Centro de Biotecnologia, Instituto Butantan, São Paulo, SP, Brazil
Programa Interunidades em Biotecnologia, Instituto de Ciências Biomédicas, USP, São Paulo, SP, Brazil
Correspondence: Ana L.T.O. Nascimento, Centro de Biotecnologia, Instituto Butantan, Avenida Vital Brazil, 1500, 05503-900 São Paulo, SP, Brazil. Tel.: 5511 2627 9829; fax: 5511 3726 9233; e-mail: email@example.com
Leptospirosis is been considered an important infectious disease that affects humans and animals worldwide. This review summarizes our current knowledge of bacterial attachment to extracellular matrix (ECM) components and discusses the possible role of these interactions for leptospiral pathogenesis. Leptospiral proteins show different binding specificity for ECM molecules: some are exclusive laminin-binding proteins (Lsa24/LfhA/LenA, Lsa27), while others have broader spectrum binding profiles (LigB, Lsa21, LipL53). These proteins may play a primary role in the colonization of host tissues. Moreover, there are multifunctional proteins that exhibit binding activities toward a number of target proteins including plasminogen/plasmin and regulators of the complement system, and as such, might also act in bacterial dissemination and immune evasion processes. Many ECM-interacting proteins are recognized by human leptospirosis serum samples indicating their expression during infection. This compilation of data should enhance our understanding of the molecular mechanisms of leptospiral pathogenesis.
The genus Leptospira encompasses both pathogenic and saprophytic species. Pathogenic Leptospira are the etiological agents of leptospirosis, while saprophytic bacteria are environment free-living organisms. Leptospira are classified according to serovar status – more than 250 pathogenic serovars have been identified. Structural heterogeneity in lipopolysaccharide moieties seems to be the basis for the great degree of antigenic variation observed among serovars. Humans are accidental hosts that become infected through contact with wild or domestic animals or exposure to contaminated soil or water (Adler & de la Pena Moctezuma, 2010).
Leptospires enter the host mainly via intact sodden or damaged skin or mucosa. The initial phase exhibits wide-ranging and nonspecific symptoms such as fever, chills, headache, and myalgia. Leptospirosis can progress to a severe condition known as Weil's syndrome, corresponding to 5–15% of the reported cases. Another severe manifestation of the disease, the leptospirosis pulmonary hemorrhage syndrome, was first described in North Korea and China and has been increasingly reported worldwide (McBride et al., 2005).
Currently available veterinarian and human vaccines are based on inactivated whole cell or membrane preparations of pathogenic leptospires. However, these vaccines do not induce long-term protection against infection and do not provide cross-protective immunity against leptospiral serovars not included in the vaccine preparation. Several recombinant proteins have been tested as potential vaccines, but only a few are able to confer partial protection against challenge with virulent leptospires (Silva et al., 2007; Yan et al., 2010; Felix et al., 2011).
In the past 10 years, we have witnessed an increase in our knowledge of Leptospira and leptospirosis. Modern molecular approaches, such as genomics, proteomics, and mutagenesis, have produced a great amount of data. Several identified proteins have been described to interact with extracellular matrix (ECM) molecules. This review aims to assemble the current knowledge of leptospiral adhesins focusing on the host interactions for this pathogen and pointing out new trends for the characterization of these attachment proteins.
The ECM is an intricate network of macromolecules, whose primary role includes supporting and connecting cells and tissues. Two main classes of macromolecules constitute the ECM: (1) polysaccharide chains of glycosaminoglycans (GAGs) (Varki et al., 2009) and (2) fibrous proteins, such as collagen, elastin, fibronectin, and laminin (Alberts et al., 2007).
The GAGs are long unbranched polysaccharides constructed of repeating disaccharide units. They are highly negatively charged molecules that permit high hydration, viscosity, and interaction with other molecules (Varki et al., 2009). The most important GAGs for the ECM composition are hyaluronic acid, chondroitin sulfate, dermatan sulfate, keratin sulfate, heparan sulfate, and heparin.
