Surface Proteins of Staphylococcus aureus

ABSTRACT

The surface of Staphylococcus aureus is decorated with over 20 proteins that are covalently anchored to peptidoglycan by the action of sortase A. These cell wall-anchored (CWA) proteins can be classified into several structural and functional groups. The largest is the MSCRAMM family, which is characterized by tandemly repeated IgG-like folded domains that bind peptide ligands by the dock lock latch mechanism or the collagen triple helix by the collagen hug. Several CWA proteins comprise modules that have different functions, and some individual domains can bind different ligands, sometimes by different mechanisms. For example, the N-terminus of the fibronectin binding proteins comprises an MSCRAMM domain which binds several ligands, while the C-terminus is composed of tandem fibronectin binding repeats. Surface proteins promote adhesion to host cells and tissue, including components of the extracellular matrix, contribute to biofilm formation by stimulating attachment to the host or indwelling medical devices followed by cell-cell accumulation via homophilic interactions between proteins on neighboring cells, help bacteria evade host innate immune responses, participate in iron acquisition from host hemoglobin, and trigger invasion of bacteria into cells that are not normally phagocytic. The study of genetically manipulated strains using animal infection models has shown that many CWA proteins contribute to pathogenesis. Fragments of CWA proteins have the potential to be used in multicomponent vaccines to prevent S. aureus infections.

INTRODUCTION

Cell wall-anchored (CWA) proteins are characterized by the presence of a sorting signal at the C-terminus which is responsible for coupling the protein covalently to peptidoglycan. The surface of Staphylococcus aureus is decorated with up to 24 CWA proteins. The precise number depends on the strain and the growth conditions. The repertoire of CWA proteins expressed by S. aureus is limited and many have evolved to perform important interactions with the host. They can be categorized into distinct structural and functional groups (Fig. 1, Table 1). Several are multifunctional due to the proteins comprising distinct domains which recognize different ligands, while for others a single domain is capable of binding different ligands by different mechanisms.

FIGURE 1.

CWA surface proteins classified based on structural motifs. The primary translation products of all CWA proteins contain a signal sequence (S) at the amino terminus and a wall-spanning region (W, Xc) and sorting signal (SS) at the carboxyl terminus. The CWA proteins that are depicted are those for which structural analysis has facilitated classification into five distinct groups. (A) Microbial surface components recognizing adhesive matrix molecules (MSCRAMMs). The clumping factor (Clf)-serine aspartate repeat (Sdr) group comprises proteins that are closely related to ClfA. ClfA and ClfB have a similar domain organization, whereas SdrC, SdrD, and SdrE contain additional B_SDR repeats that are located between the A domain and the serine-aspartate SD repeat R region. The N-terminal A region contains three separately folded domains, called N1, N2, and N3. Structurally, N2 and N3 form IgG-like folds that bind ligands by the DLL mechanism. Fibronectin-binding protein A (FnBPA) and FnBPB have A domains that are structurally and functionally similar to the A domain of the Clf-Sdr group. Located in place of the serine-aspartate repeat region are tandemly repeated fibronectin-binding domains (11 in FnBPA, 10 in FnBPB). The A region of the collagen adhesin (Cna) protein is organized differently than other MSCRAMMs, with N1 and N2 comprising IgG-like folds that bind to ligands using the collagen hug mechanism. The A region is linked to the wall-spanning and anchorage domains by variable numbers of B_CNA repeats. (B) Near iron transporter (NEAT) motif protein family. The iron-regulated surface determinant (Isd) proteins have one (for IsdA), two (for IsdB), or three (for IsdH) NEAT motifs that bind to hemoglobin or heme. The figure depicts IsdA, which has a C-terminal hydrophilic stretch that reduces cell surface hydrophobicity and contributes to resistance to bactericidal lipids and antimicrobial peptides. (C) Three-helical bundle motif protein A. The five N-terminal tandemly linked triple-helical bundle domains (known as EABCD) that bind to IgG and other ligands are followed by the repeat-containing Xr region and the nonrepetitive Xc region. (D) G5-E repeat family. The alternating repeats of the G5 and E domains of S. aureus surface protein G (SasG) (and the accumulation-associated protein [Aap] from S. epidermidis) link the N-terminal A region to the wall-spanning and anchorage domains. If the A domain is removed, the G5-E region can promote cell aggregation. (E) Legume-lectin, cadherin-like domain protein. The BR region of the serine rich adhesin of platelets (SraP) protein is flanked by serine-rich repeat domains. The BR region comprises three distinct structural domains: the legume lectin-like, the β-grasp fold (β-GF), and the cadherin-like (CHLD) domains.

TABLE 1.

