Human pathogens utilize host extracellular matrix proteins laminin and collagen for adhesion and invasion of the host


Correspondence: Kristian Riesbeck, Medical Microbiology, Department of Laboratory Medicine Malmö, Skåne University Hospital, Lund University, SE-205 02 Malmö, Sweden. Tel.: +46 40 338494; fax: +46 40 336234; e-mail:


Laminin (Ln) and collagen are multifunctional glycoproteins that play an important role in cellular morphogenesis, cell signalling, tissue repair and cell migration. These proteins are ubiquitously present in tissues as a part of the basement membrane (BM), constitute a protective layer around blood capillaries and are included in the extracellular matrix (ECM). As a component of BMs, both Lns and collagen(s), thus function as major mechanical containment molecules that protect tissues from pathogens. Invasive pathogens breach the basal lamina and degrade ECM proteins of interstitial spaces and connective tissues using various ECM-degrading proteases or surface-bound plasminogen and matrix metalloproteinases recruited from the host. Most pathogens associated with the respiratory, gastrointestinal, or urogenital tracts, as well as with the central nervous system or the skin, have the capacity to bind and degrade Lns and collagen(s) in order to adhere to and invade host tissues. In this review, we focus on the adaptability of various pathogens to utilize these ECM proteins as enhancers for adhesion to host tissues or as a targets for degradation in order to breach the cellular barriers. The major pathogens discussed are Streptococcus, Staphylococcus, Pseudomonas, Salmonella, Yersinia, Treponema, Mycobacterium, Clostridium, Listeria, Porphyromonas and Haemophilus; Candida, Aspergillus, Pneumocystis, Cryptococcus and Coccidioides; Acanthamoeba, Trypanosoma and Trichomonas; retrovirus and papilloma virus.


The extracellular matrix (ECM) is the acellular proteinaceous part of the animal tissues. ECM is involved in building structural scaffolds, regulation of physiological processes, cellular signalling, migration and transport of solutes across the body tissues and cellular barriers (Hynes, 2009). It constitutes the anchoring platform for epithelia, designated the basement membrane (BM), and also surrounds blood capillaries and neurons, and is part of the connective tissues that fill interstitial spaces of the tissue parenchyma (Fig. 1). The BMs are basically made of laminin (Ln) and collagen polymeric sheets that also contain nidogen, perlecan and agrin cross-linked proteins involved in various functions. On the other hand, connective tissues consist of intricate structures built by collagen, elastin, fibrillin, Ln, fibronectin, vitronectin, thrombospondin, proteoglycans and hyaluronic acid (Fig. 1). Being ubiquitously and profusely distributed, the ECM components are attractive targets for adherence and invasion by various human microorganisms.

Figure 1.

Overview of ECM. Human tissue sections showing ECM distribution stained with specific anti-collagen or anti-Ln antibodies (indicated in parentheses). Epithelial cell layers of tissues are indicated by arrows. E represents the interstitial spaces filled with ECM; C indicates capillaries. Major ECM components and their various properties are shown in the table. These figures were downloaded from the human protein Atlas (

Besides immunological host defence strategies, physical containments (barriers) also exist between different tissues to prevent infection by microorganisms (Lemichez et al., 2010). The ECM components Lns and collagens are targeted by microorganisms for adhesion and induction of host inflammatory responses, leading to colonization and invasion of host tissue. (Diacovich & Gorvel, 2010; Lemichez et al., 2010). Some microorganisms degrade ECM components with their secretory or surface-bound proteases, or ‘hijacked’ host proteases [e.g. plasminogen (Plg)], during inflammatory responses resulting in increased tissue damage. Moreover, during infection, partially degraded and exposed ECM components are attractive targets for adherence of pathogens, which is achieved by various microbial surface–exposed adhesive proteins (Vanlaere & Libert, 2009).

To date, several microorganisms belonging to diverse niches have been described to bind/degrade Lns and collagens. In this review, we describe several viral, bacterial, fungal and protozoal human pathogens that are known for their Ln- and collagen-binding/degradation characteristics. We conclude that usually multiple surface proteins (adhesins) interact with both Ln and collagen. Interestingly, these protein–protein interactions play significant roles during the attachment of many microorganisms to host tissues. In addition, secreted proteases produced by invasive pathogens known to cause infections in the central nervous system (CNS) can degrade Lns/collagens in order to breach tissue barriers such as the blood–brain and blood–cerebrospinal fluid (CSF) barriers. Controlling the microbial-dependent Ln/collagen interactions may thus be an attractive target for future antimicrobial therapeutics.

Structure function and distribution of Ln

Lns are multifunctional heterotrimeric molecules consisting of an α (≈ 400-kDa), a β (≈ 200-kDa) and γ chain (≈ 200-kDa). There are five α chain, three β chain and three γ chain isoforms identified in humans. In total, 15 Ln molecules (heterotrimeric isoforms) are classified on the basis of different combinations of the chains (Aumailley et al., 2005; Scheele et al., 2007; Guess & Quaranta, 2009; Durbeej, 2010). The chains interconnected with disulphide bonds at their C-terminal regions, form a triple coiled-coil region together that result in a ‘crucifix’-shaped structure (Fig. 2a). The whole molecule thus appears with three N-terminal short arms and one C-terminal long arm. The N-termini of the α1, α2, α3b, α5 and αD chains exhibit three globular domains, while the α3a and α4 chains lack these domains. In contrast, the β1, β2 and γ1, γ3 chains have two N-terminal globular domains, whereas the β3 and γ2 chains contain only one domain. The α, β and γ chains of Ln are independently expressed and assembled into trimers before extracellular delivery. The secreted trimers may be isomeric (e.g. Ln-111) or heteromeric (mixed chains) and are finally arranged into networks resulting in polymeric sheets, which form the BM (Colognato et al., 1999; McKee et al., 2009). Lns are heavily glycosylated proteins with 12–27% carbohydrate content often present as complex-type carbohydrates (Valkonen et al., 1993, 1994, 1997; Terres et al., 2003).

Figure 2.

Ultrastructure of Ln. (a) A schematic representation of the Ln heterotrimer indicating the binding regions of various proteins. Each heterotrimer consists of an α (≈ 400-kDa), β (≈ 200-kDa) and finally γ (≈ 200-kDa) chain interconnected with disulphide bonds forming a coiled-coil region. The shape of the Ln molecule resembles a ‘crucifix’. The N-terminal globular domains (LN domains) of all three chains are involved in the polymerization of Lns. The coiled-coil region of the β chain harbours the agrin-binding domain, whereas the short arm of the LE4 domain of the γ chain binds to nidogen (shown in red, yellow and green). The coiled-coil region of the LR binds to the short arm of the γ chain. The C-terminus of all α chains contains five globular domains (G1–G5), which have several biological functions. (b) Crystal structure of the LN domain of the α5 chain [protein database (PDB):2Y38]. The LN domain appears like a β jelly roll with eight antiparallel β sheets (β1–β8), five small helices and several loops (shown in magenta colour). Amino acid residues P229, L230, E231 and E234 located at the β5′–β6 loops are involved in recognition of the LN domains of the β and γ chains during polymerization. (c) The LG domains consist of 13–14 tightly packed antiparallel β-strands arranged into two sheets forming a ‘β jelly-roll’ shape. Each domain has one or two conserved disulphide bonds, and some domains have one Ca++ -binding site. The structure of the LG domain of mouse Ln α2 chain (PDB; 1QU0) is shown with Ca++ and SO4 bound as a representative domain. 3D structures were prepared using pymol (The PyMOL Molecular Graphics System, Version 1.3, Schrödinger, LLC).

LN domains

The N-terminal globular domains (approximately first 240–250 amino acid residues) are designated Ln N-terminal domains (LN) and are involved in the polymerization of the Ln molecule. LN domains of α, β and γ derived from three different molecules interact with each other to build the polymeric network (Fig. 2b). All LN domains are structurally similar and are conserved between members of the same chain subfamily (Odenthal et al., 2004). The 3D crystal structure of mouse α5-chain LN domain has recently been solved, and the structure (PDB; 2Y38) is shown in Fig. 2b (Hussain et al., 2011). The amino acid region that interacts with LN domains of the β and γ domains is indicated in Fig. 2b.

Epidermal growth factor-like domains (LE domains)

The N-terminal (short arm) of all chains encompasses a tandem distribution of Ln-type epidermal growth factor (EGF)-like domains that interconnect the globular domains of the Ln molecules. The LE domains are defined as extracellular protein modules, which are cross-linked by three or four intramolecular disulphide bonds (Green and yellow domains, Fig. 2b). EGF domains are found in several human proteins involved in a diverse array of functions such as signalling, growth and development. Other BM proteins, such as nidogen (Hopf et al., 2001), perlecan (Iozzo, 2005), agrin (Banyai et al., 2010), netrin (Schneiders et al., 2007) and usherin (Bhattacharya et al., 2002), all contain similar EGF-like domains. LE domains function as a spacer between globular domains and appear as ‘rod-like’ structures with limited flexibility.

L4 domains

In addition to LN domains, L4 domains (spanning approximately 200–250 amino acids) are other types of globular domains present in the Ln short arms (Fig. 2a). This type of globular domain is also present in perlecan. They usually flank LE domains at both ends, with 3 Cys residues at the N-terminal and 5 Cys at the C-terminal ends (Schulze et al., 1995). Thus, the L4 domains have been considered as an insert between the LE domains. The α1, α2, α3B and α5 chains consist of two L4 domains, while β1, β2 and all γ chains have only one L4 domain (Tunggal et al., 2000). The L4 domain sequences of the β1 and β2 chains are different in comparison with the L4 domains of α and γ chains and are hence named LF domain. The functions of L4 domains have not been characterized in detail, but they have been shown to directly interact with fibulin (Utani et al., 1997). Interestingly, a defect in the L4 domain α2 causes congenital muscular dystrophy (Allamand et al., 1997; Tunggal et al., 2000).

Coiled-coil region

The three Ln chains (α, β and γ) of a heterotrimer are intertwined as a C-terminal coiled-coil module making a rod-like appearance (Fig. 2a). This particular region is present in all chains and consists of approximately 600 amino acids with a series of heptad repeats involving hydrophobic and ionic interactions to hold together all three chains (Beck et al., 1993). In addition, at the N-terminal end of the long arm near the crossing point, the three chains are linked by disulphide bonds, while at the C-terminal end, only β and γ chains are linked by disulphide bonds (Tunggal et al., 2000). Theoretically, there is a possibility of 45 different combinations of the 15 Ln chains, but only 15 isoforms have been identified to date. This selectivity between chains regarding trimer formation has been shown to be dependent on the coiled-coil region interactions (Sanz et al., 2003; Scheele et al., 2007; Macdonald et al., 2010). The β chains have an additional β-knob domain (30–35 amino acids) with six Cys residues in the middle part of the coiled-coil region. Besides the important role in trimer assembly, the coiled-coil region of the β chain interacts with agrin (McFarlane & Stetefeld, 2009; Taniguchi et al., 2009), integrins (Taniguchi et al., 2009) and finally the Ln receptor (LR) (Fatehullah et al., 2010) (Fig. 2a).

Globular domains

The C-termini of all Ln α chains (approximately 900 amino acids) are arranged in five globular domains (G1–G5), displaying a length of 180–200 amino acids each (Tunggal et al., 2000) (Fig. 2a). In contrast, Ln β and γ chains lack globular domains at their C-terminal ends. These LG modules are present in many other proteins including neurexin receptors, complement protein S, agrin, perlecan, sex hormone (HSBG) and Drosophila proteins scrubs, slit and fat. Similar but distantly related domains are also present in pentraxins (serum amyloid proteins, C-reactive protein), thrombospondin and some collagens (Timpl et al., 2000; Rudenko et al., 2001). An LG module consists of 13–14 tightly packed antiparallel beta strands, arranged into two sheets forming a ‘jelly-roll’ shape (Fig. 2c). LG1-2-3 and LG4-5 have very short connecting spacers, whereas LG3 is separated from LG4 with a long spacer. Sequence alignment of the globular domains G1–G5 revealed that these domains share different degrees of homology; the α3 G1–G5 domains share 33.9% and 33.7% identity and 53.1% and 49.9% similarity with α4 and α5, respectively. On the other hand, the globular domains of α1 have 40.0% identity and 59.9% similarity with the α2 domains (Singh et al., unpublished data). The diversity of the G1–G5 sequences most likely plays a role in the selectivity of multiple receptors/ligands with various functions.

Functions and distribution

The Ln α chains play significant roles in several biological processes. For instance, α4 and α5 are vital for organogenesis during embryo development and are expressed more abundantly than α1, α2 and α3 (Erickson & Couchman, 2000). LG1-3 domains are known to interact with at least seven different integrins (α1β1, α2β1, α3β1, α6β1, α7β1, α9β1, and finally αvβ3), dystroglycan (DG), heparin sulphate proteoglycans (perlecan and agrin), ‘carbohydrate adduct of proteins-1’ (HNK-1), lutheran and syndecans (Fig. 3). Agrin functions as a bridge between Ln α chains and integrins that interact with the coiled-coil region in the β-chain (Talts et al., 1999; Sasaki et al., 2004; Yurchenco et al., 2004; Han et al., 2009; Gawlik et al., 2010). Lns also interact with a highly conserved LR, which is presented as a 67-kDa dimer on several different epithelial cell surfaces. In addition, the LR interacts with other ECM proteins including collagens. These interactions contribute to cell attachment, differentiation, cell shape, movement, maintenance of tissue phenotypes and promotion of tissue survival. Besides the physiological functions, the interactions between Lns and integrins collectively provide mechanical stability and integrity to the BM (Fig. 3) (Erickson & Couchman, 2000; Yurchenco et al., 2004).

Figure 3.

The BM with arrangement of various proteins and their interactions with cell surface receptors. The intermolecular interactions are shown by dotted lines. Epithelial cells secrete ECM components to synthesize the BM that functions as a support/platform and anchorage for the cells. The BM consists of Lns, collagen IV, nidogens and the heparin sulphate proteoglycan (HSPG) perlecan and agrin. Fibronectin, secreted protein acidic rich in cysteine (SPARC), fibulins and several other collagens are also present in the BMs (Yurchenco et al., 2004). Lns and collagen IV form independent polymeric networks. Nidogen interconnects the Ln and collagen networks with the help of perlecan and thus forms a rigid BM. The globular domains (G1–G3) of the Lns α-chain interact with integrins (α1β1, α2β1, α3β1, α6β1 and α6β4), while collagen binds to α1β1 and α2β1. In addition, Lns bind to other cellular receptors including α-DG, syndecans, lutheran, the LR (a 67-kDa surface protein), sulfatides and HNK-1 (carbohydrate adduct of protein and lipids).

Ln isoforms are differentially distributed in body tissues, organs and other infection-prone locations (Supporting Information, Table S1). Many studies in animal models have demonstrated the pattern of Ln isoforms' expression during foetus development. The expression of a particular isoform, or replacement of one isoform by another, is crucial during developmental processes, and failure cause fatal consequences (Roediger et al., 2010). However, there are only a limited number of reports available on the distribution of Lns in mature body tissues and organs. The Ln α1, β1 and γ1 chains have been detected in normal colorectal and breast tissues as well as in carcinomas, while the α2 chain is less abundant (Hewitt et al., 1997). Ln chains α1, 2, 3, 6 and β1, 2 and γ1 are found in human gastric mucosa (Virtanen et al., 1995). In contrast, another study suggested that the α1 chain is restricted in tissue distribution, whereas the α5, β1 and γ1 chains are prevalent in the BMs of normal tissues, gastric carcinoma and infiltrating breast carcinomas (Maatta et al., 2001). Ln-5 (α3 β3 γ2) is found in the colorectal epithelium (Lohi et al., 2000), whereas neuronal cells (Schwann cells) have a prominent α2-chain that is also utilized by invasive pathogens (Marques et al., 2001). Furthermore, it has been suggested that the epidermal BM contains multiple Ln isoforms including Lns 5 (α3 β3 γ2), 6 (α3 β1 γ1), 10 (α5 β1 γ1) and possibly Lns 7 (α3 β2 γ1) and 11 (α5 β2 γ1) (McMillan et al., 2006). Normal gastric epithelial cells and carcinomas express Ln-5 (α3 β3 γ2) (Lotz et al., 1997; Bouatrouss et al., 2000; Teller & Beaulieu, 2001). The same isoform, Ln-5, is also prevalent in oral epithelial cells and periodontal tissues (Thorup et al., 1997; Kinumatsu et al., 2009). Lung epithelial cells express several isoforms such as Ln-10/11 (Kikkawa et al., 1998; Carterson et al., 2005). In contrast, α1/2, β1/2 and γ2 Ln chains are expressed in liver tissues (Lietard et al., 1998) (Table S1). Most studies on Ln distribution have been carried out by histological examinations and are largely dependent on antibodies against a particular Ln isoform or chain. The human protein atlas database also has a collection of data sets that can be used for obtaining recent information regarding Ln distribution in different parenchyma (

Structure function and distribution of collagen

Collagens, the major glycoproteins (GPs) of connective tissues, account for 30% of the total protein in the human body, where they are involved in maintaining tissue architecture, cell adhesion, angiogenesis and development (Myllyharju & Kivirikko, 2004). To date, 28 collagens, encoded by more than 45 genes, have been unravelled in humans (Table S2). Human collagens are grouped in isotypes numbered with roman numerals, which are in turn classified depending on the supramolecular structures these isotypes are involved in. Hence, it is subdivided into fibril-forming collagens, network-forming collagens, fibril-associated collagen with interrupted triple helices (FACIT), transmembrane collagens also known as membrane-associated collagen with interrupted triple helices (MACITs) and finally multiplexins (Table S2). Each gene encodes a polypeptide (α chain) further related to a collagen subtype depending on its sequence and its ability to associate with each other. Three α chains associate together to form a mature protomer that can be homotrimeric (i.e. similar α chains) or heterotrimeric (i.e. different α chains). The number of α chains varies between collagen subtypes and does not correlate with abundance; for example collagen IV includes 6 α chains, type I has two α chains and type II only one α chain. Of notice is that collagens have been characterized in mammals and other vertebrates but also in many invertebrates suggesting a conservation of these molecules and their functions and roles throughout the evolution (Khoshnoodi et al., 2008; Fleury et al., 2011).

The triple helix

In spite of the large number of members included in the family and their variability in tissue distribution, all the collagens share some common features. The most significative characteristic stands in the presence of a recurring triplet Gly-X-Y repeated n times with X and Y often being proline and hydroxyproline, respectively, in the central part of each α chain of the family (van der Rest & Garrone, 1991) (Fig. 4a and b). When association of the 3 α chains occurs, the Gly-X-Y-rich region folds as a right-handed coiled-coil triple helix typical of the collagen and collagen-like molecules. As a consequence, this structure is known as the collagenous domain.

Figure 4.

The ultrastructure of collagen. (a and c) Molecular structure of the collagen triple helix (PDB; 1Cag) with the amino acid sequence (Pro-Hyp-Gly)4-(Pro-Hyp-Ala)-(Pro-Hyp-Gly)5. (a) Interactions between three chains of the triple helix. The chains are interconnected by hydrogen bonds formed between small side-chain amino ( NH ) groups of glycine and α-carbonyl ( C = O ) groups of proline (shown as dotted lines). (c) The full-length surface structure of PDB; 1Cag showing the surface structure. The total thickness of the triple helix is approximately 1–2 nm. (b and d) Molecular structure of another triple helix (PDB; 3Adm) having the sequence (Pro-Pro-Gly)4-Hyp-Ser-Gly-(Pro-Pro-Gly)4. Similarly, like in (a), the hydrogen bond pattern between Gly and Pro of the chains is shown as dotted lines. (d) Full structure of PDB; 3Adm showing surface of the triple helix. (e) Outline of the events leading to the formation of a mature protomer for fibril- and network-forming collagens. The recognition and polymerization of the α chains occur through the C-terminal domain leading to subsequent folding of the molecule, and particularly, the triple helix following a C- to N-terminal direction. In contrast to fibril-forming collagens, network-forming collagens harbour a globular C-terminal end. After secretion, the N- and the C-terminal propeptides of the protomer are cleaved by specific proteases resulting in a mature molecule.

