Chlamydia trachomatis infection prevents front–rear polarity of migrating HeLa cells

Authors


For correspondence. E-mail heuerd@rki.de; Tel. (+49) 30 18754 2967; Fax (+49) 30 18754 2328.

Summary

Chlamydiae are obligate intracellular bacterial pathogens that cause trachoma, sexually transmitted diseases and respiratory infections in humans. Fragmentation of the host cell Golgi apparatus (GA) is essential for chlamydial development, whereas the consequences for host cell functions, including cell migration are not well understood. We could show that Chlamydia trachomatis-infected cells display decelerated migration and fail to repopulate monolayer scratch wounds. Furthermore, infected cells lost the ability to reorient the fragmented GA or the microtubule organization centre (MTOC) after a migratory stimulus. Silencing of golgin-84 phenocopied this defect in the absence of the infection. Interestingly, GA stabilization via knockdown of Rab6A and Rab11A improved its reorientation in infected cells and it was fully rescued after inhibition of Golgi fragmentation with WEHD-fmk. These results show that C. trachomatis infection perturbs host cell migration on multiple levels, including the alignment of GA and MTOC.

Introduction

Chlamydia trachomatis is an obligate intracellular human pathogen and responsible for the highest burden of sexually transmitted diseases caused by bacteria. While serovars A–C infect the ocular epithelium, serovars D–K establish infections of the urogenital tract. Serious pathological outcomes, including blindness and female infertility, are often associated with chronic or reoccurring infections that promote fibrosis and scarring of infected tissue (Schachter, 1999; Wyrick, 2010). Besides their medical impact, Chlamydia spp. serve as ideal model organisms to study bacteria–host interactions; they invade a variety of different mammalian cell types and extensively reorganize the host cell to establish an intracellular niche and support their replication.

Chlamydia spp. display a unique biphasic cycle of development alternating between two functionally and morphologically distinct forms: the infectious but metabolically inactive elementary body (EB) and the non-infectious, metabolically active reticulate body (RB). The RBs replicate inside the cell within a protective membrane-bound compartment, the inclusion, that allows uptake of essential lipids from the host cell, e.g. via the interception of vesicles from the Golgi apparatus (GA) (Hackstadt et al., 1995; 1996; Carabeo et al., 2003).

The GA of mammalian cells characteristically consists of laterally linked stacks that form the Golgi ribbon (Barr and Warren, 1996). Structural Golgi matrix proteins of the golgin family are necessary for the maintenance of this interconnected and dynamic architecture (Barr and Short, 2003). Depletion of certain golgins disrupts the ribbon and results in small individual membrane ministacks (Short and Barr, 2003; Ramirez and Lowe, 2009). In particular, golgin-84 was shown to be a key factor for Golgi ribbon integrity, likely connecting the rims of cisternae (Diao et al., 2003). Chlamydia-infected cells show a dramatic fragmentation of the GA into functional ministacks that involves golgin-84 (Heuer et al., 2009). Furthermore, this process requires the small GTPases, Rab6A and Rab11A, which control cellular vesicle transport (Rejman Lipinski et al., 2009). Interestingly, during infection Golgi fragmentation occurs simultaneously with secretion of the chlamydial protease-like activity factor (CPAF) into the cytosol and can be induced by ectopic expression of CPAF in uninfected cells (Christian et al., 2011).

In addition to the GA, Chlamydia also targets the cytoskeleton at different time points during the developmental cycle. In the early phase of infection, C. trachomatis induces local actin polymerization at attachment sides via Rac1 activation and the nucleator complex Arp2/3 (Carabeo et al., 2002). Secretion of the type three effector protein Tarp promotes actin polymerization to facilitate EB uptake (Clifton et al., 2004). Later, C. trachomatis remodels actin and intermediate filaments (IF) into a cage, providing mechanical stability to the growing inclusion (Kumar and Valdivia, 2008b).

While the relevance of these profound restructuring events for the progressing infection has been investigated, the consequences for the host cell remain largely unknown. Disruption of the Golgi ribbon could affect fundamental Golgi-related aspects of eukaryotic cells, for example polarity and migration (Yadav et al., 2009).