Properties of CSA surface proteins^a

Protein structural and functional group	Ligand and binding mechanism	Function	References
MSCRAMM family
Clumping factor A (ClfA)	Fibrinogen γ chain C-terminus (DLL)	Adhesion to immobilized fibrinogen, immune evasion by binding soluble fibrinogen	3, 11
	Complement factor IMechanism unknown	Immune evasion, degradation of C3b	130, 131
Clumping factor B (ClfB)	Fibrinogen α-chain repeat 5, keratin 10 and loricrin (DLL)	Adhesion to desquamated cells from nares and from skin of AD patients; nasal colonization and colonization of AD skin	5, 7, 14, 118, 150
Serine aspartate repeat protein C (SdrC)	β-neurexin (DLL)	Unknown	23, 24
	Unknown	Adhesion to desquamated epithelial cellsNasal colonization?	93
	SdrC	Homophilic SdrC-SdrC interaction; biofilm formation	21, 23
Serine aspartate repeat protein D (SdrD)	Desmoglein-1Mechanism unknown	Adhesion to desquamated epithelial cellsNasal colonization?	93, 120
Serine aspartate repeat protein E (SdrE)	Complement factor H	Immune evasion, degradation of C3b	132
	Close DLL		13
Bone sialoprotein binding protein (isoform of SdrE)	Fibrinogen α-chain (DLL)	Adhesion to immobilized fibrinogen	151
Fibronectin binding protein A and B	FnBPA A domain binds fibrinogen γ-chain C-terminus and elastin (FnBPA A domain, DLL)	Adhesion to ECM	6, 16, 152
A domains
	FnBPB A domain also binds fibronectin but not by DLL		152
	FnBPA, FnBPB	FnBPA-FnBPA A domain homophilic interaction; biofilm formation	20, 25
Fibronectin binding repeats	Fibronectin, tandem β-zipper	Adhesion to ECMInvasion of mammalian cells	73–75, 86, 153
Collagen binding protein (Cna)	Collagen triple helix (CH)	Adhesion to collagen-rich tissue	17
	Complement protein C1q (CH)	Prevent classical pathway of complement activation	19
	Laminin	Adhesion to basement membrane	154
NEAT motif family
Iron-regulated surface protein A (IsdA)	Heme, fibrinogen, fibronectin, cytokeratin 10, loricrin (N-terminal NEAT motif region)	Heme uptake and iron acquisition	64
		Adhesion to desquamated epithelial cells	65, 119, 155
	Unknown ligand(s); C-terminal domain NEAT motif region	Resistance to lactoferrin	156
		Resistance to bactericidal lipids and antimicrobial peptidesSurvival in neutrophils	66
IsdB	Hemoglobin, heme (N-terminal NEAT motif region)	Heme uptake and iron acquisition	64, 157
	β3 integrins (NEAT motif regions)	Invasion of nonphagocytic cells	68
IsdH	Heme, hemoglobin (N-terminal and/or C-terminal NEAT motif region)	Heme uptake and iron acquisition	69
	Unknown ligand(s); N-terminal NEAT motif region)	Accelerated degradation of C3b	69
Three-helical bundle
Protein A	IgG Fc	Inhibits opsonphagocytosis	53
	IgM Fab VH3 subclass	B cell superantigen	55
	TNFR-1	Inflammation	54
	von Willebrand factor	Endovascular infection; endocarditis	56
	Unknown ligand (region Xr)	Inflammation	58
G5-E domain family
S. aureus surface protein G (SasG) and plasmin-sensitive surface protein (Pls, SasG homolog in MRSA)	A domain; no known ligand	A domain; adhesion to desquamated epithelial cells; primary attachment to unconditioned biomaterial	43, 45, 93
	G5E repeats of SasG; homophilic Zn2+-dependent homophilic interaction	Biofilm formation	48, 49, 51, 52
Legume lectin-like, cadherin-like
SraP N-terminal legume-lectin domain	N-acetyl neuraminic acid (Neu5AC)Salivary agglutinin gp340	Adhesion to and invasion of mammalian cells	71, 72
SraP CHDL	SraP CHDL	Homophilic CHDL-CHDL interaction; biofilm formationProjection of l-lectin domain	71
5′-nucleotidase
Adenosine synthase A	Converts ATP to adenosine	Survival in neutrophils; inhibits oxidative burst	87, 88
	2′-deoxyadenosine 3′-monophosphate; degradation product of DNA converted to 2′-deoxyadenosine	Macrophage and monocye apoptosis	90
EF-hand motif
Biofilm-associated proteinBovine/ovine mastitis isolates only	BiofilmCa2+ dependent	Promotes persistence in ovine mastitis	158
	Binds GP96	Promotes adhesion to epithelial cells, reduces invasion	159
Other proteinsNo known functional motifs
S. aureus surface protein X (SasX)	Ligands unknown	Biofilm, cell aggregation, squamous cell adhesion, neutrophil evasion	115
SasCN-terminal FIVAR-containing domain	Ligands unknown	Promotes primary attachment and accumulation phases of biofilm formation	160
SasB, SasD, SasF, SasJ, SasK, SasL	Unknown	SasF implicated in resistance to bactericidal effects of long chain unsaturated fatty acids	161
		Putative LPXTG proteins identified from genome sequences; no known structure or function	162

^a AD, atopic dermatitis; CH, collagen hug; CHDL, cadherin-like domain; C1q and C3a, complement components; DLL, dock lock latch; ECM, extracellular matrix; FIVAR, found in various architectures; NEAT, near iron transporter; MRSA, methicillin-resistant S. aureus; S. aureus surface protein, Sas; Pls, plasmin-sensitive; SraP, serine-rich repeat protein.

The multifunctionality of CWA proteins has resulted in some confusion over nomenclature, with the original names often referring to the first-assigned function, while in other cases proteins were named S. aureus surface protein (Sas) prior to a function being ascribed and were then assigned a new name when a function or structural feature was discovered; e.g., SasH is now called AdsA (extracellular adenosine synthase), and SasA is now SraP (serine-rich surface protein).

This review will describe our current understanding of the structure and functions of CWA proteins of S. aureus in adhesion to the extracellular matrix (ECM), invasion of host cells, evasion of immune responses, biofilm formation, and acquisition of iron. The review will include a discussion of posttranslational modifications that contribute to function and will describe the contribution of CWA proteins to colonization of the host and to pathogenesis.

CWA PROTEINS: STRUCTURE AND FUNCTION

Several distinct structural and/or functional classes of CWA protein occur in S. aureus (Fig. 1, Table 1). N-terminal domains are projected away from the cell surface and are the part of the protein that is usually engaged in interactions with the host. An important exception is the fibronectin binding repeats of fibronectin binding proteins (FnBPs), which are located toward the C-terminus.

The MSCRAMM Family

“Microbial surface components recognizing adhesive matrix molecules” (MSCRAMMs) is a term that was originally applied to CWA proteins that bind to host proteins from the ECM, such as fibrinogen and fibronectin (1). It has become widely used to encompass all CWA proteins whether or not they bind ECM molecules. It was recently proposed that the term “MSCRAMM” should be reserved for proteins with a common structure motif, viz., two adjacent IgG-like folds and similar mechanisms of ligand binding exemplified by dock lock latch (DLL) and the collagen hug (2).

Ligand binding by DLL: the basic mechanism

Proteins in the Clf-Sdr subfamily of MSCRAMMs have an N-terminal A domain comprising three subdomains called N1, N2, and N3, with N2 and N3 being the minimum required for ligand binding by DLL. The N2 and N3 subdomains each comprise two β-sheets that are arranged in a variation of the IgG fold called DEv-IgG (3). The DLL mechanism was originally defined by analyzing the structure of the Staphylococcus epidermidis protein SdrG in the apo form and in complex with the β-chain peptide of fibrinogen (4). The same mechanism was found to apply to clumping factor A (ClfA) and the A domains of FnBPA and FnBPB binding to the γ-chain peptide of fibrinogen and to ClfB binding to cytokeratin 10, loricrin, and the α-chain of fibrinogen (5–7).

