Surface-Associated Hsp60 Chaperonin of Legionella pneumophila Mediates Invasion in a HeLa Cell Model

Abstract

HeLa cells have been previously used to demonstrate that virulent strains of Legionella pneumophila (but not salt-tolerant avirulent strains) efficiently invade nonphagocytic cells. Hsp60, a member of the GroEL family of chaperonins, is displayed on the surface of virulent L. pneumophila (R. A. Garduño et al., J. Bacteriol. 180:505–513, 1988). Because Hsp60 is largely involved in protein-protein interactions, we investigated its role in adherence-invasion in the HeLa cell model. Hsp60-specific antibodies inhibited the adherence and invasiveness of two virulent L. pneumophila strains in a dose-dependent manner but had no effect on the association of their salt-tolerant avirulent derivatives with HeLa cells. A monospecific anti-OmpS (major outer membrane protein) serum inhibited the association of both virulent and avirulent strains of L. pneumophila to HeLa cells, suggesting that while both Hsp60 and OmpS may mediate bacterial association to HeLa cells, only virulent strains selectively displayed Hsp60 on their surfaces. Furthermore, the surface-associated Hsp60 of virulent bacterial cells was susceptible to the action of trypsin, which rendered the bacteria noninvasive. Additionally, pretreatment of HeLa cells with purified Hsp60 or precoating of the plastic surface where HeLa cells attached with Hsp60 reduced the adherence and invasiveness of the two virulent strains. Finally, recombinant Hsp60 covalently bound to latex beads promoted the early association of beads with HeLa cells by a factor of 20 over bovine serum albumin (BSA)-coated beads and competed with virulent strains for association with HeLa cells. Hsp60-coated beads were internalized in large numbers by HeLa cells and remained in tight endosomes that did not fuse with other vesicles, whereas internalized BSA-coated beads, for which endocytic trafficking is well established, resided in more loose or elongated endosomes. Mature intracellular forms of L. pneumophila, which were up to 100-fold more efficient than agar-grown bacteria at associating with HeLa cells, were enriched for Hsp60 on the bacterial surface, as determined by immunolocalization techniques. Collectively, these results establish a role for surface-exposed Hsp60 in invasion of HeLa cells by L. pneumophila.

Legionella pneumophila is a gram-negative, facultative intracellular parasite of freshwater protozoa (natural hosts) and an opportunistic human pathogen that causes the atypical pneumonia known as Legionnaires’ disease (17, 47, 52). The fact that alveolar macrophages are the target cells during human infection initially determined the preferential use of macrophages, monocytes, or macrophage-like cell lines to study the intracellular growth and pathogenesis of L. pneumophila. However, investigators have increasingly used natural hosts or nonphagocytic mammalian cells to study specific aspects of the early interactions between L. pneumophila and host cells (1, 2, 5, 16, 20, 23, 25, 41, 42, 55). We have characterized a HeLa cell model to specifically study the cell invasion mechanisms of L. pneumophila (22). In this model, wild-type strains of L. pneumophila efficiently invade HeLa cells, whereas nonvirulent mutants (isolated by their tolerance to NaCl) are always less invasive (12, 22); yet differences in invasiveness among salt-tolerant mutants have been detected (22). Therefore, the invasion factor(s) of L. pneumophila must be differentially displayed in virulent and avirulent strains. Also, we have determined that this invasion factor(s) must be constitutively expressed in virulent strains, since invasiveness was not significantly affected by treatment with chloramphenicol, an inhibitor of prokaryotic protein synthesis (22). Salt-tolerant avirulent strains of L. pneumophila are also typically unable to inhibit phagosome-lysosome fusion in macrophages or to target the phagosomes in which they reside to the endoplasmic reticulum (4, 15, 34, 49, 53, 54). Therefore, the multiple genetic defects associated with the phenotypes of salt tolerance and avirulence must involve (or affect) a factor that either mediates both invasiveness and phagosome trafficking or coordinately regulates these virulence traits.

