Accumulating evidence shows that many scavenger receptors (SR), including SR-A, MARCO and CD36, represent an important part of the innate immune defence by acting as pattern-recognition receptors, in particular against bacterial pathogens. Several SR are expressed on macrophages and dendritic cells, where they act as phagocytic receptors mediating non-opsonic phagocytosis of pathogenic microbes. Another important function of some SR is to act as co-receptors to Toll-like receptors (TLR), modulating the inflammatory response to TLR agonists. On bacteria, the SR ligands have commonly been reported to be lipopolysaccharide and lipoteichoic acid, but recent advances in the field indicate that bacterial surface proteins play a more important role as target molecules for SR than previously thought. Interestingly, recent data show that major pathogens, including Streptococcus pyogenes and the group B streptococcus, have evolved mechanisms to evade SR-mediated recognition. Moreover, intracellular pathogens, such as hepatitis C virus and Plasmodium falciparum, utilize the SR to gain entry into host cells, focusing interest on the importance of SR also in the molecular pathogenesis of infectious diseases. This review highlights the complex interactions between SR and pathogenic microbes, and discusses the role of these interactions in host defence and microbial pathogenesis.
The scavenger receptors (SR) comprise a large family of structurally diverse transmembrane cell surface glycoproteins restricted to macrophages (ΜΦ), dendritic cells (DC), endothelial cells and a few other cell types (Murphy et al., 2005). The concept of SR was initially put forward by Brown and Goldstein (Goldstein et al., 1979), who functionally defined the SR through their ability to bind modified low-density lipoproteins (LDL), e.g. oxidized LDL, but not native LDL. Today, the SR are divided into eight different classes (A–H) according to their overall multidomain structure (Murphy et al., 2005), and several other host molecules, including molecular chaperones and ECM proteins, have been identified as SR ligands (Plüddemann et al., 2007). Moreover, several reports show that some SR can also bind to unmodified lipoproteins. Most studies have been devoted to the role of SR-mediated endocytosis of modified LDL in foam cell formation and atherosclerosis, and the SR have also been implicated in Alzheimer's disease and to be involved in adhesion and tissue maintenance (Plüddemann et al., 2007).
The class A SR SR-A is expressed on most MΦ subpopulations and on DC and was the first SR to be cloned and characterized molecularly. It is a trimeric type II transmembrane glycoprotein with several distinct domains: a cytoplasmic tail, transmembrane, spacer, α-helical coiled coil, collagenous and C-terminal SR cystein-rich (SRCR) domains. There are two functional splice variants of SR-A (SR-AI and SR-AII), where SR-AII lacks most of the SRCR (Murphy et al., 2005). These two isoforms are collectively referred to as SR-A in this review.
The generation of SR-A knockout mice has enabled the demonstration that SR-A is important for the defence against experimental bacterial infection. Suzuki et al. (1997) showed that SR-A −/− mice are significantly more susceptible than their wild-type (WT) counterparts to infection with Listeria monocytogenes, with increased bacterial loads in liver and spleen. In subsequent studies, SR-A has been shown to protect mice against infection with S. aureus (Thomas et al., 2000), S. pneumoniae (Arredouani et al., 2006) and N. meningitidis (Plüddemann et al., 2009a). In addition, SR-A has also been implicated in protection against endotoxic shock (Haworth et al., 1997), but conflicting data have been reported (Kobayashi et al., 2000). The role of SR-A in protection against parasitic infection has also been investigated. Nogami et al. (1998) reported that SR-A −/− mice are more sensitive to infection with Plasmodium berghei compared with WT mice, but subsequent studies have failed to show a protective role of SR-A against malaria in the mouse model (Su et al., 2002; Cunha-Rodrigues et al., 2006; Inoue et al., 2006).