In essence, an unfolded peptide ligand binds to the open form of the MSCRAMM in a hydrophobic trench located between the N2 and N3 subdomains (Fig. 2). This stimulates a conformational change which results in the unfolded extension of N3 covering the peptide to lock it in place and creating the latch by forming an extra β-strand in one of the β-sheets of N2 by β-strand complementation (2, 4). The peptide ligand binds β-strand G′ in N3 through backbone interactions while side chains of residues in the peptide and residues in the binding trench of the MSCRAMM determine binding specificity.

FIGURE 2.

MSCRAMM binding to ligand by DLL. Ribbon diagram showing the structure of the N2 (green) and N3 (yellow) subdomains of SdrG in the apo form and in complex with a peptide from the β-chain (purple) of fibrinogen following ligand binding by the DLL mechanism. The C-terminal extension of N3 in the apo form undergoes a conformational change following ligand binding, resulting in an additional β-strand in a β-sheet in subdomain N2 forming the latch (red) and lock (blue). The letters refer to β-strands.

Recent studies using atomic force microscopy showed that the rupture force of SdrG in complex with fibrinogen is very high, being equivalent to that of a covalent bond (8). Similar forces were required to separate cells expressing Cna binding to collagen (9) and ClfB expressing cells binding to loricrin (10). In the latter the strength of the interaction between ClfB and loricrin increased when mechanical forces were applied, suggesting that shear forces might induce the conformational changes involved in DLL. When the MSCRAMM is attached to the surface of a staphylococcal cell, these changes must result in considerable reorientation of the protein (Fig. 3) with implications for binding to other ligands (see section FnBPs bind plasminogen).

FIGURE 3.

Schematic diagrams of MSCRAMMs before and after ligand binding. The top figure depicts an MSCRAMM in the apo form with the N2 (green) and N3 (yellow) subdomains shown as semicircles and the unstructured N1 subdomain. Serine residues in the flexible stalk are glycosylated, which prevents degradation by cathepsin. The middle diagram shows the conformational change in an MSCRAMM with respect to the bacterial cell that occurs following ligand binding by DLL. The C-terminal γ-chain peptide of fibrinogen is depicted by the red dashed line, and the gamma globule domain is in contact with the second ligand binding site in ClfA subdomain N3. It is not known if other MSCRAMMs have two binding sites on their ligands. The bottom diagram indicates that binding to fibrinogen by FnBPs exposes the N3 subdomain to plasminogen, which binds the MSCRAMM more efficiently in the presence of fibrinogen.

Ligand binding by DLL: variations on a theme

Several subtle variations on the DLL theme have been observed. Both parallel and antiparallel orientations of ligand peptide binding to the trench occur (SdrG parallel; ClfA and ClfB antiparallel), which results in the ligands binding in opposite orientations with respect to the MSCRAMM (3, 4, 7).

The ligand residues bound by SdrG and ClfB have extensive flanking residues located both C-terminally and N-terminally and thus can only bind to the open apo form of the MSCRAMM. In contrast, the ClfA ligand binding residues occur at the extreme C-terminus of the γ-chain of fibrinogen and are able to penetrate the MSCRAMM in the locked position (11).

ClfA has a second binding site on fibrinogen in addition to the γ-chain peptide (12). The epitope for the monoclonal antibody tefibazumab that blocks binding of ClfA to fibrinogen was mapped to a location on the “top” of subdomain N3 well away from the ligand binding trench involved in DLL. Molecular modeling predicted that tefibazumab partially blocks a second fibrinogen binding site involving the fibrinogen γ-globule and that this is required for high-affinity binding to the ligand (Fig. 3). Indeed, tefibazumab has a high affinity for ClfA but has only a half maximal inhibitory concentration for blocking ClfA binding to fibrinogen. Studies with ClfA variants showed that high-affinity binding requires the two sites to act cooperatively in a complex multicontact binding mechanism. It is possible that other MSCRAMMs bind ligands by similar multistep mechanisms, particularly where the affinity of a peptide is lower than the full-length protein. Ultimately, proof of two binding sites requires x-ray crystal structures of the proteins in complex.

SdrE binds to the host recognition C-terminal unit 20 of the complement control protein factor H by a variation of DLL called “closed dock lock latch” (13). In SdrE the ligand binding trench is occluded by loop A-B that protrudes from N2 forming four main chain H bonds with β-strand G′ in N3. For the factor H unit 20 to bind by DLL, it must first displace the AB loop, which is flipped out of the way (Fig. 4).

FIGURE 4.

SdrE binding to complement factor H. The top figure shows SdrE in the apo form with the unstructured N1 subdomain and the N2 and N3 subdomains (yellow and green semicircles, respectively). The loop that occludes the ligand binding trench is shown in blue. The bottom figure shows the conformational changes that occur when complement factor H binds by closed DLL. Factor H (red) can then engage nearby C3b molecules (blue) and facilitate binding and activation of the protease factor I, which cleaves C3b.

ClfB binds several ligands by DLL

ClfB binds to cytokeratin 10, loricrin, and the α-chain of fibrinogen by DLL (5, 7, 14). Thus, several different ligands can be accommodated by the same binding trench. This has allowed the definition of a consensus ligand binding sequence (7). Another example is the A region of FnBPA and FnBPB, which can bind to elastin as well as the γ-chain peptide of fibrinogen like ClfA (15, 16). However, this conclusion resulted from the study of FnBP variants with substitutions in residues predicted to be involved in DLL and not from x-ray crystal structures. It is possible that other MSCRAMMs bind several different ligands by DLL.

Collagen binding

The collagen binding protein Cna binds collagen by a variation of the DLL mechanism called the collagen hug (17). The N1 and N2 subdomains of the A region of Cna are composed of IgG-like folds and perform roles corresponding to the N2 and N3 subdomains of other MSCRAMMs, with the C-terminal extension of N2 forming the latch (Fig. 5).

FIGURE 5.

Collagen binding protein and the collagen hug. The upper part shows a schematic diagram of the Cna protein in the apo form on the left and following binding by the collagen hug to the collagen triple helix on the right. Below is a ribbon diagram of the Cna protein in complex with the collagen triple helix. The N1 subdomain (green) and N2 subdomain (yellow) are separated by a long unfolded region (blue) that forms the lock around collagen (purple). β-strand complementation by the red strand completes the hug.