In contrast to other gram-negative pathogens, few surface-exposed proteins have been identified in L. pneumophila. Among these, we have characterized the major outer membrane protein OmpS (7, 8, 32, 33), which may play a role in adherence to host cells (6), and the GroEL homologue Hsp60, an essential heat shock protein that is expressed on the surface of virulent L. pneumophila (21). A surface location is consistent with previous observations indicating a role for Hsp60 in the interaction of L. pneumophila with host cells. These include up-regulation of Hsp60 synthesis following association with host cells (15), an increased level of surface-exposed Hsp60 (15, 21), and the release of Hsp60 into newly formed and mature phagosomes (15, 21, 31), events that correlate with the ability of virulent L. pneumophila to abrogate phagosome-lysosome fusion (15). In contrast, nonvirulent L. pneumophila (also internalized by macrophages) do not up-regulate or release Hsp60 and do not abrogate phagosome-lysosome fusion (15). Surface-exposed Hsp60 may also play an immunomodulatory function, since it has been shown that L. pneumophila Hsp60 induces synthesis of interleukin-1β in macrophages, through a mechanism that involves ligand-receptor interactions in the absence of Hsp60 internalization (46). Based on these experimental results and the generalized ability of Hsp60 chaperonins to interact with proteins, we have investigated the possibility that surface-exposed Hsp60 plays a role in the adherence and invasiveness of L. pneumophila. Here we demonstrate, through five different lines of experimentation, that the surface-exposed Hsp60 of two well-characterized virulent strains of L. pneumophila acts as an adhesin-invasin and mediates the internalization and unique trafficking of latex particles in HeLa cells. Thus, the L. pneumophila Hsp60 plays previously unrecognized virulence roles that may have important implications in the L. pneumophila pathogenesis.

MATERIALS AND METHODS

L. pneumophila strains and culture media.

L. pneumophila Philadelphia 1 (Lp1-Vir) and the nonvirulent, salt-tolerant strain Lp1-AVir were obtained from the Centers for Disease Control and Prevention (Atlanta, Ga.). A previously described clinical isolate from Victoria General Hospital (Halifax, Nova Scotia, Canada), 2064 (serogroup 1, Oxford), and its avirulent isogenic derivative 2064M (14) were also used. All strains were routinely kept as frozen stocks at −70°C. Frozen stocks were grown on buffered charcoal-yeast extract agar (BCYE) (43) for 3 to 5 days at 37°C in a humid incubator and passaged once on BCYE before use as outlined previously (22). For some applications, buffered yeast extract (BYE) liquid medium (same formulation as BYCE, with charcoal and agar omitted) was used.

Buffer solutions.

Phosphate (10 mM)-buffered saline (140 mM NaCl, 10 mM KCl), pH 7.4 (PBS), was routinely used for washing bacteria and HeLa cells, and sodium cacodylate (0.1 M, pH 7.0) buffer (CB) was used for electron microscopy.

Culture of HeLa cells.

Stocks of HeLa cells were either cultured in 25-cm² flasks containing 7 ml of complete minimal essential medium or grown in suspension, in 200-ml spinner bottles containing 60 to 80 ml of complete Dulbecco’s modified Eagle’s medium (DMEM), as previously described (22). Subconfluent monolayers of freshly attached, spinner bottle-grown cells were routinely established by adding 0.5 ml of a HeLa cell suspension in DMEM with no antibiotics (10⁶ HeLa cells per ml) to each well of 12-well tissue culture multiwell plates (Falcon) (i.e., 5 × 10⁵ HeLa cells/well) and incubating the wells at 37°C and in 5% CO₂ in a humid incubator for at least 1 h.

Invasion assays.

Invasion assays were carried out as detailed elsewhere (22). Briefly, overnight BCYE-grown cultures were resuspended in BYE broth, and immediately before use, a bacterial inoculum with a final optical density (640 nm) of 0.1 was prepared in DMEM with no antibiotics (∼10⁸ bacteria/ml). Freshly attached HeLa cells were infected with 100 μl of the bacterial inoculum. Three hours after infection, monolayers were washed six times with PBS and then either lysed to determine total associated bacteria or treated for 1.5 h with gentamicin (100 μg/ml) to kill extracellular bacteria. For lysis, monolayers were treated for ∼3 min with 0.1 ml of distilled deionized water containing 0.05% Triton X-100 and then with 0.9 ml of distilled deionized water, with vigorous agitation with the aid of a pipettor. After gentamicin treatment, the antibiotic was removed with three PBS washes, and cells were lysed as described above. All cell lysates were diluted in distilled deionized water and plated on BCYE to quantify CFU after an incubation period of 4 to 5 days.

Purification of recombinant Hsp60.

Escherichia coli PSH16 (strain JM109 harboring the L. pneumophila htpAB operon in pUC19), as well as its growth and overexpression of Hsp60, has been previously described (30). Recombinant Hsp60 was purified from a crude lysate of PSH16 by ammonium sulfate precipitation and ion-exchange chromatography in a DE-52 (Whatman) column (44, 57), followed by dialysis, concentration by ultrafiltration (Amicon), and sterilization by filtration through a 0.2-μm-pore-size membrane (Nalgene). Purity of the preparation was evaluated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), Coomassie blue R-250 staining, and immunoblotting as outlined previously (21). The protein concentrations of our stocks of purified Hsp60 and of the preparations used to coat latex beads (see below) were determined by the Bradford colorimetric protein assay of Bio-Rad Laboratories, using bovine serum albumin (BSA) as the standard.

