Staphylococcus aureus is a Gram-positive human pathogen that is readily internalized by professional phagocytes such as macrophages and neutrophils but also by non-professional phagocytes such as epithelial or endothelial cells. Intracellular bacteria have been proposed to play a role in evasion of the innate immune system and may also lead to dissemination within migrating phagocytes. Further, S. aureus efficiently lyses host cells with a battery of cytolytic toxins. Recently, phenol-soluble modulins (PSM) have been identified to comprise a genus-specific family of cytolytic peptides. Of these the PSMα peptides have been implicated in killing polymorphonuclear leucocytes after phagocytosis. We questioned if the peptides were active in destroying endosomal membranes to avoid lysosomal killing of the pathogen and monitored integrity of infected host cell endosomes by measuring the acidity of the intracellular bacterial microenvironment via flow cytometry and by a reporter recruitment technique. Isogenic mutants of the methicillin-resistant S. aureus (MRSA) strains USA300 LAC, USA400 MW2 as well as the strongly cytolytic methicillin-sensitive strain 6850 were compared with their respective wild type strains. In all three genetic backgrounds, PSMα mutants were unable to escape from phagosomes in non-professional (293, HeLa, EAhy.926) and professional phagocytes (THP-1), whereas mutants in PSMβ and δ-toxin as well as β-toxin, phosphatidyl inositol-dependent phospholipase C and Panton Valentine leucotoxin escaped with efficiencies of the parental strains. S. aureus replicated intracellularly only in presence of a functional PSMα operon thereby illustrating that bacteria grow in the host cell cytoplasm upon phagosomal escape.
Staphylococcus aureus is a leading cause of severe bacterial infections. Besides healthcare-associated methicillin-resistant S. aureus (HA-MRSA), community-associated MRSA (CA-MRSA) has emerged (Maree et al., 2007), which spreads epidemically and infects patients without predisposing risk factors (Herold et al., 1998). The bacteria are actively taken up by phagocytes such as polymorphonuclear neutrophils (PMN) and macrophages during infection. Recent publications showed that the bacteria also efficiently invade non-professional phagocytes such as epithelial and endothelial cells, fibroblasts, osteoblasts and keratinocytes (Dziewanowska et al., 1999; Jevon et al., 1999; Lammers et al., 1999; Peacock et al., 1999; Sinha et al., 1999; Fowler et al., 2000; Ahmed et al., 2001; Kintarak et al., 2004; Edwards et al., 2011). Further it has been observed that S. aureus can escape from host cell phagosomes (Bayles et al., 1998). A functional agr quorum sensing system is required for this immune evasive strategy of the pathogen (Shompole et al., 2003; Schnaith et al., 2007) and agr activation precedes translocation to the host cell cytoplasm (Qazi et al., 2001). Different bacterial pathogens have acquired a multitude of strategies to avoid phagolysosomal killing: whereas some bacterial pathogens such as Mycobacterium tuberculosis, Legionella pneumophila and Chlamydia spp. avoid phagolysosomes by arresting or delaying the maturation of the endocytic vesicles (reviewed in Haas, 2007), some pathogens are able to destroy endocytic membranes thereby translocating to the host cell cytoplasm. Listeria monocytogenes is a well-characterized model organism for vacuole membrane disruption which is mediated by the pore-forming toxin listeriolysin O (LLO) and type C phospholipases (Gaillard et al., 1987; Camilli et al., 1993; Smith et al., 1995; Dramsi and Cossart, 2002; Schnupf and Portnoy, 2007). L. ivanovii (Karunasagar et al., 1993), Actinobacillus actinomycetemcomitans (Meyer et al., 1996), Shigella flexneri (Gaillard et al., 1986), Burkholderia pseudomallei (Cullinane et al., 2008; Gong et al., 2011), Francisella tularensis (Clemens et al., 2004; Clemens and Horwitz, 2007) and Rickettsia spp. (e.g. Silverman and Bond, 1984; Whitworth et al., 2005) also possess the ability to escape phagosomes by membrane disruption (reviewed in Haas, 2007; Hybiske and Stephens, 2008; Ray et al., 2009).
We and others recently demonstrated that S. aureus is capable of translocating to the host cell cytoplasm (Bayles et al., 1998; Kahl et al., 2000; Qazi et al., 2001; Shompole et al., 2003; Jarry et al., 2008; Kubica et al., 2008; Lâm et al., 2010). Whereas the agr-controlled pore-former α-toxin has been shown to contribute to phagosomal escape in cystic fibrosis cell lines (Jarry and Cheung, 2006; Jarry et al., 2008), α-toxin is not involved in escape in non-CF cells (Jarry et al., 2008; Giese et al., 2009; Lâm et al., 2010). Recently, another group of staphylococcal virulence factors has been proposed to account for increased virulence of CA-MRSA strains: phenol-soluble modulins (PSMs) are small amphipathic peptides, which have been first described in S. epidermidis (Mehlin et al., 1999). CA-MRSA strains such as USA300 LAC and USA400 MW2 show enhanced expression of these peptides (Wang et al., 2007).
Despite their common amphipathic helical structure different classes of PSMs are defined by different peptide lengths and sequences. PSMs of the α-type (e.g. PSMα peptides and δ-toxin) are about 18–26 amino acid residues (aa) long, whereas PSMs of the β-type comprise about 45 aa. PSMγ is identical to δ-toxin, an amphiphilic peptide encoded by the accessory gene regulator (agr) quorum sensing effector, RNAIII (Janzon et al., 1989). PSMs are secreted by a specific ATP-binding cassette transporter, the recently identified PSM transporter Pmt (Chatterjee et al., 2013).