The linker separating N1 and N2 is long and is able to accommodate the thick rod of the collagen triple helix. The ligand binds to a shallow trench in N2. This stimulates a conformational change, which enables the linker to wrap around collagen. Latching then takes place by β-strand complementation between the C-terminal extension of N2 and a β sheet in subdomain N1 (17).

Cna also binds to complement factor C1q

The complement pathway is initiated when complement protein C1q binds to antibodies attached to the surface of a pathogen. C1q forms a bouquet-like structure that is composed of six identical heterotrimers, each comprising a globular head that forms the antibody recognition domain and a long collagenous triple helix (Fig. 6) (18). These coalesce to form a stalk. The C1r and C1s proteins are associated with the C1q complex and are necessary for initiating the complement cascade. Cna can bind to the C1q triple helices and displace C1s-C1r, resulting in inhibition of complement fixation (19). The collagen hug mechanism is most likely to be involved because Cna variants with reduced collagen binding also bind C1q with lower affinity. Thus, Cna and other members of the Cna family can act both as adhesins to collagenous tissue and as immune evasion factors by inhibiting complement fixation.

FIGURE 6.

Complement protein C1q. C1q is a complex of six identical heterotrimers that form a bouquet-like structure. The globular domains (blue, green, and cyan ovals) make up the six IgG binding sites. Each heterotrimer forms an extended collagen-like triple-helix stalk which coalesces into a complex stem. C1r and C1s bind to the triple-helix region and are displaced when Cna binds. The figure was kindly provided by Nicole Thielens, CNRA-CFE-Université Joseph Fournier, Grenoble, France.

SdrC and FnBPs engage in homophilic interactions and promote biofilm formation

The MSCRAMMs SdrC, FnBPA, and FnBPB can promote cell-cell attachment and biofilm formation (20–22). In both cases the A domains on adjacent cells undergo homophilic interactions (Fig. 7). SdrC subdomain N2 has two interaction sites that were identified by screening phage display libraries for inhibitory peptides (21). Molecular modeling showed that a β-neurexin peptide which binds SdrC occludes one of the interaction sites and can prevent biofilm formation (23, 24). Studies with bacterial cells expressing FnBPs and variants thereof confirmed that the A domains are required for biofilm formation and that in the case of FnBPA, the ability to engage in DLL is not involved (20).

FIGURE 7.

Homophilic interactions and biofilm formation. The upper part shows a schematic diagram of the model for homophilic interactions between the A domains of the MSCRAMMs FnBPA, FnBPB, and SdrC, which promote cell-cell accumulation of staphylococcal cells (yellow spheres) during biofilm formation. See Fig. 2 and 3 for the key. The lower part shows the extended fibrillar region of SdrG and Aap (orange and blue strands), which form extended zinc-dependent zipper interactions predicted to form a twisted rope-like structure.

Atomic force microscopy has been employed to study forces involved in interactions between MSCRAMMs either with the atomic force microscopy cantilever tip conditioned with the purified A domain protein or with a single bacterial cell. It was concluded that cell-cell binding results from multiple weak interactions between proteins occurring in parallel between molecules on two individual cells (23, 25). The rupture force required to separate a single CWA protein-protein interaction was very low compared to that required to separate an MSCRAMM from a ligand bound by DLL or the collagen hug. Weak interactions might facilitate detachment of bacterial cells from the mature biofilm and their dissemination.

In the case of SdrC, the A domain can also promote attachment to hydrophobic surfaces and could thus be involved both in attachment to an indwelling medical device and in biofilm accumulation (23).

FnBPs bind plasminogen

Several CWA proteins contribute to the ability of S. aureus cells to capture plasminogen from serum (26). Mutants lacking FnBPs are only partially defective in plasminogen binding, whereas a sortase A defective mutant lacking all CWA proteins binds very little. Thus, the full repertoire of CWA proteins that bind plasminogen is not known, with the two studies published so far focusing only on FnBPs (26, 27). The A domain of all seven isoforms of FnBPB bind plasminogen with similar affinity (26). Binding of FnBPB was inhibited by ε-amino caproic acid and lysine, which implies that binding occurs between lysine residues on the MSCRAMM and negatively charged residues on plasminogen. Indeed, alanine substitutions of the two conserved patches of lysines in FnBPB N2 resulted in reduced plasminogen binding. The ability of whole cells expressing FnBPs to bind to plasminogen is greatly enhanced by fibrinogen (27). This implies that the conformational changes that occur when fibrinogen binds FnBPA by DLL reorients the cell-bound MSCRAMM and exposes the plasminogen binding domain (Fig. 3).

It is likely that MSCRAMMs and, indeed, other CWA proteins are also able to bind several ligands. Indeed, the A domain of FnBPB can also bind to fibronectin by a mechanism that does not involve DLL (15).

Functions of other regions of MSCRAMMs

The SdrC, SdrD, and SdrE proteins have two or more repeated domains of 110 to 113 residues (B_SDR) located between the A region and the SD region (Fig. 1) (28). The B_SDR repeats are folded separately and form rigid rod-like structures that depend on Ca²⁺ for their structural integrity. They have no known ligand binding activity (29), although the related B_SDR region of SdrF from S. epidermidis can bind weakly to collagen (30).

It has been suggested that the B1 domain of SdrD makes contact with the adjacent N3 subdomain of region A, resulting in the ligand binding groove between N2 and N3 opening further than seen in the structure of the A domain alone, which possibly influences binding activity (31).

Cna proteins have 2 to 4 repeats of the B_Cna domain (Fig. 1) that are not related to the B repeats of Sdr proteins. Atomic force microscopy studies of bacteria expressing Cna proteins on their surface indicated that the B repeats act as nanosprings that project the A domain from the cell surface to facilitate the conformational changes required for the A domain to engage in ligand binding by the collagen hug (9). The B_SDR repeats might have a similar function.

Flexible stalks

MSCRAMMs from the Clf-Sdr family are attached to the cell wall via long flexible unfolded peptides comprising either serine-aspartate dipeptide repeats (ClfA, ClfB, SdrC, SdrD, SdrE) or 10/11 tandem repeats of ∼38 residues that bind fibronectin by the tandem β-zipper mechanism (see below). The SD repeats of ClfA act like a stalk that projects the N-terminal ligand binding A domain away from the cell surface (32). Bacteria expressing ClfA truncates lacking SD repeats bound fibrinogen less avidly, perhaps because the A region was impaired by the proximity of the cell surface and could not undergo the conformational changes required for DLL.