Directional cell migration is of high physiological importance during many biological processes, e.g. embryonic development and chemotaxis (Ridley et al., 2003; Iden and Collard, 2008; Kurosaka and Kashina, 2008). Furthermore, epithelial regeneration, wound healing and immune surveillance strongly depend on migration processes (Hackam and Ford, 2002; Gurtner et al., 2008). Alterations in epithelial cell motility have been linked to the spread of infection and reduced tissue repair during viral infections (Morales et al., 2008; Spurzem et al. 1995). It has been shown previously that Chlamydia infection can block cell migration in an in vitro wound healing assay (Kumar and Valdivia, 2008a). However, the functional consequences of Chlamydia-induced remodelling of the GA remain elusive.

The highly orchestrated process of cell migration is still not fully understood. Previous work has established that correct orientation of the GA, and actin polymerization, at the leading edge after a migratory stimulus are crucial for cell polarity and motility (Lauffenburger and Horwitz, 1996; Pollard and Cooper, 2009). Reorientation of the GA towards the forward migrating edge depends on ribbon integrity and Golgi-derived microtubules (Vinogradova et al., 2012) and is speculated to facilitate directed secretion (Bergmann et al., 1983). A recent study has revealed that golgin-160 and GMAP210 confer membrane motility (Yadav et al., 2009). Knockdown of these golgins dispersed the GA and prevented repositioning, which ultimately reduced cellular motility.

In this study, we investigated the role of Chlamydia infection on host cell polarization during migration. We show that C. trachomatis infection blocks reorientation of the GA and the MTOC in migrating cells down-stream of Rac1 and Cdc42 activation. Golgi ministacks and MTOC remain closely associated with the inclusion and do not reposition towards the leading edge of migrating cells. This polarization defect partially depends on Rab11A and Rab6A, the two GTPases essential for Chlamydia-induced and golgin-84-dependent Golgi fragmentation and can be reverted completely by treatment with WEHD-fmk, an inhibitor blocking Golgi fragmentation and CPAF activity (Christian et al., 2011). Interestingly, in migrating golgin-84-knockdown cells, disturbance of GA positioning is rescued completely by co-silencing of Rab11A and Rab6A. Thus, our findings show that Chlamydia infection affects cell polarization on different levels including Golgi positioning. This leads to the infected cell being non-responsive to initial front-to-back polarization and showing disturbed cell migration.

Results

Chlamydia trachomatis infection blocks cell migration

We analysed the effect of C. trachomatis infection on cell migration using an in vitro scratch assay. Confluent HeLa cell monolayers were infected with C. trachomatis L2 for 24 h at different multiplicities of infection (moi) or left uninfected, and then ‘wounded’ by introducing a linear scratch using a sterile pipette tip. Migration was monitored using light microscopy at 0 h and 24 h post scratching. Uninfected control cells completely repopulated the scratch area within 24 h, whereas infected cells failed to close the wound during the same time frame (Fig. 1A). This effect was moi-dependent, since cells infected with a moderate infectious dose, e.g. moi 1, partly closed the wound. Intriguingly, in this case, mainly the uninfected cells migrated into the scratch area (Fig. 1B). Quantification confirmed these observations, revealing that wound closure was inhibited by 30% and 70% at moi 1 and moi 5, respectively, in comparison to uninfected cells (Fig. 1C). A time-course experiment showed that inhibition of migration increased with progression of the infectious cycle (Fig. 1D).

Figure 1.

Chlamydia infection blocks host cell migration.

A. In vitro scratch assay of HeLa cells infected with indicated moi of C. trachomatis. Twenty-four hours post infection, the cell monolayer was scratched to induce migration using a sterile pipette tip. Migration was monitored during the next 24 h and microscopic images were acquired at defined positions. Scale bar: 100 μm.

B. Enlargement showing the border region of the monolayer (dashed line) 24 h after infection with a moderate infectious dose (moi 1). Inclusions are indicated by asterisks. Scale bar: 100 μm.

C and D. (C) Remaining scratch area after 24 h or (D) after scratching at different time points post infection, normalized to t = 0 h. Cells were infected with indicated moi for 24 h (C) or moi 3 for 12, 16 or 24 h (D) or left uninfected for 24 h. Monolayers were then scratched and the remaining cell-free area was quantified after 24 h of migration. Error bars: SEM.

These data clearly demonstrate that cell migration is impaired in C. trachomatis-infected cells in a dose- and time-dependent manner.

