Intracellular Staphylococcus aureus Perturbs the Host Cell Ca2+ Homeostasis To Promote Cell Death

Despite being regarded as an extracellular bacterium, the pathogen Staphylococcus aureus can invade and survive within human cells. The intracellular niche is considered a hideout from the host immune system and antibiotic treatment and allows bacterial proliferation.

aureus infection, 233 shRNAs were enriched and 311 were depleted (Table S1), reminiscent of shRNAs that negatively and positively regulate infection-induced cell death. Genes targeted by shRNAs with a log 2 FC of 3 (invasion) or 3.5 (cytotoxicity), respectively, are illustrated in heatmaps ( Fig. S1D and E). We next subjected the target genes of all differentially enriched or depleted shRNAs to gene set enrichment analysis (GSEA) ( Fig. 1B and C; see also Table S2 in the supplemental material). Among the pathways significantly enriched in the cell population without intracellular infection, we identified the endocytosis pathway. Other pathways required for S. aureus cytotoxicity were gonadotropin-releasing hormone (GNRH) signaling, phenylalanine metabolism, genes related to GAP junctions, homologous recombination, lysine degradation, tryptophan metabolism, butanoate metabolism, and long-term potentiation. We identified two pathways, the "one carbon pool by folate/folate biosynthesis" and the calcium signaling pathway, where shRNAs were enriched in both S. aureus invasion and cytotoxicity.
Intracellular S. aureus bacteria induce increase in cytoplasmic Ca 21 . The GSEA analysis revealed the calcium signaling pathway to be involved in both S. aureus invasion and cytotoxicity. In addition, pathways connected to calcium signaling like longterm potentiation, GNRH signaling, and gap junctions, were enriched for the S. aureus intracellular cytotoxicity gene set. Long-term potentiation implies ER calcium channels and Ca 21 -sensing and -regulated proteins, GNRH signaling is known to increase intracellular Ca 21 levels and PKC activation (48), and gap junctions form intercellular connections that are permeable for, e.g., calcium ions. Therefore, we investigated if intracellular S. aureus interfered with host cell Ca 21 signaling and thus may activate signaling pathways and Ca 21 -sensing proteins implicated in infection-induced host cell death. We used the genetically encoded fluorescent calcium sensor R-Geco (49) and generated a transgenic HeLa cell line in order to monitor changes in cellular Ca 21 concentration after infection with S. aureus by time-lapse microscopy. Extracellular bacteria were removed by treatment with lysostaphin. We observed a massive increase of the cytosolic Ca 21 concentration after the onset of intracellular bacterial replication and host cell shrinkage, which was followed by the formation of membrane blebs and a subsequent loss of bacterial and cellular fluorescence ( Fig. 2A; see also Video S1 in the supplemental material). Relative quantification of the cytosolic Ca 21 concentration demonstrated a roughly 2-to 6-fold increment of R-Geco fluorescence in infected cells between 5 and 7 hours postinfection (h p.i.) (Fig. 2B). The high cytosolic Ca 21 level persisted for several minutes and up to 1 h. No major changes in cellular Ca 21 concentrations were detected in uninfected cells ( Fig. 2B and Video S1). This was independent of the bacterial strains used, since we observed that the methicillin-resistant and cytotoxic strain JE2 also induced a rise in cytoplasmic Ca 21 ( Fig. 3F and G).
Since several S. aureus toxins were shown to induce Ca 21 flux in human cells (37), we treated HeLa R-Geco cells with 10% sterile-filtered S. aureus culture supernatant. This led to high and frequent oscillations of cytosolic Ca 21 in most cells, which was followed by a sustained Ca 21 increase ( Fig. 2C and Video S2 in the supplemental material). Similarly to intracellular infection with S. aureus, this cytoplasmic Ca 21 overload occurred after cell contraction but prior to the formation of membrane blebs (Fig. 2D), supporting the interpretation that a secreted virulence factor may trigger Ca 21 fluxes from within the host cell.
The S. aureus pore-forming toxin alpha-hemolysin (hla) is known to form Ca 21 -permissive pores and thus cytosolic Ca 21 elevations in different types of host cells (50,51). Additionally, alpha-toxin plays a role in intracellular cytotoxicity of S. aureus (Fig. 3A) (28,52). Therefore, we investigated the Ca 21 flux-inducing properties of an isogenic S. aureus hla deletion mutant (6850 hla). Interestingly, infection of HeLa R-Geco cells with the 6850 hla mutant did not abolish the cytosolic Ca 21 increase ( Fig. 3B and C). Furthermore, the amplitude, as well as the latency, of maximal R-Geco fluorescence was not significantly different in HeLa cells infected with S. aureus 6850 hla compared to that in the wild type ( Fig. 3D and E). Another virulence factor, which we recently identified as playing a role in S. aureus intracellular cytotoxicity, is the cysteine protease staphopain A (scpA) (32). Mutation of scpA in the JE2 strain also did not impair cytoplasmic Ca 21 increase (Fig. 3F to H); however, the onset of the cytosolic Ca 21 rise was delayed by 1.5 h on average (Fig. 3I) and was shorter. Furthermore, we observed Ca 21 oscillations (Fig. 3G). This is in concordance with previous findings, in that knockout of scpA delays S. aureus intracellular cytotoxicity (32).