Posttranslational modifications

The Sdr proteins are glycosylated by the action of the glysosyltransferases SdgA and SdgB that are encoded by genes located near the sdrCDE locus (33). They attach two N-acetylglucosamine molecules to serines in the SD repeat domain in a stepwise fashion (Fig. 3). Glycosylation creates a dominant epitope for human antibodies, which can represent up to 1% of IgG. Glycosylation protects the MSCRAMMs from degradation by host proteases such as the neutrophil cathepsin G and could thus contribute to virulence (33). Indeed, Sdg mutants have reduced virulence in a mouse sepsis infection model (34).

The Pls protein encoded by the type I staphylococcal cassette chromosome mec (SCCmec) element has a short C-terminal SD repeat region that is glycosylated by a combination of glycosyltransferases encoded by the genes gtfC and gtfD, which that are closely linked to the pls gene in SCCmecI along with SdgA and SdgB (35). Proteases can remove subdomain N1 from MSCRAMMs. The staphylococcal protease aureolysin cleaves close to the boundary between subdomains N1 and N2 of ClfA and ClfB, while the host protease thrombin removes the N1 subdomain of FnBPA. Following removal of N1, ClfB can no longer bind to fibrinogen (36). It is possible that removal of subdomain N1 also reduces the capacity of S. aureus to adhere to loricrin and cytokeratin 10 since these ligands share a common binding site in ClfB. The molecular basis for the loss of ligand binding is not known (7). The biological significance of the removal of subdomain N1 from ClfA and FnBPA is unclear since it does not reduce the ability of FnBPA to promote biofilm accumulation or the ability of ClfA and FnBPA to mediate adherence to fibrinogen (20, 37). Under certain conditions, FnBPs are degraded by the S. aureus protease V8 (38).

G5-E Repeat Proteins

Three staphylococcal proteins are characterized by the presence of G5-E repeats: SasG and Pls of S. aureus and Aap of S. epidermidis. Pls is encoded within the type I SCCmec cassette and is therefore only expressed by hospital-associated methicillin-resistant S. aureus strains carrying that element (39).

The proteins form highly extended fibrillar structures due to the unusual properties of the G5-E repeats and are able to mask the ability of shorter adhesins such as ClfA and FnBPs to promote adhesion to the ECM or invasion of mammalian cells (40–43). The G5-E repeats can also engage in homophilic interactions and promote biofilm formation (see below).

The N-terminal A domains of these proteins bind to cornified squamous epithelial cells by recognizing as yet unknown ligands and thus might contribute to nasal colonization (41, 43, 44). The A domain of Aap also promotes adhesion to hydrophobic surfaces in the initiation of biofilm formation by S. epidermidis (45, 46).

The ability to form very strong and stable (long and strong) extended structures is an unusual property of the intrinsically unstable repeated region (47–49). Both the G5 and E domains comprise two triple-stranded β-sheets with a central collagen-like triple helical region (Fig. 8). When an E domain is expressed alone or combined with a single G5 domain, the proteins are intrinsically disordered and unfolded (47). However, a G5₁-E-G5₂ moiety folds to form an elongated structure. When G5₂ folds, it triggers folding of the central E domain, which in turn promotes folding of G5₁. Thus, the entire repeat region can be considered to comprise overlapping G5-E-G5 cooperative folding units.

FIGURE 8.

G5-E domains of SasG and Aap. The G5 (red) and E (blue) domains each form two triple-stranded β-helices separated by a short collagen-like triple helix.

Cell-cell adhesion and biofilm formation can be promoted by G5-E repeats. First, the N-terminal A domain must be removed by proteolysis (49–51). Only then can the folded G5-E repeats engage in the homophilic interactions involving multiple contacts along the lengths of the two molecules. Molecular modeling suggested that the G5-E domains twist around each other in a twisted rope-like structure (Fig. 7). Atomic force microscopy allowed measurement of the forces involved in cell-cell interactions, and in cases where full engagement was likely to have occurred, separation required the two proteins to become unwound by successively unfolding the E and G5 subdomains (52).

Protein A, a Protein Comprising Three-Helical Bundle Repeats

The N-terminal region of protein A (Spa) comprises a tandem array of five separately folded three-helical bundles, each of which can bind to several different ligands (Fig. 1, Table 1). The A1 domain of von Willebrand factor, tumor necrosis factor receptor 1, and the Fc region of IgG each bind to the interface between helices 1 and 2, while the Fab region of IgM binds to the interface of helices 2 and 3 (53–56). The three-helical bundles of Spa can exist in slightly different conformations, a conformational plasticity that facilitates binding to structurally diverse ligands (57). Upon binding to the Fc region of IgG, there is considerable loss of conformational heterogeneity combined with structural rearrangements at the interface.

Located between the helical bundle domains and the cell wall-spanning region Xc is a an octapeptide repeat region Xr, which is highly variable in number. Region Xr is proinflammatory by activating interferon-β expression (58).

The S. aureus binder of Ig Sbi (59) contains four tandemly arrayed three-helical bundles, two of which can bind to IgG in a way similar to Spa. Sbi associates with the cell envelope by binding noncovalently to lipoteichoic acid (60). This allows the most N-terminal repeats to be exposed on the cell surface and to contribute to immune evasion. The C-terminal repeats are buried but are biologically active in promoting fruitless consumption of complement protein C3 when released into the medium (61).

The NEAT Motif Family

The near iron transporter (NEAT) domain proteins capture heme from hemoglobin and contribute to the survival of bacteria in the human host, where iron is restricted. Heme is transported by several CWA iron-regulated surface (Isd) proteins to the membrane transporter. Once in the cytoplasm, heme oxygenases release free iron (62, 63). The defining feature of Isd proteins is the presence of one or more NEAT motifs, which bind either hemoglobin or heme. The structures of NEAT domains have been solved and the molecular mechanisms of ligand binding elucidated (64).

Isd proteins have additional functions other than transport of heme. IsdA promotes adhesion to squamous epithelial cells and promotes nasal colonization (65). The C-terminal domain of IsdA also confers resistance to bactericidal lipids (66). IsdB binds directly to β3-containing integrins and promotes platelet activation and invasion of mammalian cells (67, 68). IsdH helps S. aureus to avoid phagocytosis by promoting degradation of complement opsonin C3b by an as yet unknown mechanism (69).

