Staphylococcus aureus Alpha-Toxin Disrupts Endothelial-Cell Tight Junctions via Acid Sphingomyelinase and Ceramide

ABSTRACT

Staphylococcus aureus (S. aureus) infections are among the most common and severe infections, garnering notoriety in an era of increasing resistance to antibiotics. It is therefore important to define molecular mechanisms by which this pathogen attacks host cells. Here, we demonstrate that alpha-toxin, one of the major toxins of S. aureus, induces activation of acid sphingomyelinase and concomitant release of ceramide in endothelial cells treated with the toxin. Activation of acid sphingomyelinase by alpha-toxin is mediated via ADAM10. Infection experiments employing alpha-toxin-deficient S. aureus and the corresponding wild-type strain reveal that activation of acid sphingomyelinase in endothelial cells requires alpha-toxin expression by the pathogen. Activation of acid sphingomyelinase is linked to degradation of tight junctions in endothelial cells in vitro, which is blocked by pharmacological inhibition of acid sphingomyelinase. Most importantly, alpha-toxin induces severe degradation of tight junctions in the lung and causes lung edema in vivo, which is prevented by genetic deficiency of acid sphingomyelinase. These data indicate a novel and important role of the acid sphingomyelinase/ceramide system for the endothelial response to toxins and provide a molecular link between alpha-toxin and the degradation of tight junctions. The data also suggest that inhibition of acid sphingomyelinase may provide a novel treatment option to prevent lung edema caused by S. aureus alpha-toxin.

INTRODUCTION

Staphylococcus aureus commonly infects the skin and soft tissues, resulting in local symptoms such as purulent wound infections, but S. aureus can also cause severe infections such as pneumonia, endocarditis, osteomyelitis, sepsis, and toxic shock syndrome (1). In the United States alone, S. aureus causes approximately 500,000 infections and 20,000 deaths each year (2). The clinical situation is further complicated by the fact that many S. aureus strains have developed resistance to available antibiotics (3). In particular, methicillin-resistant S. aureus (MRSA) strains constitute a serious clinical problem (3). One of the most important consequences of S. aureus-induced sepsis is the induction of an often fatal pulmonary edema. Therefore, novel approaches to preventing and treating S. aureus-induced pulmonary edema are urgently needed.

Toxins that are released by S. aureus mediate a variety of pathophysiological effects of S. aureus. These toxins include enterotoxins and pore-forming toxins such as alpha-toxin, gamma-hemolysin, and Panton-Valentine leukocidin, which cause membrane damage, infiltration of neutrophils and macrophages, cytokine production, and increased vascular permeability, resulting in severe pulmonary edema and lung injury (4–7). The pore-forming alpha-toxin, which is one of the best-characterized virulence factors of S. aureus, is involved in the pathogenesis of skin infections, pneumonia, and sepsis. In a mouse model of S. aureus pneumonia, infection with mutant strains unable to produce alpha-toxin were much less severe, resulting in no mortality in contrast to 40% mortality with wild-type strains (8). Recently, it was reported that alpha-toxin binds to ADAM10 (a disintegrin and metalloprotease 10) as its eukaryotic receptor. Binding of alpha-toxin is necessary for alpha-toxin-induced cytotoxicity (9–11). Interaction of alpha-toxin with ADAM10 in epithelial and endothelial cells disrupts the cell barrier function, and this disruption contributes to the pathogenesis of lethal lung edema. Binding of alpha-toxin to ADAM10 is also required for proper assembly of the toxin and subsequent integration of alpha-toxin into cell membranes (12).

