Human SAP is a novel peptidoglycan recognition protein that induces complement- independent phagocytosis of Staphylococcus aureus

Abstract

The human pathogen Staphylococcus aureus is responsible for many community-acquired and hospital-associated infections and is associated with high mortality. Concern over the emergence of multidrug-resistant strains has renewed interest in the elucidation of host mechanisms that defend against S. aureus infection. We recently demonstrated that human serum mannose-binding lectin (MBL) binds to S. aureus wall teichoic acid (WTA), a cell wall glycopolymer, a discovery that prompted further screening to identify additional serum proteins that recognize S. aureus cell wall components. In this report, we incubated human serum with 10 different S. aureus mutants and determined that serum amyloid P component (SAP) bound specifically to a WTA-deficient S. aureus ΔtagO mutant, but not to tagO-complemented, WTA-expressing cells. Biochemical characterization revealed that SAP recognizes bacterial peptidoglycan as a ligand and that WTA inhibits this interaction. Although SAP binding to peptidoglycan was not observed to induce complement activation, SAP-bound ΔtagO cells were phagocytosed by human polymorphonuclear leukocytes in an Fcγ receptor-dependent manner. These results indicate that SAP functions as a host defense factor, similar to other peptidoglycan recognition proteins and nucleotide-binding oligomerization domain (NOD)-like receptors.

Introduction

Innate immunity constitutes the first line of host defense and recognizes evolutionarily conserved molecular patterns of pathogenic microbes using pattern recognition receptors (PRRs) (1). Based on their localization, PRRs are classified as either cell-associated receptors including the Toll-like receptors (2) and scavenger receptors (3) or fluid-phase molecules (4). Fluid-phase molecules, such as collectins, ficolins and pentraxins, constitute the humoral arm of the innate immune system and are generally believed to represent the functional ancestors of antibodies (5).

The pentraxin family can be divided into two subclasses, the short-chain pentraxins, which include C-reactive protein (CRP) and serum amyloid P component (SAP), and the long-chain pentraxins, which contain an additional amino-terminal domain. PTX3, a long-chain pentraxin, is produced by macrophages and myeloid dendritic cells in response to proinflammatory stimuli (6–8). Whereas human CRP and mouse SAP are major acute phase proteins (9), Human SAP is a constitutive protein in blood (10). SAP, named for its universal presence in amyloid deposits, is the precursor of amyloid P component in tissue, where it may promote the development of pathogenic amyloid deposits and prevent their degradation. Both SAP and CRP have been reported to recognize numerous pathogenic bacteria and fungi and activate the classical complement pathway via C1q (8, 11). SAP is a conserved, circulating protein that exhibits calcium-dependent binding to various ligand molecules on the surface of microbial pathogens (4). Both CRP and SAP assemble into pentameric ring structures, which are arranged with the ridge helix of each subunit on one face and microbial ligand binding sites on the opposite side (12, 13). Previous reports found that members of the pentraxin family interact with cell-surface Fcγ receptors (FcγRs) and activate leukocyte-mediated phagocytosis (14, 15). More recently, the structural basis for the binding of pentraxins to FcγRs and the mechanism of activation of FcγR-mediated phagocytosis and cytokine secretion were reported (16). Notably, because pentraxins are broadly conserved, these proteins are thought to function as ancient mediators of immunity (17).

Staphylococcus aureus is a common human pathogen responsible for hospital-associated and community-acquired infections with complications such as wound infection, bacteremia and sepsis. Recent studies have shown how this pathogen has evolved mechanisms to evade host innate immune responses and how it has acquired numerous virulence factors, which contribute to the diversity and severity of staphylococcal diseases (18). Any effort to respond to these challenges requires an examination of the molecular cross-talk between S. aureus and its host.

Like most Gram-positive bacteria, S. aureus incorporates peptidoglycan (PGN) and carbohydrate-based glycopolymers, such as wall teichoic acid (WTA) and lipoteichoic acid (LTA), into its cell envelope (19). PGN, an essential component of the bacterial cell wall, is composed of polymeric sugar chains with alternating 1,4-β-linked N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) residues; the MurNAc residues within PGN are cross-linked by shorts peptides (20). In S. aureus, WTA is also cross-linked at its C6 position of the MurNAc unit of PGN. The disaccharide, N-acetylmannosamine (ManNAc)-β-(1,3)-GlcNAc, is connected to a polymer of 11–40 repeating units of ribitol-phosphate via two glycerol-phosphates. The hydroxyl groups on the ribitol-phosphate repeats are modified with cationic d-alanine esters and GlcNAc residues (19). Unlike WTA, LTA polymers are assembled from repeating units of glycerol- phosphate and are connected to glycolipids. Though not essential for bacterial viability, WTA is necessary for adherence of S. aureus to nasal epithelial cells (21). Recent studies have demonstrated that the binding of these three glycopolymers to host PRRs activates the innate immune system and induces the release of inflammatory molecules (22). However, because of the challenges involved in purifying components of the bacterial cell wall from a complex mixture, the ligands for many host PRRs have not been identified. In addition, the diversity of molecular and structural differences among bacterial species and strains further complicates the recognition of ligand-receptor relationships (19). Despite recent advances in analytical techniques used in glycobiology, biochemical knowledge of the composition and structure of bacterial cell walls remains limited.

The complement system, which is activated by serum fluid-phase molecules, performs important functions in host defense, such as opsonization of pathogenic microbes, production of peptide mediators for phagocyte recruitment and generation of membrane-attack complexes (MAC) for killing and lysis of bacteria (4, 23). Because the processes of complement-mediated opsonophagocytosis and polymorphonuclear leukocyte (PMN)-mediated phagocytosis are crucial for innate immunity and clearance of pathogens and apoptotic cells, deficiencies in complement components are often associated with inflammatory and immunological diseases (23).

