Activation of Syndecan-1 Ectodomain Shedding by Staphylococcus aureus α-Toxin and β-Toxin*

Exploitation of host components by microbes to promote their survival in the hostile host environment has been a recurring theme in recent years. Available data indicate that bacterial pathogens activate ectodomain shedding of host cell surface molecules to enhance their virulence. We reported previously that several major bacterial pathogens activate ectodomain shedding of syndecan-1, the major heparan sulfate proteoglycan of epithelial cells. Here we define the molecular basis of how Staphylococcus aureus activates syndecan-1 shedding. We screened mutant S. aureus strains devoid of various toxin and protease genes and found that only strains lacking both α-toxin and β-toxin genes do not stimulate shedding. Mutations in the agr global regulatory locus, which positively regulates expression of α- and β-toxins and other exoproteins, also abrogated the capacity to stimulate syndecan-1 shedding. Furthermore, purified S. aureus α- and β-toxins, but not enterotoxin A and toxic shock syndrome toxin-1, rapidly potentiated shedding in a concentration-dependent manner. These results establish that S. aureus activates syndecan-1 ectodomain shedding via its two virulence factors, α- and β-toxins. Toxin-activated shedding was also selectively inhibited by antagonists of the host cell shedding mechanism, indicating that α- and β-toxins shed syndecan-1 ectodomains through stimulation of the host cell's shedding machinery. Interestingly, β-toxin, but not α-toxin, also enhanced ectodomain shedding of syndecan-4 and heparin-binding epidermal growth factor. Because shedding of these ectodomains has been implicated in promoting bacterial pathogenesis, activation of ectodomain shedding by α-toxin and β-toxin may be a previously unknown virulence mechanism of S. aureus.

by this mechanism. The diverse list of shed proteins includes cytokines, growth factors, and cell adhesion molecules, such as tumor necrosis factor ␣ (TNF␣), transforming growth factor ␣ (TGF␣), epidermal growth factors (EGFs), 1 L-selectin, CD44, and syndecans, to name a few. Ectodomain shedding is an important regulatory mechanism since it rapidly changes the surface phenotype of affected cells and generates soluble, biologically active ectodomains that can function as paracrine or autocrine effectors. A growing body of evidence indicates that these molecular and cellular features enable ectodomain shedding to regulate many pathophysiological processes, such as microbial pathogenesis, inflammation, and tissue repair.
Ectodomain shedding is highly regulated by various extracellular ligands and intracellular signaling pathways. Phorbol ester protein kinase C (PKC) agonists (i.e. PMA, TPA) enhance the shedding of most affected molecules (1). Protein-tyrosine kinase (PTK) activity is required for agonist-activated ectodomain shedding of syndecan-1 and -4 (15,25) and the cell adhesion molecule L1 (26). Furthermore, specific antagonists of mitogen-activated protein (MAP) kinase pathways inhibit ectodomain shedding of syndecan-1 and -4 (15), HB-EGF (27), TGF␣ (28), L1 (26), L-selectin (28), and TNF␣ (28) in an agonistspecific manner. At present, it is not clear if these signaling pathways are linked and a common signaling pathway exists to regulate ectodomain shedding. Nevertheless, available data clearly show that several signaling pathways can markedly affect the rate of ectodomain shedding at the cell surface, indicating that various pathways converge to modulate the cleavage event at the cell surface. The involvement of many signaling pathways in regulating ectodomain shedding may be one of the biological mechanisms that defines which cell surface protein gets shed in response to specific extracellular cues.
Recent studies have shown that microbial pathogens can stimulate ectodomain shedding by host cells. IL-6R shedding is activated by secreted products of S. aureus, Pseudomonas aeruginosa, Listeria monocytogenes, and Serratia marcescens (29). Similarly, TNF␣ shedding is provoked by culture supernatants of S. epidermidis (30). Although the physiological significance of pathogen-activated ectodomain shedding has yet to be clearly defined, several studies have shown that virulence factors enhance ectodomain shedding, suggesting that this is a pathogenic activity. For example, streptolysin O, a pore-forming toxin expressed by the majority of Group A Streptococcus clinical isolates, stimulates ectodomain shedding of L-selectin, IL-6R, and CD14 (31,32). Another study recently demonstrated that lipoteichoic acid released from the staphylococcal cell wall activates ADAM10-mediated shedding of HB-EGF ectodomains in epithelial cells via stimulation of the G proteincoupled platelet-activating factor receptor (18). Shed HB-EGF then activates the EGF receptor to induce mucin expression. This mechanism potentially increases the viscosity of mucus fluids, deters mucociliary clearance of pathogens, and promotes microbial pathogenesis.
