The staphylococcal superantigen-like proteins (SSLs) are close relatives of the superantigens but are coded for by a separate gene cluster within a 19-kb region of the pathogenicity island SaPIn2. rSSL7 (formally known as SET1) bound with high affinity (KD, 1.1 nM) to the monomeric form of human IgA1 and IgA2 plus serum IgA from primate, pig, rat, and horse. SSL7 also bound the secretory form of IgA found in milk from human, cow, and sheep, and inhibited IgA binding to cell surface FcαRI (CD89) and to a soluble form of the FcαRI protein. In addition to IgA, SSL7 bound complement factor C5 from human (KD, 18 nM), primate, sheep, pig, and rabbit serum, and inhibited complement-mediated hemolysis and serum killing of a Gram-negative organism Escherichia coli. SSL7 is a superantigen-like protein secreted from Staphylococcus aureus that blocks IgA-FcR interactions and inhibits complement, leading to increased survival of a sensitive bacterium in blood.
Staphylococcus aureus is a common human pathogen and produces an assortment of cell surface and secreted proteins that target innate and adaptive immune defenses of the host (1). Several studies (2) and the completion of eight staphylococcal genomes have revealed that the genes for many of these factors are clustered within three pathogenicity islands called staphylococcal pathogenicity island 1 (SaPIn1)3, SaPIn2, and SaPIn3 (3, 4). These discrete regions of the genome have the capacity to form infectious phage-like particles for transmission of genes required for bacterial defense and survival (4). One important class of genes located within SaPIn1 and SaPIn3 are those that code for superantigens. These potent exotoxins target Ag recognition by binding MHC class II and TCR to drive T cell activation and cytokine release (5). Superantigens have a characteristic two-domain protein structure composed of a larger C-terminal domain of the β-grasp fold and a smaller N-terminal domain of the OB-fold, an ancient fold identified in several other bacterial toxins that bind oligosaccharides and oligonucleotides (6, 7). Superantigens have variable surfaces, but the core structure between the two domains is highly conserved. Homology searching of the completed staphylococcal genomes using conserved amino acid motifs from this core region led to the identification of the staphylococcal superantigen-like proteins (SSLs) (6). The SSLs were first named staphylococcal enterotoxin-like toxins (SET) but have since been renamed to conform to a standard nomenclature (8). Twenty-six alleles of the SSL family were initially identified (3, 9, 10). Alignment by Fitzgerald et al. (11) reveals a total of 11 unique ssl genes resident within a 19-kb region of SaPIn2 across the eight completed staphylococcal genomes NCTC8325, Sanger MSSA, Sanger MRSA, MW2, N315, Mu50, Col, and NCTC6751. The ssl genes are named in sequential order as they occur along the genome. The genes formally named set1 from strain NCTC6571 (9), set11 from strains N315 and Mu50 (3), and set22 from strain MW2 (10) are now considered alleles of ssl7.
The SSL amino acid sequences align best with the superantigen toxic shock syndrome toxin (TSST-1) whose gene resides on SaPIn1 (3) (12). The three-dimensional structure of SSL5 (formally SET3) has all the major features of a superantigen molecule, but no superantigen activity has been detected suggesting an alternative role in microbial defense (9, 12, 13). The three-dimensional structure of SSL7 (14) is similar to SSL5. SSL7 was shown to bind weakly to human monocytes and to be phagocytosed by human dendritic cells in the presence of dextran, but the mechanism for uptake by DC was not determined (14). Another study has shown that the ssl genes are present in 40 of 40 strains examined from randomly collected S. aureus isolates from blood cultures and throat swabs (15), suggesting a nonredundant role in bacterial survival.
Defense against S. aureus colonization depends primarily on preventing adherence to mucosal surfaces by secretory IgA and other opsonins. IgA is the predominant Ab at the mucosa (16) and exists mostly as a dimer complexed with the J chain and the secretory component (SC). SC is the ectodomain of the poly-Ig receptor that remains bound to IgA following transcytosis and proteolytic processing. Secretory IgA is unable to activate phagocytosis via FcαRI, and the activation of a respiratory burst requires the integrin Mac-1 (complement receptor 3 or CD11b) as a cofactor (Ref.17 ; reviewed in Refs.18 and 19). Serum IgA is monomeric and binds avidly to FcαRI (CD89) on formation of immune complexes initiating phagocytosis by neutrophils, granulocytes, and monocyte/macrophages (20). Monomeric serum IgA has been proposed to provide a second line of defense against microbes such as S. aureus via FcαRI-mediated phagocytosis (21, 22). Mutagenesis (23, 24, 25) and the recent crystal structure of an IgA:FcαRI complex reveals that the FcαRI Ig-like D1 domain binds at the Cα2:Cα3 junction of the IgA H chain (26).