We along with others have previously shown that several bacterial pathogens, such as Neisseria gonorrhoeae, Listeria monocytogenes, Pseudomonas aeruginosa, Salmonella enterica serovar Typhimurium, Escherichia coli, Mycobacterium avium, Neisseria meningitidis, and S. aureus infect mammalian cells by exploiting the acid sphingomyelinase/ceramide system (13–20). The first study to show the involvement of the acid sphingomyelinase/ceramide system in bacterial infections demonstrated that acid sphingomyelinase is required for the internalization of N. gonorrhoeae into human epithelial cells (13). This notion was confirmed by many other studies demonstrating that the acid sphingomyelinase/ceramide system is also involved in the uptake of pathogens, the induction of host cell apoptosis, the maturation of phagosomes, and the fusion of phagosomes with lysosomes, as well as the controlled release of cytokines (14–20). For instance, we have previously shown that infecting endothelial cells with S. aureus or infecting epithelial cells with P. aeruginosa results in the activation of acid sphingomyelinase and a concomitant release of ceramide. The activation of the acid sphingomyelinase/ceramide system mediates the uptake of S. aureus and P. aeruginosa (15, 20–22). Additional studies showed that intracellular killing of Mycobacterium avium or L. monocytogenes after infection of macrophages with these pathogens requires expression of acid sphingomyelinase (14). Acid sphingomyelinase mediates the complete fusion of phagosomal and lysosomal membranes to allow for phagolysosomal killing and digestion of the pathogen. The absence of acid sphingomyelinase greatly impairs this defense mechanism and allows intracellular survival of the pathogen (14).

The molecular mechanisms by which ceramide mediates these diverse effects and the intracellular signaling pathways triggered by ceramide are still largely unknown.

We have previously shown that ceramide molecules released from sphingomyelin by the activity of acid sphingomyelinase spontaneously aggregate to form small microdomains that fuse to large domains, called ceramide-enriched membrane platforms (23). These platforms serve to cluster activated receptor molecules and to recruit intracellular signaling molecules mediating the effects of the cognate receptors; thereby, ceramide permits and amplifies signal transduction without being a second messenger on its own (23–29). Thus, ceramide-enriched membrane platforms serve to reorganize receptors and signaling molecules, thereby mediating stress signaling and infection (23–29). In addition to reorganizing signaling systems, ceramide has also been shown to directly bind proteins, namely, cathepsin D, phospholipase A₂, and some protein kinase C isoforms, and ceramide seems to be directly involved in the regulation of the topology or activity of these proteins (30–32).

However, the role of the acid sphingomyelinase/ceramide system in the cellular response to bacterial toxins is largely unknown. Therefore, we tested whether S. aureus alpha-toxin activates the acid sphingomyelinase/ceramide system and whether this activation is linked to the effects of the toxin on endothelial tight junctions and the development of pulmonary edema.

RESULTS

To test whether alpha-toxin, one of the major toxins of S. aureus, activates acid sphingomyelinase and triggers formation of ceramide, we treated bEnd.3 and EOMA endothelial cells with purified S. aureus alpha-toxin and determined the activity of acid sphingomyelinase and the release of ceramide in these cells over time. The results reveal a rapid and marked activation of acid sphingomyelinase (Fig. 1A) and a concomitant formation of ceramide within the endothelial cells (Fig. 2A and B). Activation of acid sphingomyelinase was already observed 5 min after stimulation with alpha-toxin; the activity peaked at 5 and 10 min after stimulation and declined thereafter (Fig. 1). Release of ceramide was slightly delayed and peaked around 10 and 15 min and also decreased thereafter. Similar results were obtained with a second endothelial cell line, i.e., EOMA cells (Fig. 2B). Interestingly, the confocal microscopy studies revealed that ceramide was predominantly formed in intracellular compartments, while only a small part seemed to be released in the plasma membrane (Fig. 2B).

FIG 1.

S. aureus alpha-toxin activates acid sphingomyelinase. bEnd.3 cells were incubated with 10 μg/ml alpha-toxin for the indicated times and lysed, and the activity of acid sphingomyelinase (Asm) was determined by measuring the consumption of [¹⁴C]sphingomyelin to [¹⁴C]phosphorylcholine and ceramide. Data shown are the means ± SD of four independent experiments. ***, P < 0.001 (ANOVA).

FIG 2.

S. aureus alpha-toxin induces predominantly intracellular ceramide. (A) bEnd.3 endothelial cells were stimulated with 10 μg/ml alpha-toxin for the indicated times and organically extracted, and ceramide was quantified using DAG kinase assays. Data are the means ± SD of four independent studies. *, P < 0.05; ***, P < 0.001 (ANOVA). (B) EOMA cells were treated with 10 μg/ml alpha-toxin for 10 min, fixed for 10 min in 1% PFA, permeabilized, stained with Cy3-labeled anti-ceramide antibodies, and examined by confocal microscopy. Shown is a result representative of four similar studies with the analysis of at least 250 cells per experiment, i.e., a total of at least 1,000 cells.