The triple helix structure that has been studied through the use of synthetic peptides shows the association between the three chains by regular hydrogen bonding pattern between Pro and Gly residues (Fig. 4a and c). The centre of the triple helix is occupied by glycine due to its small molecular size, whereas the proline and hydroxyproline residues are exposed at the surface of the molecule (Fig. 4b and d). Formation of hydroxyproline is the consequence of a modification by the prolyl 4-hydroxylase enzyme of a proline residue at the Y position of the triplet before secretion and processing of the mature collagen protomer (Gorres & Raines, 2010). The presence of proline residues at the X position and hydroxyproline at the Y position is important for the stability of the molecule. It was shown experimentally that replacing proline with an alanine or a hydroxyproline at the X position destabilizes the triple helix (Inouye et al., 1982; Shah et al., 1996; Burjanadze, 2000). In addition to the collagenous domain, every collagen molecule includes regions lacking the Gly-X-Y signature named noncollagenous (NC) domains. Collagens forming fibrils and networks possess two NCs at their N- and C-terminal ends, whereas FACITs, multiplexins and MACIT molecules often include several long NC regions scattered along the triple helix creating breaks in the triple helix conformation (Shaw & Olsen, 1991; Franzke et al., 2005) (Fig. 5).

Figure 5.

Supramolecular structures formed by some archetypal collagens. Types I and II: association of mature protomers together leads to the formation of microfibrils which in turn assemble into fibrils. Type IV: dimer formation occurs by the association of two protomers through their globular NC C-terminal domain. Dimers interact together through the 7S domain to constitute tetramers. Networks are the result of the two first steps linked to additional lateral interactions between the molecules. Regarding the type VI collagens, there is an association between the dimer and tetramer that takes place inside the cytoplasm of the cells. Connection of tetramers leads to the formation of long filaments called ‘beaded filaments’ according to their appearance in electron microscopy. FACITs: collagen IX on the one hand and collagens XIV and XII on the other hand are typical FACITs for collagen II-rich fibres and collagen I fibres, respectively. FACITs protomers attach at the surface of fibres with the C-terminal part protruding and regulate fibrillogenesis. Transmembrane collagens (XIII, XVII, XXIII and XXV) are synthesized and assembled inside the cell. The N-terminal NC domain (N-terminal NC) is located inside the cell, whereas the triple helix region is extracellular. The collagen XVIII possesses a globular C-terminal domain that can be released by proteolysis resulting in a peptide called endostatin that displays antiangiogenic properties. Multiplexin collagens XVIII and XV are found in BM. XVIII shown here has C-terminal globular domain endostatin and N-terminal NC domains.

NC domains

In the majority of fibril-forming collagens, the N-terminal NC and C-terminal NC are removed after the secretion of the mature protomer to allow the assembly into fibrils. On the contrary, the domains are retained in the molecules from the other subgroups (Canty & Kadler, 2005) (Fig. 4e). The primary structure, length and number of the NC domains are isotype specific so logically NC domains are involved in specific recognition of the chains to form the mature trimeric protomer (Fig. 4e) (Lees et al., 1997; Boutaud et al., 2000; Khoshnoodi et al., 2006). After recognition and formation of interchain disulphide bonds (Malone et al., 2005; Khoshnoodi et al., 2006), the extension of the twisting of the three chains occurs spontaneously in a zipper-like fashion from the C--terminal to the N-terminal end (Engel & Prockop, 1991; Soder & Poschl, 2004) except few exceptions like collagen XVII and XIX for which folding begins from NC domains located at the N-terminal or inside the triple helix (Areida et al., 2001; Boudko et al., 2008).

Supramolecular organization

The mature protomers, released as extracellular products, undergo further assemblage to form networks or fibrils usually in association with other ECM proteins. The best-known collagenous structures are probably the fibrils, formed by collagen types I and II in bones or the skin and the cartilage, respectively (Fig. 4e). These fibrils are relatively similar in shape: longitudinally each unit measures about 300 nm in length and when associated together they appear in electron microscopy as D-staggered bands with a periodicity of 67 nm, explained by a gap between each molecule of 67 nm. After secretion and proteolytic cleavage of N- and C-terminal NC, the molecules self-assemble in a concentration-dependent manner into microfibrils (Kadler et al., 1996; Hulmes, 2002) that, subsequently, self-associate to form fibrils (Fig. 4) strengthened by inter- and intramolecular cross-links involving lysine and hydroxylysine residues (Canty & Kadler, 2005). Although collagen I is the major component of fibrils in bone and skin, and type II is the prominent collagen in articular cartilage, the fibrils are heterotypical, containing molecules as decorin, periostin and other collagen isotypes (III, V VI, IX, XI, XII, XIV) (Eyre et al., 2002ab; Norris et al., 2007; Raspanti et al., 2007). Additional collagen types found within type I or II fibrils are either interacting at the surface of these fibrils as it is the case for collagen III (type I fibrils) and FACITs (type I and II fibrils) or embedded into the fibril with their N-terminal NC protruding, hence regulating the fibrillogenesis process (types V and XI) (Marchant et al., 1996; Blaschke et al., 2000; Birk, 2001; Wess, 2005).

Nonfibrillar collagens are involved in forming flexible networks as, for example, type IV collagen in BMs (Fig. 5). The flexibility of these networks is conferred by the presence of a number of interruptions of the triple helical pattern, which also promote self-association and interaction with ligands (Risteli et al., 1980; Prockop & Kivirikko, 1995; Bella et al., 2006; Hwang et al., 2010). Constitution of the collagen type IV network starts by the formation of dimers resulting in a head-to-head association of two protomers by their C-terminal NC and stabilization by covalent bonds (Sundaramoorthy et al., 2002; Vanacore et al., 2004, 2005, 2009). In a second phase, dimers interact through their 7S domain (N-terminal NC + N-terminal part of the triple helix) to form tetramers, which eventually form lateral interactions with each other to constitute the core of the collagen IV network (Yurchenco & Furthmayr, 1984; Yurchenco & Patton, 2009) (Fig. 5). Collagen VIII and X are separately involved in the constitution of hexagonal lattice layers in Descemet's membrane, a specialized layer between endothelial cells and the stroma, and in cartilage growth plate, respectively (Sawada et al., 1990; Kwan et al., 1991; Khoshnoodi et al., 2006). Type VI collagen forms tetramers resulting in a more complex association to create ‘beaded filaments’ that are associated with various connective tissues including type II fibrils in cartilage and type I fibrils in tendons (Baldock et al., 2003; Knupp & Squire, 2005; Guilak et al., 2006; Izu et al., 2011). Among the variety of structures collagens form, four members of the collagen family are found at the surface of cells as transmembrane proteins. All of them are present in a similar molecular organization; an intracellular N-terminal NC domain, a transmembrane domain, a collagenous domain interspersed by several interruption of varying length and a C-terminal NC domain located outside of the cells (Franzke et al., 2005).

Functions and distribution

The collagen family of proteins is present in all connective tissues of the human body (Table S2) bearing a crucial importance in many physiological processes including embryonic development where they play a role in organogenesis or in the control of angiogenesis. Nevertheless, for years, collagens were seen as a mere scaffold for the body, maintaining the various tissues and organs glued together (hence the name collagen from Kolla, glue in Greek). This vision was reinforced by the presence of tremendous quantities of collagens in tissues such as the skin, the tendons, ligaments or the bones in which collagen I accounts for 90% of the organic mass (Table S2). The major organic component of the human body, the collagen type I, is found in association with fibril-forming collagens type III and FACIT collagen XIV in the dermis and types V and XII in bone and cornea as the core of fibrils that confer resistance and elasticity to these tissues (Wess, 2005; Smith & Birk, 2012). FACITs are minor collagen in abundance associated with the main fibrillar collagens which act as regulators of fibrillogenesis (Ansorge et al., 2009; Gordon & Hahn, 2010). Several FACITs as well as collagen VI NC regions comprise one or several von Willebrand Factor type A domains (vWA), suggesting their participation in protein–protein interactions as part of multimolecule complexes (Whittaker & Hynes, 2002; Wess, 2005). All epithelia of the human body are anchored to the underlying tissues by interacting with the BM, a structure in which collagen IV accounts for a large part. Collagen IV forms three types of networks; expression of α3α4α5 and α5α5α6 is tissue-restricted (Table S2), whereas α1α1α2 is the ‘classic’ association found in all the other BMs (Mundel & Kalluri, 2007; Parkin et al., 2010).

To fulfil their role as an interface between the cells and the ECM, collagens bind to a plethora of receptors: integrins, discoidin domain containing receptors (DDR) 1 and 2, GP VI receptor, the leucocyte-associated immunoglobulin-like receptor-1 (LAIR-1) and a few mannose receptors. The four integrins (α1β1, α2β1, α10β1 and α11β1), that are ‘natural’ collagen receptors (Leitinger & Hohenester, 2007), possess the characteristic I domain responsible for collagen binding (McCall-Culbreath & Zutter, 2008). Integrin binding to collagen isotypes is specific; α1β1 and α10β1 bind preferentially network-forming collagens (IV and VI) and α2β1 (Fig. 3) and α11β1 fibril-forming collagens (Kapyla et al., 2000; Tulla et al., 2001; Zhang et al., 2003; White et al., 2004). Integrins recognize variants of a GXX GER motif located within the triple helix of various collagen types (Knight et al., 2000; Xu et al., 2000; Zhang et al., 2003; Raynal et al., 2006; Heino et al., 2009). Even though they are not considered as natural collagen receptors, integrins αvβ3 and αvβ5 are able to bind collagen type IV through RGD and non-RGD motifs and to collagen VI via an epitope localized in the triple helix (Pfaff et al., 1993; Petitclerc et al., 2000; Pedchenko et al., 2004). DDR1 and DDR2 belong to the receptor tyrosine kinase family and bind to nonfibril-forming collagen IV (DDR1) (Vogel et al., 1997; Xu et al., 2011) and to fibril-forming collagens I, II, III, V and to collagen X (DDR1 and 2) with DDR2 as the preferred receptor for collagens II and × (Vogel et al., 1997; Leitinger et al., 2004; Leitinger & Kwan, 2006). Binding of DDRs to collagens is suggested to be dependent on the structure, particularly on the conformation of the triple helix (Vogel et al., 1997). Using collagens II and III as templates, the peptide GVMGFO was determined as the main binding motif for the DDRs along with several other sequences for DDR2 (Konitsiotis et al., 2008; Xu et al., 2011). Two members of the immunoglobulin superfamily (IgSF) serve as receptor for collagens: GP VI receptor (GP VI) and the Leucocyte-associated immunoglobulin-like receptor-1 (LAIR-1). Along with α2β1 integrin, GP VI is responsible for direct binding of platelets with subendothelial collagens, an important mechanism for thrombus formation after the damage of the vasculature. Dimeric GP VI interacts mainly with fibrils of collagen III and more weakly to collagens I and II (Moroi & Jung, 2004; Jung et al., 2008) through a peptide of 30 amino acids (designated III30) which contains several hydroxyprolines (Jarvis et al., 2008). LAIR-1 binds in vitro to fibril-forming collagen (I and III) and to three of the membrane-associated collagens (i.e. XIII, XVII and XXIII) by regions located in the helical domain containing a high content in GPO triplets (Lebbink et al., 2006, 2009). The last direct collagen receptors known so far belong to the mannose receptor family, that is, the mannose receptor itself, PLA2R and Endo180 (or uPARAP). In particular, Endo180 is involved in binding and internalization of collagens, allowing the cells expressing these receptors to migrate on collagen matrices (East et al., 2003; Engelholm et al., 2003; Wienke et al., 2003). It is notable that some interactions between cell-associated receptors and collagens occur indirectly as, for instance, in the case of the platelet receptor GP I that interacts with collagens by the intermediate of von Willebrand Factor (vWF) (Leitinger & Hohenester, 2007). Finally, it is important to mention that some collagens (XIII, XVII, XXIII and XXV) act themselves as membrane receptors to components of the BM like Ln or collagen IV (Qiao et al., 2009; Van den Bergh et al., 2011).

The role of collagens as an interface is broader than cell–ECM interactions. Result in analysis of collagens I and IV interactomes has revealed a high number of recognition sites for protein–protein interactions (Sweeney et al., 2008; Parkin et al., 2010). Such a pivotal role in tissue physiology is emphasized by the disorders resulting from mutations in collagen genes or defects in the assembly process of the mature molecules. The stability of the triple helix is critical, and its disruption, because of defects in the hydroxylation mechanism of proline or in mutation of a single glycine residue, etc., results in various diseases (Kuivaniemi et al., 1991; Prockop & Kivirikko, 1995; Myllyharju & Kivirikko, 2004). Substitution of a glycine for any other amino acid in the ColI gene results in an unstable triple helix, which is the basis for osteogenesis imperfecta, a pathology characterized by brisky bones (Marini et al., 2007; Basel & Steiner, 2009). Along collagen I, defects in collagens II, IX and XI, all components of the adult cartilage, provoke defective skeletogenesis particularly at the embryonic stage (Vikkula et al., 1995; Aszodi et al., 2000; Eyre, 2002). Abnormal network-forming collagen proteins also lead to some disease states. Collagen IV as a major component of the BM is involved in numerous pathologies because of compromised chain interactions and assembly, nonsense or missense mutations or recognition by endogenous antibodies as nonself (Khoshnoodi et al., 2006). These defects lead to diseases such as the Alport syndrome, Goodpasture disease, hereditary angiopathy, nephropathy, aneurysms, and muscle cramps (HANAC) syndrome and disruption of cell–matrix interactions to name a few (Hudson et al., 2003; Parkin et al., 2010; Van Agtmael & Bruckner-Tuderman, 2010). Mutations in collagen VI genes result in structural defects like early termination of transcription and other molecular shortcomings and provoke, for example, the muscular diseases Bethlem myopathy and Ullrich congenital muscular dystrophy (Lampe & Bushby, 2005). As mentioned before, interactions of collagens with their receptors are an important mechanism for maintaining interactions between tissues and the ECM. Consequently, the loss or mutation of a receptor provokes abnormalities and pathologies, although it has been observed that experimental knockout of these receptors does not lead to lethality, suggesting that the redundancy of collagen receptors prevents a fatal issue (Hulmes, 2002; Leitinger & Hohenester, 2007; Heino et al., 2009). Impaired DDR1 function leads to failing mammary gland development (Vogel et al., 2001), inner ear function (Meyer zum Gottesberge et al., 2008) or perturbation in kidney organogenesis (Gross et al., 2004), whereas an altered function of DDR2 leads, in particular, to perturbations in the development of mineralized tissues (Labrador et al., 2001; Kano et al., 2008).

In the past 10 years, the ability of collagen-derived peptides, also known as matrikines or matricryptines, to influence and control angiogenesis processes was brought to light (Bix & Iozzo, 2005; Mundel & Kalluri, 2007; Sudhakar & Boosani, 2008). This breakthrough about a role of matrix-derived peptides was first evidenced when the C-terminal NC of collagen type XVIII, also known as endostatin, was shown to inhibit tumour growth under certain circumstances (O'Reilly et al., 1997). The proteolysis-derived C-terminal ends of these molecules, named restin (collagen XV) and endostatin (collagen XVIII), exhibit antiangiogenic properties leading to a reduced tumour growth (Marneros & Olsen, 2005; Iozzo et al., 2009). Subsequent studies demonstrated the ability of endogenously generated fragments to regulate wound healing and neovascularization (Seppinen et al., 2008). NC domains of collagen IV particularly of chains α2 and α3 also display the capacity in vitro and in vivo to regulate angiogenesis (Pasco et al., 2005; Tran et al., 2005).

The BM is breached by pathogen-bound plasminogen (Plg) and host proteases that degrade Ln and collagen

Plg is synthesized in the liver and released into the blood circulation as a zymogen. It is converted into the active plasmin protease by tissue-type Plg activator (tPA) and urokinase Plg activator (uPA) (Lahteenmaki et al., 2001ab). The main function of plasmin is to degrade fibrin (fibrinolysis), which is importantly involved in various homoeostatic processes including blood coagulation, cell migration as well as tissue and wound repair. Furthermore, plasmin activates procollagenase into collagenase that in turn degrades collagens and also activates certain complement mediators (Fig. 6). Being a wide-range protease, plasmin also degrades several ECM proteins including fibronectin (Fn), Ln and thrombospondin. To prevent excess proteolysis and establish a balanced homoeostasis, the activity of Plg is controlled by an array of regulators/ inhibitors such as Plg activator inhibitor (PAI) 1 and 2, α2-antiplasmin and α2-macroglobulin (Parry et al., 2000; Lahteenmaki et al., 2001ab, 2005). The main regulator of PAI-1 is vitronectin (Vn), an abundant serum protein. It binds directly to PAI-1, resulting in the PAI-1/Vn complex that inhibits the Plg activity by inactivating tPA and uPA (Mondino & Blasi, 2004). Vn increases the half-life of PAI-1 by 2- to 4-fold and thus causes prolonged inhibition of Plg activation (Haiko et al., 2010). It has been shown that PAI-1 expression is necessary to protect the host from Gram-negative pathogens by recruiting neutrophils to the alveolar compartment in a murine model (Renckens et al., 2007). Some pathogens such as Pseudomonas aeruginosa, Staphylococcus aureus and Bacillus subtilis cause degradation of PAI-1 during invasion (Beaufort et al., 2008, 2010). Recently, it was shown that proteins of the omptin family (bacterial surface proteases) including Plg activator (Pla) of Yersinia pestis, PgtE of Salmonella typhimurium and Kop of Klebsiella pneumoniae can degrade the PAI-1/Vn complex in vitro (Haiko et al., 2010). Thus, an increased production of plasmin at the site of infection will accelerate tissue damage that might be beneficial for pathogens in order to invade host tissues.

Figure 6.

Microorganisms utilize Plg and host proteases to degrade the host cellular barriers. (a) N-terminal (K78) and a C-terminal cleavage (between R560-V561) performed by plasmin and tPA/uPA, respectively, render the zymogen active. Activation of Plg into plasmin is controlled by the inhibitors PAI-1 and PAI-2 to prevent an excessive response. Additionally, there are plasmin inhibitors such as α2-antiplasmin and α2-macroglobulin that inhibit the proteolytic activity of free plasmin. (b) Streptokinase and staphylokinase are either secreted or cell surface–bound activators of Plg. These two proteins have different mechanisms of activating Plg as compared to tPA and uPA. In contrast, the Pla surface protein of Yersinia pestis activates Plg in a similar way as tPA and uPA, that is, by cleaving the peptide bond between R560-V561. (c) Plg is recruited by several bacteria at their surface and subsequently converted into active plasmin, either by host tPA/uPA or by their own Plg activators. Free plasmin is inhibited by α2-antiplasmin, whereas plasmin bound to the bacterial surface cannot be inhibited by α2-antiplasmin. Microorganisms thus use surface-bound plasmin as a tool to degrade the BM. (d) The inflammatory host response causes damage to the epithelial cell layer and the damage is enhanced by, for example, neutrophil infiltration and activation of MMPs. Free plasmin and bacterial surface-bound plasmin degrade Ln and fibronectin in the BM. Procollagenase is converted into active collagenase that in turn degrades collagen found in the BM. The BM is thus destroyed and bacterial pathogens will gain access to the connective tissue and ECM proteins, resulting in increased adhesion and invasion of host tissues and capillaries.