Previously, our group (24) and Nadesalingam et al. (25) have shown that human mannose-binding lectin (MBL) binds to PGN of S. aureus, activates the lectin complement pathway and promotes chemokine production by macrophages. To further characterize the binding of MBL to S. aureus, we screened S. aureus cell wall-deficient mutants and discovered that purified MBL/MBL- associated serine protease (MASP) complex binds to wild-type S. aureus but not to a WTA-deficient mutant (ΔtagO); this result suggested that WTA, specifically, is the ligand of MBL (26). This study supports the possibility that commercially available PGNs may be contaminated with bacterial WTA. Notably, the MBL/MASP complex does not bind to WTA in the serum of adults; instead, WTA is bound by anti-WTA immunoglobulin (Ig) with higher affinity. In contrast, in infants lacking a fully developed adaptive immune system, the MBL/MASP complex successfully binds S. aureus WTA and induces deposition of complement factor C4 (26). In addition, we recently purified anti-WTA Ig from human intravenous immunoglobulins (IVIG) using a WTA-coupled affinity column and demonstrated that anti-WTA Ig induces activation of the classical complement pathway, leading to opsonophagocytosis of S. aureus (27).

To understand the interactions between host defense factors and S. aureus, the identification of human serum proteins that sense specific bacterial molecules is critical. Because it is difficult to purify bacterial cell wall glycopolymers to homogeneity, the use of S. aureus mutant strains to screen for human serum proteins recognizing novel ligands presents a valuable alternative. In this report, we demonstrate that SAP binds specifically to bacterial PGNs, but this binding is abolished in the presence of bacterial WTA. In addition, we found that SAP-bound WTA-deficient ΔtagO cells were engulfed by human PMNs in a complement-independent manner, which suggests that SAP represents a novel PGN recognition protein present in human serum.

Materials and Methods

Protein, sera and bacteria

Complement component proteins and antibodies including human C1q and C1s and antibodies against human C1q and C1s were obtained from Complement Tech (Tyler, TX). Human CRP was obtained from Sigma-Aldrich. IVIG was obtained from SK Chemicals (Seoul, South Korea). Human sera were obtained from healthy volunteers who provided informed consent. SAP was purified from human serum. Detailed purification procedures and SDS-PAGE analysis patterns are summarized in Supplemental Fig. S1. Purified SAP was immunized to rabbits and anti-SAP polyclonal antibodies were obtained. Monoclonal antibodies against human FcγRs including anti-human CD64 (clone 10.1, BioLegend), anti-human CD32 (AT10, Abcam), and anti-human CD16 (clone 3G8, BioLegend) were used. Depleted serum was prepared as described previously (27) with some modifications. Briefly, a human intact serum (1 ml) was incubated on ice for 30 min with a mixture of formaldehyde-fixed S. aureus Δspa, ΔtagO double mutant cells (1.3 × 10¹⁰ cfu) and Δspa mutant cells (1.3 × 10¹⁰ cfu), and bacteria were removed by centrifugation. The same absorption process was repeated three-times for sufficient removal of S. aureus-recognizing serum factors including SAP, MBL and antibodies, which were described in supplemental Fig. S1C–S1E. Bacterial strains and functions of deleted genes in mutant strains are summarized in Table I. S. aureus RN4220 is used as a parental strain. All of the bacterial strains were cultured with Luria-Bertani (LB) medium supplemented with antibiotics wherever required.

Table 1.

Bacterial strains used in this study

Strains	Genotypes	Phenotypes	References
RN4220	Parantal strain	Parent strain	(46)
M0107	Δspa::phleo	Protein A depleted	(47)
T775	Δspa,ΔltaS::erm,km	Protin A / LTA depleted	(27)
T777	Δspa,ΔltaS/pSltaS::erm,km	Protein A depleted / ltaS complement	(27)
T258	Δspa,ΔtagO::phleo,erm	Protein A / WTA depleted	(26)
T358	Δspa,ΔtagO/pStagO::erm,cm	Protein A depleted / tagO complement	(26)
T174	ΔtagO::erm	WTA depleted	(48)
T002	ΔoatA::erm	PGN O-acetyltransferase depleted	(26)
T013	Δlgt::erm	Lipoprotein lipidation depleted	(49)
NI-1	ΔmprF::erm	Lysyl-phosphatidylglycerol depleted	(50)
M0875	Δypfp::erm	Glycolipid depleted	(48)
M0793	ΔdltA::erm	D-Ala modification of WTA and LTA	(48)
T2316	ΔsrtA::cm	Sortase A depleted	(51)

Purification of insoluble and soluble PGNs from S. aureus

Insoluble and soluble PGNs were purified as described previously (28, 29). The detailed method is described in the legend of supplemental Fig. S2.

Flow cytometric measurements of C4 and C3 deposition on S. aureus cells

Complement C4 and C3 deposition was measured as described previously (26). Briefly, 2.0 × 10⁹ of S. aureus cells were fixed with ethanol and incubated at 37°C for 1 h in 20 µl incubation buffer [10 mM Tris (pH 7.4), 140 mM NaCl, 1% human serum albumin (HSA), 2 mM CaCl₂, 1 mM MgCl₂] containing 10% human sera and purified anti-WTA Ig or SAP. After centrifugation, recovered cells were washed with washing buffer [10 mM Tris-HCl (pH 7.4), 140 mM NaCl, 2 mM CaCl₂, 1 mM MgCl₂], and then were resuspended in 20 µl incubation buffer. For detection of bound C4b, mouse anti-human C4 monoclonal antibody (mAb) (BioPorto, Denmark, diluted 1:500) and fluoresce in 5-isothiocyanate (FITC)-conjugated goat F(ab’)₂ anti-mouse IgG mAb (Beckman Coulter, Indianapolis, diluted 1:200) were used. For detection of bound C3b, FITC-conjugated mouse anti-human C3 IgG mAb (diluted 1:200) was used. Following the application of antibodies, S. aureus cells were sonicated for 15 s to disperse clumped cells before measurement of fluorescence data using flow cytometry (Model FC500, Beckman Coulter).