We have shown that shedding of syndecan-1 ectodomains is triggered by culture supernatants of P. aeruginosa and S. aureus (25). Syndecan-1 is the major heparan sulfate proteoglycan (HSPG) of epithelial cells and it binds and regulates a wide variety of biological molecules through its heparan sulfate (HS) chains (33,34). Because syndecan-1 ectodomains are replete with all of their HS chains, they can function as soluble regulators of various molecular interactions (33,34). We have identified the P. aeruginosa shedding enhancer as LasA, a known virulence factor for P. aeruginosa lung infection (25), and results from our in vivo studies indicate that P. aeruginosa activates syndecan-1 shedding to enhance its virulence in a murine model of lung infection (35). It remains unclear, however, whether the pathogenic mechanism of P. aeruginosa that exploits syndecan-1 shedding to enhance its virulence is utilized by other pathogens.
Here we report the characterization of S. aureus enhancers of syndecan-1 ectodomain shedding. S. aureus is a common and clinically important Gram-positive bacterial pathogen that is responsible for causing a wide variety of life-threatening diseases, such as pneumonia, toxic shock syndrome, osteomyelitis, endocarditis, and sepsis (36,37). The impressive pathogenic capabilities of S. aureus are mediated by the large number of virulence factors that the bacterium elaborates, including many enzymes, adhesins, and toxins. S. aureus is also one of the leading causes of nosocomial infections and the current emergence of multidrug resistant strains adds to the threat of staphylococcal infections. It is therefore imperative to define the molecular mechanisms involved in S. aureus virulence. We initially screened the effects of specific mutations of S. aureus virulence genes on syndecan-1 ectodomain shedding, and determined that mutant strains deficient in both ␣and ␤-toxin genes do not stimulate shedding. Consistent with these results, we also found that purified ␣and ␤-toxins augment syndecan-1 shedding. Surprisingly, the stimulatory effects of the S. aureus toxins on syndecan-1 shedding were sensitive to antagonists of the host cell shedding mechanism. Furthermore, ␤-toxin, but not ␣-toxin, stimulated ectodomain shedding of syndecan-4 and HB-EGF in a host sheddase-dependent manner. These results indicate that ␣and ␤-toxins stimulate syndecan-1 shedding via activation of the host cell shedding machinery and that ␤-toxin also enhances shedding of other cell surface proteins. Because ectodomain shedding of syndecan-1 and HB-EGF has been implicated in bacterial pathogenesis, these results suggest that activation of ectodomain shedding may be one of the mechanisms by which S. aureus ␣and ␤-toxins function as virulence factors.
Immunochemicals-The rat monoclonal anti-mouse syndecan-1 ectodomain (281-2) and syndecan-4 ectodomain (Ky8.2) antibodies (25) were purified from either ascites fluids or conditioned media of hybridoma cultures by protein G affinity chromatography. Rabbit anti-mouse syndecan-1 cytoplasmic domain antibody was generated by immunizing rabbits with the synthetic peptide CNGGAYQKPTKQEEFYA. Affinitypurified antibody was prepared by protein A affinity chromatography followed by synthetic peptide affinity chromatography. The affinitypurified antibody detects detergent-extracted transmembrane syndecan-1 but does not react with purified syndecan-1 ectodomains. The anti-hemagglutinin (HA) antibody 9E10 was purchased from Sigma and HRP-conjugated donkey anti-rat, -mouse, or -rabbit antibodies were obtained from Jackson ImmunoResearch Laboratories (West Grove, PA).