Invasive pathogens recruit host Plg to their surface, which is subsequently cleaved into plasmin. Pathogens then use the proteolytic activity of plasmin as a mechanism to degrade host cell barriers and ECM proteins (Fig. 6) (Korhonen et al., 1992; Lahteenmaki et al., 2001ab, 2005). Streptokinase from Streptococcus spp., staphylokinase from Saureus, Plg activator (Pla) from Y. pestis and finally Plg activator A (PauA) from Streptococcus uberis are well-known PAs (Parry et al., 2000; Lahteenmaki et al., 2001ab). Streptokinase and staphylokinase display different mechanisms of cleavage and activation of Plg, whereas Pla activates Plg by cleaving the Arg560-Val561 peptide bond similarly to tPA and uPA (Lahteenmaki et al., 2001ab). Pla also functions as an inhibitor of α2-antiplasmin, which is a major inhibitor of soluble plasmin and thus enhances uncontrolled proteolysis during inflammatory responses (Kukkonen et al., 2001). Moreover, Yersinia Pla is known to bind Ln, collagens and other ECM proteins, but does not, however, degrade these proteins. Pla functions in Ln-mediated adhesion whereas the degradation of the ECM depends on the plasmin activity (Fig. 6) (Lahteenmaki et al., 1998).

There are many other pathogens that express Plg receptors. For example, plasmin-coated Borrelia spp. penetrates human endothelial cell monolayers grown on connective tissue substrate much more efficiently as compared to uncoated bacteria (Coleman et al., 1995). Similarly, transcytosis and further invasion of S. aureus into host cells is enhanced in the presence of plasmin (Lahteenmaki et al., 2001ab). Additionally, surface-bound plasmin on several pathogens including S. typhimurium, Haemophilus influenzae and Streptococcus pneumoniae mediates enhanced penetration through reconstituted BM such as a matrigel (Lahteenmaki et al., 1995). As the field of research regarding Plg and microbial pathogenesis is rapidly growing, new Plg receptors have been unraveled in various pathogens. For example, the M-like protein PAM from streptococci groups A, C and G, fimbriae and flagella from Escherichia coli and Salmonella, aspartase from H. influenzae (Lahteenmaki et al., 2001ab), complement regulator–acquiring surface protein-1 (CRASP-1) of Borrelia burgdorferi (Hallstrom et al., 2010), PavB of S. pneumoniae (Jensch et al., 2010), skizzle protein from Streptococcus agalactiae (Wiles et al., 2010) and finally endostatin-like protein A from Leptospira interrogans (Verma et al., 2010) have been defined to bind Plg. Recently, we identified protein E (PE) of H. influenzae that acquires Plg at the surface and degrades the 3 (C3) component of the complement system to dampen the innate immune response (Barthel et al., 2012). Moreover, many fungal pathogens are known to recruit Plg at their surface, for example Candida albicans that possesses eight surface proteins that recognize Plg (Crowe et al., 2003), Aspergillus fumigatus (Zaas et al., 2008) and Cryptococcus neoformans (Stie et al., 2009).

Host matrix metalloproteinases (MMPs) are involved in tissue remodelling and repair, inducing inflammatory responses and degradation of many ECM proteins. Most of the host MMPs and their substrate specificities are shown in Table 1 (Greenlee et al., 2007). Host MMP-1, 2, 3, MMP-7, MMP-9, 10, MMP-12 and MMP-19, 20 degrade various types of collagens and Ln chains (Visse & Nagase, 2003). Metallo-, serine- and other proteases are also known to be produced by pathogens (Table 1). A number of bacterial exoproteinases can activate host MMPs by limited proteolysis in their autoregulatory domains (Okamoto et al., 2004). Expression of MMPs is often induced as an inflammatory response by the host, but some pathogens can manipulate this mechanism by suppressing the expression of selective MMPs. In contrast, some pathogens also induce MMP expression during invasion in order to degrade ECM (Kanangat et al., 2006; Salgame, 2011). For more information on the various types of MMPs and their possible roles in bacterial pathogenesis, see a comprehensive review by Vanlaere and Libert (2009).

Table 1. Host factors/enzymes that degrade and/or release Ln and collagen from the BM or ECM
Host MMPsMMP-2 (gelatinase A)Ln-5 (γ chain), collagen IV, V, VII, X, elastinPezzato et al. (2003), Pirila et al. (2003), Visse & Nagase (2003), Oku et al. (2006)
MMP-3 (stromelysin-1) MMP-10 (stromelysin-2)Ln, collagen X, XI, proteoglycans, gelatin, FnBejarano et al. (1989), Lin et al. (2009), Wilkins-Port et al. (2009)
MMP-7 (matrilysin)Ln-5, collagen, IV, X, XI, proteoglycans, gelatin, elastin, FnUitto et al. (2002), Yamamoto et al. (2004)
MMP-9 (gelatinase B)Ln, collagen IV, V, VII, X, elastinVisse & Nagase (2003), Gu et al. (2005)
MMP-12 (metalloelastase)Ln γ2, collagen I, IV, elastinFu et al. (2001), Mydel et al. (2008)
MMP-19Ln-5 (γ2), collagen IV, gelatinVisse & Nagase (2003), Sadowski et al. (2005)
MMP-20 (enamelysin)Ln-1, gelatinVaananen et al. (2001)
MMP-1 (interstitial collagenase)Collagen I, II, III, X, gelatin, proteoglycansVisse & Nagase (2003)
MMP-8 (neutrophil collagenase)Fibrillar collagen, gelatin, proteoglycansPirila et al. (2003), Visse & Nagase (2003)
MMP-13Collagen I, II, IIIVisse & Nagase (2003)
Other host factorsAlpha-thrombinLn, collagen VLiotta et al. (1981)
Cathepsin GLn-111, collagen (denatured)Capodici & Berg (1989), Heck et al. (1990)
Neutrophil elastaseLn-5, collagen I, IIIZhu et al. (2001), Mydel et al. (2008)
Plasminα-G4, γ1, collagen IVMackay et al. (1990), Tunggal et al. (2000), Skrzypiec et al. (2009)
Proteinase-3LnMydel et al. (2008)
Proteinase-3 (myeloblastin)Ln, Vn, Fn, collagen IVPezzato et al. (2003)

Alpha-enolase is a highly conserved glycolytic/gluconeogenic enzyme that is found in almost all organisms. The virulence-associated properties of enolase have been best studied in Gram-positive pathogens including S. pneumoniae, Streptococcus pyogenes, and S. aureus as well as various other commensal bacterial species (Carneiro et al., 2004; Antikainen et al., 2007). The main function of α-enolase is to convert 2-phosphoglycerate to phosphoenolpyruvate in the glycolysis and the reversed reaction in gluconeogenesis. It is primarily found in the cytoplasm, but is also present in the periplasm and associated with the bacterial outer membrane (Pancholi, 2001). Interestingly, α-enolase functions as a surface receptor of Plg, which is ultimately converted into plasmin that degrades Ln, as previously described (Bergmann et al., 2003, 2005; Antikainen et al., 2007). Enolase from C. albicans (Jong et al., 2003), Paracoccidioides brasiliensis (Nogueira et al., 2010) and finally Pneumocystis carinii (Fox & Smulian, 2001) have also been shown to bind Plg.

Proteases secreted by microorganisms degrade Ln and collagen

In order to invade the host tissue, several pathogens independently degrade Ln and collagen using endogenous proteases. A compilation of enzymes from different microorganisms that degrade Ln and collagen is shown in Table 2. In general, few bacterial proteases have been described, whereas many potential fungal and parasite proteases are commonly utilized for invasion of the host. Known bacterial Ln- and collagen-degrading proteins include elastase and alkaline proteases produced by P. aeruginosa (Heck et al., 1986ab, 1990; Bejarano et al., 1989; de Bentzmann et al., 2000; Schmidtchen et al., 2003). Both enzymes cleave the α and β chains of soluble Ln that results in different cleavage products. Moreover, during tissue invasion by bacteria that causes necrosis, immunoreactive Ln is released from the host BM after incubation with elastase and alkaline proteases in vitro. The corneal BM has also been shown to be degraded by Pseudomonas elastase and alkaline protease (Matsumoto, 2004). Treponema denticola invades BM by degrading Lns, collagen IV and Fn with a chymotrypsin-like protease (Grenier et al., 1990). ‘Dentilisin’ is another well-known and wide-range protease of T. denticola that is known to degrade various ECM components including Ln and collagen VI (Ishihara, 2010). Some Vibrio spp. and Bacilli can also produce proteases that degrade collagen in the ECM (Table 2). Porphyromonas gingivalis, a causative agent of human periodontal disease, adheres to the gum and mucosal epithelial cells using different adhesins and fimbriae. Interestingly, P. gingivalis also utilizes a protease/adhesin complex designated as gingipains (Baba et al., 2001; Olczak et al., 2001), which directly degrades Ln and Fn and thus increases the pathogen-dependent invasion (Pike et al., 1996; Potempa et al., 2000). There are also some other proteases known in P. gingivalis that plays significant roles in the degradation of Ln, collagen and other ECM proteins during host invasion (Potempa et al., 2000).

Clostridium spp. produce glycolytic and proteolytic enzymes, including collagenases, hyaluronidases, sialidases, lipases and nucleases. Collagenases and hyaluronidases degrade various ECM components. According to sequence similarity and substrate specificity, Clostridium histolyticum collagenases can be divided into (1) class I collagenases and (2) class II collagenases encoded by ColG, ColH and ColA genes (Matsushita & Okabe, 2001). Class I (ColG) has tandem distribution of homologous collagen-binding domains (CBDs) at the C-terminus, while class II (ColH) has only one CBD. On the basis of function, bacterial collagenases can be either proteases that degrade collagen macromolecules, or peptidases that recognize digested collagen-specific sequences. Host MMPs and ColG, ColH and ColA have well-defined CBDs, while other bacterial proteases lack a CBD. The CBD sequences and structures are variable among different collagenases that reflect the wide substrate specificity of collagenases. For example, ColG of C. histolyticum can bind to any kind of collagen regardless of type and diameter of fibres (Toyoshima et al., 2001). Recently, it was shown that clostridial collagenases degrade collagen types I, II, III, IV and V (Shi et al., 2010; Pruteanu et al., 2011). There are several structures of CBDs available in the PDB database (, including MMPs and Clostridial collagenases (Matsushita & Okabe, 2001; Wilson et al., 2003). The CBD of ColG appears as a beta sandwich with 10 β strands, composing five β sheets on both sides. One side of the β sandwich is rich in tyrosines that are involved in collagen binding. Binding of a metal atom alters the protein conformation and is necessary for collagenolytic activity of ColG (Wilson et al., 2003; Sakurai et al., 2009).

Among fungal pathogens, A. fumigatus is known to produce extracellular serine proteases that degrade several ECM proteins (Table 2). A protease extracted from A. fumigatus shows a proteolytic activity against major ECM components including fibronectin, lung collagens and elastin (Monod et al., 1991; Tronchin et al., 1993; Iadarola et al., 1998). Another secretory aspartic proteinase (aspergillopepsin F), a 39-kDa enzyme of A. fumigatus, has been shown to degrade ECM proteins including Ln, elastin and collagen. Homologous proteinases to aspergillopepsin F, are also produced by other species of Aspergillus. Interestingly, aspergillopepsin F was secreted actively by the invading germ tube hyphae during the invasion of mouse lung tissues (Lee & Kolattukudy, 1995). Other fungal pathogens such as C. neoformans (Salkowski & Balish, 1991; Rodrigues et al., 2003), P. brasiliensis (Puccia et al., 1998), Trichophyton schoenleinii (Ibrahim-Granet et al., 1996) and Candida spp. (Dos Santos & Soares, 2005; Parnanen et al., 2009; Portela et al., 2010) also produce extracellular proteases that can degrade various host ECM proteins. Thus, ECM degradation by these fungal proteases is a common phenomenon during host invasion (Table 2).

Several human disease-causing protozoans have been studied for proteases that are secreted during invasion of the host as a result of increasing virulence and subsequent invasion. For example, in Acanthamoeba castellanii, which causes keratitis and damage of corneal tissues, a 42-kDa protease (Cho et al., 2000) and a small 12-kDa serine protease (Na et al., 2001) have been identified. These proteases are highly active against a wide range of ECM proteins and can, in particular, degrade collagen I and, to a lesser extent, Ln, Fn, as well as non-ECM proteins such as fibrinogen, sIgA, IgG, Plg and haemoglobin. Interestingly, Na and co-workers (Na et al., 2001) showed that injection of a 12-kDa serine protease into cornea causes a direct damage and lesions that can be inhibited by the serine protease inhibitor phenylmethanesulfonyl fluoride (PMSF). Moreover, degradation of IgA and IgG suppresses immune response and protects the pathogen from host immunity. Similarly, fatal encephalitis causing pathogens Acanthamoeba healyi secretes 33- and 130-kDa serine proteases and a 150-kDa metalloprotease (Kong et al., 2000; Sissons et al., 2006), and Balamuthia mandrillaris releases a 50-kDa serine protease that has a high proteolytic activity against collagens I, III and IV (found in brain tissues), elastin, Fn, Plg, IgG, IgA, albumin, fibrinogen and haemoglobin. Interestingly, these proteases showed a direct cytotoxicity against human brain microvascular endothelial cells (HBMEC) monolayers and disintegrated them, supporting the model of breaching of the blood–brain barrier (BBB) during invasion by these pathogens (Matin et al., 2006).

Entamoeba histolytica trophozoits secrete proteases actively during invasion of host bowel epithelial cells near lesions (Keene et al., 1986). Subsequently, a 27-kDa and a 29-kDa cysteine proteases were purified from E. histolytica that have a wide range of proteolytic activities against ECM proteins including BM and ECM components such as collagens I, IV, V, elastin, Fn and Ln in vitro (Schulte & Scholze, 1989; Li et al., 1995). More recently, Que and Reed (2000) reviewed the functional and physiological characteristics of cysteine proteases of E. histolytica, suggesting that there are multiple cysteine proteases in this organism. These proteases have important roles in (1) degradation of mucus and debris overlying the intestinal mucosa; (2) degradation of ECM proteins during penetration into the host tissues; (3) degradation of immunoglobulins (IgA and IgG) to circumvent the immune response; (4) protection from complement-mediated lysis; and (5) dissemination and production of metastatic lesions by degrading ECM, mucoproteins and BM, ultimately dislodging epithelial cells for increasing colonization. Thus, inhibition of cysteine protease activity during E. histolytica pathogenesis was suggested to be an attractive target for the treatment of this infection (Que & Reed, 2000). Many other protozoal pathogens express proteases at their surface or secrete them during pathogenesis as outlined in Table 2, and further described in the next chapters.

Ln and collagen-binding proteins help microorganisms attach to the respiratory tract

The respiratory tract is composed of a multitude of connective tissues, supporting the structure of the lungs, which is composed of ECM molecules like Ln and various types of collagen (Tables S1 and S2). The major collagen found in the respiratory tract is collagen type I, which is ubiquitously distributed in the interstitial tissue of the lungs (Gil & Martinez-Hernandez, 1984). Several other collagen types are found including type II in the cartilage, type III in the interstitial tissue (and among others) types IV and VI in the BM (Table S2) (Amenta et al., 1988; Davidson, 1990). Various bacterial species utilize the ECM for adherence to surfaces of epithelial cells or to the BM after injury. It is a well-known fact that respiratory tract pathogens, for example S. pneumoniae, Moraxella catarrhalis, H. influenzae and P. aeruginosa, are commonly found in patients suffering from chronic obstructive pulmonay disease (COPD) (Sethi & Murphy, 2000). COPD is in many cases a consequence of heavy smoking, and it has been demonstrated that the Ln layer in the BM of lung epithelial cells is significantly thicker in smokers than in nonsmokers (Amin et al., 2003). Besides, a major augmentation of total collagen deposition, predominantly collagens I and III, in the bronchial airways was observed in patients suffering from COPD (Kranenburg et al., 2006). The increased Ln and collagen concentrations in the lung promote the adherence of ECM-binding respiratory tract pathogens. This higher concentration of ECM molecules is one of the possible explanations for an increased incidence of bacterial infections in patients with COPD.

Mycobacterium smegmatis Ln-binding protein (MS-LBP) is involved in adhesion to host cells. Interestingly, MS-LBP is also conserved in other Mycobacterium spp. (Table 3) (Pethe et al., 2001, 2002; Esposito et al., 2008). Moreover, M. tuberculosis has evolved an adhesive region in its housekeeping enzyme malate synthase (MS) resulting in binding of Ln and Fn by its C-terminal end. The experimental heterologous expression of MS of M. tuberculosis in M. smegmatis showed an increased adhesion of M. smegmatis to epithelial cells (Kinhikar et al., 2006). Finally, the pilin protein of M. tuberculosis, which constitutes the pilus and is produced during infection, binds to ECM proteins, and among those, the interaction with Ln is the strongest. Isogenic mutants of pilin were found to be defective in Ln binding (Alteri et al., 2007).

Table 2. Pathogen factors/enzymes that degrade and/or release Ln and collagen from the BM or ECM
Actinobacillusactinomycetemcomitans Serine protease, metalloproteaseCollagen I, FnWang et al. (1999)
Bacillus cereus 105-, 87-kDa metalloproteasesCollagen ILund & Granum (1999)
Clostridium histolyticum Col G, ColHCollagens I, II, III, IV, V, VIWatanabe (2004), Shi et al. (2010)
Clostridium perfringens ColACollagen I, IVWatanabe (2004), Pruteanu et al. (2011)
Porphyromonas gingivalis Gingipains, thiol protease, and serine proteaseLn, collagenPike et al. (1996), Potempa et al. (2000), Watanabe (2004)
Pseudomonas aeruginosa Elastase, alkaline proteaseLn, collagens III, IV, mucin Heck et al. (1986a, b), Bejarano et al. (1989), de Bentzmann et al. (2000), Schmidtchen et al. (2003), Matsumoto (2004)
Streptococcus gordonii Serine proteaseCollagen IVJuarez & Stinson (1999)
Treponema denticola DentilisinLn, collagen VI, IV, FnIshihara (2010), Grenier et al. (1990)
Vibrio alginolyticus 82-kDa metalloprotease Takeuchi et al. (1992)
Vibrio vulnificus 45-kDa metalloprotease Miyoshi et al. (1998)
Vibrio parahaemolyticus MetalloproteaseCollagen IKim et al. (2002)
Aspergillus fumigatus Serine proteases, aspergillopepsin FLn, Fn, elastin, collagen I and IIITronchin et al. (1993), Lee & Kolattukudy (1995), Iadarola et al. (1998)
Cryptococcus neoformans 75-kDa extracellular serine protease, GelatinaseLn, collagen, collagen IVSalkowski and Balish (1991), Rodrigues et al. (2003)
C. albicans 95-kDa cell wall metallopeptidase, Secreted aspartate proteinases, acid proteinaseLn, collagens I, IV, FnMorschhauser et al. (1997), Watts et al. (1998), Rodier et al. (1999)
Candida spp.Metalloproteinases, 50- and 62-kDa extracellular serine proteases, Candida acid proteinase (CAP)Ln, Ln-332, Fn, collagen, keratin, haemoglobin, casein, albuminRay & Payne (1990), Dos Santos & Soares (2005), Parnanen et al. (2008), Portela et al. (2010)
Paracoccidioides brasiliensis Serine–thiol proteinaseLn, collagen IV, Fn, proteoglycansPuccia et al. (1998)
Trichophyton schoenleinii 82-kDa metalloproteinaseCollagen IIbrahim-Granet et al. (1996)
Acanthamoeba castellanii Extracellular serine proteinaseCollagen, Fn, sIgA, IgG, Plg, fibrinogen, haemoglobinCho et al. (2000), Na et al. (2001)
Acanthamoeba healyi Serine proteinase and metalloproteaseCollagens I, III, IV, Fn, fibrinogen, IgG, IgA, albumin, haemoglobin, elastin, PlgKong et al. (2000), Sissons et al. (2006)
Balamuthia mandrillaris 40- to 50-kDa extracellular metalloproteasesCollagens I and III, elastin, Plg,Matin et al. (2006)
Entamoeba histolytica Serine/cysteine proteasesLn, collagens I, IV, V, FnKeene et al. (1986), Schulte & Scholze (1989), Rosales-Encina et al. (1992), Li et al. (1995), Que & Reed (2000)
Giardia duodenalis Collagende Carvalho et al. (2008)
Leishmania spp.Serine protease and surface metalloproteasesCollagen, Fn haemoglobin, bovine serum albumin, ovalbumin, fibrinogenMcGwire et al. (2003), Silva-Lopez et al. (2005)
Trichomonas vaginalis Cysteine proteasesCollagen IV, Fn, haemoglobinAlvarez-Sanchez et al. (2000), Mendoza-Lopez et al. (2000)
Trypanosoma spp.Cysteine protease, prolyl oligopeptidase (POP Tb)Collagen I, FnGarcia et al. (2003), Bastos et al. (2005, 2010)

Many Streptococcus spp. bind Lns and utilize this interaction for adhesion to host tissues (Table 3). Streptococcus pyogenes (or group A Streptococcus; GAS) is a common human pathogen causing infections in the tonsils, skin and occasionally lungs. During colonization and deep tissue penetration, Ln binding is a crucial step for adhesion of GAS. The GAS Ln-binding protein (Lbp) has been characterized, and its structure has been solved. Isogenic mutants of lbp are defective in adhesion to and colonization of the host cells (Terao et al., 2002; Linke et al., 2009). Lmb is another Ln-binding protein found in GAS that has been suggested as a possible future vaccine candidate because high antibody titres were found in patients priorly infected with GAS (Wahid et al., 2005). Lmb of S. agalactiae has been crystallized and the homologue (Lbp) from S. pyogenes was also resolved (Linke et al., 2009; Ragunathan et al., 2009ab). These proteins belong to the ABC-transporter family, bacterial metal-binding receptors (MBRs) of solute-binding protein family 9 (SBP_Bac9) that is described in detail ( The primary function of this family of proteins is to transport metal ions such as Fe++ and Zn++ that are required for various functions to maintain homoeostasis (Jones & George, 2004). Moreover, proteins of this family frequently function as adhesion or virulence determinants in bacterial pathogenicity. In Gram-positive bacteria, the solute-binding motif is a membrane-anchored protein, whereas in Gram-negative bacteria, it might be in the periplasm or partially exposed to membranes (Rees et al., 2009). The structure of MBRs is homologous and consists of two separate globular domains. Both globular domains are connected by a helix, and the metal-binding site exists between both the globular domains. The metal-binding active site has a very conserved amino acids arrangement that consists of two or three histidines and two or three glutamate/aspartate residues that coordinate with a central metal atom. Similar proteins also exist in, for example, Treponema pallidum and S. pneumoniae (Linke et al., 2009). The metal-binding mechanisms have been described in the SBP_Bac9 family of proteins. However, the precise mechanism behind the interactions of Ln with Lmb and Lbp at the molecular level is not yet defined.