Control studies measuring exclusion of trypan blue or propidium iodide revealed no damage (33) or even lysis of the endothelial cells upon treatment with alpha-toxin for 120 min.

To further test the role of alpha-toxin for acid sphingomyelinase activation, we infected bEnd.3 cells with a wild-type strain and alpha-toxin-deficient mutant of S. aureus, named JE2 wild-type and JE2 Δhla. The results indicate rapid activation of acid sphingomyelinase by the wild-type S. aureus strain, while the alpha-toxin-deficient JE2 Δhla mutant was without any effect on acid sphingomyelinase activity (Fig. 3A). Comparison of acid sphingomyelinase activation after S. aureus infection with the activation of the enzyme induced upon infection of bEnd.3 cells with the highly virulent P. aeruginosa strain 762 revealed similar levels of activation of acid sphingomyelinase by both pathogens (Fig. 3A). Activation of acid sphingomyelinase by purified alpha-toxin (Fig. 1) was approximately 2-fold stronger, which is most likely due to the much more synchronous activation of the cells upon treatment with the water-soluble toxin than after infection with the much larger bacteria.

FIG 3.

Expression of alpha-toxin is required for activation of acid sphingomyelinase upon infection of endothelial cells with S. aureus. (A) bEnd.3 endothelial cells were infected with a wild-type strain (JE2 wt) and the corresponding alpha-toxin-deficient mutant strain (JE2 Δhla) of S. aureus for the indicated times, lysed, and diluted, and the activity of acid sphingomyelinase (Asm) was measured. For comparison, cells were also infected with the P. aeruginosa (P.a.) strain 762. Data shown are the means ± SD from four independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (ANOVA). (B) bEnd.3 cells rapidly internalize S. aureus strain JE2. Cells were infected, extracellular bacteria were killed by incubation for 60 min with gentamicin, intracellular bacteria were released and plated, and CFU counts were determined. Data shown are the means ± SD of four independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (ANOVA).

The kinetics of the activation of acid sphingomyelinase after alpha-toxin and infection with S. aureus are slightly different. A rapid internalization of the bacteria into endothelial cells, which may terminate activation of acid sphingomyelinase after infection, might contribute to the faster kinetics of acid sphingomyelinase activation after infections with the intact pathogens. We therefore determined bacterial internalization and observed rapid and massive internalization of the bacteria (Fig. 3B).

In order to analyze mechanisms that mediate activation of acid sphingomyelinase, we preincubated bEnd.3 cells with the ADAM10 inhibitor GI254023X (34), with β-cyclodextrin that blocks the toxin by binding to the toxin heptamer (35), and with amitriptyline, a functional inhibitor of acid sphingomyelinase (36–38), and determined the activity of acid sphingomyelinase upon incubation with alpha-toxin. These results revealed that the ADAM10 inhibitor GI254023X prevented activation of acid sphingomyelinase by alpha-toxin (Fig. 4). Likewise, preincubation of the cells with cyclodextrin markedly reduced acid sphingomyelinase activation upon cellular stimulation with alpha-toxin (Fig. 4). The two inhibitors did not alter the basal activity of acid sphingomyelinase prior to treatment with alpha-toxin. Amitriptyline reduced the basal activity of acid sphingomyelinase by 46% prior to alpha-toxin treatment and also prevented the stimulation of acid sphingomyelinase activity by alpha-toxin (Fig. 4).

FIG 4.