All PSMs possess pro-inflammatory activity, activate and induce cytokine release from neutrophils (Kretschmer et al., 2010) and act as chemotactic signals for phagocytes (Wang et al., 2007; Surewaard et al., 2012). PSMs further interact with membranes and possess cytolytic activity (Haas, 2007). The lysis of neutrophils was independent of receptor binding and may be due to the amphipathic nature of the peptides (Haas, 2007). However, it has been shown that PSMs are readily inactivated by serum lipoproteins (Surewaard et al., 2012) and thus rather intracellular targets for PSMs are proposed.
PMN are killed by intracellular S. aureus in a PSMα-dependent manner (Surewaard et al., 2013). Similarly, PSMs are required for survival of intracellular S. aureus as has been shown by using a mutant in the Rel A/SpoT homologue (RSH): PSM expression is boosted by the stringent response in S. aureus and a knock-out in the synthase domain of RSH (rshSyn) specifically abolished PSMα and PSMβ expression (Geiger et al., 2012). Accordingly, the rshSyn mutant was not able to survive neutrophil phagocytosis, whereas complementation of PSMα or PSMβ in trans rescued bacterial survival (Geiger et al., 2012). In our previous work we observed that the large majority of clinical isolates and laboratory strains do at best show a weak or no phagosomal escape phenotype. However, the few strains with a strong phenotype for phagosomal escape are highly virulent (Lâm et al., 2010). We hence sought to identify the contribution of the membrane active amphiphilic peptides to phagosomal escape of MRSA strains LAC (spa type t008) and MW2 (t128) as well as the highly cytotoxic MSSA strain 6850 from a different genetic background (t185).
In this work we demonstrate that S. aureus LAC, MW2 and 6850 escape from the phagosomes of non-professional as well as professional phagocytes in a PSMα-dependent process. By contrast, PSMγ (δ-toxin) and PSMβ as well as β-toxin, and phosphatidyl inositol-dependent phospholipase C are not involved in escape. We further demonstrate that S. aureus replicates in the host cell cytoplasm after PSMα-mediated phagosomal escape.
Virulent S. aureus strains escape from phagosomes
By means of monitoring the pH microenvironment of endocytosed S. aureus (Giese et al., 2009; Lâm et al., 2010) and by microscopic evaluation of escape marker recruitment (Giese et al., 2011) we recently showed that only a minority of S. aureus strains is capable of phagosomal escape in non-professional phagocytes and that the pore-forming α-toxin was not involved in this process (Giese et al., 2009; Lâm et al., 2010). We hence were interested to identify factors involved in phagosomal escape in clinically relevant strains.
In this study we demonstrate that the community-associated MRSA USA300 LAC (CDC, 2003), USA400 MW2 (Baba et al., 2002), as well as in the strongly cytotoxic strain 6850 (Balwit et al., 1994; Proctor et al., 2002) translocate to the cytoplasm of their respective host cells, whereas a non-cytotoxic S. aureus strain, Cowan I (ATCC 12598), remained in an acidic compartment (Fig. 1A). The escape of strains 6850, LAC and MW2 is indicated by a significant changes of ΔAFU between the 2 h and 6 h time points (6850: P = 0.002; LAC: P = 0.003; MW2: P = 0.012). At 6 h p.i. ΔAFU were significantly different in the escaping strains when compared with the avirulent Cowan I (P < 0.0001). Phagosomal escape was corroborated for strains LAC (Fig. 1B, arrowhead) and 6850 (not shown) by the recruitment of YFP-Fc (Giese et al., 2011) to the cell wall of escaped staphylococci as well as the absence of a LAMP1-YFP-positive membrane around the bacteria 6 h post infection (Fig. S1).
Blocking bacterial protein biosynthesis indirectly by treating infected cells with 10 ng ml−1 rifampicin (determined MIC 16 ng ml−1) at different time points during infection (1–6 h post infection in 1 h intervals) revealed an early critical time window for de novo synthesis of required proteinaceous virulence factors for phagosomal escape in S. aureus 6850 (Fig. 1C) or LAC (Fig. 1D) to escape from phagosomes (for a schematic representation refer to Fig. 1E). Whereas rifampicin treatment at 1 or 2 h post infection blocked phagosomal escape of both strains efficiently, treatment at later time points indicated no influence of the antibiotic on bacterial translocation to the cytoplasm illustrating that a proteinaceous factor was involved in phagosomal escape and that escape commenced approximately 2.5 h after phagocytosis in both, strain 6850 and LAC (Fig. 1C and D).
α-toxin, δ-toxin, β-toxin and PIPLC do not contribute to phagosomal escape
Previously we showed that S. aureus producing large amounts of α-toxin were unable to escape the phagosome in non-professional phagocytes (Giese et al., 2009; Lâm et al., 2010), demonstrating that α-toxin had no influence on phagosomal escape of the bacteria. We corroborated this finding in the present study by infecting 293 cells with an α-toxin (hla) knock-out mutant of S. aureus LAC (Bose et al., 2013). Flow cytometric escape analysis showed that the hla mutant was able to escape just like the corresponding wild type strain (Fig. 2A). We further recently demonstrated that overexpression of both, δ-toxin and β-toxin, but not either protein alone, was able to mediate phagosomal escape of non-cytotoxic laboratory strains (Giese et al., 2011). We hence investigated the contribution of either toxin to escape of S. aureus 6850, LAC and MW2 by comparing the escape proficiencies of targeted deletion mutants with those of the wild-type strains. We found that point mutants in the initiation codon of δ-toxin (Δhld) were still capable of escaping from their phagosomal enclosures just as the negative control Δpvl, a knockout of the genes lukS and lukF, which encode for Panton Valentine leukocidin (PVL; Voyich et al., 2006; Fig. 2A and B).
The bacterial sphingomyelinase β-toxin possesses a phospholipase C activity and we demonstrated its synergistic activity in phagosomal escape when the enzyme was overexpressed together with δ-toxin (Giese et al., 2011). Most clinical strains as well as the here investigated MRSA strains LAC and MW2 do not express a functional β-toxin due to the insertion of a lysogenic prophage in the structural gene for β-toxin (hlb−). We nevertheless checked the involvement of β-toxin for phagosomal escape of strain 6850, a strong producer of the phospholipase. However, wild-type (6850) and mutant (6850 CG33) escaped with similar efficiencies indicating that β-toxin was not involved in escape (Fig. 2C).