ShrA is a secretory protein involved in GAS haemolytic functions and is also known to bind Ln and thus contributes to host cell adhesion and the infection process (Fisher et al., 2008). A recent study showed that streptococcal collagen-like protein-1 (Scl1) binds to Ln and Fn and that this interaction promotes virulence as well as bacterial internalization of host cells (Caswell et al., 2010). A lipoprotein of GAS (Lsp) also binds to Ln, and an lmb, lsp double mutant was defective in Ln binding and hence unable to colonize host cells. That study also suggested that both genes are under the transcriptional control of global transcriptional regulators such as Mga (Elsner et al., 2002). Furthermore, the global transcriptional regulator (Mga), which controls the outer membrane protein (OMP) expression and virulence of S. pyogenes, has been found to be involved in regulating proteins that bind to several ECM proteins including Ln. Another proof of evidence was that mga isogenic mutants are defective in adhesion and Ln binding (Fiedler et al., 2010). Interestingly, in a recent study, it was shown that S. pneumoniae protein pilus 1 (RgrA) plays a significant role in adherence to host cells and the ECM proteins Ln and Fn (Moschioni et al., 2010).

The pathogen S. agalactiae, or group B Streptococcus (GBS), causes neonatal sepsis, and it has been shown that all clinical GBS isolates express the Ln-binding protein Lmb in addition to the Fn-binding protein Scpb. The Ln-binding capacity of GBS is regulated by a scpb-lmb intergenic region that has a hotspot for the integration of two mobile genetic elements named IS1548 and group II intron (GBSi1) (Al Safadi et al., 2010). Streptococcus pyogenes adheres at least to collagens I, IV and VI that all are widely distributed which explain its ability to interact with ECM as well as its successful capacity to invade the host. Interaction of S. pyogenes with the fibril-forming collagen I occurs through a protein located in the FCT locus that encodes the GAS pili, and the collagen-binding protein of group A streptococci (Cpa) (Podbielski et al., 1999; Bessen & Kalia, 2002). A specific domain of Cpa is involved in the interaction with collagen I, but importantly this fragment does not exhibit any affinity for the other collagens tested (Kreikemeyer et al., 2005). Cpa is located at the tip of the pilus of S. pyogenes but its exact role in the infection process remains elusive because experimental mutations of Cpa did not show significant difference in terms of virulence (Kreikemeyer et al., 2005; Quigley et al., 2009). After breaching the epithelial cell wall, bacteria must interact with the underlying BM components, among others collagen type IV. Clinical isolates of S. pyogenes from patients suffering acute glomerulonephritis and uncomplicated infection of the respiratory tract were shown to be able to bind collagen IV (Kostrzynska et al., 1989). This interaction was proven to occur in an indirect fashion through fibronectin and the S. pyogenes fibronectin-binding protein I (SfbI) (Dinkla et al., 2003). The same mechanism also mediates the interaction between S. pyogenes and collagen I. Recent studies reported that the M proteins, which are virulence factors displayed at the surface of GAS, were involved in binding different collagen types (Nobbs et al., 2009). Thus, the M protein of S. pyogenes (M3 serotype) was reported to bind collagen IV through the PARF domain (peptide associated with rheumatic fever) (Dinkla et al., 2007; Dinkla et al., 2009). In addition, collagen VI was reported to be the target of M1. It was therefore hypothesized that this interaction favours the invasion of the respiratory tract by S. pyogenes (Bober et al., 2010). It is likely that some mechanisms of invasion of streptococcal species are conserved as S. pneumoniae also binds collagen type VI (Bober et al., 2010). Another illustration of a conserved ECM–bacteria adhesion process in Streptococcus spp. implies RrgA, a pili protein containing MSCRAMM (Microbial Surface Components Recognizing Adhesive Matrix Molecules) motifs and domains homologue to the vWA, which interacts with collagen type I (Hilleringmann et al., 2008). These comprehensive studies proved the importance of Ln/collagen binding in the pathogenesis of Streptococcus spp. (Table 3).

Another example of an ECM-binding species is the respiratory tract pathogen M. catarrhalis (Table 3) that efficiently attaches to Ln via its ubiquitous surface proteins (Usp) A1 and A2. UspAs are surface-exposed, multifunctional trimeric autotransporter adhesins (TAA) involved in serum resistance and adhesion to epithelial cells, both directly and via ECM proteins (Singh et al., 2010ab; Tan et al., 2005, 2006). The appearance of these autotransporters resembles fibrils at the bacterial surface with a C-terminal conserved transmembrane domain and a long ‘stalk’ that connects to the N-terminal ‘head’, thus forming a ‘lollipop-like’ structure (see section ‘Gastrointestinal pathogens and their interactions with Ln and collagen’). The N-terminal head region of the UspAs interacts with Ln, and the stalk with Fn, thus contributing to adhesion (Tan et al., 2006). Another Moraxella autotransporter, MID also called Hag, is a multifunctional protein involved in adherence to respiratory cells, binding of IgD and mediating agglutination (Perez Vidakovics & Riesbeck, 2009). This adhesin appears to bind collagen type IV, through a specific domain distinct of the ones involved in cell adhesion (Bullard et al., 2007). More bacterial TAA, similar to the UspA family, are described in detail in section ‘Gastrointestinal pathogens and their interactions with Ln and collagen’.

Haemophilus adhesion penetration protein (Hap), which also belongs to the bacterial autotransporter family, is known to bind several ECM proteins including Ln and collagens (Table 3) (Bresser et al., 2000). The Hap adhesive domain is responsible for binding to Ln, and this interaction contributes to adhesion to epithelial cells (Virkola et al., 1996; Fink et al., 2002). Recently, we found that H. influenzae binds to soluble as well as immobilized Ln via another adhesin, PE (Singh et al., 2010c; Hallstrom et al., 2011). Haemophilus influenzae isogenic mutants lacking hap and pe have a similarly reduced binding to Ln, and a double (pe/hap) mutant exhibited an even greater reduced Ln binding. We have thus suggested that both PE and Hap contribute to the Ln-binding property of H. influenzae. The N-terminal domain of PE (amino acids PE 41–68) binds Ln, and more precisely, the heparin-binding domain of Ln (LG4) interacts with PE (Fig. 7a). Interestingly, PE functions as a multifunctional adhesin and binds vitronectin and Ln at the same time. The binding affinity of PE to Ln was measured to K D ≈1.0 μM (Hallstrom et al., 2011). Several H. influenzae clinical strains were shown to bind collagen I and more weakly collagen IV (Bresser et al., 2000). Adhesion to the latter is mediated by the Hap autotransporter, whereas Hap does not bind to collagen II in vitro proving certain specificity in recognition and binding (Fink et al., 2002). The epitope for the Hap-dependent interaction with collagen I was localized in the C-terminal of the passenger domain, and the results suggested the implication of two distinct subdomains (Fink et al., 2003).

Figure 7.

Models of Ln-binding bacterial receptors. (a) Transmission electron microscopy (negative staining) showing binding of gold-labelled PE of Haemophilus influenzae to the C-terminal globular domains (LG1–5) of the Ln-111 molecule. (b) Mycobacterium leprae binds to Ln α2 present in Schwann cells. The phenolic glycolipid-1 (PGL-1) binds to LG1, 4 and 5 domains, and a 21-kDa Ln-binding protein binds to the LG5 module of Ln α2. At the same time, Ln α2 binds to DG surface receptors of Schwann cells and thus cross-links M. leprae to host.

ECM interactions are not limited to bacteria but exist in all microorganisms including fungi. Aspergillus fumigatus is an opportunistic fungal pathogen and the main causative agent of human aspergillosis, which predominantly affects immunocompromised patients (Gil et al., 1996; Upadhyay et al., 2009). To initiate colonization and subsequent infection, A. fumigatus conidia bind to Ln and collagen (Gil et al., 1996). Several surface structures involved in these interactions have consequently been identified and characterized (Table 3). A 37-kDa surface protein and a 72-kDa surface glycoprotein (GP) were identified as conidial Ln binders in vitro (Gil et al., 1996; Tronchin et al., 1997). In addition, a major allergen of A. fumigatus, Asp f2, was also found to specifically bind Ln (Banerjee et al., 1998). More recently, an in silico method was used to predict novel possible A. fumigatus adhesins, and one surface protein, AfCalAp, was identified as a novel Ln-binding protein that could be detected at the conidial surface (Upadhyay et al., 2009). AfCalAp furthermore displayed affinity for murine pneumocytes. The Ln isotypes and/or region(s) that interact with these various proteins have not yet been elucidated. A. fumigatus also expresses a variety of proteases that have been shown to degrade ECM components, leading to possible dissemination. Aspergillopepsin F is an aspartic proteinase that is secreted by the fungus and has been shown to degrade Ln, collagen and elastin (Lee & Kolattukudy, 1995). It has furthermore been shown to be secreted in vivo in the lungs of immunocompromised mice, further indicating an active role of the protease in the course of infection. Other proteases, such as A. fumigatus serine proteases, have been identified to degrade Ln, type I and III collagens, elastin and Fn (Tronchin et al., 1993; Iadarola et al., 1998). An extracellular collagenolytic 33-kDa alkaline protease, initially identified in A. fumigatus, was found to have orthologs in the pathogenic species Aspergillus flavus, Aspergillus terreus and Aspergillus nidulans but not in the nonpathogenic Aspergillus glaucus, Aspergillus versicolor and Aspergillus clavatus (Monod et al., 1991; Hanzi et al., 1993). These findings collectively point towards an important role of proteases in the pathogenicity of aspergillosis.

Coccidioides immitis and P. brasiliensis are fungal pathogens that cause pulmonary infections in humans known as coccidioidomycosis (or ‘valley fever’, because of the organism's natural desert-soil habitat) and paracoccidioidomycosis, respectively (Hung et al., 2002). These pathogens are known to bind ECM proteins during pathogenesis (Table 3). Coccidioides immitis and P. brasiliensis are at times known to disseminate from the primary site of infection, the lungs, and cause systemic infections. In C. immitis, several extracellular serine proteases and metalloproteases have been implicated in the dissemination mechanism and, when isolated, been shown to degrade collagen and elastin, as well as secretory IgA and IgG (Resnick et al., 1987; Yuan & Cole, 1987). Moreover, an 82-kDa adhesin of C. immitis, SOWgp82, has been identified as a parasitic phase–specific (as opposed to saprobic phase) Ln-, type IV collagen- and Fn-binding adhesin (Hung et al., 2002). Deletion of the SOWgp gene leads to marked reduction in virulence in an experimental murine model. Paracoccidioides brasiliensis also produces extracellular proteases that potentially promote tissue invasion, for example a serine–thiol proteinase with ability to degrade Ln, type IV collagen, Fn and proteoglycans (Puccia et al., 1998). Furthermore, there are many reports on paracoccidioidal adhesins involved in interactions with the ECM. Cell wall–associated glyceraldehyde-3-phosphate dehydrogenase (GAPDH) binds Ln, type I collagen and Fn, and the treatment of yeast with α-GAPDH polyclonal antibodies caused the inhibition of adherence to and invasion of pneumocytes in vitro (Barbosa et al., 2006). Surface GP gp43 has been shown to bind Fn and Ln, and the interactions could be inhibited by the peptides YIGRS (Ln) and RGD (Ln and Fn), suggesting that gp43 binds to the integrin-binding region of these ECM proteins (Mendes-Giannini et al., 2006). Moreover, adhesins of P. brasiliensis include 47-kDa and 80-kDa surface components that bind type I collagen (Mendes-Giannini et al., 2006), a 30-kDa adhesin that binds Ln (Andreotti et al., 2005) and 19- and 32-kDa cell wall proteins that bind Ln, Fn and fibrinogen (Gonzalez et al., 2005).

Cryptococcus neoformans is, in analogy with C. immitis and P. brasiliensis, a primary pulmonary fungal pathogen with potential to disseminate. A 75-kDa extracellular serine proteinase secreted by the yeast has been shown to hydrolyse the ECM proteins Fn, Ln and type IV collagen, which provide clues as to how the mechanisms of dissemination function (Rodrigues et al., 2003). Histoplasma capsulatum (McMahon et al., 1995) and Pneumocystis jiroveci (carinii) (Narasimhan et al., 1994) are other respiratory tract fungal pathogens that produce Ln-binding proteins. Interestingly, the H. capsulatum 50-kDa adhesin was shown to cross-react with α-human LR antibodies, suggesting that the proteins are related and interact with the same domain in Ln.

In addition to bacteria and fungi, envelope GPs extracted from the influenza A virus have affinity for Fn and Ln (Table 3), thus potentially enhancing the viral–host tissue interaction during the course of infection (Julkunen et al., 1984). Taken together, these data suggest that attachment to, and degradation of, ECM components seems to play an important role for all classes of microorganisms, irrespective of origin and complexity.

Ln and collagen are target molecules in bacterial meningitis and other infections of the CNS

Meningitis is an inflammation of the lining that protects the brain and spinal cord (meninges; dura mater, arachnoid and pia mater). During childhood, bacterial meningitis is caused by common respiratory tract pathogens including S. pneumoniae, Neisseria meningitidis and H. influenzae type b (Hib). The mainly intracellular organisms Listeria monocytogenes and M. tuberculosis are also occasionally isolated from patients with meningitis, the latter nowadays particularly in developing countries. In addition, Mycobacterium leprae and some protozoal pathogens are also known to colonize the CNS during infection. Pathogens enter the CNS by interacting with the luminal side of the cerebral endothelium, which constitutes the BBB (Fig. 8). Crossing the BBB is thus indispensible for bacterial tropism in the CNS, which means that pathogens cannot invade the CNS without crossing the microvascular endothelial or brain epithelial cells. Invasive pathogens have many different strategies to cross the BBB (Nassif et al., 2002). The exact mechanism of how pathogens breach the BBB is not fully understood. There are three steps in the pathogenesis of meningitis: (1) breaching of the BM and basal lamina is the first step in infection and is discussed in detail in the previous section ‘Proteases secreted by microorganisms degrade Ln and collagen’. Pathogens use their own adhesins, proteases and host Plg as well as MMPs to invade the epithelial barrier and BM; (2) after degrading the BM, bacteria will attack the capillary endothelial cells, cross them, enter the blood stream and continue multiplying; (3) pathogens will breach the blood–CSF barrier (brain endothelial lining; choroid plexus) and colonize the underlying tissues. Bacteria will subsequently induce the permeability of the BBB and release of pro-inflammatory molecules, which in turn cause inflammation and tissue damage (Kim, 2003; Abbott et al., 2006). These different events will in the worst-case scenario cause a rise in intracranial pressure, consequently leading to death.

Figure 8.

Schematic overview of blood–tissue, blood–brain and blood–CSF barriers breached by pathogens. Three different barriers are confronted by invasive pathogens in order to invade host tissues. As the BM constitutes a major part of these barriers, Lns and collagen are important target proteins. (1) The tissue–blood barrier; the BM beneath the epithelial cells will be the first obstacle for pathogens after penetrating the epithelial cell layer. This crossing occurs via transcytosis or pinocytosis, which are well-defined virulence mechanisms. Pathogens have their own proteases to degrade the Lns and collagens of the BM, but also utilize host factors such as Plg and MMPs. Invasive blood microorganisms penetrate the blood capillaries that are covered by BMs, and subsequently cross the endothelial cell layer reaching the blood stream. To survive, invasive pathogens have usually developed resistance to complement-mediated killing and adopted various immune evasion strategies. (2) The BBB; the brain parenchyma is covered with a protective membrane called meninges, which consists of three different layers: outer dura mater, middle arachnoid and inner thin membrane pia mater. Meninges (all three layers) have blood microcapillaries, and thus, microorganisms in the blood stream can reach the meninges and breach the microcapillaries. (3) The blood–CSF barrier; the blood capillaries present in the choroid plexus (ventricles) are fenestrated (permeable), where the capillaries and basal layer of epithelial cells control nutrient transport between blood and CSF (Abbott et al., 2006). The CSF is produced in the choroid plexus by selective secretion of plasma. The neural pathogens are most likely to invade the brain tissues through the choroid plexus, where the blood–CSF barrier exists. The endothelial layer of blood capillaries in the brain is covered by an outer sheet (BM), followed by contact of foot processes of astrocytes that separate capillaries from the neurons. Microglial cells are resident macrophages of the brain and the spinal cord and function as a defence in the CNS. The blood–brain and blood–CSF barriers can only be penetrated by neural pathogens, for example Neisseria meningitidis, Mycobacterium tuberculosis, Streptococcus pneumoniae, L. monocytogenes, Haemophilus influenzae serotype b (Hib) and some virulent strains of Escherichia coli.

Streptococcus agalactiae (GBS) is one of the most pathogenic bacteria that causes neonatal meningitis. It has been shown in vitro that mutation in the lmb gene, which encodes the Ln-binding surface protein, causes a 69–74% decrease in host cell (HBMEC) invasion (Spellerberg et al., 1999; Tenenbaum et al., 2007; Nobbs et al., 2009). More recently, it was shown that S. agalactiae displayed a several-fold higher expression of Lmb in meningitis isolates in comparison with other isolates, whereas the expression of other bacterial ECM-binding proteins was unchanged. Thus, it was hypothesized that the Ln binding may be crucial for S. agalactiae adhesion during meningitis (Al Safadi et al., 2010).

