Mammals have a family of four PGRPs, which were initially named PGRP-S, PGRP-L, and PGRP-Iα and PGRP-Iβ (for ‘short’, ‘long’ or ‘intermediate’ transcripts respectively), by analogy to insect PGRPs (Liu et al., 2001). Subsequently, the Human Genome Organization Gene Nomenclature Committee changed their names to Peptidoglycan Recognition Protein 1, 2, 3 and 4 (PGLYRP-1, PGLYRP-2, PGLYRP-3 and PGLYRP-4) respectively (Table 1). This terminology is also beginning to be used for mouse and all vertebrate PGRPs.
Mammalian PGLYRPs were initially hypothesized to function as pattern recognition receptors based on analogy to mammalian Toll-like receptors (TLRs) or signal-transducing insect PGRPs, and also based on initial computer predictions of transmembrane domains in mammalian PGLYRP-2, PGLYRP-3 and PGLYRP-4 (Liu et al., 2001). However, recent crystallographic (Guan et al., 2004a,b; 2005) and biochemical (Lu et al., 2006) data show that all four mammalian PGLYRPs are secreted, and are not transmembrane molecules. Functional studies also failed to reveal any role for mammalian PGLYRPs in TLR-induced cell activation, and instead, showed direct bacterial recognition and effector functions for mammalian PGLYRPs.
Mammalian PGLYRP-2 is an N-acetylmuramoyl-L-alanine amidase that hydrolyses the lactyl bond between the MurNAc and L-Ala in bacterial peptidoglycan (Fig. 2) (Gelius et al., 2003; Wang et al., 2003). PGLYRP-2 is constitutively produced in the liver and is secreted from the liver into blood (Zhang et al., 2005). This liver PGLYRP-2 and serum N-acetylmuramoyl-L-alanine amidase (that was earlier identified but not cloned) are the same protein encoded by the pglyrp2 gene (Zhang et al., 2005).
PGLYRP-2 is also expressed in the intestinal follicle-associated epithelial cells (Lo et al., 2003). PGLYRP-2 is not expressed in healthy human skin, but its expression in keratinocytes is induced by bacteria and cytokines (Wang et al., 2005). This induction is limited to epithelial cells, does not involve TLR2 or TLR4, and correlates with keratinocyte differentiation and stress responses that proceed through the activation of the p38 mitogen-activated protein kinase (Wang et al., 2005).
Some mammals express multiple splice forms of PGLYRP-2 that may have different expression and possibly multiple functions. For example, pigs have two PGLYRP-2 splice forms, short and long. They both have N-acetylmuramoyl-L-alanine amidase activity, and the long form has a similar expression to human PGLYRP-2, whereas the short form is constitutively expressed in several tissues, including bone marrow, intestine, liver, spleen, kidney and skin (Sang et al., 2005).
The in vivo role of PGLYRP-2 amidase activity could be threefold. First, it could function as a scavenger of proinflammatory peptidoglycan. In mammals, recognition of extracellular polymeric peptidoglycan occurs through TLR2 (Schwandner et al., 1999; Takeuchi et al., 1999; Yoshimura et al., 1999; Dziarski and Gupta, 2005), which likely requires glycan chains and stem peptides. Digestion of peptidoglycan with amidase reduces or eliminates cell-activating proinflammatory activity of polymeric peptidoglycan (Hoijer et al., 1997; Mellroth et al., 2003). Recognition of intracellular peptidoglycan fragments occurs through Nod1 and Nod2. Digestion of peptidoglycan with amidase would abolish its Nod2-activating capacity, which requires, at minimum, a muramyl dipeptide peptidoglycan fragment (Chamaillard et al., 2003; Girardin et al., 2003a). Second, PGLYRP-2, could have direct antibacterial activity or, similarly to lysozyme, could enhance the killing activity by antibacterial peptides. And third, PGLYRP-2 digestion of polymeric peptidoglycan could generate a Nod1-activating peptide, because the minimum structure that activates Nod1 is a peptide derived from the stem peptide of DAP-type peptidoglycan (without the glycan) (Chamaillard et al., 2003; Girardin et al., 2003b). Therefore, digestion of peptidoglycan with amidase could have a scavenger function to reduce proinflammatory activity of peptidoglycan, direct antibacterial function, or could generate Nod1-activating peptides, which would enhance antimicrobial responses.