Infected cells display reduced migration velocity

To obtain better temporal resolution of the migration process, phase-contrast live cell microscopy was performed for 16 h with infected and control HeLa cells. During this time period, the cells were tracked in 4 min intervals. As expected, uninfected control cells persistently migrated into the gap with a mean velocity of approximately 10 μm h−1 (Fig. 2A and C). In striking contrast, C. trachomatis-infected cells moved significantly slower (2 μm h−1, Fig. 2B and C). As a consequence, infected cells failed to close the scratch wound in the indicated time frame.

Figure 2.

Live tracking of infected cells reveals reduced migration velocity. HeLa cells were seeded into cell culture inserts with two opposing chambers separated by a 500 μm gap. To start migration, the chambers were removed 24 h p.i. Migration of individual uninfected cells (A) or cells infected with moi 3 (B) was tracked until full scratch closure by control cells (16 h post insert removal). Time-lapse images were acquired every 4 min. Scale bar: 50 μm. (C) Quantification of mean migration velocity. Migration speed of individual cells was analysed measuring the distance covered between time points. Data were taken from at least nine individual cells per movie. Error bars: SD. ***P < 0.001.

Thus, C. trachomatis infection reduces the velocity of cell migration, suggesting that the infection alters sensing of the scratch wound, cytoskeletal rearrangements and/or polarization of infected host cells.

Rho GTPase-induced protrusions are formed in infected migrating cells

The Rho GTPases, Rac1, Cdc42 and RhoA cycle between a GDP-bound (inactive) state and a GTP-bound (active) state and thereby control many migratory processes. The first two are essential for the formation of lamellipodia and filopodia (Jaffe and Hall, 2005). In addition, Cdc42 is crucial for the establishment of cell polarity and important for directional cell migration (Stowers et al., 1995; Etienne-Manneville and Hall, 2001; 2003). RhoA, on the other hand, promotes migration by limiting membrane protrusions at the trailing edge and is required for tail retraction (Worthylake et al., 2001; Worthylake and Burridge, 2003). To test whether infected cells still respond to migration signals induced by scratch wounding, we analysed the activation status of the Rho GTPases Rac1, Cdc42 and RhoA in infected and control cells.

Twenty-four hours post infection (p.i.) active Rac1, Cdc42 and RhoA were precipitated from infected and control cells via PAK1-PBD (Cdc42, Rac1) or Rhotekin-RBD (RhoA) bound to agarose beads and analysed by Western blotting using specific antibodies. Surprisingly, in C. trachomatis-infected cells the amount of GTP-bound Rac1 and Cdc42 was considerably higher compared with uninfected control cells, while the active form of RhoA did not increase significantly (Fig. 3A). The same observation was made in cells that had not been wounded (Fig. S2).

Figure 3.

Rho family GTPase-dependent membrane protrusions are formed in infected cells. Infected HeLa cells (moi 3) were examined for Rho family GTPase-dependent migration responses.

A. Co-precipitation of active Rho family GTPases. Infected and uninfected HeLa cell layers were scratched multiple times (three directions, 15 times each, per 10 cm dish) 24 h p.i. and harvested 90 min after scratching. GTP-bound Rac1, Cdc42 and RhoA were co-precipitated from whole cell lysates using PAK1-PBD agarose beads (Rac1, Cdc42) or GST-Rhotekin-RBD beads (RhoA).

B. Membrane protrusions of border cells. Twenty-four hours post infection, migration was initiated by scratch and cells were fixed after another 6 h. Scale bar: 20 μm.

C. Quantification of leading edges (distinct hem-like actin polymerization along the frontal cell margin). Cells were fixed 6 h post scratching and actin was visualized by fluorescent phalloidin to score the percentage of cells with a visibly developed leading edge via confocal microscopy. Error bars: SEM.

We addressed the formation of lamellipodia to analyse how C. trachomatis infection blocks cell migration down-stream of these GTPases. Cells were scratched and fixed after 6 h. Lamellipodia were visualized by phase-contrast microscopy, while leading edge formation (visible hem-like actin polymerization at the cell front) was detected with fluorescent phalloidin. Lamellipodia developed normally in infected and control cells and leading edges were formed to the same extent (Fig. 3B and C).

