|Home | About | Journals | Submit | Contact Us | Français|
Listeria monocytogenes is a bacterial pathogen that replicates within the cytosol of infected host cells. The ability to rapidly escape the phagocytic vacuole is essential for efficient intracellular replication. In the murine model of infection, the pore-forming cytolysin listeriolysin O (LLO) is absolutely required for vacuolar dissolution, as LLO-deficient (ΔLLO) mutants remain trapped within vacuoles. In contrast, in many human cell types ΔLLO L. monocytogenes are capable of vacuolar escape at moderate to high frequencies. To better characterize the mechanism of LLO-independent vacuolar escape in human cells, we conducted an RNA interference (RNAi) screen to identify vesicular trafficking factors that play a role in altering vacuolar escape efficiency of ΔLLO L. monocytogenes. RNAi knockdown of 18 vesicular trafficking factors resulted in increased LLO-independent vacuolar escape. Our results suggest that knockdown of one factor, RABEP1 (rabaptin-5), decreased the maturation of vacuoles containing ΔLLO L. monocytogenes. Thus, we provide evidence that increased vacuolar escape of ΔLLO L. monocytogenes in human cells correlates with slower vacuolar maturation. We also determined that increased LLO-independent dissolution of vacuoles during RABEP1 knockdown required the bacterial broad-range phospholipase PC-PLC. We hypothesize that slowing the kinetics of vacuolar maturation generates an environment conducive for vacuolar escape mediated by the bacterial phospholipases.
Listeria monocytogenes is a facultative intracellular bacterial pathogen capable of infecting a wide variety of species. Upon entry into host cells, L. monocytogenes escapes the entry vacuole, replicates within the cytosol, and uses actin-based motility to spread to neighboring cells resulting in formation of secondary double-membrane vacuoles. Once in neighboring cells, L. monocytogenes escapes the secondary vacuole and begins replicating again in the cytosol (Tilney and Portnoy, 1989). Vacuolar dissolution is essential for intracellular growth as L. monocytogenes fails to replicate or replicates inefficiently within vacuoles (Birmingham et al., 2008; Cheng and Portnoy, 2003; Portnoy et al., 1988). In all murine cells examined, the bacterial pore-forming cytolysin listeriolysin O (LLO, encoded by hly) is absolutely required for vacuolar escape. Consistent with vacuolar escape being essential for intracellular growth, LLO-deficient L. monocytogenes strains (ΔLLO) are avirulent in the murine infection model (Portnoy et al., 1988). Based on these results, inhibitors of LLO have been proposed as alternative or supplemental treatments for L. monocytogenes infections (Edelson and Unanue, 2001).
However, in contrast to what is observed in the murine system, some human epithelial cells, such as HeLa and HEp-2 cells, permit efficient vacuolar escape of ΔLLO L. monocytogenes (Grundling et al., 2003). Additionally, primary human dendritic cells and human WS-1 skin fibroblasts have been previously shown to allow vacuolar escape of ΔLLO L. monocytogenes, but at much lower levels than wild-type bacteria (Paschen et al., 2000; Portnoy et al., 1988). Similar results have been observed recently in our lab for a number of other human cell lines including HepG2 (hepatocytes), Hct116 (colon epithelial cells) and HEK-293 (kidney epithelial cells) (Burrack and Higgins, unpublished). These results suggest that inhibitors of LLO may be an effective treatment for L. monocytogenes infections in the murine infection model, but not in humans. Therefore, we sought to better understand the role of LLO and additional bacterial/host cell factors mediating efficient vacuolar escape in human cells.
The observation that some human cells are permissive for vacuolar escape of ΔLLO L. monocytogenes was first made over 20 years ago (Portnoy et al., 1988), but the differences in the vacuolar biology of human cells that allow LLO-independent vacuolar escape have not been described. In human cell lines permissive for vacuolar escape of ΔLLO L. monocytogenes, the bacterial broad-range phospholipase PC-PLC was shown to be essential for vacuolar escape (Alberti-Segui et al., 2007; Grundling et al., 2003; Marquis et al., 1995). L. monocytogenes also encodes a phosphatidylinositol-specific phospholipase, PI-PLC (Camilli et al., 1993). PI-PLC assists LLO and PC-PLC in lysis of primary vacuoles and secondary double-membrane vacuoles (Smith et al., 1995). The bacterial phospholipases appear to work synergistically during infection as deletion of both phospholipases results in a more severe defect in vacuolar dissolution than would be expected from the phenotypes observed following deletion of either phospholipase alone (Smith et al., 1995).
LLO is a member of the cholesterol-dependent cytolysin family and has optimum pore-forming activity at an acidic pH (Beauregard et al., 1997). The pH dependence of the pore-forming activity of LLO is important for compartmentalizing the activity of the protein to the vacuole and thus preventing cytotoxicity by LLO secreted from bacteria replicating in the cytosol (Glomski et al., 2002). However, the requirement of an acidic pH environment for proper function of LLO makes L. monocytogenes dependent on partial maturation of the vacuole to facilitate efficient lysis of the vacuole. Treatment of murine macrophages with bafilomycin A1, an inhibitor of vacuolar ATPase (v-ATPase) activity, or knockdown of v-ATPase subunits by RNA interference (RNAi) in Drosophila hemocytes results in vacuoles that do not acidify and inefficient vacuolar escape by L. monocytogenes (Agaisse et al., 2005; Beauregard et al., 1997). In murine macrophages, L. monocytogenes is thought to efficiently escape from RAB7-positive compartments resembling late-endosomes. However, the ability of bacteria to escape from mature vacuoles containing lysosomal-associated membrane protein 1 (LAMP1) has been shown to be much less efficient than from RAB7-positive late endosome-like vacuoles (Henry et al., 2006). This result suggests that maturation beyond a late-endosome-like compartment is detrimental for vacuolar dissolution by L. monocytogenes.
Recent studies demonstrate that LLO activity affects vacuolar maturation during infection. Swanson and colleagues have shown that in murine RAW 264.7 cells, vacuoles containing wild-type L. monocytogenes exhibit slower maturation than vacuoles containing ΔLLO bacteria (Henry et al., 2006). Furthermore, it was shown that LLO causes small perforations in the vacuolar membrane prior to bacterial escape causing uncoupling of pH and calcium gradients (Shaughnessy et al., 2006). Based on these observations, we hypothesized that one of the functions of LLO during infection is to slow vacuolar maturation and that alterations in vacuolar maturation kinetics in some human cell lines provides a more conducive environment for PC-PLC to mediate vacuolar escape of L. monocytogenes in the absence of LLO activity.