Infection with intracellular S. aureus causes influx of extracellular Ca 21 in the host cell. We next studied the origin of the calcium ions. A preliminary test showed that a combination of Ca 21 withdrawal and chelation with BAPTA [1,2-bis(2-aminophenoxy)ethane-N,N,N9,N9-tetraacetic acid] resulted in a strong reduction of R-Geco fluorescence upon addition of the calcium ionophore ionomycin (see Fig. S2A in the supplemental material). Thus, HeLa R-Geco cells were infected with S. aureus 6850, and the cell culture medium was replaced with Ca 21 -containing or Ca 21 -free medium (including BAPTA) after bacterial invasion and phagosomal escape (3 h p.i.). Subsequently, infected cells were monitored for changes in Ca 21 concentration ( Fig. 4A and B). When Ca 21 was omitted from the medium and additionally chelated with BAPTA, the relative R-Geco fluorescence showed a significantly reduced maximal amplitude (P = 0.0004) in S. aureus-infected cells (Fig. 4C), whereas no temporal changes of the cytosolic Ca 21 peak were noticed upon omission of extracellular Ca 21 (Fig. 4D). Differences in cell morphology were not observed under the tested conditions. Additionally, Ca 21 withdrawal from the medium significantly reduced S. aureus intracellular cytotoxicity (P = 0.0085) by 61% (Fig. 4E).
We next depleted intracellular Ca 21 with the cell-permeable chelator BAPTA-AM [1,2-bis(2-aminophenoxy)ethane-N,N,N9,N9-tetraacetic acid tetrakis(acetoxymethyl ester)]. Increasing concentrations of BAPTA-AM delayed, but did not reduce, an ionomycin-induced cytosolic Ca 21 increase (Fig. S2B). Furthermore, concentrations greater than 10 mM BAPTA-AM were cytotoxic to HeLa cells after several hours of incubation (data not shown). Therefore, we used 5 mM BAPTA-AM to treat HeLa R-Geco cells 2 h after S. aureus infection, and R-Geco fluorescence was recorded over time. However, no differences in the cytoplasmic Ca 21 increase between BAPTA-AM-treated and dimethyl sulfoxide (DMSO)-treated cells were observed ( Fig. 4F and G). Also, amplitude and latency of the maximal Ca 21 concentration were not significantly altered by BAPTA-AM treatment ( Fig. 4H and I). Thus, the Ca 21 -chelating effect of BAPTA-AM was likely too small to interfere with S. aureus-induced cytosolic Ca 21 increase. Intracellular S. aureus perturbs ER Ca 21 homeostasis. Since Ca 21 release from intracellular stores of the host cell may also contribute to the cytoplasmic Ca 21 increase triggered by intracellular S. aureus, we generated a reporter cell line stably expressing a red fluorescent endoplasmic reticulum (ER)-targeted fluorescent Ca 21 sensor (ER-LAR-Geco) (53), in addition to a green fluorescent cytosolic Ca 21 indicator (G-Geco). The localization of ER-LAR-Geco was compartment specific, with a typical distribution of the fluorophores to the ER (Fig. S2C). We infected HeLa reporter cells with S. aureus 6850 and monitored fluorescence over time. In contrast to the cytoplasm, only small changes in ER Ca 21 concentration were observed in infected host cells ( Fig. 5A and B and Fig. S2D). Quantification of ER-LAR-Geco fluorescence disclosed a decrease in ER Ca 21 concentration concomitant with the onset of cytosolic Ca 21 rise, whereas ER-LAR-Geco fluorescence increased minutes later, then dropped again with the decrease of G-Geco fluorescence. Compared to the changes in cytosolic Ca 21 concentration, the shift of the ER-LAR Geco fluorescence was minor. To determine the dynamic range of the ER Ca 21 indicator, we treated HeLa ER-LAR-Geco G-Geco cells with Ca 21 -perturbing agents. Ionomycin and thapsigargin also induced only small changes in ER-LAR-Geco fluorescence compared to G-Geco fluorescence (Fig. S2E). These results show that the ER Ca 21 indicator was functional and that even high concentrations of Ca 21 -perturbing agents only trigger a small decrease in ER-LAR-Geco fluorescence.
Our observation of a simultaneous decrease in ER Ca 21 concentration and increase in cytosolic Ca 21 levels ( Fig. 5B and Fig. S2D) thus suggest that store-operated calcium entry (SOCE) may be implicated in S. aureus-induced cell death. We therefore blocked SOCE with 2-aminoethyl diphenylborinate (2-APB). HeLa R-Geco cells were treated with 30 mM 2-APB 1 h before infection and were subsequently infected with S. aureus 6850 in the presence of the inhibitor. Cytosolic Ca 21 concentrations were monitored over the time course of infection. We found that 2-APB did not interfere with the amplitude of the bacterially induced cytoplasmic Ca 21 increase, but the inhibitor significantly (P , 0.0001) delayed the S. aureus-triggered cytoplasmic Ca 21 rise (Fig. 5C to F). Quantification of lactate dehydrogenase (LDH) release revealed that 2-APB treatment significantly (P = 0.0015) reduced the cytotoxicity of S. aureus at 6 h p.i. by 99% compared to that in DMSO-treated cells (Fig. 5G). This effect of 2-APB was reproducible in other epithelial cell lines, such as A549 and 16HBE14o 2 cells, infected with S. aureus   Fig. S3A and B in the supplemental material). Similarly, the cytotoxicity of S. aureus JE2 in HeLa cells was abrogated by 2-APB treatment (Fig. S3C).
Host cell invasion and phagosomal escape of S. aureus are prerequisites of intracellular cytotoxicity. We found that both were not significantly affected by 2-APB treatment ( Fig. S3D to F), although we noticed a small, nonsignificant difference in intracellular bacteria at 3 h p.i. in 2-APB-treated samples compared to that in cells treated with the solvent control (Fig. S3F). Live cell imaging also showed attenuated intracellular replication of S. aureus upon 2-APB treatment (Fig. 5C). In cell culture medium, 30 mM 2-APB had only a mild inhibitory effect on S. aureus replication (Fig. S3G) and did not affect expression of virulence factors, such as alpha-toxin and RNAIII (Fig. S3H).
To minimize effects on bacterial intracellular replication, HeLa R-Geco cells were infected with S. aureus, and 2-APB was only added to infected cells at 3 h p.i. (Fig. 5H to J). Time-lapse imaging showed that intracellular S. aureus was replicating despite 2-APB treatment (Fig. 5H). However, addition of the inhibitor did not abolish the S. aureus-triggered cytosolic Ca 21 increase. Both amplitude and latency of the cytosolic Ca 21 increase were not significantly altered by 2-APB treatment under these conditions ( Fig. 5I and J).