Collagens are the most abundant proteins found in the animal kingdom and the major protein constituents in the ECM, providing tensile strength, regulating cell adhesion, supporting chemotaxis and migration. Laminin is a multi-adhesive protein that binds to numerous proteoglycans, type IV collagen, and certain receptors on the cells surface, such as integrins, an important class of cell adhesion molecules. Fibronectins are complex glycoproteins present in plasma and in ECM that display important multi-adhesive properties. Fibronectin contains different domains of high-affinity binding sites for collagen, heparin, gelatin, and integrins (via RGD motif), thereby attaching cells to the ECM (Chagnot et al., 2012). Plasma fibronectin circulating in the blood can bind to fibrin, activated platelets and enhance blood clotting (Lodish et al., 2000). The elastin results in the assembly of the monomeric precursor tropoelastin (Muiznieks et al., 2010). Elastin proportionally provides a degree of elasticity to the ECM, and further permits molecular interaction with ECM components such as proteoglycans, fibulin, and fibrillin (Hayes et al., 2011).
Leptospiral adhesion to ECM
The adhesion of Leptospira to cultured mammalian cells was reported, and the efficiency of attachment was correlated with bacterial virulence (Thomas & Higbie, 1990). The direct attachment of Leptospira interrogans to the ECM proteins laminin, collagen type I, collagen type IV, cellular and plasma fibronectin was reported indicating the existence of multiple adhesion molecules (Barbosa et al., 2006). Leptospiral binding to elastin and its precursor, tropoelastin, was described further stressing the presence of a broad spectrum of binding proteins (Lin et al., 2009). Moreover, the interaction of L. interrogans to the GAGs chondroitin sulfate and heparan sulfate has been reported (Breiner et al., 2009).
The first described leptospiral adhesin
The first leptospiral adhesin was a 36-kDa fibronectin-binding protein isolated from the outer sheath of a virulent variant of pathogenic leptospires (Merien et al., 2000). This adhesin was described to be expressed in virulent L. interrogans, totally lost during virulence attenuation and absent in the saprophytic strain Leptospira biflexa serovar Patoc I. Evidence of surface-exposed localization of this 36-kDa protein was provided by proteinase K treatment that abolished its fibronectin-binding activity. Table 1 and Supporting Information, Table S1 depict features of the published ECM-binding proteins, and Fig. 1 illustrates a model of membrane architecture featuring the leptospiral adhesins described by our group. The drawing illustrates an outer membrane (OM), a cytoplasmic or inner membrane (IM), and the peptidoglycan (PG) cell wall closely associated with the IM. Protein location was based on previous individually published results (Table S1).
Table 1. Features of identified ECM-binding proteins of Leptospira
Studies performed in our laboratory described the first leptospiral protein that exhibited attachment to purified laminin (Barbosa et al., 2006). The protein encoded by the gene LIC12906 was named Lsa24 for leptospiral surface adhesin of 24 kDa. Lsa24 binds strongly to laminin in a specific, dose-dependent, and saturable fashion. Complementary experiments showed that Lsa24 could partially block live leptospires from binding to immobilized laminin, evidencing the participation of this protein in attachment, but also implying the expression of other proteins involved in the interaction. This protein was previously described as a binding protein of both, factor H and factor H-related protein 1 of L. interrogans, named LfhA (leptospiral factor H-binding protein A; Verma et al., 2006). Subsequently, Stevenson et al. (2007) reported that this gene possesses five paralogs, distributed and limited to virulent leptospires (Stevenson et al., 2007). Because of their similarities to mammalian endostatins, these proteins were called Len proteins, for leptospiral endostatin-like. LenA protein (formerly called LfhA and Lsa24) and the homologs were named LenB, LenC, LenD, LenE, and LenF. LenB was found to bind human factor H, and all Len proteins showed laminin-binding capacity. In addition, LenB, LenC, LenD, LenE, and LenF exhibited affinity for fibronectin. However, lenB and lenE leptospiral mutant strains did not present an attenuated phenotype, suggesting that the corresponding proteins are not crucial for virulence (Murray et al., 2009a).
LipL32 was first described as an OM lipoprotein only found in pathogenic leptospires (Haake et al., 2000). The ECM-binding capacity of LipL32 has been shown by independent studies. One showed that recombinant LipL32 selectively binds to the individual components laminin and collagens I and V, where the carboxy-terminal is designated as the binding region (Hoke et al., 2008), while the other reported that LipL32 interacts with collagen IV and plasma fibronectin and confirmed that the carboxy-terminal portion was responsible for the binding (Hauk et al., 2008). Calcium binding to LipL32 was suggested to be important for the interaction with fibronectin (Tung et al., 2009). However, these results were not confirmed by other authors that showed that calcium binding to LipL32, although important to stabilize the protein structure, is not required to mediate interaction with host ECM proteins (Hauk et al., 2012).