Inhibition of ADAM10 or alpha-toxin assembly prevents activation of acid sphingomyelinase. bEnd.3 cells were treated for 18 h with the ADAM10 inhibitor GI254023X (10 μM), for 30 min with β-cyclodextrin (1 mM), which interferes with alpha-toxin assembly, or for 30 min with amitriptyline (30 μM), a functional inhibitor of acid sphingomyelinase, and treated with 10 μg/ml alpha-toxin for 10 min. Cells were then lysed, and the activity of acid sphingomyelinase was measured. The ADAM10 inhibitor and β-cyclodextrin inhibit activation of acid sphingomyelinase but do not change the basal activity of acid sphingomyelinase in endothelial cells. Amitriptyline reduces the basal activity, consistent with its mode of action, and also prevents stimulation of the enzyme upon treatment of the cells with alpha-toxin. Data shown are the means ± SDs (n = 4). **, P < 0.01; ***, P < 0.001 (ANOVA).

We have previously shown that the acid sphingomyelinase/ceramide system is involved in the destruction of tight junction proteins upon infection of endothelial cells with S. aureus (20). To test whether alpha-toxin is already sufficient to mediate destruction of tight junctions in endothelial cells, we treated bEnd.3 cells with alpha-toxin and determined the integrity of tight junctions by staining with anti-zonula occludens-1 (ZO-1) antibodies and by confocal microscopy analysis. These studies revealed a marked downregulation of ZO-1 in endothelial cells upon treatment with alpha-toxin, indicating destruction of tight junctions already by the toxin itself (Fig. 5A and B). Pretreatment of endothelial cells with the functional acid sphingomyelinase inhibitor amitriptyline almost completely prevented the degradation of the tight junction protein ZO-1 (Fig. 5A and B), indicating that alpha-toxin mediates the degradation of tight junctions via activation of the acid sphingomyelinase/ceramide system. Similar results were obtained for ZO-2 and occludin.

FIG 5.

S. aureus alpha-toxin induces a degradation of tight junctions in vitro, which is prevented by inhibition of acid sphingomyelinase or ADAM10. (A) bEnd.3 endothelial cells were grown on coverslips, incubated with 10 μg/ml alpha-toxin for 120 min, fixed in 1% buffered PFA, and immunostained with Cy3-coupled anti-ZO-1 antibodies. As indicated, cells were incubated with 30 μM amitriptyline for 30 min prior to application of the toxin to functionally inhibit acid sphingomyelinase. Shown is a representative result of four independent experiments with similar results. (B) The fluorescence intensity of ZO-1 was quantified in 25 randomly chosen cells/sample. The values from the four independent samples were averaged. Thus, a total of 100 values were used in calculations. Data shown are the means ± SD. ***, P < 0.001 (ANOVA). (C and D) Preincubation of bEnd.3 cells with the ADAM10 inhibitor (Inh) GI254023X (10 μM) prevents alpha-toxin-induced degradation of ZO-1. Panel C shows a typical result from four independent studies, and panel D shows the means ± SD of the quantitative analysis of the ZO-1 fluorescence in each set of 25 cells per sample. The values are from four independent experiments. ***, P < 0.001 (ANOVA).

We also investigated the effect of the ADAM10 inhibitor GI254023X on ZO-1 expression and topology after treatment of bEnd.3 cells with alpha-toxin. The results show that inhibition of ADAM10 prevents alpha-toxin-induced changes of ZO-1 (Fig. 5C and D).

To test the significance of these in vitro findings for the in vivo development of lung edema after treatment of mice with alpha-toxin, we treated mice with intravenous (i.v.) injections of alpha-toxin and determined the integrity of tight junctions in the lung by staining of lung sections with anti-ZO-1, anti-ZO-2, or anti-occludin antibody. To unambiguously define the role of acid sphingomyelinase in the effect of alpha-toxin in vivo, we employed acid sphingomyelinase-deficient mice and compared the effect of alpha-toxin in these mice with that in wild-type mice. While injection of alpha-toxin into wild-type mice resulted in severe lung edema (Fig. 6A) and a marked degradation of the tight junction proteins ZO-2 (Fig. 6B), ZO-1, and occludin (data not shown) in endothelial cells, the effects of the toxin were almost completely abrogated in acid sphingomyelinase-deficient mice (Fig. 6A and B).

FIG 6.