Intracellular S. aureus induces a mitochondrial Ca 21 increase and cell lysis. The second major intracellular Ca 21 store in addition to the endoplasmic reticulum (ER) is the mitochondria. To investigate changes in mitochondrial Ca 21 concentration during S. aureus infection, we generated a dual Ca 21 reporter cell line stably expressing a red fluorescent mitochondrially targeted fluorescent Ca 21 sensor (Mito-LAR-Geco) (53) in addition to a green fluorescent cytosolic Ca 21 indicator (G-Geco) (see Fig. S4A in the supplemental material), and we infected these cells with S. aureus. Within S. aureusinfected cells, we detected an increase in mitochondrial Ca 21 concentration, which ensued minutes after the increase in cytosolic Ca 21 ( Fig. 6A and B and Fig. S4B). Interestingly, after the initial rise, no or only a partial decrease of mitochondrial Ca 21 was detected in contrast to G-Geco fluorescence.
Mitochondrial Ca 21 overload is associated with necrotic cell death. We therefore tested for plasma membrane permeability in S. aureus-infected HeLa R-Geco cells by addition of the small membrane-impermeant fluorophore Alexa Fluor 647 hydrazide (AF647) to the culture medium. We observed that in infected cells, influx of the fluorescent dye ensued after the increase in cytosolic Ca 21 ( Fig. 6C and Fig. S4C). Quantification of the relative fluorescence of single infected cells revealed that AF647 fluorescence signal started to increase when R-Geco fluorescence reached maximal levels (Fig. 6D). In contrast, uninfected cells or host cells infected with the noncytotoxic S. aureus strain Cowan 1 did not show evidence of AF647 influx and cytoplasmic Ca 21 increase ( Fig. 6C and E and Fig. S4C).
S. aureus-induced cytosolic Ca 21 rise is followed by caspase and calpain activation. S. aureus-induced cell death displays features of both necrosis and apoptosis (6), and Ca 21 is also involved in regulation of apoptosis (54). Therefore, activation of effector caspases 3 and 7 visualized by CellEvent caspase-3/7 green detection reagent was added to HeLa R-Geco cells infected with S. aureus 6850. We observed that the increase in R-Geco fluorescence was followed by an increase in CellEvent caspase 3/7 mbio.asm.org 10 fluorescence ( Fig. 7A and B). Fluorescence quantification on a single-cell level demonstrated that the effector caspases were activated in infected cells when the cytoplasmic Ca 21 signal peaked ( Fig. 7B and Fig. S4D). In contrast, fluorescence of CellEvent caspase 3/7 and R-Geco did not increase notably in uninfected cells (Fig. S4D).
Calpains are Ca 21 -activated cysteine proteases known to be implicated in cell death processes such as apoptosis (43). Since calpain activation involves autocatalytic proteolysis (55), we investigated the degree of calpain 4 and calpain 1 proteolysis by Western blotting at 1, 3, and 4.5 h after S. aureus infection (Fig. 7C). An increase in autocatalytically processed calpains was not observed in uninfected cells or in cells infected with the noncytotoxic control strain S. aureus Cowan 1, or well in cells infected with cytotoxic S. aureus at early time points (1 and 3 h p.i.). At 4.5 h. p.i., in contrast, calpain autolysis products were detected. At this time, we also found reduced protein levels of the calpain-specific endogenous inhibitor calpastatin (Fig. 7C). This suggested that calpains were activated in S. aureus-infected cells.
Since calpain-specific cleavage of the cytoskeletal protein a-spectrin is also used as a biomarker for calpain activation (56), we tested S. aureus 6850-infected HeLa cells by Western blotting for the calpain-specific breakdown products of a-spectrin (150 and 145 kDa). Protein bands with these molecular masses could be detected at 5 and 6 h p.i. (Fig. 7D). In contrast, infection of HeLa cells with the noncytotoxic Cowan 1 strain did not result in cleavage of a-spectrin. Additionally, a fluorogenic calpain substrate (Boc-Leu-Met-CMAC [7-amino-4-chloromethylcoumarin, t-BOC-Lleucyl-L-methioninamid]) specifically demonstrated calpain activity in host cells infected with the cytotoxic 6850 strain, but not in those infected with S. aureus Cowan 1 (Fig. 7E).
We therefore deleted the genes for calpain 1 and calpain 4 (CAPN1 and CAPN4, respectively) by a CRISPR/Cas9-based approach within the haploid HAP1 cell line, in which S. aureus demonstrates an intracellular pathogenicity comparable to that in HeLa cells (see Fig. S5A to E in the supplemental material). As observed within HeLa cells, HAP1 cells demonstrated calpain activation, as well as an S. aureus-induced cytoplasmic Ca 21 increase (Fig. S5F to H). CAPN1 and CAPN4 gene knockouts were verified by immunoblotting (Fig. S5I). However, the gene deletions did not reduce the cytotoxicity of S. aureus (Fig. 7F).

