Ton Lisman, Surgical Research Laboratory, Department of Surgery, CMC V, Y2144, University Medical Center Groningen, Hanzeplein 1, 9713 GC, Groningen, the Netherlands. Tel.: +31 50 361 9028; fax: +31 50 363 2796. E-mail: email@example.com
Summary. Objectives: Staphylococcal superantigen-like 5 (SSL5) is an exoprotein secreted by Staphylococcus aureus that has been shown to inhibit neutrophil rolling over activated endothelial cells via a direct interaction with P-selectin glycoprotein ligand 1 (PSGL-1). Methods and Results: When purified recombinant SSL5 was added to washed platelets in an aggregometry set-up, complete and irreversible aggregation was observed. Proteolysis of the extracellular part of GPIbα or the addition of dRGDW abrogated platelet aggregation. When a mixture of isolated platelets and red cells was perfused over immobilized SSL5 at a shear rate of 300 s−1, stable platelet aggregates were observed, and platelet deposition was substantially reduced after proteolysis of GPIb or after addition of dRGDW. SSL5 was shown to interact with glycocalicin, a soluble GPIbα fragment, and binding of SSL5 to platelets resulted in GPIb-mediated signal transduction as evidenced by translocation of 14-3-3ζ. In addition, SSL5 was shown to interact with endothelial cell matrix (ECM) and this interaction enhanced aggregation of platelets from whole blood to this ECM. Conclusions: SSL5 activates and aggregates platelets in a GPIbα-dependent manner, which could be important in colonization of the vascular bed and evasion of the immune system by S. aureus.
Staphylococcus aureus (S. aureus) is a common human pathogen that induces both community-acquired as well as nosocomial infections. Infections caused by S. aureus are a growing concern, considering the increasing incidence of antibiotic-resistant strains, such as methicillin-resistant S. aureus (MRSA) . S. aureus causes infections that can be accompanied by infective endocarditis, sepsis and toxic shock syndrome [2,3]. Furthermore, infection with S. aureus may be accompanied by thrombocytopenia as a result of local platelet activation, and disseminated intravascular coagulation may occur when activation of hemostasis by the bacterium becomes systemic. In addition, infective endocarditis is regularly accompanied by embolic events such as stroke . The interaction of S. aureus with platelets has been studied extensively and it has been shown that S. aureus can activate platelets, and this is mediated by several surface-expressed proteins (reviewed by Fitzgerald et al.), such as clumping factor A and fibronectin-binding protein A. Furthermore, S. aureus is able to bind to adhered platelets, which might be an important mechanism contributing to the colonization of the vascular bed or damaged heart valves [6,7].
The invasiveness of S. aureus is dependent on a combination of surface-expressed virulence factors, such as clumping factor A and fibronectin-binding protein A, as well as excreted virulence factors, such as coagulase and alpha-toxin [3,8]. We and others have described several excreted virulence factors, such as staphylococcal complement inhibitor (SCIN) and the chemotaxis inhibitory protein of S. aureus (CHIPS), to be important in inhibiting pathogen clearance by the immune system. SCIN was shown to interfere with the complement system, whereas we have shown that CHIPS inhibits neutrophil chemotaxis [9,10]. CHIPS is closely homologous to another family of excreted factors referred to as staphylococcal superantigen-like proteins (SSLs). The SSL-family of proteins is encoded on staphylococcal pathogenicity island 2 and each strain of S. aureus expresses at least seven to a maximum of eleven SSLs. Recently, it has been suggested that SSL members 5 and 7 play a role in staphylococcal evasion of the immune system [11,12]. We previously showed that SSL5 inhibits PSGL-1-mediated neutrophil rolling on activated endothelium, and we speculated that this mechanism is important for immune evasion . Similar to PSGL-1, glycoprotein Ibα (GPIbα), one of the most abundant receptors on the platelet surface, is a tyrosine sulphated sialomucin. GPIbα is widely known as the receptor for von Willebrand factor (VWF) , but an increasing number of ligands, including coagulation factors IIa, VIIa, XII, XI(a), high-molecular weight kininogen, (activated) protein C, the leukocyte receptor MAC-1 and a dimeric form of beta2 glycoprotein I, have been described in recent years [14–21]. Recently, GPIbα was shown to interact with P-selectin , the main ligand of PSGL-1, and conversely, PSGL-1 was shown to interact with the main GPIbα ligand VWF . These observations led us to the hypothesis that more ligands may be able to bind both GPIbα and PSGL-1. As platelets are known to be an important mediator in the colonization of S. aureus, we hypothesized that SSL5 could have an effect on platelets via GPIbα. Here, we show that SSL5 is capable of activation, adhesion and aggregation of platelets, in which both GPIbα and αIIbβ3 play a role.