Determination of serum levels of SAP, MBL and anti-S. aureus Ig

Western blot analyses to determine levels of human SAP, MBL and anti-S. aureus Ig were determined by western blot analysis as described previously (26). The detailed methods of western blot analyses are described in the legend of supplemental Fig. S1. ELISA for human SAP and MBL were as described previously (26) and are shown in the legend of supplemental Fig. S1.

PMN preparation

PMNs were prepared as previously described (27) with some modification. The detailed method is described in the legend of supplemental Fig. S2. The FcγRII-expressing HEK293T cells were prepared as described previously (30). The detailed method was shown on the legend of supplemental Fig. S3.

SAP-mediated phagocytosis assay

The phagocytosis experiment was performed as previously described (27) with some modification. Briefly, Δspa mutant and ΔtagO, Δspa double mutant S. aureus cells grown in LB medium to post-exponential phase were washed, killed with 70% ethanol, labeled with 0.1 mM FITC (Sigma) in 100 mM Na₂CO₃ buffer (pH 8.5) for 30 min at room temperature and resuspended in Hanks' balanced salt solution (HBSS). FITC-labeled bacteria (equivalent to 1.5 × 10⁷ cfu) or FITC-labeled PGNs (40 µg) were incubated with 10% depleted serum with or without 1 µg SAP or CRP in 20 µl HBSS containing 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, 150 mM NaCl and 0.4% HSA for 30 min at 37°C with shaking. Then, the PMN preparation (1.5 × 10⁵ cells in 35 µl) was added to 5 µl resuspended FITC-labeled bacteria (corresponding to 3.8 × 10⁶ cfu, MOI of approximately 25), and the mixture was incubated at 37°C for 60 min with shaking. Extracellular FITC-labeled S. aureus were quenched by 0.2% trypan blue. Finally, phagocytosed FITC-labeled S. aureus cells within the 100 PMNs or 100 FcγRII expressing HEK293T cells were counted using fluorescent/phase-contrast microscopy.

Biacore analysis

The two types of PGN fragments used in the study [monomeric PGN (MurNAc-GlcNAc-l-Ala- d-isoGln) and dimeric PGN (MurNAc-GlcNAc-l-Ala-d-isoGln-L-Lys-d-Ala)₂], were synthesized based on the methods described previously (31). For determination of the dissociation constants (K_D) for SAP binding to the different forms of PGN, biotinylated SAP was immobilized onto a streptavidin-coated biosensor chip (SA chip; Biacore AB, Neuchâtel, Switzerland). Subsequently, the different concentrations of PGN fragments (10–40 µM in 10 mM HEPES pH 7.4, 150 mM NaCl) were passed over the surface of the sensor chip at a flow rate of 25 µl/min. The change in surface plasmon resonance (SPR) at 25°C was measured using Biacore 2000 software (Biacore AB). After 500 s of monitoring, buffer without PGN was passed over the chip to initiate dissociation. At the end of each cycle, regeneration of the chip was accomplished by washing away the bound PGN fragment using 12.5 µl of 10 mM EDTA solution. Both the association rate constant (k_a) and the dissociation rate constant (k_d) were determined using the SPR binding data with BIAevaluation software (version 3.2; Biacore AB) and used to calculate the K_D (defined as k_d/k_a).

Data processing and statistical analysis

Results from quantitative analysis of the data are expressed as the mean ± SD from at least three independent experiments, unless otherwise stated. Other data are representative of at least three independent experiments that yielded similar results. Statistical analysis was performed using Student’s t-test, and p values less than 0.05 were considered significant.

Results

SAP binds specifically to WTA-deficient S. aureus ΔtagO mutant cells

To identify serum factors capable of binding to S. aureus cell surface components, we used 9 different S. aureus gene mutations (Table 1). Human serum was incubated with each mutant, and serum proteins bound to the bacteria were eluted with 0.1 M glycine (pH 2.5) and analyzed by SDS-PAGE under non-reducing conditions. As shown in Fig. 1A, the WTA-deficient ΔtagO mutant bound strongly to a 25-kDa human serum protein (lane 5); however, when the ΔtagO mutant was complemented with a tagO-expressing plasmid (i.e., ΔtagO, Δspa/pStagO), it no longer bound this protein (lane 6). By contrast, a human apolipoprotein A bound similarly to all mutant strains (band b). The N-terminal amino acid sequence of this 25-kDa protein matched that of SAP (Fig. 1B). Because SAP is known to exhibit calcium -dependent, lectin-like binding activity toward monosaccharides and polyanionic carbohydrate biopolymers such as lipopolysaccharides, glycosaminoglycans and DNA (4), we examined the calcium-dependence of SAP binding to S. aureus ΔtagO mutant cells (Fig. 1C). SAP binds to ΔtagO mutant cells in a calcium-dependent manner (lanes 3 and 4). To confirm this binding, purified SAP was labeled with FITC and incubated with the ΔtagO mutant. FITC-labeled SAP showed stronger binding to the ΔtagO mutant than to wild-type cells (Fig. 1D). Taken together, these results establish that SAP specifically recognizes the WTA-deficient ΔtagO cells in a calcium-dependent manner.