DISCUSSION
Previously conducted genome-wide screens that investigated staphylococcal virulence factors mainly involved the addition of purified toxins (57)(58)(59). However, S. aureus is readily internalized even by nonprofessional phagocytic host cells and is cytotoxic after phagosomal escape. Here, we employed a lentiviral shRNA library to identify host genes that contribute to host cell death caused by the intracellular bacteria. Host cells without intracellular infection were enriched for shRNAs targeting components of the endocytic machinery thereby supporting the hypothesis that S. aureus host cell invasion is a process driven by the host endocytosis machinery and verifying the validity of our screening procedure.  Aside from the calcium signaling pathway, the "one carbon pool by folate" pathway and "folate biosynthesis" were identified as playing a role in S. aureus invasion and intracellular cytotoxicity. Knockdown of factors involved in these pathways may result in metabolic and bioenergetic changes of the host cell and thus impair intracellular survival of S. aureus, which previously was shown to exploit the host cell carbon metabolism for intracellular replication (60,61). Furthermore, it was shown that the intracellular bacterium is able to alter the amino acid metabolism of its host cell (61). "Phenylalanine metabolism." "tryptophan metabolism," and "lysine degradation" were identified by our screen to be crucial for cytotoxicity of intracellular S. aureus. Phenylalanine levels are increased in HeLa cells infected with S. aureus USA300 (60). Tryptophan metabolism is mediated by an enzyme called indoleamine 2,3-dioxygenase-1 (IDO1), which induces the conversion of tryptophan into kynurenine and downstream metabolites (62,63). Kynurenine was shown to promote apoptosis of human and mouse neutrophils and to inhibit production of reactive oxygen species (ROS) (64). Similarly, kynurenine may be utilized by intracellular S. aureus to induce cell death in epithelial cells. The specific relationship between these pathways and intracellular infection with S. aureus remains to be elucidated.
When comparing our results to a previously published shRNA screen of S. aureusinfected cells (65), genes involved in long term potentiation were identified in both studies. Inhibition of ER Ca 21 refill by the SERCA inhibitor thapsigargin impaired bacterial internalization but not S. aureus intracellular replication or cytotoxicity (65). Our study, which additionally differentiated between invasion-and cytotoxicity-relevant factors, demonstrated that intracellular S. aureus induced a perturbance of intracellular Ca 21 homeostasis in epithelial cells that culminated in host cell death. So far, bacterial manipulation of the host cell Ca 21 signaling and homeostasis were only shown to be involved in the disintegration of epithelial barriers, cytoskeletal rearrangements required for cell binding or internalization of the pathogen or control of innate defense cells (66).
Secreted S. aureus virulence factors induced strong cytosolic Ca 21 oscillations and elevations when added extracellularly to HeLa cells (Fig. 2C). We observed that a cytosolic Ca 21 overload occurred after cell contraction but before the formation of plasma membrane blebs, which morphologically resembled those of intracellular S. aureus infection (compare Fig. 2A and D). Thus, virulence factors secreted by intracellular S. aureus may trigger the Ca 21 overload when released into the cytoplasm of the host cell. Although the inner leaflet of the plasma membrane is composed of different proteins and phospholipids compared to the outer leaflet, our results suggest that receptor-independent pore formation of S. aureus toxins may still occur from the inner side of the plasma membrane. Alpha-hemolysin, for instance, has been shown to nonspecifically integrate into membranes at high concentrations and thereby induce Ca 21permissive pores (50,51,(67)(68)(69)(70)(71). However, infection of epithelial cells with an alphatoxin mutant did not significantly reduce or delay the cytosolic Ca 21 overload (Fig. 3B to E), suggesting that other factors are responsible for the observed phenotype. Furthermore, inactivation of the S. aureus cysteine protease staphopain A, which plays a role in S. aureus intracellular cytotoxicity (32), did not abolish cytosolic Ca 21 overload (Fig. 3F to I).
We further observed that, in S. aureus-infected cells, ER Ca 21 levels decreased together with a cytosolic Ca 21 rise (Fig. 5B and Fig. S2D) and that the latter also depended on calcium ions in the culture medium ( Fig. 4A to C). Removal of extracellular Ca 21 also attenuated S. aureus intracellular cytotoxicity (Fig. 4E). This points to a  role of store-operated Ca 21 entry (SOCE). During SOCE, a release of Ca 21 from the ER triggers an influx of the ion from the extracellular space to refill the internal Ca 21 stores (42). In order to investigate the role of SOCE in the process of S. aureus-induced Ca 21 perturbations and host cell death, we applied 2-APB, which inhibits SOCE at higher concentrations ($10 to 50 mM) (72,73). Treatment of infected cells with 2-APB led to strong reduction of S. aureus intracellular cytotoxicity and delay of bacteria-induced cytosolic Ca 21 increase (Fig. 5C to G), whereas it did not impair S. aureus invasion and phagosomal escape (see Fig. S6D to F in the supplemental material). S. aureus intracellular growth was attenuated upon inhibitor treatment (Fig. 5C and Fig. S3F), which may imply that SOCE is not only required for induction of cytotoxicity but also for intracellular replication of S. aureus. 2-APB had only a minor direct effect on bacterial growth, even over extended times (Fig. S3G), and it did not affect the expression of the S. aureus major virulence factors alpha-toxin or RNAIII (Fig. S3H). When infected epithelial cells were treated with 2-APB only later during S. aureus intracellular infection (approximately 3 h p.i.), intracellular S. aureus was able to replicate (Fig. 5H) but no effect of 2-APB on cytosolic Ca 21 increase was observed ( Fig. 5I and J). Therefore, SOCE may be involved in the S. aureus intracellular lifestyle at an early time point of infection (before 3 h p.i.) and interfere with S. aureus replication. No clear conclusion regarding the involvement of SOCE in S. aureus-induced host cell death can be drawn.
The rise in cytosolic Ca 21 was succeeded by increased Ca 21 levels in mitochondria ( Fig. 6B and Fig. S7B in the supplemental material). Mitochondria are involved in Ca 21 compartmentalization and are able to buffer cytosolic Ca 21 to a certain extent. Under high cytosolic Ca 21 concentrations, larger amounts of Ca 21 can accumulate in the mitochondria (74), and excessive Ca 21 uptake by mitochondria may lead to mitochondrial permeability transition (MPT)-driven necrosis, which is characterized by the abrupt loss of inner mitochondrial membrane osmotic homeostasis (75,76). Cell lysis as a characteristic of a necrotic type of cell death was detected as a consequence of prolonged high cytosolic Ca 21 concentrations (Fig. 6C and D). Accordingly, the cytosolic Ca 21 rise induced by intracellular S. aureus may lead to cell death by subsequent mitochondrial Ca 21 overload and MPT-driven necrosis.