These data demonstrate that in infected cells, Rac1 and Cdc42 are activated, while RhoA activity is not affected. Thus, in C. trachomatis infected cells, Rho GTPase activity is altered, but not blocked.

Cell polarization is inhibited in Chlamydia-infected migrating cells

Directional cell migration is known to depend on correct cell polarization. In migrating fibroblasts, the GA and MTOC is positioned in front of the nucleus to face the leading edge and this is controlled by Cdc42 (Kupfer et al., 1982; Gomes et al., 2005). To assess whether Chlamydia infection interferes with cell polarization we determined GA and MTOC positioning in infected and control cells during cell migration. Cells were seeded into small chambers separated by a 500 μm gap and then infected with C. trachomatis. After 12, 16 and 24 h of infection, chambers were removed to allow migration of cells into the gap. Cells were fixed after another 6 h and stained to detect GA, MTOC and nucleus by fluorescence microscopy analysis. In uninfected cells, the GA was positioned in front of the nucleus facing the leading edge during the first 6 h of migration (Fig. 4A, Figs S1 and S3A). In infected cells, the GA failed to position in front of the nucleus (Fig. 4A, Fig. S3B). GA fragments were found around the bacterial inclusions and could even be detected close to the inclusions behind the nucleus in relation to the migration front starting from 16 h p.i. Quantification revealed that in the majority of control cells (80%) the GA faced the front of the migrating cells (Fig. 4B, Fig. S3C). After infection, less than 30% of the cells exhibited a normal GA positioning. Infected cells also displayed reduced MTOC positioning during migration in a time-dependent manner (Fig. 4C). Taken together, Chlamydia infection blocks cell polarization during the initial phase of migration.

Figure 4.

Golgi apparatus and MTOC reorientation is perturbed in Chlamydia-infected migrating cells. (A) Reorientation of the Golgi apparatus (GA) in cells infected for different time periods. Cells were infected with moi 3 for 12, 16 or 24 h or left uninfected for 24 h. Following induction of migration by scratch, cells were fixed after 6 h and the GA was visualized via immunostaining. Red: giantin. Scale bar: 20 μm. (B) Relative amount of cells with correctly oriented GA or (C) MTOC, scored by means of orientation within 120° facing the scratch. Error bars: SEM. *P < 0.05; **P < 0.01.

Depletion of golgin-84 prevents correct Golgi positioning during cell migration

Chlamydia infection induces Golgi fragmentation which has been linked to golgin-84 (Diao et al., 2003; Heuer et al., 2009). To elucidate the contribution of golgin-84 and golgin-84-dependent Golgi fragmentation to correct Golgi positioning, we directly compared polarization of stable golgin-84 and control knockdown cells in a Golgi reorientation assay. Control cells exhibited correct Golgi positioning during the first 6 h of migration (Fig. 5A); by contrast, stable golgin-84 knockdown cells failed to reorient the GA towards the leading edge (Fig. 5B). After 6 h, 80% of the control knockdown cells were polarized, whereas only 30% of golgin-84 knockdown cells exhibited a correctly oriented GA (Fig. 5C). MTOC relocation was also dependent on the presence of golgin-84 (Fig. 5D and E). These results demonstrate that correct cell polarization during migration strongly depends on Golgi structure, which in turn is stabilized by golgin-84. Consistently, the migration of golgin-84-depleted cells into scratch wounds was significantly reduced (Fig. 5F and G). Thus, the polarization defect and parts of the motility defect observed in Chlamydia-infected cells can be phenocopied by golgin-84 knockdown cells. These results indicate that the modulation of golgin-84 during infection affects polarization and contributes to the migration defect.

Figure 5.

Knockdown of golgin-84 reduces migration in uninfected cells and prevents anterior-posterior cell polarization. Reorientation of the GA (A) and MTOC (D) during the polarization of uninfected HeLa cells stably transfected with Luciferase control shRNA or golgin-84 shRNA. Cells were cultured in cell culture inserts. After insert removal, cells were allowed to initiate migration and fixed after 6 h. The position of the GA was determined via immunostaining. Correct reorientation of the GA (C) or MTOC (E) was defined as localized within a 120° angle facing the border (indicated by dashed line). Red: giantin (A, B) or pericentrin (D), blue: DRAQ5 DNA staining. Error bars: SEM. Scale bar: 20 μm. *P < 0.05; **P < 0.01; ***P < 0.001. (F) Migration of golgin-84 depleted cells. Scratch wound assay comparing the migration of a HeLa cell line stably depleted for golgin-84 (sh golgin-84) and control cells (sh Luci). Cells were scratched and allowed to migrate for 20 h. Scale bar: 100 μm. (G) Quantification of remaining scratch area after 20 h of migration. Error bars: SEM. *P < 0.05.