Complement is also an important defense against S. aureus infection. Mice that have been depleted of complement by the injection of cobra venom are substantially more susceptible to severe staphylococcal infection (27). Extracellular fibrinogen binding protein (EfB) is one staphylococcal protein that binds C3 and inhibits opsonophagocytosis and C3 surface deposition (28). Complement C5 is 189 kDa and is synthesized as a single-chain precursor that is cleaved to a disulfide linked two-chain glycoprotein consisting of a 115-kDa C5α and a 75-kDa N-terminal C5β fragment (29, 30). Surface-bound C5 convertase cleaves soluble C5 to generate C5a and C5b. C5a is a potent anaphylotoxin that binds the G protein-coupled receptor C5aR to stimulate proinflammatory and chemotactic responses such as oxidative burst, phagocytosis, and leukocyte recruitment that contribute substantially to killing of S. aureus (31). The C5b fragment initiates formation of the membrane attack complex that forms water-permeable membrane channels leading to cell lysis.
In this study, rSSL7 is shown to selectively bind both IgA and complement C5 and inhibit IgA-FcαRI binding, and complement-mediated lytic activity.
Materials and Methods
SSL7 protein expression and purification
The ssl7 gene was amplified by PCR from the genomic DNA of two local clinical isolates of S. aureus (designated GL1 and GL10) obtained from Greenlane Hospital (Auckland, New Zealand) using the primer sequences (forward-BamH1) (reverse-EcoRI) CGGGATCCAAAGAAAAGCAAGAGAGAG and GGAATTCTTAAATTTGTTTCAAAGTCAC. The PCR fragment was subcloned into the expression vector pET32a-3C and expressed in Escherichia coli (AD494(DE3)pLysS) as an N-terminal thioredoxin fusion protein. The fusion protein was purified by Ni2+ affinity chromatography, cleaved using 3C protease, and subjected to Ni2+ affinity chromatography again to separate SSL7 from thioredoxin as previously described (12). Purified SSL7 protein was passed through a polymyxin B-agarose column to remove residual E. coli endotoxins (Sigma-Aldrich).
Serum and secreted samples
Primate sera were donated by the Auckland Zoo. Fresh sera from rabbit, cow, mice, pigs, horse, sheep, and rat were obtained from the Auckland University Animal Research Unit. Human breast milk was kindly donated by a lactating female laboratory technician. Colostrums from cow and sheep were kindly provided by Dr. C. Prosser (AgResearch, Hamilton, New Zealand). Saliva and tears were collected from laboratory individuals and stored under sterile conditions at 4°C.
Staphylococcal strains GL1 and GL10 are part of a large collection of staphylococcal strains isolated from patients with endocarditis or bacteremia from Green Lane Hospital (Auckland, New Zealand).
Production of rFcαRI protein
Soluble (s)FcαRI protein was produced by recombinant baculovirus infection of SF21 insect cells and purified by nickel affinity chromatography as previously described (32).
Coupling of proteins to Sepharose
Protein in PBS (pH 8.0) was added to CNBr-activated Sepharose 4B (Pharmacia Biotech) at 5–7 mg of protein/ml of wet gel Sepharose and mixed by rotation at room temperature (RT) for 2 h. Remaining active groups were blocked with the addition of Tris (pH 8.0) to 100 mM and incubating for 2 h. The protein Sepharose was washed a total of six times in PBS then suspended 1:1 in PBS/0.025% NaN3 and stored at 4°C. Coupling typically resulted in concentrations of ∼5 mg of protein/ml of Sepharose gel. Coupled gel was stored in PBS, 0.1% azide, at 50% v/v at 4°C.
SSL7 binding assays
Ten microliters of SSL7-Sepharose (50% v/v equivalent to ∼50 μg of SSL7) or protein A-Sepharose (50% v/v) suspension was added to 10 μl of serum or 100 μl of secretion, respectively, and the volume made up to 500 μl with lysis buffer (1% Triton X-100, 1% bovine hemoglobin, 140 mM NaCl, 10 mM Tris-Cl (pH 8.0), 0.025% NaN3, 1 mM PMSF, 1 mM iodoacetamide) and incubated with rotation for 1 h at 4°C. Sepharose without any protein coupled was used as a negative control to determine proteins bound independently of SSL7. The samples were washed three times using wash buffer (1% Triton X-100, 0.1% SDS, 1% deoxycholate, 500 mM NaCl, 10 mM Tris-Cl (pH 8.0)), once with TSA (10 mM Tris-Cl (pH 8.0), 140 mM NaCl, 0.025% NaN3), and once with 50 mM Tris (pH 6.8). Proteins were solubilized by boiling for 2–5 min in 10 μl of SDS-PAGE sample buffer before running on a 10 or 12.5% SDS-PAGE gel.