Effect of antisera on cell association and invasion.

Monospecific rabbit antisera raised against Hsp60 (30) and OmpS (7) were diluted in heat-inactivated neonatal bovine serum (GIBCO), and a predetermined amount of bacterial cells (pelleted by centrifugation from a suspension in BYE broth) was resuspended into the corresponding antiserum dilutions. The amounts of bacterial cells added to the different serum dilutions were adjusted to yield a final inoculum of 10⁷ bacteria per well and a ratio of 10⁹ bacteria per ml of diluted serum. The mixtures were incubated for 1 h at 37°C before infection of the HeLa cell monolayers. Control bacterial suspensions were routinely preincubated in undiluted heat-inactivated neonatal bovine serum. However, a number of assays run with neat rabbit preimmune serum as controls showed no differences between these controls and the heat-inactivated neonatal bovine serum controls. In some experiments, the anti-Hsp60 monoclonal antibody GW2X4B8B2H6 (27) was used as neat hybridoma cell culture supernatant; control suspensions for these experiments were made in DMEM with 10% heat-inactivated neonatal bovine serum. Antisera or control sera were not removed from the bacterial inoculum before the infection of HeLa cells.

Effect of purified Hsp60 in cell association and invasion.

Before infection with our virulent L. pneumophila strains, HeLa cells were preincubated at 37°C for 1 h with different amounts of Hsp60 solubilized in complete DMEM with no antibiotics. We then performed standard invasion assays, maintaining the concentration of soluble Hsp60 throughout the 3-h infection period. Control wells were preincubated and run in the presence of BSA at a concentration equal to the maximum concentration of soluble Hsp60 used.

The possible existence of specific receptors for Hsp60 was examined in a receptor modulation assay. In this type of assay, the cell membrane of host cells attached onto a ligand-coated surface is polarized into two regions: a receptor-rich, adherent region facing the ligand-coated surface, and a receptor-depleted region facing the extracellular medium (40). This down-modulation of receptors on the exposed membrane inhibits subsequent interactions of host cells with bacteria (51). Subconfluent monolayers of freshly attached HeLa cells were established in wells precoated with different amounts of purified Hsp60, and then standard invasion assays were run. Precoating of wells was carried out by adding 1.0 ml of the corresponding protein solution in PBS to each well; the total amount of protein per well was maintained at 120 μg by the addition of different amounts of BSA. Control wells were coated with BSA only. Binding of protein to the wells was allowed to occur overnight at 4°C and then for 2 h at 37°C.

Assays with latex beads.

Carboxylated latex beads (0.8 μm in diameter) were obtained from Polysciences (Warrington, Pa.) at a density of 2.5% solids, ∼10¹¹ beads per ml. One milliliter of beads was coated in the presence of 600 μg of purified Hsp60 or BSA, using a carbodiimide kit for carboxylated microparticles (Polysciences) as recommended by the manufacturer. Similar amounts of Hsp60 or BSA (i.e., 200 to 250 μg of protein per 10¹¹ beads, or 2 to 2.5 fg per bead) were bound to the activated beads. As a reference, we estimated from SDS-polyacrylamide gels and immunoblots run with known numbers of bacteria and known amounts of Hsp60 that each bacterial cell of L. pneumophila would contain ∼10 fg of Hsp60. After thorough washing, the coated beads were resuspended and stored in PBS containing 0.1% BSA. Subconfluent monolayers of 10⁶ freshly attached HeLa cells were established in six-well plates. Beads were added at a ratio of 100 to 200 per HeLa cell, and plates were incubated for different times at 37°C in a 5% CO₂ atmosphere. Cells were then washed once in PBS and fixed for 2 h in 2.5% glutaraldehyde in CB. One half of each well was scraped, and the scraped cells were transferred to separate tubes for preparation for transmission electron microscopy (TEM) (see below). The remaining cells were prepared for light microscopy (see below).

Beads were also used in standard invasion assays to test their ability to compete with our virulent strains of L. pneumophila. In these assays, HeLa cells were preincubated for 1 h at 37°C with 10 μl of Hsp60-coated or BSA-coated beads (∼1,000 to 2,000 beads/cell) before addition of the bacterial inoculum. Beads were not removed upon addition of bacteria but kept throughout the infection period.

Trypsin treatment of L. pneumophila.