Bactericidal and bacteriostatic activity
What is the function of the three remaining mammalian PGLYRPs? PGLYRP-1 was the first cloned mammalian PGLYRP (Kang et al., 1998). Initially, PGLYRP-1 was shown to be highly expressed in the bone marrow (Kang et al., 1998; Liu et al., 2001), and later it was shown to be almost exclusively present as a soluble protein in the granules of polymorphonuclear leucocytes (PMNs) (Liu et al., 2000; Tydell et al., 2002; 2006; Dziarski et al., 2003; Cho et al., 2005). This suggested its role in antibacterial activity, and indeed PGLYRP-1 knockout mice are more susceptible to infection with some Gram-positive bacteria, and their PMNs are defective in killing and digesting Gram-positive bacteria (Dziarski et al., 2003). The rate of bacterial uptake is not impaired in PGLYRP-1 knockout mice, and exogenous PGLYRP-1 added to bacteria and leucocytes has no effect on the rate of bacterial uptake by PMNs, which indicates that PGLYRP-1 is not an opsonin. However, PGLYRP-1 restores the normal killing and digestion of PMNs from PGLYRP-1 knockout mice (Dziarski et al., 2003). Although in initial studies purified mouse or human PGLYRP-1 were only bacteriostatic (Liu et al., 2000; Dziarski et al., 2003; Cho et al., 2005), the latest results demonstrate that human PGLYRP-1 is bactericidal and that the cidal activity requires Ca2+ (Lu et al., 2006); the earlier preparations were not bactericidal because they did not contain Ca2+. Bovine PGLYRP-1, purified from leucocyte granules, is also directly bactericidal for both Gram-positive and Gram-negative bacteria (Tydell et al., 2002; 2006). Therefore, PGLYRP-1 is a part of PMNs’ antibacterial arsenal, but it is also present in milk (Kappeler et al., 2004) and in intestinal M cells (Lo et al., 2003).
The functions of the remaining two mammalian PGLYRPs, PGLYRP-3 and PGLYRP-4, until very recently, were still unknown. The initial results showed very selective expression of PGLYRP-3 and PGLYRP-4 in esophagus (out of 76 tissues and cell types tested) (Liu et al., 2001), and computer analysis initially identified them as transmembrane molecules (Liu et al., 2001), leading to the hypothesis that they could function as cell surface receptors. However, recent crystallization of the C-terminal half of PGLYRP-3 demonstrated the lack of transmembrane domain, suggesting that PGLYRP-3 and PGLYRP-4 may be secreted (Guan et al., 2004a,b). Indeed, the most recent biochemical results revealed that both PGLYRP-3 and PGLYRP-4, as well as PGLYRP-1, are secreted primarily as disulphide-linked homodimers (Lu et al., 2006). Moreover, when PGLYRP-3 and PGLYRP-4 are coexpressed in the same cells, they are exclusively secreted as PGLYRP-3:4 heterodimers (Lu et al., 2006).
Because pglyrp3 and pglyrp4 genes are located in the epidermal differentiation complex gene cluster on chromosome 1q21 (Liu et al., 2001; Mathur et al., 2004; Sun et al., 2006) that encodes genes co-ordinately expressed in specialized epithelial cells, such as keratinocytes, PGLYRP-3 and PGLYRP-4 were also likely to be expressed in differentiated keratinocytes and other differentiated epithelial cells. Indeed, human PGLYRP-3 and PGLYRP-4 proteins are selectively expressed in tissues that come in contact with the environment (Lu et al., 2006): in the skin epidermis, hair follicles, sebaceous glands and eccrine sweat glands; in the eye’s ciliary body (which produces aqueous humour that fills anterior and posterior chambers of the eye); in the eye’s corneal epithelium; in the mucus-secreting cells of the main salivary (submandibular) gland and in mucus-secreting glands in the throat (both mucus-secreting glands selectively express PGLYRP-4, but not PGLYRP-3); in the tongue and esophagus in squamous epithelial cells; in the stomach in acid-secreting Parietal cells (PGLYRP-3) and glycoprotein-secreting neck mucous cells (PGLYRP-4); and in the small and large intestine in the columnar absorptive cells, but not in mucus-secreting goblet cells and not in the crypts in Paneth cells, which produce antimicrobial peptides (Lu et al., 2006). Mouse PGLYRP-3 and PGLYRP-4 are also expressed in the intestinal tract and salivary glands (Mathur et al., 2004). Bacteria and their products increase the expression of PGLYRP-3 and PGLYRP-4 in keratinocytes (Lu et al., 2006) and oral epithelial cells (Uehara et al., 2005), likely through activation of TLR2, TLR4, Nod1 and Nod2.