Western blot analysis
Proteins were resolved on either 10 or 12.5% SDS-PAGE gels under reducing conditions, and then transferred to a nitrocellulose membrane using a Bio-Rad transblot apparatus (Bio-Rad Laboratories). The membrane was incubated for 1 h at RT in TTBS (140 mM NaCl, 10 mM Tris-Cl (pH 7.6), 0.1% Tween 20) containing 5% (w/v) nonfat milk powder to block nonspecific binding. Goat anti-IgA, anti-IgD, anti-IgG, or anti-IgM (Kallestad) was incubated at 1/5000 dilution with the membranes for 1 h at RT in TTBS. After three 15-min washes in 5 ml of TTBS, the membrane was incubated with a 1/4000 dilution (in TTBS) of biotin-labeled rabbit anti-goat IgG (Sigma-Aldrich) for 1 h at RT followed by three 5-ml washes in TTBS. The membranes were incubated with a 1/4000 dilution of peroxidase-labeled avidin (DakoCytomation), washed three times in TTBS, and analyzed using an ECL Western blotting detection kit (Amersham Biosciences).
Flow cytometric analysis
Polyclonal human IgA was purified from normal human serum by SSL7-Sepharose affinity chromatography. One milligram of IgA was incubated with 1 mg FITC (Sigma-Aldrich) for 1 h at 4°C, and then separated from free FITC by chromatography on a 5.0-ml column of Sephadex G25 (Pharmacia). Fresh leukocytes were prepared from whole human blood by first diluting 10-fold with 0.85% ammonium chloride solution to lyse erythrocytes and centrifuging remaining white cells at 2500 rpm for 10 min. Leukocytes were suspended at 1 × 107 cells/ml in PBS/2% FCS. A 2-fold dilution series from 100 to 6.25 μg of SSL7 or staphylococcal enterotoxin A (SEA) was prepared in 100 μl of PBS/2% FCS. Ten micrograms of polyclonal IgA-FITC was added to each dilution of SSL7 and incubated for 15 min at RT in the dark. Fifty microliters of freshly prepared leukocytes (5 × 105 cells) was added and incubated for 15 min at RT in the dark. Cells were then fixed with the addition of 300 μl of 8% formaldehyde and incubating for 3 min at RT in the dark. Cells were washed twice with 1 ml of PBS/2% FCS, centrifuged at 400 × g for 1 min, and resuspended in 0.5 ml of PBS/2% FCS. The cells were analyzed by flow cytometry (FACS Analyser; BD Biosciences) with selective gating on the granulocyte population.
HPLC affinity purification of secretory IgA from human breast milk, bovine and ovine colostrums
An HPLC affinity column was generated by coupling 5 mg of purified rSSL7 to 0.7 g of POROS CNBr-activated medium (PerSeptive Biosystems) overnight at RT in 50 mM PO4 (pH 8.0). The 50-mm × 5-cm HPLC column was mounted on a Biocad HPLC (PerSeptive Biosystems). Aliquots of human breast milk or bovine or ovine colostrums diluted to 40 mg/ml and filtered were loaded at 10 ml/min in 50 mM PO4 (pH 6.8). Specifically bound protein was eluted with two column volumes of 50 mM glycine (pH 11.0). Collected peaks were analyzed by SDS-PAGE. Bands were excised and subjected to peptide mass fingerprinting.
SSL7 binding to IgA.
Purified human serum IgA (Calbiochem) was immobilized on a CM5 carboxyldextran chip using carbodiimide chemistry to a level of ∼300 response units (RU) as described previously (32) using a BIAcore2000 (BIAcore). Purified SSL7 in the concentration range of 10–200 nM was injected over the chip at a flow rate of 30 μl/min. The binding and dissociation kinetics were globally fitted using the BIAevaluation, version 2.1, software. The analysis of the equilibrium binding used 120-min injections of SSL7 in the concentration range of 0.25–400 nM SSL7. The equilibrium binding response (Req) at 120 min was fitted to the two-site binding model:
where B1, B2, KD1, and KD2 are the respective binding capacities and dissociation constants of the two sites, and A is the free analyte concentration.
Equilibrium C5 binding analysis.