Interestingly, it is speculated that the facultative intracellular pathogen Shigella induces mitochondrial Ca 21 overload by a sustained cytosolic Ca 21 increase due to plasma membrane permeabilization and thus MPT-driven necrotic cell death (77)(78)(79). However, so far, local and global Ca 21 fluxes were detected only during host cell invasion of Shigella (80)(81)(82). Calpains, which are activated by Ca 21 release from intracellular stores likely act as executioner proteases during for MPT-driven necrosis. Additionally, in S. aureus-infected keratinocytes, degradation of the endogenous calpain inhibitor calpastatin and reduced bacterial cytotoxicity upon calpain inhibitor treatment were observed (83).
We found that calpains were also activated in S. aureus-infected HeLa cells (Fig. 7C  to E). However, knockout of calpains 1 and 4 did not abolish or reduce S. aureusinduced host cell death (Fig. 7F). The high and sustained concentration of cytosolic Ca 21 induced by intracellular S. aureus may induce several cell death pathways in parallel by excess stimulation of Ca 21 -sensitive targets, such as phospholipases, proteases, and endonucleases, in the cytosol (84)(85)(86). The Ca 21 -induced damage may also trigger necrotic cell death without involving specific signaling. In this scenario, knockout of calpains cannot restrain host cell death.
Indications for involvement of an apoptotic mode of cell death in S. aureus-infected cells were found, too. The activation of effector caspases 3/7 followed after the increase in cytosolic Ca 21 ( Fig. 7A and B). S. aureus-infected macrophages are also killed through a pathway characterized by membrane blebbing and activation of caspases 3/7, followed by cell lysis (87). Therefore, an apoptotic mode of cell death, which readily transits to a necrotic state, such as secondary necrosis, is also likely (88).
In summary, our study shows that intracellular S. aureus induces Ca 21 perturbations in the host cell (Fig. 8). Both plasma and ER membranes become permeable for calcium ions and induce a cytoplasmic, and subsequently a mitochondrial, Ca 21 overload, which results in breakdown of the plasma barrier function and cell death. Furthermore, we demonstrate that S. aureus activated calpains 1/4 and effector caspases after translocation from the endocytic compartment to the host cytosol. Several cell death pathways thus are executed in parallel by activation of multiple Ca 21 -sensitive proteins, and we found evidence for MPT-driven necrosis as well as for apoptotic pathways. However, the damage induced by strong perturbations of the cellular Ca 21 homeostasis and Ca 21 overload may not allow reliable detection of specific signaling pathways and therefore may explain the confusing picture in the literature of host cell death pathways activated by intracellular S. aureus (6).

MATERIALS AND METHODS
Bacterial culture conditions. Escherichia coli strains were grown in lysogeny broth (LB), and Staphylococcus aureus strains were grown in tryptic soy broth (TSB; Sigma), if not stated otherwise. Media were supplemented with appropriate antibiotics when necessary, and broth cultures were grown aerobically at 37°C overnight at 180 rpm. E. coli was selected on LB plates containing 100 mg/ml ampicillin, and selective TSB plates for S. aureus were prepared using 10 mg/ml chloramphenicol and/or 10 mg/ ml erythromycin.
For preparation of sterile supernatant, bacteria were grown overnight at 180 rpm in TSB, and cultures were adjusted to an optical density at 600 nm (OD 600 ) of 7. Subsequently, bacterial cultures were centrifuged, and the supernatant was sterile-filtered (0.22-mm pore size) and diluted to 10% with cell culture medium.
Bacterial growth curves were measured using a Tecan plate reader. Bacterial cultures were inoculated in triplicates to an OD 600 of 0.1 in 400 ml cell culture medium without fetal bovine serum (FBS) and grown for intended periods at 37°C in a 48-well microtiter plate. Absorbance was recorded every 10 min at 600 nm.
Generation of constructs. All used cell lines, strains, plasmids, and oligonucleotides can be found in Table S3 in the supplemental material.
In order to constitutively express the codon-adapted fluorescent reporter proteins mRFPmars and Cerulean, we amplified the SarAP1 promoter from genomic S. aureus DNA by PCR using oligonucleotides SarAP1-f and SarAP1-r. The resulting PCR product was cloned into pCR2.1-Topo (Invitrogen) and was sequence verified, restricted with SalI and KpnI, and eventually ligated into the plasmids pmRFPmars and pCerulean (89), opened with SalI and KpnI. We thereby exchanged the anhydrous tetracycline-inducible promoter, including the open reading frame (ORF), for the repressor TetR with the constitutively active SarAP1 promoter, yielding pSarAP1-mRFP and pSarAP1-Cerulean, respectively. All assembled vectors were transformed into chemically competent E. coli DH5a cells and confirmed via PCR and Sanger sequencing (SeqLab, Göttingen, Germany). Subsequently, vectors were electroporated into S. aureus RN4220 and, if required, further either electroporated or transduced (using phage w 11) into the respective strains (Table S3).
For generation of stably integrating reporter gene expression vectors, we used the lentiviral vector pLVTHM (90). Generation of the phagosomal escape reporter has been previously described (19,91). Lentiviral vectors for expression of R-Geco, G-Geco, ER-LAR-Geco, and Mito-LAR-Geco used the following plasmids as PCR templates: CMV-R-GECO1, CMV-G-GECO1.1, CMV-ER-LAR-GECO1, and CMV-mito-LAR-GECO1.2, respectively (Table S3). The reporter genes were PCR amplified with oligonucleotides PmeI-pCMV-pLV and SpeI-BGHpolyA-pLV, thereby introducing PmeI and SpeI restriction sites. PCR products were hydrolyzed with the respective restriction enzymes (Fermentas) and were ligated into the correspondingly opened pLVTHM vector (90), thereby replacing the original enhanced green fluorescent protein (eGFP) open reading frame with the respective sensor proteins. A 20-mg aliquot of a plasmid preparation of these pLVTHM-derived plasmids was used in calcium phosphate-based cotransformation of a 15-cm dish of 293T cells along with 10 mg psPAX and 10 mg pVSVG. Dulbecco's minimal essential medium (DMEM) growth medium was exchanged after 4 to 8 h. Two days after transfection, the supernatant was harvested and sterile-filtered (0.45-mm filter). HeLa and HAP1 cell lines were transfected with the virus particle preparations in the presence of 10 mg/ml Polybrene, and after three subculturings, the resulting transgenic cell lines were sorted on a FACSAria III cell sorter (BD).