Our group and others have previously used RNAi as a method to identify host factors required for intracellular bacterial infection. RNAi in Drosophila tissue culture cells has been used to identify host factors required for infection by L. monocytogenes, Mycobacterium fortuitum, Legionella pneumophila and Chlamydia caviae (Derre et al., 2007; Dorer et al., 2006; Agaisse et al., 2005; Cheng et al., 2005; Philips et al., 2005). Additionally, Portnoy and co-workers used a combination of L. monocytogenes mutants with a genome-scale Drosophila RNAi screen to further characterize the interaction of bacterial virulence factors and host cells (Cheng et al., 2005). One of the bacterial strains used was ΔLLO L. monocytogenes. Their results indicated that RNAi knockdown of several members of the host multi-vesicular body required for transition between late endosomes and lysosomes led to increased vacuolar escape of ΔLLO L. monocytogenes in Drosophila cells (Cheng et al., 2005). However, these results were not confirmed in human cells nor was the mechanism permitting increased vacuolar escape characterized. Compared to the Drosophila system, large-scale RNAi screens have been used to a much more limited extent in human cells for studies of interactions between bacterial pathogens and their hosts, as most research using RNAi techniques in mammalian cells has focused on knockdown of specific host cell proteins (Knodler et al., 2005; Ohya et al., 2005). Recently, Kuijl et al. used a sub-genomic library of small interfering RNAs (siRNAs) targeting the human kinome to identify kinases required for efficient intracellular infection of Salmonella enterica serovar Typhimurium and Mycobacterium tuberculosis (Kuijl et al., 2007).
In this work, we designed and completed an RNAi screen in human cells to identify vesicular trafficking host cell factors that play a role in altering vacuolar escape of ΔLLO L. monocytogenes. We provide evidence that shRNA treatments resulting in a significant increase in vacuolar escape by ΔLLO bacteria also slow the vacuolar maturation process. Our work supports a role for LLO in modulating vacuolar maturation in host cells and suggests that the ability of LLO to alter the vacuolar maturation process is indeed functionally important during L. monocytogenes infection of mammalian hosts.
To identify host cell vesicular trafficking components that play a role in preventing efficient vacuolar escape of ΔLLO L. monocytogenes in human cells, we first compiled a small hairpin RNA (shRNA) library containing 211 hairpins targeting 72 gene products with functional roles in vesicular trafficking (Table S1). The majority of the human homologs for the vesicular trafficking proteins identified in the genome-wide Drosophila RNAi screen performed in our lab (Agaisse et al., 2005) and a similar screen performed by Portnoy and co-workers (Cheng et al., 2005) were included along with other selected targets predicted to be in the same vesicular trafficking pathways. The sub-genomic library was constructed by selecting individual shRNA clones from a previously constructed genome-wide pSMC2 shRNA library (Silva et al., 2005). The pSMC2 shRNAs were designed to more closely mimic naturally occurring microRNA (miRNA) transcripts. This design has been shown to produce shRNAs that are more potent than constructs that do not mimic miRNAs. The increased potency is likely due to better processing of the shRNA by Dicer into duplexes capable of interacting with RISC in a manner that results in efficient mRNA degradation (Silva et al., 2005).
We then used this library to screen for shRNAs targeting vesicular trafficking gene products that caused increased vacuolar escape of ΔLLO L. monocytogenes in human cells. For our screen, we used HEK-293 human kidney epithelial cells. HEK-293 cells are easily transfectable. Also, in untreated HEK-293 cells, a very low level of vacuolar escape (<1%) of ΔLLO L. monocytogenes is observed. For the screen, HEK-293 cells were transfected with each shRNA in black, clear bottom 96-well plates. A small amount of dsRed-encoding plasmid (1/10 total DNA amount) was co-transfected to estimate the transfection efficiency. Typically, greater than 90% of cells in each well were dsRed-positive. After incubating the transfected HEK-293 cells for 60 h to allow for RNAi-mediated knockdown of target gene products, transfected cells were infected with GFP-expressing ΔLLO L. monocytogenes for 10 h. Host cell nuclei were then stained with Hoechst dye to allow for the determination of the number of host cells per well. Intracellular growth of ΔLLO L. monocytogenes was monitored by automated fluorescence microscopy (Fig. 1). Replication of ΔLLO L. monocytogenes within membrane-bound vacuoles of HEK-293 cells is inefficient. Therefore, high levels of bacteria within host cells indicate that L. monocytogenes has escaped the vacuole and has replicated efficiently in the cytosol. We tested this assumption by staining infected cells with Texas Red-phalloidin. In host cells with large numbers, L. monocytogenes bacteria were stained with Texas Red-phalloidin indicating association with host cell F-actin, an established marker of L. monocytogenes vacuolar escape. To estimate vacuolar escape, we counted the number of cells in each image that were filled with bacteria (escape events) as shown by white arrows in Figure 1. Escape events were quantified in two images per well and the screen was performed in duplicate (4 images/shRNA in total). For untreated HEK-293 cells, the mean number of escape events observed was 1.8 ± 0.8 per image. If the mean number of escape events in shRNA-transfected wells was greater than 2 standard deviations above the mean number of escape events observed in untreated cells (>3.4 events per image), the shRNA treatment was determined to increase vacuolar escape.
We identified 32 hairpins targeting 18 gene products that increased vacuolar escape of ΔLLO L. monocytogenes (Table 1). All hairpins identified were sequenced to ensure that the inserts were correct and that the shRNA targeted the intended gene product. We chose 10 gene products for further analysis (ARCN1, COPG, EEA1, RAB21, RAB4A, RAB5C, RABEP1, SEC15L1, CTSL, and VPS28). Because the pBlueScript-based vector pBS-mir is a more stable vector for repeated propagation in E. coli than pSMC2 (Li and Elledge, 2005; Silva et al., 2005), the shRNA inserts targeting the selected gene products were transferred from the pSMC2 vesicular trafficking library plasmids to pBS-mir. Each shRNA insert that was successfully transferred to a pBS-mir vector was then analyzed and the vacuolar escape frequency of ΔLLO L. monocytogenes was quantified by determining the mean number of escape events in 4 images per well in three separate experiments. Control shRNAs targeting GFP and firefly luciferase were also included in these experiments. The mean number of escape events per image for all experiments performed with each shRNA insert is indicated in Table 1.