For methods, please see the data in Supporting Information.
Aggregation of washed platelets by SSL5 requires GPIb and αIIbβ3
When purified recombinant SSL5 was added to washed platelets in an aggregometry set-up, irreversible aggregation was observed as shown in Fig. 1(A). Maximal aggregation was observed with 400 nm of SSL5. Aggregation occurred in the absence of exogenously added fibrinogen, and addition of fibrinogen did not alter the aggregation profiles (data not shown). Aggregation was inhibited by dRGDW (200 μm), a peptide that blocks ligand binding to integrin αIIbβ3 (Fig. 1B). In addition, preincubation of platelets with the specific αIIbβ3-antagonist tirofiban inhibited platelet aggregation induced by SSL5 (Fig. 1B). Furthermore, treatment of platelets with O-sialoglycoprotein endopeptidase (OSE, 30 μg mL−1), which cleaves the extracellular part of GPIbα, reduced platelet aggregation substantially (Fig. 1B). Platelet aggregation was dependent on intracellular signaling, as treatment with a stable prostacyclin analog, iloprost (20 ng mL−1), inhibited platelet aggregation induced by SSL5 (Fig. 1B). As the interaction between SSL5 and PSGL-1 was shown to be critically dependent on sialic acid residues present on PSGL-1, we investigated the effect of neuraminidase treatment of platelets on aggregation induced by SSL5. As shown in Fig. 1(B), platelets treated with neuraminidase (0.2 U mL−1, 45 min) showed reduced aggregation upon addition of SSL5. SSL11, which was previously shown also to interact with the same sialic acid residues as described for SSL5 , did not result in any platelet aggregation at a concentration of 2 μm, suggesting a specific interaction between SSL5 and platelets (data not shown).
Although SSL5 was able to fully aggregate platelets at 400 nm in a washed platelet suspension, SSL5 did not induce any platelet aggregation in platelet-rich plasma (PRP), even after 60 min of incubation, with concentrations up to 4 μm of SSL5 (data not shown). This was not due to antibodies against SSL5, which are present in human plasma, as preincubation of washed platelets with pooled human immunoglobulins (IVIg, 5 mg mL−1) did not inhibit platelet aggregation induced by SSL5 (Fig. 1B). Also, the addition of fibrinogen (1 mg mL−1) had no effect on platelet aggregation induced by SSL5. In contrast, the presence of human serum albumin (HSA, 0.1%) inhibited platelet aggregation induced by SSL5 (Fig. 1B). However, subsequent experiments demonstrated that SSL5 was active towards platelets in a more physiological environment (see below).
SSL5 interacts with platelet GPIbα
Because OSE treatment reduced platelet aggregation by SSL5, we assessed a direct interaction between SSL5 and GPIbα. First, we immobilized SSL5 on a 96-well plate and incubated it with increasing concentrations of glycocalicin (GC), a plasma-purified extracellular part of GPIbα. As shown in Fig. 2(A), GC readily bound to SSL5. However, a recombinant truncated GPIbα (recGPIbα, residues 1–290) did not bind SSL5 (data not shown). Pretreatment of GC with neuraminidase resulted in a significant decrease of binding of GC to immobilized SSL5 (Fig. 2A). These results suggest that the binding site on GC for SSL5 is located in the highly glycosylated stack of GPIbα.