Bacterial cells harvested from BCYE plates were suspended in BYE broth to a final optical density (640 nm) of 1 (∼10⁹ bacteria/ml) and split in several 1-ml aliquots in Eppendorf microcentrifuge tubes. Bacteria were pelleted by centrifugation, and pellets were then resuspended in 90 μl of trypsin solutions of different concentrations, previously prepared in DMEM with no serum and no antibiotics. After a 30-min incubation, each tube received 10 μl of trypsin inhibitor from soybean (Sigma) (at a concentration of 25 mg/ml in DMEM) and 25 μl of 5× Laemmli sample buffer (37). Sample lysates (15 μl/lane) were then subjected to SDS-PAGE in 12% acrylamide, vertical slab minigels (Bio-Rad). Gels were stained with Coomassie blue G250 or immunoblotted, as recently described, to immunostain Hsp60 with monoclonal antibody GW2X4B8B2H6 (21).

To assess the effect of trypsin treatment on the ability of L. pneumophila to associate with HeLa cells, standard invasion assays were run with bacterial cells pretreated for 1 h with 2.5 mg of trypsin per ml of DMEM (final concentration) in the presence of chloramphenicol (100 μg/ml). Control samples were pretreated with chloramphenicol only. Trypsin-treated and control bacteria were washed once with PBS and resuspended in DMEM with 10% heat-inactivated neonatal calf serum before infection of HeLa cells. A chloramphenicol concentration of 50 μg/ml was maintained throughout the invasion assays. Trypsin did not affect the viability of bacteria, in the presence or absence of chloramphenicol, as determined by a standard dilution plate method and colony counting.

Light microscopy.

Live, unstained cells were routinely examined under phase-contrast microscopy in a Nikon Diaphot inverted microscope. Cells incubated with latex beads were processed directly in the multiwell plates as follows. Cells were washed twice in PBS, permeabilized in 80% acetone for 10 min, air dried, and stained with 0.44% Giemsa stain in distilled water. After staining, circles (5 mm in diameter) were cut from the bottom of the wells with a cork-boring machine and mounted on glass slides with Permount. Association of beads with HeLa cells was evaluated by directly quantifying the percentage of HeLa cells bearing beads and the average number of beads per cell with beads. These two indexes were combined to calculate the number of beads per each cell in the monolayer.

TEM.

Cells that had been fixed in 2.5% glutaraldehyde in CB and scraped from the wells were pelleted by centrifugation (20 min at 300 × g) and washed three times with CB. Pellets were then postfixed, stained en bloc, dehydrated, embedded in TAAB resin (Marivac), ultrathin sectioned, and stained, all as detailed previously (21). Specimens of HeLa cells infected with mature intracellular bacteria (see below), and to be used for immunogold labeling, were also processed as previously described (21). Specimens were observed in a Philips EM300 electron microscope at an accelerating voltage of 60 kV.

Assays with mature intracellular L. pneumophila.

Mature intracellular L. pneumophila (bacteria that had grown intracellularly in HeLa cells and differentiated into a distinct developmental form) were isolated from infected HeLa cell cultures in a continuous density gradient of Percoll as previously reported (22). Mature bacteria were allowed to infect HeLa cells in standard invasion assays, and their adherence to and invasiveness of HeLa cells were assessed in relation to BCYE-grown legionellae. In some instances, specimens of HeLa cells infected with mature intracellular bacteria were processed for immunogold TEM as described above.

RESULTS

Hsp60 antibodies selectively inhibited the association of virulent L. pneumophila with HeLa cells.

Preincubation of the virulent strains Lp1-Vir and 2064 with different dilutions of an anti-Hsp60 rabbit serum caused a decrease in their ability to associate with HeLa cells (Fig. 1a). However, no inhibition was apparent upon preincubation of the avirulent strains Lp1-AVir and 2064M with the Hsp60 antiserum (Fig. 1a). A similar result was obtained by the pretreatment of Lp1-Vir and Lp1-AVir with monoclonal antibody GW2X4B8B2H6 (Fig. 1a, inset). On the other hand, treatment of bacterial cells with the OmpS antiserum caused a decrease in the cell association abilities of all strains (Fig. 1b), suggesting that OmpS was accessible to antibodies in all the strains. Importantly, simultaneous treatment of our virulent legionellae with both antisera did not have an additive inhibitory effect (not shown). We determined the effects of antibodies on the invasiveness of strains 2064 (virulent) and 2064M (avirulent), the invasiveness of strain 2064M being always much lower and variable than that of 2064 (22). The invasiveness of 2064 was clearly inhibited by both the Hsp60 and OmpS antisera in a dose-dependent manner, whereas the inhibitory effects of both antisera on the invasiveness of 2064M were neither significant nor dose dependent (Fig. 1c). Collectively, these neutralization assays showed the differential effects of Hsp60 and OmpS antisera on the association and invasion of the virulent and isogenic avirulent strains tested.

FIG. 1.