In untreated wells and wells transfected with GFP and firefly luciferase control shRNAs, escape events occurred in approximately 1% of HEK-293 cells. Among all of the shRNA treatments examined, knockdown of several COP-I coatomer subunits and RABEP1 resulted in the highest levels of vacuolar escape of ΔLLO L. monocytogenes with vacuolar escape events occurring in approximately 5–10% of HEK-293 cells. Based on these results and our interest in the effect of alterations in early stages of vacuolar maturation on L. monocytogenes infection, we decided to further characterize the role of RABEP1 in modulating vacuolar escape of ΔLLO L. monocytogenes. RABEP1 is an effector of RAB5A and RAB4A (Pagano et al., 2004) and functions in early endosome and recycling endosome formation. RAB4A was also identified in the RNAi screen. Several other gene products with roles in early endosomes and endosome recycling were also identified (Table 1). The significant increase in escape events observed following knockdown of RABEP1 might occur because RABEP1 functions in the regulation of both the early endosome and recycling endosome pathways.
Three shRNAs targeting RABEP1 were in the vesicular trafficking sub-genomic library (Table S1). Two of these hairpins, v2HS_37113 and v2HS_37115 (RABEP1-1 and RABEP1-3, respectively) resulted in an increased frequency of HEK-293 cells in which efficient growth of ΔLLO L. monocytogenes was observed (Table 1). To further demonstrate that knockdown of RABEP1 resulted in increased vacuolar escape, we performed Western blots and infection experiments in parallel with shRNA-transfected HEK-293 cells. Depletion of RABEP1 protein was observed upon treatment with the RABEP1-1 and RABEP1-3 hairpins (Fig. 2A), which also resulted in increased vacuolar escape (Fig. 2B). Knockdown of RABEP1 was most efficient following transfection of the RABEP1-1 shRNA. However, transfection of the RABEP1-2 hairpin (v2HS_37116) failed to decrease RABEP1 protein levels or alter vacuolar escape of ΔLLO L. monocytogenes (Fig. 2A and 2B). To further show that ΔLLO bacteria in highly infected host cells were located within the cytosol, we infected shRNA-transfected HEK-293 cells with GFP-expressing ΔLLO bacteria and then stained host cell actin with Texas Red-phalloidin. L. monocytogenes in highly infected cells transfected with the RABEP1-1 and RABEP1-3 hairpins were associated with cytosolic host cell F-actin indicating that bacteria had escaped the vacuole (Fig. 2C).
To ensure that the results we observed were not specific to HEK-293 cells, we determined the efficiency of vacuolar escape in Hct116 human colon epithelial cells following knockdown of vesicular trafficking gene products identified in the RNAi screen. We transfected Hct116 human epithelial colon cells with shRNAs targeting COPG, CTSL and RABEP1. These gene products were chosen because they represent three distinct functional categories of vesicular trafficking gene products identified (Table 1). Since untreated Hct116 cells are more permissive for vacuolar escape of ΔLLO L. monocytogenes than HEK-293 cells, the increase in vacuolar escape observed for shRNA-treated Hct116 cells was less dramatic than that in HEK-293 cells. However, similar results were observed in both human-derived cell lines, as knockdown of COPG, CTSL, and RABEP1 (data not shown and Fig. 3) significantly increased vacuolar escape and intracellular replication of ΔLLO L. monocytogenes in both HEK-293 cells (Fig. 3A and 3B) and Hct116 cells (Fig. 3C and 3D) indicating that the results are likely applicable to multiple human-derived cell lines.
In human cervical epithelial cell lines (HeLa, HEp-2 and Henle 407), the L. monocytogenes broad range phospholipase PC-PLC is required for vacuolar escape in the absence of LLO (Grundling et al., 2003). L. monocytogenes also encodes a phosphatidylinositol-specific phospholipase, PI-PLC, that assists in lysis of primary vacuoles and double-membrane secondary vacuoles (Smith et al., 1995). To determine the roles of the phospholipases for LLO-independent vacuolar escape in HEK-293 cells, we used a GFP-expressing strain lacking LLO and both phospholipases (ΔLLO ΔPI-PLC ΔPC-PLC) as well as GFP-expressing strains lacking LLO and either PI-PLC or PC-PLC (ΔLLO ΔPI-PLC and ΔLLO ΔPC-PLC, respectively) to infect HEK-293 cells treated with RABEP1-1 shRNA. PC-PLC was required for the observed increase in vacuolar escape following knockdown of RABEP1 in HEK-293 cells, as infection with ΔLLO ΔPI-PLC ΔPC-PLC or ΔLLO ΔPC-PLC bacteria resulted in no escape events being observed (Fig. 4). PI-PLC was shown to have an important synergistic role in allowing vacuolar escape of ΔLLO L. monocytogenes as the ΔLLO ΔPI-PLC strain showed lower levels of vacuolar escape than ΔLLO L. monocytogenes (Fig. 4, compare ΔLLO and ΔLLO ΔPI-PLC). Infection of RABEP1-1 shRNA-treated HEK-293 cells with ΔLLO L. monocytogenes resulted in ~7.5-fold more escape events per image than infection with ΔLLO ΔPI-PLC L. monocytogenes as 34.0 ± 7.1 escape events per image were observed with ΔLLO L. monocytogenes infection while 4.5 ± 0.7 escape events per image were observed with ΔLLO ΔPI-PLC L. monocytogenes infection.
The requirement of PC-PLC for vacuolar escape of ΔLLO L. monocytogenes provides support for the hypothesis that the increase in vacuolar escape observed following shRNA treatment is due to alterations in vacuole maturation that facilitate function of the bacterial phospholipases. To further characterize the mechanism permitting increased vacuolar escape, we next focused on analyzing the effects of knockdown of RABEP1 on the maturation of L. monocytogenes-containing vacuoles. We first examined association of LAMP1, a marker of mature/late endosomes and lysosomes, with ΔLLO L. monocytogenes. To examine co-localization with LAMP1, we transfected HEK-293 cells with RABEP1-1 shRNA as previously described and then transfected the cells with a LAMP1-CFP fusion construct 24 h prior to infection. Infected HEK-293 cells were fixed and ΔLLO L. monocytogenes were detected using wide-field immunofluorescence microscopy. A positive association of LAMP1-CFP with a L. monocytogenes-containing vacuole was characterized as a ring of increased CFP fluorescence intensity surrounding the intracellular ΔLLO L. monocytogenes.
