Intracellular staphylococcus aureus: Live-in and let die

1Most manifestations of Staphylococcus aureus disease involve extracellular bacteria (furuncles, carbuncles, impetigo, abscesses, septicemia, necrotizing pneumonia) or biofilm formation (catheter-induced infective endocarditis, atherosclerosis). Aside from this S. aureus infections have a second face: there is accumulating evidence that S. aureus is able to survive within its hosts cells and thus might be termed a facultative intracellular pathogen. Intracellularity of S. aureus has been implied as immune-evasive strategy thereby escaping detection by professional phagocytes.

2Invasion of non-professional phagocytes by S. aureus is mediated by a zipper-type mechanism. To date many bacterial adhesins have been identified with Fibronectin (Fn)-binding proteins A and B (FnBPA, FnBPB) constituting the major staphylococcal adhesins for non-professional phagocytes such as epithelial, endothelial cells, fibroblasts, osteoblasts, and keratinocytes (;;;;;;;;;; Figure 1, Map Item 1). Fibronectin-bridging between FnBPs and α5β1 integrins on the host cell surface is sufficient to induce zipper-type uptake of staphylococci. However, FnBPs also have been shown to directly bind to human heat shock protein 60 (Hsp60) exposed on the cellular surface. FnBP-independent invasion was observed in S. aureus Newman, which produces C-terminally truncated FnBPs that are not covalently anchored to the cell wall of S. aureus. It has been shown that strain Newman gets internalized by epithelial cells and fibroblasts mediated by extracellular adherence protein (Eap; ) with its cellular receptor still not identified to date. Recently, the staphylococcal autolysin (Atl) was identified to function as adhesin/invasin with heat shock cognate protein Hsc70 being the direct cellular receptor. Further, wall teichoic acids (WTA) seem to be important for establishment of nasal colonization and there is evidence that a scavenger receptor is involved in WTA binding. Clumping factor B (ClfB) has been shown to bind to cytokeratins in the extracellular matrix (ECM) of host cells, and staphylococcal protein A can directly interact with tumor necrosis factor α receptor 1 (TNFR1; ). To what extent the internalization of the pathogen is mediated by WTA, ClfB, protein A, and a body of other molecules interacting with the ECM of host cells is not known thus far.

3Since FnBPs contribute to the adherence of S. aureus to intact endothelium in vivo, we can assume that staphylococcal invasion of epithelia or the endothelium is relevant in natural infections. The interaction of FnBP with ECM Fn is mediated by tandem β zipper structures via the binding of multiple fibronectin molecules by the repetitively arranged modules within a single FnBP. As a result FnBP/Fn sequester α5β1 integrins on the host cell surface. The resulting receptor clustering relays signals that result in cytoskeletal rearrangements. The rearrangements initiated at focal adhesions, which are remodeled to fibrillar adhesions by loss of focal adhesion kinase (FAK), paxillin, and vinculin. The rearrangement is accompanied by a centripetal movement of S. aureus on the host cell surface that were observed by videomicroscopy. The repeated generation of actin comet tails beneath adherent staphylococci or FnbA-coated beads and the formation of actin cups without internalization of staphylococci is interpreted by the authors as a delay of phagocytosis (; Figure 1, Map Item 2). Invasion signaling further involves src kinase. Extracellular signal-regulated kinases (ERK) and c-Jun N-terminal kinase (JNK) but not mitogen-activated protein kinase (MAPK) p38 are required in osteoblasts, whereas in HEp-2 cells p38 MAPK was found to be upregulated alongside ERK. Further, phosphorylation of transcription factor c-Jun, but not of Elk-1 or ATF-2 has been demonstrated during invasion of osteoblasts. The phosphoinositide-3-kinase (PI3K)-Akt pathway is active during S. aureus internalization by bovine endothelial cells.

4Eventually, S. aureus gets endocytosed by professional as well as non-professional cells where the pathogen faces a variety of intracellular fates.