SSL7 (1 μM) was reacted with the immobilized IgA layer (30 μl; flow rate of 10 μl/min) and a subsequent injection (122 μl; flow rate of 1 μl/min) was made of human C5 (Sigma-Aldrich) in the concentration range of 2–130 nM. For each concentration of C5, the equilibrium binding response to the SSL7:IgA was obtained after 120 min as the response above that of the injection of buffer alone. The data were fitted to the single binding site model: Req = B × A/(KD + [A]).
rsFcαRI was immobilized on a CM5 carboxyldextran chip for high-affinity 2:1 receptor capture of serum IgA (32). Serum IgA (Calbiochem) at 100 nM was injected (20 μl, 10 μl/min) as the analyte, and subtraction from a chemically coupled flow cell was used to correct for bulk refractive effects. The inhibition of IgA binding was observed by incubating 100 nM IgA with 0–150 nM SSL7 for 30 min at 25°C before injection on the BIAcore. A binding response report point 1 min after the injection was used to analyze SSL7 inhibition.
Complement-mediated hemolytic assay
Complement hemolytic assay was followed according to protocols in Shevach (33). Normal human sera from lab personnel were tested for spontaneous hemolytic activity toward allogeneic human RBC. Two sera (IB and JL) displayed spontaneously lysis via the alternate activation pathway. Activity was destroyed by heat inactivation (56°C for 30 min) or addition of EDTA but not EGTA indicating that lysis was via the alternate pathway. Fresh human RBC were prepared by repeated washing in gelatin/veronal buffered saline (GVB = 50 mM diethyl barbiturate (pH 7.4), 1% gelatin, 0.15M NaCl). GVB2+ buffer was GVB with 0.06 mM CaCl2 and 0.4 mM MgCl2. GVBE buffer was GVB with 0.4 mM MgCl2 and 10 mM EGTA. Erythrocytes were standardized at 2 × 108 cells/ml in ice-cold GVB2+. Immediately before use, 50 μl of serum diluted 1/2 with GVB2+ buffer (25 μl equivalent), was incubated for 2 h at 37°C with varying concentrations of SSL7 protein in duplicate in 12 × 75 mm borosilicate glass tubes. Volumes of 100 μl of human RBC and 50 μl of GVB2+ buffer were added and the tubes were incubated for 1 h at 37°C. Total hemolysis was measured by adding 100 μl of water instead of GVB2+. After 1 h, 1.2 ml of ice-cold 0.15 M NaCl was added, tubes were centrifuged at 1250 rpm, and hemolysis was determined by measuring the absorbance at 412 nm of the supernatant.
Serum bactericidal assay
Aliquots (500 μl) of 1/2 diluted fresh human serum diluted with HBSS were preincubated in duplicate for 30 min at 37°C with varying concentrations of rSSL7 or SEA diluted with HBSS in 5-ml borosilicate glass tubes. A volume of 10 μl (∼107 CFU) of a fresh mid-log phase (OD620, 0.15) culture of E. coli K12 (DH5α strain) was washed in HBSS and added to each sample, and then incubated without mixing for 90 min at 37°C. Tubes were placed on ice, and then 50 μl of appropriate dilutions (up to 1,000,000-fold) were plated onto LB plates and cultured for 24 h at 37°C. Surviving colonies were counted the following day.
Peptide mass fingerprinting and protein identification
Polypeptides were excised from Coomassie blue-stained SDS-PAGE gel, cut into small cubes, and subjected to overnight digestion with trypsin (Promega protein sequencing grade; 500 IU/mg; Promega; V511A 11652007) as described previously with minor modifications (34). Digests were dried down in a vacuum desiccator, and then resuspended in 50 μl of 0.1% acetic acid/0.005% heptafluorobutyric acid in high purity water for injection into an Agilent 1100 series ion-trap mass spectrometer. Peptide mass profiles were analyzed on the MASCOT database (〈www.matrixscience.com〉) for protein identification.
SSL7 binds IgA and complement C5 from human serum
rSSL7 (isolated from the GL1 strain) and rsFcαRI CD89 (Fig. 1) used for these studies were produced to a high level of purity from E. coli- and baculovirus-infected Sf21 cells, respectively (32). rSSL7 was incubated with normal human PBLs to test for superantigen activity using a standard 3-day PBL proliferation assay. No proliferative activity was observed over the entire range of SSL7 concentrations used (0.1 pg/ml to 10 μg/ml; data not shown). Binding to a soluble form of HLA-DR1 was undetectable by BIAcore analysis (data not shown).
Immobilized SSL7 routinely bound four polypeptides from human sera. The two predominant polypeptides were ∼60 and ∼27 kDa under reducing conditions and a single ∼170 kDa under nonreducing SDS-PAGE consistent with Ig (Fig. 2,A, lane 3). These polypeptides were identified as human IgA H chain and Ig L chain, respectively, by peptide mass spectrum fingerprinting (performed by the Australia Proteome Analysis Facility; 〈www.proteome.org.au〉). In addition to IgA H and L chains, two additional polypeptides with molecular masses of 110 and 75 kDa were also purified by SSL7. Although these were less predominant than IgA, their relative abundance to IgA was the same in all individuals tested (except one—see Fig 3 B). These were unambiguously identified as human complement component C5α chain (115 kDa) and C5β chain (75 kDa) by mass spectrometry of tryptic peptides.