S. aureus 6850 further displays a strong phosphatidyl inositol-dependent phospholipase C (PIPLC; plc) activity (data not shown). Since homologues of PIPLC are involved in phagosomal escape of L. monocytogenes (Goldfine et al., 1995), we generated a plc knockout, 6850 Δplc, and measured phagosomal escape. Both, wild-type as well as knock-out strains, recruited the escape marker with similar efficiencies (Fig. 2D). We confirmed the inability of PIPLC to mediate phagosomal escape by overexpressing PIPLC in recombinant S. aureus 113 (Iordanescu and Surdeanu, 1976; Fig. S2). The transgenic strains displayed very low escape rates when only Plc (0.08 ± 0.12%) or δ-toxin and the phospholipase (HLD-PIPLC; 0.32 ± 0.14%) were overexpressed. By contrast, the positive control pHld-Hlb-Cerulean (Giese et al., 2011) expressing δ-toxin and β-toxin escaped at high rates (29.58 ± 2.3%; Fig. S2). Invasion assays by flow cytometric measurements showed no significant differences between the recombinant strains (not shown).
Thus, our results demonstrate that δ-toxin, β-toxin as well as Plc were not required for phagosomal escape of clinically relevant S. aureus.
PSMα but not PSMβ is required for phagosomal escape of S. aureus in professional and non-professional phagocytes
S. aureus LAC produces a large amount of phenol-soluble modulins with PSMα being the most virulent gene cluster in a mouse model of infection (Wang et al., 2007). We hence investigated the contribution of the amphipathic peptides for phagosomal escape in epithelial cells as well as professional phagocytes. PSMα mutants of MRSA strains LAC and MW2, as well as 6850 were unable to reduce the acidification signal in 293T (Fig. 3A and B), HeLa and EA.hy926 (Fig. S3A–F) regardless of the genetic background. Complementation of PSMα restored the wild type phenotype in 6850 and partially in LAC (Fig. 3A) and MW2 (not shown). Accordingly, LAC Δpsmα showed a significantly reduced escape rate in YFP-Fc recruitment assays within HeLa (2.35 ± 0.92%) when compared with the wild type strain (20.85 ± 0.63%; Fig. S3G). By contrast PSMβ-deficient or δ-toxin-deficient mutants escape with efficiencies similar to the wild type strains (Fig. 3C and D and Fig. S3), whereas the invasiveness of the strains did not differ (Fig. S6).
To investigate the relevance of phagosomal escape in professional phagocytes, we cloned a recruitment marker recognizing Gram-positive peptidoglycan, YFP-CWT (Gründling and Schneewind, 2006), into a pseudotyped lentiviral vector (Wiznerowicz and Trono, 2003), and transduced the marker in the monocytic cell line THP-1 (Tsuchiya et al., 1980), thereby leading to stable expression of YFP-CWT in these cells that allowed to assay YFP-CWT marker recruitment (Fig. 4A). After infection of phorbol-12-myristate-13-acetate-differentiated THP-1 with S. aureus 6850 (Fig. 4B) or LAC (Fig. 4C) at an moi (multiplicity of infection) of 10 recruitment of YFP-CWT was observed by confocal microscopy. 25.6 ± 0.3% of wild-type S. aureus 6850 escaped from their membrane enclosures, whereas the Δpsmα mutant was translocating to the cytoplasm in 5.5 ± 0% of the detected bacteria. Introduction of the plasmid pTXpsmα1–4 complemented escape (12.2 ± 2.8%; Fig. 4A). S. aureus 6850 is therefore capable of escaping macrophage phagosomes. Similarly, strain LAC wild-type (5.4 ± 1.6%) and complemented mutant (3.9 ± 1.6%) escaped from THP-1 phagosomes although with reduced efficiency when compared with strain 6850, whereas for the psmα mutant an escape signal was barely detected (0.8 ± 0.3%; Fig. 4C). The ability of strain 6850 to escape from THP-1 phagosomes further was corroborated in pH escape assays (Fig. 4D).
S. aureus 6850 produces PSMα and PSMβ transcripts at high levels relatively early (2 h) during planktonic growth (Fig. S4), whereas PSM production in strain LAC is delayed with respect to expression levels in 6850. Also of note, strain 6850 produced higher levels of PSMβ transcripts than LAC in vitro (Fig. S4).
Next we demonstrated that PSM transcription is active in the intracellular environment. In wild-type S. aureus LAC (wt) intracellular psmα expression levels (RQ) increased from inoculum (0.57 ± 0.20) to 2 h p.i. (3.01 ± 1.13) and were found to be at low levels at 6 h p.i. (0.29 ± 0.09). A similar dynamics was found for LAC Δpsmβ (inoculum: 0.69 ± 0.19; 2 h: 3.65 ± 0.18; 6 h: 0.53 ± 0.13). Both strains showed only moderate activity of agrA at 2 h (wt 1.12 ± 0.20; Δpsmβ 1.75 ± 0.40) and 6 h p.i. (wt 0.49 ± 0.09; Δpsmβ 0.6 ± 0.14) and in the wild-type strain psmβ levels significantly decreased during the time-course (2 h: 1.02 ± 0.55; 6 h: 0.08 ± 0.05; Fig. S5A; P = 0.017).