Interestingly, two clinically important bacterial pathogens, Streptococcus pyogenes and the group B streptococcus (GBS; Streptococcus agalactiae), have evolved mechanisms to evade recognition by SR-A on MΦ (Fig. 2) (Areschoug et al., 2008). S. pyogenes express the surface M protein, which is an important virulence factor that prevents opsonin-dependent phagocytosis of S. pyogenes in human blood by inhibiting complement deposition at the bacterial surface (Carlsson et al., 2005). Similarly, all strains of GBS express a major surface virulence factor, the polysaccharide capsule, that prevents complement deposition and phagocytosis of GBS by neutrophils in presence of serum (Wessels et al., 1989). In addition to their role in preventing opsonin-dependent phagocytosis by neutrophils, recent data show that the GBS polysaccharide capsule and the S. pyogenes M protein prevent recognition and non-opsonic phagocytosis of streptococci by SR-A on MΦ, most probably by masking the SR-A ligand(s) at the bacterial surface (Areschoug et al., 2008). This is the first example of a microbial strategy to evade recognition by SR-A.
In vivo evidence that MARCO is involved in host defence against bacterial infection was provided by Arredouani et al. (2004), who showed that MARCO −/− mice are significantly more sensitive to infection with S. pneumoniae compared with WT mice. Interestingly, a recent study suggests that pulmonary IFN-γ produced during T cell responses to influenza infection in mice inhibits bacterial clearance from lungs by alveolar MΦ, which was correlated with downregulation of MARCO on the MΦ (Sun and Metzger, 2008). Thus, downregulation of MARCO on alveolar MΦ may contribute to the enhanced susceptibility to secondary pneumococcal infection after an influenza infection.
In another interesting paper, it was reported that E. coli is able to engage CD16 (FcγRIII) on MΦ, which induces recruitment of SHP-1 to MARCO, resulting in inhibition of MARCO-mediated phagocytosis of these bacteria (Pinheiro da Silva et al., 2007). This may represent a novel microbial SR subversion mechanism, but the bacterial factor(s) interacting with CD16 has not been identified.
Other class A SR
In 2001, two independent research groups identified an SR-A-like receptor designated SRCL-I (Nakamura et al., 2001) or CL-PI (Ohtani et al., 2001), expressed by placenta, umbilical venous and arterial endothelium. Instead of an SRCR domain in the C-terminus, this receptor contains a C-type lectin domain. The role of SRCL-I/CL-PI as a PRR is unclear, but it has been shown to bind to heat-inactivated E. coli and S. aureus bacteria and Saccharomyces cerevisiae yeast particles (Ohtani et al., 2001). Interestingly, a recent study indicates that this receptor mediates non-opsonic phagocytosis of yeast zymosan by vascular endothelial cells, implicating a possible role of SRCL-I/CL-P1 in antifungal defence (Jang et al., 2009). In addition, SCARA5 is a recently identified class A SR expressed in epithelial cells that binds to heat-killed E. coli and S. aureus, indicating a role of this SR as a PRR (Jiang et al., 2006).
Class B SR
The first class B SR to be cloned and molecularly characterized was CD36, which is a type III transmembrane receptor with two transmembrane domains, an extracellular loop with multiple glycosylation sites and two short intracellular tails. It is expressed on MΦ, platelets, adipocytes and some endothelial and epithelial cells (Murphy et al., 2005). CD36 has been shown to be a sensor for LTA and a diacylated lipopeptide (MALP-2) (Hoebe et al., 2005). Hoebe et al. (2005) showed that CD36 acts as a co-receptor for TLR2 in responses to microbial diacylglycerides (Fig. 1), as CD36 was found to be essential for TLR2 responses to both MALP-2 and LTA. Further studies suggested that upon stimulation with diacylglycerides, CD36 associates with TLR2/6 heterodimers in lipid rafts at the cellular surface, an interaction believed to be crucial for signalling in response to these microbial molecules (Triantafilou et al., 2006). Importantly, two independent groups showed that CD36-deficient mice are more susceptible to infection with S. aureus compared with WT mice, demonstrating that CD36 is important for the defence against this bacterial pathogen (Hoebe et al., 2005; Stuart et al., 2005). Interestingly, CD36 expressed on MΦ acts as a phagocytic receptor (Fig. 1) for S. aureus, but in contrast to pure diacylglycerides, TLR2/6 responses to whole S. aureus bacteria is dependent on CD36-mediated internalization of S. aureus into the phagosome (Stuart et al., 2005).