However, despite LipL32 abundance in bacteria cell (Malmstrom et al., 2009) and its presence limited to pathogenic species, mutagenesis experiments with mutants lacking the lipL32 gene surprisingly showed that this protein is not essential for the bacterial survival during acute or chronic infection of L. interrogans (Murray et al., 2009b). In addition, LipL32 cellular position was recently reexamined, and evidence supporting subsurface location was presented in place of surface exposure (Pinne & Haake, 2013). Thus, it appears that more studies are needed to establish the role of LipL32 in mediating Leptospira–host interaction.
Leptospiral immunoglobulin-like proteins
Members of the Lig family of genes in L. interrogans that encode OM proteins with immunoglobulin-like repeats were identified (Matsunaga et al., 2003). LigA was first cloned and characterized by Palaniappan et al. (2002) that showed its expression during infection in vivo. The L. interrogans Lig proteins are present only in pathogenic species, expressed during infection, absent after attenuation by culture passage (Matsunaga et al., 2003), and upregulated under physiological osmolarity stimuli (Matsunaga et al., 2005). Based on this evidence, the Lig proteins were thought to be important for leptospirosis pathogenesis. The potential of the Lig proteins as vaccine candidates and diagnostic markers has also been proposed (Yan et al., 2009).
Choy et al. (2007) first described LigA and LigB as leptospiral ECM-binding proteins capable of interacting with fibronectin, laminin, and collagen IV and I. The adhesive regions were attributed to the unique regions of each protein, instead of the shared bacterial immunoglobulin-like (Big) repeats. Further studies have also shown LigB to interact with collagen type III (Choy et al., 2011). The high-affinity fibronectin-binding region was identified and found to comprise amino acid residues from 1014 to 1165. This portion includes part of the Big domain and a nonrepeated region in the variable C-terminal portion, which also binds laminin, but not collagen (Lin & Chang, 2008).
Although leptospiral strains with mutation of ligB does not attenuated the virulence in the hamster model of infection (Croda et al., 2008), heterologous expression of ligA and ligB genes using the saprophyte L. biflexa as a surrogate host improved the adhesion to fibronectin and to cultured cells. Although LigB has been shown to interact with collagen and elastin (Choy et al., 2007; Lin et al., 2009), L. biflexa lig mutants were not capable of binding collagens (type I and IV) or elastin more efficient than wild-type cells (Figueira et al., 2011). Thus, it appears that more investigations are needed to establish the function of Lig proteins in virulence.
Outer membrane protein A-like proteins
Outer membrane protein A (OmpA) is a major heat-modifiable surface-exposed protein in Escherichia coli and is one of the best-characterized OMPs (Confer & Ayalew, 2013). OmpA protein of E. coli and other enterobacteria has been reported to act as adhesin/invasin (Smith et al., 2007).
The first protein identified in pathogenic Leptospira containing a carboxy-terminal OmpA domain was a lipoprotein of 22 kDa, named Loa22 (Koizumi & Watanabe, 2003). The OmpA-like protein Loa22 was reported to be essential for the leptospiral virulence, as mutants lacking Loa22 expression resulted in attenuated virulence in an animal model of acute infection (Ristow et al., 2007). However, work performed in our laboratory showed that Loa22 had a modest interaction with the ECM molecules tested: laminin, collagen I, collagen IV, cellular and plasma fibronectin (Barbosa et al., 2006). Hence, although a potential leptospiral virulence factor, Loa22 is not an ECM-interacting protein.
Oliveira et al. (2011) identified a novel leptospiral hypothetical protein, encoded by the gene LIC10258, which has a putative OmpA-like domain comprising 504–583 amino acid residues in the carboxy-terminal region. The recombinant protein, named Lsa66, showed laminin and plasma fibronectin-binding properties. Moreover, antibodies in serum samples of leptospirosis cases recognize Lsa66, suggesting its expression during infection. The presence of this protein in Leptospira, as a fibronectin ligand, was further corroborated by other studies using high-throughput microarray strategies (Pinne et al., 2012).