S. aureus Strains and Culture Conditions-The S. aureus strains used in this study are described in Table I. Glycerol stocks of each strain were made from cultures grown in TSB with erythromycin (10 g/ml) or tetracycline (2 g/ml) where appropriate. Culture supernatants were prepared by growing glycerol stocks in TSB overnight without antibiotics at 37°C with agitation and filter sterilization. The culture supernatants were tested at 5, 10, or 20% (v/v) for their ability to activate ectodomain shedding.
Syndecan Assays-Syndecan shedding assays were performed as described previously (25). Briefly, confluent or 1 day post-confluent cultures of normal murine mammary gland (NMuMG) epithelial cells in 96-well plates were incubated with various test samples and inhibitors diluted in culture media (DMEM with 7.5% fetal bovine serum and 10 g/ml insulin) for the indicated times at 37°C. NMuMG cells passaged between 16 -20 times were used for all assays. To quantify shedding, the culture supernatants were collected, spun down to remove cells, and acidified by addition of NaOAc (pH 4.5), NaCl, and Tween 20 to a final concentration of 50 mM, 150 mM and 0.1% (v/v), respectively. Various volumes of the acidified samples were dot blotted onto Immobilon Nyϩ and developed with specific antibodies and the ECL development reagent. Blots were scanned and quantified using the NIH Image software (V. 1.62) as described previously. Cell viability after incubation with the test samples was measured by the tetrazolium salt MTT conversion assay (25).
HB-EGF Shedding Assay-Chinese hamster ovary-K1 (CHO-K1) cells stably expressing the transmembrane construct of HB-EGF (proHB-EGF) tagged with HA at the N terminus were generated previously (27). These cells were pretreated with or without KB-R8301, a hydroxamate-based HB-EGF shedding inhibitor, for 15 min and then incubated with PMA (1 M), ␣-toxin (10 g/ml), or ␤-toxin (10 g/ml) for 30 min at 37°C. Conditioned media were collected, and cells were lysed after extensive washes. The cleaved HB-EGF ectodomain in the conditioned medium was precipitated with heparin-Sepharose beads. HB-EGF ectodomains and transmembrane proHB-EGF in total cell lysates were resolved by 15% SDS-PAGE and detected by Western immunoblotting with the 9E10 anti-HA monoclonal antibody.

S. aureus Mutant Strains Lacking ␣-Toxin and ␤-Toxin Do
Not Shed Syndecan-1 Ectodomains-We reported previously that the majority of S. aureus strains secrete factors that stimulate syndecan-1 ectodomain shedding by cultured epithelial cells (25). To identify the S. aureus shedding enhancers, we initially screened a panel of S. aureus strains with specific mutations in their toxin, enzyme, or regulatory genes (Table I) for their ability to activate syndecan-1 shedding. Confluent cultures of NMuMG epithelia were incubated with 20% (v/v) TSB (media), culture supernatants obtained from the parent S. aureus 8325-4 strain (WT), or supernatants from mutant S. aureus 8325-4 strains lacking ␣-toxin (Hla Ϫ ), ␤-toxin (Hlb Ϫ ), ␥-toxin (Hlg Ϫ ), ␣and ␤-toxins (Hla Ϫ Hlb Ϫ ), ␣-, ␤-, and ␥-toxins (Hla Ϫ Hlb Ϫ Hlg Ϫ ), metalloprotease (Aur Ϫ ), or the Agr regulatory locus (Agr Ϫ ) (Fig. 1). S. aureus ␣-toxin is a pore-forming cytolytic toxin and it is one of the most potent bacterial toxins known (38). S. aureus ␣-toxin is secreted as a water soluble monomer, binds to target membranes as a soluble monomer, and then forms heptameric transmembrane pores in target cell membranes (39 -41). S. aureus ␥-toxin is also a pore-forming toxin (38). In contrast, ␤-toxin exerts its cytotoxic effects through its neutral sphingomyelinase activity (38). The role of the staphylococcal metalloprotease in pathogenesis is not fully understood, but it can digest both host and bacterial factors (42). The Agr locus specifies a density-dependent regulatory system that stimulates transcription of genes encoding exoproteins in the stationary phase of growth. Agr Ϫ mutants express significantly reduced levels of S. aureus exoproteins, including ␣and ␤-toxins (43).