Subsequently, we investigated the binding of SSL5 to platelets using flow cytometry analysis. As shown in Fig. 2(B), FITC-labeled SSL5 (SSL5-FITC) was able to bind to washed platelets. Binding of SSL5-FITC to platelets was not affected by dRGDW, but pretreatment of platelets with OSE reduced the binding of SSL5-FITC by 65%, indicating a role for GPIbα in the initial interaction of SSL5 with platelets (Fig. 2C).
In addition, direct interaction between SSL5 and GC was investigated using surface plasmon resonance. GC was coated to a CM5 sensor chip via amine-coupling with an adsorption of 2400 RU and perfused with different concentrations of SSL5. Non-linear regression analysis resulted in an affinity constant (Kd) of 0.34 ± 0.11 μm (mean ± SD, Fig. 2D).
SSL5 triggers translocation of the GPIbα-associated adapter protein 14-3-3ζ
To assess consequences of the interaction of SSL5 with GPIbα on platelets, we examined 14-3-3ζ translocation to the actin cytoskeleton, which was previously shown to indicate a GPIbα-dependent signaling event [25,26]. As shown in Fig. 2(E), increased 14-3-3ζ association with the cytoskeleton was observed upon stimulation with SSL5. Removal of GPIbα by OSE partially prevented 14-3-3ζ translocation, whereas combined treatment with OSE and dRGDW fully abrogated SSL5-induced 14-3-3ζ translocation.
SSL5 interacts with platelet integrin αIIbβ3
As dRGDW blocked both SSL5-induced aggregation and 14-3-3ζ translocation, we assessed a possible interaction between SSL5 and platelet integrin αIIbβ3, using CHO cells transfected with αIIbβ3 (CHO-αIIbβ3). As shown in Fig. 3(A), CHO-αIIbβ3 adhered to immobilized SSL5, whereas untransfected CHO cells did not. Furthermore, direct interaction between SSL5 and αIIbβ3 was investigated using surface plasmon resonance. αIIbβ3 was coated to a CM5 sensor chip via amine-coupling with a maximum adsorption of 1500 RU and perfused with different concentrations of SSL5. Non-linear regression analysis resulted in an affinity constant (Kd) of 0.82 ± 0.13 μm (mean ± SD, Fig. 3B).
Platelet adhesion to immobilized SSL5 under flow conditions is dependent on GPIb and αIIbβ3
To examine whether platelets can interact with SSL5 under flow conditions, we perfused reconstituted blood (a mixture of washed platelets in HEPES-Tyrode buffer at pH 7.35 and isolated red cells, 40% hematocrit, 200 000 platelets per μL) over immobilized SSL5. Platelets readily adhered to immobilized SSL5 and were able to form large and stable aggregates (Fig. 4A, Movie S1). Platelet adhesion and aggregate size was dependent on shear rate and adhesion was observed with shear rates up to 1600 s−1 (data not shown). Maximal surface coverage was observed at 300 s−1, and therefore we continued investigation of platelet adhesion to SSL5 at 300 s−1. Proteolysis of GPIbα by OSE resulted in formation of much smaller aggregates as compared with adhesion of platelets with functional GPIbα (Fig. 4B,E). Furthermore, addition of dRGDW abrogated aggregate formation, and only single adhered platelets were observed (Fig. 4C,E). When platelets were pretreated with both OSE and dRGDW, platelet adhesion was almost completely abolished (Fig. 4D,E). Platelet activation was necessary for the interaction with immobilized SSL5, as iloprost (20 ng mL−1) completely abolished platelet adhesion (Fig. 4E). In contrast, an antibody against GPIbα (AK2, 10 μg mL−1), interfering with the binding site for VWF on GPIbα, did not inhibit the interaction of platelets with SSL5 (Fig. 4E).