By contrast in the psmα mutant, psmβ as well as agrA RQ levels peaked at 2 h p.i. (psmβ: 2.8 ± 0.7; agrA: 4.52 ± 2.1) and remained high at 6 h p.i. (psmβ: 1.31 ± 0.83; agrA: 2.3 ± 1.3) when compared with either wild-type or psmβ mutant. This illustrated that in LAC Δpsmα, which was not able to escape from phagosomes (Fig. 3), agr as well as psmβ was still transcriptionally active. A similar tendency was observed when we assayed the temporal expression dynamics of strain 6850 (Fig. S4B). Again intracellular psmα expression levels (RQ) increased in the wild-type from inoculum (0.16 ± 0.20) to 2 h p.i. (0.78 ± 0.36) and dropped to very low levels at 6 h p.i. (0.14 ± 0.07). psmβ expression followed that profile at slightly elevated RQ (inoculum: 0.34 ± 0.19, 2 h: 2.1 ± 0.83; 6 h: 0.27 ± 0.11) with a significant drop in psmβ expression levels from 2 to 6 h p.i. (P = 0.025). By contrast, in the escape-deficient psmα mutant psmβ expression already started at an RQ of 2.04 ± 0.72, peaked with 9.2 ± 2.56 at 2 h p.i and remained at elevated levels at 6 h p.i. (4.96 ± 2.35) illustrating that conditions persisted, which led to activation of PSM transcription persisted (Fig. S5B). Our data thus support the requirement of PSMα for phagosomal escape of the clinically relevant strains LAC and 6850 within professional as well as non-professional phagocytes.
Intracellular replication of S. aureus LAC and 6850 in the host cell cytoplasm
Next we infected 293T with wild type or psmα mutants in strains LAC and 6850 and tested intracellular survival of S. aureus by lysostaphin protection assays (Fig. 5). Cowan I, 6850 Δpsmα as well as LAC Δpsmα, which are unable to escape from phagosomes, did not display significant changes of recovered colony-forming units (cfu) between 2 and 6 h p.i (6850 Δpsmα 2 h: 16 ± 4.94; 6 h: 25 ± 5.58; LAC Δpsmα 2 h: 8 ± 0.87; 6 h: 14 ± 3.27), illustrating that bacteria neither are efficiently killed during this time nor replicate. By contrast, the escape-proficient strains 6850 (2 h: 20 ± 7.8; 6 h: 59 ± 12.84), LAC (2 h: 9 ± 2.37; 6 h: 45 ± 10.18) and LAC Δpsmβ (2 h: 5 ± 0.73; 6 h: 34 ± 5.12) were recovered with 3- to 5-fold increased cfu at 6 h p.i. when compared with the 2 h time point (Fig. 5).
This indicates that intracellular S. aureus replicates within the host cell cytoplasm upon phagosomal escape in a relatively short time span.
A variety of pathogenic bacteria are able to survive inside the host cell using different virulence strategies, such as modulation of phagolysosomal maturation or disruption of the phagosomal membrane. For instance, L. monocytogenes escapes the phagosome by phospholipase C-dependent pore formation with listeriolysin O (Gaillard et al., 1987; Camilli et al., 1993; Goldfine et al., 1995; Smith et al., 1995). Similarly Rickettsia spp. disrupts the membrane using hemolysin C and phospholipases (Silverman and Bond, 1984; Whitworth et al., 2005; Ray et al., 2009). S. aureus, which has been shown to be readily internalized by mammalian cells, also translocates to the cytoplasm of its host cells (Bayles et al., 1998; Kahl et al., 2000; Jarry and Cheung, 2006; Lâm et al., 2010; Rasigade et al., 2013), however, only few strains are capable to phagosomal escape (Lâm et al., 2010). Only little is known about the involved virulence factors: whereas the pore-forming staphylococcal α-toxin has previously been demonstrated to lyse phagosomes of cystic fibrosis cells (Jarry and Cheung, 2006; Jarry et al., 2008), it is not sufficient for phagosomal escape within non-CF cells (Jarry et al., 2008; Giese et al., 2009; Lâm et al., 2010). We corroborated these observations by infecting 293 cells with S. aureus LAC Δhla. (Fig. 2A). As expected LAC Δhla does not escape from host cell phagosomes supporting our previous findings (Giese et al., 2009; Lâm et al., 2010).
In this study we show that the MRSA strains LAC and MW2 as well as the highly cytotoxic MSSA 6850 are able to escape from phagosomes of epithelial and endothelial cells (Fig. 1) and macrophages (Fig. 4) thus possibly contributing to the observed virulence of the strains. We determined the timing of translocation to the host cell cytosol by adding rifampicin to infected cells at different time points after infection thus blocking bacterial RNA-polymerase (Calvori et al., 1965) and hence synthesis of virulence factors. We found that phagosomal escape of such diverse strains as LAC and 6850 commenced at similar time points: approximately 2.5 h after infection (Fig. 1C and D). This is comparatively late during endosome maturation, since vacuole perforation of Listeria-infected macrophages has been observed already as early as 17 min after phagocytosis (Beauregard et al., 1997; Henry et al., 2006). Similarly, escape of Rickettsiae is a fast process with 50% of internalized bacteria located freely in the host cytosol about 12 min after internalization into Vero cells (Teysseire et al., 1995; reviewed in Ray et al., 2009).
As indicated by our results (Fig. 1C and D) the phagosomal escape of S. aureus must be governed by proteinaceous factors. We thus set out to identify virulence factors involved in escape of the staphylococcal strains. Since we previously showed that δ-toxin can act in a synergistic manner with β-toxin when recombinantly overexpressed in avirulent laboratory strains of S. aureus (Giese et al., 2011), we initially focused on these genes. However, when we analysed the escape proficiency of LAC, MW2, 6850 and their respective toxin mutants, neither δ-toxin nor β-toxin proved to be necessary for phagosomal escape (Fig. 2A–C). Also, a S. aureus 6850 knockout of plc, of which a L. monocytogenes homologue had been shown to be required for phagosomal escape (Goldfine et al., 1995), did not reveal an escape phenotype (Fig. 2D).