The latest results demonstrate that human PGLYRP-3, PGLYRP-4, PGLYRP-3:4, as well as PGLYRP-1, are bactericidal for pathogenic (Listeria monocytogenes and S. aureus) and non-pathogenic (Bacillus and Lactobacillus) Gram-positive bacteria (Lu et al., 2006). They are markedly less bactericidal or bacteriostatic for all other bacteria tested, including normal flora Gram-positive bacteria (Micrococcus luteus, Staphylococcus epidermidis and several Streptococci) and Gram-negative bacteria.
PGLYRP-3 and PGLYRP-4 are active in vivo on mucous membranes (which is the site of their production) in an intranasal S. aureus lung infection in mice (Lu et al., 2006). PGLYRPs are bactericidal (kill 99% of bacteria) at 0.1–1 µM, and thus are more active than most antibacterial peptides, such as defensins or magainin, but less active than phospholipase A2, which is the most active human bactericidal peptide (Lu et al., 2006). PGLYRPs act synergistically with bacteriolytic enzymes (lysozyme and lysostaphin) and antimicrobial peptides (defensins and phospholipase A2) (Cho et al., 2005; Lu et al., 2006). This synergism should greatly enhance their effectiveness in vivo, because PGLYRPs and these antimicrobial enzymes and peptides are often found in the same secretions or on the same cell surfaces that come in contact with bacteria.
Spectrum of activity and specificity of mammalian PGLYRPs
Human PGLYRP-1, PGLYRP-3, PGLYRP-4 and PGLYRP-3:4 are bactericidal for Bacillus, Lactobacillus, L. monocytogenes and S. aureus, which are both non-pathogenic and pathogenic Gram-positive bacteria. They are markedly less bactericidal or bacteriostatic for all the remaining Gram-positive bacteria tested (S. epidermidis, M. luteus, Streptococcus pyogenes, Streptococcus agalactiae, Enterococcus faecalis) and for Gram-negative bacteria (E. coli, Enterobacter cloacae, Proteus vulgaris), but they have no effect on fungi (Candida albicans, Saccharomyces cerevisiae). Each PGLYRP, however, has a different spectrum of bactericidal activity (Lu et al., 2006). These results suggest that normal flora bacteria have developed resistance to bactericidal mechanisms constitutively present in the skin, eyes and mucous membranes (such as PGLYRP-3 and PGLYRP-4), and can colonize these areas. Bacteriostatic effect of PGLYRPs on normal flora bacteria makes perfect sense for host physiology – normal flora bacteria are not killed, but their overgrowth is limited. Non-pathogenic not normal flora bacteria (transient flora, such as Bacillus and Lactobacillus) are uniformly sensitive to killing by PGLYRPs. L. monocytogenes, a pathogen that does not infect skin and mucous membranes, is also highly sensitive. S. aureus, a pathogen that often infects skin, has intermediate sensitivity to killing by PGLYRPs, with some strains highly sensitive and some less sensitive, whereas S. pyogenes, a pathogen that frequently infects throat and skin and has high carrier rate in the throat, is more resistant to killing by PGLYRPs. These results demonstrate how normal flora adapt to their environment and how successful pathogens evade the immune system at the site of infection.
Mouse PGLYRP-1 seems to have similar spectrum of activity to human PGLYRP-1, whereas bovine PGLYRP-1 is bactericidal for both Gram-positive (L. monocytogenes, S. aureus) and Gram-negative (Salmonella typhimurium) bacteria, and also has some microbicidal activity against a fungus, Cryptococcus neoformans (Tydell et al., 2002; 2006). The broader spectrum of cidal activity of bovine PGLYRP-1 may reflect a true difference between the human and bovine orthologues, or may simply reflect a difference in the protein purification methods and assay conditions.