While CD36 is an important PRR for Gram-positive pathogens, its role in recognition of Gram-negative bacteria is unclear. In contrast to S. aureus, CD36 is not a major phagocytic receptor for E. coli in MΦ (Stuart et al., 2005). Other studies, however, have shown that when expressed in HEK293 cells, CD36 can bind to several Gram-negative pathogens, including E. coli, Klebsiella pneumoniae and Salmonella typhimurium (Philips et al., 2005; Baranova et al., 2008). Indeed, CD36 may act as a sensor for LPS (Baranova et al., 2008), but there are as yet no in vivo data supporting a host-protective role of CD36 against infection with a Gram-negative bacterium.
In addition to its role as a PRR for bacterial pathogens, CD36 has also received substantial attention in Plasmodium falciparum malaria as a receptor for parasitized erythrocytes (PE), but it is somewhat unclear whether CD36 contributes primarily to innate host defence or rather the opposite, malaria pathophysiology (Serghides et al., 2003). CD36 was initially identified as a sequestration receptor for PE, contributing to the adhesion of PE to the venular endothelium within various organs, including heart, lung, liver and brain (Ockenhouse et al., 1991; Newbold et al., 1997; Serghides et al., 2003). Adherence of PE to the endothelium may cause microvascular occlusion, which is believed to contribute to the acute pathology of malaria (Serghides et al., 2003; Lovegrove et al., 2008). Baruch et al. (1996) identified the P. falciparum erythrocyte membrane protein-1 as the main ligand for CD36 on PE, an interaction suggested to be important for sequestration. However, CD36 has also been reported to act as a non-opsonic phagocytic receptor for PE on MΦ, mediating phagocytic clearance of PE (McGilvray et al., 2000; Patel et al., 2004). In a mouse model of malaria infection, CD36 −/− mice showed more severe and fatal malaria compared with WT mice when challenged with P. chabaudi chabaudi AS, providing in vivo evidence for a protective role of CD36 against malaria (Patel et al., 2007). Possibly, some parasite isolates may evade CD36-mediated clearance of PE by the expression of novel variable surface antigens, including polymorphic forms of P. falciparum erythrocyte membrane protein-1 (Serghides et al., 2006). CD36 has also been shown to act as a sensor for P. falciparum glycosylphosphatidylinositol (pfGPI), resulting in production of pro-inflammatory cytokines by MΦ (Patel et al., 2007). Interestingly, the pro-inflammatory response to pfGPI from MΦ is largely dependent on TLR2 (Krishnegowda et al., 2005). Thus, CD36 may cooperate with TLR2 in pro-inflammatory cytokine responses to pfGPI, similar to the proposed role in responses to diacylglycerides from Gram-positive bacteria. However, it is not clear whether the pro-inflammatory response elicited by pfGPI contributes primarily to malaria pathogenesis or host defence.
Of note, CD36 may also be involved in antifungal defence, as it was recently suggested to bind β-glucan and to contribute to in vivo protection in mice against experimental Cryptococcus neoformans infection (Means et al., 2009).