We initially examined co-localization of LAMP1-CFP with ΔLLO L. monocytogenes at 2 h post-infection, as phospholipase-mediated vacuolar escape appeared to occur shortly after this time. Co-localization of LAMP1-CFP with ΔLLO L. monocytogenes was determined for at least 50 bacteria per transfection condition in each of three independent experiments. In cells treated with FuGene 6 or transfected with firefly luciferase control shRNAs, 79.9% and 82.2% of bacteria, respectively, were positively associated with LAMP1-CFP at 2 h post-infection. In cells treated with RABEP1-1 shRNA, a slight decrease in LAMP1-CFP association with ΔLLO L. monocytogenes was observed with 71.6% of bacteria associated with LAMP1-CFP. A one-way ANOVA and Bonferronni post-tests indicated that RABEP1 knockdown significantly decreased LAMP1-CFP association compared to FuGene 6 (p<0.01) and luciferase (p<0.001) shRNA treatments. There was no statistical difference between FuGene 6 and luciferase shRNA treatments.
Based on the observation of a small, but consistent decrease in the association of ΔLLO L. monocytogenes with LAMP1-CFP in cells treated with shRNA targeting RABEP1 at 2 h post-infection, we further examined association of LAMP1-CFP with ΔLLO L. monocytogenes at various time points post-infection. We determined that the association of LAMP1-CFP with intracellular ΔLLO L. monocytogenes is significantly decreased at several time points post-infection in cells treated with shRNAs targeting RABEP1 (Fig. 5A and 5B). Association of LAMP1-CFP in control cells and cells treated with shRNAs targeting RABEP1 was similar at 15 min post-infection. However, the percentage of LAMP1-CFP association with ΔLLO L. monocytogenes was less in cells treated with RABEP1 shRNA than in control cells at later time points (Fig. 5A). Furthermore, knockdown of RABEP1 primarily had an observable effect on the development of LAMP1-associated phagolysosome-like vacuoles, as the fraction of ΔLLO L. monocytogenes associated with the late-endosome protein marker RAB7-CFP was not altered following treatment of HEK-293 cells with RABEP1-1 shRNA (Fig. 5C and 5D). Similarly, the association of the late-endosome lipid lysobisphosphatidic acid (LBPA) with ΔLLO L. monocytogenes in RABEP1-1 shRNA-treated HEK-293 cells was similar to the association of LBPA with ΔLLO L. monocytogenes in control HEK-293 cells (Fig. S1).
The observation that ΔLLO L. monocytogenes can escape from vacuoles in human-derived cell lines was first made over 20 years ago (Portnoy et al., 1988), however the mechanism permitting vacuolar escape of ΔLLO bacteria has not been characterized extensively. In this work, we performed a sub-genomic RNAi screen in human cells to identify vesicular trafficking factors that play a role in facilitating vacuolar escape of ΔLLO L. monocytogenes. Knockdown of 18 different gene products increased vacuolar escape of ΔLLO L. monocytogenes. The apparent enhancement of vacuolar escape could be caused by increased bacterial uptake by shRNA-treated host cells or by increased LLO-independent vacuolar dissolution. Quantification of intracellular ΔLLO L. monocytogenes at early time points post-infection did not show an increase in bacterial uptake following treatment with shRNAs identified in the screen (Fig. 3A, Fig. 3C, Fig. S2, and data not shown). In contrast, the data presented here are consistent with an increased efficiency of vacuolar disruption by ΔLLO L. monocytogenes following shRNA treatment. Furthermore, our results suggest that LLO-independent vacuolar dissolution is correlated with reduced vacuolar maturation (Fig. 6).
The bacterial phospholipase PC-PLC was found to be essential for vacuolar escape of ΔLLO L. monocytogenes following knockdown of several host cell vesicular trafficking gene products, including RABEP1 (Fig. 4). In recent studies examining the vacuolar biology of L. monocytogenes, several different outcomes have been observed depending upon the infection model and how vacuole maturation was altered. In Drosophila cells, RNAi knockdown of host cell proteins with roles in lysosomal transport resulted in growth of wild-type and ΔLLO L. monocytogenes within vacuoles (Agaisse et al., 2005). Recent work in murine macrophages has shown that LLO can promote replication of L. monocytogenes within vacuoles in an autophagy-dependent manner (Birmingham et al., 2008). In both cases, replication within membrane-bound compartments is significantly slower than that observed within the cytosol. In this study, we identified conditions in which phospholipase-mediated vacuolar escape of ΔLLO L. monocytogenes occurs at elevated frequencies to allow replication in the cytosol.
Using RNAi in Drosophila cells, Portnoy and co-workers determined that knockdown of several components of the multi-vesicular body (MVB) and endosomal sorting complex required for transport (ESCRT) complexes allowed bypass of the requirement of LLO for vacuolar escape (Cheng et al., 2005). MVB and ESCRT complexes function in the transition of endosomes to lysosomes. We also observed increased vacuolar escape of ΔLLO L. monocytogenes in HEK-293 cells following knockdown of the ESCRT complex member VPS28 (Table 1). In addition to confirming that knockdown of an ESCRT complex subunit can cause increased LLO-independent vacuolar escape within a human cell line, we identified several other functional categories of vesicular trafficking proteins that have roles in facilitating vacuolar escape of ΔLLO L. monocytogenes (Table 1). Treatment with shRNAs targeting a lysosomal trafficking protein (LYST) and a lysosomal protease (CTSL) resulted in increased vacuolar escape of ΔLLO bacteria. Knockdown of several members of the COP-I coatomer, a complex involved primarily in Golgi-related vesicle transport (Cai et al., 2007), and three additional gene products (RAB2A, SARA2, and YKT6) with roles in endoplasmic reticulum (ER) to Golgi vesicular trafficking increased vacuolar escape of ΔLLO L. monocytogenes. We also observed increased vacuolar escape following treatment with shRNAs targeting two exocyst components, SEC3L1 and SEC15L1. The exocyst complex is required for protein secretion and functions in vesicle transport between the Golgi apparatus, endosomes, and the plasma membrane (Wang and Hsu, 2006). Two other functional categories of gene products that when targeted by shRNA treatment caused increased LLO-independent vacuolar escape were those involved in regulation and formation of early endosomes and recycling endosomes. These results suggest that several vesicular trafficking pathways can be altered in human cells that permit efficient intracellular infection by ΔLLO L. monocytogenes.