5The fates of the pathogen and the infected host cell depend on staphylococcal isolate and genotype as well as differential susceptibility of host cells to virulence factors, host cell gene expression, etc. For example, S. aureus produces different hemolysins. The majority of bovine mastitis strains were phenotypically positive for the sphingomyelinase β-toxin, whereas only a minority of human strains isolated from cases of septicemia or nasal carriage was positive for β-toxin. There seems to exist a selective pressure for S. aureus strains colonizing humans to acquire β-toxin converting phage. This is most likely due to staphylococcal complement inhibitor SCIn and the chemotaxis inhibitor protein CHIPS (and additional factors) that are usually found to be encoded by the respective phage genomes. Purified β-toxin, selectively kills monocytes and destroys platelets, but barely affects other cell types. Similarly, most human cell types are fairly insensitive to the pore-former α-toxin, whereas human leukocytes and cells from other mammalian species are highly susceptible. The metalloprotease ADAM10 has been recently identified as receptor of α-toxin monomers (;; Figure 1, Map Item 3). The differential specificities of α-toxin possibly reflect ADAM10 expression differences in the respective cells or might result from differential capability of host cells to remove α-toxin pores by endocytosis.

6Staphylococcus aureus survival within host cells was highly dependent on multiplicity of infection (MOI; e.g.,;; ) and also the growth phase of the bacteria used for infection. Green-fluorescent protein (GFP)-expressing S. aureus displayed bleaching of the fluorescent protein, which indicated degradation of the bacteria in polymorphonuclear neutrophils (PMN; Figure 1, Map Item 4a). The loss of fluorescence was not strain-specific and was seen in each of several different strains of S. aureus, including nosocomial and community-associated methicillin-resistant strains. When rapidly growing S. aureus was used for infection, the bacteria were found to be more susceptible to GFP bleaching. This indicated that these bacteria were cleared more efficiently. Bacterial disinfection was mainly dependent on hypochlorous acid (HOCl; ). Phagosomal acidification and digestion of S. aureus within professional phagocytes is required for MyD88-dependent toll-like receptor (TLR) responses to infection.

7Not all bacterial cells are disinfected by the phagolysosomes. S. aureus has been reported to persist inside phagocytes or endothelial cells for prolonged periods (;;;;;;;;; reviewed in ). Persistence is most often attributed to small colony variants (SCVs) of S. aureus. SCVs often present a metabolically quiescent, non-hemolytic, non-pigmented phenotype characterized by reversible auxotrophies in heme biosynthetic pathways or in oxidative phosphorylation as well as a defined transcriptome and proteome. Also, SCVs are generally found to be mutants in the accessory gene regulator locus (agr), thus failing to produce a variety of quorum sensing-controlled virulence factors. SCVs grow slowly and hence are more resistant to a variety of antibiotics (reviewed in ). Further, SCVs display a thick cell wall and an up-regulation of alternative sigma-factor σB, which enables S. aureus to cope with a variety of environmental stressors. There is increasing evidence that S. aureus can persist in vivo in human infections presumably owing to its extreme durability and resistance against a variety of environmental conditions and thus can serve as a potential source for recurrent infection (;;;; reviewed in; ). SCVs of S. aureus even have been shown to survive and grow within host cell phagosomes (; Figure 1, Map Item 4b). Also, the complementation of rsbU in laboratory strains restored activity of the alternative sigma-factor σB and led to intracellular growth of S. aureus within phagolysosomes of THP-1 phagocytes. Contrasting these reports, staphylococcal growth has been described after pathogen translo-cation to the host cell cytoplasm (Figure 1, Map Item 4c): S. aureus strain Newman is able to escape the phagosome and persists within human monocyte-derived macrophages (hMDM) which resulted in host cell lysis on day 5 after infection. The authors postulate that this survival within phagocytes might constitute a route for dissemination of staphylococcal infection. This is further corroborated by the identification of cytopro-tective effects on macrophages after phagocytosis of S. aureus. Thus, the up-regulation of anti-apoptotic factors upon staphy-lococcal infection is responsible for extended phagocyte lifetime. Both studies suggest that S. aureus might penetrate deeper into the tissue and even disseminate to different sites within “Trojan horse” phagocytes. Survival within PMN is reported to depend on the accessory regulator Sar1, which was crucial to S. aureus survival inside spacious vacuoles, whereas sar- strains were localizing to so-called “tight vacuoles”. Such different vacuoles can also be observed innon-professional phagocytes. The large vacuoles also are reminiscent of spacious Liste-ria-containing phagosomes, which were found to be non-acidified and non-degradative niches in macrophages. However, a more thorough characterization of the different vacuolar locations in intracellular S. aureus infections is lacking.

