozup was supported by a grant from the West Virginia Graduate Student Fellowship in Science, Technology, Engineering and Mathematics (STEM). Confocal microscopy experiments were performed in the West Virginia University Microscope Imaging Facility, which has been supported by the Mary Babb Randolph Cancer Center and NIH grants P20 RR016440, P30 RR032138/GM103488 and P20 RR016477. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the National Institute of Occupational Safety and Health.