Mycobacterium leprae, a causative agent of leprosy, binds to Ln and uses this interaction for adhesion (Table 3). Nerve cells (Schwann cells) constitute the principle target for M. leprae, where bacteria survive for an extended period of time and disseminate to other body tissues. This kind of specific relationship between Schwann cells and M. leprae is an example of neural tropism. Schwann cells are covered with BM composed of Lns, collagen IV, nidogens and proteoglycans (Barker, 2006). Ln-2 (α2, β1, γ1 chains) is the most common form present on Schwann cells, as well as in the placenta and striated muscle cells; hence, these tissues constitute the natural niches for M. leprae (Barker, 2006). Ln is anchored to the Schwann cells by the Ln- receptor α-DGs (Rambukkana, 2001). In a Schwann cell infection model, Ln-α2 globular domains have been shown to bind to M. leprae (Rambukkana et al., 1997; de Melo Marques et al., 2000; Marques et al., 2001; Ottenhoff et al., 2005). The M. leprae Ln-binding protein has been identified as a 21-kDa histone-like protein (Hlp), a highly conserved cationic protein also present in other species of mycobacteria (de Melo Marques et al., 2000). Hlp also interacts with heparin-like Glycosaminoglycans (GAGs). The binding domain of Hlp lies in its C-terminal end and is involved in binding to Ln as well as to host cells (Soares de Lima et al., 2005). Moreover, the cell wall phenolic glycolipid-1 (PGL-1) of M. leprae is involved in attachment to Schwann cells via the Ln molecule. The globular domains G1, G4 and G5 of the Ln α2 chain directly bind to PGL-1 molecules in vitro (Ng et al., 2000). A putative Ln-binding model of M. leprae has been presented by Rambukkana (2001) (Fig. 7b). It suggests that the M. leprae cell surface contains the Ln-binding phenolic glycolipid-1 (PGL-1) that recognizes domains LG1, 4 and 5. Additionally, a 21-kDa Ln-binding protein of M. leprae binds to the LG5 module of Ln α2. Simultaneously, Ln α2 binds to DG surface receptors of Schwann cells, and consequently, Ln cross-links M. leprae to Schwann cells (Rambukkana, 2001). Thus, these above studies indicate a major role of Ln-α2 in M. leprae attachment to nervous tissue (Table 3), survival and pathogenesis.

Other meningitis-causing pathogens such as L. monocytogenes and M. tuberculosis also bind Ln directly at their surface, but their specific Ln-binding proteins (proteins that interact with nervous tissues) are presently unknown. Interestingly, these pathogens do not cause only meningitis but have their original niches in other tissues. Therefore, most of the Ln-binding studies have been performed on their natural habitats, such as the lung and the intestine. There is thus a partial lack of in vivo evidence on the involvement of pathogen-dependent Ln interactions in the meninges or with ex vivo tissue specimens directly originating from the brain endothelial lining. Only a few studies are available on CNS infections and interactions of pathogens with collagens. It has, however, been reported that a single amino acid mutation of FimH of the E. coli serotype O18:K1:H7 results in abolished binding to collagens I and IV (Pouttu et al., 1999).

Besides bacterial interactions with ECM constituents, the free-living soil-amoeba B. mandrillaris (Table 3), which is associated with granulomatous amoebic encephalitis, has been shown to bind Ln, type I collagen and Fn, as well as to degrade types I and III collagen, elastin and Plg (Matin et al., 2006; Rocha-Azevedo et al., 2007). It is hypothesized that these mechanisms are important for the CNS tropism of B. mandrillaris as the BM has to be breached when the pathogen migrates through the olfactory nerve pathway or via haematogenous dissemination to the final site of infection.

The genus Trypanosoma harbours several human pathogens, such as Trypanosoma brucei (causative agent of human African trypanosomiasis or ‘sleeping sickness’) and Trypanosoma cruzi (causative agent of American trypanosomiasis or ‘Chagas disease’) (Bastos et al., 2005, 2010). These protozoans are transmitted between vertebrate hosts via the saliva or faeces of blood-sucking insects and, upon entry to new host, have to attach to and degrade BM components in order to access host cells where they affect tissues such as the CNS, the gastro-intestinal tract and the heart. Proteases such as prolyl oligopeptidase isolated from both species have been shown to degrade type I collagen and, to a lesser degree, Fn (Bastos et al., 2005, 2010). It was also shown that the enzyme is secreted in vivo in an experimental animal model, and treatment with inhibitory compounds diminished or nullified the cell infectivity of T. brucei and T. cruzi trypomastigotes. Trypanosoma cruzi has also a vast array of adhesins such as gp 58/68 (Fischer et al., 1988), LAG bp (Velge et al., 1988), penetrin (Ortega-Barria & Pereira, 1991) and Ln-binding GP (Giordano et al., 1994) that bind ECM components such as collagen, Fn, heparin sulphate and Ln (Table 3).

The free-living protozoal opportunistic parasites Acanthamoeba spp. that cause human infections such as Acanthamoeba granulomatous encephalitis and amoebic keratitis (potentially leading to the loss of vision) have been shown to interact with components of the BM (Table 3) (Rocha-Azevedo et al., 2009). Three A. castellanii serine proteases (130, 42 and 12 kDa) and one metalloprotease (150 kDa) have been reported to possess collagenolytic activity (Cho et al., 2000; Na et al., 2001; Sissons et al., 2006). Furthermore, the 42- and 12-kDa proteases were demonstrated to degrade rabbit corneal extracts and to be cytopathic to human cells including, but not limited to, corneal cell lines in vitro. The 12-kDa protease was also found to degrade Fn, sIgA, IgG, Plg, fibrinogen and haemoglobin, possibly designating it as a major virulence factor. In contrast, the two larger proteases did not exhibit any cytotoxic effect against human brain microvascular endothelial cell lines, but did, however, disrupt the monolayer in a trypsin-like fashion, which could point towards a role in CNS invasion. Similarly, the closely related species A. healyi produces a serine protease that exhibits proteolysis of collagen types I and III, Fn, fibrinogen, IgA, IgG and even albumin and haemoglobin (Kong et al., 2000). Acanthamoeba healyi and Acanthamoeba culbertsoni have also been reported to display Ln binding via a characterized 55-kDa adhesin (Rocha-Azevedo et al., 2009) and AhLBP (Hong et al., 2004). Interestingly, both proteins were found in either low or nonexisting quantities in nonpathogenic Acanthamoeba. The sequence and probably structure similarities between AhLBP and the 67-kDa human LR were also significant, suggesting a shared binding site on the Ln molecule.

Attachment of oral pathogens to Ln and collagen

Lns are widely distributed in dental tissues and more precisely at the mucosal epithelial BMs (Masaoka et al., 2009). The expression of Lns varies during tooth morphogenesis, for example, mesenchymal cells produce Ln α1–4 chains, while the dental epithelium produces Ln α5 chain. Ln-5 (α3β3γ2) most prominently plays a role during morphogenesis and promotes enamel formation (Table S1) (Emingil et al., 2005, 2006; Fried et al., 2005). The mineralized part of the teeth, the dentin localized just beneath the enamel contains collagen I fibres and collagen type V. Similar to the process observed in bone formation, collagen I synthesized by odontoblasts serves as a template for hydroxyapatite crystal deposition (Butler, 1995; Butler et al., 2003). Moreover, the tooth pulp contains connective tissues constituting ECM proteins secreted by odontoblasts and fibroblasts. The pulp fibroblasts secrete several Ln isoforms and collagen type I among others (Table S1 and S2) (Garcia et al., 2003; Fried et al., 2005; Fukumoto et al., 2006). Oral pathogens may thus have an abundance of Ln and collagen available for adhesion. Adherence of oral pathogens to oral epithelial cells is necessary for colonization as the bacteria otherwise become part of the salivary flow and consequently will be swallowed. They have unique adhesins that bind to host cell receptors, including the ECM proteins and salivary agglutinins, and provide a firm anchorage (Whittaker et al., 1996).

Periodontal pathogens utilize host MMPs for degrading host epithelial cells and the BM for subsequent invasion. Most studies have been conducted on matrilysin, which is a host metalloproteinase (MMP-7) expressed in the junctional epithelium facing tooth side and the ‘epithelial cell rests of Malassez’ (ERM). Several oral pathogens can bind matrilysin at their surface and use it to degrade a variety of ECM proteins including Ln and collagen. Uitto et al. (2002) demonstrated that some oral pathogenic species such as Fusobacterium nucleatum, Fusobacterium necrophorun, Prevotella denticola and Prevotella endodentalis induce the expression of matrilysin in porcine periodontal ligament epithelial cells. The elevated levels of matrilysin could subsequently promote the invasion of the junctional tissue by these bacteria (Uitto et al., 2002). More recently, it was shown using human keratinocytes in vitro that F. nucleatum can modulate the expression of several host proteinases including collagenase 3 (MMP-13) and gelatinase B (MMP-9) (Uitto et al., 2005). In parallel, it was also found that F. nucleatum attracts pro-MMP-9 at its surface, which is converted into activated MMP-9. This interaction leads to increased invasion of the pathogens in a matrigel model (i.e. a reconstituted BM) (Gendron et al., 2004).

Porphyromonas gingivalis causes periodontal disease and adheres to gum and mucosal epithelial cells via its fimbriae and other virulence factors (Table 4). Purified fimbriae have comparatively similar binding characteristics to the ECM proteins Ln, Fn and Vn with a binding association constant of K a  = 2.15 × 106 M−1 for Ln (as measured by surface plasmon resonance) and K a  = 2.76 × 106 M−1 for type I collagen (Nakamura et al., 1999). Fimbriae particularly bind collagen I, through adhesins FimA, C, D and E, and it was suggested that this binding was made possible by partial proteolysis of ECM components (Kontani et al., 1996; Hamada et al., 1998; Nishiyama et al., 2007). Recently FimA expression level has been found to increase in response to tobacco smoke extract. This upregulated FimA expression was strongly correlated with increased biofilm formation. Thus, this might provide a direct clue for the increased incidence of periodontitis observed in smokers (Bagaitkar et al., 2010). Proteases are part of P. gingivalis proteinase–adhesin complexes (gingipains) present in both membrane-bound and soluble secreted forms. Gingipains are cysteine proteinases rich in arginine (RgpA) and lysine (Kgp) motifs capable of binding ECM proteins. RgpA is a high-affinity Ln-binding complex with a dissociation constant of K d  = 4.7 ± 0.17 nM (Pathirana et al., 2006). In another study, the Rgp complex was demonstrated to degrade integrins α2, β1 and β3, which are known to bind Fn and Ln. Thus, degradation of those integrins releases fibroblasts that ultimately undergo apoptosis thereby contributing to the damage of the periodontal tissues (Baba et al., 2001). RgpA binds specifically to collagen V which probably allows P. gingivalis to adhere to the interstitial connective tissue of oral mucosa in order to invade these tissues (O-Brien-Simpson et al., 2003).

Table 3. Microorganisms with affinity for the respiratory tract and the CNS and their interactions with various ECM proteins. The type of collagen bound by adhesins is specified when experimental data are available to support the specificity. If not, it is stated “collagen”, either reflecting an affinity for a collagen extract or a lack of information about the collagen considered in the study. (The same remark applies to Tables 4 and 5)
PathogensLn/collagen-interacting protein/systemLnCollagenOthersReferences
Burkholderia cepacia BCAM0224 (trimeric autotransporter adhesion) I Mil-Homens et al. (2010)
Escherichia coli Ln Wells et al. (2009), Erdem et al. (2007)
Haemophilus influenzae PELn Vn, PlgHallstrom et al. (2010)
H. influenzae nontypable (NTHi) Haemophilus adhesion penetration (Hap)LnIV Fink et al. (2002)
I Bresser et al. (2000)
I, III Virkola et al. (1996)
Klebsiella pneumoniae MrkD1P V Schurtz et al. (1994), Sebghati et al. (1998)
Legionella pneumophila Mip I, VI Wagner et al. (2007)
Moraxella catarrhalis UspA1/UspA2Ln Tan et al. (2006)
MID/Hag IV Bullard et al., 2007
Mycobacterium bovis Ln-binding protein (LBP)Ln Pethe et al. (2001)
M. leprae

Histone-like protein (Hlp)


Ln de Melo Marques et al. (2000)
Ln Rambukkana et al. (1997), de Melo Marques et al. (2000), Marques et al. (2001), Ottenhoff & Klein (2005)
M. smegmatis LBPLn Pethe et al. (2002)
M. tuberculosis M. tuberculosis pili (Mtp)Ln Alteri et al. (2007)
Malate synthaseLn Kinhikar et al. (2006)
Mycoplasma pneumoniae Ln Roberts et al. (1989), Yavlovich & Rottem (2007)
Neisseria meningitidis PiLQ, PorALn Orihuela et al. (2009)
I, III, V Eberhard et al. (1998)
Pseudomonas aeruginosa Saccharides I, II Stepinska & Trafny (1995)
Streptococcus agalactiae Ln-binding protein (Lmb)Ln Spellerberg et al. (1999), Tenenbaum et al. (2007), Nobbs et al. (2009)
S. anginosus Putative Ln-binding protein (Plbp)Ln Allen & Hook (2002)
LnI, IV Allen et al. (2002)
S. pneumoniae Pilus I (RgrA)Ln Moschioni et al.(2010)
RrgA I Hilleringmann et al. (2008)
VI Bober et al. (2010)
IV Kostrzynska & Wadstrom (1992), Eberhard et al. (1999)
S. pyogenes Ln-binding protein (Lbp),Ln Terao et al. (2002), Linke et al. (2009)
SpeBLn Hytonen et al. (2001)
Lipoprotein of S. pyogenes (Lsp)Ln Elsner et al. (2002)
Streptococcal collagen-like protein (Scl-1), Ln Caswell et al. (2010)
ShrALn Ouattara et al. (2010)
M1 VI Bober et al. (2010)
Cpa I Podbielski et al. (1999), Kreikemeyer et al. (2005)
VI Glurich et al. (1991), Visai et al. (1995), Dinkla et al. (2003)
Yersinia pestis Plg activator (Pla)LnI, IV, V Lahteenmaki et al. (1998)
Ln Benedek et al. (2003, 2005)
Influenza virus type AEnvelope GPLn FnJulkunen et al. (1984)
Aspergillus fumigatus AfCalApLn Murine pneumocytesUpadhyay et al. (2009)
Asp f 2Ln IgEBanerjee et al. (1998)
72-kDa surface GPLn Tronchin et al. (1997)
37-kDa surface proteinLn Gil et al. (1996)
Extracellular serine proteaseLn Tronchin et al. (1993)
Extracellular serine proteinase I, IIIElastin, FnIadarola et al. (1998)
Aspergillopepsin FLnCollagenElastinLee & Kolattukudy (1995)
Aspergillus spp.33-kDa alkaline protease CollagenElastin, fibrinogenMonod et al. (1991), Hanzi et al. (1993)
Coccidioides immitis SOWgp-encoded cell surface GP LnIVFnHung et al. (2002)
Cryptococcus neoformans 75-kDa extracellular serine proteaseLnIV Rodrigues et al. (2003)
Gelatinase Collagen Salkowski & Balish (1991)
Histoplasma capsulatum 50-kDa Ln-binding proteinLn McMahon et al. (1995)
Paracoccidioides brasiliensis GAPDHLnIFnBarbosa et al. (2006)
47- and 80-kDa surface components I Mendes-Giannini et al. (2006)
30-kDa adhesinLn Andreotti et al. (2005)
Gp43Ln FnVicentini et al. (1994), Mendes-Giannini et al. (2006)
Serine–thiol proteinaseLnIV Fn, proteoglycansPuccia et al. (1998)
19-kDa and 32-kDa cell wall proteinsLn Fn, fibrinogenGonzalez et al. (2005)
Pneumocystis carinii 33-kDa adhesinLn Fn, type 1 pneumocytesNarasimhan et al. (1994)
Acanthamoeba castellanii 42-kDa extracellular serine proteinase Collagen Cho et al. (2000)
12-kDa extracellular serine proteinase CollagenFn, sIgA, IgG, Plg, fibrinogen, haemoglobinNa et al. (2001)
150-kDa metalloprotease and 130-kDa serine protease I, IIIElastin, Plg, haemoglobinSissons et al. (2006)
Acanthamoeba culbertsoni 55-kDa Ln-binding proteinLn-1 Rocha-Azevedo et al. (2009)
Acanthamoeba healyi 33-kDa extracellular serine proteinase I, IVFn, fibrinogen, IgG, IgA, albumin, haemoglobinKong et al. (2000)
A. healyi Ln-binding protein AhLBPLn Hong et al. (2004)
Balamuthia mandrillaris ?Ln-1IFnRocha-Azevedo et al. (2007), (Rocha-Azevedo et al. (2007)
40- to 50-kDa extracellular metalloproteases I, IIIElastin, PlgMatin et al. (2006)
Trypanosoma brucei T. brucei prolyl oligopeptidase (POP Tb) I, native collagen in rat mesentery Bastos et al. (2010)
Trypanosoma cruzi T. cruzi prolyl oligopeptidase (POP Tc80) CollagenFnBastos et al. (2005)
Gp 58/68 or HAG bp CollagenFnFischer et al. (1988), Velge et al. (1988)
LAG bp CollagenFnVelge et al. (1988)
Penetrin CollagenHeparin, heparin sulphateOrtega-Barria & Pereira (1991)
30-kDa cysteine protease I Garcia et al. (2003)
Ln-binding GPLn Giordano et al. (1994)

Treponema denticola and Prevotella intermedia are spirochaetes found in human gingival crevices that cause human periodontal diseases. Adhesins of T. denticola and P. intermedia have a strong binding capacity to several ECM proteins that are used for increased adhesion to host tissues (Table 4) (Haapasalo et al., 1991). It includes a 72-kDa OMP and in particular, the major sheath protein (Msp), a porin displaying a considerable variability in sequence between the strains studied (52–63 kDa) (Haapasalo et al., 1992; Fenno et al., 1996; Edwards et al., 2005). Msp is a versatile protein mediating adherence to host cells and ECM components collagens I, III, IV and V and Ln, but also to other oral pathogens such as, P. gingivalis and F. nucleatum (Umemoto & Namikawa, 1994; Fenno et al., 1996; Umemoto et al., 1997; Rosen et al., 2008). Furthermore, the surface protein OppA binds Plg that can degrade ECM proteins in order to increase the invasive capacity of the pathogen (Fenno et al., 2000; Ishihara, 2010). Prevotella intermedia has an aetiologic role in gingivitis, destructive periodontitis and root canal infections. P. intermedia binds to many ECM proteins (Kalfas et al., 1992; Alugupalli & Kalfas, 1997; Eiring et al., 1998), and recently, the surface-exposed AdpB (29 kDa) was characterized as a high-affinity Fn- and Ln-binding protein. P. intermedia has also the ability to bind to collagen I found in teeth, and to collagen IV located in the BM, but although it was suggested that porins are involved in these interactions, no specific collagen-binding adhesins have been unraveled so far (Alugupalli et al., 1994; Grenier, 1996). Thus, it has been hypothesized that the abundance of ECM proteins such as Fn and Ln promotes the attachment of P. intermedia to host tissues in the oral cavity (Yu et al., 2006).