Figure 1. Biochemical characterization of a ΔtagO mutant-binding 25-kDa human serum protein.

(A) Screening of human serum for factors that bind S. aureus cell wall mutants. Twelve ethanol-fixed S. aureus cells (1.0 × 10¹⁰ cells) were incubated with 1 ml human serum. Cells were recovered by centrifugation and washed. Bound proteins were eluted with 0.1 M glycine (pH 2.5) and analyzed by SDS-PAGE on a 15% gel under non-reducing conditions. Protein bands were stained with Coomassie Brilliant Blue. The S. aureus mutants are described in Table 1. Lane 1, parental S. aureus; lane 2, Δspa; lane 3, ΔltaS, Δspa; lane 4, ΔltaS, Δspa/pSltaS; lane 5, ΔtagO, Δspa; lane 6, ΔtagO, Δspa/pStagO; lane 7, ΔoatA; lane 8, Δlgt; lane 9, ΔmprF; lane 10, ΔypfP; lane 11, ΔdltA; lane 12, ΔsrtA. The ΔtagO, Δspa double mutant sample (lane 5) was enriched for a 25-kDa serum protein, and when complemented with a tagO-expressing plasmid, this enrichment was not observed (lane 6). The arrow indicates SAP; HSA, human serum albumin; Ig, immunoglobulin. (B) Comparison of the N-terminal amino acid sequence of "band a” with that of human SAP. (C) Binding of SAP to the parental S. aureus RN4220 strain (parent) or the ΔtagO, Δspa double mutant (ΔtagO) in the presence or absence of calcium ion. (D) Binding of FITC-labeled SAP to the parental S. aureus RN4220 strain (parent) or the ΔtagO by flow cytometry as described in Materials and Methods. The curves with gray areas underneath represent cell-only controls.

SAP recognizes PGNs

Bacterial PGNs are covalently modified with WTA via a phosphodiester linkage (19). Because of previous data indicating that staphylococcal WTA blocks PGN recognition protein-SA (PGRP-SA) from binding to PGN during induction of the innate immune system in insects (32, 33), together with the inability of ΔtagO mutant cells to synthesize WTA, we hypothesized that S. aureus PGN exposed on the ΔtagO mutant cell surface may serve as a ligand of SAP. To examine this possibility, we used 10 preparations of insoluble bacterial cell wall components, each one depleted for a different set of PGN-associated surface proteins or WTA through gene mutation or treatment with trypsin or trichloroacetic acid (TCA) (Fig. 2A). For use as a control, intact PGN non-depleted for WTA and surface proteins was obtained from the parental S. aureus RN4220 strain without trypsin or TCA treatment (Fig. 2A, lane 1). WTA and surface proteins were removed from the crude cell wall using both trypsin and TCA treatments (lane 4). When SAP was incubated with the 10 different S. aureus cell wall preparations, SAP bound only to WTA-depleted S. aureus PGNs (Fig. 2A, lanes 4, 5, 6, 8 and 10) and not to WTA-containing PGNs (lanes 1, 3, 7 and 9). Additionally, surface proteins were apparently capable of blocking the interaction between SAP and the cell wall. The srtA gene, which encodes sortase A, an enzyme that covalently attaches surface proteins-such as Protein A, the product of the spa gene-to PGN, seems to be involved in this surface protein-mediated inhibition. To investigate this observation, we measured the binding between FITC-labeled SAP and WTA-linked or WTA-depleted PGNs using flow cytometry (Fig. 2B). SAP was confirmed to bind specifically and with high affinity to WTA-depleted insoluble bacterial PGN but not to WTA-linked PGN (Fig. 2B). These results suggest that SAP can recognize bacterial PGN, unless prevented from doing so by WTA.

Figure 2. Purified SAP binds specifically to bacterial PGNs.

(A) Insoluble (is)-PGNs were purified from four different S. aureus strains and treated with trypsin or TCA. Next, is-PGN preparations (1 mg) from sortase- or WTA-depleted were incubated with 1 µg SAP for 1 h at 4°C, recovered by centrifugation, and washed. Proteins bound to the is-PGNs were eluted with Tris-buffered saline (TBS, pH 8.0) containing 10 mM EDTA and analyzed by SDS-PAGE on a 15% gel under non-reducing conditions. (B) Measurement of the binding between FITC-labeled SAP and WTA-linked or WTA-depleted PGNs using flow cytometry. The curves with gray areas underneath represent cell-only controls. (C) Competitive inhibition experiments; various concentrations of purified soluble (s)-PGN were added to reaction mixtures containing 3 mg of is-PGNs and 2 µg of SAP and incubated for 1 h at 4°C. Is-PGNs were recovered by centrifugation and washed; SAP bound to is-PGNs was eluted in TBS (pH 8.0) containing 10 mM EDTA, and SAP released from is-PGN was recovered by TCA treatment. Samples of SAP bound and released were analyzed by SDS-PAGE on a 15% gel under non-reducing conditions.

