α-MEM cell culture medium was purchased from Gibco (Invitrogen, 11965-175) and fetal bovine serum (A15-101) from PAA (GE Healthcare Argentina S.A., Argentina). Nocodazole, Chloramphenicol, Gentamicin, Wortmannin and α-hemolysin from Staphylococcus aureus (H9395) were obtained from Sigma Aldrich (Buenos Aires, Argentina). Geneticin (G418) was purchased from Gibco (Invitrogen, Argentina).

The pEGFP-LC3, pRFP-LC3, pEGFP-Rab1b, pEGFP-Rab11, and pEGFP-Rab24 constructs were obtained as described in earlier publications (Munafó and Colombo, 2002; Savina et al., 2005; Zoppino et al., 2010). The plasmid pEGFP- Rab7wt was a generous gift from Bo van Deurs (University of Copenhagen) and the plasmid encoding enhanced GFP (EGFP)-Rab5wt was kindly provided by Dr. Philip D. Stahl (Washington University). The pCDNA Kif5wt, pCDNA Kif5T93N were kindly provided by Dr. Alfredo Cáceres (Mercedes and Martín Ferreyra Institute, Córdoba, Argentina). The pRFP-RILP and mCherry-RILP ΔN was gently provided by Dr. Walter Berón (IHEM, Mendoza, Argentina). The pEGFP-Rab34 was gently provided by Dr. Maximiliano Gutiérrez (Helmholtz Centre for Infection Research, Braunschweig, Germany). The pEGFP-Rab29 and pEYFP-Rab32 were gently provided by Drs. Jorge Galán and Stefania Spano (Yale University School of Medicine, New Haven).

CHO-K1 (ATCC) were grown in α-MEM supplemented with 10% FBS, streptomycin (50 μg/ml) and penicillin (50 U/ml). Stably transfected CHO cells overexpressing pEGFP-LC3, pEGFP-Rab1b or pEGFP-Rab7 wild type and pEGFP-Rab7T22N were selected with 0.5 mg/ml of geneticin. Cells were transfected or cotransfected using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions.

Staphylococcus aureus strain wild type (8325-4), the mutant deficient for α-hemolysin Hla (–) (DU1090) or the Hla (–) mutant complemented with a α-hemolysin plasmid (DU1090/pDU1212) [Hla(+)] were kindly provided by Dr. Claudia Sola (CBICI-CONICET, Córdoba, Argentina). These bacterial strains were grown overnight at 37°C in 10 ml of Luria-Bertani (LB) broth with appropriate antibiotics. Bacteria were resuspended in infection medium (αMEM containing 10% FBS) at an OD₆₅₀ of 0.4 (approximately 4 × 10⁸ CFU) and then were diluted to achieve a multiplicity of infection (MOI) of 10:1 (bacteria:cell) in the infection medium. After 1 h of infection, the medium was washed to remove extracellular bacteria and new infection medium was added to cells. Subsequently, cells were incubated for different periods of times (2, 4, or 6 h) and fixed with 4% paraformaldehyde (PFA) or processed as living cells by confocal microscopy. In all cases, after 1 h of infection cells were washed and gentamicin (100 μg/ml) was added for 30 min to kill extracellular bacteria.

For analysis in fixed samples, bacterial DNA was labeled with 45 nM Hoechst in Mowiol. For live imaging, bacteria were incubated with Rhodamine-Red (5 μg/ml) for 1 h at 37°C, subsequently washed with PBS twice and then added to the cells. In some live experiments Hoechst or TOPRO was added to the infection medium and visualized directly on the microscopy chamber.

Staphylococcus aureus strains wt, the mutant deficient for α-hemolysin Hla(–) or the Hla(+) mutant complemented with an α-hemolysin plasmid were grown at 37°C with shaking in 10 ml of LB broth with appropriate antibiotics to an optical density at 600 nm (OD600) of 2.0. After removal of the bacterial cells by centrifugation at 12,000 g for 15 min at 4°C, the obtained supernatants were filtered using a 0.22 mm Millipore filter. The resulting filtrate was used as supernatant for experiments. Supernatants were checked for absence of bacteria by cultivating small aliquots on LB agar. Protein concentration was determined by the Bradford assay.