Previous work has suggested that vacuolar escape of L. monocytogenes is inhibited by maturation of vacuoles to lysosome-like compartments. Mature LAMP1-positive vacuoles are not as permissive for vacuolar escape of L. monocytogenes (Henry et al., 2006). Additionally, the lysosomal protease cathepsin-D (CTSD) has been implicated in degrading LLO and inhibiting L. monocytogenes intracellular growth (del Cerro-Vadillo et al., 2006). Moreover, Cheng et al. suggested that knockdown of MVB components allowed increased vacuolar escape of ΔLLO L. monocytogenes in Drosophila cells due to a block of vacuolar maturation (Cheng et al., 2005). Our results examining vacuolar maturation following knockdown of the early endosome/recycling endosome regulator RABEP1 support the hypothesis that LLO-independent vacuolar escape is inversely correlated with rates of vacuolar maturation. In five independent experiments, we showed a decreased association of the lysosomal protein LAMP1 with ΔLLO L. monocytogenes-containing vacuoles following knockdown of RABEP1 (Fig. 5A). While the transfection procedure used allowed for a limited number of host cells to be evaluated per experiment, the data were consistent and suggest that knockdown of RABEP1 alters the maturation of ΔLLO L. monocytogenes-containing vacuoles and results in increased vacuolar escape of ΔLLO L. monocytogenes (Fig. 5A).
Our analysis of the maturation of ΔLLO L. monocytogenes vacuoles in cell lines in which varying levels of LLO-independent vacuolar escape occurs also suggests that LLO-independent vacuolar dissolution may be inversely correlated with rates of vacuolar maturation. Swanson and co-workers previously showed that 90% of vacuoles containing ΔLLO L. monocytogenes were LAMP1-positive by 15 min post-infection in RAW 264.7 cells (Henry et al., 2006), a cell line where vacuolar escape is absolutely dependent upon LLO. We determined that in HEK-293 cells, where ΔLLO L. monocytogenes can undergo vacuolar escape at a low frequency, ~5% of intracellular ΔLLO L. monocytogenes were LAMP1-postive 15 min post-infection (Fig. 5A). In HEK-293 cells, ΔLLO L. monocytogenes vacuoles eventually mature to lysosomes, but the association of LAMP1 occurs at later time points. The percentage of ΔLLO L. monocytogenes associated with LAMP1 in HEK-293 cells increases from ~5% LAMP1-positive vacuoles 15 min post-infection to a maximum of ~80% LAMP1-positive vacuoles at 2 h post-infection. The efficiency of vacuolar maturation and the vacuolar escape efficiency of ΔLLO L. monocytogenes appear to be correlated in murine RAW 264.7 macrophage cells and in human HEK-293 epithelial cells. In human HEp-2 cervical epithelial cells, ΔLLO L. monocytogenes escape the vacuole efficiently (Grundling et al., 2003). Consistent with the correlation that LLO-independent vacuolar escape occurs more readily in cells with slower vacuolar maturation rates, preliminary results suggest that LAMP1 association with ΔLLO L. monocytogenes is much slower in HEp-2 cells than in HEK-293 cells yielding ~18% LAMP1-positive vacuoles in HEp-2 cells 60 min post-infection compared to ~70% LAMP1-positive vacuoles in HEK-293 cells 60 min post-infection (data not shown and Fig. 5A).
Our results suggest that knockdown of host cell vesicular trafficking proteins within each of the identified functional categories (Table 1) could result in a block in vacuolar maturation or a general slowing of vacuolar maturation (Fig. 6). Enhanced vacuolar dissolution following knockdown of MVB components and lysosomal proteins is likely due to blocking the formation of a degradative, LAMP1-positive, lysosomal compartment (Piper and Katzmann, 2007). Interestingly, knockdown of ESCRT complex members in Drosophila S2 cells and murine RAW 264.7 macrophage-like cells was recently shown to make cells permissive for intracellular infection by normally non-pathogenic Mycobacterium smegmatis (Philips et al., 2008). Knockdown of ESCRT complexes also prevented expression of GFP from an acid-activated promoter in M. fortuitum indicating that knockdown of ESCRT complexes resulted in the bacteria residing in a less acidic, less mature vacuolar compartment (Philips et al., 2008). Knockdown of COP-I coatomer subunits may be disrupting protein trafficking from the Golgi apparatus to the ER and between Golgi cisternae resulting in pleiotropic effects on vesicle dynamics within the cell. However, knocking down COP-I components may result in a more specific inhibition of vacuolar maturation, as some subunits of the COP-I complex have been localized to endosomes and have functional roles in the formation of vesicles that allow transport between early and late endosomes (Aniento et al., 1996). Exocyst components have been recently identified as being associated with phagosomes in Drosophila cells and murine macrophages (Stuart et al., 2007). We hypothesize that exocyst components may also be a part of the induced phagosome-like vacuoles formed to allow L. monocytogenes uptake in non-professional phagocytic cells, such as epithelial cells, and that these proteins may function in the maturation of the vacuole, perhaps by assisting in the removal of proteins associated with early stages of maturation. Knockdown of early endosome or recycling endosome components likely affect the vacuole protein composition early after bacterial entry. These altered vacuoles may be missing signaling proteins required for acquisition of late endosome and lysosomal characteristics. Indeed, inhibition of endosome recycling by expression of a dominant negative version of RAB11 has previously been shown to slow maturation of Salmonella-containing vacuoles (SCVs) (Smith et al., 2005).
Slower vacuolar maturation may be more permissive for efficient vacuolar escape of L. monocytogenes for multiple reasons. First, slower vacuolar maturation may provide the bacterial phospholipases with an environment more conducive for optimal activity for a longer period of time. It is known that PC-PLC activity is activated at pH 7.0 and lower, but not at pH 7.3 (Marquis and Hager, 2000). Therefore, slightly acidic endosome-like compartments are likely to promote activation of PC-PLC, but do not contain the high concentrations of degradative proteases found in more mature lysosomal compartments. The concept that lysosomal compartments are inhibitory for L. monocytogenes protein function is supported by the observation that wild-type L. monocytogenes escape inefficiently from LAMP1-positive lysosomal vacuoles (Henry et al., 2006). Also, since plcA and plcB, the genes encoding PI-PLC and PC-PLC, are up-regulated during intracellular infection (Chatterjee et al., 2006), the altered vacuolar maturation may allow for the bacteria to produce increased levels of the phospholipases before the vacuole matures to a degradative lysosomal compartment. Additionally, there may be differences in the membrane composition of the endosome-like L. monocytogenes vacuoles that are better substrates for PC-PLC. Conditions that increase vacuolar escape of ΔLLO L. monocytogenes do not alter the concentration of the late-endosome specific lipid LBPA in the vacuolar membrane (Fig. S1), but the concentrations of other phospholipids may affect LLO-independent vacuolar escape efficiencies. PC-PLC is a broad-range phospholipase, but has been shown to have the highest level of activity on vesicles containing phosphatidylserine (PS) and phosphatidylethanolamine (PE) than on sphingomyelin (Sph) and phosphatidylcholine (PC) containing membranes (Montes et al., 2004). The relative levels of these phospholipids may be altered by changes in vesicular trafficking (Yeung and Grinstein, 2007).