All mammalian PGLYRPs tested so far bind peptidoglycan and bacteria (Kang et al., 1998; Werner et al., 2000; Liu et al., 2000; 2001; Mellroth et al., 2003; Lu et al., 2006; Tydell et al., 2006). Mammalian PGLYRPs bind to both Gram-positive and Gram-negative bacteria (Liu et al., 2000; 2001; Cho et al., 2005; Lu et al., 2006; Tydell et al., 2006). Binding to peptidoglycan may not be always responsible for binding to intact bacteria, because in Gram-negative bacteria peptidoglycan is located underneath the outer membrane and is not easily accessible on the bacterial surface (although penetration of PGLYRPs through the outer membrane cannot be excluded). PGLYRPs also bind to fungi, which do not have peptidoglycan. Therefore, PGLYRPs may also bind to other polymers, such as lipoteichoic acid and lipopolysaccharide (Liu et al., 2000; Lu et al., 2006; Tydell et al., 2006). However, human and mouse PGLYRPs have the highest affinity for peptidoglycan and much lower for lipoteichoic acid and lipopolysaccharide (Liu et al., 2000; Lu et al., 2006), whereas bovine PGLYRP-1 seems to have high affinity for lipoteichoic acid and lipopolysaccharide (Tydell et al., 2006). It is not clear, however, whether these other ligands bind to the peptidoglycan binding groove (see next section) or to another portion of the PGLYRP molecule, such as the hydrophobic region on the opposite side of the molecule.
Also, the binding specificities of the amidase-active PGRPs may be broader than their enzymatic specificity, which is reminiscent of the vertebrate lysozyme. This lysozyme has highly specific lytic activity for peptidoglycan’s polysaccharide backbone, but it nevertheless also binds to other compounds, such as Gram-negative bacteria and their lipopolysaccharide component (Imoto et al., 1972; Ohno and Morrison, 1989).
Molecular basis of recognition and function of mammalian PGLYRPs
All PGRPs have a ∼165-amino-acid-long PGRP domain. In PGLYRP-1, one PGRP domain comprises most of the entire ∼190-amino-acid sequence, and in PGLYRP-2, one C-terminal PGRP domain comprises ∼1/3 of the sequence. PGLYRP-3 and PGLYRP-4 both have two PGRP domains. Each PGRP domain has one ligand binding site (Guan et al., 2004b). Thus, PGLYRP-3 and PGLYRP-4 monomers and dimers have two and four ligand binding sites respectively, whereas PGLYRP-1 and PGLYRP-2 monomers and dimers have one and two identical ligand binding sites respectively (Fig. 2). However, because these PGRP domains are not identical (have 37–43% identity), the fine binding specificity or affinity of each PGRP domain in each PGLYRP molecule is likely different. The diversification of PGLYRPs’ specificities is then further increased by formation of PGLYRP-3:4 heterodimers, which have four different binding sites. This way the host can fine-tune the specificities of PGLYRPs by expressing PGLYRP-3 and PGLYRP-4 either in the same or in separate cells, to form hetero- or homodimers respectively. In addition, PGLYRPs have hydrophobic domains on the opposite side of the molecule from the ligand binding groove, which were previously hypothesized to interact with signal transduction molecules (Kim et al., 2003). In mammalian PGLYRPs, however, these hydrophobic domains may either play a role in interaction of PGLYRPs with bacteria, or in formation of dimers.
Crystallographic analysis of human PGLYRP-1 and C-terminal PGRP domain of PGLYRP-3, as well as insect PGRP-LB, -SA and -LCa, show that all these PGRPs have a ligand binding groove that binds peptidoglycan and is specific for muramyl-tripeptide (Kim et al., 2003; Chang et al., 2004; 2005; Guan et al., 2004a, b; 2005; Reiser et al., 2004). It can accommodate a larger structure, such as GlcNAc-MurNAc-tetrapeptide, but it does not bind muramyl-dipeptide or a peptide without MurNAc (Kumar et al., 2005; Swaminathan et al., 2006). These results are consistent with the specificity of human PGLYRP-2 for muramyl-tripeptide, which is the minimum peptidoglycan fragment hydrolysed by PGLYRP-2 (Wang et al., 2003), and with the specificity and high affinity (Kd = 13 nM) of murine PGLYRP-1 for uncross-linked polymeric peptidoglycan, but not muramyl-dipeptide or pentapeptide (Liu et al., 2000). The high-affinity binding of peptidoglycan to PGLYRP is achieved by burying both the peptide and MurNAc portions of peptidoglycan in a deep cleft that completely excludes solvent (Guan et al., 2004b).