A group of streptococci including Streptococcus viridans, S. anginosus (Allen et al., 2002), S. gallolyticus (Vollmer et al., 2010) and Streptococcus gordonii (Sommer et al., 1992) are commonly found in the oral cavity and gums where they interact with ECM components (Table 4). Under pathogenic conditions, these species can also cause systemic heart valve infection known as endocarditis. Most of the S. viridans isolates from endocarditis patients have a high Ln-binding capacity (Switalski et al., 1987). The majority of endocarditis isolates of S. anginosus bind with a low affinity to collagens I and IV, whereas they interact more strongly with Ln and tissues of the porcine heart valve explants, suggesting that Ln is exposed during damage of the valves and hence might promote bacterial colonization (Allen et al., 2002). In another study, it was shown that S. anginosus adheres to exposed Ln in the BM via an 80-kDa putative Ln-binding protein (PLBP) (Allen et al., 2002). Several adhesins have been reported in S. gordonii resulting in bacterial adhesion to host tissues and an improved colonization. Hence, a 145-kDa Ln-binding OMP presenting some similarities with collagen I interacts with Ln, while Ssp A and B, two members of the streptococcal antigen I/II family, have been shown to interact with collagen type I (Sommer et al., 1992; Love et al., 1997). Interestingly, it was observed that Ssp(s) are up-regulated upon bacterial exposure to type I collagen or collagen-derived peptides leading to an increased adhesion and invasion (Heddle et al., 2003). Adhesins that were primarily characterized as Fn binding, such as CshA and FbpA, and the cell wall receptor polysaccharide also have a role in binding to collagens I and II (Giomarelli et al., 2006). Moreover, the aetiologic agent responsible for dental caries, Streptococcus mutans, has the ability to bind Ln and collagen I, a major component of the dentine (Beg et al., 2002). Streptococcus mutans Cnm and WapA were identified as collagen-binding molecules on the basis of the presence of a CBD in their respective sequences (Sato et al., 2004; Han et al., 2006). Recombinant Cnm interacts with collagen I and Ln in vitro, and the presence of the molecule at the surface of clinical strains has been correlated with their ability to bind collagen and Ln (Sato et al., 2004). AgA, the CBD of WapA, binds collagen I but is unable to interact with collagen IV or Ln, suggesting a role in interaction with teeth rather than with underlying tissues (Han et al., 2006). In Actinobacillus actinomycetemcomitans, EmaA, an orthologous of YadA, the Yersinia spp. adhesin, encodes an OMP that leads to the formation of antenna-like structures responsible for binding to fibril-forming collagens (Table 4) and subsequently to the invasion of extraoral tissues (Mintz, 2004; Ruiz et al., 2006; Tang et al., 2008). A. actinomycetemans has also been reported to bind Ln but details on the adhesins involved are unknown (Alugupalli & Kalfas, 1997).

Gastrointestinal pathogens and their interactions with Ln and collagen

The intestinal immune system is dysregulated in chronic bowel disease, for example Crohn's disease, leading to chronic inflammation. The intraluminal antigens of bacterial origin lead to the activation and release of inflammatory molecules, such as cytokines, reactive oxygen species, eicosanoids and proteolytic enzymes (MMPs) in genetically predisposed patients. This causes tissue injury and exposure of the ECM (Medina & Radomski, 2006). The intestinal epithelium has a prominent BM (Tables S1 and S2) consisting of Ln 3, Ln-2, Ln-5 and Ln-1 and several collagen subtypes (Leivo et al., 1996; Lotz et al., 1997; Bouatrouss et al., 2000; Teller & Beaulieu, 2001). The exposed BM provides an attractive target for intestinal pathogens such as Salmonella typhimurium, Helicobacter pylori, Campylobacter jejuni, Shigella spp., and Clostridium difficile leading to colonization and deep tissue penetration (Westerlund & Korhonen, 1993).

Fimbriae-expressing S. enterica serovar Typhimurium strains bind Ln at a significantly higher degree in comparison with nonfimbriated strains, whereas no difference in collagen binding was observed. A detailed study on fimbriae-dependent Ln binding suggested that purified Ln interacts with fimbriae via Ln carbohydrate side chains, as the removal of carbohydrate moieties by periodate treatment significantly reduced the Ln binding to fimbriae. Furthermore, using reconstituted BM ‘matrigel’ that mimics in vivo conditions also revealed fimbriae mediating Ln binding of S. enterica (Kukkonen et al., 1993). Additionally, the fimbriae of S. enterica bind Plg that degrades the noncollagenous ECM components of the BM thus rendering Ln available for the pathogen (Kukkonen et al., 1998) (details on these mechanisms are described in section ‘The BM is breached by pathogen-bound Plg and host proteases that degrade Ln and collagens’). Crago & Koronakis (1999) demonstrated that two virulence factors of S. enterica, Rck and PagC, also bind to carbohydrate side chains of Ln and are directly involved in adhesion to host cells. Attachment of S. typhimurium to collagens is mediated by SdhA, an autotransporter required for intestinal persistence of Salmonella in the mouse (Kingsley & Baumler, 2000; Kingsley et al., 2004), and by MisL, another autotransporter necessary for intestinal colonization in an avian model (Dorsey et al., 2005). MisL that mainly binds Fn exhibits a weak avidity for collagen IV, whereas the interaction of Sdh with collagen is stronger and partly inhibited by the addition of heparin (Kingsley et al., 2004; Dorsey et al., 2005). Thus, S. typhimurium has multiple systems to interact with Lns and collagens (Table 4).

Helicobacter pylori colonizes the human gastric mucosa and often causes gastritis. Because the gastric ventricle has a highly acidic pH, the survival conditions are harsher for H. pylori compared to pathogens residing in the smaller intestine or colon (Dubreuil et al., 2002). H. pylori is known to bind several ECM proteins including Ln, an interaction mediated by lectins that recognize the carbohydrate side chains of Ln (Table 4). Several H. pylori surface proteins are involved in the interaction with Ln. One such molecule is a 25-kDa protein that functions as a surface antigen and gives rise to antibodies detectable in saliva and sera of H. pylori-positive patients (Trust et al., 1991; Valkonen et al., 1994; Moran et al., 2005). Additionally, it has been predicted that haemagglutinating H. pylori isolates bind Ln two-fold higher (K d  = 4.1 picomole) in comparison with nonhaemagglutinating isolates (K d  = 8.2 picomole). This is attributable to the presence of more binding sites for Ln in haemagglutinating H. pylori isolates (Valkonen et al., 1997). Walz et al. (2005) have shown that the sialic acid–binding adhesin (SabA) interacts with Ln and that sabA isogenic H. pylori mutants lose their Ln-binding capacity. Sialidase treatment of SabA also reduced Ln-SabA binding, which further confirms the involvement of lectin-like interactions (Walz et al., 2005). Recently, AlpA and AlpB membrane proteins (adhesins) of H. pylori have been characterized as Ln binding. Isogenic mutants of alpA and alpB showed defective Ln binding, and ectopic expression of AlpA and AlpB in E. coli causes the acquirement of Ln-binding (Senkovich et al., 2011). It has also been demonstrated in a cell culture model that H. pylori releases soluble protein components that interfere with the attachment of T84 epithelial cells to a polystyrene surface. This soluble extract induces ‘dome-shaped’ blister formations in the cell monolayer, causing cell detachment from the surface. Only filters coated with Ln showed dome formation, while Fn- and collagen-coated filters displayed poor dome formation that was prevented by preincubation of the bacterial extract with Ln. Epithelial cells had a decreased expression of the Ln-binding α6β4 integrins (involved in cell–ECM adhesion) as well as the 67-kDa LR. Moreover, it has been shown that these dome-shaped cells have a disbalanced Na+/K+ ATPAse activity, which in turn causes Na+ depletion inside cells, ultimately leading to shrinkage and loosening of the cell-to-cell adherence. This interesting finding suggested that H. pylori-secreted proteins can alter the physiological balance and integrity of epithelial cells using Ln (Terres et al., 2003). Finally, infection with H. pylori results in an increased Ln-1 production, and this has been suggested to result in a more efficient adherence of the pathogen (Chan et al., 2006). Thus, this phenomenon might play a crucial role in H. pylori pathogenesis. The invasive potential of H. pylori is also attributable to its ability to bind collagen components. Attachment of both spiral and coccoid forms of H. pylori to fibril-forming collagen V has been reported as well as a weak binding to the other fibril-forming collagen types I, II and III (Khin et al., 1996). Besides interacting with Ln, invasion of subepithelial layers by H. pylori resides in the capacity of the pathogen to bind collagen IV (Trust et al., 1991; Ringner et al., 1994). However, a recent study showed that only two of 92 clinical isolates were able to bind collagen IV. These two strains were the only ones unable to produce BabA and SabA, two OMPs, suggesting the possibility that these two molecules ‘hide’ the adhesin(s) responsible for collagen IV (Odenbreit et al., 2009). Surprisingly, despite these experiments with whole bacteria, no studies revealed the specific molecules involved in collagen attachment of H. pylori, although an implication of fimbriae or at least OMP complexes has been evoked (Trust et al., 1991).

Enterococcus faecalis is a commensal that colonizes the gastrointestinal and genitourinary tracts. In patients compromised with severe underlying diseases, E. faecalis acts as an opportunistic pathogen and causes a wide range of infections, such as hospital-acquired infections, urinary tract infections, wound infection and endocarditis or infections related to implanted devices (Tomita & Ike, 2004). Clinical E. faecalis isolates show a wide range of adhesive capacity to ECM proteins including Fn, Vn, collagens I, II, IV and V as well as Ln (Table 4) (Xiao et al., 1998; Tomita & Ike, 2004). The OMP and adhesin Ace (Adhesin for collagen of E. faecalis) were, at first, identified as a collagen I-binding protein because of structural features similar to collagen binding adhesin (CNA), the collagen-binding adhesin from S. aureus (Rich et al., 1999). Subsequent studies, including the generation of an Ace mutant and anti-Ace antibody blocking experiments, revealed a broader substrate specificity including other ECM molecules such as Ln and also collagen IV (Nallapareddy et al., 2000ab). The interaction of Ace with collagen involves a collagen Hug model as firstly characterized between S. aureus CNA and collagen (Liu et al., 2007). Ace is ubiquitously expressed in all strains of E. faecalis. Interestingly, patients with E. faecalis endocarditis have high amounts of Ace and anti-Ace antibodies detectable in serum (Nallapareddy et al., 2000ab; Nallapareddy & Murray, 2006). Enterococcus faecalis urinary tract isolates proved that Ace is important for virulence and onset of infection in experimental models. Interestingly, Ace expression at the surface of bacteria is regulated by exogenous factors, in particular signals from the host. Thus, the oxidative stress regulator and repressor protein ‘enterococcal regulator for survival’ (Ers) control the expression of Ace during pathogenesis, for example the protein is upregulated at high bile salt concentrations (Lebreton et al., 2009). Moreover, Ace expression at the surface of enterococci is enhanced in vitro by the exposure to collagen type IV and to a lesser extent collagen I (Nallapareddy & Murray, 2006). Recently, Singh et al. (2010d) established in an experimental endocarditis mouse model that E. faecalisace mutants, exhibiting a reduced ability to bind collagen types I and IV, were shown to be defective in infecting mice as compared to the Ace-expressing wild types. Moreover, mice immunized with recombinant Ace showed a significant protection against E. faecalis infections. It has also been observed that Ace is important for the initial adherence of bacteria to mouse tissues. These results suggest that successful infection of tissues by E. faecalis requires an efficient Ace-mediated attachment to the ECM proteins, in particular collagen and/or tissues (Singh et al., 2010d). Enterococcus faecium that is closely related to E. faecalis is a commensal naturally localized in the gastrointestinal tract responsible for, among other infections, peritonitis and endocarditis. The capability of bacteria to interact with host connective tissues is a possible explanation for the success of the infection as evidenced by the existence of several MSCRAMMs (described in more detailed in section ‘Urogenital and skin pathogens also interact with Ln and collagen’) with affinity to ECM components (Hendrickx et al., 2009ab). ‘Adhesin for collagen of E. faecium’ (Acm) was characterized in the majority of clinical isolates and shows a particular affinity for collagen (Nallapareddy et al., 2003). Intriguingly, transformation with Acm of clinical isolates unable to bind ECM triggers in these strains an ability to bind collagen at levels comparable with the naturally collagen-binding strains (Nallapareddy et al., 2003). In addition, 62% of clinical isolates but only 6% of nonclinical strains were able to adhere to collagen proving the importance of Acm in E. faecium pathogenesis (Nallapareddy et al., 2008). Scm and EcbA are other proteins binding ECM and display a high affinity to collagen V and a low affinity to collagen I (Nallapareddy et al., 2008; Sillanpaa et al., 2008; Hendrickx et al., 2009ab). In silico studies revealed 15 putative MSCRAMMs from E. faecium, some of which could be adhesion for other ECM proteins such as Ln because isolates from patients with nosocomial infections were shown to efficiently bind to Ln and Fn (Sillanpaa et al., 2008; Zhao et al., 2009).

Similar to enterococci, enteropathogenic E. coli O157:H7 (EPEC) immobilized on a dextran surface (Biacore) binds collagen I, Ln, Fn and glycosaminoglycans (Medina & Fratamico, 1998; Medina, 2001). This attachment of EPEC to host cells is often mediated by the surface-exposed fimbriae, and these are known to bind Plg involved in the degradation of the epithelial BM (described in section ‘The BM is breached by pathogen-bound Plg and host proteases that degrade Ln and collagens’) (Korhonen et al., 1992; Kukkonen et al., 1998). In the Shiga-toxigenic E. coli (STEC) O157:H7, two proteins present at the surface have been reported as adhesins. The fimbrial component protein ELF encoded by the operon ycbQRST mediates the interaction with cells and Ln but not with the collagen types tested (Samadder et al., 2009). In contrast, the autotransporter EhaB displays an affinity for collagen I and Ln (Wells et al., 2009). Flagella from O157:H7 shows no adhesive properties to ECM molecules, whereas flagella from the EPEC O127:H6 serotype interacts with collagen and Ln probably through the hypervariable region of FliC (Erdem et al., 2007); notably, some adhesins are widely distributed in various E. coli including gastrointestinal pathogens such as the serine protease autotransporter family and Ag43 (Reidl et al., 2009; Tsang et al., 2010). Ag 43 has been described primarily for its requirement in biofilm formation but recently its adhesiveness to collagen types I, III, IV and V was highlighted (Reidl et al., 2009).

Yersina enterocolitica and Y. pseudotuberculosis are two human enteropathogenic bacteria. These species share a similar ability to bind ECM molecules, in particular because of the presence of the TAA YadA at their surface, which is involved in many processes such as serum resistance, neutrophil binding and haemagglutination to name a few (El Tahir & Skurnik, 2001). YadA binds to collagens I, II, III, IV, V, and XI and with lower affinity Ln (Schulze-Koops et al., 1992; Tamm et al., 1993; Flugel et al., 1994). Binding of YadA to collagen has been studied using the Y. enterocolitica isoform of YadA because its affinity for ECM components (fibronectin) is higher as compared to Y. pseudotuberculosis YadA. This difference is due to 30 amino acids located in the N-terminal sequence of Y. pseudotuberculosis YadA. This region is essential for fibronectin-binding and integrin-mediated bacterial internalization while its deletion leads to a molecule displaying an affinity to collagens and Ln (Heise & Dersch, 2006). The interaction of YadA with collagen occurs through the triple helix in a structure-dependent manner but without sequence specificity (Leo et al., 2008, 2010). The structural and binding analysis of YadA to collagen has been characterized using crystallographic and site-directed mutagenesis approaches. The full structural model for YadA is shown in detail in Fig. 9a (Koretke et al., 2006). All of the TAAs, including YadA, contain an N-terminal short signal peptide, an internal passenger domain and a C-terminal translocator domain. At the bacterial surface, they appear as a ‘lollypop-like’ structure and the proteins have distinct head, neck, coiled-coil (stalk) and transmembrane regions (Fig. 9a). The head regions of TAAs are structurally diverse and contain the major ligand-binding domains. In recent years, N-terminal head domains of H. influenzae Hia (Linke et al., 2006; Meng et al., 2008) and Hsf (Cotter et al., 2005), BpaA from Burkholderia pseudomallei (Edwards et al., 2010), BadA (Szczesny et al., 2008) and YadA (Nummelin et al., 2004) have been crystallized. The head regions are diverse in structure, shape and size, and are thus consequently involved in several different functions. The coiled-coil region in YadA (Nummelin et al., 2004), UspAs (Brooks et al., 2008; Conners et al., 2008; Agnew et al., 2011), and BpaA (Edwards et al., 2010) has similarities in structural folding, but the length of coils varies among different TAAs. Hsf, Hia and BpaA have a distribution of multiple structural domains between the stalks, rather than a long coiled-coil stalk (Meng et al., 2008), like in UspAs and YadA. The transmembrane-anchoring domain consists of a β barrel and is homologous in TAAs (Cotter et al., 2005).

Figure 9.

Trimeric autotransporter YadA of Yersina enterocolitica and collagen-binding model. (a) A model that shows the full structure of YadA encompassing membrane-anchoring, coiled-coil and head domains. The PDB file for this model was downloaded from Coordinate file; Yad_Mod_V2. (b) The head domain YadA (amino acids 24–196) is composed of beta sheets arranged in nine coiled LPBR, surrounded by a partly disordered random coil at N-terminal and a C-terminal neck followed by coiled-coil stalk (PDB; 1P9H). (c) Docked model of the YadA head domain with a collagen triple helix. Site-directed mutagenesis of identified residues suggested that E80, K83, V98, N99, K108, L110, H159, H162, D180, E182, N166, Y169, R133 are involved in collagen binding. Mutation of the amino acid residues (red colour) abolished the binding to collagen, whereas that of amino acid residues (orange and magenta) only partly attenuated the binding. Mutated D180 and E182 also completely abolished binding to collagen and were also involved in the trimerization. The model was taken from Nummelin et al. (2004) with permission. Structures (a and b) were prepared using pymol.

The head domain of YadA (amino acids 24–196) is solely composed of β sheets arranged in nine coiled left-handed parallel β rolls (LPBR), surrounded by partly disordered random coils at the N-terminal and a C-terminal neck region (amino acids 195–219) with a random coil and short helix (Fig. 9b). The LBPR structure consists of a 13–16 residues repeated motif, and alignment of these repeated motifs reveals an NSVAI G XXS motif where G is completely conserved. The neck region connects the LPBR domain to the narrower stalk region (Fig. 9b). Residues 194–200 of one monomer run below the inner sheet of an adjacent monomer in an anticlockwise direction, and the 201-209 -residue loop under the third monomer. These two loops together make a ‘safety pin’, with some interchain hydrophobic and ionic interactions that are responsible for holding three chains together into a stable trimeric structure (Fig. 9b). The collagen triple helix (collagen I) was docked with the head structure of YadA using a bioinformatics approach (Fig. 9c). Further, site-directed mutagenesis suggested that several residues of YadA play important roles in collagen binding. Mutation of D180 and E182 (shown in boldface) completely abolished the collagen binding (Fig. 9c) (Nummelin et al., 2004).

Other bacterial gastrointestinal pathogens that are outlined in Table 4, for example, C. jejuni (Kuusela et al., 1989), Shigella spp. (Menard et al., 1996), Clostridium spp. (Kreutz & Jurgens, 1995), C. difficile (Calabi et al., 2002), Clostridium perfringens (Martin & Smyth, 2010) and Lactobacillus brevis (de Leeuw et al., 2006), are also known to bind ECM molecules, but more specific details are not yet available.

Entamoeba histolytica is a causative agent of amoebic dysentery and causes severe damage of the intestinal epithelium and ECM (Table 4) during the course of infection which can lead to systemic dissemination (Li et al., 1995). Proteolytic enzymes involved in the invasion of host tissue have been identified, and a 56-kDa proteinase as well as a 27-kDa cysteine protease that degrades Ln, collagen and Fn has been characterized in detail (Keene et al., 1986; Schulte & Scholze, 1989; Li et al., 1995). The 56-kDa proteinase was also found to activate Plg, thus potentially promoting the invasiveness of the amoeba. In addition to proteases, E. histolytica expresses adhesins that bind to collagen (Rosales-Encina et al., 1992), as well as Ln and Fn (Sengupta et al., 2001). Other amoebal pathogens including Giardia duodenalis have been reported to secrete enzymes that degrade BM (de Carvalho et al., 2008).