Although no other Ig isotype was detected by peptide mass fingerprinting in the 60-kDa band excised from the SDS-PAGE gel, it was possible that other Ig isotypes might still occur in lower abundance in the SSL7-purified material, and which migrated at slightly different mobilities. Very little IgG H chain (identified by band excision and mass fingerprinting), which migrates below IgA H chain at 50 kDa (Fig. 2,B, lane 1), was evident in the SSL7-purified material when compared with IgG purified by staphylococcal protein A (SpA)-Sepharose. To confirm this, SSL7-purified polypeptides were tested by Western blot against isotype-specific antisera. The SSL7-purified 60-kDa polypeptide reacted strongly to anti-human IgA but not to anti-IgG, anti-IgD, or anti-IgM (Fig. 2,C, lane 2). In comparison, the anti-human IgG, IgA, IgD, and IgM antiserum were all weakly cross-reactive to the SpA-purified 50-kDa polypeptide (Fig. 2 C, lane 1), confirming that SSL7 was specific for the IgA isotype.
SSL7 binds both serum and secreted IgA and C5 from different species
Human secretions were tested for secretory IgA binding to SSL7. In these experiments, 10 μl of serum and 100 μl of human breast milk, saliva, and tears were reacted with 10 μl of 50% (v/v) SSL7-Sepharose (an equivalent of ∼50 μg of coupled SSL7). SSL7 bound IgA from all samples (Fig. 3 A). Monomeric IgA along with the C5 α- and β-chain polypeptides were bound from the serum sample, whereas secretory IgA with the additional 75-kDa SC were observed in the breast milk, saliva, and tear samples. The SC from human breast milk was noticeably smaller than SC from saliva or tears. Nevertheless, its identity was confirmed by peptide mass fingerprinting, and so the size difference was presumed to be a result of either an alternative proteolytic cleavage of the poly-Ig receptor in mammary tissue and/or variable glycosylation differences.
IgA binding by SSL7 might conceivably result from a profound individual seroconversion through naturally acquired SSL7 immunity. This was unlikely because the amount of SSL7-reactive IgA would be expected to vary among individuals, yet was consistently strong across 13 of 14 individuals tested. This suggested that SSL7 bound to an invariant region of IgA. Further studies confirmed that an excess of SSL7 extracted all the IgA available in serum (data not shown). Some individuals (lanes 2, 10, 11, and 14, for example) displayed additional IgG reactivity seen as a 50-kDa H chain migrating below the 60-kDa IgA H chain (Fig. 3,B). The IgG in these samples was attributed to seroconversion acquired through natural exposure to SSL7-producing S. aureus. The clear exception was volunteer 12 (Fig. 3 B), who had no detectable SSL7-reactive IgA and compensatory levels of anti-SSL7 IgG. Individual 12 was later found to have a genetic IgA deficiency.
SSL7 bound C5 in equivalent amounts from all sera, although individuals 5 and 7 showed slightly lower reactivity that may reflect lower serum levels of C5. Notably, SSL7 bound C5 from individual 12, confirming that reactivity to C5 was independent of IgA.
Binding of SSL7 to IgA-like polypeptides was observed for human, chimpanzee, baboon, and pig IgA sera. Weaker but still detectable binding was observed for horse and rat sera, but no detectable binding was seen for cow, sheep, mouse, rabbit, or goat IgA (Fig. 3,C). In contrast, SpA bound serum IgG from all species (Fig. 3,C). SSL7 bound polypeptides consistent with complement C5 from human, chimp, baboon, pig, sheep, goat, and most conspicuously rabbit serum where SSL7 appeared highly selective (Fig. 3 C). This experiment took no account of the known differences in serum levels of IgA and C5 across species, but nonetheless provided initial evidence that SSL7 reactivity with both IgA and C5 was not limited to humans. Although the polypeptides bound by SSL7 from other species requires formal confirmation, their sizes and SDS migrations under reducing and nonreducing conditions are entirely consistent with IgA and C5.