To characterize the specificity of SAP binding to WTA-depleted PGN, we performed a competitive inhibition assay using soluble PGN (s-PGN) (Fig. 2C). The s-PGN was prepared from WTA-depleted insoluble PGN (is-PGN) by digestion with lysostaphin followed by size exclusion column. Addition of 0.2 to 0.8 mg s-PGN into 3 mg of is-PGN, s-PGN liberated approximately half the bound SAP into the supernatant (lane 8), confirming the interaction of SAP with S. aureus PGN. To determine whether SAP can bind to additional bacterial PGNs, SAP was incubated with five other WTA-depleted is-PGN preparations from Bacillus subtilis, Enterococcus faecalis, Lactobacillus bulgaricus, Micrococcus luteus and Escherichia coli (Fig. S2A) and bound to all five (lanes 2–6). Among them, E. coli and E. faecalis PGNs interacted relatively weakly with SAP (lanes 3 and 6). Because the amino acid sequence of SAP shares 50% sequence homology with that of human CRP, we examined whether CRP might also bind S. aureus is-PGN; however, CRP was not observed to bind (Fig. S2B, lanes 2 and 5). When we incubated a mixture of CRP and SAP (2 µg each) with is-PGN, SAP alone localized to the is-PGN fraction (lane 3), whereas CRP remained in the supernatant (lane 6). Taken together, these results suggest strongly that SAP binds specifically to WTA-depleted S. aureus PGN and that the presence of WTA inhibits SAP binding in vitro.

SAP forms a high-molecular mass complex in the presence of soluble PGN

We previously demonstrated that clustering of insect PGRP-SA on the soluble polymeric form of S. aureus PGN is required for sensing bacterial PGN during activation of the melanin synthesis cascade, a major host innate immune response in insects (29). Because of this observation, we hypothesized that SAP binding to soluble polymeric S. aureus PGN might perform a similar biological function. As expected, when a mixture of SAP and soluble polymeric S. aureus PGN was injected onto the size exclusion column, a peak with an apparent molecular weight of 560 kDa was observed; SAP alone elutes at a volume consistent with a molecular mass of 235 kDa (Fig. 3A). Western blot analysis using anti-SAP antibodies confirmed that SAP elution coincided with the peaks (Fig. 3B), indicating that the SAP monomer (25 kDa) can assemble into a decamer (235 kDa) in the presence of calcium ion (13); moreover, SAP can form a larger-mass complex (560 kDa) with soluble PGN.

Figure 3. SAP forms a high-molecular mass adduct with soluble PGN and binds to the PGN monomer during SPR analysis.

(A) A mixture of SAP (5 µg) and polymeric soluble-PGN (s-PGN) (200 µg) was injected onto a Toyopearl HW55-S column (2.6 × 155 cm) equilibrated with 10 mM Tris-HCl (pH 8.0) containing 140 mM NaCl and 10 mM CaCl₂. Two peaks (red trace) generated were corresponded to molecular masses of 560- and 235-kDa. When SAP alone (black trace) was injected onto the same column, only the 235-kDa peak was observed. (B) The presence of SAP in each fraction was assayed by western blot using an anti-SAP Ab. The intensity of the SAP band from each fraction was directly correlated with the magnitude of the trace. (C) SPR sensorgrams were obtained by injecting various concentrations of either dimeric PGN [(MurNAc-(l-Ala-d-isoGln-l-Lys-d-Ala)-GlcNAc)₂, left panel] or monomeric [(MurNAc-(l-Ala-d-isoGln)-GlcNAc), right panel] or over an SA chip containing immobilized SAP for 500 s.

SPR analysis reveals SAP binding to synthetic PGN fragments

To determine the number of repeating MurNAc-GlcNAc disaccharides in PGN required for SAP binding, the interaction of SAP with two different synthetic PGN fragments [monomeric PGN, (MurNAc-(l-Ala-d-isoGln)-GlcNAc) and dimeric PGN (MurNAc-(l-Ala-d-isoGln-l-Lys-d-Ala)- GlcNAc)₂] was evaluated using SPR. SAP was immobilized on the surface of an SA chip, and varying concentrations of the two synthetic PGN fragments were passed over the chip (Fig. 3C). The estimated dissociation constant (K_D) for dimeric PGN binding to SAP was calculated to be 8.00 × 10⁻⁸ ± 8.65 × 10⁻⁹ M; the K_D for monomeric PGN binding to SAP was determined to be 5.97 × 10⁻⁸ ± 7.31 × 10⁻⁹ M. In a previous study, the K_D value between an identical dimeric PGN fragment and human PGRP-SA was calculated to be 3.69 × 10⁻⁶ ± 2.65 × 10⁻⁶ M (34); therefore, SAP appears to possess a greater affinity than human PGRP-SA for the dimeric PGN fragment. Taken together, these results indicate that SAP has the ability to interact with bacterial PGN fragments ranging in size from monomers to oligomers.

SAP does not induce complement activation onto S. aureus cells

In previous studies, SAP was found to activate the classical complement pathway (11). To examine whether SAP is capable of activating the complement system upon binding to PGNs, we prepared human sera depleted of S. aureus-recognizing proteins including SAP, MBL or anti-S. aureus Ig. Recently, we prepared a human serum that was depleted of both anti-S. aureus Ig and MBL and demonstrated that addition of purified anti-S. aureus Ig or purified MBL to this depleted serum induces specific C3 deposition (26, 27). In this study, we needed also to deplete SAP. A serum depleted of all three components was prepared by adsorption of the intact serum with a mixture of WTA-intact parental S. aureus cells and WTA-depleted S. aureus ΔtagO mutant cells. Depletion of MBL and SAP was firstly confirmed by western blot analysis (Supplemental Fig. S1C and S1D). Also, to confirm the depletion of serum SAP and MBL by absorption, we measured the amounts of SAP and MBL of the intact sera and depleted sera by ELISA (Supplemental Fig. S1E). As expected, before absorption, the amount of SAP and MBL in the intact serum was calculated to be 27.3 ng/µl and 3.7 ng/µl, respectively. But, SAP and MBL were not detected in the depleted serum (Fig. S1E). Furthermore, to exclude the possibility of reduced C1q concentration by depletion process, we examined the concentrations of serum C1q and C1s proteins before and after depletion by western blot analyses (Fig. S1F and S1G), confirming that there is little reduction of serum C1q and C1s proteins by depletion process. Next, we examined whether the depleted serum retained the necessary and functional complement factors. Whereas the depleted serum itself did not induce C3 deposition (Fig. 4c), inclusion of anti-S. aureus Ig or MBL into the depleted serum did induce C3 depositions on the surface of parental S. aureus cells (Fig. 4e and 4f). This result suggests that all necessary serum complement factors are present and functional in the depleted serum. Nevertheless, addition of SAP (1 µg) to the depleted serum did not induce C3 deposition on either parental (Fig. 4d) or ΔtagO cells (Fig. 4j). Similarly, when we added SAP to complete human serum, no additional C3 deposition was observed (Fig. 4b and 4h) when compared with a control (Fig. 4a and 4g). Whereas SAP can bind to S. aureus ΔtagO mutant cells, SAP did not stimulate C4 deposition on these cells (Fig. S2D). Taken together, these results reveal that SAP bound to S. aureus PGN does not induce activation of the complement system under these conditions.