S. aureus alpha-toxin triggers degradation of tight junctions via acid sphingomyelinase in vivo. Mice were given intravenous injections of 50 μg/kg alpha-toxin. We determined extravasation of Evans blue into lung tissue and the integrity of the tight junction protein ZO-2. (A) Mice were injected with 4% Evans blue dye (20 mg/kg; Sigma) and sacrificed after 30 min, and lungs were flushed with saline via the right heart, removed, dried, extracted in formamide (Sigma), and centrifuged (A). Supernatants were measured at the absorbance at 620 nm. Values shown are means ± SD or representative data from at least four mice. ***, P < 0.001 (ANOVA). (B) S. aureus alpha-toxin induces degradation of tight junctions in the lung. Wild-type (wt) and acid sphingomyelinase-deficient (Asm^−/−) mice were i.v. injected with alpha-toxin, paraffin sections were obtained, and ZO-2 was stained with Cy3-coupled anti-ZO-2 antibodies. Displayed are representative results from five mice per group.

DISCUSSION

Our findings indicate that S. aureus alpha-toxin induces massive changes in the lung, causing severe pulmonary edema, hypoxia, and vascular dysfunction. All of these events strictly depend on expression of acid sphingomyelinase and are absent from acid sphingomyelinase-deficient mice or endothelial cells treated with the functional acid sphingomyelinase inhibitor amitriptyline, a finding indicating that this enzyme is necessary for development of the pathophysiological effects of S. aureus toxins.

We have previously shown that genetic deficiency or pharmacologic inhibition of the acid sphingomyelinase/ceramide system protects against pulmonary edema induced by systemic infection with intact S. aureus (20). Here, we extend these studies and demonstrate that alpha-toxin is one of the factors synthesized by S. aureus that mediates the activation of acid sphingomyelinase and thereby the detrimental effects of the pathogen on endothelial cells. The experiments performed with purified alpha-toxin were confirmed by the infection studies employing wild-type and alpha-toxin-deficient S. aureus strains. These studies reveal the central role of alpha-toxin for activation of acid sphingomyelinase by S. aureus. Other factors besides alpha-toxin might also contribute to the stimulation of acid sphingomyelinase upon infection of endothelial cells with S. aureus; however, alpha-toxin is clearly the most important factor.

Degradation of tight junctions is one of the most important steps in the multistep process of disruption of endothelial cell integrity, the resultant increase in vascular permeability, and the development of pulmonary edema during systemic infections with S. aureus. Although our studies certainly do not exclude pulmonary edema caused by either direct interaction of the pathogen with the host cell or other toxins, such as Panton-Valentine leukocidin or enterotoxin B and A, which are also known to cause membrane damage (38, 39), cytokine production (40, 41), and increased vascular permeability (42, 43), they clearly indicate that alpha-toxin is one of the major factors involved in this process.

At present it is unknown how a bacterial toxin activates acid sphingomyelinase. It has been previously shown that alpha-toxin binds to ADAM10 (9). Our data employing an ADAM10 inhibitor and β-cyclodextrin suggest that alpha-toxin induces activation of acid sphingomyelinase via binding to ADAM10, ADAM10-mediated assembly and oligomerization of the toxin (12), and integration into the plasma membrane. This is consistent with the data demonstrating a predominantly intracellular release of ceramide, suggesting that uptake of the toxin is required for the action of the toxin on acid sphingomyelinase. Acid sphingomyelinase localizes to lysosomes, secretory lysosomes, and the cell surface (44). Secretory lysosomes do not participate in uptake pathways, and it seems unlikely that acid sphingomyelinase is targeted by alpha-toxin within these vesicles. It is therefore possible that alpha-toxin activates acid sphingomyelinase in lysosomes. Alternatively, it might be possible that incorporation of the toxin into membranes triggers a repair response with the fusion of secretory lysosomes with the plasma membrane and concomitant surface exposure of acid sphingomyelinase. Surface acid sphingomyelinase might be rapidly internalized, resulting in the formation of intracellular ceramide. The activation mechanisms of intracellular acid sphingomyelinase are poorly defined. Reactive oxygen species have been previously shown to stimulate acid sphingomyelinase via redox mechanisms (45–47), a process that might also occur in intracellular vesicles. In addition, the enzyme has been shown to be activated by limited cleavage (48), which might occur in multivesicular bodies. However, it seems to be unlikely that alpha-toxin triggers the formation of multivesicular bodies. ADAM10 is a crucial factor for alpha-toxin-induced cytotoxicity (9–12), and its activation induces disruption of tight junctions and degradation of tight junction proteins such as E-cadherin (10, 49). Since ADAM10 is a metalloprotease, it is tempting to speculate that activation of ADAM10 by alpha-toxin results in limited cleavage of acid sphingomyelinase in internalized vesicles and, thereby, in stimulation of acid sphingomyelinase and release of ceramide to execute tight junction disruption.