These results suggest a role for GPIbα in initial adhesion to SSL5, after which αIIbβ3 is responsible for stable adhesion. To test this hypothesis, we performed real time analysis using differential interference contrast (DIC) microscopy and investigated the number of transient contacts as a measure for initial adhesion. As seen in Fig. 4(F), pretreatment of platelets with OSE resulted in a decrease in the number of transient contacts with SSL5 (Movie S2A). In contrast, dRGDW resulted in an increase in the number of transient contacts (Fig. 4F, Movie S2B), although stable platelet adhesion was slightly reduced (Fig. 4C,E). These results support our hypothesis that GPIbα is necessary for initial adhesion and αIIbβ3 is necessary for stable adhesion of platelets to immobilized SSL5.
The results discussed so far were obtained using plasma-free, reconstituted blood. When citrated or low-molecular weight heparin-anticoagulated whole blood was perfused over immobilized SSL5, no adhesion was observed. To investigate whether platelets present in whole blood interacted at all with immobilized SSL5, we analyzed platelet adhesion by real-time video microscopy. We observed only minor interaction of platelets from whole blood with immobilized SSL5. However, when whole blood was preactivated with the PAR-1 activating peptide SFLLRN (15 μm, 10 min), substantial platelet rolling and adhesion was observed, although these interactions did not lead to the formation of stable platelet aggregates (Fig. 4G, Movie S3). Again, pretreatment of platelets with OSE substantially reduced the number of transient contacts (25% ± 8% compared with control situation), while addition of dRGDW markedly increased the number of transient contacts (> 400% compared with control situation).
SSL5 increases platelet adhesion to endothelial cell matrix
To investigate the relevance of the SSL5-platelet interaction in a more physiologic context, we assessed platelet adhesion to endothelial cell matrix (ECM). First, we prepared ECM in a 96-well plate and subsequently incubated the ECM with increasing concentrations of SSL5. As shown in Fig. 5(A), SSL5 readily bound to ECM with an apparent Kd of 0.8 ± 0.3 μm (mean ± SD). In subsequent experiments we investigated SSL5 binding to purified components of the ECM and observed that SSL5 bound to vitronectin (Kd = 0.29 ± 0.08 μm, mean ± SD), but did not interact with fibronectin, fibrinogen, laminin or collagens type I to V (data not shown).
To investigate the effect of SSL5 on platelet adhesion to ECM, we preincubated ECM-coated coverslips with SSL5 (1 μm, 60 min) before perfusing with whole blood for 5 min at 300 s−1. As shown in Fig. 5(B), platelets primarily showed spreading on ECM, whereas they additionally formed large aggregates when ECM was pretreated with SSL5. Interestingly, aggregates could only be observed on locations of initial platelet adhesion, indicating that the platelets did not show stable adhesion to SSL5 itself, but that SSL5 increases platelet aggregation rather than platelet adhesion. This was further demonstrated by calculating the amount of surface that was covered by the formed aggregates. Total surface coverage was not significantly different in the presence or absence of SSL5. However, when the surface coverage of areas with high density of staining was calculated as a measure for formed aggregates, preincubation of ECM with SSL5 showed a significant increase in the amount of aggregates formed as compared with ECM alone. Figure 5(C) shows the amount of aggregates as a percentage of total surface coverage, which was significantly increased upon preincubation with SSL5.
S. aureus excretes a variety of soluble proteins that are pivotal in host infection. Specifically, many of these proteins facilitate evasion of the host immune system. Virulence factors such as CHIPS and SCIN have been extensively characterized, and recently, we have identified SSL5 as an additional immune modulator excreted by S. aureus. SSL5 was shown to inhibit neutrophil rolling by interference with the PSGL-1-P-selectin interaction . Here, we show that SSL5 interacts with GPIbα, resulting in activation of platelets. In addition, SSL5 interacts with αIIbβ3 and the combined interaction of SSL5 with both adhesive proteins is sufficiently strong to support platelet adhesion under conditions of flow.