FIG. 1. Activation of syndecan-1 ectodomain shedding by toxin, protease, and regulatory gene mutant strains of S. aureus. S. aureus strains were grown overnight in TSB at 37°C to stationary growth phase, and their culture supernatants were collected and filter-sterilized. S. aureus strains tested were the parent 8325-4 strain (WT) and mutant 8325-4 strains lacking ␣-toxin (Hla Ϫ ), ␤-toxin (Hlb Ϫ ), ␥-toxin (Hlg Ϫ ), ␣and ␤-toxin (Hla Ϫ Hlb Ϫ ), ␣-, ␤-, and ␥-toxin (Hla Ϫ Hlb Ϫ Hlg Ϫ ), metalloprotease (Aur Ϫ ), or Agr (Agr Ϫ ). Confluent NMuMG epithelial cells in 96-well plates were incubated for 4 h at 37°C with bacterial supernatants diluted with NMuMG culture media. The conditioned media were collected, centrifuged to remove cells, acidified, and dot blotted onto cationic Immobilon Nyϩ nylon membranes. The extent of shedding was quantified by immunoblotting using the 281-2 anti-syndecan-1 ectodomain antibody as described under "Experimental Procedures." Various concentrations of supernatants from each S. aureus strain were tested in at least three independent experiments. Data shown are the result of a representative experiment testing the effects of 20% (v/v) S. aureus supernatants or fresh TSB (media). The concentration of syndecan-1 ectodomains in conditioned media was determined using purified NMuMG syndecan-1 ectodomains as standards and each data point represents the mean of triplicate measurements Ϯ S.E. Syndecan-1 shedding activity was partially reduced (ϳ40%) in the ␣-toxin and ␤-toxin single mutant strains relative to the WT strain and was completely absent in the ␣and ␤-toxin double and ␣-, ␤and ␥-toxin triple mutant strains (Fig. 1). In contrast, shedding activity was not affected in the ␥-toxin and metalloprotease single mutant strains. Furthermore, consistent with the role of Agr in positively regulating the expression of exoproteins, the Agr Ϫ mutant strain also did not stimulate syndecan-1 ectodomain shedding (Fig. 1). Cell viability was not significantly different among epithelial cells incubated with different S. aureus supernatants (ϳ85-95%) and syndecan-1 was not detected in the conditioned media of S. aureus supernatant-treated cells when probed with an affinity-purified antibody directed against the syndecan-1 cytoplasmic domain (data not shown), verifying that the ectodomains were shed and not released intact from damaged cells. These results suggest that S. aureus ␣and ␤-toxins are the enhancers of syndecan-1 shedding.
Purified S. aureus ␣-Toxin and ␤-Toxin Stimulate Syndecan-1 Ectodomain Shedding-To determine whether ␣and ␤-toxins are indeed the S. aureus shedding enhancers, we tested the effects of several purified S. aureus toxins on syndecan-1 shedding. NMuMG epithelia were incubated with various concentrations of purified ␣-toxin, ␤-toxin, toxic shock syndrome toxin-1 (TSST-1), or staphylococcal enterotoxin A (SEA) and the extent of syndecan-1 ectodomain shedding was measured ( Fig. 2A). SEA is the major cause of staphylococcal food poisoning and TSST-1 is one of the causative toxins of staphylococcal toxic shock syndrome (38). Both SEA and TSST-1 can function as superantigens by binding to class II major histocompatibility complexes (MHC) and stimulating the class II MHC-restricted immune response. Because of these functional similarities, SEA, and TSST-1 are classified in the family of staphylococcal pyrogenic superantigen toxins, along with other staphylococcal enterotoxins.