Urogenital and skin pathogens also interact with Ln and collagen

Lns are widely distributed in the renal parenchyma including glomeruli and tubules. A moderate to weak expression of Ln-1, Ln-3 and Ln-4 can be seen in the kidneys (Table S1), whereas the epithelial cells of the urinary bladder express high concentrations of Ln-4 ( Ln-5 is also distributed in the renal parenchyma (Zuk & Matlin, 2002; Joly et al., 2003). Collagen type IV is an important component of the renal glomerular BM where its action to stabilize the structure of the filtering unit is essential to maintain a functional glomeruli (Yurchenco & Patton, 2009). Uro-epithelial cells express Ln isoforms composed of α1, α3, α5, β1, β2, β3, γ1 and γ2 chains, while stromal cells express α1, α2, α4, α5, β1, β2 and γ1 chains (Table S1) (Hattori et al., 2003). Collagen type IV is also present in the BM underlying the urogenital epithelium and the bladder epithelium (Knupp & Squire, 2005). The collagen protein family is largely present in adult skin in quantity and in variety. Indeed, if collagen I is a prominent component of the dermis, fibril-forming collagens III and V, FACITs XII and XIV and collagens IV, VI and VII (Table S2) to name a few are also constituents of the various layers of the skin (Keene et al., 1997; Kielty & Shuttleworth, 1997). On the other hand, Ln-2 and Ln-5 are the dominant isoforms that are distributed in the skin (Shibuya et al., 2003; Schneider et al., 2007), and others such as Ln6, Ln2/4, Ln-10 and Ln-332 also exist in the genitourinary tract (Table S1) (Gorelik et al., 2001; McMillan et al., 2006; Sugawara et al., 2008; DeRouen et al., 2010). BMs of the urogenital system and skin are thus equipped with Lns and collagens (Tables S1 and S2) that various pathogens can adhere to.

Haemophilus ducreyi causes chancroid and genital ulceration and is categorized as an agent of sexually transmitted infections (STI). The species has adhesive capacity leading to binding of many ECM proteins including fibrinogen, Fn, gelatin, collagen and Ln (Table 5). Pili of H. ducreyi have been suspected to be Ln binding because clinical isolates devoid of pili are unable to adhere to Ln, while these isolates still efficiently bind other ECM proteins (Abeck et al., 1992; Brentjens et al., 1996). Binding studies with ECM proteins including Ln have been conducted on pili mutants of H. ducreyi, but no significantly decreased binding was observed. Thus, at present, it is not clear which mechanism is responsible for Ln binding in H. ducreyi (Bauer & Spinola, 1999). In order to infect its host, H. ducreyi interacts specifically with collagen types I and III, major components of the dermis connective tissue (Bauer & Spinola, 1999). This interaction has been shown in vitro and in vivo suggesting that subsequently to wounding, H. ducreyi binds to the temporary exposed collagen fibrils of the dermis (Bauer et al., 2001). This interaction is mediated by NcA, an OMP showing between 20–25% identity with UspA from M. catarrhalis and YadA of Y. enterocolitica, two adhesins binding among other proteins collagens (Fulcher et al., 2006). In addition, Chlamydia trachomatis (Kihlstrom et al., 1992), Lactobacillus crispatus (Antikainen et al., 2002) and Brucella abortus (Castaneda-Roldan et al., 2004) are also known to bind Ln and collagen (Table 5) but details on these interactions in pathogenesis have not yet been published.

Table 4. Oral and GI tract microorganisms that interact with Ln, collagens and other proteins in the ECM
PathogensLn/ collagen-interacting protein/systemLnCollagenOthersReferences
Actinobacillus actinomycetemcomitans Protein ALn Alugupalli et al. (1994)
EmaA Fibril-forming Mintz (2004)
I, II, III, V Mintz & Fives-Taylor (1999)
Campylobacter jejuni OMPs I Moser & Schroder (1995)

I, III, V (strain dependent)

IV (weak)

Kuusela et al. (1989)
Clostridium spp.-Ln Kreutz & Jurgens (1995)
C. difficile Surface layer proteins (SLPs)LnI Calabi et al. (2002)
C. perfringens LnIII, IV Martin & Smyth (2010)
Enterococcus faecalis Adhesin of collagen from E. faecalis (Ace)LnI, IV Singh et al. (2010c), Nallapareddy et al. (2000a, b), Nallapareddy & Murray (2006)
I Rich et al. (1999)
E. faecium Adhesin of collagen from E. faecium (Acm) I, IV Nallapareddy et al. (2003)
Scm V Sillanpaa et al. (2008)
EcbA V Hendrickx et al. (2009a, b)
Ln Zhao et al. (2009)
Escherichia coli FimbriaeLn Korhonen et al. (1992), Kukkonen et al. (1993), Ramirez et al. (2009)
Flagella (serotype specific)LnCollagen Erdem et al. (2007)
AaFLnIV Farfan et al. (2008)
ELFLn Samadder et al. (2009)
EhaBLnI Wells et al. (2009)
Ag43 I, III, IV, V Reidl et al. (2009)
LnI Medina (2001)

Fusobacterium nucleatum

Fusobacterium spp.

Reconstituted BMReconstituted BM Gendron et al. (2004)
Helicobacter pylori 25 kDaLn Valkonen et al. (1997), Moran et al. (2005)
LPSLn Valkonen et al. (1994)
Ln Valkonen et al. (1993), Terres et al. (2003)
BabA, SabALn Walz et al. (2005)
AlpA, AlpBLn Senkovich et al. (2011)
IV Trust et al. (1991), Ringner et al. (1994)
V Khin et al. (1996)
Porphyromonas gingivalis Fimbriae, proteinase–adhesin complex gingipainLnV Pathirana et al. (2006)
RgpA V O'Brien-Simpson et al. (2003)
FimA I Hamada et al. (1998)
FimE, FimC, FimD I Nishiyama et al. (2007)
LnI Nakamura et al. (1999)
Prevotella spp.Ln FnEiring et al. (1998)
P. intermedia 62 kDaLn Kalfas et al. (1992)
AdpBLn Yu et al. (2006)
LnI, IV Alugupalli et al. (1994), Grenier (1996)
Shigella spp.LnIV Watarai et al. (1995), Menard et al. (1996)
Salmonella typhimurium Type I fimbriae, Ln Kukkonen et al. (1993, 1998)
Rck and PagCLn Crago & Koronakis (1999)
T2544Ln Ghosh et al. (2011)
SdhA I Kingsley et al. (2004)
MisL IV Dorsey et al. (2005)
Streptococcus mutans Cnm I Sato et al. (2004)
Wap A (AgA) I Han et al. (2006)
LnI Beg et al. (2002)
S. gallolyticus Acb Collagen Sillanpaa et al. (2008, 2008)
LnI, II, IV, V (strain specific) Vollmer et al. (2010), Sillanpaa et al. (2008)
S. gordonii SspA & SspB I Love et al. (1997)
CshA, FbpA, polysaccharide I, II Giomarelli et al. (2006)
Treponema denticola MspLn Haapasalo et al. (1991, 1992), Fenno et al. (1996)
I, III, IV, V Umemoto & Namikawa (1994), Umemoto et al. (1997), Edwards et al. (2005)
Yersinia enterocolitica Yad ALnI, II, III, IV, V, XI Schulze-Koops et al. (1992), Tamm et al. (1993)
Entamoeba histolytica 56-kDa extracellular proteinaseLnIFn, PlgKeene et al. (1986)
27-kDa extracellular cysteine proteaseLnIV, VFnSchulte & Scholze (1989), Li et al. (1995)
105-, 56- and 30-kDa collagen-binding proteins I Rosales-Encina et al. (1992)
140-kDa beta1 integrin-like moleculeLnCollagenFnSengupta et al. (2001)
Giardia duodenalis Uncharacterized extracellular proteases Collagen de Carvalho et al. (2008)

Treponema pallidum causes syphilis that is also categorized as an STI. Interestingly, only pathogenic strains bind to host cells as opposed to nonpathogenic isolates. Treponema pallidum colonizes epithelial surfaces, invades the tight junctions of the epithelial cells, traverses tissue barriers and enters the circulation, which results in a systemic infection. Recently, a protein named Tp0751 was identified as an Ln-binding protein in T. pallidum. Binding assays in vitro revealed that the carbohydrate residues of Ln are involved in this interaction. Furthermore, Tp0751 is immunogenic because high titres of anti-Tp0751 antibodies are found in convalescent sera from patients with syphilis (Cameron, 2003). In a follow-up study, Cameron et al. (2008) showed that Tp0751 has a broad specificity for the Ln-1, Ln-2/4, Ln-8 and Ln-10 isoforms. The Ln-binding region of Tp0751 was mapped using synthetic peptides, and the results showed that a region only comprising 10 amino acids was responsible for the interaction with Ln. Moreover, heterologous expression of Tp0751 in the related non-Ln-binding Treponema phagedenis results in the acquisition of the Ln-binding capacity (Cameron et al., 2008). Thus, Ln binding is a potential adherence mechanism utilized by T. pallidum to initiate host invasion, and hence inhibitory drugs against the interaction with Ln may be used for prevention. Treponema spp. also showed a capacity to interact with fibril-forming collagens I, III and V and the network-forming collagen IV (Umemoto & Namikawa, 1994). Although it was determined that the binding was mediated by several OMPs, the adhesions (Table 5) and their interactions were not further studied.

Leptospira interrogans causes interstitial nephritis, tubular necrosis and liver dysfunction. Interaction with humans through adhesion to the skin or via ingestion leads to the regulation of over 200 genes in the bacteria, including several adhesins (Matsunaga et al., 2007). Efforts have recently been made in order to identify the Ln- and collagen-interacting surface proteins of L. interrogans (Table 5). Leptospiral surface protein 21 (Lsa21) is an adhesin that binds to two major BM components, Ln via the glycosylated side chains of Ln and collagen type IV (Atzingen et al., 2008). Other proteins of L. interrogans designated Lsa24 (Stevenson et al., 2007), Lsa27 (Longhi et al., 2009) and Lsa63 also bind several ECM proteins including Ln and collagen IV (Vieira et al. 2010). Most of these protein–protein interaction studies were performed in vitro and have not been confirmed by isogenic mutants. In a more recent study, Oliveira et al. (2010) identified another protein, LipL53, that is regulated by temperature changes. At 37 °C, all isolates of L. interrogans express this protein, whereas expression is exclusive to virulent strains at 28 °C. Interestingly, LipL53 binds Ln in a dose-dependent manner, and immunization of hamsters with LipL53 provides protection against infection. In addition, the protein named LipL32 has been reported to bind several ECM proteins and subsequently induces pro-inflammatory responses via the host Toll-like receptor-2 (TLR-2) (Hsu et al. 2010; Hauk et al., 2008). These reports indicate that the adhesive mechanism of L. interrogans very effectively utilizes ECM proteins in order to cause disease, but details of the interaction mechanisms are still unknown. LigA and B are two OMPs with Ig-like domains from L. interrogans with the ability to bind several ECM components (Choy et al., 2007, 2011; Lin et al., 2009). Experiments in vitro showed a stronger affinity of LigB for collagens I and IV as compared to LigA (Choy et al., 2007). Although both molecules bind Ln, the strength of this interaction was weaker than the one observed for collagens. Interestingly, LigA and B are induced by exposure to human physiological osmolarity conditions, which also results in an increase in adhesion of the whole bacteria to ECM. This observation suggests a role of Ligs in the invasion process of the host (Choy et al., 2007) (Tables 5).

Table 5. Known interactions with Ln, collagens and other ECM proteins for microorganisms found in the urogenital tract and the skin
PathogensLn/ collagen-interacting protein/systemLnCollagenOthersReferences
Bacillus anthracis BA0871A, BA5258A I Xu et al. (2004)
Borrelia burgdorferi ErpX (OMP), Ln Brissette et al. (2009)
BmpALn Verma et al. (2009)
CRASP-1LnI, III, IV Hallstrom et al. (2010)
I (native) Zambrano et al. (2004)
Brucella abortus LnCollagens Castaneda-Roldan et al. (2004)
Chlamydia pneumoniae Ln Roberts et al. (1989)
C. trachomatis I Froman et al. (1994)
LnI, IV Kihlstrom et al. (1992)
E. coli (uropathogenic)Dr haemagglutinin, DraE IV Miettinen et al. (1993), Nowicki et al., (1993), Carnoy & Moseley (1997), Selvarangan et al. (2004)
Curli fimbriaeLn Antao et al. (2009)
Haemophilus ducreyi NcaA I Fulcher et al. (2006)
LnI, III Bauer & Spinola (1999), Bauer et al. (2001)
Lactobacillus crispatus CbsALnI, IV Sillanpaa et al. (2000), Antikainen et al. (2002)
EnolaseLnI Antikainen et al. (2007)
I, IV (high affinity) and V (low affinity) Toba et al. (1995)
L. brevis SlpALnIV de Leeuw et al. (2006)
Leptospira interrogans Len A to F Ln Stevenson et al. (2007)
LipL32 IV Hauk et al. (2008)
Lsa 21LnIV Atzingen et al. (2008)
Lsa 27Ln Longhi et al. (2009)
Lsa63LnIV Vieira et al. (2010)
LipL53LnIV Oliveira et al. (2010)
Lig A & BLnI, III, IV Choy et al. (2007), Choy et al. (2011)
Staphylococcus aureus Eap I, IV, VI Hansen et al. (2006)
Emp I Hussain et al. (2001)
α-Enolase I Carneiro et al. (2004)
CNA Patti et al. (1992, 1995)
Treponema pallidum Tp0751 Cameron (2003, 2005, 2008), Houston et al. (2011)
45-, 49- and 62-kDa proteins I, III, IV, V Umemoto & Namikawa (1994)
I, IV Fitzgerald et al. (1984)
Human papillomavirus CapsidLn-5 Culp et al. (2006)
Retrovirus HIV-1gp120/gp160Ln Fn, VnBozzini et al. (1998)
Polyomavirus Large T-antigen (T-ag)LnIV Fenton et al. (2008)
Candida albicans Als3p IV Nobbs et al. (2010)
95-kDa cell wall metallopeptidaseLnI, IVFnRodier et al. (1999)
21-kDa componentLn Sakata et al. (1999)
Cell wall–associated GAPDHLn FnGozalbo et al. (1998), Gil et al. (1999)
Secreted aspartate proteinasesLn (matrigel) Watts et al. (1998)
Ala1LnIVFnGaur & Klotz (1997)
Acid proteinaseLnCollagenECM proteinsMorschhauser et al. (1997)
p37Ln Lopez-Ribot & Chaffin (1994), Lopez-Ribot et al. (1996)
60- and 105-kDa surface GPs IFnKlotz et al. (1993)
60- and 68-kDa mannoproteinsLn Fibrinogen, C3dTronchin et al. (1989), Bouchara et al. (1990)
46-kDa extracellular protease Collagen Kaminishi et al. (1986)
Candida spp.MetalloproteinasesLn-332 Parnanen et al. (2008)
50- and 62-kDa extracellular serine proteasesLn FnDos Santos & Soares (2005), Portela et al. (2010)
Candida acid proteinase (CAP) CollagenKeratin, haemoglobin, casein, albuminRay & Payne (1990)
Trichophyton schoenleinii 82-kDa metalloproteinase I Ibrahim-Granet et al. (1996)
Trichomonas vaginalis 118-kDa surface proteinLn Casta e Silva Filho et al. (1988)
Surface cysteine proteinase (CP65) IVFnAlvarez-Sanchez et al. (2000)
Surface-associated GAPDH CollagenFn, PlgLama et al. (2009)
30-kDa cysteine proteinase (CP30) IVFn, haemoglobinMendoza-Lopez et al. (2000)
Leishmania amazonensis 115-kDa extracellular dimeric serine protease CollagenHaemoglobin, bovine serum albumin ovalbumin, fibrinogenSilva-Lopez et al. (2005)
L. donovani LBPLn Bandyopadhyay et al. (2002)
L. mexicana ? I Lira et al. (1997)
Leishmania spp.Surface metalloprotease gp63 IVFnMcGwire et al. (2003)

The aetiologic agent of lyme disease, B. burgdorferi, is transmitted by ticks that deposit pathogens in the dermis of the host while having a bloodmeal. Consequently, to invade the human host, spirochaetes will bind to connective tissue components of the skin, particularly collagen type I (Zambrano et al., 2004). That study emphasizes the affinity of B. burgdorferi for native collagen rather than isolated' protomers. However, recently in vitro studies demonstrated the ability of the complement regulator–acquiring surface protein-1 (CRASP-1) to bind collagens I, III and IV and Ln, suggesting a role of this OMP in the adhesion and invasion processes (Hallstrom et al. 2010). Recently, the outer surface protein (ErpX) of B. burgdorferi was reported as the first Ln-binding candidate (Brissette et al., 2009). Another group of paralogous OMPs, that is, BmpA, B, C and D, have the capacity to bind Ln (Verma et al., 2009). Thus, B. burgdorferi interactions with Ln and collagen (Table 5) may contribute to adhesion, colonization and dissemination of this pathogen.

In parallel to most bacterial species, L. crispatus is able to bind both Ln and collagen types I and IV by the intermediate of CbsA, a molecule also involved in anchoring the S layer to the bacterial cell wall (Sillanpaa et al., 2000; Antikainen et al., 2002). The N-terminal part of CbsA is responsible for interaction with immobilized and soluble collagen and with immobilized Ln (Antikainen et al., 2002). Chlamydia trachomatis (Kihlstrom et al., 1992) and B. abortus (Castaneda-Roldan et al., 2004) bind Ln and various collagen types (Table 5) but details on these interactions during pathogenesis are currently unknown.

Bacillus anthracis most commonly causes cutaneous infections (95% of cases) and localized infections (black eschar), but is also associated with gastrointestinal and pulmonary infections that may be fatal. It has been demonstrated that Bacillus collagen-like protein (BclA) is a surface protein that binds Ln at the surface of B. anthracis spores. BclA acts as a shield for the spores and controls the germination and attachment to host ECM proteins (Brahmbhatt et al., 2007). More recently, it was shown that the surface protein α-enolase binds Ln and Plg, which is subsequently converted into plasmin that increases the invasiveness of the B. anthracis (Agarwal et al., 2008). In B. anthracis, two collagen-binding adhesins displaying a similar domain organization to the S. aureus adhesion CNA have been unraveled. These proteins named BA0871A and BA5258A bind specifically to the dermal collagen type I (Xu et al., 2004).

Urinary tract infections such as cystitis or glomerulonephritis are commonly associated with the presence of various E. coli strains (Servin, 2005). At their surface, these uropathogenic E. coli (UPEC) exhibit the Dr haemagglutinin and F1845 fimbriae, two members of the Dr adhesin family that mediate infection (Bilge et al., 1989; Westerlund et al., 1989; Servin, 2005). Dr haemagglutinin binds either to decay accelerating factor (DAF; CD55), an epithelial cell receptor, which allows the bacteria to interact with epithelial cells of the urogenital mucous membrane, or to type IV collagen located in the subepithelial BM which leads to a persistent infection (Miettinen et al., 1993; Nowicki et al., 1993; Carnoy & Moseley, 1997; Selvarangan et al., 2004). Generation of DraE mutants demonstrated the importance of the presence of DraE in the Dr haemagglutinin complex to achieve binding to DAF but not to collagen IV, and subsequently the importance of the interaction of fimbriae with collagen IV to achieve an efficient infection (Van Loy et al., 2002; Selvarangan et al., 2004). Moreover, type-G fimbriae have been defined to bind Plg, and other surface proteins such as curli fimbriae and the UPEC trimeric autotransporter were found to bind Ln (Korhonen et al., 1992; Kukkonen et al., 1998; Antao et al., 2009).