Effects of antibodies on interactions of different strains of L. pneumophila with HeLa cells. (a) Effects of different dilutions of an Hsp60 rabbit antiserum on the association of the virulent strains (solid lines) Lp1-Vir (○) and 2064 (•), or the avirulent strains (broken lines) Lp1-AVir (○) and 2064M (•), with HeLa cells. The inset indicates the effect of monoclonal antibody GW2X4B8B2H6 on the association of Lp1-Vir and Lp1-AVir with HeLa cells. (b) Effects of different dilutions of an OmpS rabbit antiserum on the association of the four L. pneumophila strains described above. (c) Effects of different dilutions of an Hsp60 antiserum (solid lines) or an OmpS antiserum (broken lines) on the invasiveness of the virulent strain 2064 (•) and the avirulent strain 2064M (○). Graph points represent the means ± standard deviations (vertical bars) from single experiments run in duplicate (n = 2). Asterisks indicate the level of significance of the differences between the indicated graph points and their corresponding controls at P values of ≤0.05 (∗), ≤0.01 (∗∗), and ≤0.001 (∗∗∗). P values were calculated by comparing mean CFU/monolayer values (not the relative percentages), and their corresponding standard deviations, in Student t tests.

Trypsin treatment selectively inhibited the invasiveness of virulent L. pneumophila.

The selective presence of Hsp60 on the surface of virulent strains of L. pneumophila may constitute an explanation for the much higher invasiveness of such strains in relation to avirulent ones (12, 22). We have shown that avirulent strains of L. pneumophila adhere well to HeLa cells (22), even in the presence of Hsp60 antiserum (see above), but not in the presence of OmpS antiserum (Fig. 1b). Therefore, we hypothesized that Hsp60 could play a role in adherence and invasion, while OmpS could play a role in adherence only. We tested this hypothesis by treating whole bacterial cells with trypsin. Because OmpS is highly resistant to proteolytic degradation, we predicted that trypsin treatment of virulent legionellae would not affect adherence but would significantly decrease invasiveness, mimicking the avirulent phenotype. Indeed, treatment of whole bacterial cells led to both a selective degradation of Hsp60 (Fig. 2a) and a selective inhibition of the invasiveness (but not adherence) (Fig. 2b) of the virulent strains. Two further observations seem important to emphasize here. First, the low level of Hsp60 proteolysis in whole 2064M cells was due to lack of exposure and not to an intrinsic resistance to trypsin (Fig. 2a, control). Second, the weak invasiveness of our nonvirulent strains was trypsin insensitive. OmpS was confirmed to be resistant to proteolysis (in trypsin-treated virulent or avirulent legionellae) by analysis of SDS-polyacrylamide gels stained with Coomassie blue or immunostained with anti-OmpS serum (not shown).

FIG. 2.

Effects of trypsin treatment on interactions of different strains of L. pneumophila with HeLa cells. (a) Immunoblot showing differential proteolysis of Hsp60 in whole cells of virulent 2064 and avirulent 2064M strains. Numbers above the lanes indicate the amounts (in micrograms) of trypsin added to ∼10⁹ whole bacterial cells (in a final volume of 100 μl) to effect the proteolysis of exposed proteins. Coomassie blue staining detected no apparent degradation of other proteins. The lane marked Control shows the degradation of Hsp60 in lysates of 2064M to which no trypsin inhibitor was added. (b) Levels of association and invasion of L. pneumophila strains treated with trypsin before infection of HeLa cells. For levels of significance, see the legend to Fig. 1.

Because the existence of a trypsin-sensitive invasion factor(s), different from Hsp60, could not be ruled out at this point, the specific role for Hsp60 in invasion was further assayed in competition assays with purified Hsp60.

Effects of purified Hsp60 on the adherence and invasiveness of virulent legionellae.

Preincubation of HeLa cells with different amounts of soluble Hsp60 (up to 200 μg per well) caused a highly significant reduction in the ability of Lp1-Vir to associate with or invade HeLa cells (Fig. 3a). A less apparent inhibitory effect was observed for the invasiveness of the virulent strain 2064, but in this case, a differential effect upon invasion and adherence further suggested a role for Hsp60 in the invasion of HeLa cells. Interestingly, a dose-dependent decrease of invasiveness (but not adherence) of Lp1-Vir was clearly observed in the receptor modulation assays (Fig. 3b). In the case of strain 2064, adherence was significantly inhibited, albeit not as strongly as the invasiveness of this strain (Fig. 3b). The inhibitory effects of substratum-bound Hsp60 were much more apparent than those caused by soluble Hsp60 and were achieved with much lower concentrations of purified Hsp60. These results suggested that specific receptors for Hsp60 mediate the association of HeLa cells with virulent L. pneumophila.

FIG. 3.