The hla deletion mutant of strain 6850 was constructed using allelic replacement using the plasmid pKOR1 (92). A gene replacement cassette encompassing genomic regions upstream and downstream of the hla ORF was amplified by PCR with oligonucleotides attB1-hla-up-f and hla-up-r and hla-down-f and hla-down-r, respectively. Purified PCR products were restricted with SacII, ligated, and subsequently Ca 21 -Induced Host Cell Death by Intracellular S. aureus ® recombined into pKOR1 using BP Clonase (Invitrogen) and were subsequently transformed into competent E. coli cells. The resulting vector, pKOR1-Dhla, was electroporated in S. aureus RN4220 (93), which accepts foreign DNA, and eventually in the target strain 6850. Selection of mutants was performed as described by Bae and Schneewind (92), and mutants were screened for successful gene replacement by PCR with the oligonucleotides hla-test-f and hla-test-r. Mutants were further checked for the phenotypic loss of beta-hemolysis on sheep blood agar.
shRNA screen. For generation of the HeLa-shRNA library, 293T cells were cultivated in five 15-cm dishes to a density of 80%. Every dish was transfected with 10 mg of pGIPZ lentiviral shRNAmir library encoding 10,000 different shRNAs, 7 mg of psPAX2, and 3.5 mg of pMD2.G using polyethylenimine (1 ml/ 1 mg of total DNA). Virus-containing supernatant was collected at 48 h and 72 h posttransduction and sterile-filtered, and 1 mg/ml Polybrene was added. HeLa cells cultivated in five 15-cm dishes were infected with the produced virus at a surname multiplicity of infection (MOI) of 0.1. At 24 h postinfection, medium was changed, and cells were cultivated with puromycin-containing medium for selection. These five HeLa-shRNA sublibraries encoding 10,000 different shRNAs were cultivated under puromycin pressure until they were used for infection.
One day before infection, 3 Â 10 6 HeLa-shRNA cells of each of the five subgroups were seeded in seven 15-cm dishes. Six dishes were infected with S. aureus 6850 mRFP at an MOI of 30, and 20 mg/ml lysostaphin was added 1 h p.i. in order to kill extracellular bacteria. At 4.5 h p.i., dead cells were removed by two phosphate-buffered saline (PBS) washing steps, and adherent viable cells were collected for FACS. Viable cells were selected via forward and sideward scatter (FSC-A and SSC-A) (see Fig. S1A in the supplemental material). HeLa-shRNA cells, which express green fluorescent protein (GFP), were sorted for the mRFP-positive (infection with S. aureus 6850) and mRFP-negative (uninfected) population using a FACSAria III instrument. GFP fluorescence was measured using a 488-nm laser and a 530/30-nm band pass filter for detection, and mRFP fluorescence was measured using a 561-nm laser and a 610/20-nm band pass filter. One 15-cm dish of uninfected cells served as the input control.
Genomic DNA of the sorted uninfected and infected cells and input cells was isolated with a multisource genomic DNA miniprep kit (Axygen Biosciences). The shRNA cassettes were amplified via PCR (for primers, see Table S3), and separation from the genomic DNA was obtained via agarose gel electrophoresis. Samples were diluted to a final concentration of 2 ng/ml, and Experion analysis confirmed the purity and concentration of the DNA. Samples were prepared for Illumina sequencing as suggested by the company.
High-throughput sequencing was performed with the Illumina GAIIx genome analyzer. FASTQ files were aligned with Bowtie 1 to a reference library containing the expected 10,000 shRNA sequences of each sublibrary. Alignment of the sequencing reads to the reference library was quantified. Gene features without hits in any of the libraries were eliminated prior to normalization in order to avoid bias.  1 mM). Intracellular organelles, like the endoplasmic reticulum (ER) and mitochondria, known as the internal Ca 21 stores, accumulate Ca 21 and thus exhibit a higher Ca 21 concentration (100 to 500 mM) compared to that of the cytoplasm. Upon S. aureus infection, calcium ions from the extracellular space and the ER are recruited to promote a cytoplasmic and subsequent mitochondrial calcium overload, which likely results in necrotic cell death. In parallel, excess stimulation of Ca 21 -sensing proteins contributes to this effect. Additionally, effector caspases, which execute apoptotic cell death, are activated by intracellular S. aureus.

®
Data quality control and normalization were implemented in R. A batch effect among different libraries was evident by the relative log expression (RLE) plot and was carefully removed by remove unwanted variation (RUV) methods (94, 95) Further normalization and differential expression analysis were conducted with edgeR (96,97), whereby the trimmed mean of M values (TMM) method was applied.
To expose in more detail the systematic impact of the different pathways, we further conducted a gene set enrichment analysis (98) according to the pathways of the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Hallmark (gene set obtained from MSigDB ver. 7.0 [99]). We ranked all the genes with a score defined as follows: score = 2log 10 (P value) Â log 2 FC. The enrichment was computed using a permutation number of 5,000. The pathways with high positive enrichment scores together with P values lower than 0.02 have been carefully investigated.
Bacterial overnight cultures were diluted to an OD 600 of 0.4 and incubated for 1 h at 37°C and 180 rpm to reach the exponential growth phase. Then, bacteria were washed twice by centrifugation and used to infect HeLa cells at a multiplicity of infection (MOI) of 50, if not stated otherwise. After 1 h of cocultivation, extracellular bacteria were removed by 30 min of treatment with 20 mg/ml lysostaphin (AMBI), followed by washing and further incubation in medium containing 2 mg/ml lysostaphin and, if required, 30 mM 2-APB, until the end of experiment.
