Following uptake into target cells, the bacterium is internalized into a compartment that escapes the endocytic pathway (8, 9). Fifteen minutes after internalization, the vacuolar membrane lacks plasma membrane or endocytic markers and is surrounded by mitochondria and small vesicles (7). Based on the thickness of their membranes these vesicles are thought to be derived from the endoplasmic reticulum (ER) (22). In addition, L. pneumophila vacuoles intercept vesicular traffic from ER exit sites and vesicle budding from the ER appears to be required for establishment of the replication vacuole (11). Six hours after internalization, ER surrounds the vacuole and the bacterium starts to multiply (21, 22). As replication proceeds, greater than 50% of the L. pneumophila vacuoles acquire markers of endocytic organelles (20) and the host cell is eventually lysed.

The kinetics of ER recruitment and the nature of the vesicles recruited to the L. pneumophila vacuole are still unclear. ER markers have been detected around the vacuoles 6 h after infection by immunofluorescence and electron microscopy analysis (21, 22). However, it has not been shown that ER markers are associated with the vacuolar membrane at early stages of vacuolar biogenesis. In this study, we analyzed preparations of isolated L. pneumophila vacuoles to determine how soon after bacterial contact with host cells that ER-derived components are associated with these vacuoles and to identify the host factor(s) involved in this process.

A sucrose gradient-based procedure was used to isolate the L. pneumophila vacuoles. A total of 10⁸ differentiated U937 cells were incubated for 10 min or 1 h with either wild-type L. pneumophila or with the avirulent dot/icm mutant (dotA) (1), both expressing green fluorescent protein (GFP) (19), at a multiplicity of infection (MOI) of 5 to maximize the yield of vacuoles. As L. pneumophila has maximum potential for promoting intracellular growth after entering the postexponential phase, we used bacteria grown to this density for all the infections described in this study. The infected cells were suspended in 2 ml of homogenization buffer (20 mM HEPES-KOH [pH 7.2], 250 mM sucrose, 0.5 mM EGTA) and lysed in a Dounce homogenizer, and the unbroken cells and nuclei were pelleted by centrifugation (3 min, 1,500 rpm). The postnuclear supernatant (PNS) was loaded onto a discontinuous sucrose gradient (Beckman centrifuge tubes, 14 by 89 mm) consisting of 6 ml of 14% (wt/wt) sucrose layered on top of 2 ml of 50% (wt/wt) sucrose in HEPES buffer (20 mM HEPES-KOH [pH 7.2], 0.5 mM EGTA) and subjected to centrifugation at 600 × g for 90 min at 4°C. The L. pneumophila vacuoles were identified in the 14 to 50% interface fraction of the sucrose gradient based on the presence of the GFP marker (Fig. (Fig.1A).1A). The vacuoles were allowed to bind poly-l-lysine-coated coverslips prior to fixation (12) and probing with the indicated antibody. Immunofluorescence analysis of L. pneumophila vacuoles was the same throughout the study. The vacuolar membrane integrity was verified by immunostaining of the vacuoles containing bacteria by using a polyclonal anti-L. pneumophila antibody (Fig. (Fig.1B).1B). In the absence of vacuolar membrane permeabilization, less than 20% of the L. pneumophila vacuoles stained positively for the bacteria. In contrast, pretreatment with a permeabilization reagent (−20°C methanol for 5 s) resulted in staining of 100% of the vacuoles (Fig. (Fig.1B).1B). We also tested if the lysosomal marker LAMP1 was recruited to the L. pneumophila vacuole (anti-human CD107a; BD Pharmingen). Only 15% of vacuoles harboring bacteria with an intact Dot/Icm system were positive for LAMP1, whereas 45% of the vacuoles harboring a dotA mutant strain showed positive staining (data not shown). As the isolation procedure left the vacuoles largely intact and since the lack of LAMP1 recruitment to the vacuole is similar to what is observed in infected bone marrow macrophages, we next analyzed the recruitment of proteins associated with the early secretory system to nonpermeabilized L. pneumophila vacuoles.