Effects of purified Hsp60 on interactions of our virulent strains of L. pneumophila with HeLa cells. (a) Competition for Hsp60 binding sites by soluble Hsp60 added to the HeLa cell supernatant before infection with the virulent strain Lp1-Vir (○) or 2064 (•). Cell association (solid lines) and invasion (broken lines) were determined in standard invasion assays. (b) Receptor modulation assays carried in the presence of different amounts of substratum-bound Hsp60. Symbols are as described for panel a. Points represent means ± standard deviations (vertical bars) of single experiments run in duplicate. Statistical significance was calculated in relation to controls run in the presence of soluble BSA (a) or substratum-bound BSA (b). Levels of significance are indicated as detailed in the legend to Fig. 1.

Association of latex beads with HeLa cells.

Hsp60-coated (but not BSA-coated) latex beads associated well with HeLa cells, as determined by bead quantitation via light microscopy (Fig. 4a). Interestingly, only Hsp60-coated beads effectively competed with our virulent strains of L. pneumophila for receptor sites on HeLa cells, mediating significant decreases in the ability of strains Lp1-Vir and 2064 to associate to or invade HeLa cells (Fig. 4b).

FIG. 4.

Interaction of HeLa cells with latex beads coated with Hsp60 or BSA. (a) Number of coated beads/cell in HeLa monolayers to which beads were added at a bead-to-HeLa cell ratio of 100:1 to 200:1. (b) Competition for available binding sites between Hsp60-coated beads (Hsp60) or BSA-coated beads (BSA) and our virulent L. pneumophila strains. Columns represent means + standard deviations (vertical bars) of single experiments run in triplicate (a) or in duplicate (b). In panel b, the percentages were calculated in relation to a common control to which no beads were added. Levels of significance are as detailed in the legend to Fig. 1 and refer to differences between the corresponding BSA and Hsp60 columns.

TEM showed that Hsp60 promoted bead uptake.

Because the cells processed for TEM and light microscopy were taken from the same monolayers, we roughly estimated that 65 to 75% of the Hsp60-coated beads shown in Fig. 4a, at the 3-h time point, were intracellular. That is, for each bead observed resting on the surface of a HeLa cell, two or three beads were observed inside the cell, surrounded by a membrane in very tight apposition to the bead surface (Fig. 5a). Frequently, an electron-dense line (Fig. 5b) defined the bead-membrane interface. In contrast to the above results, in the TEM specimens of HeLa cells to which BSA-coated beads were added, no intracellular beads were observed at the 3-h time point, and only a few beads were spotted overall. By 24 h after the addition of beads, virtually 100% of the Hsp60-coated beads were intracellular and still enclosed in tight endosomes, often associated with what appeared to be a defined type of vesicle (Fig. 5c). The tightness of the endosomes suggested that they had not fused with lysosomes or with other vesicles. In contrast, BSA-coated beads that were intracellular at the 24-h time point were often contained in enlarged endosomes and did not closely associate with the endosomal membrane (Fig. 5d).

FIG. 5.

Ultrastructural view of the interaction of coated beads with HeLa cells. Electron micrographs show intracellular Hsp60-coated beads (a to c) or BSA-coated beads (d) taken up by HeLa cells. Micrographs were taken from specimens fixed 3 h (a and b) or 24 h (c and d) after addition of the beads. Arrowheads in panel b point to the tight association between the endosomal membrane and the surface of beads. Arrows in panel d point to the loose endosomes containing BSA-coated beads. V, unidentified vesicle commonly found in association with endosomes containing Hsp60-coated beads. Bars represent 1 μm in all panels.

Assays with mature intracellular L. pneumophila.

Standard invasion assays confirmed that mature forms of strain 2064 (recovered from a lysate of infected HeLa cells) associated with HeLa cells up to 100-fold more efficiently than BCYE-grown 2064 (Table 1). These results are in agreement with previously reported observations with HeLa cell-grown (22) and amoeba-grown (9) L. pneumophila. Because significant amounts of Hsp60 are produced and released during the intracellular residence of L. pneumophila (15, 21, 31), we wanted to determine whether Hsp60 would be present on the surface of mature bacteria. Immunogold labeling of mature bacteria infecting HeLa cells showed that Hsp60 epitopes were quite abundant and had a prominent surface location (Fig. 6a), suggesting that Hsp60 may indeed play a role in the enhanced association of mature bacteria with HeLa cells. Interestingly, there were fewer OmpS epitopes than Hsp60 epitopes on infecting mature bacteria (Fig. 6b).

TABLE 1.