The Legume Lectin Domain of SraP

The serine-rich adhesin of platelets (SraP) is a member of a family of glycoproteins in Gram-positive cocci (70). The structural gene sraP is included in a locus comprising genes that encode glycosyltransferases (GtfA and GtfB) and accessory secretion factors (SecY2 and SecA2). A short serine-rich repeat region (SSR1) is located at the N-terminus of the CWA protein followed by the ligand binding BR domain (Fig. 1). The BR domain is composed of four separately folded subdomains that form a rigid bent rod (71). Two cadherin-like domains and a β-grasp fold domain act to project the legume-lectin domain. This has structural similarity to legume lectins and binds N-acetyl neuraminic acid (Neu5Ac) containing glycoproteins that are attached to the surface of mammalian cells. It also binds to salivary glycoprotein gp340, which is rich in a 5NeuAc containing trisaccharide (72). SraP promotes bacterial adhesion to and invasion of mammalian cells (71). This was apparent using an experimental system where there was little or no fibronectin present and employing a strain that expressed low levels of FnBPs. Thus, there was minimal involvement of the highly efficient FnBP-mediated uptake (see below). SraP, like IsdB, might act as an accessory invasin. The cadherin-like domains dimerize in solution and promote cell-cell accumulation and biofilm formation (71).

The Tandem β-Zipper Binding to Fibronectin Promotes Invasion of Mammalian Cells

S. aureus can invade many types of mammalian cells that do not normally engage in phagocytosis, such as various epithelial cells, endothelial cells, fibroblasts, osteoblasts, and keratinocytes (73–80). This allows bacteria to evade host immune defenses and antibiotics. Bacteria can escape into the cytosol by lysing the phagosomal membrane and then multiply before eventually destroying the integrity of the cell and being released. In some cases staphylococci enter a semidormant state called the small colony variant (81). Small colony variants do not express cytolytic toxins, so the invaded cell stays intact. Small colony variants express FnBPs at high levels, which facilitates efficient uptake into nearby cells when they are released (82).

Fibronectin is a dimeric glycoprotein comprising three distinct domains: FI, FII, and FIII (83). In plasma fibronectin exists in a compact state which is held together by intramolecular interactions between F1 modules in the N-terminal domain and FIII modules in the C-terminal cell binding domain. The binding of FnBP to FI modules in the N-terminal domain of fibronectin by the tandem β-zipper mechanism (Fig. 9) forces a conformational change in fibronectin (84). This exposes a cryptic RGD integrin binding site in the 10th FIII module. This RGD motif is recognized by α5β1 integrins on the surface of mammalian cells. In this way, fibronectin acts as a bridge between S. aureus and host cells. Clustering of integrins triggers intracellular signaling by the focal adhesion kinase and Src kinase and subsequently endocytosis (77, 85, 86).

FIGURE 9.

Fibronectin binding by FnBPs. The figure shows how one fibronectin binding repeat of the unstructured fibronectin binding region of FnBP binds to N-terminal type I modules of fibronectin by the tandem β zipper mechanism. Potentially up to 10 such interactions can occur per molecule of FnBP. Intramolecular interactions between the N-terminal type I modules and C-terminal type III modules result in allosteric activation of the 10th type III module, exposing an RGD motif which engages an α₅β₁ integrin on the surface of a mammalian cell to promote invasion by endocytosis.

Nucleotidase Motif

A nucleotidase motif was discovered in a previously uncharacterized CWA protein called SasH. It was shown to be enzymatically active and, importantly, to contribute to the ability of S. aureus to evade innate immune responses in the infected host (87, 88). The ability of SasH (now called adenosine synthase [AdsA]) to convert intracellular ATP to the potent immunoregulatory molecule adenosine inhibited the oxidative burst and promoted survival within neutrophils.

Neutrophils also contribute to the host defenses by releasing DNA, which forms neutrophil extracellular traps (89). These can entrap bacteria and facilitate killing by their attached granule contents and histones. S. aureus expresses an extracellular deoxyribonuclease which degrades neutrophil extracellular trap DNA producing phosphomononucleotides, including 2′ deoxyadenosine 3′ monophosphate (dAMP) (90, 91). AdsA converts dAMP to 2′ deoxyadenosine, which triggers caspase 3-promoted apoptosis in macrophages and monocytes (90). By excluding macrophages from a developing abscess, AdsA helps to promote persistent infection.

CWA PROTEINS AS COLONIZATION AND VIRULENCE FACTORS

Investigating the Contribution of CWA Proteins to the Virulence of S. aureus

To investigate if a CWA protein can act as a virulence factor, a mutant defective in the protein can be isolated and its virulence compared to the wild type in an appropriate animal infection model. However, functional redundancy sometimes makes it difficult to show conclusively that a mutant lacking a single factor has reduced virulence. For example, S. aureus expresses several CWA proteins that bind fibrinogen, and most strains elaborate two fibronectin binding proteins. Another approach is to express the CWA protein in a surrogate host such as Lactococcus lactis (14, 92–94) or Staphylococcus carnosus (95). Until recently, it was difficult to isolate mutations in most clinically relevant strains, so earlier studies of the role of CWA proteins were performed with laboratory strains such as Newman and derivatives of NCTC8325. However, it is now possible to circumvent the restriction barriers that prevent transfer of plasmid DNA from Escherichia coli cloning hosts into S. aureus, rendering feasible genetic manipulation of diverse clinical isolates (96–98).

Another important consideration is differences in animals used for infection models compared to humans. Results of experimental infection studies must be interpreted with caution if the surface protein being analyzed has a lower affinity for ligands in the animal compared to humans. One way to circumvent such difficulties is to engineer a humanized mouse expressing the human version of the ligand (99). For example, IsdB does not bind mouse hemoglobin, so a transgenic mouse expressing the human version of the protein has been employed in infection studies (100). Another option is to murinize the pathogen by engineering the virulence factor to bind to the murine version of the ligand as has been achieved with internalin A of Listeria monocytogenes (101).