Recent studies have claimed a prominent role for phenol-soluble modulins in staphylococcal virulence because of their ability to activate and lyse neutrophils (Wang et al., 2007; Geiger et al., 2012; Surewaard et al., 2013) or osteoblasts (Rasigade et al., 2013). For instance, S. aureus LAC Δpsmα caused less skin lesions in mice. Further, a bacteremia model revealed that mice infected with S. aureus MW2 Δpsmα survived at significantly higher rates when compared with mice infected with wild-type MW2 (Wang et al., 2007). The virulence of the wild-type strains is attributed to the ability of S. aureus to avoid the innate immune system since PSMα peptides produced by intracellular S. aureus caused lysis of human neutrophils (Surewaard et al., 2013). In addition, there is a clear correlation between psm expression and bacterial survival in neutrophils (Geiger et al., 2012). The α-helical and amphipathic PSMs possess a membrane-damaging activity (Wang et al., 2007), which is similar to that of non-ionic detergents as has been shown for δ-toxin (Kreger and Bernheimer, 1971; Kreger et al., 1971; Rahal, 1972; Kapral, 1976; Verdon et al., 2009). In this study we demonstrate that S. aureus phagosomal escape proceeds in a PSMα-dependent fashion in non-professional phagocytes (HeLa, Ea.Hy926, 293; Fig. 3 and Fig. S3) as well as THP-1 macrophages (Fig. 4). By contrast, mutants in PSMβ as well as PSMγ (δ-toxin) had no effect on escape efficiencies (Fig. 3).
When we analysed transcriptional activity of psmα, psmβ and agrA of internalized staphylococci (Fig. S5), we found that expression of the genes was strongest at 2 h p.i. (before phagosomal escape) for all tested strains and decreased after translocation to the cytosol at 6 h p.i. in escape-proficient strains (Fig. S5). Thus the observed timing of escape (Fig. 1) is caused by PSM expression. The similarity of intracellular expression dynamics of both strains, LAC and 6850, was striking since the strains showed significantly different propensities to produce psmα transcripts during planktonic growth (Fig. S4). Further, although agrA transcript was detected at higher levels before escape, the levels did not reach that of the psm genes. Whereas PSMs are agr-controlled genes (Queck et al., 2008), a recent study showed that PSM expression is requiring basal levels of agr but is boosted by the stringent response (Geiger et al., 2012). Thus we hypothesize that the observed boost in psm expression and the subsequent phagosomal escape was caused by the stringent response, possibly caused by nutritional limitation or other environmental stressors such as an acidic environment, reactive oxygen species, or the presence of other antimicrobial agents within the phagosome (reviewed in Flannagan et al., 2009).
Our results clearly demonstrate that phagosomal escape of clinically relevant S. aureus strains such as the epidemic MRSA S. aureus strains LAC (Diep et al., 2006), MW2 (Baba et al., 2002) or the highly cytotoxic MSSA strain 6850 (Balwit et al., 1994) is mediated by a common PSMα-dependent mechanism despite of the distant genetic backgrounds (e.g. S. aureus LAC and 6850).
We further show that only escape-proficient strains (6850, LAC and LAC Δpsmβ) were able to replicate inside 293T cells, whereas non-escaping strains without PSMα (Cowan I, 6850 Δpsmα or LAC Δpsmα) were not (Fig. 5A and B). Although there is no replication of the latter bacterial strains, the staphylococci are able to survive within the 293T phagosome for several hours. This was not surprising as it has been demonstrated before, that S. aureus withstands conditions of phagoendosomes of non-professional phagocytes (Schröder et al., 2006; Giese et al., 2009). Further it is known that S. aureus is able to form persister cells, so-called small colony variants (SCV), which are metabolically quiescent and do not produce toxins. Such SCV have been described to persist in cells for extended periods (Hamill et al., 1986; Lowy et al., 1988; Vann and Proctor, 1988; Buisman et al., 1991; Hiemstra et al., 1992; Schröder et al., 2006; Garzoni et al., 2007; Kubica et al., 2008; Sendi and Proctor, 2009; Tuchscherr et al., 2011).
However, it was striking that in our analyses escape-proficient S. aureus strains were recovered at 3- to 5-fold increased cfu when comparing the 2 h and 6 h time points. This indicates that S. aureus strains LAC and 6850 replicate in the cytoplasm of their host cells only after phagosomal escape (Fig. 5A and B). Replication of S. aureus in the cytoplasm of host cells has been described before (Kahl et al., 2000; Jarry et al., 2008). Evidence most often came from the absence of enclosing membranes in transmission electron micrographs. However, the failure to visualize a membrane does not necessarily indicate the functional absence of a membrane (Lâm et al., 2010). With our assays we unequivocally show that phagosomal escape is a prerequisite for replication of S. aureus LAC and 6850. It has to be noted, however, that for different strain backgrounds alternative strategies can exists. For instance, strain Newman has been shown to persist in intracellular vacuoles for multiple days before lysing the host cells (Kubica et al., 2008).
S. aureus also has been shown to modify the autophagic pathway after host cell invasion and thus promotes pathogen survival (Schnaith et al., 2007; Mestre et al., 2010). Hence future experimentation will have to determine the timing and the interdependencies of autophagy and phagosomal escape in intracellular staphylococcal infections.