As shown in Fig. 2A, syndecan-1 shedding was increased ϳ4-fold when epithelial cells were incubated with 1 g/ml purified ␣or ␤-toxin for 4 h at 37°C. Maximum stimulation was reached at a concentration of 5 g/ml for both ␣-toxin (ϳ8-fold increase) and ␤-toxin (ϳ6-fold increase). In contrast, both SEA and TSST-1 failed to activate shedding at all concentrations tested. Furthermore, shedding activation was rapid (Ͼ2-fold increase by 30 min) and saturable (by 2 h) when incubated with 5 g/ml ␣or ␤-toxin (Fig. 2B). At the range of concentrations tested in our shedding assays, both ␣-toxin (44) and ␤-toxin (31) have been shown to have minimal toxic effects on host cells. Consistent with these findings, we found that NMuMG epithelial cells were Ͼ95% viable after a 4-h incubation with 5 g/ml ␣or ␤-toxin (data not shown). These results establish that S. aureus activates syndecan-1 ectodomain shedding via ␣-and ␤-toxins.
S. aureus ␣-Toxin and ␤-Toxin Activate Syndecan-1 Shedding through Stimulation of the Host Cell's Shedding Mechanism-We next analyzed the size of the syndecan-1 ectodomains shed by S. aureus culture supernatants and ␣and ␤-toxins, and found that both intact ectodomains (Fig. 3A) and heparinase II-and chondroitinase ABC-digested core proteins ( Fig. 3B) shed by these S. aureus factors are similar in size to that of the constitutively shed ectodomain. These findings suggested that S. aureus activates a shedding mechanism that is similar to that used for the endogenous shedding of syndecan-1 ectodomains. To test this hypothesis and to better understand how S. aureus stimulates syndecan-1 shedding, we examined the effects of various antagonists of the host cell's shedding mechanism on S. aureus-activated ectodomain shedding (Table II).
Ectodomain shedding by host cells is inhibited by peptide hydroxamate sheddase inhibitors. The two peptide hydroxamate inhibitors, GM6001 and TAPI-1, significantly inhibited syndecan-1 ectodomain shedding triggered by S. aureus culture supernatant (20%, v/v) and purified ␣and ␤-toxins (both at 5 g/ml) (Table II). This effect was specific because inhibitors of serine and aspartic acid proteases did not affect shedding (data not shown). Furthermore, the inhibitory effect of the sheddase inhibitors was on the host cell and not on the S. aureus toxins because GM6001 did not inhibit syndecan-1 shedding if it was preincubated with the toxins and removed prior to incubation with the cells (Table II).
PKC agonists have been shown to activate ectodomain shedding of the majority of affected proteins, whereas MAP kinase antagonists have been shown to inhibit agonist-activated shedding of various cell surface molecules, including syndecan-1 (1,15). In view of these findings, we tested the effects of specific inhibitors of PKC (bisindolylmaleimide I) and the ERK (PD98059) and p38 (SB203580) MAP kinase pathways, and found that these do not significantly inhibit shedding activated by S. aureus factors (Table II). We also tested the effects of Tyrphostin A25, a general inhibitor of PTKs that interferes with binding of PTKs to target Tyr residues, because it has been shown to inhibit syndecan-1 shedding activated by various agonists (15,25). As shown in Table II, Tyrphostin A25 significantly inhibited syndecan-1 shedding activated by S. aureus culture supernatant and purified toxins, indicating that PTK activity is essential for S. aureus-activated shedding. Taken together, these results indicate that S. aureus exploits a PTK-dependent shedding mechanism of host cells to enhance syndecan-1 ectodomain shedding.
To further study the role of PTKs in S. aureus-activated syndecan-1 shedding, we next tested the effects of the following specific PTK inhibitors: herbimycin A, an inhibitor of Src family PTKs; PP2, an inhibitor of Lck, Fyn, and Hck PTKs belonging to the Src family PTKs; AG490, an inhibitor of JAK family PTKs; and piceatannol, an inhibitor of Syk family PTKs (45). As shown in Table II, piceatannol abrogated syndecan-1 ectodomain shedding augmented by S. aureus culture supernatant and purified ␣and ␤-toxins. Interestingly, AG490-inhibited ␤-toxin-activated, but not ␣-toxin-activated, syndecan-1 shedding. Herbimycin A and PP2 had no significant effect. Furthermore, AG490 and piceatannol did not inhibit shedding if they were preincubated with S. aureus toxins and removed prior to incubation with cells, indicating that the effects of AG490 and piceatannol were on the host cell and not on the toxins. The differential effects of specific PTK inhibitors on shedding suggest that ␣and ␤-toxins stimulate related, yet distinct, signaling pathways that converge to activate a common cleavage mechanism at the cell surface.