Bioinformatics have predicted 184 coding sequences that appeared to be exported to the leptospiral surface, and thus, with the potential to participate in host–pathogen interactions (Nascimento et al., 2004). Most of these proteins are annotated as hypothetical with no assigned function. By picking coding sequences from this pool of sequences, our group has cloned and expressed various proteins that were identified as novel leptospiral adhesins. Some of these proteins exhibited only laminin-binding property when screened against several individual ECM components. These are as follows: Lsa27 (Longhi et al., 2009), Lsa20 (Mendes et al., 2011), Lsa25 and Lsa33 (Domingos et al., 2012), and Lsa26 (Siqueira et al., 2013). All these proteins are most probably surface-exposed, as assessed by indirect liquid-phase immunofluorescence or proteinase K accessibility. In addition to mediating ECM interactions through laminin, Lsa33, Lsa20, and Lsa26 are also plasminogen-interacting proteins and, hence, may participate to leptospiral invasiveness and dissemination (Mendes et al., 2011; Domingos et al., 2012). Moreover, Lsa25 and Lsa33 are capable to capture C4BP (Domingos et al., 2012), a regulator of the classical pathway complement system, and may have a role in leptospiral immune evasion.
Adhesins with multiple ECM-binding profile
Several leptospiral adhesins with broader spectrum binding to ECM have been described. LipL53 and Lsa21 were reported to interact with laminin, collagen type IV, cellular and plasma fibronectin and are thought to have roles in pathogenesis and virulence of leptospires (Atzingen et al., 2008; Oliveira et al., 2010). Lsa21 and LipL53 share some similarities with the Lig proteins and hence are virulence factor candidates: their protein expression seems to be restricted to highly virulent low-passage strains (Matsunaga et al., 2003; Atzingen et al., 2008; Oliveira et al., 2010). The gene encoding Lsa21 is restricted to pathogenic strains of L. interrogans, and the protein is upregulated at the transcriptional level by physiological osmolarity and temperature (Atzingen et al., 2008).
Proteomics studies of L. interrogans have identified a protein encoded by the gene LIC10314 that is differentially expressed under virulence conditions (Vieira et al., 2009). The recombinant protein, named Lsa63, was characterized as an adhesin that binds strongly to laminin and collagen type IV (Vieira et al., 2010). The protein Lsa63 is highly conserved among pathogenic strains and absent in L. biflexa. This adhesin is probably expressed during infection because the protein is recognized by human leptospirosis serum samples, mostly at the convalescent phase of the disease.
The proteins encoded by the genes LIC11087, LIC11360, and LIC11975 were genome annotated as hypothetical and predicted to be surface-exposed. We have evaluated the recombinant proteins, named Lsa30, Lsa23, and Lsa36, respectively, and shown that they are also ECM-interacting proteins with the ability to mediate attachment to laminin and plasma fibronectin (Souza et al., 2012; Siqueira et al., 2013). Furthermore, the proteins Lsa23 and Lsa30 may interfere with the complement cascade by interacting with C4BP regulator, and in the case of Lsa23, also with factor H. The proteins Lsa23, Lsa36, and Lsa30 are probably expressed during infection because antibodies present in serum samples of hamsters experimentally infected (Lsa30) or human leptospirosis sera (Lsa23 and Lsa36) can recognize these proteins. The OmpL1 has been described as a serological marker for the diagnosis of human leptospirosis and as a vaccine against the disease (Haake et al., 1999; Natarajaseenivasan et al., 2008). Although the gene is restricted to pathogenic serovars of Leptospira, its function is uncertain. Recently, it has been shown that OmpL1 has the ability to mediate attachment to laminin and plasma fibronectin. Like previously described adhesins, OmpL1 inhibited the binding of live Leptospira to the same ECM components (Fernandes et al., 2012). These are versatile leptospiral adhesins that may play a role in mediating adhesion to hosts.
Also included in this category of multiple ECM binding are the adhesins OmpL37 (LIC12263) and OmpL47 (LIC13050; Pinne et al., 2010), and the above discussed, LipL32, LigA/LigB and LenB, LenC, LenD, LenE, and LenF proteins (Table S1).
Lp95, an adhesin that activates cellular adhesion molecules
Cellular adhesion molecules (CAMs) are surface receptors present in eukaryotic cells that mediate cell–cell or cell–ECM interactions, and several microorganisms are reported to employ CAMs during their pathogenesis (Boyle & Finlay, 2003). The leptospiral proteins encoded by the genes LIC10365, LIC10507, LIC10508, and LIC10509 have been shown to be present in cellular infiltrate of liver and kidney of hamsters experimentally infected with virulent L. interrogans, suggesting that these proteins are expressed during infection. Most importantly, the recombinant proteins were capable of promoting the upregulation of ICAM-1 and E-selectin in human umbilical vein endothelial cells (HUVECs; Vieira et al., 2007; Gomez et al., 2008). Subsequently, we demonstrated for the first time that virulent L. interrogans, but also the saprophyte L. biflexa, induced an upregulation of CAMs in HUVECs, although the virulent one was more effective on activation (Atzingen et al., 2009). Furthermore, the same work showed a novel leptospiral protein encoded by LIC12690, named Lp95 (leptospiral protein of 95 kDa), that is capable of specifically activating E-selectin in HUVECs. In addition to CAM induction, Lp95 binds to laminin and fibronectin, being the carboxy-terminal portion responsible for the interaction. Thus, Lp95 with its dual activities could act as a protein that contributes to leptospiral attachment to ECM and cell receptors during Leptospira infection (Atzingen et al., 2009).