Staphylococcus aureus is a Gram-positive bacterium residing on the skin and in the nose, causing a wide range of infections including skin infections, endocarditis and pneumonia to name a few. Therefore, it is not surprising that S. aureus exhibits a wide range of adhesins with a broad spectrum of ECM ligands such as Ln, collagens, elastin, fibrinogen and fibronectin (Table 5) (Lopes et al., 1985; Speziale et al., 1986; Foster & Hook, 1998). Several studied adhesins of S. aureus belongs to the microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) that are involved in adhesion and immune evasion. MSCRAMMs of Gram-positive pathogens are covalently linked to the bacterial surface. The anchoring of these molecules at the surface of the pathogen is performed by the action of sortase enzyme, lipid II and penicillin-binding protein (PBP) in four different steps: (1) MSCRAMMs have an N-terminal signal for the translocation of protein to the membrane and a C-terminal cell wall sorting signal. The N-terminal signal is cleaved off after translocation of the proteins across the plasma membrane by a signal peptidase. The N-terminal domain is displayed at bacterial surface, whereas the C-terminal domain is buried in the cell wall. The C-terminal sorting signal encompasses a 35-amino-acid sequence including an LPXTG motif with hydrophobic residues and a positively charged tail; (2) the enzyme sortase through its active site cleaves the amide bond between Thr and Gly of the LPXT↓G motif and generates an acyl enzyme (thioester intermediate); (3) as a third step, the Thr residue of LPXT is bridged with lipid II (peptidoglycan synthesis precursor) as a consequence of nucleophilic attack by the amino group within the pentaglycine of lipid II; (4) in the final step, the lipid II-protein precursor functions as a substrate for PBP that incorporates this precursor into cell wall by catalysing a transpeptidation reaction. This pathway is conserved in most of the Gram-positive pathogens. For a detailed mechanism of sortase-mediated linkage of MMSCRAMs, see the detailed reviews (Marraffini et al., 2006; Hendrickx et al., 2011)

The first adhesin isolated from S. aureus was the collagen-binding adhesin CNA exhibiting an affinity for fibrillar collagens (Patti et al., 1992). CNA has been reported as a versatile adhesin that establishes contact with host cells and tissues under different pathogenic conditions. S. aureus is frequently associated with bacterial arthritis (Shirtliff & Mader, 2002). A very interested finding is that S. aureus expressing CNA have the capability to cause arthritis-like symptoms when injected in joints of animal models. However, Ccna mutants caused less frequent arthritis-like symptoms. This study directly correlated the collagen-binding capacity of S. aureus with arthritis causing symptoms (Patti & Hook, 1994). In another recent study, it has been shown that the collagen-binding capacity mediated by CNA is important for S. aureus colonization in joints (Xu et al., 2004). Moreover, during systemic infection in an endocarditis model, it has been observed that CNA is important for the attachment of bacteria to the aortic valve (Hienz et al., 1996). CNA has also been observed as a direct virulence factor for S. aureus in experimental model of keratitis (eye lens infection), but isogenic cna mutants were not able to cause keratitis (Rhem et al., 2000).

Ponnuraj et al. (2003) described a unique mechanism of binding of MSCRAMMs; SdrG with fibrinogen by a dock, lock and latch mechanism, which also has been observed in other homologous MSCRAMMs of Gram-positive pathogens (Ponnuraj et al., 2003). To resolve collagen binding mechanism, CNA of S. aureus has been co-crystallized with synthetic collagen peptides. These experiments revealed the very first model of a collagen interaction with a bacterial receptor. The MSCRAMM family members basically consist of two domains (N1 and N2) with IgG-like folds designated DE variants, hence DEv-IgG folds that are connected by a loop (Deivanayagam et al., 2002; Zong et al., 2005) (Fig. 10a). The N1 and N2 domains of S. aureus CNA consist of antiparallel 9–11 β strands with one or two α helices (Fig. 10a). Both domains, N1 (amino acids 31–163) and N2 (amino acids 174–329), are connected by a 9-amino-acid-long (164–173) flexible loop (Zong et al., 2005). The binding mechanism of CNA to the collagen triple helix was described as the Hug model, where N1 and N2 domains exist in an open conformation (apoprotein) with a binding trench in the N2 subdomain being accessible for a collagen triple helix to dock. As a result, the N1 and N2 domains wrap around the collagen helix and make a tunnel-like structure that locks the collagen ligand (closed conformation). Finally, the C-terminal extended β sheet of the N2 domain acts as a latch by being inserted into a trench present in the N1 domain by β strand complementation (Zong et al., 2005) (Fig 10a). The collagen triple helix synthetic peptide-CNA complexed has a major binding region in the N2 domain and a few more interactions in the linker region. In the first step, a collagen triple helix binds to an open confirmation of CNA by low affinity that is established by polar and hydrophobic interactions. The residues Y175, T177, R189, Y233, N223, N278 and N193 of N2 domain are involved in interaction with proline and hydroxyproline of the collagen chains (Fig. 10b). In the following step, reorganization of the N1 domain occurs where the residue V172 of the linker region interacts with a proximal proline of the collagen leading chain and repositions the N1 domain. This repositioning is stabilized by a hydrophobic interaction between Y88 and P108 of the N1 and P182 and M180 of the N2 domains (Fig. 10b). Thus the repositioning step is significant in CNA wrapping around the bound collagen. Locking takes place by reorientation of the C-terminal extension of the N2 domain and finally insertion in between the N1 domain. The collagen peptide ligand is further secured by conformational changes occurring in the linker region, which causes the shrinkage of the hole between the N1 and N2 domains to snugly fit around the collagen. Thus, the collagen triple helix is enclosed in the hole created by the two domains followed by a linker that joins them (Zong et al., 2005). A similar kind of Hug model for binding of E. faecalis protein Ace to collagen has been reported (Liu et al., 2007; Ponnuraj & Narayana, 2007). CNA and Ace are homologous proteins with similar collagen-binding properties and conserved amino acids in the N2 domain. A superimposed model for both molecules is presented in Fig. 10a, and the collagen complex (CNA-collagen) with the active binding site are presented in Fig 10b. Homologous proteins with similar structures to CNA are S. aureus clumping factor (ClfA) (Deivanayagam et al., 2002), E. faecalis Ace (Liu et al., 2007), Staphylococcus epidermidis SdrG (Bowden et al., 2008) and finally Acranobacterium pyogenes collagen-binding protein A (CbpA) (Pietrocola et al., 2007).

Figure 10.

Structure of microbial surface components recognizing adhesive matrix molecules (MSCRAMM) present in Gram-positive bacteria. (a) Superimposed models of protein CNA from Staphylococcus aureus (PDB; 2F6A) and Ace (PDB; 2Z1P) from Enterococcus faecalis. MSCRAMM basically consists of two domains (N1 and N2) with IgG-like folds designated DE variants (DEv-IgG folds) that are connected by a loop. Each domain consists of antiparallel 9–11 β strands with one or two α helices. Synthetic collagen triple helical peptide (GPO)4 GPRGRT (GPO)4 is shown between the N1 and N2 domains. (b) Molecular interactions between CNA and a collagen triple helix (PDB; 2F6A). The residues Y175, T177, R189, Y233, N223, N278, N193 (shown in sticks) of the N2 domain are involved in interaction with proline and hydroxyproline of the collagen chains. Hydrophobic interactions between Y88, P108 of N1 and P182 and M180 of N2 domains have also been observed. Structures were prepared using PyMol.

In addition to the staphylococcal MSCRAMMs mentioned above, some staphylococcal adhesins have been reported to be anchorless, therefore named ‘secretable expanded repertoire adhesive molecules’ (SERAMs) (Chavakis et al., 2005). Two of these SERAMs have been shown to bind collagens. The extracellular adherence protein (Eap) is an adhesin with a wide range of substrate including ECM molecules and cells (Hussain et al., 2008). Eap has the ability to bind at least three different collagen types (I, IV and VI) that are present in a multitude of tissues and structures (Hansen et al., 2006). Similar to Eap, Emp belongs to the SERAM family and displays a vast collection of substrates including Vn, Fn and collagen I (Hussain et al., 2001). SERAMs offer a versatility in substrate interactions that partially explain the capacity of S. aureus to infect so many different host tissues.

Candida albicans is a pleiomorphic fungal opportunistic microorganism, commonly found on mucosal surfaces of the urogenital tract, on the skin and in the oral cavity. If dissemination occurs, the yeast can cause a wide range of severe infections in various tissues such as the heart, kidney and brain (Nobbs et al., 2010). Interactions of C. albicans with proteins of the BM are varied and well studied (Table 5). Adhesins that bind to the ECM components include Als3p (Nobbs et al., 2010), cell wall–associated GAPDH (Gil et al., 1999), Ala1 (Gaur & Klotz, 1997), p37 (Lopez-Ribot & Chaffin, 1994; Lopez-Ribot et al., 1996), a 21-kDa component (Sakata et al., 1999), two 60-kDa and 68-kDa mannoproteins (Tronchin et al., 1989) and two 60-kDa and 105-kDa surface GPs (Klotz et al., 1993). Moreover, Candida spp. produce cell wall–tethered peptidases (Rodier et al., 1999) and an array of secreted proteases (Kaminishi et al., 1986; Ray & Payne, 1990; Morschhauser et al., 1997; Watts et al., 1998; Dos Santos & Soares, 2005; Parnanen et al., 2008) with affinity for Ln, collagen and Fn. These reports implicate a highly important role of the ECM-degrading/binding proteins for Candida virulence and dissemination. Other fungal skin pathogens known to secrete ECM-degrading enzymes include T. schoenleinii, which produces a collagenolytic 82-kDa metalloproteinase (Ibrahim-Granet et al., 1996).

Protozoal pathogens belonging to the genus Leishmania cause a variety of diseases known as leishmaniasis, including cutaneous ulcers. They spread via sand fly vectors, in which they replicate in flagellated forms, and are transferred to humans via insect bites (Silva-Lopez et al., 2005). In human hosts, Leishmania becomes intracellular parasitic amastigotes residing in mononuclear phagocytes before eventually lysing the phagocytes and disseminating to other tissues (Bandyopadhyay et al., 2002). During this process, the pathogen has to attach to and cross the BM in order to reach the target tissues (Table 5). A Ln-binding cell surface protein, LBP, has been identified and characterized in Leishmania donovani (Bandyopadhyay et al., 2002). It was shown that the interaction is ionic and that LBP binds to the Ln β1-chain YIGSR pentapeptide sequence. Leishmania spp. have also been shown to bind type I collagen, possibly conferring skin tropism on the parasite as type I collagen is abundant in cutaneous tissues (Lira et al., 1997). Furthermore, two collagenolytic proteases have been identified in the genus. A 115-kDa extracellular dimeric serine protease was found to degrade collagen as well as fibrinogen, haemoglobin, insulin, gelatin, bovine serum albumin and ovalbumin (Silva-Lopez et al., 2005). The surface metalloprotease gp63 was also shown to degrade type IV collagen, and strains designed to superexpress the protein had enhanced capacity to migrate through the ECM in vitro (McGwire et al., 2003). These findings indicate that ECM interactions with Leishmania spp. could be of much importance for the virulence.

Trichomonas vaginalis is a flagellated protozoal parasite and the causative agent of trichomoniasis, one of the most common STI, usually manifesting as vaginitis (Lama et al., 2009). In the initial step of infection, protozoa colonizes the cervicovaginal epithelium and the underlying BM. Implicated in this interaction is a 118-kDa surface adhesin, found to bind Ln (Casta e Silva Filho et al., 1988). Furthermore, much like the fungal pathogens C. albicans and Paracoccidioides, T. vaginalis produces cell surface–associated GAPDH that binds to collagen, Fn and Plg (Lama et al., 2009). The pathogen also produces CP65 and CP30, proteins with both adhesive qualities (binding of HeLa cells) and proteolytic activity (degradation of type IV collagen and Fn) (Alvarez-Sanchez et al., 2000; Mendoza-Lopez et al., 2000).

Several viral cervicogenital pathogens have also been shown to bind ECM components. Human papilloma virus (HPV) is a sexually transmitted virus and the most prevalent cause of cervical cancer. HPV replicates in the epithelium of the skin and mucosa and can after wounding of the epithelial cell layer spread to basal keratinocytes (Culp et al., 2006). HPV type 11 capsids have been shown to bind Ln-5, which is secreted by migrating keratinocytes, consequently bridging the virus to the basal keratinocytes via Ln interaction with the α6 integrin. Indeed, it was shown that mutant keratinocytes lacking the α6 integrin were much less susceptible to infections by virions preadsorbed to a Ln-containing substrate, compared to α6 integrin-expressing keratinocytes (Culp et al., 2006). The authors suggested a model in which virions adsorbed to Ln-5 (expressed by migrating keratinocytes) would target proliferating, α6 integrin-expressing keratinocytes after epithelial wounding, implicating an important role of the pathogen–ECM interaction during the course of infection.

Renal productive polyomavirus has a strong association with systemic lupus erythematosus, which results in lupus nephritis and severe damage of the kidneys (Fenton et al., 2008). It was shown in an experimental murine model that glomerular expression of polyomaviral large T-antigen (T-ag) would result in nucleosome–T-ag complexes with high affinity for Ln and type IV collagen on the glomerular membrane. The following humoral response could then cause the onset of lupus nephritis, as suggested by Fenton et al. (2008).

Retroviruses have also been shown to display interactions with the ECM. Human immunodeficiency virus type 1 (HIV-1) is the causative agent of the blood and sexually transmitted disease such as acquired immune deficiency syndrome (AIDS), and HIV-1 proteins gp120 and gp160 bind to the C-terminal heparin-binding domain of Fn, as well as to Ln and Vn (Bozzini et al., 1998). Interestingly, this interaction may be beneficial to the host, as circulating Vn and Fn could help clear virions from the bloodstream. It was indeed reported that soluble Fn would significantly inhibit the CD4 binding of gp120/160 and neutralize the HIV-1 infectivity of CD4+ T cells. These interesting findings show that not all interactions between pathogens and ECM components are solely beneficial to the pathogen.

Concluding remarks and future perspectives

The first step of a successful colonization is attachment to host tissues, for all microbes including viral, bacterial, fungal and protozoal pathogens, and interactions with ECM are of prime importance in this regard (Fig. 8). This, of course, depends on the particular microorganism in question and its specific anatomical niches. Invasive pathogens cross different barriers, and in contrast, noninvasive counterparts might only colonize the epithelia. The microbial attachment to the host ECM is performed by surface proteins (adhesins). These adhesins are highly variable in nature, with a wide range of sizes (10–250 kDa), shapes (globular, fibrillar, single or multiple domains) and functions. A detailed literature study shows that microorganisms usually have multiple adhesins to recognize various ECM proteins. Thus, pathogen–host attachment via Lns and collagens is multifactorial.

The invasive capacity of various microorganisms is directly linked to the characteristics of the species in addition to host proteases that degrade structural ECM proteins including Ln and collagens. Several bacteria, many fungi and protozoans secrete collagenases, serine proteases and metalloproteases that eventually degrade Ln and collagens (Table 2). During infection, host MMPs can also be activated as a consequence of the pro-inflammatory response, consequently increasing the degradation of the ECM (Table 1). In addition, several microorganisms bind to host Plg, which is ultimately used to degrade the ECM (Fig. 6). Moreover, infection triggers the activation of neutrophils that migrate towards the site of infection which ultimately results in damaging of the tissue and inflammation. Thus, the degradation of Lns and collagens by proteases can pave the way for infection deeper into host tissues and helps breaching the cellular barriers. In recent years, protease inhibitors have been considered as antimicrobial therapeutics to control the invasive capacity (virulence) of pathogens and also to limit host ECM degradation (Hiemstra, 2002; Doyle et al., 2007; Craik et al., 2011).

The 15 Ln isoforms and 28 collagens have an enormous variability in structure that is difficult to estimate in terms of interactions with microorganisms. To date, most studies have been performed on the basis of purified proteins available from commercial sources. The results obtained in vitro have not, however, been verified in vivo in experimental animal models. The structural and compositional variability of ECM regarding Ln and collagen isoforms raises numerous questions related to their interactions with microorganisms. Do all Lns have similar binding affinity to all microorganisms? Do microorganisms that belong to a specific niche bind more efficiently to the abundant isoforms available in that particular tissue or organ? Has a specific ECM-binding capacity resulted in an increase of fitness to microorganisms under the evolutionary selection pressure?

Lns are large GPs with structural multidomain arrangements. The binding of bacteria should be directed to a certain domain, either to a single chain or to 3D structures involving several chains. In some cases, the globular domains (G1-G5) of the Ln α chains are known to interact with microorganisms (Fig. 7). It remains to investigate whether there is a conserved microorganism-binding region on the Ln α chains. The microorganism-binding domain of the Ln molecule is likely to be separated from the cellular adhesion domain (integrin and DG-binding domains) hence permitting the formation of a bridge between the pathogen and the host. On the other hand, adhesion of bacteria to the BM does not require distinct functional cellular adhesion domains, as Lns are arranged in the ‘cell-free’ BM network with other ECM molecules. In the case of collagen, several pathogens displayed a specificity of binding to particular isotypes, suggesting the recognition of specific domains. Intriguingly, collagens are also highly multifaceted proteins with a profusion of putative sites of interactions and yet, so far, only a few studies have defined the peptidic region targeted by pathogens and their specific surface receptors (Leo et al., 2010). An increased understanding of the nature of microorganism and Ln/collagen (protein–protein) interactions may lead to a future design of effective inhibitors and innovative antimicrobial therapies. Therefore, elucidation of additional biological structures will be important for the future expansion of the research field.

Is ECM binding to bacterial surface receptors only mediated by protein–protein interactions? Because Lns and collagens are glycosylated, the abundant amount of carbohydrate chains might be available for biological interactions. It has been demonstrated in several studies that deglycosylation of Ln by sodium metaperiodate treatment reduces the Ln binding to bacterial surface receptors (Kukkonen et al., 1993; Cameron, 2003; Terres et al., 2003; Atzingen et al., 2008). Collectively, these studies indicate the involvement of the Ln carbohydrate side chains. M. leprae phenolic glycolipid-1 (PGL-1) is involved in attachment to Schwann cells by recognizing the globular domains G1, G4 and G5 of Ln in vitro (Fig. 7) (Soares de Lima et al., 2005). In our laboratory, we recently observed that the H. influenzae PE binds to Ln and that heparin efficiently blocks the binding. (Hallstrom et al., 2011). Interestingly, the existence of heparin-binding regions has also been demonstrated on collagens (Koliakos et al., 1989; Zatterstrom et al., 2000) suggesting the existence of a similar mechanism, although their involvement in host–collagen interactions remains to be proven. It is probable that a conserved sequence pattern is used by several microorganisms and identification of such a motif is an interesting challenge.

It has recently been observed that heparin-binding peptides derived from the ECM have an in vitro antimicrobial activity and might be used to kill both Gram-negative and Gram-positive bacteria. Interestingly, the C-terminal heparin-binding regions of the Ln α1, α5 and β1 chains, as well as Vn and Fn heparin-binding peptides exhibit, on their own, an antimicrobial activity against E. coli, P. aeruginosa, S. aureus and E. faecalis (Andersson et al., 2004; Malmsten et al., 2006; Kobayashi & Yoshida, 2007). Ln peptides and their derivatives may therefore also have a potential use as novel antimicrobial agents.

Lns and collagens are known to bind several epithelial cell surface receptors (Fig. 3) and modulate signalling processes during cell–cell attachment/detachment (Leitinger & Hohenester, 2007; Marinkovich, 2007). Integrin-mediated cellular signalling is used by many microorganisms (direct binding to integrins) for internalization into host cells (Pizarro-Cerda & Cossart, 2006). In addition, Vn- and Fn-mediated binding of microorganisms to integrins also causes the internalization of microorganisms (Singh et al., 2010a; Bergmann et al., 2009). As microbes seem to use ECM-components for immune evasion and host invasion strategies, it would be interesting to study whether an alteration of Ln or collagen expression is induced by invading pathogens. Indeed, reports describing such processes have already been published for H. pylori (Chan et al., 2006).

In conclusion, it is known that numerous microorganisms bind to Lns and collagens to help them to better adhere/attach to host tissues. There is, however, a lack of knowledge on the precise role of ECM molecules in pathogenesis. Most of the recent studies have been performed in vitro to evaluate the role of these ECM components in bacterial invasion. These interesting findings must be extended to suitable in vivo models to better understand the exact mechanism of involvement of Ln and collagen in microbial pathogenesis. Finally, inhibitors of bacterial adhesins/receptors that bind ECM molecules may be potentially useful as future antimicrobial therapies.


This work was supported by grants from the Alfred Österlund, the Anna and Edwin Berger, the Gyllenstierna Krapperup, the Lars Hierta, the Greta and Johan Kock Foundations, the Swedish Medical Research Council, the Physiographical Society, the Cancer Foundation at the University Hospital in Malmö and Skane county council′s research and development foundation. We are thankful to Dr. Saraboji Kadhirvel, Department of Biology, Lund University, for helping in the preparation of figures by the pymol software.

Conflict of interest

None reported.