CHO GFP-LC3 cells were incubated for 1 h with 10 μg/ml of purified α-hemolysin from S. aureus (Sigma). Live cells were analyzed by fluorescence microscopy.

CHO cells were analyzed by fluorescence microscopy using an Olympus Confocal FV1000. The program FV10-ASW 3.0 was used for all acquisitions and settings for the live imaging. For spinning disk acquisition, we used a Quorum Technologies WaveFX Confocal Microscope with a Yokogawa CSU-X1 Spinning Disc Unit and a LCI Temperature Controller and CO₂ Mixer for Live Cell Imaging (Neurobiology of Stress CNS, University of Toronto at Scarborough, Canada). The resulting movies series were corrected to improve the contrast and resolution using the deconvolution plugin on ImageJ. The image analysis of Saf growth and contraction was carried out using ImageJ.

For colocalization studies images were processed using the plugin Colocalization analysis from ImageJ. Images were deconvolved and then analyzed using Mander's Overlap Coefficient (MOC) and Pearson Correlation Coefficient (PCC) to determine the degree of overlap between images using the green and red channel. Data are representative of two different experiment (n = 20 cells/condition).

Cells were cultured in cover glasses of 25 mm of diameter and transfected when required. Infection with the S. aureus strains was performed as described above. After the proper infection time the medium was replaced by new infection medium and the cover glass was imaged at the required time-points after infection. The cover glasses were mounted on the microscope chamber equipped with a heating unit to maintain the temperature at 37°C.

CHO cells were infected for 2 or 4 h with S. aureus wt. In some experiments CHO cells stably expressing EGPF-Rab7 or EGFP-Rab7T22N were used or CHO cells were transiently transfected with GFP-Rab1b wt or GFP-Rab1b S22N. In all cases, after 1 h of infection cells were washed and gentamicin (100 μg/ml) was added for 30 min to kill extracellular bacteria. After the infection period (i.e., 2 h) cells were washed with PBS and lysed with sterile double distilled water for 10 min at room temperature. Lysates were diluted with PBS, plated in Brain Heart agar and incubated for 12 h at 37°C. Colonies were counted on the plate with dilutions yielding 50–100 visible colonies.

It has been previously demonstrated that S. aureus reside in an unique phagosomal compartment that poorly localizes with LAMP but is highly decorated by Rab7 (Schnaith et al., 2007; Seto et al., 2011). Both publications present some contradictory observations regarding the staining with lysosomal markers however both conclude that the S. aureus-residing compartment acquires Rab7. In order to analyze the sequence of events that lead to the recruitment of Rab7 to the S. aureus phagosome, CHO cells stably overexpressing GFP-Rab7 were infected with S. aureus wild type (wt) and analyzed at four different post-infection times (1, 2, 4, and 6 h; Figure 1A). From early post-infection times the recruitment of Rab7 to the S. aureus containing-phagosome was observed. Interestingly, at these post-infection times we also observed tubular like structures that were marked with Rab7. At later post-infection times (4 h.p.i) these tubular structures were no longer visible. Instead, infected CHO cells presented S. aureus containing-phagosomes in a “honeycomb-like” distribution. Interestingly, the late endosomal/lysosomal protein LAMP-1 was recruited to S. aureus phagosome but no tubular structures labeled with LAMP-1 were observed at 1–2 h post-infection (Figure S1A).