Live-cell imaging. Phase contrast and fluorescence microscopic images of live cells were acquired with a Leica DMR microscope connected to a SPOT camera using a 10Â (Leica HC PL Fluotar, numerical aperture [NA] = 0.32) or 20Â (Leica C Plan, NA = 0.3) objective and VisiView software.
For time-lapse imaging, HeLa cells were seeded in 8-well chamber m-slides (ibidi) 24 h prior to infection. Infection with fluorescent protein-expressing bacterial strains was performed at an MOI of 5, as described above. Time-lapse imaging of the samples was performed on a Leica TCS SP5 confocal microscope using a 20Â (Leica HC PL APO, NA = 0.7) dry objective or 40Â (Leica HC PL APO, NA = 1.3) or 63Â (Leica HCX PL APO, NA = 1.3 to 0.6) oil immersion objective. Prior to imaging, cell culture medium was substituted with imaging medium (RPMI 1640 without phenol red and containing 10% FBS and 2 mg/ml lysostaphin) containing inhibitors, fluorogenic enzyme substrates, and/or chemicals, if indicated. For Ca 21 depleted conditions, DMEM without calcium (catalog no. 21068028; Thermo Fisher Scientific) substituted with 0.2 mM BAPTA (Merck Millipore) was used. As a positive control, 1.8 mM CaCl 2 was added to this medium. Chelation of intracellular calcium ions was performed using BAPTA-AM (Merck Millipore). CellEvent caspase-3/7 green detection reagent (5 mM; Thermo Fisher Scientific) was applied for monitoring effector caspase activation, and 2.5 mM Alexa Fluor 647 hydrazide was applied for detecting cell lysis.
The m-slides were transferred to a prewarmed live-cell incubation chamber surrounding the confocal microscope, and a temperature of 37°C was applied during imaging. LAS AF software was used for setting adjustment and image acquisition. All images were acquired at a resolution of 1,024 Â 1,024 pixels and recorded in 8-bit mode at predefined time intervals. In certain cases, chemical treatment was performed during imaging in situ at the live-cell incubation chamber. All image processing steps were performed using Fiji (100). For quantification of fluorescence intensities, raw imaging data were used. For single-cell analysis, one region of interest (ROI) was defined for all recorded time frames. Fluorescence intensities (mean of relative fluorescence units [RFU]) were measured, background was subtracted, and data were normalized to time point zero (R 0 ) to obtain relative fluorescence values. The maximum amplitude was determined as the highest relative fluorescence value measured, and peak latency was defined as the time point of maximum amplitude.
Cytotoxicity assays. HeLa cells were infected as described above. At the desired time point after infection, medium, which possibly contained detached dead cells, was collected from the wells, and adherent cells were detached using TrypLE (Thermo Fisher Scientific). Adherent and suspension cells of each sample were pooled, and after centrifugation for 5 min at 800 Â g, cells were carefully resuspended in cell culture medium with 10 ml/ml 7-aminoactinomycin D (7-AAD; BD Biosciences). After 10 min of incubation in the dark, cells were immediately analyzed by flow cytometry using a FACSAria III instrument (BD Biosciences) and BD FACSDiva Software (BD Biosciences). Forward and sideward scatter (FSC-A and SSC-A) were used to identify the cell population, and doublet discrimination was performed via an FSC-H versus FSC-W and SSC-H versus SSC-W gating strategy. 7-AAD fluorescence was measured using a 561-nm laser for excitation and a 610/20-nm band pass filter for detection. In total, 10,000 events were recorded for each sample. Uninfected cells served as a negative control.
The LDH assay was performed as described previously (32). Briefly, HeLa cells were infected as described above, and at 1.5 h after infection medium was replaced by RPMI 1640 without phenol red containing 1% FBS and 2 mg/ml lysostaphin. At 6 h after infection, medium was removed from the wells, briefly centrifuged, and 100 ml of supernatant of each sample were transferred into the well of a 96-well microtiter plate in triplicates. LDH release was measured using the Cytotoxicity Detection Kit Plus (Roche) according to the manufacturer's instruction.
Invasion and intracellular replication assay. HeLa cells were infected with GFP-expressing strains and prepared for flow cytometry as described above. Adherent and suspension cells were resuspended in fresh medium without phenol red containing 1% FBS and 2 mg/ml lysostaphin. To determine invasion, Ca 21 -Induced Host Cell Death by Intracellular S. aureus ® infected cells at 1 h p.i. were treated with 20 mg/ml lysostaphin for 10 min to remove extracellular bacteria, and the percentage of GFP-positive cells representing the infected cells was measured by flow cytometry. Intracellular replication was determined by measurement of GFP fluorescence (arbitrary units [AU]) of the infected cells at 1 and 3 h after infection by flow cytometry using a FACSAria III instrument (BD Biosciences). The intensity of GFP fluorescence corresponds to the amount of intracellular bacteria. Gating and analysis were performed as described above. GFP fluorescence was measured using a 488nm laser and a 530/30-nm band pass filter for detection. Uninfected cells served as a negative control.
Phagosomal escape assay. Phagosomal escape was determined as described previously (32). HeLa YFP-CWT cells, which encode a cytoplasmic fluorophore (yellow fluorescent protein [YFP]) coupled to the staphylococcal cell wall binding domain (CWT) of lysostaphin allowing detection of cytoplasmic S. aureus, were infected with mRFP-expressing bacterial strains at an MOI of 10 in a 24-well m-plate (ibidi). After synchronization of infection by centrifugation, infected cells were incubated for 1 h to allow bacterial invasion. Subsequently, a 30-min treatment with 20 mg/ml lysostaphin removed extracellular bacteria, after which the cells were washed and medium with 2 mg/ml lysostaphin was added. At 3 hours after infection, cells were washed, fixed with 4% paraformaldehyde overnight at 4°C, and permeabilized with 0.1% Triton X-100, and nuclei were stained with Hoechst 34580. Images were acquired with an Operetta automated microscopy system (PerkinElmer) and analyzed with the included Harmony Software. Colocalization of YFP-CWT and mRFP signals indicated phagosomal escape.