S. aureus ␤-Toxin, but Not ␣-Toxin, Activates Ectodomain Shedding of Syndecan-4 and HB-EGF-Ectodomain shedding of affected cell surface molecules is mediated by metalloproteinase sheddases (1). Our results indicate that S. aureusactivated syndecan-1 shedding is also mediated by the host cell sheddase, suggesting that S. aureus toxins may stimulate shedding of other host cell surface molecules. To test this hypothesis, we examined whether S. aureus ␣and ␤-toxins activate shedding of syndecan-4 and HB-EGF ectodomains. HB-EGF is a member of the EGF ligand family and binds to the EGF receptors ErbB-1 and -4 (46,47). Ectodomain shedding of HB-EGF has been implicated in various pathophysiological processes, such as wound repair (48), cardiac hypertrophy (19), and staphylococcal infections (18). Syndecan-4 is a member of the syndecan family of HSPGs and is expressed in most adult tissues, including epithelia, albeit at a lower level than that of syndecan-1 (33,34). The physiological function of syndecan-4 ectodomain shedding is not known. However, functions of syndecan-1 ectodomains are mediated by their HS chains (33,34), and HS chains of syndecan-1 and syndecan-4 obtained from the same cell type have been shown to be structurally similar (49). Thus, we speculate that syndecan-4 ectodomains may possess functions similar to that of syndecan-1 ectodomains.
To determine the effects of S. aureus toxins on syndecan-4 shedding, confluent cultures of NMuMG cells were incubated with various concentrations of S. aureus 8325-4 culture supernatant, purified ␣-toxin, or purified ␤-toxin and shed syndecan-4 was quantified by immunoblotting using the Ky8.2 antisyndecan-4 ectodomain antibody. S. aureus culture supernatant and purified ␤-toxin significantly stimulated syndecan-4 ectodomain shedding in a concentration-dependent manner and this shedding was inhibited by the hydroxamate sheddase inhibitor GM6001 (Fig. 4). In contrast, purified ␣-toxin did not augment syndecan-4 shedding at the three concentrations shown to enhance syndecan-1 shedding (Fig. 4).
The effect of S. aureus toxins on ectodomain shedding of HB-EGF was assessed by incubating CHO-K1 cells stably expressing an HA-tagged transmembrane construct of HB-EGF (proHB-EGF) with purified ␣and ␤-toxins (both at 10 g/ml). The extent of shedding was visualized by detecting HB-EGF ectodomains in the conditioned media and transmembrane proHB-EGF in cell lysates by Western immunoblotting using an anti-HA antibody. As shown in Fig. 5, the intensity of the bands corresponding to HB-EGF ectodomains (21-25 kDa) and proHB-EGF (25-32 kDa) were increased and decreased, respectively, in CHO-K1 cells incubated with 1 M PMA (positive control) or purified ␤-toxin. These results indicate that ␤-toxin stimulates HB-EGF ectodomain shedding. Furthermore, ␤-toxin-and PMA-activated HB-EGF shedding was inhibited by the peptide hydroxamate sheddase inhibitor. However, similar to its effect on syndecan-4 shedding, purified ␣-toxin did not stimulate HB-EGF ectodomain shedding. These findings indicate that ␣-toxin is a specific enhancer of syndecan-1 ectodomain shedding, whereas ␤-toxin can potentiate ectodomain shedding of several cell surface proteins. DISCUSSION Our study defines the molecular basis of how S. aureus activates syndecan-1 ectodomain shedding. We have shown previously that the majority of S. aureus strains secrete factors that stimulate syndecan-1 shedding (25). Several independent criteria provided by this study establish that S. aureus ␣-and ␤-toxins are the secreted enhancers of syndecan-1 shedding. First, analyses of several S. aureus mutant strains showed that the capacity to stimulate syndecan-1 shedding is absent in strains with mutations that eliminate both ␣and ␤-toxins and partially reduced in single ␣-toxin and ␤-toxin-deficient mutants. Second, inactivation of the Agr global regulatory locus, which positively regulates expression of ␣and ␤-toxin genes along with those encoding other exoproteins, abrogated the ability to stimulate syndecan-1 ectodomain shedding. Third, purified ␣and ␤-toxins rapidly activated ectodomain shedding in a concentration-dependent manner. Furthermore, activation of syndecan-1 shedding by both ␣ and ␤-toxins was inhibited by specific antagonists of the host cell shedding mechanism and the effects of these inhibitors were determined to be on the host cell and not on the staphylococcal toxins. These findings demonstrate that S. aureus ␣and ␤-toxins activate syndecan-1 ectodomain shedding via stimulation of the host cell shedding machinery.