Affinity purification of secretory IgA from bovine or ovine colostrums and human breast milk
Species reactivity in those that showed poor binding of serum IgA was further tested using milk as a source of secretory IgA. Samples of diluted human breast milk (40 mg/ml) and diluted bovine or ovine colostrums (also diluted to 40 mg/ml) were passed through a rSSL7 HPLC affinity column (Fig. 4). Bound protein was eluted using high pH, giving a typical profile shown in Fig. 4. The single peak eluted with high pH was subjected to nonreducing and reducing SDS-PAGE (displayed as an inset of Fig. 4). Under nonreducing conditions, a protein with an estimated molecular mass of ∼400 kDa was routinely observed in the three species corresponding to the predicted size of the IgA dimer (2 × 170-kDa IgA molecules plus a single 75-kDa secretory chain along with the 15-kDa J chain). Under reducing conditions, this resolved into the secretory chain and the IgA H chain (the L chain and J chain were eluted from the bottom of this 7.5% SDS-PAGE gel to better separate the high-molecular-mass polypeptides). In the bovine sample, the SC was slightly lower in molecular mass than the human and ovine SC, presumed to be due to either differences in proteolytic processing or alternatively altered SC glycosylation between these species. The 75- and 60-kDa polypeptides were confirmed as SC and IgA H chain, respectively, by peptide mass fingerprinting.
Allelic differences in SSL7 do not alter IgA or C5 binding
To date, five alleles of the ssl7 gene have been published (MW2, NCTC8325, N315, NCTC6571, and Mu50) revealing amino acid sequence variation of up to 16% (Fig. 5,A). Two alleles of ssl7 were obtained from local clinical isolates called GL1 and GL10 obtained from patients suffering endocarditis. GL1-SSL7 and GL10-SSL7 differed at 30 of 210 aa positions (14%). Both bound similar amounts of human IgA and C5 from two different individuals (Fig. 5 B), indicating that both C5 and IgA binding sites were intact in these two variants. Slightly better binding of GL10-SSL7 to C5 and conversely GL1-SSL7 to IgA was observed, suggesting that some of the variable residues are sufficiently close to alter binding affinity. Quantitative affinity studies of these and other SSL7 alleles are currently in progress in an attempt to map each of the binding sites.
SSL7 has nanomolar affinity for both human IgA and C5
Biosensor analysis (BIAcore biosensor) confirmed binding to both IgA and C5 and established the affinity of both interactions. rSSL7 protein was reacted with immobilized serum IgA (Calbiochem), and global analysis of the kinetic data fitted well (χ2 = 0.184) to a model of parallel binding to a heterogeneous ligand with KD1 = 1.0 nM and for the second site KD2 = 330 nM (Fig. 6,A). The fit was poor to other models including a single site model. The random coupling of ligand by carbodiimide chemistry to a biosensor chip can result in some orientations of ligand unfavorable for interaction with the analyte. Thus, the high-affinity site, KD1 = 1.0 nM, most probably reflected the IgA:SSL7 interaction. Because serum IgA consists of both IgA1 and IgA2 subclasses, this also may contribute to the heterogeneity of the binding behavior observed. Equilibrium binding analysis, like the kinetic data, best fitted a two-site binding model, yielding KD1 and KD2 values of 1.1 ± 0.2 and 80 ± 40 nM, respectively (Fig. 6 B), consistent with nanomolar affinity for serum IgA. The binding to rIgA1-Fc region produced in baculovirus and to rIgA2 (gift of Drs. M. Goodall and R. Jefferis, University of Birmingham, Birmingham, U.K.) confirmed that SSL7 binds to both IgA1 and IgA2 isotypes in the Fc region of the molecule (data not shown).
The interaction of SSL7 with human C5 was also analyzed using the biosensor. SSL7 protein was first injected over the immobilized IgA followed by an injection of purified human C5 (Sigma-Aldrich) to demonstrate C5 binding to the SSL7 (Fig. 6,C). Simultaneous binding of IgA and C5 to SSL7 thus indicates that these interactions occur through separate sites, consistent with the serum binding studies that showed C5 from some species (e.g., rabbit) bound SSL7 independent of binding to IgA (Fig. 6 C) and volunteer 12 whose C5 bound to SSL7 in the total absence of serum IgA.
The affinity of the SSL7 interaction with human C5 was ascertained using IgA to first capture SSL7, which was then reacted with C5 as the analyte. In this case, kinetic analysis was not possible because of the dissociation of the captured SSL7 from the IgA layer. However, equilibrium analysis was possible with the assumption that the binding of C5 to SSL7 did not affect the rate of dissociation of SSL7 from the IgA layer. The data obtained fitted well to a single binding site, as often occurs with orientated presentation of ligands to analyte on the biosensor, with a KD = 18 ± 1 nM (Fig. 6 D).