Our results suggest that knockdown of RABEP1 altered LAMP1 acquisition on ΔLLO L. monocytogenes-containing vacuoles in HEK-293 cells (Fig. 5A). However, we did not observe differences in the late-endosome stage of vacuolar maturation either when measuring RAB7 or LBPA association with intracellular ΔLLO L. monocytogenes following treatment with shRNAs targeting RABEP1 (Fig. 5C and Fig. S1). We initially found this result surprising since RAB7 generally functions at an intermediate step between early endosomes and lysosomes. One possible explanation is that since the late endosome stage is transient, while mature LAMP1 vacuoles persist in cells, the changes in maturation are simply not sufficient to observe a difference at the late endosome stage. Another possibility is that, since RABEP1 functions at the junction of whether endosomes are recycled or mature, knockdown of RABEP1 may alter endosome formation and shift the balance toward recycling endosome-like compartments while simultaneously slowing down the maturation process of endosomes (Deneka et al., 2003). This phenotype would result in fewer L. monocytogenes vacuoles becoming late-endosome-like, but with those still formed having slower kinetics of LAMP1 acquisition. When RAB7-association was examined at fixed time points, the reduced formation efficiency and increased duration of the late endosome-like stage may result in approximately the same fraction of membrane-bound compartments being RAB7-positive as in control cells.
RABEP1 is a Rab effector protein that has been shown to interact with both RAB4A and RAB5A to regulate passage of cargo in early endosomes to either the recycling pathway (back to the surface or Golgi) or the endocytic maturation pathway (towards lysosomes) (Pagano et al., 2004; Deneka et al., 2003). RAB4A was also identified in our RNAi screen (Table 1). Furthermore, murine Rab5a has been implicated in regulating L. monocytogenes infection and the L. monocytogenes p40 protein has been shown to inhibit Rab5a GDP/GTP exchange activity (Alvarez-Dominguez et al., 2008). Degradation of RABEP1 by caspase-3 following infection with L. pneumophila has also been implicated in halting maturation of the Legionella-containing vacuole (LCV) (Molmeret et al., 2004). Collectively, these data suggest that the switch between endosome maturation and endosome recycling may be important for modulating the L. monocytogenes vacuole, both in our infection conditions where LLO is absent and during wild-type infections. The importance of the regulation of endosome recycling during L. monocytogenes infection and for the proper formation of SCVs and LCVs suggests that modifications of endosome recycling may alter vacuolar maturation during infection by a variety of intracellular pathogens. In L. monocytogenes infections, LLO has been shown to slow vacuolar maturation kinetics by uncoupling pH and calcium gradients that act as maturation signals (Shaughnessy et al., 2006). Our work demonstrating that knockdown of host cell factors resulting in altered vacuolar maturation can compensate for the vacuolar escape defect of ΔLLO L. monocytogenes suggests that the ability of LLO to slow vacuolar maturation is functionally important during infection.
L. monocytogenes strains used in this study were 10403S ΔLLO (DH-L489), pPL3-bGFP ΔLLO (DH-L1288), pPL3-bGFP ΔLLO ΔPI-PLC ΔPC-PLC (DH-L1546), pPL3-bGFP ΔLLO ΔPI-PLC (DH-L1544), and pPL3-bGFP ΔLLO ΔPC-PLC (DH-L1545). pPL3-bGFP is a pPL3-based integration vector (Grundling et al., 2004) that expresses GFPmut2 from the tRNAArg locus on the L. monocytogenes chromosome. gfpmut2 in pPL3-bGFP is transcribed from the constitutive pHyper promoter as a fusion to the 5' UTR of the L. monocytogenes hly gene (Shen and Higgins, 2005). L. monocytogenes strains were routinely grown at 37°C in brain heart infusion (BHI) broth or on BHI agar (Difco Laboratories, Detroit, MI). L. monocytogenes were typically grown in 2 ml BHI at 30°C without shaking for 14–16 h prior to infection. All L. monocytogenes strains were stored at −80°C in BHI containing 40% glycerol. Chloramphenicol was used at 7.5 μg/ml for selection of integrated pPL3 derivatives in L. monocytogenes.
HEK-293 human kidney epithelial cells (ATCC CRL-1573) and Hct116 human colon epithelial cells (ATCC CCL-247) were cultured in Dulbecco's modified Eagle's medium (DMEM; Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT), 2 mM glutamine, 1 mM sodium pyruvate, and 100 μg/ml penicillin and streptomycin (P/S) (DMEM-10-P/S). For infection assays, host cell cultures were maintained in antibiotic-free DMEM-10 medium (DMEM-10). All host cell cultures were maintained at 37°C in a 5% CO2-air atmosphere.
E. coli clones containing the pSMC2 shRNAs were isolated from the human genome library described in Silva et al. (Silva et al., 2005) and grown in LB with 40 μg/ml chloramphenicol. Aliquots were frozen in 20% glycerol at −80°C and shRNA plasmids were prepared using a 96-well mini-prep kit (Qiagen, Valencia, CA). Plasmid concentrations were tested from representative wells and were approximately 200 ng/μl. For sequencing, additional plasmid DNA was prepared from 5 ml cultures started from the frozen aliquots using a single-column mini-prep kit (Qiagen). Sequences for all hairpin inserts listed in Table 1 can be found at http://codex.cshl.org. Mating was used to move shRNA inserts from the pSMC2 shRNA vector to the pBS-mir vector and was done as previously described (Li and Elledge, 2005).