S. aureus ␣-toxin is a major virulence factor of this bacterium. It exerts its cytotoxic effects by spontaneously forming heptameric pores on target cell membranes (39 -41). When ␣-toxin pore formation is extensive, cells are killed by loss of ATP and an imbalance of critical ions. Our results demonstrate that syndecan-1 ectodomain shedding is activated at ␣-toxin concentrations that are non-cytotoxic for epithelial cells, indicating that extensive pore formation is not required for shedding activation. Furthermore, pore formation alone is not a signal for syndecan-1 ectodomain shedding since our results suggest that ␥-toxin and other non-␣ pore forming toxins of S. aureus do not stimulate shedding. At low concentrations, ␣-toxin forms discrete pores in target cell membranes (44) so it is possible that formation of small, discrete pores signals to trigger syndecan-1 shedding. At non-cytotoxic concentrations, ␣-toxin stimulates phosphatidylinositol hydrolysis to generate inositol phosphate signaling mediators in A549 lung epithelial cells (44). Our results show that ␣-toxin-induced syndecan-1 shedding involves PTKs, possibly Syk, but not PKC and MAP kinase signaling pathways. Interestingly, streptolysin O has been shown to stimulate L-selectin shedding via discrete pore formation and this activity was not affected by PKC inhibitors (31). These findings suggest that, similar to streptolysin O, ␣-toxin activates syndecan-1 ectodomain shedding via small, discrete pore formation. Alternatively, pore formation may not be required and binding of ␣-toxin to a putative ␣-toxin receptor may trigger signals essential for syndecan-1 shedding. Low and high affinity binding sites on host cells have been described for ␣-toxin and the high affinity binding site is thought to be an ␣-toxin receptor protein (38). Therefore, an important issue of future studies will be to test these possibilities using mutant ␣-toxin constructs that lack the pore forming activity but retain the binding function or one that lacks binding to the high affinity site.
S. aureus ␤-toxin exerts its cytotoxic effects through its neutral sphingomyelinase activity. Hydrolysis of membrane sphingomyelin by ␤-toxin generates ceramide, a potent lipid second messenger, that regulates many signaling pathways (50). Results from this study show that ␤-toxin-activated syndecan-1 shedding is PTK-and sheddase-dependent. Importantly, we have previously shown that incubation of epithelial cells with membrane-permeable C8-ceramide also activates syndecan-1 shedding in a PTK-and sheddase-dependent manner (15). Thus, it is likely that ␤-toxin activates syndecan-1 ectodomain shedding via hydrolysis of membrane sphingomyelin and subsequent generation of ceramide.