Golgi apparatus stabilization improves its reorientation in infected cells

To elucidate the impact of Golgi fragmentation on cell polarization in infected cells, Golgi fragmentation was inhibited by WEHD-fmk or Rab GTPase knockdown. In contrast to untreated cells, WEHD-fmk-treated cells displayed a compact GA in spite of infection and could relocate the GA towards the leading edge (Fig. 6A and B). Unfortunately, the inhibitor was not feasible for migration studies (Fig. S4).

Figure 6.

Golgi apparatus stabilization improves Golgi reorientation in infected cells.

A. GA reorientation of border cells in the presence of WEHD-fmk. Cells were cultured and infected in culture inserts. The inhibitor (80 μm) was added 8 h p.i. and migration was induced by insert removal 24 h p.i. Cells were fixed for immunostaining 30 h p.i. Red: giantin, green: chlamydial Hsp60. Scale bar: 20 μm.

B. Quantification of uninfected (− L2) and infected (+ L2) cells with a correctly reoriented GA, 6 h after insert removal. Error bars: SEM. ***P < 0.001.

C. HeLa cells with a stable double knockdown of Rab6A and Rab11A were infected with moi 3 (+ L2) and migration was induced by scratch 24 h p.i. Golgi orientation was assessed 6 h post scratching via immunostaining. Red: giantin, blue: chlamydial Hsp60. Scale bar: 20 μm.

D. Quantification of cells with a correctly positioned Golgi apparatus. Error bars: SEM.

In addition, Rab6A and Rab11A were depleted using RNA interference and polarization of the GA was scored 6 h post induction of migration by scratch. As shown previously, fragmentation of the GA by silencing of golgin-84 was reverted by knockdown of either Rab6A or Rab11A (Rejman Lipinski et al., 2009, Fig. S5). Silencing of Rab6A or Rab11A furthermore completely rescued GA polarization after golgin-84 knockdown (Fig. S5A and B). Interestingly, in infected cells, knockdown of Rab11A and Rab6A only partially restored orientation of the GA towards the leading edge although Chlamydia-induced GA fragmentation was fully inhibited (Fig. 6C and D). Interestingly, migration of infected cells was partially rescued by Rab6A and Rab11A silencing (Fig. S6A and B). These data indicate that during Chlamydia infection, an intact GA structure is necessary for cell polarization and suggests that infection blocks cell migration on multiple levels including cell polarization.

Discussion

After infection, C. trachomatis dramatically rearranges the architecture of the host cell by secreting bacterial effectors including Tarp and CPAF into the host cytoplasm (Zhong et al., 2001; Carabeo et al., 2002; Mehlitz et al., 2008; 2010; Kumar and Valdivia, 2008a; Christian et al., 2011). The roles of some of the secreted effectors in chlamydial development as well as their molecular modes of action have been extensively studied. However, the broad consequences of interfering with the cytoskeleton and cellular organelles for higher order cellular functions, including cell motility, are currently not well understood.

In this work, we confirmed and further characterized that infection blocks cell migration (Kumar and Valdivia, 2008a). Live cell microscopy of infected and control cells revealed that the velocity of motile epithelial cells is reduced upon C. trachomatis infection. We further demonstrated that relocation of the GA and MTOC towards the leading edge is inhibited in infected cells. This polarization defect is caused down-stream of Rac1 and Cdc42 activation and is reverted by WEHD-fmk treatment. Interestingly, knockdown of Rab6A and Rab11A partially restored Golgi reorientation and cell migration in infected cells. Our data show that C. trachomatis infection prevents GA and MTOC positioning, a prerequisite for directional cell migration, by Golgi fragmentation.

Cell migration is a complex process that requires activation of Rho GTPases, actin polymerization and microtubule-dependent positioning of the GA and centrosome. The Rho GTPases Rac1, RhoA and Cdc42 control these processes during early stages of cell migration by spatially regulated signalling.