Association to and invasion of HeLa cells by BCYE-grown and HeLa cell-grown (mature intracellular bacteria) L. pneumophila 2064

Strain	Viable cell counts (CFU/monolayer)a
	Expt 1,b association	Expt 2c
		Association	Invasion
BCYE grown	(1.5 ± 0.65) × 103	(1.2 ± 0.4) × 104	5.0 × 103
HeLa cell grown	(1.6 ± 0.53) × 104	(2.0 ± 0.3) × 106	(4.2 ± 2.5) × 105

^aAssociation refers to adherent and intracellular bacteria; invasion refers exclusively to bacteria that survived treatment with gentamicin (100 μg/ml) and were assumed to be intracellular. 

^bInvasion assay run at a bacterium-to-HeLa cell ratio of 10:1. Inoculum was ∼10⁶ bacteria on a monolayer of ∼10⁵ HeLa cells. Results are shown as mean ± standard deviation from a single experiment run in duplicate. 

^cInvasion assay run at a bacterium-to-HeLa cell ratio of 20:1. Inoculum was ∼10⁷ bacteria on a monolayer of ∼5 × 10⁵ HeLa cells. Results are shown as mean ± standard deviation from a single experiment run in duplicate, except for the invasion sample from BCYE-grown bacteria, which represents a single determination. 

FIG. 6.

Immunolocalization of Hsp60 or OmpS epitopes in mature forms of L. pneumophila 2064 infecting HeLa cells. Ultrathin sections of HeLa cells infected with mature bacteria were immunolabeled with rabbit anti-Hsp60 serum (a) or rabbit anti-OmpS serum (b) and a secondary antibody (goat anti-rabbit immunoglobulin G) conjugated to 10-nm gold particles (Sigma Immunochemicals). Bars represent 0.1 μm in both panels. H, HeLa cells; B, mature bacteria isolated from infected HeLa cells.

DISCUSSION

We have used a previously characterized HeLa cell model (22) to test the hypothesis that the 60-kDa heat shock protein of L. pneumophila, which is displayed on the surface of virulent strains (21), plays a role in adherence to and invasion of nonphagocytic cells. Several lines of evidence, derived from five different experimental approaches, indicated that Hsp60 promoted binding to HeLa cells and subsequent internalization of both virulent L. pneumophila and coated latex beads. Our studies also identified a putative role for OmpS, a novel disulfide bond cross-linked outer membrane protein, in attachment but not invasion of HeLa cells. Of particular interest was the preliminary observation that endosomes containing Hsp60-coated beads did not fuse with secondary lysosomes or become associated with the endoplasmic reticulum, intracellular events that are well documented in macrophages for avirulent or virulent strains of L. pneumophila, respectively (15, 34, 53). The Hsp60-coated beads may prove useful in more rigorous studies aimed at dissecting the early events associated with organelle trafficking decisions.

Roles for heat shock proteins in bacterial attachment to host cells are unusual but not unprecedented. The Hsp70 of Haemophilus influenzae has been described as a new class of adhesin that binds to cell surface sulfoglycolipids (26). Also, Hsp60 chaperonins have been reported to mediate the attachment of Salmonella typhimurium to intestinal mucus (13), Helicobacter pylori to gastric epithelial membrane sulfatides (35), and Haemophilus ducreyi to HEp-2 cells (18). However, to our knowledge, the involvement of chaperonins as invasion factors constitutes an unprecedented virulence role for Hsp60. The receptor modulation assay suggested that the L. pneumophila Hsp60 exerts these novel functions through its interaction with specific receptors on the HeLa cell membrane. Indeed, the existence of specific receptors for the L. pneumophila Hsp60 has been previously documented in macrophages, where the Hsp60-receptor interaction triggers a signaling cascade mediated by protein kinase C that results in enhanced synthesis of interleukin-1β (46). The presence of Hsp60-specific receptors on host cells may be related to the surveillance function of the mammalian immune system, since Hsp60 is recognized as one of the key signatures of microbial infection (29, 58).

We have previously proposed that surface-associated Hsp60 promotes the establishment of intracellular L. pneumophila infections (21). A role for Hsp60 as an invasion factor is compatible with this idea, since the fate of intracellular L. pneumophila is determined during or immediately after host cell entry (48) by a factor or factors that are preexisting on virulent strains (29). The role of Hsp60 in adherence and invasion may not be exclusive, since the nonvirulent legionellae, which virtually did not display Hsp60 on their surface, were still able to attach well to HeLa cells except in the presence of OmpS antiserum. Avirulent bacteria are also efficiently taken up but do not replicate in macrophages, suggesting that Hsp60 is not required for phagocytosis by macrophages. We also observed that neither large concentrations of antibodies nor the simultaneous addition of Hsp60 and OmpS antisera completely abolished adherence and invasion of HeLa cells. However, we cannot exclude the possibility that surface-bound antibodies could have mediated low-level adherence and internalization through Fc receptors, as previously demonstrated for Chlamydia trachomatis in HeLa cells (45). Even in our competition assays with purified Hsp60, adherence-invasion was not completely abolished. The last observation and the early settlement of the intracellular fate of virulent and avirulent legionellae in macrophages (15, 34, 48, 54) or HeLa cells (22) suggest the existence of different entry pathways, perhaps mediated by different ligands. In any instance, it is common for intracellular bacterial pathogens to possess multiple adhesins and/or invasins as has been demonstrated for Yersinia spp. (36), Neisseria gonorrhoeae (24), and Listeria monocytogenes (11, 19, 28, 38). Because loss or blockage of surface-exposed Hsp60 significantly reduced the adherence and invasiveness of virulent strains, we have concluded that Hsp60 must be an important adhesin-invasin in L. pneumophila.