1.0 × 104 HEK-293 cells per well were seeded in a black, clear bottom 96-well plate (Corning, Lowell, MA) in 100 μl DMEM-10. Twenty-four hours later, cells were transfected with shRNAs using FuGene 6 (Roche Diagnostics, Basel, Switzerland). For transfection of each well, 0.6 μl FuGene 6 was diluted in 20 μl OPTI-MEM, followed by addition of approximately 0.2 μg shRNA plasmid DNA and incubation for 20 min at room temperature (RT; 20–25°C). Complexes were added to cells and plates were centrifuged for 5 min at 200 × g in a table-top centrifuge. Approximately 60 h after transfection, cells were infected with GFP-expressing L. monocytogenes strains at an MOI=5 by addition of bacteria in 25 μl DMEM-10, followed by centrifuging the plates for 1 min at 200 × g. Twenty-five microliters of DMEM-10 with 350 μg/ml gentamicin was added 1 h post-infection to kill extracellular bacteria (final concentration of 50 μg/ml gentamicin). At 10 h post-infection, cells were fixed with 3.2% paraformaldehyde in phosphate-buffered saline, pH 7.1 (PBS) for 14–18 h at 4°C. Host cell nuclei were then stained with Hoechst dye by washing wells once with PBS, incubating cells for 15 min at RT with 1:5000 Hoechst 33342 (Invitrogen, Carlsbad, CA) in PBS-0.1% Triton X-100 and washing once with PBS. The plates were imaged using automated microscopy at DAPI and FITC wavelengths at 200X total magnification. Automated microscopy was performed as previously described (Agaisse et al., 2005).
The infections for the secondary RNAi screens were performed as described above, however at 2 h post-infection and 8 h post-infection cells from three wells of each shRNA treatment were lysed in 100 μl PBS+0.1% Triton X-100. Intracellular bacteria from the lysed cells were diluted in PBS and plated on LB agar at 37°C. After 24 h, the number of colony forming units (cfu) was determined.
1.0 × 105 HEK-293 cells per well were seeded in a 24-well plate. After 24 h, cells were transfected with shRNAs using FuGene 6. After 48 h, cells were transferred to a 6-well dish containing collagen-coated glass coverslips. Approximately 60 h after transfection, cells were infected with GFP-expressing ΔLLO L. monocytogenes at an MOI=5 in 2 ml DMEM-10. Cells were washed with PBS and DMEM-10 with 50 μg/ml gentamicin was added 1 h post-infection to kill extracellular bacteria. At 12 h post-infection, cells were fixed with 3.2% paraformaldehyde in PBS. Fixed coverslips were washed with Tris-buffered saline, pH 8.0 (TBS) supplemented with 0.1% Triton X-100 (TBS-TX) and stained for 30 min with 33 nM Texas-Red phalloidin (Molecular Probes, Eugene, OR) in TBS-TX supplemented with 1% BSA. Following an additional wash, samples were stained with Hoechst 33342 (Molecular Probes, Eugene, OR) in TBS-TX for 10 min. Coverslips were washed with TBS-TX and mounted with mounting media containing 20 mM Tris pH 8.0, 0.5% N-propyl gallate (Sigma) and 90% glycerol. Cells were imaged with a TE-300 inverted microscope (Nikon Instruments, Melville, NY) and 100X Plan Fluor objective using MetaMorph (Molecular Devices, Downington, PA) software for image acquisition and analysis.
1.0 × 105 Hct116 cells per well were seeded in a 24-well tissue culture plate in 500 μl DMEM-10. Twenty-fours hours later, cells were transfected with shRNAs using Lipofectamine 2000 (Invitrogen). For transfection of each well, 1.5 μl Lipofectamine 2000 was diluted in 50 μl OPTI-MEM and incubated at RT for 5 min. 0.5 μg shRNA plasmid DNA was diluted in 50 μl OPTI-MEM aliquots, then 50 μl of Lipofectamine 2000 mixture was added to each shRNA plasmid mixture followed by a 20 min incubation at RT. Complexes were added to cells and plates rocked to mix. Approximately 48 h after transfection, cells were resuspended in 1.5 ml DMEM-10. For microscopy infections, 100 μl/well of cells were seeded in wells of a black, clear bottom 96-well plate. For intracellular growth assays, 500 μl/well of cells were seeded in wells of a 24-well plate. Cells were incubated for 12 h, then infected with GFP-expressing L. monocytogenes strains at MOI=5. For microscopy infections, bacteria were added in 25 μl DMEM-10, followed by centrifuging plates for 1 min at 200 × g. Twenty-five microliters of DMEM-10 with 350 μg/ml gentamicin was added 1 h post-infection to kill extracellular bacteria (final concentration of 50 μg/ml gentamicin). At 10 h post-infection, cells were fixed with 3.2% paraformaldehyde in PBS for 14–18 h at 4°C. Host cell nuclei were stained with Hoechst dye by washing wells once with PBS, incubating cells for 15 min at RT with 1:5000 Hoechst 33342 in PBS-0.1% Triton X-100 and washing once with PBS. The plates were imaged using automated microscopy at DAPI and FITC wavelengths at 200X total magnification. For intracellular growth assays, media was removed from wells and bacteria were added in 500 μl DMEM-10, followed by centrifuging plates for 1 min at 200 × g. At 1 h post-infection, media with bacteria was removed and 500 μl DMEM-10 with 50 μg/ml gentamicin were added. At 2 h post-infection and 8 h post-infection, cells from three wells of each shRNA treatment were lysed in 500 μl PBS+0.1% Triton X-100. Intracellular bacteria from the lysed cells were diluted in PBS and plated on LB agar at 37°C. After 24 h, the number of cfu was determined.
HEK-293 cells were transfected as described above in a 96-well plate. Protein samples were prepared by first resuspending cells from one well in 100 μl cold PBS and then pelleting at 800 × g, 5 min, at 4°C. Cells were lysed by resuspending the pellet in lysis buffer (420 mM KCl, 50 mM HEPES, pH 8.3, 1 mM EDTA, 0.1% NP-40), incubating on ice for 15 min and pelleting the cell debris at 16,000 × g, 10 min, at 4°C. Total protein concentrations were determined using a Bradford assay (Sigma, St. Louis, MO) and equal amounts of whole cell lysate proteins were loaded in each well of an SDS-PAGE gel. Proteins were separated, transferred to PVDF membranes and Western blots were performed using a rabbit anti-RABEP1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and mouse anti-actin antibodies (Sigma) as a loading control. HRP-conjugated anti-rabbit and anti-mouse antibodies were used to detect primary antibodies (BioRad, Hercules, CA).