In order to analyze the dynamics of these structures we observed living cells using confocal microscopy. For this purpose, CHO cells stably overexpressing GFP-Rab7 were infected with S. aureus wt and living cells were imaged with confocal microscopy in a temperature controlled chamber (Figure 1B and Movie S1). We observed the formation of highly dynamic tubules that extended from the S. aureus containing phagosome, thus we named these tubular structures S. aureus-induced filaments (Safs). Quantification analysis indicates that approximately 70% of the infected cells showed one or more Safs at 2 h.p.i. (Figure 1C). We next analyzed time-lapse series of living CHO cells stably transfected with GFP-Rab7 and infected with S. aureus wt and a representative cell is shown in Figure 1D (see also Movie S1). We noted the formation of tubular GFP-Rab7 positive structures extending from the S. aureus phagosomes toward the cell periphery as early as 1 h after infection (Figure 1D). The number and the extension of these tubular structures increased over time. Closer analyses of the time-lapse movies indicated that individual Safs were highly dynamic showing not only extension but also contraction, branching or contact with another S. aureus-containing vacuoles. Linear growth of Safs was observed as well as apparently random changes in the direction of Saf extensions. As a control, uninfected CHO GFP-Rab7 living cells were observed to assure that the formation of the tubules was restricted to the infected cells (Figure S1B).

Rab proteins are small GTPases regulated by specific guanine nucleotide exchange factors (GEFs), which catalyze the displacement of GDP and binding of GTP, by GTPase activating proteins (GAPs) which stimulate the GTP hydrolysis (Novick, 2016). We therefore studied the involvement of a functional Rab7 in Saf generation and elongation. For this purpose, CHO cells stably overexpressing an EGFP wild-type Rab7 or the dominant-negative mutant (Rab7-T22N) were infected with S. aureus wt (Figure 1E). After 1 h post-infection in Rab7 wt expressing cells we observed Saf formation; however, in cells overexpressing Rab7-T22N no Rab7 tubular structures were observed (Figure 1F), indicating that an active Rab7 is required for the formation of the filamentous structures.

With the purpose of determining the importance of a functional Rab7 on S. aureus replication, CHO cells stably overexpressing GFP-Rab7 wt or GFP-Rab7 T22N were infected with S. aureus wt. After 1 h post-infection cells were washed with PBS and gentamicin (100 μg/ml) was added to kill extracellular bacteria. Subsequently, cells were incubated an additional hour (2 h.p.i) or 3 h (4 h.p.i) and finally were lysed with sterile water and plated on Brain Heart Agar for colony forming units (CFU) determination. As shown in Figure 1G, cells overexpressing Rab7-T22N presented a significant decrease in bacterial replication compared to cells overexpressing Rab7 wt. These results indicate that Rab7 is essential for S. aureus replication and suggest that Saf formation is also important for efficient bacteria replication.

The role of the GTPase Rab7 has been extensively investigated and it is well recognized to facilitate endosomal maturation, transport from the late endosome to the lysosome, and positioning of the endosomes and lysosomes by regulating their movement along the cytoskeleton (Wang et al., 2011). Nocodazole is a pharmacological inhibitor of microtubules (MT) that affects the cytoskeleton integrity. We tested the effects of nocodazole on the dynamic of Safs marked with Rab7 (Figure 2A); treatment with nocodazole did result in the loss of a large number of existing Safs (Figure 2B). It is important to take into account that cells were treated with nocodazole after 30 min of infection to guarantee a proper internalization of S. aureus, as microtubules and actin microfilaments are necessary for bacterial phagocytosis (Jevon et al., 1999; Shi and Zhang, 2012).

With the purpose of determining if the inhibition of Safs caused by nocodazole treatment affected bacteria replication we infected CHO cells with S. aureus wt for 30 min and then 2 μM of nocodazole was added. After 2 or 4 h post-infection cells were lysed and CFU was determined. As shown in Figure 2B a marked decrease in bacterial replication was observed in cells treated with the microtubule depolymerizing agent in contrast to control conditions (Figure 2B).

Microtubule-based motors include kinesins, which move toward the MT plus end, and dyneins that moves to the minus end (Akhmanova and Hammer, 2010). Kinesins are a family of molecular motors that use the energy of ATP hydrolysis to travel along the surface of microtubule filaments (Verhey and Hammond, 2009). In order to determine if Kinesin-1 is involved in Saf elongation, CHO cells stably expressing Rab7 were cotransfected with pcDNA-Kif5 wt or pcDNA-Kif5 T93N, a dominant negative mutant that has no ATPase activity. After protein expression cells were infected with S. aureus wt for 1 h and images from living cells were analyzed (Figure 2C). As shown in Figure 2C, when Kif5 wt was overexpressed we observed 70% of infected cells with Safs decorated with Rab7. In contrast, in cells overexpressing Kif5-T93N mutant Saf elongation was inhibited and no tubular structures marked with Rab7 were observed (Figure 2E). These results indicated that Saf formation and elongation requires kinesin-1 motor to extend toward the cellular periphery.