8Phagocytosed bacterial pathogens evade lysosomal killing, e.g., by disintegration of the organelle membrane in order to translocate into the host cell cytoplasm. Listeria monocytogenes co-opts the pore-forming toxin (PFT) listeriolysin O (LLO) and phos-pholipases (reviewed in; ), and Group A streptococci use the PFT strep-tolysin O. Phagosomal escape of S. aureus initially has been described by and the agr-dependency of this process has been demonstrated. The membrane destructive function of staphylococcal a-toxin thus suggested an involvement of the pore-former in phagosomal membrane disruption. The requirement for phago-somal escape and intracellular bacterial survival has been reported in CFT-1, a cystic fibrosis (CF) lung cell line as well as in macrophages (; Figure 1, Map Item 5a). In CFT-1, S. aureus translocates into the cytoplasm in an α-toxin-dependent manner and the bacteria replicate within the cytoplasm. However, in the LCSFN cell line complemented with the wild-type CF transmem-brane conductance regulator, CFTR, α-toxin has no effect. Further, it has been demonstrated that neither α-toxin overexpressing strains nor laboratory strains inducibly expressing α-toxin are capable of releasing staphylococci into the host cell cytoplasm. By expression of the amphiphilic 26 amino-acid (AA) peptide δ-toxin in the non-cytotoxic laboratory strain S. aureus RN4220 S. aureus was capable of escape in rates similar to that of heterolo-gously expressed LLO. δ-Toxin is encoded by the agr-effector RNAIII and is translated about 1 h after transcription of RNAIII. It is capable of lysing bacterial protoplasts, lysosomes, lipid spherules, mitochondria, and erythrocytes in vitro in a temperature-independent manner. Its activity and mode of action is comparable to that of non-ionic detergents (;;;; reviewed in ). δ-Toxin is encoded by RNAIII, the agr effector, and thus might constitute an immediate response to space limitation by phagosomal engulfment (Figure 1, Map Item 5b). However, membrane disruption by δ-toxin depended on the presence of the staphylococcal sphingomyelinase, β-toxin, which cleaves sph-ingomyelins (SM) to phosphorylcholine and ceramide moieties. δ-Toxin hardly binds to negatively charged phospholipids, binds strongly to liquid-disordered domains and poorly to cholesterol and sphingomyelin liquid-ordered raft domains. In one model, β-toxin thus may cleave SM to ceramides, which tend to accumulate in membrane microdomains. The hydrophobic nature of ceramide-rich domains thus might constitute regions of δ-toxin assembly, which eventually lead to target membrane permeabilization (Figure 2). S. aureus strain USA300 LAC, however, is escape proficient yet does not encode a functional β-toxin due to lysogeny of a β-converting phage. We thus have to hypothesize alternative factors that can act in phagosomal escape, such as a variety of lipases encoded by the staphylococcal genome or phenol-soluble modulins (PSMs; see below). Alternatively, prophages might be lost, e.g., during exposure to phagosomal reactive oxygen species, and then might contribute to phagosomal escape. A similar activation mechanism is used by Streptococcus pneumoniae, which produces hydrogen peroxide and thereby lyses S. aureus by a “remote control” prophage activation.