Multiple studies indicate that S. aureus induces host cell death in a variety of host cells from an intracellular location (Bayles et al., 1998; Menzies and Kourteva, 1998; Wesson et al., 1998; Kahl et al., 2000; Nuzzo et al., 2000; Tucker et al., 2000; Krut et al., 2003; Haslinger-Löffler et al., 2005; Jarry et al., 2008; Kubica et al., 2008; Lâm et al., 2010), and it has been reported that host cell death is dependent on phagosomal escape by S. aureus (Menzies and Kourteva, 1998). Human endothelial cells are virtually insensitive to the action of S. aureus α-toxin, however, comparatively low numbers of S. aureus cells with a combined invasive and strongly hemolytic phenotype readily induce apoptotic cell death in HUVEC (Haslinger-Löffler et al., 2005). This suggests that cell death mechanisms are activated from within their intracellular location. Surewaard and colleagues observed that an intracellular PSMα mutant of strain MW2 did not kill human PMN (Surewaard et al., 2013). Further, a S. aureus LAC lukAB mutant LAC was unable to efficiently kill the immune cells (Dumont et al., 2013). Since our data provide a clear link between PSMα and phagosomal escape, we hypothesize that the inability of phagocytosed S. aureus LAC Δpsmα to kill host cells from within (Surewaard et al., 2013) is caused by the lack of phagosomal escape. Our data further suggest that escaped bacteria replicate within the host cell cytosol and switch off expression of PSM after escape. Hence we hypothesize that cell death may be triggered by virulence factors other than PSMs and that phagosomal escape and host cell death are independent processes in S. aureus infected cells. Our previous observations using a gain-of-function approach demonstrated that overexpression of PSMα alone did not lead to translocation of laboratory S. aureus strains (Giese et al., 2011). Thus, we cannot exclude that PSMα requires a synergistic toxin to mediate phagosomal escape. We are currently conducting experiments to identify potential candidate factors.
In summary, we show that PSMα is required for phagosomal escape of S. aureus strains from genetically divergent backgrounds such as CA-MRSA LAC (USA300), MW2 (USA400) and MSSA 6850 and that phagosomal escape progresses at similar time scales in the strains. Phagosomal escape takes place in a PSMα-dependent fashion in both, professional as well as non-professional phagocytes and we demonstrate that S. aureus is able to replicate in the cytoplasm of its host cells. We therefore hypothesize that PSMα constitutes a general principle of S. aureus phagosomal escape and virulence.
Bacterial and host cell culture
S. aureus strains were grown in trypticase soy broth (TSB) or Mueller–Hinton (MH) unless indicated otherwise. Selective antibiotics were added where appropriate for overnight cultivation of genetically engineered strains but were omitted for cultures directly used in infections. For phenotypic control of hemolysis S. aureus strains were grown on sheep blood agar at 37°C overnight and hemolytic activity was inspected visually. For a list of strains used in this study please refer to Table S1.
All cell lines were grown Mycoplasma-free in T75 tissue culture flasks (Sarstedt, Nümbrecht, Germany), in media supplemented with 10% fetal calf serum and penicillin/streptomycin (50 U ml−1 and 50 μg ml−1 respectively). 293 cells (human embryonic kidney; DSMZ Cat. No. ACC-305) were grown in DMEM/F12 (1:1) GlutaMAXTM-I. HeLa (DSMZ Cat. No. ACC-57) and the human endothelial cell line EA.hy926 (Edgell et al., 1983) were grown in RPMI 1640 supplemented with 1× GlutaMAXTM-I. The monocytic cell line THP-1 (Tsuchiya et al., 1980) was grown in RPMI1640 supplemented with 10% FCS. Monocytes were differentiated by addition of 0.025 μg ml−1 PMA for 24 h prior to infection. All media and supplements were purchased from Gibco® Life Technologies (Darmstadt, Germany) or PAA (Cölbe, Germany).
Generation of constructs
For generation of stably integrating YFP-CWT and YFP-Fc escape markers (Giese et al., 2011) in growing and non-growing cell lines we used the lentiviral vector pLVTHM (Wiznerowicz and Trono, 2003). The GFP-marker of pLVTHM was removed by restriction with PmeI and SpeI and an accordingly processed DNA-Fragment encoding YFP-Fc (Giese et al., 2011) or YFP-CWT was introduced by ligation and subsequent transformation in E. coli DH5α. Twenty micrograms of a plasmid preparation of the resulting vector was used in calcium phosphate-based co-transformation of a 15 cm dish of 293T cells along with 10 μg psPAX and 10 μg pVSVG. DMEM growth medium was exchanged after 4–8 h. Two days after transfection, the supernatant was harvested and sterile-filtered (0.45 μm filter). Target cells, such as THP-1, were infected in presence of 10 μg ml−1 polybrene and were sorted on a FACSAria III cell sorter (BD).
The plc as well as the psmα operon deletion mutant of strain 6850 was constructed using an allelic replacement procedure (Bae and Schneewind, 2006). For generation of 6850 Δpsmα the plasmid pKOR1psma was used (Joo et al., 2011). For the generation of S. aureus 6850 Δplc we used pBASE6 (B. Krismer, Tübingen, Germany) which is derived from pKOR1 (Bae and Schneewind, 2006). A gene replacement cassette with up-, downstream and tetracycline resistance parts was constructed by amplifying the upstream and downstream regions of plc with PLC-up-f and PLC-up-r or PLC-down-f and PLC-down-r respectively. TetBD was amplified with tetBD-f and tetBD-r and restricted with XhoI and BglII. The purified PCR fragments were ligated in BamHI-opened pBASE6 and transformed into E. coli DH5α. The resulting vector, pBASE6-Δplc, was transformed in S. aureus RN4220 (Kreiswirth et al., 1983) and eventually in the target strain 6850. Selection of mutants was performed as described (Bae and Schneewind, 2006) and mutants were screened for successful gene replacement by PCR with plc-up656 and plc-down740. Mutants were further checked for the loss of opaque halos on diacylglycerol-containing TSB plates.
Generation of pHld-Hlb-Cerulean, pCerulean, or pmRFPmars is described elsewhere (Paprotka et al., 2010; Giese et al., 2011). pHld-Plc-Cerulean, pPlc-Cerulean were generated by amplifying plc from S. aureus gDNA using oligonucleotides plc-Av-f and plc-r. The PCR product was cloned into pCR2.1-TOPO (Invitrogen) and sequenced. An AvrII fragment containing plc was ligated in AvrII-opened pHld-Cerulean (Giese et al., 2011). All oligonucleotide sequences are listed in Table S3.