Rab7-interacting lysosomal protein (RILP) regulates microtubule minus-end directed transport by recruiting the dynein/dynactin motor complex (van der Kant et al., 2013). Thus, we transfected CHO cells stably overexpressing GFP-Rab7 with RFP-RILP wt or mCherry-RILPΔN, a truncated mutant with an N-terminal deletion that retains the ability to bind to active Rab7 but is unable to bind to microtubule motors (Jordens et al., 2001). Subsequently, cells were infected with S. aureus wt for 1 h and images of living cells were analyzed (Figure 2D). We observed colocalization of Rab7 and RILP in Safs membranes, suggesting that Rab7 is interacting with RILP to promote Saf elongation. Interestingly, we observed inhibition of Saf elongation when RILPΔN was used. Taken together these findings indicate that S. aureus uses the MT-motor kinesin-1 and the protein RILP of the host cell to promote Saf elongation.

Next we were interested in determining the nature of the tubular-like structures using different sub-cellular markers. To determine whether S. aureus-filaments are enriched in endosomal associated membrane proteins, CHO cells were transiently transfected with different Rab proteins (Rab1b, Rab5, Rab7, Rab11, Rab24, Rab29, Rab32, or Rab34) and infected with S. aureus wt. Living cells were analyzed after 1 h post-infection by fluorescence microscopy. Rab24 and Rab29 were not recruited to S. aureus-containing vacuoles neither to Safs whereas Rab5, Rab11, Rab32, and Rab34 were recruited to S. aureus phagosome but not to Saf membranes (Figure S2). To our surprise, as depicted in Figure 3A S. aureus phagosomes were decorated with Rab1b and they presented numerous tubular structures also marked by the protein in approximately 60% of the infected cells (Figure 3C). Analysis of video microscopy showed dynamic tubules marked with Rab1b emerging form S. aureus phagosomes that were able to branch, contract, elongate and reach other bacteria-containing phagosomes (Figure 3D and Movie S2). In order to study the participation of a functional Rab1b in Saf generation and elongation, CHO cells were transfected with EGFP-Rab1b S22N (a dominant negative mutant locked in a GDP-bound conformation) and infected with S. aureus wt. As show in Figure 3B cells overexpressing Rab1b S22N did not present Rab1b positive Safs after 1 h post-infection. Next, we analyzed the importance of Rab1b in S. aureus replication. Therefore, CHO cells overexpressing GFP-Rab1b wt or GFP-Rab1b S22N were infected with S. aureus for 2 or 4 h and CFU were determined. As shown in Figure 3E, cells overexpressing Rab1b S22N presented a significant decrease in bacterial replication compared to cells overexpressing Rab1b wt. These results indicate that Rab1b is present in Saf membranes and that the activity of this Rab GTPase is necessary for Saf formation and efficient bacteria replication.

We have previously demonstrated that Rab7 is a critical protein involved in the maturation of autophagic vacuoles and required for the progression of the pathway (Gutierrez et al., 2004). In addition, we have also shown that the small GTPase Rab1b is involved in the formation of autophagosomes (Zoppino et al., 2010). Therefore, we were next interested in determining whether the filamentous structures generated by S. aureus were related to the autophagy pathway by assessing the autophagic protein LC3. To analyze the presence of LC3 in Saf membranes CHO cells stably overexpressing GFP-LC3 were infected with S. aureus wt for 1 h and living cells were observed by confocal microscopy. As shown in Figure 4A we noted the formation of tubular GFP-LC3 positive structures extending from S. aureus autophagosomes (Movie S3). Time-lapse series show the extension of tubular membrane emerging form the autophagosome that contains S. aureus (Figure 4B). Quantification analysis indicates that approximately 35% of the infected cells showed one or more Safs marked with LC3 after 2 h.p.i. (Figure 4C). In order to analyze in more detail Safs structure, CHO GFP-LC3 cells were infected with S. aureus wt for 1 h and Z stacks were acquired with a confocal microscopy, images were processed with the ImageJ program, and a three dimensional reconstruction was performed (Figure 4D).