Rho GTPases and the actin cytoskeleton are targeted by C. trachomatis at important stages of the chlamydial cycle. Early in infection, C. trachomatis secretes type three effectors, including Tarp, that promote invasion of the pathogen into host cells. After translocation, Tarp is phosphorylated by multiple cellular kinases and thereby activates Rac1 via Sos and Vav2 (Lane et al., 2008; Mehlitz et al., 2008). Activation of Rac1 seems to be transient because another bacterial effector, CT166, can glycosylate and inactive Rac1 (Thalmann et al., 2010). At later time points of the infection, the C. trachomatis inclusion is encased in an actin scaffold. Formation of this cage is RhoA-dependent and stabilizes the inclusion (Kumar and Valdivia, 2008b). These Chlamydia-induced changes to the host actin cytoskeleton have been postulated to be responsible for defects in cell migration of infected cells (Kumar and Valdivia, 2008a).

Interestingly, we demonstrated that at 24 h p.i. the two GTPases, Rac1 and Cdc42, are activated and actin-dependent protrusions are still formed in infected migrating cells. These observations suggest C. trachomatis infection blocks cell migration by targeting down-stream functions of Rho GTPases other than actin polarization at the leading edge. However, global activation of migration-associated GTPases, as observed here, could just as well negatively influence motility. In epithelial cells, a gradient of active Cdc42-GTP at the cell front and inactive Cdc42-GDP at the back provides for the required asymmetry (Machacek et al., 2009) and constitutively active Cdc42 has been previously reported to hinder microtubule organising centre polarization (Etienne-Manneville and Hall, 2001; 2002). It needs to be investigated whether the overall activation of the Rac1 and Cdc42 leads to deregulated, or erroneous, recruitment of effectors and polarity proteins that normally accumulate the leading edge after local GTPase activation.

Based on our findings, we propose that C. trachomatis infection blocks polarization of migrating host cells and thereby eventually affects cell migration. Stabilization of the GA in infected cells by knockdown of Rab6A and Rab11A partially restored Golgi positioning and it was fully rescued by WEHD-fmk treatment. Fragmentation of the Golgi ribbon after silencing of golgin-84 in uninfected cells phenocopied the Golgi orientation defect, which supports the hypothesis that Golgi integrity plays an essential role in cell polarization (Yadav et al., 2009). However, we only observed a partial motility reduction after golgin-84 knockdown and stabilization of the GA in infected cells by Rab knockdown only partially rescued cell migration. This indicates that Chlamydia inhibits migration on additional levels, beyond loss of Golgi integrity and cell polarization. However, it remains to be determined which additional factors affect cell motility of infected cells.

For example the distribution of IF, including vimentin and keratin-18, likely contributes to this process. Vimentin interacts with the GA and is recruited to the chlamydial inclusion (Gao and Sztul, 2001; Kumar and Valdivia, 2008b) and it also has previously been found to be important for cell polarization and migration in vitro and in vivo. For example, fibroblasts from vimentin null mice display reduced wound healing and cell migration (Eckes et al., 1998; 2000). Furthermore, silencing of vimentin in MDCK cells decreased relocalization of the GA towards the leading edge (Osmani et al., 2006; Phua et al., 2009).

The data presented here strongly speak for a role of Golgi structure and CPAF activity in polarization during cell migration. We are aware that the extent of CPAF-mediated cleavage in infected cells and/or even the identity of golgin-84 as a CPAF substrates is currently questioned although Chlamydia-induced Golgi fragmentation has been documented by different groups (Chen et al., 2012). Furthermore, the functional link between golgin-84, Chlamydia-induced Golgi fragmentation, cell polarization and migration, all being sensitive to WEHD-fmk treatment, supports the view that CPAF is involved in these processes in vivo.

Chlamydial pathology is characterized by fibrosis and scarring of infected tissues, a process that is thought to be a consequence of a prolonged immune reaction. The identification of cell polarization defects in infected cells could have implications for chlamydial pathology especially in persistent infections, as cell polarization and migration are essential for wound healing. Unfortunately, treatment of uninfected cells with WEHD-fmk inhibited cell migration in a dose-dependent manner preventing us from addressing the consequences of Golgi stabilization on cell migration of infected cells (Fig. S4). The identification of Rab6A and Rab11A as important players in blocking cell migration of infected cells is an interesting start to better understand the underlying mechanisms of Golgi orientation and cell migration during Chlamydia infection, which may enhance our perception of how changes in cell architecture could contribute to Chlamydia pathogenesis.