It is unknown how activation of acid sphingomyelinase is linked to degradation of tight junction proteins. It might be possible that acid sphingomyelinase and ceramide activate lysosomal proteases, for instance, cathepsin D (30), that might be released from lysosomes and that directly or indirectly initiate tight junction degradation.

Incubation of the cells with amitriptyline results in functional inhibition of acid sphingomyelinase, consistent with previous reports (36, 37, 50). Amitriptyline displaces acid sphingomyelinase from the inner lysosomal membrane, resulting in degradation of the enzyme. Interestingly, amitriptyline treatment resulted in almost complete abrogation of the activation of acid sphingomyelinase by alpha-toxin. This suggests that different pools of acid sphingomyelinase exist that might be involved either in the response of the enzyme to exogenous stimuli and thereby the signaling functions of this enzyme or in lysosomal maintenance and degradation of sphingomyelin. The different pools seem to have differential sensitivities to amitriptyline.

S. aureus is one of the major pathogens responsible for the development of sepsis (1), and therefore S. aureus infections are clinically very important. Even with the use of appropriate antibiotics, many patients still develop fatal pulmonary edema (51–53). Here, we provide a mechanism for this phenomenon via circulating alpha-toxin released after antibiotic-induced bacterial death. Amitriptyline could potentially prevent these adverse effects of S. aureus toxins on endothelial cell layers and development of pulmonary edema. The drug, which is routinely used to treat major depression, might be a novel approach, in conjunction with the appropriate antibiotics, to treat systemic S. aureus infections.

MATERIALS AND METHODS

Mice, cells, and treatments.

Wild-type and acid sphingomyelinase-deficient mice (Smpd1^−/−) littermates (54) were originally from R. Kolesnick, Memorial Sloan-Kettering Cancer Center, New York, NY. Mice were bred on a C57BL/6H background. Acid sphingomyelinase-deficient mice were used at an age of 6 weeks to avoid interference with sphingomyelin accumulation (55, 56). The alpha-toxin was applied intravenously at a sublethal dose of 50 μg/kg (Sigma). Amitriptyline was administered intraperitoneally (i.p.) at 10 mg/kg twice daily for 2.5 days prior to a single i.p. injection of 50 μg/kg alpha-toxin (Sigma). To determine Evans blue leakage into lung tissue, we injected mice intravenously with 4% Evans blue dye (20 mg/kg; Sigma) 30 min prior to sacrificing the mice; the lungs were thoroughly flushed with saline via the right heart, removed, dried, extracted in formamide (Sigma), and centrifuged, and supernatants were measured at 620 nm with a fluorescence microplate reader (BMG Labtech). Background was determined by measuring absorbance at 740 nm. All studies were performed in accordance with animal permissions of the Regierungspraesidium Düsseldorf and the Institutional Animal Care and Use Committee, Cincinnati, OH.

For in vitro studies we used two endothelial cell lines, i.e., bEnd.3 and EOMA cells (both from ATCC). Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 mM HEPES (pH 7.4) (Carl Roth GmbH), 2 mM l-glutamine, 1 mM sodium pyruvate, 100 μM nonessential amino acids, 100 units/ml penicillin, and 100 μg/ml streptomycin (all from Invitrogen). Cells were cultured in 24-well plates for 2 days prior to stimulation, washed twice in HEPES-saline (H-S; 132 mM NaCl, 20 mM HEPES [pH 7.4], 5 mM KCl, 1 mM CaCl₂, 0.7 mM MgCl₂, 0.8 mM MgSO₄) and stimulated in 300 μl of H-S with 10 μg/ml alpha-toxin for 10 min. Amitriptyline and cyclodextrin were added 30 min prior to stimulation at 30 μM and 1 mM, respectively (15, 36, 50, 57, 58). Inhibition of ADAM10 was achieved by incubation with 10 μM GI254023X for 18 h (34). All inhibitors were present during the treatment with alpha-toxin, and they were added again after the cells were washed.