1.0 × 105 HEK-293 cells were seeded in a 24-well plate in 500 μl DMEM-10. Twenty-four hours later, cells were treated with FuGene 6 or transfected with luciferase or RABEP1-1 shRNA using FuGene 6 for 40 h, then transfected with LAMP1-CFP plasmid (JS106) or RAB7-CFP plasmid (JS036) DNA (Henry et al., 2006) for 8 h, then transferred to collagen-treated 18 mm coverslips in 6-well dishes for 16 h. The transfected cells were then infected with ΔLLO L. monocytogenes grown at 30°C for 3 h (OD600=0.5) at an MOI=50 in 2 ml DMEM-10. Bacteria were centrifuged onto cells at 600 × g for 5 min at RT. For infections of 60 min of less, cells were washed extensively and cells were fixed with 3.2% paraformaldehyde in PBS at indicated time points post-infection. For the 2-h infections, cells were washed with PBS at 60 min post-infection and DMEM-10 with 50 μg/ml gentamicin was added to kill extracellular bacteria. At 2 h post-infection, cells were fixed in 3.2% paraformaldehyde in PBS. Cells from all infections were fixed for 16 to 18 h. Bacteria in cells were stained with a rabbit anti-L. monocytogenes antibody (Difco) and Rhodamine Red X-conjugated donkey anti-rabbit IgG secondary antibody (Jackson ImmunoResearch, West Grove, PA). Coverslips were mounted with mounting media containing 20 mM Tris pH 8.0, 0.5% N-propyl gallate (Sigma) and 90% glycerol. Cells were imaged with a Nikon TE-2000 inverted microscope using CFP and RFP filter sets with a 100X Plan Apo 1.4 NA objective in z-series with 0.25 μm z-steps according to the recommended Nyquist z-sampling distance. Images were acquired using MetaMorph software. CFP images were processed with AutoDeblur deconvolution software using adaptive (blind) point spread function (PSF) prior to analysis (MediaCybernetics, Bethesda, MD). Bacteria were determined to be CFP-positive if there was a region of enrichment (defined as increased fluorescence intensity) encircling the location of the bacteria in the CFP images. In establishing the criteria for CFP-positive bacteria, we measured pixel intensity of regions immediately surrounding the bacteria and regions in random locations throughout the cell. Additionally, to check accuracy, other individuals scored selected images without knowledge of the experimental condition of the images. Quantitative results are presented as the mean ± SEM. For the 2-h infections, approximately 50 intracellular bacteria were examined for each condition in at least three independent experiments. The total numbers of intracellular bacteria counted were as follows: 216 (FuGene), 139 (Luciferase shRNA), and 176 (RABEP1 shRNA). For the 60-min time course experiment, all intracellular bacteria were examined from at least 10 CFP-transfected cells at each time point for each vacuolar maturation marker in 5 independent experiments. The total numbers of intracellular bacteria counted for the LAMP1-CFP experiments with FuGene treatment were: 19 (15 min), 49 (30 min), 56 (45 min), and 99 (60 min). For the LAMP1-CFP experiments with RABEP1 shRNA treatment the numbers counted were: 25 (15 min), 45 (30 min), 65 (45 min), and 80 (60 min). The total numbers of intracellular bacteria counted for the RAB7-CFP experiments with FuGene treatment were: 40 (15 min), 51 (30 min), 73 (45 min), and 119 (60 min). For the RAB7-CFP experiments with RABEP1 shRNA treatment the numbers counted were: 15 (15 min), 51 (30 min), 76 (45 min), and 106 (60 min).
1.0 × 105 HEK-293 cells were seeded in a 24-well plate in 500 μl DMEM-10. Twenty-four hours later, cells were treated with FuGene 6 or transfected with RABEP1-1 shRNA using FuGene 6 for 60 h, then transferred to collagen-treated 18 mm coverslips in 6-well dishes for 16–18 h. The transfected cells were then infected with ΔLLO L. monocytogenes grown at 30°C for 3 h (OD600=0.5) at an MOI=50 in 2 ml DMEM-10. Bacteria were centrifuged onto cells at 600 × g for 5 min at RT. For infections of 60 min or less, cells were washed extensively and fixed with 3.2% paraformaldehyde in PBS at the indicated time points post-infection. For the 2-h infections, cells were washed with PBS at 60 min post-infection and DMEM-10 with 50 μg/ml gentamicin was added to kill extracellular bacteria. At 2 h post-infection, cells were fixed in 3.2% paraformaldehyde in PBS. Cells from all infections were fixed for 16–18 h at 4°C. Any remaining extracellular bacteria were stained without cell permeabilization in PBS + 0.1% BSA with a rabbit anti-L. monocytogenes antibody (Difco) and goat anti-rabbit IgG AlexaFluor350-conjugated secondary antibody (Invitrogen). Following extensive washing, the cells were permeabilized with saponin buffer and stained for intracellular L. monocytogenes and LBPA simultaneously. First, cells were incubated with rabbit anti-L. monocytogenes antibody and mouse anti-LBPA antibody (Kobayashi et al., 1998) in PBS + 0.05% saponin (for permeabilization). After washing with PBS, cells were incubated with donkey anti-rabbit IgG FITC-conjugated secondary antibody (Jackson ImmunoResearch) and donkey anti-mouse Rhodamine Red X-conjugated secondary antibody (Jackson ImmunoResearch) in PBS. Coverslips were mounted with mounting media containing 20 mM Tris, pH 8.0, 0.5% N-propyl gallate (Sigma) and 90% glycerol. Cells were imaged with a Nikon TE-2000 inverted microscope with a 100X Plan Apo objective in z-series with 0.25 μm z-distance steps using DAPI (AlexaFluor 350), FITC and Texas Red filter sets. Images were acquired using MetaMorph software. AlexaFluor 350-negative (intracellular) bacteria were determined to be LBPA-positive if there was a region of enrichment (defined as increased fluorescence intensity) encircling the location of the bacteria in the Texas Red images. All intracellular bacteria were analyzed from at least 20 host cells at each time point in 4 independent experiments. The total numbers of intracellular bacteria counted for the LBPA experiments with FuGene treatment were: 75 (30 min), 155 (60 min), and 162 (120 min). For the LBPA experiments with RABEP1 shRNA treatment the numbers counted were: 67 (15 min), 153 (30 min), and 146 (60 min).
We thank current and past members of the Higgins lab for helpful discussions, especially Heather Kamp and Christine Alberti-Segui. We would also like to thank Jing Chen for assistance with the construction of the vesicular trafficking shRNA library and Jianping Jin for construction of the pBS-mir vector. We thank Dr. Joel Swanson for the kind gift of LAMP1-CFP and RAB7-CFP constructs and Dr. Jean Gruenberg for making the LBPA antibody available to us. Microscopy images to determine the association of markers of vacuolar maturation for this study were acquired and analyzed in the Nikon Imaging Center at Harvard Medical School. We would like to thank the imaging center staff for their advice and assistance. This work was supported by U.S. Public Health Service grants AI-053669 and AI-056446 (DEH) and grant AG-011085 (JWH) from the National Institutes of Health. LSB was a Howard Hughes Medical Institute Predoctoral Fellow.