Figure 4. SAP does not induce C3 deposition on Δspa mutant or ΔtagO, Δspa double mutant cells, but induces complement-independent phagocytosis of S. aureus cells by PMNs.

(A) Panels (a) and (g), C3 deposition on parental and ΔtagO mutant cells after incubation with intact serum; (b) and (h), C3 deposition by incubation of SAP (1 µg) with intact serum; (c) and (i), C3 deposition in depleted serum; (d) and (j), C3 deposition after addition of 1 µg SAP into depleted serum; (e) and (k), C3 deposition after addition of S. aureus-recognizing Ig (1 µg) into depleted serum; (f) and (l), 50 µg MBL added into panels (c) and (i), respectively. The serum concentration used was 10%. C3b bound to S. aureus cells was detected by flow cytometry as described in Materials and Methods. The curves with gray areas underneath represent cell-only controls. (B) Ethanol-killed ΔtagO mutant (columns 1–5) and parental cells (columns 6–10) were labeled with FITC (0.1 mM) and incubated with depleted serum (10%) in a 20 µl buffer. SAP (1 µg) or CRP (1 µg) and FITC-labeled S. aureus cells (3.8 × 10⁶ cells) were incubated with human PMNs (1.5 × 10⁵ PMNs) at multiplicity of infection of 25 in 40 ml RPMI 1640 medium at 37°C for 1 h. Phagocytosed S. aureus cells in 100 PMNs were counted under fluorescent phase-contrast microscopy. Data are the means ± SD of results of three independent experiments. *p < 0.01.

SAP induces complement-independent phagocytosis of WTA-depleted ΔtagO mutant cells

Though SAP failed to activate the complement system, SAP might mediate phagocytosis via association with FcγRs. One earlier observation that supports this supposition is that SAP- or CRP-mediated opsonization of apoptotic cells enhances their phagocytosis by macrophages (15, 16). To examine the effect of SAP (1 µg) on phagocytosis of S. aureus cells by human PMNs, we directly counted the number of FITC-labeled ΔtagO cells engulfed by 100 PMNs (Fig. 4B). In the absence of depleted serum, SAP increased the number of phagocytosed ΔtagO mutant cells from 10 ± 4 (Fig. 4B, column 1) to 111 ± 13 (column 2), suggesting that SAP may activate FcγR-mediated phagocytosis. Addition of depleted serum did not increase the number of phagocytosed ΔtagO mutant cells (column 4), suggesting that serum complement factors do not play a role in SAP-mediated phagocytosis of ΔtagO mutant cells, a result consistent with the failure of SAP to activate the complement system against S. aureus ΔtagO cells (Fig. 4A). In the absence of depleted serum, SAP (1 µg) increased the number of parental S. aureus cells engulfed by 100 PMNs from 4 ± 2 (Fig. 4B, column 6) to 33 ± 9 (column 7), indicating that SAP only weakly induces phagocytosis of WTA-intact S. aureus cells. Again, addition of depleted serum had little effect on phagocytosis of parental cells (column 9). Notably, CRP (1 µg) did not induce phagocytosis of either parental or S. aureus ΔtagO cells under identical conditions (columns 5 and 10).

To expand upon these results, we investigated the possibility that SAP stimulates phagocytosis of WTA-depleted PGN (Fig. S2C). Regardless of the presence or absence of depleted serum, SAP enhanced greater engulfment of FITC-labeled, WTA-depleted PGN than WTA-intact PGN (Fig. S2C, columns 2, 4 and columns 6, 8, respectively). Taken together, these results support the notion that SAP induces complement-independent phagocytosis of WTA-depleted S. aureus ΔtagO mutant cells as a result of its interaction with exposed PGN; in contrast, SAP does not affect phagocytosis of WTA-intact Δspa mutant cells.

SAP-mediated phagocytosis was FcγR-dependent

Finally, to confirm whether the FcγRs of PMNs are involved in SAP-enhanced phagocytosis, commercially available mAbs targeting three different FcγRs-CD64 (FcγRI), CD32 (FcγRII) and CD16 (FcγRIII)-were tested. These three receptors are known to differ in their abilities to bind Ig and Ig-containing immune complexes (35). As a control, a combination of three anti-FcγR mAbs clearly inhibited anti-WTA Ig-mediated phagocytosis of parental S. aureus cells (Fig. 5, columns 11 and 12); however, the antibody cocktail did not affect MBL-mediated opsonophagocytosis (columns 14 and 15). Using ΔtagO mutant cells, the mixture of three mAbs was able to inhibit SAP-mediated phagocytosis (columns 2 and 3). By using different pairs of mAbs, the anti-CD64/CD32 combination exhibited more potent inhibition compared with the anti-CD32/CD16 and anti-CD64/16 combinations (columns 4 to 6). When each anti-FcγR mAb was assayed independently, anti-CD64 exhibited a greater effect than either anti-CD32 or anti-CD16 (columns 7 to 9). These results reveal that SAP-mediated phagocytosis is mediated by FcγRs present on the surface of PMNs.