9For complete hemolysis of erythrocytes in sheep blood agar plates either a shift to 4°C (hot-cold hemolysis), osmotic stress, or synergistic toxins are necessary. δ-Toxin belong to the class of PSMs, which initially had been shown to be present within a hot-phenol extraction of S. epidermidis extracts and only later had been identified in S. aureus by sequence homology. The major groups of PSMs are divided into two operons referred to as PSMα and PSMβ. The PSMα operon is comprised of four open reading frames (ORFs) with approximately 20 AA. PSMβ encodes two ORFs, which are about 40 AA in length. The expression of PSMβ has been demonstrated to result in phagosomal escape in a gain-of-function study. Just like δ-toxin, PSMα and PSMβ are agr-dependently expressed. It recently was shown that the staphylococcal agr system is confinement induced and thus comprises not only a quorum sensing system but also a diffusion sensing system active at the single cell level. δ-Toxin and PSMs are hence produced upon phagosomal confinement and thus also low numbers of endocytosed staphy-lococci should be able to mount a toxin response in order to avoid lysosomal killing. Whereas gain of function assays demonstrated PSMβ activity in phagosomal escape, in vivo studies suggests a prominent role of PSMα, with the third ORF of the operon, PSMα3, being the most virulent principle. Despite these incongruities a common theme emerges: amphiphilic PSMs are involved in phagosomal escape. One possibility to explain the observed differences is the postu-lation of alternative pathways of phagosomal escape mechanisms for S. aureus.

10Molecular patterns of pathogens that reside in the cytoplasm of their hosts are detected by nucleotide-binding and oligomerization domain proteins NOD1 and NOD2, which detect γ-D-glutamyl-diaminopimelic acid and muramyl dipeptide, pep-tidoglycan components of Gram-negative and Gram-positive bacteria. Peptidoglycan binding leads to a conformation change in NOD proteins, which initiates the recruitment of ubiquitin lig-ases and kinases and ultimately results in nuclear translocation of NFκB and activator protein 1 and expression of inflammatory genes (reviewed in ). NOD2 signaling upon S. aureus infection has been found to induce cytokine production and thus might contribute to induction of inflammation, e.g., in the lung (; Figure 1; Map Item 6). NOD2-deficient mice exhibit a delayed inflammatory response and impaired bacterial clearance after infection with S. aureus. α-Toxin facilitates NOD2-dependent recognition of S. aureus muramyl dipeptide, possibly by interfering with phagosomal integrity.

11The observation that S. aureus can translocate into the cytoplasm of host cells and grow without an immediately ensuing cell death (e.g., ) illustrates that phagosomal escape is not identical with cytotoxicity Thus, the link between phagosomal escape and host cell death still needs to be elucidated.

12The classical separation of host cell death into programed cell death (PCD; apoptosis) and accidental cell death or necrosis has long been superseded after the identifying a large variety of cell death mechanisms (for a reviews, see;; ). Many of which are defined by key factors that allow the assessment of death pathway activation following bacterial infection (e.g., reviewed in ), although the synchronous activity of virulence factors from the staphylococcal arsenal renders unequivocal correlations between toxin and associated mode of cell death a daunting task. When virulent S. aureus strains are added to host cells in tissue culture, host cell death occurs via mechanisms that have been mainly identified as apoptotic (;,;,;;;;;; Figure 1, Map Item 7). α-Toxin is both required and sufficient for induction of leukocyte cell death, either apoptotic or necrotic. Leukocytes are sensitive to staphylococcal α-toxin. Already low toxin doses induce apoptosis accompanied by a breakdown of the mitochondrial transmembrane potential. A recent study suggests that caspase-2 acts as an initiator caspase during cell death of non-professional phagocytes. Caspase 2 was induced by potassium efflux due to pore-forming toxins such as staphylococcal alpha-toxin and aerolysin. By contrast, high doses induce necrotic cell death. 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. This suggests that cell death mechanisms are activated from within their intracellular location. The effect is highly specific, since fixed, non-hemolytic, rifampin-treated or weakly invasive staphylococci are not cytotoxic toward endothelial cells. Multiple other studies indicate that S. aureus might kill its other types of host cells from within. The virulence factors required for S. aureus-induced apoptosis in endothelial cells depend on agr and the alternative stress–response sigma-factor σB, but mainly seem to be independent of SarA - although there are contradicting results on the involvement of SarA.