Experimental procedures

Generation of stable cell lines

Stable golgin-84, Rab6A/Rab11A and luciferase knockdown cell lines were created via lentiviral transduction of a shRNA cassette as described previously (Heuer et al., 2009; Rejman Lipinski et al., 2009). The lentiviral vector pLVTHM and the packaging vectors pMD2G and psPAX2 were a generous gift from D. Trono, Lausanne, Switzerland.

Cell culture and infection

HeLa cells and stable HeLa knockdown cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen, Carlsbad, USA) supplemented with 10% heat-inactivated FCS, 1 mM sodium pyruvate and 2 mM l-glutamine at 37°C in a humidified incubator containing 5% CO2. C. trachomatis serovar L2 was propagated in HeLa cells. Infection was performed in DMEM supplemented with 5% FCS (infection medium). Cells were washed twice with infection medium and the inoculum containing the desired number of infectious particles per cell (moi, multiplicity of infection) was added (if not designated otherwise, the moi was 3). Cells were then incubated for 2 h at 35°C and 5% CO2. After washing, cells were kept in infection medium at 35°C.

Antibodies and dyes

Antibodies were obtained from the following sources: mouse anti Rac1 and mouse anti Cdc42: provided with the RhoA/Rac1/Cdc42-Activation Assay Combo Kit by Cell Biolabs, San Diego, USA. Rabbit anti RhoA: provided with the Active Rho Pull-Down and Detection Kit by Thermo Fisher Scientific, Rockford, USA. Rabbit anti giantin: Covance, Princeton, USA. Mouse anti chlamydial Hsp60: Alexis Biochemicals, Farmingdale, USA. Rabbit anti Pericentrin: Abcam, Cambridge, UK. Secondary antibodies conjugated to Cy2 and Cy3: Molecular Probes, Invitrogen, Carlsbad, USA. Secondary antibodies conjugated to horseradish peroxidase: Amersham, GE Healthcare, Piscataway, USA. DNA dye: DRAQ5 (Biostatus Ltd, Leicestershire, UK) or DAPI (Invitrogen, Life Technologies, Grand Island, USA). Actin staining: Alexa Fluor 647-phalloidine (Invitrogen, Life Technologies, Grand Island, USA).

In vitro scratch assay

Cells were seeded into 6-well plates at a density of 1.5 × 105 cells ml−1 (corresponding to 3 × 105 cells per well) and infected the next day with C. trachomatis at the indicated moi. Wounding was performed as previously described (Liang et al., 2007). Briefly, after reaching confluency, the monolayer was scratched multiple times with a sterile pipette tip. Detached cells were washed away with medium and fresh medium was added. Microscopic images were acquired at defined positions after 0 and 24 h of migration using an Olympus light microscope with 10× magnification and a camera with Visicapture software. Analysis of the scratch area was performed with an ImageJ macro containing the ‘analyse particles’ function. The unpopulated area was selected via a fixed threshold (isolating pixels with grey values between 0 and 30 in greyscale microscopic images) and quantified in relation to the original scratch area.

Live cell microscopy and tracking

Cells were seeded into live cell microdishes containing silicone cell culture inserts with two opposing chambers (ibidi, Martinsried, Germany). Live cell imaging of migrating cells was performed using an Olympus IX81 microscope equipped with a climate chamber and using 10× magnification (Olympus, Hamburg, Germany). All samples were treated with 50 ng ml−1 epidermal growth factor (EGF) after chamber removal to globally accelerate migration. Phase-contrast images were acquired every 4 min for 16 h using a C9100-02 CCD camera (Hamamatsu, Bridgewater, USA). Movie processing was done using Metamorph software (Molecular Devices, Sunnyvale, USA). Tracking of individual cells and determination of mean velocity was performed manually in representative movies with the ‘manual tracking’ function of the ImageJ software. A minimum of nine border cells was analysed for each condition.