A summary of studies demonstrating that bacterial mutants defective in CWA proteins have reduced virulence in different infection and colonization models is shown in Table 2. A systematic analysis of mutants defective in individual surface proteins was performed with mutants of strain Newman using a mouse model of bacteremia, dissemination, and abscess formation (102). Mice were injected intravenously with bacteria, and the bacterial load in the kidneys and the number of abscesses were measured. This model measured the ability of bacteria to survive in the bloodstream and to invade kidney tissue and establish an abscess. Several mutants had a statistically significant >1 log reduction in viable counts in kidney tissue, indicating the importance of CWA proteins, although the precise role of each protein is not known (103). The study was limited by the fact that strain Newman does not express Cna and lacks the ability to display FnBPs on its surface due to mutations in the fnbA and fnbB genes, which prevent anchorage of the proteins to the cell wall (104).

TABLE 2.

CWA proteins as colonization and virulence factors studied using animal models

Role in colonization or infection	CWA protein	Mechanism	References
Nasal or skin colonization	ClfB	Adhesion to loricrin on squames	14
	IsdA	Adhesion to squames	119
	SasX	Adhesion to squames	115
Endocarditis	ClfA	Adhesion to thrombus	163
	FnBPA	Adhesion to thrombus; invasion of adjacent endothelium	94
	ClfB	Adhesion to thrombus	164
	SraP	Adhesion to platelets; colonization of thrombus	165
Mastitis	FnBP	Invasion of epithelial cells in mammary gland	(166)
Pneumonia	Protein A	Enhanced inflammation of lung epithelium	167
Foreign body infection	FnBP	MRSAa biofilm-promoted infection	168
		Adhesion to intra-aortic patch	169
Ocular keratitis	Cna	Enhanced colonization and infection	170
Septic death	ClfA	Reduced opsonophagocytosis	102, 107, 141
Survival in blood	Protein A		171, 172
	IsdH		69
	AdsA		87
	SasX		115
Kidney abscess	AdsA, ClfA, ClfB Spa, IsdC. IsdA, IsdB, SdrD SdrE	Increased survival in blood stream prior to kidney infection	87, 102
Septic arthritis	ClfA, protein A	Enhanced survival in bloodstream prior to invasion of joint	141, 172
	Cna	Enhanced survival in blood stream; adhesion to cartilage within joint	173
Joint infection	Fibrinogen binding MSCRAMMs	Biofilm formation in synovial fluid	174
Subcutaneous abscess	Spa	Abscess development, bacterial load	171
	FnBPs, ClfA		175
	SasF		161

^a MRSA, methicillin-resistant S. aureus

The collagen binding protein promotes virulence in infections where adhesion to collagen-rich tissue is important, such as ocular keratitis and septic arthritis. The strength of adhesion of the recombinant protein to immobilized collagen in vitro correlated with the severity of the disease (105). Cna also interferes with complement fixation and most likely contributes to innate immune evasion during infection (19).

Studies of S. aureus strains that colonized cardiac devices in human patients compared to those isolated from cases of uncomplicated bacteremia identified polymorphisms in the fnbA gene encoding FnBPA associated with complicated bacteremia that increased the affinity of the protein for fibronectin in vitro (106). This implies that selection occurred in vivo for strains that had a high affinity for fibronectin that coated the device implanted in the patient.

Rather than isolate null mutants that lack the ability to express a surface protein bacteria expressing variants, the protein on the cell surface with amino acid substitutions that cause loss of function can be studied. For example, mutants expressing mutants of ClfA (107) and Cna (105) with reduced ligand binding were shown to lack virulence in a model of septic arthritis. A mutant expressing AdsA that was defective in adenosine synthase survived less well than the wild type in blood stream infection (87). An advantage of this approach is that possible pleiotropic effects of the missing surface protein on the elaboration of other CWA proteins are avoided.

Infection studies with the surrogate host L. lactis expressing CWA proteins have been used to study aspects of the infection process. The ability of L. lactis to colonize a sterile thrombus on damaged heart valves in a rat model of endocarditis was dependent on bacteria being able to adhere to fibrinogen/fibrin on the damaged tissue. This could be mediated either by ClfA or the A domain of FnBPA (94). Subsequent bacterial proliferation and expansion of the lesion required FnBP-promoted invasion of cells of the surrounding endothelium.

CWA Proteins Promote Colonization of the Host

Approximately 20 to 30% of the population are colonized persistently by S. aureus in the nares, with the remainder being intermittently colonized (108, 109). The factors that determine whether an individual is a persistent carrier are not completely understood (110). S. aureus colonizes the nares by adhering both to ciliated epithelial cells in the posterior region of the nares and to squamous epithelial cells in the anterior nares (111).

Adhesion to ciliated epithelial cells is mediated in part by the zwitterionic bacterial wall teichoic acid (WTA) binding to the scavenger receptor SREC-1 (112). This occurs through interactions involving positively charged d-alanine residues that decorate WTA binding to negatively charged residues on the host protein (112). Glycosylation of WTA contributes to the ability of S. aureus to colonize the nares of cotton rats, although the receptor for the N-acetyl glucosamine residues is not known (113).

Several CWA proteins promote adhesion of S. aureus to the cornified envelope of squamous cells (corneocytes). ClfB and IsdA have been shown to promote nasal colonization in rodent models and, in the case of ClfB, humans (14, 65, 114). SdrC, SdrD, SasG, Pls, and SasX contribute to bacterial adhesion to corneocytes in vitro, and SasX can also promote colonization of the nares of mice (43, 93, 115). Pls and SasX are encoded by genes located within mobile genetic elements and are thus only expressed by a subset of methicillin-resistant S. aureus strains carrying the relevant element. Pls is encoded by a gene carried within the type I SCCmec cassette (116), while the sasX gene is carried by a lysogenic phage found only in ST239 methicillin-resistant S. aureus strains that are prevalent in the Far East (115). The sasG gene is carried by 50% of the most prevalent lineages (clonal complex 1 [CC1], CC5, CC8, CC15, and CC22) but is absent from CC12, CC25, CC51, CC45, and CC30 strains (117).

It is well established that ClfB can bind to cytokeratin 10 and loricrin, proteins that only occur in corneocytes (5, 7, 118). Loricrin is probably the more important ligand in vivo; loricrin knockout mice are defective in colonization (14). ClfB binds to loricrin by the high-affinity DLL mechanism (14). IsdA can also bind to loricin, cytokeratin 10, and involucrin, but the mechanistic basis of these relatively weak interactions is not known (119). SdrD binds to desmoglein 1, a component of both the desmosomes that hold keratinocytes together and the corneodesmosones that regulate desquamation in the stratum corneum (120, 121).