Antibodies directed against the Golgi-associated proteins p115 (G. Watters) and GM130 (BD Bioscience) failed to reveal recruitment of Golgi material at any time point after infection (data not shown). In contrast, calnexin (rabbit polyclonal antibody from Stressgen), a membrane ER protein, was detected on the surface of 77% of the vacuoles isolated from macrophages infected with wild-type L. pneumophila as early as 10 min after infection (Fig. (Fig.2A).2A). The punctate pattern of the staining was consistent with the previous hypothesis that vesicles, recruited to the L. pneumophila vacuoles 15 min after infection, are likely to be ER derived (22) (Fig. (Fig.2B).2B). The number of calnexin-positive vacuoles increased to 92% 1 h after infection (Fig. (Fig.2A),2A), and the pattern of staining changed significantly from the earlier time, as the L. pneumophila vacuoles were completely surrounded by a thick layer of calnexin-containing material (compare Fig. Fig.2B2B to C). In contrast, less than 15% of the vacuoles isolated from macrophages infected with the dotA mutant showed a positive staining at either time point (Fig. (Fig.2),2), indicating that ER recruitment was dependent on an intact Dot/Icm secretion system. This lack of coating by ER material may explain why vacuoles harboring a dotA mutant were twice as likely to be damaged during isolation as Dot/Icm-intact L. pneumophila (Fig. (Fig.1B).1B). ER material surrounding the vacuole may provide significant protection against mechanical or osmotic lysis.

The rapid accumulation of calnexin about the vacuoles is in apparent contradiction with previous reports indicating that calnexin was only associated with intracellular L. pneumophila 6 h after internalization (11). This discrepancy could be due to the fact that the previous study used murine bone marrow-derived macrophages instead of human macrophages, but we favor the hypothesis that the small amount of calnexin recruited during the earliest times after uptake can only be detected when the vacuoles are isolated from other cellular material. Our results also indicate that ER recruitment continues throughout the first hour of infection, as ER accumulates around the L. pneumophila vacuole during this period of time. Moreover, this process is dependent on the presence of an intact Dot/Icm system, indicating that the effector molecule(s) secreted by this apparatus is responsible for the proper recruitment of ER-derived material.

We believe that the recruitment of ER-derived material to the L. pneumophila vacuole differs from the observed ER-mediated phagocytosis previously seen with latex beads (5), since the process described here is Dot/Icm dependent (Fig. (Fig.2).2). The absence of calnexin recruitment to the dotA mutant vacuoles could be due to rapid interaction of the vacuoles with the endocytic pathway, or it could be that there was very little recruitment of ER. However, calnexin was reported to be easily detected on the surface of latex bead phagosomes as early as 15 min after uptake (5), strongly suggesting that the dotA mutant vacuoles never interact significantly with ER-derived material. ER-mediated phagocytosis is also proposed to play a role in presentation of antigens by major histocompatibility complex class I molecules (10). However, in the case of wild-type L. pneumophila, the bacteria are not degraded in the vacuole, suggesting that the ER-derived compartments harboring latex beads are very different in nature from vacuoles bearing L. pneumophila.

Recruitment of ER-associated material to the L. pneumophila vacuole may involve hijacking of host proteins that regulate vesicular traffic between the ER and the Golgi apparatus. The Rabs and the SNAREs are the largest families of proteins known to be involved in membrane trafficking (6, 17). They respectively regulate the traffic of vesicles and participate in their docking and fusion to their target membranes. Two Rab proteins, Rab1a/b and Rab2 (13, 23), and at least one SNARE complex, composed of the v-SNARE Sec22b and three t-SNAREs (syntaxin 5, membrin, and rBet1), have been shown to regulate the ER-to-Golgi traffic (26). Therefore, we investigated the recruitment of these proteins to the L. pneumophila vacuoles.

One hour after infection, 43% of the vacuoles isolated from macrophages infected with wild-type L. pneumophila were positive for the small GTPase Rab1 (rabbit polyclonal antibody from Calbiochem) (Fig. (Fig.3A).3A). In contrast, no staining for Rab2 could be detected (data not shown) (the mouse monoclonal anti-Rab2 antibody was a kind gift from E. J. Tisdale, Wayne State University School of Medicine). At the same time point, 75% of phagosomes harboring a wild-type L. pneumophila strain were also positive for the v-SNARE protein Sec22b (Fig. (Fig.3A)3A) (the rabbit polyclonal anti-Sec22b was a kind gift from J. Hay, University of Michigan). Moreover, the recruitment of both Rab1 and Sec22b to the L. pneumophila vacuole was dependent on the presence of the Dot/Icm complex, since less than 10% of the vacuoles isolated from macrophages infected with the dotA mutant showed positive staining for either protein (Fig. (Fig.3A).3A). The pattern of both Rab1 and Sec22b staining was punctate even 1 h after infection (Fig. (Fig.3B3B and C), but the staining pattern of Rab1 was significantly less densely packed than that observed with calnexin or Sec22b. The vacuoles were also analyzed 14 h after the initiation of the infection. For this purpose, the cells were infected and lysed as described for the PNS preparation but unbroken cells and nuclei were not removed. The cell lysates were allowed to directly bind coverslips and analyzed as described for isolated vacuoles. As shown in Fig. Fig.3D,3D, wild-type L. pneumophila vacuoles in which the bacteria were undergoing extensive replication were still labeled with calnexin, Rab1, and Sec22b, suggesting that these markers are associated with L. pneumophila vacuole throughout the replication cycle of the bacteria.