SSL7 blocks IgA binding to myeloid cells
Flow cytometric analysis of human granulocytes with FITC-labeled polyclonal IgA was performed to determine whether SSL7 blocked cell surface binding of IgA to FcαRI (CD89) expressed on myeloid cells such as neutrophils, granulocytes, eosinophils, monocytes, and macrophages (35). SSL7 was incubated at serial concentrations with a fixed concentration of FITC-IgA (10 μg) before incubation with freshly prepared human leukocytes. The superantigen SEA was used as a negative control. Dose-dependent inhibition of FITC-IgA binding to cells was observed with SSL7 but not with SEA. An inhibition of 90% was achieved with a 10-fold (w/w) excess of SSL7 (100 μg) over FITC-IgA (10 μg), equivalent to a 70-fold molar excess of SSL7 (Fig. 7). This suggested that SSL7 most likely bound to a site on IgA overlapping the FcαR1 binding site in the Cα2:Cα3 domains of the IgA H chain (26).
SSL7 blocks binding of IgA to purified FcαRI
Purified rsFcαRI was immobilized to a CM5 biosensor chip at a level of 500 RU. No binding of SSL7 (200 nM) to the FcαRI was detected (Fig. 8,A). Binding of 100 nM serum IgA to the FcαRI was observed (∼220 RU) with kinetics as reported previously (Fig. 8,B). Prior incubation of IgA with rSSL7 inhibited the binding of IgA to immobilized FcαRI. An approximately equimolar concentration of SSL7 (100 nM) to IgA (100 nM) reduced the receptor binding by 70% (62 RU; Fig. 8 C). Thus, SSL7 inhibited IgA binding to both soluble and cell-bound FcαRI with consistent inhibitory kinetics.
SSL7 inhibits complement-mediated lysis and serum bactericidal activity
Serum from normal individuals was tested for naturally occurring complement lysis of allogeneic human RBC. Two sera, IB and JL, were found that lysed allogeneic RBC from donor (JF) in a standard in vitro complement hemolytic assay in the presence of EGTA. IB serum was significantly more hemolytic (100% hemolysis at 2-fold dilution) than JL (40% hemolysis at 2-fold dilution). Hemolysis was negligible following serum heat inactivation (56°C for 30 min) or addition of EDTA but was unaffected by EGTA confirming the activation of the alternate complement pathway (not shown).
rSSL7 inhibited both IB and JL serum-initiated hemolysis in a dose-dependent fashion (Fig. 9). Maximum inhibition by SSL occurred at 1.0 nM (0.63 μg/25 μl of serum) for donor IB and 0.26 nM (0.16 μg/25 μl of serum) for donor JL. Purified rSEA that was used as a control had no detectable effect on lysis even at the highest concentration tested. The slightly lower maximum lysis of IB serum between the SSL7 inhibition experiment (∼80%) and the SEA inhibition experiment (∼50%) was most likely due to aging of the serum because these two assays were performed 2 days apart.
SSL7 enhances survival of bacteria in serum
rSSL7 was incubated with fresh human serum over a range of SSL7 concentrations before the addition of a fixed number (∼107) of E. coli K12 bacteria. E. coli is susceptible to complement lysis. In the absence of SSL7, no cells survived after a 90-min incubation of 107 bacteria at 37°C with 500 μl of fresh serum diluted 1/2 in HBSS. Addition of rSSL7 at concentrations >0.26 nM (7 μg/ml) profoundly inhibited serum antimicrobial activity and increased the number of surviving E. coli cells to a maximum level of 14% at 2.1 nM (50 μg/ml) (Fig. 10). The control rSEA had no effect on lysis even at the highest concentration used (50 μg/ml).
The structural similarity between the SSLs and superantigens argued a compelling case for similar roles, but this study reveals that the SSLs have very different functions relating to microbial defense. This functional diversity may relate to the versatility of the OB-fold domain shared by both families as well as a number of other microbial toxins (7). Neither the IgA nor the C5 binding site on SSL7 has been located, but a likely position of at least one of these sites is the external face of the OB domain. This has been deduced by analyzing the known alleles of SSL7 against the predicted protein structure of SSL7. When the 30 aa variations between the GL1 and GL10 variants, which both bind IgA and C5, were mapped onto a modeled SSL7 structure, the only region completely conserved was the external face of the N-terminal OB-folded domain.