Host cells were prepared one day prior to infection. The cells were counted and 0.7 × 105 cells per well were seeded in a 12-well plate. For subsequent microscopic analysis cells were grown on 18 mm coverslips. One hour before infection cell culture medium was replaced by antibiotic-free medium (infection medium). S. aureus was grown overnight in TSB broth. The cultures were diluted to an OD540 of 0.4 and grown at 37°C in TSB containing the appropriate antibiotics. At an OD540 of 0.6 bacteria were harvested by centrifugation [3000 g; 2 min; room temperature (RT)], resuspended in infection medium and used for infection of the host cells. Infection of host cells was synchronized by centrifugation (500 g; 10 min; RT) and cells were subsequently incubated for additional 50 min (37°C, 5% CO2). The supernatant was aspirated and replaced with infection medium supplemented with 20 μg ml−1 lysostaphin (AMBI, Lawrence, NY, USA) and 100 μg ml−1 gentamicin and the assays were incubated for an additional 30 min. Cells were rinsed twice with phosphate-buffered saline (PBS, pH 7.4) and incubated in gentamicin-containing medium to indicated time points for further analysis.
Lysostaphin protection assay (cfu assays)
Cells and bacteria were grown and co-cultivated as described above. After 2 and 6 h, respectively, infected cells were washed 2× with PBS and trypsinized. Reaction was stopped with DMEM/F12 with 10% FCS, the infected cells were transferred into 1.5 ml tubes and quickly spun down. Pellets were resuspended in sterile water, mixed and incubated at 37°C for 10 min for cell disruption. The cell lysates were serially diluted in PBS supplemented with 1% human serum albumin (HSA), plated on MH agar and incubated at 37°C for one day.
Sample preparation for microscopic escape assays was performed largely as published previously (Giese et al., 2011). Briefly, for sample fixation the cells were rinsed twice with PBS supplemented with 1 mM MgCl2 and 0.1 mM CaCl2 (PBS2+) and subsequently fixed at RT for 30 min using 4% paraformaldehyde (PFA) in PBS2+. The cells were rinsed once and quenched with 50 mM NH4Cl2 in PBS2+. After washing three times with PBS2+ the actin cytoskeleton and DNA were counterstained with fluorophore-conjugated phalloidin (Invitrogen, Karlsruhe, Germany) and 5 μg ml−1 Hoechst 34580 or 4′,6-diamidino-2-phenylindole (DAPI; Sigma, Taufkirchen, Germany). The samples were incubated for 30 min in staining solution, were rinsed three times with PBS and mounted on glass slides with Mowiol or ProLong Gold reagent (Invitrogen, Karlsruhe, Germany).
For semi-automatic escape quantification image series were obtained from three biological replicates and two technical duplicates each. Image acquisition was performed with either a 40×/1.3 objective on a LSM510 Meta confocal laser scanning microscope (Carl Zeiss, Jena, Germany) or 63×/1.4 objective on a Leica TCS SP5 confocal laser scanning microscope (Leica, Wetzlar, Germany). The escape signal is generated by the recruitment of cytoplasmic YFP-Fc to protein A or YFP-CWT to S. aureus peptidoglycan (Gründling and Schneewind, 2006) upon disruption of the phagosomal membrane barrier (Giese et al., 2011). The colocalization of fluorescent bacteria with yellow fluorescent escape marker indicated translocation of the pathogen to the host cell cytosol (Giese et al., 2011) and was analysed with ImageJ (Abramoff et al., 2004). Localization to LAMP1-YFP-positive vesicles and immunofluorescence analyses were performed as described elsewhere (Giese et al., 2009).
Flow cytometric escape analysis
Bacterial overnight cultures were performed in MH broth at 37°C without agitation using appropriate antibiotics. Flow cytometric escape analysis was based on pH monitoring of the environment of intracellular S. aureus essentially as described previously (Lâm et al., 2010). Shortly, host cells were seeded (3 × 105 cells per well) and grown over night in a 24-well culture dish to sub-confluency. Staphylococci were labelled for 30 min with fluorescein isothiocyanate (FITC) dissolved in dimethylsulphoxide (DMSO) (Sinha et al., 1999). FITC-labelled bacteria were resuspended in PBS supplemented with 1% HSA. Bacterial suspensions were adjusted to an OD540 of 1. Prior to infection, cells were washed with invasion medium (10 mM Hepes, pH 7 in DMEM/F12 GlutaMaxTM-I supplemented with 1% HSA). The infection was performed in 500 μl invasion medium with 50 μl of the FITC-labelled bacterial suspension (control cells were treated with 50 μl HSA/PBS). Next, the staphylococcal suspensions were spun down (7 min at 800 rpm) to synchronize the infection. After an incubation for 55 min (37°C and 5% CO2), the supernatant was aspirated and remaining extracellular bacteria were removed by incubating the cells with medium supplemented with lysostaphin (20 μg ml−1; AMBI, Lawrence, NY, USA) at 37°C for 10 min. Cells were further incubated in complete medium for additional 55 min and 4 h 55 min, which resulted in overall infection times of 2 and 6 h respectively. Cells were washed with PBS, trypsinized and transferred into 5 ml round-bottom polystyrene tubes (Falcon; BD, Heidelberg, Germany), pelleted, and resuspended in PBS with 1% HSA. Each sample was aliquoted and either incubated with 50 μM monensin or 1 μl 100% ethanol for 15 min at 37°C. Then each sample was treated with 5 mg ml−1 propidium iodide (PI). Samples were analysed with a FACScalibur or an Accuri C6 (Becton Dickinson, Heidelberg, Germany). For each cell type, a pre-set forward scatter and side scatter (FSC-SSC) gating as well as amplification were used. The FITC fluorescence was analysed in Fl-1 while cell debris was excluded for the analysis by PI measurement in Fl-3. As FITC fluorescence is quenched by low pH, quenching can be used as a marker for acidification of endosomes (Sinha et al., 1999). Thus, we measured the difference in Fl-1 fluorescence signals in samples with and without monensin treatment. The total internalized bacteria, expressed as numbers of arbitrary fluorescence units (AFU), was determined according to the formula AFU = (mean Fl-1 intensity of infected cells) × (% of infected cells) × 10−3 and normalized to the mean fluorescence intensity of each corresponding bacterial preparation. The difference in AFU (ΔAFU) in the presence or absence of monensin, normalized to the invasion rate was used as readout for pH according to the formula ΔAFU = (AFU−monensin − AFU+monensin)/AFU+monensin × 105. Higher (i.e. less negative) ΔAFU values represent less acidic pH values and thus differences of ΔAFU at 2 and 6 h post infection indicate the translocation of bacteria from an endosomal to a cytoplasmic environment (Lâm et al., 2010).