Infection experiments with Staphylococcus aureus and Pseudomonas aeruginosa.

The S. aureus strain JE2 and an alpha-toxin-deficient mutant (JE2 hla::ΦNΣ, termed JE2 Δhla) were employed to infect bEnd.3 cells as previously described (15, 59–62). A total of 10⁵ bEnd.3 cells were grown in 24-well plates for 48 h. Bacteria were grown overnight on 5% sheep blood-Trypticase soy agar (TSA) plates (BD), removed from the plates, and suspended in 40 ml of prewarmed tryptic soy broth (TSB) (BD). The density of the bacteria was adjusted to an optical density of 0.2 to 0.25. The bacteria were then grown to an early logarithmic phase for 75 min at 37°C with shaking at 125 rpm. The bacteria were pelleted at 3,000 rpm for 10 min, washed twice in DMEM (Invitrogen) supplemented with 10 mM HEPES, and finally suspended in the same buffer. Cells were infected at a multiplicity of infection (MOI) of 1:500. The bacteria were centrifuged onto the cells to promote synchronization of the infection. Infection was terminated by removing the medium containing the bacteria and lysis in 300 μl of 250 mM sodium acetate (pH 5.0) and 1% NP-40 on ice for 10 min. Cells were then scraped off the plates, transferred into Eppendorf tubes, and sonicated once with a tip sonicator to achieve complete lysis. The lysates were then diluted to 0.1% NP-40 and used to determine the activity of acid sphingomyelinase as described below.

Infection with Pseudomonas aeruginosa (P. aeruginosa) strain 762 (15, 59–61) was performed as described for S. aureus with a few modifications: P. aeruginosa was grown overnight on TSA plates (BD), grown to early logarithmic phase in TSB for 60 min, and centrifuged at 2,800 rpm. All other steps were the same.

Ceramide measurements.

Endothelial cells were stimulated and lysed in 250 mM sodium acetate (pH 5.0) and 1% NP-40 as described above, shock frozen, and thawed, and 100 μl of the lysates plus 100 μl of H₂O was added to CHCl₃-CH₃OH–1 N HCl (100:100:1, vol/vol/vol). Phases were separated by 5 min of centrifugation at 14,000 rpm, and the lower phase was dried. The pellet was suspended in 7.5% (wt/vol) n-octyl glucopyranoside and 5 mM cardiolipin in 1 mM diethylenetriaminepentaacetic acid (DTPA). The kinase reaction was initiated by addition of 0.01 units of diacylglycerol (DAG) kinase (GE Healthcare Europe) in 0.1 M imidazole-HCl (pH 6.6), 0.2 mM DTPA (pH 6.6), 70 mM NaCl, 17 mM MgCl₂, 1.4 mM ethylene glycol tetraacetic acid, 2 mM dithiothreitol, 1 μM ATP, and 10 μCi of [γ-³²P]ATP. The kinase reaction was performed for 60 min at 30°C and terminated by addition of 1 ml of CHCl₃-CH₃OH–1 N HCl (100:100:1, vol/vol/vol) followed by addition of 170 μl of buffered saline solution (135 mM NaCl, 1.5 mM CaCl₂, 0.5 mM MgCl₂, 5.6 mM glucose, 10 mM HEPES, pH 7.2) and 30 μl of 100 mM EDTA. The lower phase was collected, dried, dissolved in 20 μl of CHCl₃-CH₃OH (1:1, vol/vol), and separated on silica G60 thin-layer chromatography (TLC) plates using CHCl₃-acetone-CH₃OH-acetic acid-H₂O (50:20:15:10:5, vol/vol/vol/vol/vol) as the developing solvent for ceramide. The TLC plates were analyzed using a phosphorimager (Fuji). Ceramide was determined using a standard curve of C₁₆ to C₂₄ ceramides.

Acid sphingomyelinase measurements.