13Phage-encoded Panton-Valentine leukocidin (PVL) predominantly destroys leukocytes, although there also is some species specificity for human and rabbit PMN. In PMN, PVL induced a rapid caspase-9/3-dependent cell death in vitro. The authors further identified to a mitochondrial localization of the PVL toxin. Isolated mitochondria were permeabilized for pro-apoptotic factors such as cytochrome c (CytC) by PVL which suggested that PVL is able to create pores in the mitochondrial outer membrane and thus triggers a Bax-independent mitochondrial pathway of host cell apoptosis (; Figure 1, Map Item 8). During the intrinsic pathway of apoptosis release of CytC from injured mitochondria leads to activation of apoptosis-activating factor-1 (APAF-1). Oligomerizing CytC/APAF-1 recruits and subsequently activates pro-caspase 9. Caspase 9 then proteolytically activates effector caspases, which finally cleave their respective substrates resulting in membrane blebbing, and DNA fragmentation (reviewed in ). Similarly, α-toxin has been described to activate caspases via the intrinsic death pathway independently of death receptor signaling (CD95/Fas/APO-1). Bcl-2 overexpressing Jurkat cells were protected from α-toxin mediated cell death although recent result suggest that the function of Bcl-2 in autophagy might be responsible for the observed phenomena.

14Interestingly, S. aureus also seems to be able to exert anti-apoptotic host cell responses. The pathogen was shown to suppress staurosporine-induced apoptosis in hMDM although early apoptotic features such as phosphatidylserine display at the outer plasma membrane leaflet, reduced mitochondrial membrane potential, CytC release, and caspase-3 activation are still observed. S. aureus infection strongly upregulated the expression of mitochondrial membrane potential stabilizing Bcl-2 and Mcl-1 gene products. Since also heat-killed S. aureus was able to suppress apoptosis in the host cells staphylococcal products such as lipoteichoic acid or peptidoglycan might activate the macrophages through intracellular pattern recognition sensors such as nucleotide oligomerization domain receptors (NOD;; ).

15Aside from apoptosis S. aureus is also able to induce pyronecrosis (Figure 1; Map Item 9). There, caspase-1 is activated as part of an inflammasome, which further consists of NOD-like receptor protein 3 (NLRP3) and the adaptor protein, apoptosis-associated speck-like protein containing a caspase-associated recruitment domain (ASC). S. aureus can function as a stimulus for NLRP3, however, the molecular identity of the stimulating signal is not known thus far. Lysosomal permeabilization is one NLRP3-activating principle, wherein release of the lysosomal protease cathepsin B into the cytoplasm contributes to NLRP3 activation (reviewed in; ). Thus, it seems likely, that lysosomal rupture or permeabilization by S. aureus toxins releases cathepsin which then activates the inflammasome. Indeed staphylococcal α, β, and γ-hemolysins have been shown to be important activation of the NLRP3 inflammasome. The pore-forming α-and γ-toxins of S. aureus permeabilize membranes and thus might be involved in cathepsin release and subsequent inflammasome activation. α-Toxin further is known to permeabilize the plasma membrane for potassium ions. In turn, potassium efflux activates the inflammasome. β-Toxin has been shown to be involved in phagosomal escape and thus also might act in cathepsin release.