We next studied if Safs marked with LC3 corresponded to the same structures marked with Rab7 or Rab1b. For this purpose, CHO cells stably overexpressing EGFP-Rab7 were transfected with RFP-LC3 and infected with S. aureus wt for 1 h. Rab7 and LC3 showed a colocalization of 50% approximately, indicating that both proteins are recruited to the same Saf membrane. When cells were co-transfected with GFP-Rab1b and RFP-LC3, and subsequently infected with S. aureus wt for 1 h, we also observed a high degree of colocalization (90% approximately) of both proteins at the Saf membranes (Figures 5A,B). In contrast, in cells co-expressing LC3 and Rab7 T22N infected with S. aureus wt for 1 h no tubules labeled with LC3 were observed reflecting an impaired Saf biogenesis (Figure 5C). A similar effect was observed when CHO cells were cotransfected with RFP-LC3 and GFP Rab1b S22N and infected with S. aureus wt for 1 h (Figure 5D). Taken together, these results describe a particular phenotype induced by intracellular S. aureus with the formation of tubular membrane structures that recruit Rab1b, Rab7, and LC3 proteins and that depends on active conformation of the GTPases Rab1b and Rab7.

Staphylococcus aureus produces an extensive variety of exoproteins that are secreted in the last exponential phase of growth in a culture media (Dinges et al., 2000). In order to determine if the formation of tubular Safs is induced by an S. aureus secreted factor, we treated S. aureus with chloramphenicol (Cm), a bacteriostatic antibiotic that inhibits protein synthesis. As depicted in Figure 6A, in CHO cells stably expressing GFP-LC3 that were infected with S. aureus pre-treated with Cm, Saf formation was drastically reduced (only 7% of the cells presented tubules) compared to the control condition (40%). To corroborate that Saf formation is induced by a S. aureus secreted protein we treated CHO GFP-LC3 cells with different concentrations of filtered supernatant of a S. aureus culture (Figure 6B). Interestingly, cells treated with increasing concentrations of the supernatant (S) of S. aureus showed Saf formation after 30 min of incubation. We performed a dose curve of undiluted supernatant of a protein concentration of 34 μg/ml final volume and determined that 0.1 ml was sufficient to induce Saf formation (Figure 6B). As a control, sterile culture media was added to CHO cells in the same concentration and no Safs were observed (Figure S3A). Hence, we next tested the culture supernatant of three S. aureus strains: wild type, Hla (–) (a mutant deficient for α-toxin), and Hla (+) (a mutant complemented with the α-toxin-encoded plasmid). Strikingly, we observed that in the cells treated with the supernatant of the Hla (–) strain no Safs were visualized, in contrast the strains that produce α-toxin did induce tubular structures marked with LC3 (Figure 6C). To determine if this effect was also observed in cells overexpressing Rab7 or Rab1b, CHO cells stably expressing GFP-Rab7 or cells transfected with EGFP-Rab1b were treated with the supernatant obtained from the three strains: wt, Hla (–) and Hla (+) and we observed that in concordance with the results obtained above, Hla (–) supernatant did not induced Saf formation marked with Rab7 neither Rab1b. These findings strongly suggest that the α-hemolysin is the S. aureus protein responsible for Saf formation.