As the two Rab1 effectors p115 and GM130 are absent from the vacuole, other factors may be involved in the recruitment of ER-derived material. ER-derived vesicles dock and fuse with the Golgi apparatus by formation of a complex between Sec22b and three t-SNAREs (syntaxin 5, membrin, and rBet1) present on the Golgi target membrane (26). We performed immunofluorescence analysis using antibodies directed against these t-SNARES (the rabbit polyclonal anti-syntaxin 5, antimembrin, and anti-rBet1 antibodies were kind gifts from J. Hay, University of Michigan), but none was found associated with the surface of the vacuole (data not shown). Therefore, if Sec22b plays a role in the formation of the L. pneumophila replication vacuole, it may bind to unidentified t-SNARE or t-SNARE-like proteins encoded by the host or translocated by the bacterium across the vacuolar membrane.

Rab proteins are known to cycle between a GTP-bound active form and a GDP-bound inactive form, the latter of which acts as a dominant-negative inhibitor when overexpressed in mammalian cells such as COS1 cells (African green monkey kidney cells, transformed by an origin-defective mutant of simian virus 40). To test for growth in COS1 cells, postexponential-phase L. pneumophila strains were introduced onto monolayers of 10⁵ cells in 24-well dishes at an MOI of 0.05 by centrifugation for 5 min at 1,000 rpm. A low MOI was used to maximize the efficiency of initiation of intracellular growth (3). After incubation for 2 h at 37°C in the presence of 5% CO₂, the monolayers were washed three times to remove any extracellular bacteria and intracellular growth of L. pneumophila was quantified (1). As shown in Fig. Fig.4A,4A, L. pneumophila replication within COS1 cells was found to be similar to that observed in macrophage-like cell types, with intracellular growth dependent on a functional Dot/Icm secretion apparatus. This result indicated that COS1 cells are suitable to test the effect of overexpression of Rab1 isoforms on the intracellular multiplication of L. pneumophila. For this purpose, 3 × 10⁴ COS1 cells were seeded onto glass coverslips in 24-well dishes and were transfected the next day with 1 μg of plasmid DNA, encoding GFP fusions to Rab1a derivatives having either the wild-type or dominant inhibitory (Rab1aS25N) sequences, using the calcium phosphate transfection method according to the manufacturer's instructions (Invitrogen). Twenty-four hours after transfection, the bacteria were incubated with the cells, as described for the intracellular growth assay, except that the MOI was 5. After incubation for 2 h at 37°C in the presence of 5% CO₂, the monolayers were washed to synchronize the infection and were further incubated for 14 h at 37°C in 5% CO₂ prior to fixation. The samples were stained for extracellular and total bacteria and processed for immunofluorescence microscopy. The efficiency of replication vacuole formation was determined by visual inspection by counting the number of bacteria per vacuole, and the data were quantified by counting 100 phagosomes per coverslip. The average and standard error were calculated from three coverslips, and the P value was calculated by using the unpaired Student's t test (two samples assuming unequal variances). As shown in Fig. Fig.4B,4B, L. pneumophila growth was not affected by overexpression of either GFP or wild-type GFP-Rab1a. However, overexpression of the GDP-bound form of GFP-Rab1a (Rab1aS25N) led to a significant (P < 0.005) reduction of large replication vacuole formation. In comparison to the results from analysis of isolated vacuoles from U937 cells (see above), it was difficult to detect recruitment of the wild-type form of GFP-Rab1a to the L. pneumophila vacuole when the protein was overexpressed in intact COS1 cells, although the vacuoles were found in an area of the cell that was rich in Rab1. In contrast, the GDP-bound form of GFP-Rab1a (Rab1aS25N) strongly accumulated and labeled the L. pneumophila vacuole (Fig. (Fig.4C4C).

Our results indicate that Rab1 is required for the optimal intracellular growth of L. pneumophila in COS1 cells. Since Rab1 has an essential role in the docking of ER-derived vesicles with their target membrane, it is possible that Rab1 provides a similar role for L. pneumophila. Rab1 may direct the association of ER-derived vesicles with the L. pneumophila vacuole, thus assisting the formation of an environment permissive for growth. It has been previously reported that the GDP-bound form of Rab1 localizes to the cytoplasm of the cell as well as to punctate structures that are likely to be fragments of ER, ERGIC, and Golgi membranes (14). It is possible that when the inactive form of the Rab1 is overexpressed, ER-derived material rich in GDP-bound Rab1 is recruited to the L. pneumophila vacuole. The irreversible association of inactive Rab1 to the vacuole may interfere with cycles of recruitment of ER and, consequently, with the ability of L. pneumophila to multiply.