Platelets play an important role in the colonization of subendothelial tissue by S. aureus. Therefore, the enhancement of platelet aggregation to ECM by SSL5 might be an important mechanism to attract S. aureus to ‘weak’ spots in the endothelial layer. Subsequently, increased platelet adhesion will cause the attraction of additional neutrophils to the site of injury. However, SSL5 has already been shown to inhibit the attraction of neutrophils. In this way, SSL5 will first cause enhanced platelet aggregation at the site of injury, which is beneficial for the colonization of subendothelial tissue by S. aureus, while at the same time it anticipates the counter effect of the immune system by interfering with the attraction of neutrophils (Fig. 6).
We showed SSL5 to interact directly with GPIbα, which was anticipated from the notion that due to analogy in structure, multiple proteins that interact with GPIbα are also able to interact with PSGL-1. The interaction of SSL5 with GPIbα triggers signaling that is sufficiently strong to result in complete platelet aggregation even in the absence of exogenous fibrinogen, as evidenced by inhibition of aggregation by specific proteolysis of GPIbα by OSE, and translocation of 14-3-3ζ. In addition, αIIbβ3 was shown to interact with SSL5, and this interaction contributes to the adhesion of platelets to immobilized SSL5 under conditions of flow. However, a potential role for yet another receptor in the interaction with SSL5 can not be ruled out. Taken together, SSL5 might interact with initially deposited platelets and thereby contribute to attraction of additional platelets, a mechanism we have previously postulated for GPIb-bound thrombin .
In our initial experiments, we studied platelet responses to SSL5 in plasma-free systems. Surprisingly, in platelet-rich plasma or whole blood, the potent activatory effects of SSL5 initially could not be replicated. At present, the component(s) blocking SSL5 function in plasma are unknown but may involve albumin, as the addition of human serum albumin significantly inhibited platelet aggregation induced by SSL5 (Fig. 1B). To investigate whether SSL5 may exert more subtle effects on platelets in plasma-containing environments, we performed experiments in a more physiologically relevant context. First we showed that although platelets in a whole blood system did not interact with immobilized SSL5 under flow conditions, substantial interactions were observed when platelets were preactivated by the PAR-1 activating peptide SFLLRN. In this model, no permanent platelet-SSL5 contacts could be demonstrated. In contrast, when we studied platelet adhesion to extracellular matrix generated from cultured primary endothelial cells, we observed SSL5 to significantly enhance platelet aggregate formation. Thus, in a plasma environment, SSL5 is not capable of directly activating platelets, but does facilitate platelet adhesion to a damaged vasculature. This implies that in vivo, platelets might need prior stimulation before bacterial proteins such as SSL5 can exert their activity, which prevents systemic platelet activation.
In conclusion, we have identified SSL5 as a novel ligand of GPIbα. In a purified system, the SSL5–GPIbα interaction is sufficiently potent to result in complete platelet aggregation. However, in a whole blood system, effects of SSL5 are more subtle, but still contribute significantly to a local increase in platelet adhesion and aggregation. This mechanism could stimulate the colonization of S. aureus, but may also contribute to some extent to S. aureus-induced thrombocytopenia. Inhibition of SSL5 could therefore be a potential target for treatment of S. aureus infection.
Contributions: C.W. performed and designed research and wrote the article. M.M.V. collected data. C.J.C. de H. and T.L. designed research and wrote the article. J.A. v. S and Ph.G. de G. critically read the manuscript
This research was supported in part by grants from the Netherlands Organisation for Scientific Research (NWO) (VENI 916.56.076) to T. Lisman and a scientific stimulating grant from the Division of Laboratory and Pharmacy of the University Medical Center Utrecht, the Netherlands, to T. Lisman and C.J.C. de Haas.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interests.