Human PGLYRP-1 and C-terminal fragment of PGLYRP-3 bind muramyl-tetrapeptide and muramyl-pentapeptide with higher affinity than muramyl-tripeptide (Kumar et al., 2005). C-terminal fragment of PGLYRP-3 has a preference for binding the Lys-type over the DAP-type peptidoglycan, whereas PGLYRP-1 binds DAP-type peptidoglycan with higher affinity than Lys-type peptidoglycan (Kumar et al., 2005; Swaminathan et al., 2006). Moreover, both human and insect PGRPs employ a dual strategy for discrimination among different types of peptidoglycan, based on detection of Lys or DAP in the stem peptide and the type of peptide cross-bridge (Swaminathan et al., 2006). Discrimination between Lys- and DAP-type peptidoglycan is based on two amino acids in the peptidoglycan binding groove in position 236 and 237 in PGLYRP-3. The validity of this model is verified by mutations in these positions that can change the specificity of the binding from Lys to DAP or DAP to Lys (Swaminathan et al., 2006). This allows prediction of binding specificity of various PGRP domains for Lys- or DAP-type peptidoglycan. Detection of peptide-cross-linked peptidoglycan would require engagement of two peptidoglycan binding sites in two PGRP domains, which could be accomplished by PGLYRPs with two PGRP domains and/or by dimeric PGLYRPs, which is consistent with recent demonstration of dimeric PGRPs in mammals (Lu et al., 2006) and insects (Mellroth et al., 2005).
Although the amino acids that contact peptidoglycan in the C-terminal portion of human PGLYRP-3 have been established, there is likely to be considerable variation in the fine specificity of different PGRPs, because both contact and non-contact residues in and around peptidoglycan binding groove are quite variable and less than 50% conserved among PGRPs (Kim et al., 2003; Guan et al., 2004a,b). Therefore, this structural variation may correspond to different ligand specificities of different PGRPs.
Amidase-active PGLYRP-2 have a conserved Zn2+ binding site in the peptidoglycan binding groove, which consists of two His, one Tyr and one Cys (C530 in human PGLYRP-2), and which is also present in all amidase-active insect PGRPs and in homologous bacteriophage type 2 amidases. In non-amidase PGLYRPs, this Cys is substituted with Ser. PGLYRP-2 have at least four other amino acids that are required for their activity, because mutations in these amino acids greatly reduce or abolish their amidase activity (Gelius et al., 2003; Wang et al., 2003).
Mechanism of antibacterial activity
To determine the mechanism of bactericidal activity of PGLYRPs, their effects on bacteria were compared with the effects of membrane-permeabilizing antibacterial peptides, peptidoglycan-lytic enzymes, and inhibitors of peptidoglycan biosynthesis. Vertebrate antibacterial peptides (e.g. defensins) kill bacteria by traversing bacterial cell wall and permeabilizing bacterial cytoplasmic membrane. Therefore, kinetics of membrane permeabilization by antibacterial peptides correlates with the kinetics of bacterial killing. Peptidoglycan-lytic enzymes kill bacteria by destroying the physical integrity of peptidoglycan, which causes osmotic lysis, because the function of peptidoglycan in bacterial cell wall is to counteract the high osmotic pressure of bacterial protoplast. This osmotic lysis results in rapid membrane permeabilization, and the kinetics of membrane permeabilization correlates with the kinetics of bacterial killing. Antibiotics (such as penicillin) that target cell wall kill bacteria by inhibiting peptidoglycan synthesis. However, in contrast to peptidoglycan-lytic enzymes, penicillin does not cause early permeabilization of bacterial membranes, because penicillin-treated bacteria only die when they start to grow and are unable to synthesize peptidoglycan. Therefore, in penicillin-killed bacteria membrane permeabilization is substantially delayed until the bacteria have grown without synthesizing peptidoglycan. The kinetics of bacterial killing and membrane permeabilization by PGLYRPs resembles those by penicillin, thus suggesting that PGLYRPs target cell wall peptidoglycan rather than cell membranes (Lu et al., 2006).