16Autophagy sequesters cytoplasmic contents via an isolation membrane. Engulfment of cargo by the so-called phagophore forms double membrane-bound autophagic vesicles that eventually fuse with lysosomes to yield autolysosomes. Autophagy serves for degradation of organelles or self-digestion during nutrient limiting conditions such as starvation and is generally thought to constitute a cellular survival mechanism. During bacterial infections autophagy disposes of leaky vesicles or intracellular bacteria, however, bacterial pathogens have found multiple ways to subvert autophagy (reviewed in;;;;; ).

17Staphylococcus aureus was shown to interact with autophagosomes in a rather unique way when compared to autophagy-subvertive strategies of other bacterial pathogens. S. aureus inhibits fusion of phagosomes with lysosomes. It permeabilizes HeLa phagosomes by a mechanism dependent on Staphylococcus-secreted toxins. The leaky phagosomes are targeted by autophagy and within autophagosomes S. aureus replicates. Eventually the bacteria escape from their intracellular confinement into the host cell cytoplasm in an agr-dependent manner (; Figure 1, Map Item 9). Finally, host cell death is induced, which is independent of a caspase activation cascade but was blocked by overexpression of anti-autophagic Bcl-2. Induction of autophagy by rapamycin resulted in an increased number of recovered colony-forming units, whereas inhibition with wortmann in reduced the colonies recovered from the intracellular environment. S. aureus-induced autophagy resulted in a vacuolization of the host cell cytoplasm (“Swiss cheese phenotype”). agr-deficient S. aureus fail to induce autophagy, which results in maturation of bacteria-containing phagosomes followed by lysosomal degradation of the pathogens. α-Toxin is able to permeabilize membranes for Ca2+, an inducer of autophagy and autophagy targets phagosomes perforated by α-toxin in Chinese hamster ovary cells. Whereas latter observation is in line with the results obtained by it contrasts findings that α-toxin is not sufficient to permeabilize HeLa phagosomes.

18With about 200–300 virulence factors, S. aureus is able to exert a multitude of effects upon its eukaryotic host cells. Although many details have emerged through ground-breaking and recent research, only a minority of pathogenicity factors of S. aureus has been functionally annotated to date. Particularly the assessment of intracellular staphylococcal virulence is often hampered by the difficulty to discriminate between toxin effects that result from extracellular or intracellular bacteria, although inducible toxin-expressing might provide valuable tools for molecular dissection of host–pathogen interactions.

19With our advancing knowledge of cell death mechanisms ground-breaking experiments will have to be re-evaluated in order to understand the mechanisms of S. aureus-induced host cell killing. When comparing experimental research originating from different labs the influence of a body of factors should be taken into account that could lead to different infection outcomes:

20Staphylococcus aureus strain used for infection, its growth phase at the time of infection (and hence the bacterial growth medium), as well as MOI are important, whereas on the host side the cell type and hence the protein profile will drastically influence infection outcome by altering host cell susceptibility to bacterial toxins, response to pathogen-associated molecular patterns, expression of receptors, caspases, and other host factors.

21In order to identify activities of single toxins or virulence factors gain-of-function studies can be useful, e.g., using the a pathogenic S. carnosus as toxin delivery vehicle. However, such strategies will not easily identify pathways during which an orchestrated interplay of multiple virulence factors is required. However, novel high-throughput sequencing technologies of transposon insertions open up new possibilities for analysis of genome-wide mutant libraries of clinically relevant strains. Using such novel tools we will be able to address a lot of open questions with regard to intracellular staphylococcal infections: do different adhesins result in employment of different uptake mechanisms into non-professional phagocytes as these would subsequently result in different infection outcomes? What is the nature of tight and spacious vacuoles that are occupied by, for example, sarA- and sarA+ S. aureus, respectively? Which alternative pathways for phagosomal escape do exist? Which modes of cell death are activated by a single strain in different host cell types or different strains in a single host cell line? There is still a lot to learn about the versatile facultative intracellular pathogen, Staphylococcus aureus.

22The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.