The second receptor in this class, SR-BI, has a similar loop structure to that of CD36 and consists of two splice variants designated SR-BI and SR-BII. SR-BI is primarily expressed on MΦ, DC, hepatocytes and steroidigenic tissue (Murphy et al., 2005). The first report describing the binding of a PAMP to SR-BI was presented by Vishnyakova et al. (2003), who showed that human SR-BI (designated CLA-1) binds pure LPS when expressed in HeLa cells. In a subsequent study, it was shown that CLA-1 mediates binding and uptake of both Gram-negative (E. coli and S. typhimurium) and Gram-positive (S. aureus and L. monocytogenes) bacteria in CLA-1 HeLa cells (Vishnyakova et al., 2006). However, peritoneal MΦ isolated from SR-BI −/− mice demonstrated only a 30% decrease in bacterial uptake when compared with WT MΦ. In addition, when expressed in HEK293 cells, human SR-BI also binds to Mycobacterium fortuitum (Philips et al., 2005). The binding and uptake of bacterial pathogens into host cells via CLA-1/SR-BI has been suggested to represent a bacterial virulence mechanism (Philips et al., 2005; Vishnyakova et al., 2006), but in vivo evidence for this hypothesis is lacking.
SR-BI has received significant attention as a possible target receptor for hepatitis C virus (HCV) entry into host hepatocytes (Fig. 1) (Dubuisson et al., 2008). Scarselli et al. (2002) first reported human SR-BI as a putative receptor for the E2 glycoprotein of HCV. In this study, the investigators used soluble E2 glycoprotein to pull down a putative receptor on human hepatoma cells following coimmunoprecipitation. Several subsequent studies using HCV pseudoparticles, which consist of unmodified HCV envelope glycoproteins assembled onto retroviral core particles, support a role of SR-BI in HCV entry (Bartosch et al., 2003; Bartosch et al., 2005; Lavillette et al., 2005; Kapadia et al., 2007; Dreux et al., 2009). However, the exact role of SR-BI in HCV entry is not understood. For example, HCV does not only rely on SR-BI for entry, but also other receptors such as CD81 and possibly members of the Claudin (CLDN) family (Dubuisson et al., 2008), but it is not known at what stage of entry SR-BI is required. Moreover, ectopic expression of SR-BI and CD81 in non-liver cell lines does not lead to HCV pseudoparticle entry, suggesting that additional receptors are required for HCV entry (Bartosch et al., 2003). Of note, SR-BI-mediated uptake of HCV has also been reported in human DC (Barth et al., 2008), resulting in cross-presentation of viral antigens. Thus, when expressed in DC, SR-BI may primarily mediate recognition and uptake of viral structures and contribute to host defence.
Whereas CD36 has been extensively studied in malaria pathogenesis, less is known about SR-BI in this context. However, two recent studies (Rodrigues et al., 2008; Yalaoui et al., 2008) report that SR-BI plays an important role in Plasmodium infection, as it promotes sporozoite invasion in hepatocytes and subsequent intracellular parasite development into exoerythrocytic forms and merozoites (Fig. 1). Thus, in malaria, SR-BI may mainly act as a target molecule for parasite infection, rather than in host defence. However, as described for HCV, CD81 is also required for hepatocyte entry of Plasmodium sporozoites (Yalaoui et al., 2008).
Other SR classes
Several other mammalian SR have been shown to recognize bacteria or isolated bacterial PAMP. The class E SR LOX-1 was originally identified as an endothelial cell-specific SR, but it can also be detected on MΦ, smooth muscle cells and platelets. It is a type II membrane protein with a short cytoplasmic N-terminus, a transmembrane domain, a neck domain and a C-proximal C-type lectin domain (Murphy et al., 2005). SREC-I is a class F SR expressed on endothelial cells and MΦ, and is composed of a large N-terminal extracellular domain containing 10 cysteine-rich repeats, five of which form EGF-like domains (Murphy et al., 2005). Both of these SR have been reported to bind heat-inactivated S. aureus and E. coli bacteria (Shimaoka et al., 2001; Jeannin et al., 2005). Interestingly, as described for SR-A and MARCO, LOX-1 and SREC-I can recognize bacterial surface proteins (Jeannin et al., 2005).