We considered the possibility that Hsp60 may be targeted to the cell surface as a complex with other proteins that mediate invasion, perhaps by playing an important role in cell surface organization or the proper display of surface proteins. However, the ability of recombinant Hsp60, which had been purified from an E. coli background, to promote cell adherence and internalization of Hsp60-coated beads (see below) argues against a supporting role for Hsp60 in invasion. Moreover, the abundance of surface-exposed and released Hsp60 in host cell-grown legionellae (Fig. 6 and reference 21) suggests its importance in pathogen-host interactions and may also explain the enhanced ability of mature legionellae to associate with host cells.

Purified Hsp60 promoted the uptake of coated beads by HeLa cells, a function recently documented for the Ipa complex, a well-recognized invasin of Shigella flexneri (39). HeLa cells efficiently internalized latex beads, on which the Ipa complex was immunoprecipitated. In contrast, BSA-coated beads were poorly internalized (39), as we also observed in our studies. BSA-coated latex beads are also commonly used in endocytic trafficking studies, since endosomes containing BSA-coated beads fuse with secondary lysosomes in mammalian cells (10). Therefore, it is noteworthy that Hsp60-coated beads apparently followed an endocytic pathway different from that of BSA-coated beads and remained in tight endosomes that came into close contact with vesicles without fusing with them (Fig. 5). Since L. pneumophila appears to replicate in a ribosome-studded phagosome surrounded by the endoplasmic reticulum (replicative phagosome) in HeLa cells (22), the inability of the Hsp60-coated beads to become enveloped by the endoplasmic reticulum suggests that additional factors or steps may be involved in formation of the replicative phagosome.

The early interactions of virulent L. pneumophila with monocytes or L929 cells are characterized by an up-regulation of Hsp60 synthesis, increased expression of Hsp60 at the bacterial cell surface, and release of the protein into the phagosome (15). Similar observations have been made for L. pneumophila in the HeLa cell model (22). These events correlate with the ability of virulent, but not avirulent, L. pneumophila strains to invade HeLa cells and to abrogate phagosome-lysosome fusion in host cells, which is one of the early steps of the degradative phagocytic pathway. Collectively, these observations warrant further attention to experimentally determine whether Hsp60 is capable of influencing the fate of endocytosed particles, modifying organelle trafficking, and/or altering the normal evolution of the degradative pathway, e.g., by phagosome-lysosome fusion.

Because the basal levels of Hsp60 are similar between the virulent and avirulent isogenic pairs of L. pneumophila, avirulent strains must be defective not in the synthesis of Hsp60 but in the ability to display this protein on the bacterial surface and/or respond to the interaction with host cells (15). Immunolocalization studies have shown that the Hsp60 of the avirulent strains 2064M and Lp1-AVir associates with the cell envelope (inner membrane, periplasm, and outer membrane) (unpublished results), but it is neither efficiently expressed on the cell surface nor released (15, 31) (Fig. 2a). Thus, avirulent strains appear to have a molecular defect that prevents completion of the release process. In this respect, it is known that virtually all sodium-tolerant avirulent L. pneumophila strains characterized to date harbor chromosomal mutations in either a 22-kb region or in the dotA/icmA locus (3, 50). Surprisingly, many of the genes within this virulence-related region are involved in transport of DNA during bacterial conjugation, and it has been suggested that the products of these genes most likely form a multimeric complex that may also participate in the transport of proteins (50, 56). It has recently been demonstrated that DotA, a cytoplasmic membrane protein, is required for early phagosome trafficking decisions but not for intracellular multiplication in macrophages (48). The implied role of the conjugation-protein secretion complex in early events associated with macrophage infection is certainly consistent with our findings that Hsp60 not only mediates invasion of HeLa cells but, perhaps more importantly, may play a role in altering organelle trafficking, a feature common to all permissive host cells. Thus, it will be interesting to determine whether this transport system, which appears to be distantly related to the ptl operon of Bordetella pertussis involved in the secretion of pertussis toxin (50, 56), is responsible for the mobilization of Hsp60 to the bacterial surface.