Cells were stimulated with alpha-toxin and shock frozen, and 50-μl aliquots were added to 250 mM sodium acetate (pH 5.0) and 1% NP-40 for 10 min. The tissues were then homogenized by 10 s of sonication using a tip sonicator. Preparation of lysates after infection with wild-type JE2 and JE2 Δhla S. aureus and with P. aeruginosa has been described above. All lysates were diluted to 0.1% NP-40–250 mM sodium acetate (pH 5.0) and were then incubated with 50 nCi per sample of [¹⁴C]sphingomyelin for 30 min at 37°C. The enzyme reaction was terminated by addition of 800 μl of chloroform-methanol (2:1, vol/vol), phases were separated by centrifugation, aliquots of the upper phase containing the released [¹⁴C]phosphorylcholine were removed, and liquid scintillation counting was performed.

Immunohistochemical analysis of tight junction proteins ZO-1, ZO-2, and occludin.

Animals were injected i.v. with 50 μg/kg alpha-toxin, a sublethal dose of the toxin; they were sacrificed after 48 h, and the lungs were immediately removed and fixed for 42 h in 4% paraformaldehyde (PFA) buffered in phosphate-buffered saline (PBS; pH 7.4). The specimens were then dehydrated through an alcohol series and xylol, embedded in paraffin, and trimmed to 6 μm. Sections were dewaxed and washed in PBS, and antigens were retrieved by incubation for 30 min with Pepsin Digest All (Invitrogen Life Technologies) at 37°C. The specimens were washed again in PBS, and unspecific antibody binding sites were blocked for 10 min with 10% fetal calf serum (FCS) in PBS. Sections were washed once in H-S and immunostained with anti-ZO-1, anti-ZO-2 (Santa Cruz Inc.), or anti-occludin (Invitrogen) antibody for 45 min at room temperature. Antibodies were diluted 1:100 in H-S plus 1% FCS. The samples were then washed three times in PBS–0.05% Tween 20 and once in PBS. Thereafter, the specimens were incubated for 45 min at room temperature with Cy3-coupled anti-rabbit IgG F(ab)₂ fragments (Jackson ImmunoResearch) and washed again three times in PBS–0.05% Tween 20 and once in PBS before being embedded in Mowiol.

Cells were cultured on glass slides in 24-well plates, washed two times in H-S, and stimulated in 200 μl of H-S with 10 μg/ml alpha-toxin for 120 min. Stimulation was terminated by removal of the medium and 10 min of fixation in 1% PBS-buffered PFA (pH 7.4). Coverslips were washed in PBS and stained with the primary antibodies as described for the tissue sections. Finally, the glass coverslips were embedded in Mowiol. All samples were analyzed on a Leica TCS SL confocal microscope (Leica).

Fluorescence signals were quantified using ImageJ and expressed as arbitrary units (au). For each slide, 25 randomly selected cells were investigated, and the fluorescence staining of tight junction proteins was quantified and averaged with the values obtained in the other photos of the corresponding studies.

S. aureus internalization.

bEnd.3 cells were infected with S. aureus strain JE2, washed, and incubated for 60 min with 100 μg/ml gentamicin, which kills extracellular bacteria while the cells are impermeable for gentamicin, at least for the short incubation time. The samples were washed and lysed in 5 mg/ml saponin for 10 min to release intracellular bacteria; the samples were centrifuged at 3,000 rpm for 10 min, and the pellets were resuspended in sterile H-S. Aliquots were plated on LB plates and cultured overnight, and CFU were counted.

Statistical analysis.

Data are expressed as arithmetic means ± standard deviations (SD). We employed Student's t test to compare two groups and one-way analysis of variance (ANOVA) for more than two groups, followed by post hoc Student's t tests for all pairwise comparisons and applying a Bonferroni correction for multiple testing. The Bonferroni correction was applied prior to calculation of P values for the pairwise comparisons. All values were normally distributed. Sample size planning for in vivo infections was based on two-sided Wilcoxon-Mann-Whitney tests (G*Power, version 3.1.7; University of Düsseldorf, Germany). Investigators were blinded for histology experiments, animal identity, and quantification of all histology studies.