Assessment of protrusion formation and leading edge quantification

Cells were grown in ibidi cell culture inserts on coverslips and infected the next day. Twenty-four hours post infection, the inserts were removed to start migration. After 6 h, cells were fixed in PBS, 2% PFA and protrusions were detected by confocal microscopy. The percentage of established leading edges was quantified microscopically after actin staining with Alexa Fluor 647-phalloidine (Invitrogen, Life Technologies, Grand Island, USA). A leading edge was defined by a visible rim of polymerized actin at the cell front.

Detection of active Rho GTPases

Precipitation of GTP-bound Rac1 and Cdc42 was performed with the Rho/Rac/Cdc42 Activation Assay Combo Kit by Cell Biolabs, San Diego, USA. Cells were seeded into 10 cm cell culture dishes and infected with C. trachomatis (moi 3). Twenty-four hours post infection, cell layers were scratched multiple times vertically and horizontally, as well as diagonally with a sterile pipet tip and detached cells were washed away with medium. Cells were allowed to migrate for 1.5 h before harvesting on ice. Cell lysis and affinity precipitation were performed according to the manufacturer's instructions. Briefly, cells were lysed for 20 min in the dish and the lysate was passed through a 26G syringe needle. Cleared lysates were loaded onto PAK-1 PBD beads and incubated for 1 h at 4°C. After washing, proteins were eluted by boiling in 6× SDS PAGE-sample buffer and subjected to gel electrophoresis followed by Western blot analysis. GTP-bound RhoA was precipitated using the Active Rho Pull-Down and Detection Kit by Thermo Fisher Scientific, Rockford, USA. Samples were treated as described above. Cells were lysed for 5 min and cleared lysates were loaded onto glutathione agarose resin together with purified GST-Rhotekin Rho-binding domain (RBD) and incubated for 1 h at 4°C. All three GTPases were detected via Western blot with specific antibodies supplied with the kits.

Golgi apparatus/MTOC repositioning assay and immunostaining

Cells were seeded into silicone cell culture inserts with two opposing chambers (Ibidi GmbH, Martinsried, Germany) with a density of 1.5 × 104 cells per ml in a total volume of 70 μl (1000–1200 cells) per chamber and infected the next day with C. trachomatis (moi 3). The insert was removed 24 h p.i. and 1 ml fresh infection medium was added. Removal of inserts resulted in two patches of cell monolayer divided by a cell-free area of 500 μm. After the indicated time points, cells were washed with PBS and fixed for immunostaining with 2% PFA for 30 min at room temperature. Fixed cells were washed with PBS and permeabilized for 30 min with 0.2% TX-100 in PBS and 0.2% BSA. Samples were washed with PBS and incubated with primary antibodies for 1 h at room temperature. After washing, the cells were exposed to secondary antibodies and DNA staining, washed and mounted in Mowiol. Images were acquired using a Leica TCS-SP confocal microscope (Leica, Solms, Germany) and a Zeiss LSM 780 confocal microscope (Zeiss, Jena, Germany) and processed with Adobe Photoshop 11.0. Figures showing MTOC staining consist of stacked images. Analysis was done manually. Each cell reaching the edge of the cell layer was divided into three sections of 120 degrees, one facing the scratch (Fig. S1). Cells were counted as being polarized if the MTOC, or more than 50% of the GA staining, respectively, was located in this region. A minimum of 90 cells were analysed for each condition. Indicated samples were treated with 80 μM WEHD-fmk (R&D Systems, Minneapolis, USA) from 8 h p.i.

siRNA experiments

For protein depletion prior to microscopic assays, cells were seeded into 12-well plates with a density of 1.5 × 105 and transfected with specific siRNA. One microgram of total RNA and 6 μl of RNAiFect (Invitrogen, Life Technologies, Grand Island, USA) per 12-well were incubated in 200 μl OptiMEM (Invitrogen, Life Technologies, Grand Island, USA) for 20 min before the mixture was added to the cells with 600 μl DMEM (Rejman Lipinski et al., 2009). The next day, the cells were split into appropriate volumes and grown for 24 h before infection.

Acknowledgements

The authors would like to thank Anja Greiser for excellent technical assistance. J.H. and D.H. would like to thank Anton Aebischer, Sebastian Banhart, Lukas Aeberhard and Sophia Koch for critical comments on the manuscript. The authors furthermore declare to have no conflict of interest.

Ancillary