We have previously demonstrated that Hla secretion is necessary for autophagy activation and the process occurs through a PI3K/Beclin1-independent mechanism (Mestre et al., 2010). To assess the role of Hla in Saf formation, we tested the effect of the purified α-hemolysin in CHO cells stably expressing GFP-LC3 (Figure 7A). For this purpose, cells were treated with 10 μg/ml of the toxin for 1 h (Mestre et al., 2010) and living cells were analyzed by confocal microscopy. As depicted in Figure 7A, 35% of cells show tubular structures decorated with LC3 protein just by the addition of the purified toxin. Analysis of time laps series show dynamic Safs marked with LC3 induced by α-hemolysin (Figure 7B). To rule out that Saf formation could be an effect of the activation of the autophagy pathway CHO GFP-LC3 cells were incubated with starvation media and no Safs were visualized (Figure S3A). It has been previously shown that alpha hemolysin is recognized, in epithelial cells, by it superficial receptor ADAM10 (Wilke and Bubeck Wardenburg, 2010). Indeed, ADAM10 is internalized by clathrin-dependent endocytosis specifically interacting with AP2 in neuronal cells (Marcello et al., 2013). Taking these findings into account, we next analyzed if the induction of Saf formation by Hla is affected when clathrin-dependent endocytosis is inhibited. Thus, we treated CHO cells stably expressing GFP-LC3 with dynasore, a GTPase inhibitor that targets dynamin (Macia et al., 2006) or PitStop 2, a clathrin terminal domain (TD) inhibitor (von Kleist et al., 2011) and then added purified Hla. As shown in Figure S3B no Saf formation was observed when cells were treated with the inhibitors. This result clearly indicates Hla induces tubular compartments only when is endocytosed by a clathrin-dependent process.

Previously we have reported that cAMP and the effector proteins, EPAC (exchange protein activated by cAMP) and Rap2b, inhibit the autophagic response induced by α-hemolysin (Mestre and Colombo, 2012). Hence, we next assessed the effect of wortmannin (PI3K inhibitor) and dbcAMP (a membrane permeable cAMP analog) on the formation of Safs (Figure 7C). As expected wortmannin had no effect, on the contrary, dbcAMP inhibited the formation of S. aureus filaments induced by the bacteria or by the added supernatant. Thus, the inhibition of the autophagic response generated by Hla seems to be sufficient to inhibit the formation of LC3 tubular structures. In addition, we analyzed the effect of dbcAMP on Safs labeled with Rab7 or Rab1b, for this purpose CHO cells stably expressing GFP-Rab7 or GFP-Rab1b were treated with 1 mM of dbcAMP for 30 min and then infected with S. aureus wt. After 1 h post-infection living cells were analyzed by confocal microscopy and no Safs were visualized (Figure 7D). These findings strongly indicate that cAMP is a key regulator of the S. aureus induced filaments and that they are related to the autophagy pathway induced by the toxin.

Next, we performed live cell fluorescence microscopy using a spinning disk microscope to follow living cells infected with S. aureus for prolonged periods of time (i.e., up to 4 h). CHO GFP-LC3 cells were infected with S. aureus wt for 30 min [indicated as time point 0 (00:00:00 s)], to allow bacteria internalization, and then cells were plated in a specific chamber at 37°C with 5% of CO_2. Images were acquired every 30 s during 4 h. As depicted in Figure 8A, at initial times we observed LC3 recruited to S. aureus phagosomes and after 10 min approximately we visualized Saf formation emerging from the phagosome. This phagosome after 15 min showed another Saf that contacted a neighbor phagosome. As visualized on Figure 8A and Movie S4 after 40 min more Safs were visualized extending in different directions and continued until 1 h and 50 min of infection approximately. The S. aureus containing-phagosome was observed during the next 2 h and no Saf formation was detected until the end of the movie (4 h).

Staphylococcus aureus scape from the phagosome depends on bacterial virulence factors, such as toxins, and it has been described to occur between 3 and 6 h post-infection (Giese et al., 2011; Münzenmayer et al., 2016). However, there is a discrepancy whether the bacteria only replicates in the cytoplasm (Grosz et al., 2014) or in the phagosomal compartment (Schröder et al., 2006; Kubica et al., 2008; Flannagan et al., 2016). Here we have observed bacteria replication within the phagosomal compartment that shows Saf formation at early infection times (Figure 8). Taken together these results describe a particular intracellular phenotype induced by S. aureus characterized by the formation of dynamic tubular membrane structures labeled with Rab1b, Rab7, and LC3 proteins that occur at early infection times (30 min–2 h post-infection) which are induced by α-hemolysin and regulated by cAMP levels.

The 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. The reviewer RF and handling Editor declared their shared affiliation.