Flow cytometric invasion assays
To compare the invasion behaviour of different bacterial strains, the bacteria were labelled with carboxyfluorescein diacetate (CFDA, Invitrogen) according to the manufacturer's instructions. Uninfected cells served as negative control. Cells were infected with CFDA-labelled bacteria essentially as outlined above. Cells were incubated for 1 h at 37°C, the supernatant was aspirated and 300 μl medium containing 10 μg ml−1 lysostaphin and 100 μg ml−1 gentamicin was added in order to remove extracellular staphylococci. After half an hour, wells were washed twice with PBS, the cells were trypsinized and analysed on either an Accuri C6 flow cytometer or a FACSAria III (BD). Intact cells were gated with FSC and SSC and fluorescence was analysed in the Fl-1 channel. Histograms of fluorescence of FSC/SSC-gated cells (10 000 events) were obtained. Gates for infected cells were set to include less than 1% of the cells of the negative control. Invasion efficiency was measured by determining the proportion of infected cells per sample.
Real-time PCR analysis of expression
For RNA isolation of intracellular S. aureus, host cells were seeded in 6-well plates (1.5 × 106 cells per well) one day prior to the infection and were grown to sub-confluency. The infection was performed using 2 ml invasion medium and 200 μl bacterial suspensions adjusted to an OD of 1. Further procedures were performed as described above. After 2 and 6 h, respectively, the cells were washed twice with PBS, trypsinized, transferred to 1.5 ml tubes and collected by centrifugation (2000 rpm, 5 min). Supernatants were discarded, cells were resuspended in ice-cold killing buffer (20 mM Tris, pH 7.5; 5 mM MgCl2; 20 mM NaN3 and 150 mM NaCl) and incubated on ice. Pellets were recovered by centrifugation and cell material from two wells was pooled, resuspended in 600 μl RLT buffer and transferred into lysing matrix B tubes (MP). The homogenization was done in a FastPrep (MP) for 45 s at a setting of 6.5. The samples were immediately put on ice. Supernatants were transferred into a fresh RNase-free tube, mixed with one volume of 70% ethanol and RNA was purified by RNeasy columns (Qiagen, Hilden, Germany) following the manufacturer's instructions.
For RNA extraction from in vitro shaking cultures of S. aureus bacteria were grown in MH for 2 h and 5 h after inoculation to OD 0.05 from overnight cultures. Prior to extraction, bacteria were pelleted (5000 rpm, 5 min) and washed by resuspending within PBS. Pellets were then resuspended in 600 μl RLT buffer, transferred into lysing matrix B tubes and RNA was prepared with RNeasy following the manufacturer's instructions.
RNA concentrations were determined with a NanoDrop spectrometer (Thermo Scientific). The reverse transcription of 1000 ng RNA was performed using the Quantitec Kit (Qiagen) including the digestion of genomic DNA with 7× gDNA Wipeout Buffer and following the instructions of the manufacturer. Successful removal of gDNA was tested with PCR using primers gyrB-F and gyrB-R (Table S2) for samples treated either with or without Quantiscript Reverse Transcriptase.
Primer pairs for qRT-PCR were designed to yield products between 145 bp and 176 bp and were tested in a standard PCR with Taq DNA polymerase. SYBR® Green PCR Master Mix was used for qRT-PCR containing SYBR® Green I dye, AmpliTaq Gold® DNA Polymerase, dNTPs with dUTP, Passive Reference 1, and optimized buffer. The reactions were carried out in 96-well plates and each sample was run in triplicates. A single reaction of 20 μl contained 10 μl Master Mix (2×), 900 nM forward or reverse primer respectively, 6.2 μl H2O and 2 μl (10 ng) of cDNA. qRT-PCR was performed on a StepOne Plus cycler (Applied Biosystems, Darmstadt, Germany). The initial denaturation was performed at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 65°C for 1 min. For each primer pair and for each run a non-template control was prepared to exclude contamination.
For quantification, the relative amount of target gene in comparison to an internal control gene was determined. The housekeeping gene gyrase (gyrB) was used for the normalization step. For the calculation of the relative expression [RQ] of a gene, the cycle threshold (CT) of the target gene and housekeeping gene were subtracted using following formula: 2ΔCT.
All statistical analyses were performed using a two-tailed Student's t-test.
We thank the German Science Foundation (DFG; http://www.dfg.de) for funding this project within the Transregional Research Collaborative TRR34, project C6 (M.G., B.S.), B1 (C.W., T.G.), C11 (A.-C.W., T.R.) and within FR1504/2-1 (M.F.) and the Intramural Research Program of the National Institute of Allergy and Infectious Diseases (NIAID) , US National Institutes of Health (http://www.nih.gov; M.O.). We are grateful to Barbara Conrad and Heidi Linß for excellent technical assistance, Christiane Goerke and Bernhard Krismer (Tübingen, Germany) for kindly supplying strain 6850 pCG33 and pBASE6, respectively, and Andreas Demuth for critically reading the manuscript. The α-toxin mutant (NE1354) was obtained through the Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA) Program supported under NIAID/ NIH Contract No. HHSN272200700055C. The authors declare no conflict of interests.