RNA isolation. Bacteria were grown in infection medium without FBS with 30 mM 2-APB or DMSO as a solvent control at 37°C under shaking conditions (180 rpm) for 1.5 h (exponential phase) and 6 h (stationary phase), and OD 600 was adjusted to 0.14 or 0.46, respectively. Bacterial pellets were frozen in liquid nitrogen and stored at 280°C. RNA was extracted using the previously described TRIzol method (101). Briefly, bacterial pellets were resuspended in 400 ml buffer A (10% glucose, 12.5 mM Tris [pH 7.6], and 10 mM EDTA) and mixed with 60 ml 0.5 M EDTA. The samples were transferred to lysing matrix B tubes (MP Biomedicals) containing 500 ml Roti aquaphenol (Roth) and mechanically lysed at 6 m/sec for 45 s at 4°C using the FastPrep cell disruptor (FP120). After centrifugation, the aqueous phase was collected, mixed with 1 ml TRI reagent solution (Invitrogen), and incubated for 5 min at room temperature. Chloroform (100 ml) was added, mixed, and incubated for 3 min. After centrifugation, the aqueous phase was collected and chloroform extraction was repeated. RNA was precipitated using 500 ml isopropanol. After 15 min of incubation, the RNA pellet was washed with 75% ethanol, and the dried pellet was resuspended in diethyl pyrocarbonate water.
The Turbo DNA-free DNase kit (Invitrogen) was used to remove DNA contamination in RNA samples according to the manufacturer's instructions. Briefly, 10 mg isolated RNA was mixed with DNase I in Turbo DNase buffer and incubated for 30 min at 37°C, followed by addition of a bead-based DNase inactivation reagent to stop the reaction. After 5 min of incubation time, the mixture was sedimented by centrifugation and the aqueous layer was collected. DNA digestion was verified by PCR using quantitative real-time PCR (qRT-PCR) primers against gyrB (Table S3).
Quantitative real-time PCR (qRT-PCR). Complementary DNA (cDNA) was generated using the RevertAid first-strand cDNA synthesis kit (Thermo Scientific). Briefly, 1,000 ng DNA was mixed with random hexamer primers and incubated for 5 min at 65°C. Reaction buffer, nucleotide mix, RNase inhibitor, and reverse transcriptase were added and incubated for 5 min at 25°C, followed by 60 min at 42°C and 5 min at 70°C. Quantitative real-time PCR was performed on a StepOne Plus PCR system (Applied Biosystems) in a 96-well plate format. The reaction mixture was composed of Green mastermix (Genaxxon), 300 nM qRT-PCR primers (Table S3), and 100 ng cDNA in a total volume of 20 ml. Reactions were run for 40 cycles at 95°C for 15 s and 60°C for 1 min after an initial holding stage of 10 min at 95°C. At the end of each run, a melting curve was performed with 0.3°C temperature increments. Analysis of relative gene expression was done according to the threshold cycle (2 2DDCT ) method (102). Gene expression was normalized to the expression levels of the housekeeping gene gyrB. Nontemplate controls were run to verify the lack of contamination.
Gene knockout using CRISPR/Cas9. For designing gene-specific single guide RNAs (sgRNAs), the CRISPOR online tool (crispor.tefor.net) was used, and gene sequences were retrieved from Ensemble (www.ensembl.org). Exons coding for essential function(s) of the protein were chosen for genetic manipulation. Synthetized sgRNA oligonucleotides (Table S3) were cloned into pSp-Cas9(BB)-2A-GFP according to the protocol from Ran et al. (103). Insertion of the sgRNA was verified by Sanger sequencing (SeqLab) using primer U6-fwd.
For plasmid transfection, HAP1 cells with a low passage number were seeded into 6-well microtiter plates at a density of 5 Â 10 5 cells/well in IMDM (catalog no. 21980032; Thermo Fisher Scientific) containing 10% FBS (Sigma-Aldrich) and antibiotic 24 h before transfection. pSp-Cas9(BB)-2A-GFP-sgRNA (3.6 mg) and pSc-TIA-CMV-BSR-TIA (400 ng) were mixed with polyethylenimine and Opti-MEM (catalog no. 51985026; Thermo Fisher Scientific) for transfection of cells. After 24 h, cells were diluted and seeded in cell culture dishes (100 Â 20 mm). The next day, selection was applied by treatment with 30 mg/ml blasticidin. Subsequently, cells were monitored over 2 to 3 weeks and, if necessary, cells were washed and medium was renewed. When cell colonies were large enough, single colonies were each transferred to a well of a 12-well plate using trypsin-soaked, sterile Whatman paper. Gene knockout was verified by Western blotting.
Calpain activity assay. HeLa cells were infected with S. aureus 6850 GFP as described above. At 4 h p.i., 10 mM CMAC peptidase substrate t-BOC-Leu-Met (Thermo Fisher Scientific) was added to the cells and incubated for 30 min at 37°C and 5% CO 2 . Subsequently, phase contrast and fluorescence microscopic images of live cells were acquired with a Leica DMR camera.
Statistical analyses. Data were analyzed using Prism ver. 6.01 (GraphPad Software). For statistical analysis, three biological replicates were performed, if not indicated otherwise. All data are presented as means with standard deviation (SD). P values of #0.05 were considered significant. Pairwise comparisons were assessed using the unpaired Student's t test. Analysis of variance (ANOVA) was performed to determine whether means in a group was significantly different from each other. ANOVA was performed with Tukey's post hoc analysis for defining individual differences.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. VIDEO S1, MPG file, 1.