Figure 5. Anti-FcγR mAbs inhibit SAP-mediated phagocytosis.

Ethanol-killed ΔtagO mutant (columns 1–9) and parental S. aureus cells (columns 10–15) were labeled with FITC (0.1 mM). As a positive control of FcγRs-mediated phagocytosis, anti-WTA Ig (1 µg) was incubated with FITC-labeled parental cells (3.8 × 10⁶ cells) and human PMNs (1.5 × 10⁵ PMNs) in the absence (column 11) or presence (column 12) of three different anti-FcγRs mAbs, anti-CD64 (anti-FcγRI mAb), anti-CD32 (anti-FcγRII mAb) and anti-CD16 (anti-FcγRIII mAb). MBL-mediated opsonophagocytosis used as a negative control. MBL (50 ng) was incubated with FITC-labeled parental cells (3.8 × 10⁶ cells) and depleted serum (2 µl) in the absence (column 14) or presence (column 15) of three different anti-FcγRs mAbs. FITC-labeled ΔtagO mutant S. aureus cells (3.8 × 10⁶ cells) were incubated with SAP (1 µg) and PMNs (1.5 × 10⁵ cells) in the presence of different combinations of anti- FcγRs mAbs (each 0.35 µg) in 40 µl RPMI1640 medium at 37°C for 1 h (columns 3–9). Phagocytosed S. aureus cells in 100 PMNs were counted under fluorescent- phase contrast microscopy. Data are the means ± SD of results of three independent experiments. *p < 0.01.

To further confirm the requirement of FcγRs for SAP-mediated phagocytosis, we transfected the FcγRII gene into HEK293T cells. The expression of FcγRII on the HEK293T cells was confirmed by flow cytometry analyses as like as described previously (30). As a positive control, anti-WTA IgG induced phagocytosis of parental S. aureus cells by the FcγRII-expressing HEK293T cells (Fig. S3, column 12) and anti-CD32 (FcγRII) mAb clearly inhibited the anti-WTA IgG-mediated phagocytosis of parental S. aureus cells (column 13). In this condition, SAP increased the number of ΔtagO mutant cells engulfed by 100 FcγRII-expressing HEK293T cells from 8 ± 5 (Fig. S3, column 1) to 86 ± 37 (column 2), indicating that SAP only weakly induces phagocytosis of WTA-deficient ΔtagO S. aureus cells. Addition of anti-CD32 mAb into column 2 solution inhibited SAP-mediated phagocytosis (column 3), but anti-CD64 mAb did not (column 4). As like in PMNs, FcγRII expressing HEK293T cells did not induce SAP-mediated phagocytosis toward to parent cells (column 9). Taken together, these results demonstrated that the involvement of FcγRII for SAP-mediated phagocytosis against ΔtagO mutant cells.

Phagocytosed S. aureus cells lost WTA and were recognized by SAP

Next, we tried to get an insight into when WTA-deficient S. aureus cells can be generated in vivo. Since it was known that S. aureus WTA is easily released from PGN at the acidic pH (36), we supposed that S. aureus WTA will be released when S. aureus cells were transferred into acidic phagolysosomes after engulfment by PMNs. Before checking this possibility, we firstly incubated parent S. aureus cells in vitro with different pH conditions at 37°C for 2 h. Bacterial cells were recovered by centrifugation, washed, and incubated with SAP whether it can bind to bacterial cells pre-incubated with acidic conditions (Fig. S4A). As expected, SAP bound to S. aureus cells pre-incubated at pH 2, 3 and 4 (Fig.S4A, lanes 2–4), but SAP weakly bound bacteria pre-incubated at pH 5, 6 and 7 conditions (lanes 5–7), indicating that S. aureus WTA was removed at acidic conditions, leading to SAP binding to the exposed PGN of S. aureus. Then, we tested SAP-binding to engulfed S. aureus cells. Parental S. aureus cells were firstly opsonized by incubation of Δspa-treated serum (10%) and anti-WTA Ig, and were incubated with PMNs for the indicated times (Fig. S4B). Then, cell wall fraction of phagocytosed S. aureus cells were recovered by lysis of PMNs and incubation with 0.5% SDS solution to remove the proteins originated from PMNs. When we examined SAP binding abilities toward cell walls derived from phagocytosized parental S. aureus cells, SAP bound to recovered cell walls with longer incubation time (Fig. S4B, lanes 3–5), but not to non-phagocytosized S. aureus cells (Fig. S4B, lane 2). Reversely, unbound SAP was gradually decreased by increase of incubation time with PMNs (Fig. S4B, lanes 12–15). To confirm this observation, we performed the same experiments in the presence of NH₄Cl, an inhibitor of endosome acidification (37). As expected, SAP could not bind to bacterial cell walls recovered from phagocyosized S. aureus cells in the presence of NH₄Cl (lanes 7–10) and the amounts of unbound SAP were not changed (lanes 17–20). These results demonstrated that inhibition of acidification by NH₄Cl prevented both WTA release and SAP-binding to bacterial cell walls derived from phagocytosed S. aureus cells. Also, removal of WTA in phagocytosed S. aureus cells was confirmed by FACS analyses (Fig. S4C). As expected, anti-WTA IgG bound to cell walls recovered from parental S. aureus cells (panel a), but its binding ability was almost lost after 3 h phagocytosis (panel d). Therefore, these results also suggest that phagocytosed S. aureus cells lost WTA and were able to be recognized by SAP.