When bacteria lose their cell wall peptidoglycan due to digestion with an enzyme or due to inhibition of peptidoglycan synthesis by antibiotics, such as penicillin, they can be kept alive in the protoplast medium containing 0.75 M sucrose, until they rebuild their cell walls. Consistent with this model, killing of bacteria and membrane permeabilization by peptidoglycan-lytic enzymes, penicillin, and also by PGLYRPs are prevented in the protoplast medium containing 0.75 M sucrose, because sucrose protects the cells from osmotic lysis (Lu et al., 2006), which further supports that PGLYRPs target cell wall peptidoglycan rather than cell membranes. By contrast, neither membrane permeabilization nor killing by membrane-damaging antibacterial peptides is prevented in the protoplast medium.
Therefore, the mechanism of bacterial killing by PGLYRPs is different from the mechanism of killing by peptidoglycan-lytic enzymes and membrane-permeabilizing antibacterial peptides, and it most resembles the effect of antibiotics that inhibit peptidoglycan synthesis. Antibiotics inhibit peptidoglycan synthesis by two basic mechanisms: (i) they either block the active site of peptidoglycan-synthesizing enzymes (e.g. β-lactams); or (ii) they bind to metabolic precursors of peptidoglycan and prevent their use in peptidoglycan synthesis (e.g. vancomycin, mersacidin and actagardine) (Lu et al., 2006). Because PGLYRPs avidly bind to peptidoglycan or its fragments (e.g. GlcNAc-MurNAc-pentapeptide), they may exert their antibacterial effect by inhibiting peptidoglycan synthesis through binding to peptidoglycan biosynthetic precursors or by limiting access to peptidoglycan or its precursors by enzymes that participate in cell wall synthesis. This mechanism of action of PGLYRPs is consistent with the ability of exogenous peptidoglycan to inhibit bactericidal activity of PGLYRPs (Lu et al., 2006).
Another possible mechanism of bacterial killing by PGLYRPs is enzymatic digestion of peptidoglycan by PGLYRPs, which would also be consistent with inhibition of killing by exogenous peptidoglycan. This is a less likely mechanism for several reasons (Lu et al., 2006). First, as already mentioned, peptidoglycan-lytic enzymes kill bacteria by destroying physical integrity of peptidoglycan, which causes immediate osmotic lysis of bacteria, manifested as permeabilization of cytoplasmic membrane that correlates with bacterial killing. PGLYRPs, however, do not cause early permeabilization of cytoplasmic membranes. Second, PGLYRP-1, PGLYRP-3 and PGLYRP-4, in contrast to PGLYRP-2, do not have any detectable amidase activity (Wang et al., 2003; Lu et al., 2006), likely because they do not have a conserved Cys, which corresponds to C530 in PGLYRP-2 and which is required for Zn2+ binding and amidase activity. Third, no significant bacteriolytic activity of PGLYRP-1, PGLYRP-3 and PGLYRP-4 was detected under the conditions that killed 99% of bacteria. And fourth, although bovine PGLYRP-1 does have bacteriolytic activity, this activity does not correlate with bactericidal activity, because the former is heat-labile and the latter is heat-stable (Tydell et al., 2006). However, it is still possible that peptidoglycan hydrolytic activity could be responsible for the bactericidal activity of PGLYRPs, if the binding of PGLYRPs to peptidoglycan was of high affinity (essentially irreversible), but the rate of hydrolysis was very slow. Such a slow rate of peptidoglycan hydrolysis is exhibited by insect PGRP-SA, which is a D-Ala-carboxypeptidase, and which, similarly to mammalian PGLYRP-1, PGLYRP-3 and PGLYRP-4, does not have the conserved Zn2+ binding Cys (Chang et al., 2004). Such a carboxypeptidase activity of PGLYRPs would not have to be bacteriolytic, as it could convert the peptidoglycan biosynthetic precursor, GlcNAc-MurNAc-pentapeptide, to GlcNAc-MurNAc-tetrapeptide, and make it unsuitable for further steps in the biosynthetic pathway, such as transpeptidation. These possibilities will have to be explored experimentally.
Finally, it is also possible that bactericidal activity of PGLYRPs is unrelated to their peptidoglycan binding ability. This mechanism is the least likely, although it is consistent with the inhibition of PGLYRPs binding to bacteria (Lu et al., 2006) and killing of bacteria (Tydell et al., 2006) by lipoteichoic acid and lipopolysaccharide.