The class G SR SR-PSOX was originally identified as a transmembrane-type chemokine designated CXCL16 and is expressed in the endothelium, MΦ, DC, smooth muscle cells (Murphy et al., 2005), and has also been found in keratinocytes (Tohyama et al., 2007). It is a type I transmembrane protein with a chemokine domain adjoining a glycosylated mucin-like region (Murphy et al., 2005). When expressed in COS cells, SR-PSOX/CXCL16 mediates binding of S. aureus and E. coli, and antibodies to SR-PSOX/CXCL16 block uptake of bacteria by human DC, indicating that it may act as a phagocytic receptor (Shimaoka et al., 2003). Moreover, a study by Gursel et al. (2006) suggests that SR-PSOX/CXCL16 may act as an uptake receptor for bacterial CpG DNA, enhancing TLR9-mediated responses to this PAMP. Interestingly, a recent study suggests that the chemokine domain of SR-PSOX/CXCL16 may have direct antimicrobial activity against bacteria (Tohyama et al., 2007). Thus, SR-PSOX/CXCL16 might have dual roles in the innate immune defence, acting both as a PRR and a bactericidal host molecule.
The class H SR FEEL-1, and its paralogue FEEL-2, are very large type I transmembrane proteins containing Fasciclin (Fas-1), EGF-like, laminin-type EGF-like and a hyaluronan-binding link domains. Both FEEL-1 and FEEL-2 bind to heat-killed bacteria (Adachi and Tsujimoto, 2002), but their role as PRR remains unclear.
Bacterial surface protein ligands for SR
As outlined above, many SR bind to purified LPS and/or LTA, but it is unclear whether these molecules act as SR ligands at the surface of intact bacteria. Indeed, SR-A-dependent binding and non-opsonic phagocytosis of whole N. meningitidis bacteria is independent of surface LPS, as shown for a lipid A-deficient isogenic mutant of N. meningitidis (Peiser et al., 2002). Thus, SR-A is dependent on surface structures distinct from LPS on N. meningitidis for its ability to directly bind and phagocytose these bacteria. Peiser et al. (2006) identified several conserved meningococcal surface proteins, including NMB0278 and NMB1220, as ligands for SR-A, focusing interest on microbial surface proteins as major target molecules for SR-A. To investigate the role of the meningococcal surface proteins in SR-A-mediated recognition of these bacteria in vivo, Plüddemann et al. (2009a) constructed N. meningitidis mutants lacking surface expression of NMB0278 and NMB1220. In a murine model of meningococcal septicaemia employing WT and SR-A−/− mice, it was shown that SR-A contributes to recognition of the serogroup B WT strain MC58, with significantly lower health score and survival in SR-A −/− mice. In contrast, when mice were infected with an isogenic double-mutant (MC58-278-1220), lacking both NMB0278 and NMB1220, there was no difference in survival or health score between WT and SR-A −/− mice, indicating that SR-A lost the ability to recognize N. meningitidis when NMB0278 and NMB1220 are not surface-exposed (Plüddemann et al., 2009a). Thus, at least for N. meningitidis, the major ligands for SR-A are most likely surface proteins, rather than LPS. As shown for SR-A, MARCO also recognize meningococcal surface proteins (Plüddemann et al., 2009b). The binding specificity to N. meningitidis surface proteins varies somewhat between SR-A and MARCO, adding to the complexity of ligand recognition of these SR (Plüddemann et al., 2009b).
The ligand for SR-A in Gram-positive bacteria has been reported to be LTA (Dunne et al., 1994), and this interaction was shown to be affected by D-alanyl moieties on the LTA structure (Greenberg et al., 1996). However, a recent study on GBS identified a surface lipoprotein, designated Blr, as an SR-A ligand (Areschoug et al., 2008). The Blr protein contains a leucine-rich repeat region, indicating that it may be involved in protein–protein interactions, but the biological role of Blr is not known (Waldemarsson et al., 2006). The ability of SR-A to recognize GBS was significantly reduced by employing an isogenic Blr-negative mutant compared with the WT GBS strain. Thus, the Blr protein is a ligand for SR-A at the surface of GBS, but SR-A also recognize bacterial structures distinct from Blr, which might be LTA or other surface proteins (Areschoug et al., 2008).