Discussion

Screening of S. aureus mutant strains for microbial molecular patterns enabled the identification of a novel ligand of SAP. SAP binding to bacterial PGN was abolished by the presence of S. aureus WTA. We were unable to reproduce a SAP-mediated complement activation reported in previous studies. This inconsistency may be attributable to the different methods used to prepare human sera and/or the bacterial species tested. However, in this report, we clearly demonstrate that SAP opsonization enhances phagocytosis, specifically, of ΔtagO mutant cells by PMNs in an FcγRs-dependent manner. The requirement of FcγRs for SAP-mediated phagocytosis was confirmed independently using the FcγRII expressing HEK293T cells and anti-FcγRII (CD32) mAb. Also, we provided the evidences that WTA-deficient S. aureus can be generated in phagolysosomes of PMNs. Recently, Sun et al. reported that serum anti-PGN antibodies induce to deliver S. aureus PGN into the phagocytes via FcγRs to activate intracellular nucleotide-binding and oligomerization domain (NOD) proteins that are known as NOD-like receptors (NLRs) (30). In this study, since SAP is also functioning as an S. aureus PGN recognition protein and an inducer of FcγRs-dependent phagocytosis in serum as like anti-PGN antibodies, it will be quite plausible for serum SAP plays a similar role to anti-PGN antibodies.

From these results, together with those from our recent studies (26, 27), we developed a model describing the recognition mechanisms of the host innate immune response against S. aureus invasion (Fig. 6). In human infants, who have not developed adaptive immunity, the serum MBL/MASP complex recognizes S. aureus WTA and induces activation of the complement via the lectin pathway, leading to clearance of S. aureus by opsonophagocytosis. Human adults have developed anti-WTA Igs that directly recognize S. aureus WTA and activate the classical complement pathway, which triggers opsonophagocytosis of S. aureus (right). Clearance of S. aureus can also be achieved following SAP recognition of PGN in the absence of WTA; SAP-mediated clearance occurs via FcγRs-dependent and complement-independent phagocytosis (left). One possible explanation for the function of SAP in immunity is that recognition of a cell wall component PGNs that are common to both Gram-negative and Gram-positive bacteria-namely, PGN recognition by a serum PRR represents an ancient means of defense. In response, Gram-positive bacteria may have evolved WTA to conceal PGN in order to escape the SAP-induced host innate immune response. An alternative model is that a host may possess an ability to remove WTA to allow SAP-mediated opsonophagocytosis.

Figure 6. Two putative host immune responses against S. aureus infection.

(Left) SAP recognizes the exposed PGN of S. aureus ΔtagO mutant cells and then induces FcγRs-dependent phagocytosis by PMNs. (Right) Serum MBL/MASP complex in infants who have not yet fully developed their adaptive immunity recognizes S. aureus WTA and then induces lectin complement pathway, leading to the clearance of S. aureus by opsonophagocytosis. But, anti-WTA Igs in adults who have developed their adaptive immunity recognize S. aureus WTA and then induce C1q-mediated classical complement pathway, leading to the clearance of S. aureus by opsonophagocytosis. Green pentameric moieties indicate SAP.

At present, two PGN-recognizing protein families have been characterized for the mammalian immune system. Members of the first family, the so-called PGRPs, bind to and, in some cases, hydrolyze PGNs of the bacterial cell wall (38). A combined genomic and experimental approach has led to the identification of four PGRPs family members, which are conserved in insects, mice and humans (39). These proteins share a conserved 160-amino acid-long PGRP domain with significant sequence similarity to a family of bacteriophage and bacterial type 2 amidases that catalyzes hydrolysis of the amide bonds in PGNs (40). The PGN-binding domain of PGRPs binds the muramyl pentapeptide or tetrapeptide fragment of PGN with high affinity. Mammalian PGRPs were initially identified as PRRs regulating host innate immunity. Notably, recent studies have demonstrated that all four known PGRPs modulate the acquisition and maintenance of normal gut microbiota, which protect the host from inflammation, tissue damage and colitis (41). Members of the second PGN-recognizing protein family contain NOD and NLRs (42). The NLRs, NOD1 and NOD2, are intracellular receptors for bacterial PGN fragments (42). The ligands of NOD1 and NOD2 were determined to be d-glutamyl-meso-diaminopimelic acid (DAP) (43) and muramyl dipeptide (MDP), respectively; MDP is a PGN motif conserved widely among both Gram-positive and Gram-negative bacteria (44). Recent studies have demonstrated that NOD1 and NOD2 recognize a subset of pathogenic microorganisms able to invade a host cell and multiply intracellularly. Once activated, NOD1 and NOD2 trigger intracellular signaling pathways that stimulate expression of inflammatory genes (45). In this report, we add SAP to the list of PGRP-like protein that induces phagocytosis of bacteria by PMNs. Our SPR analysis revealed that SAP recognizes the monomeric unit of the PGN fragment. Despite this, the functional significance of SAP remains ambiguous. Our most significant observation is that WTA blocks the recognition of PGN by SAP; therefore, PGN is a cryptic ligand of SAP. Biochemical evidence that a host removes WTA to allow detection of S. aureus PGN by SAP is provided by showing intracellular generation of WTA-deficient S. aureus cells inside of PMNs after phagocytosis.