SSL7 bound with high affinity to IgA and inhibited IgA:FcαRI complex formation most probably through competition for binding at the IgA-Fc Cα2-Cα3 interface where the first ectodomain of FcαRI binds (23, 24, 25, 26, 32). SSL7 inhibition of IgA binding to FcαRI was seen in a biosensor assay configured for high-affinity bivalent capture of IgA (26, 32), and 70% inhibition was achieved by 1:1 addition of 100 nM SSL7 with 100 nM IgA. Moreover, SSL7 bound human secretory IgA and momomeric serum IgA equally well, indicating that the SSL7 targets both humoral and mucosal environments to effect survival of the organism. Competition of SSL7 and rsFcαRI for binding to IgA suggests that the most likely scenario is two SSL7 binding sites per IgA-Fc. Furthermore, SSL7 binding of secretory IgA suggests that either the SSL7 and SC binding sites are separate, or alternatively, the single SC in the sIgA incompletely occupies the SSL7 binding sites. It is predicted that SSL7 binds at the Cα2:Cα3 hinge region where FcαRI binds, and preliminary mutation analysis supports this (B. Wines, unpublished observations). One curious result from this study that requires further investigation is the difference in binding of SSL7 to bovine and ovine IgA. In this study, SSL7 bound serum IgA poorly from both bovine or ovine serum, yet bound sIgA well from bovine and ovine colostrums. This difference might be due to the lower concentrations of serum IgA in these two species. Alternatively, it may suggest that the SC and/or dimerization found in sIgA may enhance SSL7 binding affinity. Quantitative binding studies comparing affinities of both monomeric and sIgA are currently underway.
SSL7 is the first example of a staphylococcal Ig binding protein that targets IgA at the site where it binds to FcαRI. One possible role is therefore in preventing FcR-mediated leukocyte activation. Streptococcus pyogenes and group B streptococci also produce IgA binding proteins, suggesting that there is a selective advantage for these microbes to evade IgA-mediated defense mechanisms. These proteins, Arp4 and Sir22, are M proteins from S. pyogenes and the unrelated β-protein from group B streptococcus all bind IgA-Fc at the same site as FcαRI, and like SSL7, also block IgA binding to FcαRI-bearing granulocytes (36).
SSL7 was shown to effectively inhibit complement-mediated hemolysis and enhance bacterial survival of E. coli at concentrations similar to the predicted serum concentration of C5. Inhibition of C5 was first observed at concentrations >0.3 nM and increased to a maximum of ∼2.1 nM, entirely consistent with the direct one-to-one binding and blocking of C5, which is estimated to be ∼0.4 nM (75 μg/ml) in serum. At its maximum concentration of 2.1 nM, ∼14% of the initial inoculum of 107 E. coli survived the 90-min incubation in human serum. The SSL7 binding site on C5 remains to be identified, but one logical location would be that which prevents C5 activation and cleavage to the C5 α-chain to the C5a anaphylotoxin and C5b. Notably, the selective inhibition of C5 by SSL7 indicates the importance of C5 in the defense against the Gram-negative organism E. coli and highlights the usefulness of SSL7 as a novel reagent to selectively examine the importance of C5 in various immune defenses. The role of C5 against Gram-positive organisms such as S. aureus must wait until an isogenic mutant defective in SSL7 production can be produced.
Each strain of S. aureus carries just one or a few (or sometimes none at all) superantigen genes. In contrast, every strain of S. aureus so far examined carries a cluster of many ssl genes (3). This suggests that, unlike superantigens, the SSL have distinct and possibly nonredundant functions. A number of other rSSLs have been tested for IgA binding and/or complement inhibition, but only SSL7 has displayed these activities. Further investigation of the other SSLs is likely to reveal complementary and perhaps synergistic functions involved in pathogen defense. SSL7 is likely to be part of a larger defense armamentarium that includes some or all of the other SSL family members whose relative expression might vary under different circumstances. Development of S. aureus isogenic ssl7− mutants is currently underway to examine the role of SSL7 in staphylococcal infection. However, using mouse infection assays may prove to be problematic, because SSL7 does not appear to bind either murine IgA or murine C5, necessitating the use of animals where binding of SSL7 to IgA and/or C5 has been clearly established. Despite the absence of in vivo relevance of SSL7 to staphylococcal infection, this study nevertheless describes for the first time, the activity of one of the superantigen-like proteins and suggests that superantigens are but a subgroup of a much larger family of molecules with varied activities designed to bind components of host immunity.
The authors have no financial conflict of interest.
Primate serum was kindly provided by the Auckland Zoo (Auckland, New Zealand). Bovine and ovine colostrums was kindly provided by Dr. Colin Prosser (AgResearch, Hamilton, New Zealand). This research has been facilitated by access to the Australian Proteome Analysis Facility established under the Australian Government’s Major National Research Facilities program.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work is supported by a project grant from the Health Research Council of New Zealand.
Abbreviations used in this paper: SaPIn, staphylococcal pathogenicity island; SSL, staphylococcal superantigen-like protein; SET, staphylococcal enterotoxin-like toxin; SC, secretory component; s, soluble; SEA, staphylococcal enterotoxin A; RT, room temperature; RU, response unit; GVB, gelatin/veronal buffered saline; SpA, staphylococcal protein A.