In an interesting paper by Jeannin et al. (2005), it was shown that LOX-1 and SREC-I both bind to recombinant outer membrane protein A (OmpA) from K. pneumoniae. OmpA is a highly conserved surface protein in the Enterobacteriaceae family believed to be important for structural cell integrity and virulence (Jeannin et al., 2002). Employing an OmpA-negative mutant of E. coli, it was demonstrated that OmpA also acts as a target molecule for LOX-1 and SREC-I at the bacterial surface (Jeannin et al., 2005). Interestingly, OmpA activates antigen-presenting cells via TLR2, but it does not bind to TLR2 (Jeannin et al., 2000), suggesting that LOX-1 and SREC-I may act as sensors for OmpA in TLR2 signalling. Thus, LOX-1 and SREC-I may, as described for CD36, act as co-receptors for a TLR2 agonist. In addition, SREC-I has been reported to interact with the PorB protein of Neisseria gonorrhoeae (Rechner et al., 2007), which also is a TLR2 agonist (Massari et al., 2006), but in this case it was suggested that the binding of PorB to SREC-I represents a bacterial invasion mechanism into epithelial cells.
A growing body of evidence demonstrates that several SR represent an important part in the innate immune defence by acting as phagocytic receptors or as co-receptors to TLR, especially in responses to bacteria. Several SR have been shown to recognize the same pathogens, e.g. S. aureus, but it is often difficult to evaluate the individual contributions of different SR in recognition of specific pathogens. Moreover, the molecular basis for recognition is still poorly understood for many SR, as these receptors have rather broad ligand-binding specificity, including LPS, LTA, nucleic acids, β-glucan and proteins. However, the emerging picture of the molecular basis for the ability of SR to recognize pathogens indicates that microbial surface proteins are major ligands, at least for SR-A, MARCO, CD36, LOX-1 and SREC-I. Interestingly, this finding may have implications for vaccine development, as several SR have been shown to promote antigen presentation (Nicoletti et al., 1999; Barth et al., 2008; Harvey et al., 2008). Thus, identification of microbial surface proteins recognized by SR might represent a novel method to identify vaccine candidates against important pathogens, as targeting of SR may result in efficient delivery of protein antigens to antigen-presenting cells. Of note, many of the ligands for SR and other PRR are not only found on pathogens, but also on commensal microbes. Therefore, instead of PAMP, ‘microbe associated molecular patterns’ has sometimes been used as an alternative term to describe microbial ligands for PRR (Fritz et al., 2008). It is not known, however, if recognition of pathogens and commensals, respectively, by SR result in different host responses.
The importance of SR in the immune defence is further underlined by the finding that several pathogens utilize their virulence factors to avoid recognition and phagocytosis via SR. Studies of the interactions between SR and pathogenic microbes may therefore become important model systems for general mechanisms by which pathogens can subvert pattern recognition and establish infections. Thus, to fully understand the interplay between microbes and SR, it is of importance to use clinically relevant strains, rather than attenuated laboratory strains, which have commonly been used in these studies. Moreover, the interest of SR in the molecular pathogenesis of infectious diseases has increased significantly in recent years by the findings that major intracellular pathogens, such as HCV and Plasmodium spp., utilize SR-BI to enter host cells, mechanisms believed to represent important virulence mechanisms by these pathogens.
T.A. was supported by the Swedish Society for Medical Research, the Anna-Greta Crafoord Foundation for Rheumatological Research, the PE Lindahl Foundation, Swedish Government Support for Clinical Research (ALF), the Royal Physiographic Society in Lund and the trusts of Magnus Bergvall, Emil and Wera Cornell, Crafoord, Ollie and Elof Ericsson, Golje, Lars Hierta, Thelma Zoéga and Alfred Österlund. Work in the laboratory of S.G. was supported by the Medical Research Council, UK.