High-throughput screening microarrays
In addition to in silico identification of putative surface proteins and their production, such as recombinant proteins for the characterization of leptospiral ECM-binding proteins, recent work has employed high-throughput microarray strategy to identify novel host ligand-binding proteins (Pinne et al., 2012). These studies have employed OMP microarray containing 401 leptospiral proteins expressed in a cell-free expression system. These cited authors identified 15 candidates as fibronectin-binding proteins, but only six of them were confirmed (Table S1). Among them, only LIC10258, called Lsa66, had been previously characterized as a fibronectin adhesin (Oliveira et al., 2011). However, discrepancies between the array data and the published with purified recombinant proteins are evident. Several proteins with reported fibronectin adhesion properties were identified below the threshold, including the vastly studied LigA, LigB, and LipL32 and therefore admitted as nonbinding proteins (Pinne et al., 2012). LigA and LigB showed an intermediate binding value, while LipL32 a low one. Interestingly, Lsa33, which was shown to bind uniquely to laminin (Domingos et al., 2012), showed a binding value for fibronectin (c. 9.3) near the admitted threshold (10) (Pinne et al., 2012). LIC13050 (OmpL47), detected by the same group as fibronectin-interacting protein, had a binding value below 0. The other proteins included in the same group are as follows: LipL53 (Oliveira et al., 2010), Lsa21 (Atzingen et al., 2008), Lp95 (Atzingen et al., 2009), OmpL1 (Fernandes et al., 2012), and all Len proteins (Stevenson et al., 2007). Thus, although the array can rapidly identify ECM-binding proteins, the data will have to be further confirmed with the isolated proteins.
In recent years, major advances were made in the leptospiral field. Studies aiming to elucidate the pathogenesis of Leptospira based on available genome sequences, proteomics, and microarrays have been published, and the remarkable number of novel leptospiral ECM-interacting proteins recognized demonstrates this progress. This work assembles the current knowledge of leptospiral adhesins and describes their characterizations. These ECM-binding proteins are capable of mediating the attachment of Leptospira to mammalian hosts and of starting the process of invasion/colonization. Several laminin- and plasma fibronectin-binding proteins were identified indicating that leptospires have a redundant repertoire of adhesion molecules that are probably part of their invasion strategies. Proteins with exclusive or larger ligand capacity to ECM were described. In addition, there are multifunctional proteins that exhibit binding activities toward several target proteins, such as plasminogen/plasmin and regulators of the complement system, and may have several tasks in leptospiral pathogenesis. The presence of multiple adherence proteins in pathogens such as Yersinia and Neisseria has been discussed in terms of compensation mechanisms or the need for cooperative action in adhesin–ligand interactions (Kline et al., 2009; Mikula et al., 2013). This redundancy may explain why mutants lacking virulence factors do not change their phenotypes and point out the heterologous expression of virulent proteins in the saprophyte, L. biflexa serovar Patoc, as an attractive model. In vitro studies using leptospiral recombinant proteins represent only a model to evaluate possible tasks of these novel proteins that no function could be assigned during genome annotation. However, it is possible that some of these proteins might be expressed only in small amounts in Leptospira and therefore could not perform these interactions. In any event, these ECM-binding proteins will have to be further investigated to gain information on their precise role in leptospiral pathogenesis. These studies should contribute to identify a multicomponent acellular vaccine in addition to potential targets for antimicrobial therapy that may help fight leptospirosis.
This work was supported by FAPESP, CNPq, and Fundacao Butantan, Brazil; MLV, MVA, RO, RFD, GHS, NMS, ARFT, and LGF have scholarships from FAPESP and CAPES (Brazil). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We thank Dr. A. Leyva for helpful discussion and with English editing of the manuscript. The authors report that they have no conflict of interest.