S. aureus is regarded as a minor or transitory resident of healthy skin. However, in eczema (atopic dermatitis [AD]) patients suffering from a flare-up, the diversity of normal skin microbiota is reduced and S. aureus proliferates (122). The bacterium releases several proteins that can exacerbate inflammation in AD skin (123). Colonization of AD skin is at least in part promoted by the ability of bacteria to adhere to corneocytes in a ClfB-dependent fashion (124). Adhesion to loricrin and/or to ligands that become exposed on the surface of the deformed corneocytes where corneodesmosome proteins are present on villous-like protrusions is an important first step in colonization (125).

IMMUNE EVASION PROMOTED BY CWA PROTEINS

Evasion of Innate Immune Responses: Inhibition of Complement Activation

Protein A is a multifunctional CWA protein of S. aureus that promotes evasion of innate responses by its ability to capture IgG from serum. Bacterial cells become coated with IgG that is bound incorrectly so that the Fc region of the antibody cannot be recognized by the Fc receptor on neutrophils (126, 127). To initiate the classical pathway of complement fixation, the complement protein C1q must be able to bind to the Fc region of bound IgG (128, 129). Thus, opsonophagocytsosis is comprehensively inhibited by preventing recognition of the opsonins IgG and C3b by both the Fc receptor and the complement receptor on neutrophils.

The collagen binding protein also blocks initiation of the classical pathway of complement fixation by binding to the collagen triple-helix structures of C1q, displacing C1r-C1s and preventing complement activation (19).

Evasion of Innate Immune Responses: Enhanced Destruction of C3b

Complement activation results in the complement protein C3b being deposited on the bacterial cell surface, allowing it to be recognized by the complement receptor on phagocytic cells. Accumulation of C3b begins the amplification process of the alternative activation pathways and also triggers activation of C5 convertase (129). Three CWA proteins, SdrE, ClfA, and IsdH, contribute to enhanced complement degradation (69, 130–132). Only in the case of SdrE is the mechanistic basis understood. SdrE binds the host complement regulatory protein factor H, which in turn recruits and activates the specific C3b-degrading protease factor I (13). It is possible that ClfA can bind and activate factor I directly without the requirement of factor H, but evidence for this is incomplete and further work is required.

Interference with Adaptive Immune Responses

A hallmark of S. aureus infections is the elaboration of superantigens which interfere with adaptive immunity. Most strains have the ability to secrete one or more small proteins (enterotoxins and toxic shock syndrome toxin 1) that have the ability to bind the MHC-II receptor of antigen-presenting cells and promote binding to T cell receptors (133). This triggers activation and proliferation of a plethora of T cells without appropriate guidance and recognition by bound antigen fragments. The results is uncontrolled expression of cytokines, which in the worst case scenario can cause toxic shock syndrome. Depletion of T cells also results in failure to mount an appropriate adaptive immune response.

Recently, it has been recognized that protein A is a crucially important factor in the interaction of S. aureus with its host. Protein A is expressed by all clinical strains of S. aureus and is important in persistent nasal carriage and in disrupting adaptive immune responses during infection (134, 135). Protein A is released from the cell surface by the action of autolysins, resulting in soluble molecules with fragments of peptidoglycan bound to the C-terminus (136, 137). The Ig binding repeats of Spa bind to and cross-link the variant V_H3 chain of receptors on B cells in lymphoid tissue (55). This results in B cell proliferation. During infection there is a rise in the level of V_H3 clonal IgM and IgG in serum, but these antibodies lack the ability to bind to S. aureus antigens (137, 138). This inappropriate expression of Ig requires the cooperation of CD4⁺ T cells (137), although it is not clear if antigen presentation is involved. The C-terminal peptidoglycan fragment is implicated in activation of (nucleotide-binding oligomerization domain) NOD1/NOD2 signaling, which in turn triggers maturation of the B cells into antibody-secreting plasma cells (55, 137, 139). Depletion of B cells from lymphoid tissues causes anergy, and the lack of immunological memory results in S. aureus being able to reinfect and is probably responsible for bacteria being able to proliferate in the nares of persistent carriers.

CONCLUDING REMARKS

CWA proteins provide opportunities for S. aureus to interact with the host, both in the commensal state and during infection. Given that the repertoire of CWA proteins is restricted, it is perhaps not surprising that several of these proteins have been found to be multifunctional.

Recently, it was proposed that CWA proteins should be categorized based on structural motifs and functional attributes (2, 140). There has been a tendency to use the acronym MSCRAMM inappropriately for all CWA proteins regardless of function and whether or not they bind to components of the ECM.

The functions of some CWA proteins have yet to be established, and it is likely that new ligands will be discovered for those with known functions. Structural analysis will provide clues about function as well as the application of high-throughput techniques for discovering binding partners. Far Western blotting, phage display, and two-hybrid analysis have identified potential ligands which must then be validated by detailed analysis to show biological relevance (21, 24, 26, 120).

Until recently, most studies were performed with laboratory strains. It is now possible to genetically manipulate strains from all genetic lineages so that clinically relevant bacteria can more easily be studied (96–98). This is important because since variations in the repertoire of surface proteins occur in different lineages, sequence variations in ligand binding domains might lead to different affinities for ligands, and the timing of and regulation of gene expression can result in different levels of protein abundance on the cell surface.

Several CWA proteins have been investigated as potential antigens for vaccination to protect against S. aureus infections and have been shown to offer protection in murine models of infection (65, 141–143). Furthermore, recombinant truncates that have been engineered so as not to bind host ligands offer the potential for improved antigenicity (107, 144). Several combinations of antigens including fragments of CWA proteins and other surface components have been formulated (145–147). However, the only phase III clinical trial performed to date was with a single-antigen IsdB and resulted in failure for reasons that are not clear (148, 149). Any future clinical trials will require the use of an adjuvant that triggers a cellular response involving T_H1 cells and/or T_H17 cells as well as interleukin-17 and interferon-γ-mediated recruitment of neutrophils and must avoid the immune dysregulation caused by expression of superantigens (146, 149).

In conclusion, CWA proteins carry out a range of important functions and are essential for successful colonization of the host and for infection. Structural and functional analysis has provided a rational framework for classification of CWA proteins and has helped to define the mechanistic basis of host-pathogen interactions.