Our results indicate that ␤-toxin also activates syndecan-4 and HB-EGF ectodomain shedding. Walev et al. (31) have shown that ␤-toxin stimulates L-selectin shedding via its sphingomyelinase activity. These observations indicate that ␤-toxin, via ceramide generation, likely enhances ectodomain shedding of a wide variety of cell surface proteins. However, how ceramide activates ectodomain shedding in a sheddase-dependent manner is not clear. Although available results indicate that TACE is not the syndecan sheddase, TACE can shed HB-EGF (8) and L-selectin (7), and the cytoplasmic domain of TACE can be Ser-phosphorylated (51). Because ceramide activates a Ser/ Thr kinase (50), Ser phosphorylation of TACE by this kinase may activate TACE-mediated shedding of HB-EGF and L-selectin ectodomains. Another possibility is stabilization of the putative syndecan sheddase and TACE since ceramide has been shown to stabilize ␤-secretase and augment its ability to shed ␤-amyloid precursor protein (52). Alternatively, the ability of ceramide to re-organize the cell membrane into a signaling platform and to cluster cell surface proteins (53) may augment shedding by increasing the physical proximity of cell Syndecan-4 ectodomain shedding is enhanced by S. aureus ␤-toxin, but not ␣-toxin. NMuMG cells were incubated with S. aureus 8325-4 supernatant at 5, 10, or 20% (v/v) or 10% supernatant plus 5 M GM6001 or purified ␣or ␤-toxin at 5, 10, or 20 g/ml or 10 g/ml plus 5 M GM6001 for 4 h at 37°C. Conditioned media were harvested and the relative amount of syndecan-4 ectodomains was measured using the Ky8.2 anti-syndecan-4 ectodomain antibody. Syndecan-4 ectodomains, shown as absorbance units (AU), were quantified by densitometric scanning of dot blots using the NIH Image software as described. Each bar represents the mean Ϯ S.E. of triplicate determinations.
FIG. 5. S. aureus ␤-toxin, but not ␣-toxin, stimulates the ectodomain shedding of HB-EGF. CHO-K1 cells stably expressing an HA-tagged transmembrane construct of HB-EGF (proHB-EGF) were pretreated with or without the KB-R8301 hydroxamate shedding inhibitor and then incubated with media (Ϫ), PMA (1 M), ␣-toxin (10 g/ml), or ␤-toxin (10 g/ml) for 30 min at 37°C. Conditioned media were collected and cells were lysed after extensive washes. HB-EGF ectodomains in the conditioned media were precipitated with heparin-Sepharose beads. HB-EGF ectodomains and proHB-EGF in total cell lysates were fractionated by 15% SDS-PAGE and detected by Western immunoblotting using the 9E10 anti-HA antibody. surface proteins and their sheddases.
Stimulation of syndecan-1 ectodomain shedding by ␤-toxin is regulated by PTKs, possibly JAK2 and Syk, based on our inhibitor studies. In fact, available data indicate that PTK activity is essential for all known agonists of syndecan-1 shedding (15,25). How PTKs regulate syndecan-1 shedding is not clear. However, all four mammalian syndecans are shed and contain three highly conserved Tyr residues in their cytoplasmic domain (33), and so far, Tyr phosphorylation of the cytoplasmic domains of syndecan-1 and -3 have been documented (54,55). It is not known whether syndecan shedding can be regulated by Tyr phosphorylation of its cytoplasmic domain, but ectodomain shedding of L1 has been shown to be regulated by Tyr phosphorylation of its cytoplasmic tail (26). These findings suggest that regulation of syndecan shedding by Tyr phosphorylation of their cytoplasmic domain warrants further investigation.
In summary, we have determined that S. aureus activates syndecan-1 ectodomain shedding via its two virulence factors, ␣and ␤-toxins. Both toxins activate shedding by exploiting the host cell's shedding machinery. Our results also revealed that ␣-toxin specifically activates syndecan-1 shedding, whereas ␤-toxin enhances ectodomain shedding of not only syndecan-1, but also of syndecan-4 and HB-EGF. The significance of these to the pathophysiology of S. aureus infection remains largely undefined. If S. aureus uses cell surface HSPGs as its attachment receptor, syndecan shedding may inhibit pathogenesis by interfering with its adhesion to host tissue components. However, available data indicate that syndecan-1 ectodomains facilitate P. aeruginosa lung infection by inhibiting various host defense factors (e.g. antimicrobials) through their HS chains (35), suggesting that syndecan-1 and -4 ectodomains may function similarly in S. aureus pathogenesis. Activation of HB-EGF ectodomain shedding has been suggested to contribute to the generation of viscous mucus fluids (18), which is a major deterrent to mucociliary clearance of microbial pathogens. These observations suggest that S. aureus activates ectodomain shedding of host components to promote its pathogenesis.